1 A LIQUID CONDENSED PHASE (LCP) OF CALCIUM CARBONATE AND ITS RELEVANCE TO THE POLYMER INDUCED LIQUID PRECURSOR (PILP) PROCESS: FUNDAMENTALS AND APPLICATION S By MARK ALAN BEWERNITZ A DISSERTATION PRESENTED TO THE GRADUATE SC HOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012
2 2012 Mark Alan Bewernitz
3 To God almighty, particularly the Holy Spirit for supplying the grace wisdom, and patience n eeded to complete this endeavor
4 ACKNOWLEDGMENTS I thank my parents, my brother, and my sister for their support me through out the years which has been so vital. I want to thank my wife for the c ounsel and wisdom, patience and love, she offered me throughout this process. I am blessed and graced to have been given such wonderful loved ones in my life. I d like to thank my advisor, Dr. Laurie Gower, for her mentorship, without which, this disserta tion would not have been possible.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF FIGURES ................................ ................................ ................................ .......... 7 LIST OF OBJECTS ................................ ................................ ................................ ....... 11 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTERS 1 INTRODUCTION TO CALCIUM CARBONATE PRECURSOR PHASES ............... 14 Nanocolloidal Microstructure is Evidence of a L iquid Precursor Coalescence Mechanism of Calcium Carbonate Biominerals ................................ ................... 14 Classical Nucleation Theory ................................ ................................ ................... 16 Calcium Carbonate Non C lassical Nucleation ................................ ........................ 17 Polymer Induced Liquid Precursor (PILP) Process: A Non Classical Means to Direct Biomineral Nucleation ................................ ................................ ............... 18 2 A METASTABLE, LIQUID CONDENSED PHASE OF CALCIUM CARBONATE .... 29 Introduction ................................ ................................ ................................ ............. 29 Materials/Methods ................................ ................................ ................................ .. 30 Generation of Super Saturated Solution in the Absence of Polymer Additive .. 30 Generation of Super Saturated Solution in the Presence of Polymer Additive ................................ ................................ ................................ ......... 31 Ca 2+ Electrode and pH Electrode Measurements ................................ ............. 31 Isothermal Titration Calorimetry (ITC) of Phase Transition .............................. 32 Nanoparticle Tracking Analysis (NTA) of Emergent Phase .............................. 32 Analytical Ultracentrifugation (AUC) ................................ ................................ 33 NMR, PFG STE, Spin Spin (T 2 ) Relaxation Time Measurement ...................... 33 Results ................................ ................................ ................................ .................... 34 Metastable Liquid Condensed Phase (LCP) ................................ ..................... 34 Experiments in the Presence of Polyaspartate ................................ ................. 45 Discussion ................................ ................................ ................................ .............. 50 3 CALCIUM CARBONATE LCP ................................ ................................ ................ 67 Introduction ................................ ................................ ................................ ............. 67 Removal of Counterion from the Mother Solution ................................ ............. 72 Activity Reduction Due to Ionic Shielding ................................ ......................... 74 Materials/Methods ................................ ................................ ................................ .. 78 Generation of Super Saturated Calcium Carbonate Solution ........................... 78
6 Ca 2+ Electrode and pH Electrode Measurements ................................ ............. 78 Nanoparticle Tracking Analysis (NTA) of Emergent Phase .............................. 79 NMR, Spin Spin (T 2 ) Relaxation Time Measurement ................................ ....... 79 Results ................................ ................................ ................................ .................... 80 Discussion ................................ ................................ ................................ .............. 82 Conclusion ................................ ................................ ................................ .............. 84 4 BIOMEDICAL APPLICATION OF THE PILP PROCE SS: cALCIUM CARBOANTE COATED EMULSIONS AND LIPOSOMES FOR CONTROLED RELEASE APPLICATIONS ................................ ................................ .................... 92 Introduction ................................ ................................ ................................ ............. 92 Materials/Methods ................................ ................................ ................................ .. 95 Emulsion Preparation ................................ ................................ ....................... 95 Liposome Preparation ................................ ................................ ...................... 95 Core Shell Microcapsule Synthesis ................................ ................................ .. 96 Fluorescence Imaging of Encapsulated Model Compounds ............................. 97 Polarized Light Microscopy for Characteriza tion of CaCO 3 Shell Crystallinity .. 97 Scanning Electron Microscopy (SEM) for Morphological Analysis ................... 98 X ray Diffraction (XRD) ................................ ................................ ..................... 99 Results ................................ ................................ ................................ .................... 99 Discussion ................................ ................................ ................................ ............ 103 Conclusion ................................ ................................ ................................ ............ 108 5 CONCLUSIONS AND OUTLOOK ................................ ................................ ........ 119 APPENDIX : SUPPLEMENTARY MATERIAL ................................ .............................. 129 LIST OF REFEREN CES ................................ ................................ ............................. 139 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 147
7 LIST OF FIGURES Figure page 1 1. Examples of Biomi nerals with non equilibrium morphologies. .............................. 23 1 2. Biominerals display evidence of colloidal assembly. ................................ ............. 24 1 3. The free energy of cluster formation vs. the cluster radius for nucleus formation. ................................ ................................ ................................ ........... 25 1 4. A comparison of calcite film topology when PILP phase is deposited on a silicon wafer patterned with a SAM (mercaptohex adecanoic acid SH(CH 2 ) 15 COOH, grid) vs. gold (interior square region). (A) Polarized light micrographs show PILP phase deposited on both the COO -terminated SAMs and on the bare gold surface. ................................ ................................ .. 26 1 5. Liquid precursor droplets obtained by Wolf et al. by means of sonic levitation of a supersaturated CaCO 3 solution. ................................ ................................ ...... 27 1 6. An SEM micrograph of the state of the CaCO 3 precipitation 1 m in after gently mixing 0.01 M solutions of CaCl 2 and Na 2 CO 3 ................................ .................. 28 2 1. The bound [Ca 2+ ] evolution and solution pH evolution during titration of Ca 2+ (aq) into 20 mM bicarbonate buffer with an in itial pH of 8.5. ............................... 53 2 2. The enthalpy of reaction during the titration of CaCl 2 (aq) into 20 mM carbonate buffer, pH 8.5. ................................ ................................ .................... 54 2 3. The emergent phase size distribution, according to nanoparticle tracking analysis (NTA), of the 20 mM bicarbonate, pH 8.5, solution at and after the phase transition (injection 10 and 13, respectively for a 6 mM CaCl 2 (aq) injection). ................................ ................................ ................................ ............ 55 2 4. Sedimentation coefficients (25 C) for the detected species in solution by means of analytical ultracentrifugation (AUC) for the injections indicated in the inset. ................................ ................................ ................................ ............. 56 2 5. A 1 D 13 C spectrum of the solution after the phase transition (17 th injection of 6 mM CaCl 2 (aq) into 20 mM carbonate buffer, pH 8.5). ................................ ....... 57 2 6. A comparison betw een the 13 C 1 D NMR spectrum for bicarbonate solution (control), the bicarbonate solution with CaCl 2 (aq) prior to the phase transition, and the bicarbonate solution with CaCl 2 (aq) after the phase transition. ................................ ................................ ................................ ............ 58 2 7. The results of the CPMG T 2 relaxation measurement and the 13 C PFG STE self diffusion measurement of the deconvoluted NMR peaks. ............................ 59
8 2 8. The [Ca 2+ ] profile of a system where 300 mM bicarbonate buffer, pH 8.5 was 2 PAsp (no PAsp for control). ................................ ................................ ................ 60 2 9. The pH evolution of a 10 mM CaCl 2 solution containing 20 g/mL Pasp due to the punctuated titration of 300 mM bicarbonate buffer, pH 8.5. .......................... 61 2 10. A comparison between the 1 D NMR spectra of bicarbonate buffer with polyasparti c acid sodium salt and the separated bicarbonate biased ions. ........ 62 2 11. The time evolution of a 1 D NMR spectrum of the phase separated PILP containing solution shown in Figure 2 10. ................................ .......................... 63 2 12. The results of the CPMG T 2 relaxation measurement and the 13 C PFG STE self diffusion measurement of the PILP phase (suspected polymer stabilized LCP), bulk solution, and bicarbonate buffer with polyaspartic acid (control). ...... 64 2 13. An o verview of the energetics (not to scale) of calcium carbonate precipitation from supersaturated solution, putting LCP into a global context with earl ier findings. ................................ ................................ ................................ .............. 65 3 1. Example of a phase diagram describing the partitioning of species A and B, with an upper consolute temperature limit (left) and a lower consolute temperature limit (right ). ................................ ................................ ..................... 86 3 2. The evolution of [Ca 2+ ] Bound during titration of 10 mM CaCl 2 (aq) into 20 mM bicarbonate buffer, pH 8.5, as measured by a calcium ion selective electrode. 87 3 3. The NTA obtained size profile of the LCP droplets in a 20 mM bicarbonate solution, pH 8.5, with an initial concentration of Cl of 20 mM after 17 injections of CaCl 2 (aq). ................................ ................................ ...................... 88 3 4. The results of the NMR T 2 relaxation experiment of the 20 mM bicarbonate solution, pH 8.5, with an initial concentration of Cl of 20 mM after 17 injections of CaCl 2 (aq). ................................ ................................ ...................... 89 3 5. The general solution of Equation 3 13 and its use to measure the partitioning of ions between the LCP phase and the mother solution. ................................ .. 90 3 6. By expressing the T 2 relaxati on data in terms of area of the Gaussian models used to deconvolute the data, the fraction of ions in the LCP phase at the beginning of the experiment ( = 0) is estimated. ................................ ............... 91 4 1. A schematic describing the synthesis of CaCO 3 core shell microcapsules. ........ 110 4 2. Light microscopy images of CaCO 3 coated emulsion droplets and liposomes. ... 111 4 3. Scanning Electron Microscopy (SEM) images of the core shell microcapsules. 112
9 4 4. Scanning Electron Microscopy (SEM) images of intact and intentionally fractured CaCO 3 coated emulsion microcapsules. ................................ ........... 113 4 5 Potential tailoring of microcapsule crystallinity as determined by polarized light microscopy. ................................ ................................ ................................ ...... 114 4 6. X ray diffraction (XRD) analysis of the CaCO 3 shells of liposome derived microcapsules. ................................ ................................ ................................ .. 115 4 7. Confocal fluorescence microscopy of microcapsules from CaCO 3 coated emulsion droplets with Nile Red fluorescent dye entrapped within the oily interior of the emulsion. ................................ ................................ .................... 116 4 8. Confocal fluorescence microscopy demonstrating the entrapment capability of the CaCO 3 coated liposomes. ................................ ................................ .......... 117 4 9. SEM micrographs of water and acid dissolved microcapsules. ........................... 118 5 1. Chemical shift anisotropy can affect th e width and shape of the NMR spectrum. ................................ ................................ ................................ .......... 126 5 2. NMR experiments at various spin rates and varying sample and the bulk magnetic field, B 0 ) other than the magic angle of 54.7 could lead to valuable information regarding the tumbling rate and complex structure of the carbonate/bicarbonate solutes that form during LCP emergence. ................................ ................................ ................................ ....... 127 5 3. A calcium bicarbonate complex would have a large amount of rotational and protonization freedom making it entropically unfavorable to solidify. ................ 128 A 1. NMR pulse sequence used to acquire the 13 C diffusion data. ............................. 129 A 2. A comparison between the same titrations conducted with injection nozzle (micropipette tip) in the solution (as conducted in this paper) vs. with the nozzle just above the solution. ................................ ................................ .......... 130 A 3. An example of our modeling technique. ................................ .............................. 131 A 4. induced attenuation of the 13 C peak during the Carr Purcell Meiboom Gill (CPMG) T 2 relaxation experiment of the bicarbonate peak of the 20 mM, pH 8.5, bicarbonate control solution titrated with injections of nanopure water. 132 A 5. The real data and the modeling of the 13 C diffusion during the 17 th injection without polyaspartic acid additive. ................................ ................................ .... 133 A 6. induced attenua tion of the 13 C peak during the Carr Purcell Meiboom Gill (CPMG) T 2 relaxation experiment of the bicarbonate peak of the 20
10 g/mL M.W. 14,000 g/mol Pasp control solution titrated with 30.15 mM bicarbonate buffer, pH 8.5. ................................ ................................ ............... 134 A 7. The gradient induced attenuation during the Pulse Field Gradient Stimulated Echo (PFG STE) T 2 relaxation experiment of the bicarbonate peak of the 20 g/mL polyaspartic acid sodium salt (M.W. 14,000 g/mol) control s olution titrated with 300 mM bicarbonate buffer, pH 8.5, yielding a final concentration of 18 g/mL pasp and 30.15 mM bicarbonate. ................................ ................. 135 A 8. induced attenuation of the 13 C peak during the Carr Purcell Meiboom Gill (CPMG) T 2 relaxation experiment of the bicarbonate peak of the 10 mM CaCl 2 titrated with 17 injections of 30 0 mM bicarbonate buffer, pH 8.5. ..................... 136 A 9. The gradient induced attenuation during the Pulse Field Gradient Stimulated Echo (PFG STE) diffusion measurement experiment of the bicarbonate peak of the 20 g/mL polyaspartic acid sodium salt (14,000 g/mol M.W.) solution titrated with 300 mM bicarbonate buffer, pH 8.5, yielding a final concentration of 18 g/mL pasp and 30.15 mM bicarbonate buffer. ................................ ....... 137
11 LIST OF OBJECT S Object page A 1 First emergent phase experiment, 6mM CaCl 2 into 20 mM bicarbonate buffer, 10 th injection ................................ ................................ ................................ ..... 138 A 2 Second emergent phase experiment, 6mM CaCl 2 into 20 mM bicarbonate buffer, 10 th inje ction ................................ ................................ .......................... 138 A 3 10 mM CaCl 2 into 20 mM bicarbonate buffer with 20 mM Cl initial concentration, 17 th injec tion ................................ ................................ .............. 138
12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doc tor of Philosophy A LIQUID CONDENSED PHASE (LCP) OF CALCIUM CARBONATE AND ITS RELEVANCE TO THE POLYMER INDUCED LIQUID PRECURSOR (PILP) PROCESS: FUNDAMENTALS AND APPLICATIONS By Mark Alan Bewernitz December 2012 Chair: Laurie Gower Major: Biomedical Eng ineering I nvertebrate organisms display a high degree of control over the deposition of their calcium carbonate biominerals. Work in our group has led to the hypothesis that c harged polyelectrolytes, like acidic proteins, may be employed by organisms to direct crystal growth through an intermediate liquid phase in a process called the polymer induced liquid precursor (PILP) process. Recently, it has been proposed that calcium carbonate crystallization, even in the absence of any additives, follows a non classical, multi step crystallization process by first associating into prenucleation clusters In contrast, others have found evidence of liquid liquid phase separation occurring in calcium carbonate solutions (even without addition of polymer) The goal of this dissertation was to determine what species might be present under reaction conditions used for the PILP process, and whether or not these species might play a role in the PILP process. To achieve this goal, w e have employed Ca 2+ ion selective ele ctrodes, pH electrodes, isothermal titration calorimetry, nanoparticle tracking analysis, 13 C T 2 relaxation measurements, and 13 C PFG STE diffusion NMR measurements. These studies provide evidence that, in the absences of additives, and at near neutral pH a
13 liquid pr ecursor phase does indeed exist in the form of a bicarbonate rich liquid condensed phase (LCP). The data further shows that addition of acid ic polymer promote s /stabilize s the LCP in a distinct and pronounced fashion providing a mechanistic u nderstanding of the so called PILP process. As a demonstration of the utility of the biomimetic approach, we have used the PILP process to synthesize m icron sized core shell particles for potential use in biomedical drug delivery applications. Calcium c arbonate coatings were deposited on the curved surfaces of micro droplets of either emulsions or liposomes to make non toxic, inexpensive, biodegradable microcapsules which can encapsulate chemicals of interest within the fluidic core. T he microcapsules c an be dried down to a powder, providing a convenient means of storage and transport of entrapped agents The dissolution of the mineral shell is pH dependent, enabling a rapid triggered release of the active agent. These CaCO 3 microcapsules could make ide al systems for drug/chemical delivery in their own right, or could provide a convenient system for which polymeric or other inorganic coatings could be further applied In addition to drug delivery, other c ontrolled release applications could be envisione d
14 CHAPTER 1 INTRODUCTION TO CALCIUM CARBONATE PRECURSOR PHASES Nanocolloidal Microstructure is Evidence of a Liquid Precursor Coalescence Mechanism of Calcium Carbonate Biominerals Calcium carbonate is one of the most abundant minerals on Earth 1 4 Calcium carbonate is very important in many industrial processes such as in filler for papermaking and other composites, in carbon dioxide storage for environment al concerns, 5 and in biomedical applications such as drug deliv ery, 6 due to its low cost and biocompatibi lity 7 From an engineering standpoint, understanding t he biological formation and regulation of calcium carbonate biominerals is very intriguing because b io logically derived calcium carbonate exhibits an incredible array of crystals with complex and non faceted morphologies. In nature, it is usually found as the mineral component of the skeletal structures in invertebrates, such as in the spine of sea urchins 8 and in the nacre of mollusk shell s 9 but it plays a role in the otoliths and bones of vertebrates as well 10 11 Figure 1 1 shows images of a sea urchin spine and sheet nacre, both of which demonstrate the incredible extent to which organisms can control and utilize calcium carbonate. Sheet nacre, shown in Fig 1 1C, is comprised of thin tablets of aragonite, which is a non equilibrium polymorph (calcite should be favored at room temperature), displaying non equilibrium morphology (synthetic aragonite forms needles) embedded in between thin layers of an organic matrix. 9 12 20 These tablets also have a nanoscale structural motif, exhibiting a nanocolloidal texture. 3 The sea urchin spine (see Figure 1 1A and 1 1B ) is a large, microporous biomineral with a non equilibrium molded morphology with smoothly curved surfaces. In addition, the microporous structure produces a calcite single crystalline diffraction pattern even though it also has a nanocolloidal texture. Some have described this texture as
15 mesocrystalline, 21 28 c olloidal crystals composed of individual nanocrystals that are aligned in a common crystallographic o Niederberger et al.. 29 Similar colloidal internal structures, shown in Figu re 1 2, have been found in other biomineral structures such as in calcareous sponge spicules 30 and octocoral sclerites, 31 which also diffract as a single crystal. The surface of the biominerals shown in Figure 1 2 reveal a mesocrystalline internal arrangement of partially coalesced and continuous, yet outlined and non homogenous, nanogranular constituents. This data hints that biominerals form t hrough a colloidal precipitation mechanism. An element of biological control over mineral products is also believed to be due to cellular manipulation of an amorphous calcium carbonate (ACC) precursor phase which may act as a reactive intermediate in ge nerating complex functional materials 32 33 This may be due to the fact that ACC is not subject to constraint s of crystal lattice equilibrium morphologies. 3 M anipulation of the ACC precursor by templating techniques and altering environmental influences can direct the formation of a crystalline calcium carbonate product which may be calcite, aragonite, and even vaterite 34 (although the latter case often serves as a crystalline intermediate). 35 38 E vidence of an ACC precursor has been reported in th e sea urchin tooth 39 sea urchin spine, 40 mollusk shell prismatic layer 16 larval mollusk shell 41 sea urchin larval spicule during formation 42 and on the surface of mature sheet nacre at the protein/mineral interface. 43 Ac 44
16 energetically unfavorable ACC may form initially and then ripen to form a more stable polymorph, but the mechanism by which this ACC phase forms is not fully understood. The evidence presente d above suggests t hat there is a mechanism where biomineral crystals are formed from accumulation and/or a coalescence of colloidal ACC precursors a non classical nucleation and growth mechanism. Somehow, there is a robust ACC nucleation mechanism that can occur at biolo gically relevant conditions that generates large amounts of energetically unfavorable amorphous mineral that is molded into non equilibrium morphologies by precipitation within vesicles or on organic matrices. Organisms require an incredible amount of con trol over the timing, as well as location and phase of the non classical solid calcium carbonate nucleation. This has led many researchers to investigate the prenucleation behavior of supersaturated calcium carbonate, and how biological systems might infl uence the eventual solid critical nucleus formation. Classical Nucleation Theory Classical nucleation theory (CNT) is a n old concept 45 that has been studied and improved upon over decades. It is based upon the premise that the energetic driving force of form ation and growth of a crystal nucleus is derived from the bulk energy (volume) of the nucleus and that the destabilizing force is due to the concurrent emergence of an unfavorable surface interfacial energy (surface area) with respect to the mother solutio n. Below a certain critical size, the destabilizing influence of the emergent interface is dominant to the stabilizing influence of the bulk energy. At this point the energetically unfavorable nucleus exists very transiently due to random thermal fluctu ations but quickly dissolves again. However, as the concentration of the ion and counterion are increased above the saturation point for the mineral, the average
17 size of the transient nuclei can grow to a size where the stabilizing bulk energy term super sedes the destabilizing interfacial term and a so called critical nucleus is formed. Further growth is continued by ion by ion addition A f igure demonstrating the relationship between stabilizing bulk energy and destabilizing interfacial energy and how this leads to a critical nucleus radius is shown in Figure 1 3 The nucleation event requires three 1. The pre nucleus and critical nucleus must have the same structural arrangement and chem istry of the final mineral product it initiates. The only difference is size. 2. Nucleation occurs in a single step. 3. The fluctuating pre critical nucleus grows by monomeric growth of the constituents (i.e., ion by ion, molecule by molecule, etc.) Calcium C arbonate Non Classical Nucleation Traditionally it was assumed that the crystallization of CaCO 3 proceeds via a classical nucleation and growth mechanism. R ecently, it has been found that CaCO 3 nucleation occurs through a non classical process where a sta ble prenucleation solute interaction, called a prenucleation cluster (PNC) forms. Gebau er et al. demonstrated that Ca 2+ and CO 3 2 ions associate into clusters of ions (having a typical diameter on the order of ca. 2 3 nm) in undersaturated, saturated and supersaturated conditions. 46 The presence of PNC s has been corroborated by cryo TEM experiments in solutions saturated with respect to calcite. 47 Evidence has been obtained in the form of computer simulations that suggest that CaCO 3 2+ and CO 3 2 ions. 48 Th is d ynamically ordered liquid like oxyanion polymer Gale et al. 49 further supports the notion that CaCO 3 nucleation may occur through a non classical nucleation mechanism due to non monomeric (PNC) growth of the
18 anging from 9.0 to 10.0 and has been reported to be s table with respect to the initial ions in solution, prior to the formation of a metastable solid nucleus of ACC. 50 Gebauer et al. demonstrat ed that structure that lead to the ACC nucleation. 51 This PNC description initially sounds like an ideal mechanism for generating direction of ACC to a specific polymorp 2 3 nm size scale, which is much too small to account for large scale molten morphologies of calcium carbonate found in biominerals. Furthermore, molten morphologies do not occur in calcium carbonate formed without additives, even though in any given conformation is on the order of nanoseconds. 48 This leads some in the field to question whether this is truly a non classical phenomenon consideri ng that classical nucleation also allows for ions to briefly form unstable clusters through thermal fluctuations. Even though PNCs were an interesting finding, there seemingly must be some other non classical means utilized by nature to generate the biomi nerals with non equilibrium morphologies. Polymer Induced Liquid Precursor (PILP) Process : A Non Classical Means to Direct Biomineral Nucleation It has been demonstrated with in vitro model systems that non equilibrium crystal morpholog ies resembling biomi nerals can be reproduced by the inclusion of negatively charged polyelectrolytes (that are believed to e mulate the acidic proteins of biominerals)
19 during the precipitation process, such as was proposed by our group for the polymer induced liquid precursor (PILP) process where a pseudomorphic transformation of a fluidic ACC precursor can lead to crystals with non equilibrium morphologies 52 53 During the PILP process, CaCO 3 appears to nucleate through a multistep step process, where the polymer (anionic polyelectrolyte, often polyaspartic acid) associates with Ca 2+ and CO 3 2 ions to form an intermediate liquid ph ase prior to solid nucleation. The films is because the precursor nanodroplets meld together to form smooth mineral products that lack the facets found in crystals grown by the classical crystallization process. Figure 1 4 shows an illustration and micrograph evidence of the PILP process. Optical micrographs and AFM data in the figure demonstrate the colloidal coalescence property of the liquid precursor. Additionally, it tends to have a memory of the PILP of a nanocolloidal texture which resembles the nanocolloidal texture seen in biominerals. (compare Figure 1 4 to Figure 1 2). It is clear that the proteins involved in biomineralization play an important role in modulating the crystallization process to a non classical pathway, and that this can be nicely emulated in vitro using simple polypeptides. It is difficult, however, to theo retically explain how a true liquid metastable phase can be induced by the presence of polymer while assuming that, in the absence of polymer additives, the nucleation follows the classical view of crystallization. There has been considerable evidence re cently that CaCO 3 transitions through a liquid precursor non classical nucleation pathway in the absence of additives as well. 35
20 38 54 By analyzing an acoustically levitated droplet of saturated CaCO 3 (with respect to calcite) at a pH of 6.3, Wolf et al. 54 demonstrated that a liquid phase of CaCO 3 is formed in supersaturated conditions that arise from an increase in pH upon outgassing of CO 2 The liquid precursor phase was observed to form wi th droplet diameters of up to several hundred nanometers. Figure 1 5 is a TEM micrograph of these droplets. Rieger et al. 35 quenched and freeze captured a solution after inducing precipitation by rapidly mixing CaCl 2 (aq) and Na 2 CO 3 TEM to visualize a presumed liquid CaCO 3 precursor up to 2 m in diameter. A figure of the results of the freeze capture is shown in Figure 1 6. Faatz et al. 38 used a CO 2 outgassing technique at pH 7 to demonstrate that ACC with diameters of hundreds of nanometers can be obtained through a suspected liquid liquid phase separation prior to solidification into the amorphous state. Figures 1 5 and 1 6 show a fluidic CaCO 3 precursor phase that has been described by Rieger et al. 35 3 phase and by Wolf et al. as a colloidal liquid precursor of CaCO 3 54 These results are further evidence that CaCO 3 nucleates through a multistep, non classical pathway which begins with the formation of a liquid precursor. Evidence of non classical multi st ep mechanisms is not limited only to CaCO 3 mineralization, but can occur in other metal carbonates, 38 54 56 in ca lcium oxalate systems 57 p hosphate coordinated systems, 58 62 as well as in organics, such as bi opolymers, 63 64 amino acids 65 68 and organic pigments. 69 Although not well known, t he concept that the ions in a supersaturated calcium carbonate solution may transition through a liquid phase prior to solid nucle ation is not
21 phase, but the one that is closest in free energy to the parent . 44 expande d on this proposed principle to include the concept of a metastable liquid phase when a substance begins to separate from a solution, so making its first appearance as a 70 Such a process would be classified as non classical nucleation because it would occur as a cascade of phases from dissolved ions (gaseous) to an ion condensed phase dropl et (liquid) to a solid critical nucleus (solid), rather than a singular event. However, direct evidence of this was lacking and the concept was lost for many years until it was recently resurrected. In the more recent works by Wolf, 54 Rieger, 35 55 and Faatz, the detected liquid precursor in their experiments appear to be metastable with respect to the phases that occur at later stages, and that the structural transitions toward a solid, ACC like form follows a downhill energetic sequence, similar to the dehydration and subsequent crystallization of ACC. 71 As these studies emerged, we begin to wonder, could the PILP mechanism be a derivative of this very fundamental non classical calcium carbonate by the polymer in the PILP mechanism the same liquid phase that forms in the absence of additives ? Is the role of the polymer to induce a liquid phase, or to stabilize an existing one? Are the PNCs involved in the nucleation pathway? Answering these que stions is a fundamental focus of this report. If the PILP mechanism is correlated with the liquid precursor phenomenon in the absence of additives, the theory is attractive because it successfully reconciles many of the shortcomings of other proposed biol ogical mechanisms of crystallographic control. The PILP process simultaneously accounts for nucleation inhibition in the bulk solution,
22 and nucleation promotion in the PILP phase. The coalescence and subsequent solidification of PILP precursor droplets ex plains the molten, non equilibrium morphologies of many biominerals, the nanocolloidal texture, as well as the embedded proteins found in most CaCO 3 biominerals. It would be a robust mechanism that is dependent on an intrinsic fundamental process rather t han a sensitive one, such as protein temptation, that would be at the mercy of something as simple as a single genetic mutation. The fundamentals related goals of the collective research results presented in this disse rtation are to demonstrate (1) that c alcium carbonate transitions through a bicarbonate rich, metastable, liquid condensed phase (LCP) prior to solid nucleation even in the absence of additives ; (2) that calcium carbonate polymer induced liquid precursor (PILP) appears to be comprised of poly mer stabilized LCP ; (3) that s pectator ions, such as chloride, effect the extent of formation of LCP. From an engineering standpoint, calcium carbonate minerals are very attractive potential materials for biomedical applications. CaCO 3 is biocompatible, f ormed under benign conditions, and is relatively inexpensive. The applications related goal of this report is to demonstrate an application of the PILP directed non classical calcium carbonate mechanism by synthesizing novel core shell microcapsules by me ans of the PILP process.
23 Figure 1 1. Examples of Biominerals with non equilibrium morphologies. A) A scanning electron micrograph of the cross section of a spine from the urchi n Arbacia tribuloides s hows the typical microporous architecture. The structure diffracts as a single crystal of calcite. B) Higher magnification of the internal spine in part A shows the morphology as being a molded and continuous mineral phase which lacks facets. C) Cryo SEM images of a fracture surface of nacre from t he top most 5 layers (growth front) of Pinctada margaritifera shell The tablet morphology is energetically unfavorable for aragonite. (A and B) Reprinted with permission from X. G. Cheng and L. B. Gower, Biotechnology Progress 2006, 22 141 149. Copyw right 2006 American Chemical Society. (c) F. Nudelman, B. A. Gotliv, L. Addadi and S. Weiner, Journal of Structural Biology 2006, 153 176 187. r eproduced by permision of the Royal Society of Chemistry.
24 Figure 1 2. Biominerals display evidence of colloidal assembly. A ) Optical micrograph of the center of a large triactine spicule imaged in reflected light with the image focus beneath the spicule surface to show th e finely segmented features that lie perpendicular to the long axes (a1*,a2*,a3*) B ) Phase image of the surface of the a1* region shown in part a reveals a nanogranular texture, suggesting a colloidal assembly growth mechanism. C ) AFM deflection image of the surface of a c alcitic sea urchin skelet on trabeculae. The nano structure sugg ests a colloid coalescence mechanism of formation ( A and B ) reprinted from Journal of In organic Biochemistry, 100, I. Sethmann, R. Hinrichs, G. Worheide and A. Putnis, Nano cluster composite structure of calcitic sponge spicules A case study of basic characteristics of biominerals, 88, 2006 with permission for Elsevier. (c) reprinted from Micron, 39, I. Sethmann, G. Worheide and A. Putnis, Structure and composition of calcareous sponge spicules: A review and comparison to structurally related biominerals 209 228, 2008 with permission for Elsevier. A C B
25 Figure 1 3 The free energy of cluster formation vs. the cluster radius for nucleus formation As the transient clusters grow, the bulk stabilization (green curve) overcomes the interfacial destabilization ( red curve) at a critical radius size. Beyond this size, the growth of the nucleus is energetically favorable.
26 Figure 1 4. A c omparison of calcite film topology when PILP phase is deposited on a silicon wafer patterned with a SAM (mercaptohexadecanoic acid SH(CH 2 ) 15 COOH, grid) vs. gold (interior square region). (A) Polarized light micrographs show PILP phase deposited on both the COO -terminated SAMs and on the bare gold surface. (B) and (C) AFM images scanned across the (as indicated with the blue boxes in A) within each patterned region (left, 3 colloidal surface, showing colloidal particle morphology with different size Reprinted with permission from Y. Y. Kim, E. P. Douglas and L. B. Gower, Langmuir, 2007, 23, 4862 4870 Copyright 2006 American Chemical So ciety.
27 Figure 1 5. Liquid precursor droplets obtained by Wolf et al. by means of sonic levitation of a supersaturated CaCO 3 solution. A B ) TEM of liquid precursor droplets after 400 seconds of development. C D ) SEM of the dried and solidifie d solution. Solid mineral has a colloidal structure with some traces of calcite rhombahedrons (circle). The dried and solidified cluster of liquid droplets has a colloidal structure similar to the poorly coalesced PILP phase Scale bars: A ) 500 nm, B ) 20 0 nm, C ) 20 m, D ) 10 m Reprinted with permission from S. E. Wolf, J. Leiterer, V. Pipich, R. Barrea, F. Emmerling and W. Tremel, J. Am. Chem. Soc. 2011, 133 12642 12649 Copyright 20 11 American Chemi cal Society. D C A B
28 Figure 1 6. An SEM micrograph of the state of the CaCO 3 precipitation 1 min after gently mixing 0.01 M solutions of CaCl 2 and Na 2 CO 3 Rieger et al. 35 obtained this image and described this as a hydrated CaCO 3 precursor phase to solid ACC. The partial coalescence and distinct domains are similar to the poorly coalesced PILP phase J. Rieger, T. Frechen, G. Cox, W. Heckmann, C. Schmidt and J. Thieme, Faraday Discussions 2007, 136 265 277. reproduced by permision of the Royal Society of Chemistry.
29 CHAPTER 2 A METASTABLE, LIQUID CONDENSED PHASE OF CALCIUM CARBONATE Introduction Our group has primarily focused on the polymer induced liquid precursor (PILP) process and its relevance to CaCO 3 biomineralization. In this case, the PILP phase exists long enough to be manipulated into non equilib rium morphologies, the hallmark of biominerals. Thus, it is highly desirable to understand what the mechanistic role of the polymer is in this process in light of these other new findings of various liquid precursor phases in the absence of additives. For example, rather than the polymer interacting with ions, as we had originally assumed, one must now consider if the polymer might actually be interacting with any of these other species that have been detected in CaCO 3 solutions. 35 38 54 Thus, the focus of this chapter can be expressed in two parts: ( 1) Without po lymer additives, can the CaCO 3 liquid precursor be detected at a more neutral and biologically friendly pH of 8.5? ( 2) Does PILP form through polymer interaction with a pre 72 or by some other interaction? Given this focus, we carried out studies o f the early stages of precipitation in a fashion similar to Gebauer et al. except that we conducted our studies at pH 8.5 and allowed the ever increasing supersaturation to evolve by using punctuated injections of aqueous CaCl 2 into bi carbonate buffer all owing for time between injections to allow for system equilibration. At various points in the evolution, we analyzed the state of the solution using a Ca 2+ ion selective electrode, Isothermal Titration Calorimetry (ITC), Nanoparticle Tracking Analysis (NT A) light scattering Analytical Ultra Centrifugation (AUC) and carbon specific nuclear magnetic resonance ( 13 C NMR) spectroscopic techniques such as Carr Purcell Meibloom Gill (CPMG) T 2
30 measurement and Pulsed Field Gradient Stimulated Echo (PFG STE) Diffus ion NMR. It should be noted that we did not modulate pH during the supersaturation evolution as did Gebauer et al.. The PILP generating technique used in our lab that we wished to model does not maintain a constant pH. Rather, we allowed the pH to evolv e with the titration and monitored it throughout the titration. Using these methods at a moderate pH of ~8.5, we discovered a new phase transition of a liquid condensed phase (LCP) that occurs at a critical concentration of bound calcium. We then used th e same techniques with polymer (polyaspartic acid sodium salt) present to determine if the formation of polymer induced liquid precursor (PILP) phase is indeed a polymer stabilized LCP phenomenon. Ma terials/Methods Generation of Super Saturated Solution in the Absence of Polymer Additive Using a micropipette, 20 mM sodium carbonate (Fisher) was titrated into a 20 mM sodium bicarbonate (Fisher) solution to generate a 20 mM, pH 8.5 carbonate buffer solution. Calcium chloride (Fisher) solution with a conce ntration of 4.5, 6, or 10 mM was titrated into 29 mL of carbonate buffer, which was stirred at 100 rpm using a magnetic stir bar. The volumes of titration were 200 L each except for the first injection which was only 40 L to account for infinite dilution phenomenon. Titrations were injected at an approximate rate of 20 L /sec immediately over the rotating stir bar to ensure adequate mixing The titrations were made in a punctuated fashion, 2 minutes of constant pH, and [Ca 2+ ] Free measurements were acqui red before injecting more CaCl 2 (aq). The solutions were prepared with nanopure water and all were filtered using a 0.22 m Millipore syringe filter prior to any titrations. Each of these titrations
31 was conducted in triplicate and the error expressed in the results is plus or minus two standard deviations. Generation of Super Saturated Solution in the P resence of Polymer Additive Using a micropipette, 300 mM sodium carbonate was titrated into a 300 mM sodium bicarbonate solution to generate a 300 mM, pH 8 .5 carbonate buffer solution. This solution was then titrated into 29 mL of a stirred 10 mM CaCl 2 solution which may mL polyaspartic acid sodium salt (monodispersed, Alamanda Polymers), depending on the experiment. The volumes of titration were 200 L each except for the first injection which was only 40 L matching the titration conditions for the non additive experiments. Titrations were injected at an approximate rate of 10 20 L /sec, always over the rotating stir bar to ensu re adequate mixing and to minimize the formation of strong concentration gradients that might affect the experiments. The titrations we re made in a punctuated fashion; 2 minutes of constant pH, and [Ca 2+ ] Free measurements were acquired before injecting mo re titrant. The injections were made using disposable micropipette tips that were disposed of after each injection and replaced prior to a new injection. Each experiment was conducted in triplicate, including the control, and the error shown i s plus or m inus two standard deviations. Ca 2+ Electrode and pH Electrode Measurements The free Ca 2+ concentration, [Ca 2+ ] Free in the titrated solution was obtained using a Ca 2+ ion selective electrode (Radiometer Analytical, ISE K Ca, E11M006) in conjunction with a reference electrode (Radiometer Analytical E21M009). A calibration standard curve for calculating free Ca 2+ concentration was generated by titrating the experiment appropriate concentration of CaCl 2 (aq) into nanopure water which had been brought to pH 8. 5 by the addition of trace amounts of NaOH (Fisher) (aq). The pH
32 evolution of the titration was obtained using a standard pH electrode. Both pH and [Ca 2+ ] Free values had to remain constant for at least 2 minutes of mixing before adding another titration injection to verify that the solution equilibrated and that solid nucleation had not yet occurred. It is important to note that we are not quantitatively accounting for CO 2 net diffusion out of the solutions during our experiments. Isothermal Titration Calorimetry (ITC) of Phase Transition All measurements were made using a VP ITC (MicroCal Tm ). Carbonate/bicarbonate buffer and CaCl 2 (aq) was generated as described above. The injection of CaCl 2 (aq) into the reservoir carbonate/bicarbonate buffer was ma intained at exactly the same ratio as the titration experiments to give exact punctuated enthalpies that correspond to the titration experiment. The experiments were conducted at a controlled temperature of 298 Kelvin. The rest time between injections wa s adjusted to allow for complete thermodynamic equilibration between the reaction vessel and the reference vessel. A stir speed of 180 rpm was used for all the experiments. A reference power of 2 cal/sec was used due to the very subtle, endothermic natu re of the reaction. The solutions were not degassed due to the fraction of CO 2 that is in equilibrium within the carbonate/bicarbonate buffer at pH 8.5. No bubbling phenomena were observed during the course of the ITC experimentation. Control enthalpy p rofiles of CaCl 2 (aq) injections into water, water injections into bicarbonate buffer, and water injections into water were subtracted from the raw data to normalize the data. Nanoparticle Tracking Analysis (NTA) of Emergent Phase The number count and the hydrodynamic radius of the emergent phase droplets were analyzed using the NTA light scattering technique. Samples were analyzed using an LM20 analyzer (Nanosight TM ) and the data was process ed using an NTA analytical
33 software suite (Nanosight TM ). Samples for analysis were prepared as described in the ection. 0.3 mL of sample was used in each analysis. Analytical Ultracentrifugation (AUC) AUC was performed on an Optima XL I (Beckman Coulter, Palo Alto, CA) using the Rayleigh interference optics at 60,000 rpm and 25 C. The experiments were performed in 12 mm Titanium double sector cells (Nanolytics, Potsdam, D). All experiments were evaluated using the software SEDFIT applying Lamm equation modelling for 1 4 non interacting species to determine sedimentation and diffusion coefficients as well as concentration of up to 4 species in a solution. 73 74 NMR, PFG STE, Spin Spin (T 2 ) Relaxation Time Measuremen t All NMR experiments were conducted on a Bruker Avance DRX 500 MHz vertical bore system using a xyz gradient TXI probe with a 1H and 2H interior coil, 13 C and 15 N exterior c oil, and xyz gradients. All carbonate/bicarbonate buffer solutions were generated as described above except using 100% 13 C enriched sodium carbonate and sodium bicarbonate ingredients (Cambridge Isotopes) to enhance signal/noise. All experiments were con ducted at 298 Kelvin. Deuterium oxide was used to obtain a lock at a volume fraction of 2.5% of the total sample. Data was processed using overlapping spectral peaks was required. The T 2 relaxation times of the various species in solution were obtained using a Carr Purcell Meibloom Gill (CPMG) sequence with msec. PFG STE 13 C diffusion experiments were con ducted using the variation of the Bruker stegs1s pulse sequence based upon the Pulsed Field Gradient Spin Echo (PFG
34 SE) technique for measuring diffusion. 1 H was decoupled from 13 C for the entirety of the pulse sequence. We used 1.8 second diffusion time s and 1 msec gradient pulse times. All processing was zero filled twice and was done with 0.3 Hz line broadening to allow for characterization of NMR spectral features. We used a gradient with strength of 50 g/cm for the gradient pulses which were varied equidistantly between 2% and 95% to generate 16 1 D slices for analysis. The PFG STE pulse sequence used, as well as many of the relevant variables chosen, is shown in Figure A 1 of the A ppendix Results Metastable Liquid Condensed Phase (LCP) The bind ing behavior of Ca 2+ ion as aqueous CaCl 2 was titrated at various concentrations into 20 mM carbonate buffer (pH 8.5) is shown in Figure 2 1 Even though care was taken to minimize temporary concentration gradients when introducing CaCl 2 (aq) to the bicar bonate buffer, there still appear to be some kinetic effects due to the injection. This is evident by the slight difference in binding fraction between the various concentrations of CaCl 2 (aq) injectant. The bound fraction (slope of the displayed line) w ould be expected to be very similar if the system were at true thermodynamic equilibrium. Therefore, interpretation of the data should be approached from a qualitative standpoint. To additionally demonstrate the kinetic binding effects that can occur due to imperfect addition of ion to counterion, and to eliminate the possibility that the character of the Ca 2+ binding profiles are due to nucleation of mineral at the nozzle (micropipette tip), an additional titration was performed with the nozzle just abov e the solution. The results, shown in Figure A 2 of the A ppendix demonstrate that the qualitative character of the Ca 2+ binding profile is conserved and not due to nucleation at the nozzle even if kinetic binding effects are enhanced. Additionally, we
35 d id not quantifiably account for carbon dioxide gassing out that would occur with this solution over time. This can lead to changes in binding affinity and pH over time. For these reasons, we interpreted the data qualitatively. Initially during the titrati on, the fraction of calcium that binds is constant with each additional injection, which is expected for the formation of calcium carbonate prenucleation clusters. 46 However, at a critical bound calcium concentration ( ~0.125 mM in this case) the constant Ca 2+ binding affinity decreases yielding a new linear binding affinity slope, suggesting a change in the types of products forming Note this change in binding behaviour occurs prior to solid nucleus formation, which would be evidenced by the inability to achieve apparent equilibrium due to decreasing pH (massive carbonate binding) and decreasing free calcium concentration (massive binding of Ca 2+ to fo rm solid mineral). Interestingly, at the point of the transition, the [Ca 2+ ] Free [CO 3 2 ] Free ion product is different for all three CaCl 2 (aq) titration concentrations ( 0.4, 0.7, 0.1 mm2 for 4.5, 6, and 10 mM injections, respectively ). Solid nucleation and growth leading to precipitation eventually occurred at concentrations of bound Ca 2+ between 0.2 and 0.22 mM. The pH evolution of the solution was monitored during the titration and the results are show n in Figure 2 1b The pH of the solution was decreasing initially, as would be expected, prior to the phase transition due to the sequestering of CO 3 2 carbonate ions as predicted to occur during PNC formation. The pH evolution flattened considerably at the same bound Ca 2+ concentration (~0.125 mM) as the change in binding affinity was observed indicating that the Ca 2+ binding affinity to bicarbonate has increased relative to carbonate. As nucleation is a singular event, the discontinuity in the pH deve lopment
36 and/or the change in Ca 2+ binding development strongly points toward the nucleation of a second phase. Because the pH development becomes flatter less carbonate, or more bicarbonate ions are b inding after nucleation of this phase suggesting that the bicarbonate ion is participating in the emergent phase. The data displayed in Figure 2 1 suggest that there is a critical bound calcium concentration that leads to a phase transition which involves the bicarbonate ion. In order to gain further insigh t into the thermodynamics of this phase transition, isothermal titration calorimetry (ITC) was conducted using the same punctuated titration method as was used for the Ca 2+ binding and pH development experiments to measure the enthalpy change upon nucleati on of the emergent phase. All three titration concentrations were tested, and as can be seen in Figure 2 2 a first order phase transition identified by the apparent enthalpic discontinuities, is occurring at similar conditions as identified in the titra tion experiments further corroborating the notion that a phase transition has occurred The phase transition occurred 3 injections earlier for the 4.5 mM injection and 2 injections early for the 6 mM injection in the ITC experiment as compared to the des ktop titrations, possibly due to the more controlled environment of the ITC reaction chamber. The enthalpic discontinuity is positive demonstrating that the driving force of the phase transition is entropic, which is often a sign of liberation of hydrati on waters. Nanoparticle Tracking Analysis (NTA) was used to examine the size of the emergent phase species and to estimate the amount of species in solution after the phase transformation. Samples were analyzed prior to, at, and after the observed phase transition (the 7 th 10 th and 13 th injection, respectively) for the case of the 6 mM
37 CaCl 2 titrant. The results of the analysis are shown in Figure 2 3 Prior to the detected phase transition (7 th injection), no phase was detectable since pre nucleation clusters of ca. 2 3 nm in diameter are well below the 20 nm diameter threshold for the NTA method At the detected phase transition (the 10 th injection), large species emerged with a diamet er size distribution of ~60 nm. Additional CaCl 2 injections (13 th injection) resulted in an increase in size of up to 100 nm diameter, suggesting that increasing Ca 2+ concentration leads to growth and/or coalescence of the emergent phase. This strongly su ggests that the emergent phase is stable with respect to the solution species, because if it were metastable, it could only occur in the form of microscopic fluctuations in the solution and would not grow with further Ca 2+ injections. V ideo of the raw data used to track the scattering phase is shown in Object A 1 and A 2 of the A ppendix In the video one can see that the droplets showed somewhat unusual behavior almost as if the species were warping and ad o pting a temporary aspect ratio. The species were not observed to coalesce or split permanently; they came close to doing so and then re established the former interfaces prior to the interaction. This behavior seems consistent with the Wolf et al. observation of CaCO 3 precursor droplets having emulsion like behavior with a resistance to coalescence due to weak electrostatic stabilization. 54 According to the NTA results, the emergent species is very dilute (on the order of 10 12 species/liter), which suggests that the aggregation/coalescence of particles/dro p lets that yield phase arises due to weak interactions. Note this phase woul d be difficult to detect using standard dynamic light
38 scattering techniques which require a much higher particle/droplet count and a known refractive index. Analytical Ultracentrifugation (AUC) was used to detect the species present in solution for the var ious points of [Ca 2+ ] evolution for the case of the 6 mM CaCl 2 (aq) injection. The results are presented in Figure 2 4 Prior to the emergent phase formation, ions / ion pairs are present as indicated by the squares and circles, respectively. A t the expected point of phase emergence larger species suddenly emerge (D & E in Figure 2 4 ) After a few more injections only ions / ion pairs are detected. This result seems counterintuitive after a phase transformation, but if the volumes of the phas es are considered, it becomes understandable. As determined by NTA analysis the LCP droplets are very dilute and represent a very small fraction of the overall volume Literature reports on emulsion systems demonstrate coalescence with subsequent demixing allowing for a determination of the respective phase volumina. 75 However, the small volume fraction of this newly formed liquid phase with respect to the mother solution, would lead to an undetectable phase boundary and therefore, only the ions and prenucleation clusters in the major liquid phase can be seen. Due to the apparent liquid nature of this emergent phase as inferred from the NTA and AUC analyses we refer to these as liquid condensed phase (LCP) droplets. Nuclear magnetic resonance was used to further analyse the constituents involved in the LCP in compar ison to the solvated carbonate/bicarbonate species in the bulk solution. Specifically, Carr Purcell Meiboom Gill (CPMG) spin echo and Pulse Field Gradient Stimulated Echo (PFG STE ) diffusion NMR techniques were employed. CPMG measures the T 2 relaxations, which investigates the immediate surrounding
39 chemical environment of the species. PFG STE measures the self diffusion of the carbonate/bicarbonate species. Self diffusion is the diffusion coefficient of a species in the absence of any chemical potential gradient but for our situation it is synonymous to the translational diffusion coefficient. Firstly, it should be noted that a 1D spectrum of the system after the phase transition ( 17 injections of 6 mM CaCl 2 (aq) titrant shown in Figure 2 5 ) does not in dicate the presence of solid precipitation products at the CaCO 3 or Ca(HCO 3 ) 2 predicted chemical shifts. Solids would result is a broad NMR spectrum with a carbonate peak at ~170 ppm or a bicarbonate peak at ~160 ppm, which are not observed. However, if t he carbonate/bicarbonate is behaving as a solvated solute or a dynamic liquid, then they can interconvert through proton exchange and the result is an intermediate NMR spectrum, with one peak in between the two chemical shifts due to rapid (de)protonation interconversion between the two states. 76 77 Thus, NMR provides additional independent evidence of the liquid c haracter of the nucleated LCP phase The emergent LCP phase is further evident by the distortion of the resultant peak resulting in a n asymmetric broadening and in the up field direction (bicarbonate direction) (Figure 2 6 ). There i s a slight widening seen in the 7 th injection prior to the phase transition as compared to the buffer control but it is minor and very symmetric, suggesting the presence of only one phase. The asymmetric up field direction bulging of the solution with 17 injections of CaCl 2 (aq) suggests that this portion of ions is favouring bicarbonate in the bicarbonate carbonate interconversion to a greater extent than the bulk solution and that there may be an additional phase in the system This is consistent wit h the pH measurement shown in Figure 2 1b, which also suggests a favo ring of bicarbonate binding (or a less pronounced carbonate
40 binding) after the phase transition. We do not consider this emergent asymmetric bulging alone, because of its bicarbonate favo red g on the order of pico and nano seconds, which would be far too short to resolve from the bulk solution over time average d data of 5.5 seconds for the se NMR experiments. The broadened peak was deconvoluted for analysis using Gaussian distributions to model overlapping peaks, resulting in a bulk solution peak (main peak) which resembles in width and chemical shift the carbonate buffer control, and an em ergent peak (small peak) which is suspected to be the emergent liquid condensed precursor phase (see Figure 2 6 ). To see an example of our Gaussian model i ng of overlapping peaks, see Figure A 3 of the A ppendix Measuring the T 2 relaxation and diffusion pr operties of the carbonates and bicarbonates in the buffer solution by means of NMR is an excellent way to establish that there are distinctly different species present after the phase transition. The T 2 relaxations of both deconvoluted peaks were determine d by using the CPMG technique, which measures the reduction in peak intensity (M /M 0 ) vs. the change in the tau delay according to the following equation: ( 2 1) The result is a linear relationship between the natural log o f the signal attenuation and the tau delay with a slope of 1/ T 2 which can be found by fitting the data. The results, shown in Figure 2 7 A show that the average T 2 relaxation of 0.9 seconds for the ions associated with the LCP is much faster than the ave rage bulk solution ions T 2 relaxation of 1.5 seconds, suggesting that the emergent ions expressing bicarbonate bias are tumbling at a different rate and/or are in a different chemical environment than
41 the ions making up the bulk solution over several secon d time averages The bulk solution T 2 relaxation of 1.5 seconds is identical to the T 2 relaxation that was obtained for pure carbonate buffer suggesting that the carbonate/bicarbonate ions making up the bulk solution are tumbling and/or are in a chemical environment similar to bicarbonate buffer control (control) (shown in Figure A 4 of the A ppendix ). T 2 relaxations of solvated, 13 C enriched, carbonate/bicarbonate solutions like the ones discussed in this paper are primarily due to chemical shift anisotop y (CSA). Chemical shift anisotropy occurs if the electron shielding around the dipole is anisotropic, as is the case of asymmetric carbonate and bicarbonate ions. When CSA is present, the extent in which the dipole interacts with an applied magnetic fiel d depends on the rate of molecular tumbling of a dipole. In dilute, non viscous solution states, the effect of CSA is often averaged out due to rapid tumbling of molecules. When carbonates/bicarbonates rotations are inhibited by interactions/binding with Ca 2+ the reduced tumble rate of molecules enhances the CSA effect and leads to a reduction of T 2 relaxation times. The relaxation rates of the Ca 2+ bound/interacting carbonates and bicarbonates are on the order of solution state relaxations. Solids hav e much shorter T 2 relaxations, on the order of less than a millisecond, and therefore, the detected bicarbonate biased ions exhibit solute character, which may rely on the internal rapid dynamics of the liquid calcium carbonate precursor phase. As mentione d above, 13 C PFG STE diffusion NMR is an excellent non invasive technique for establishing the co presence of two distinct diffusions within a solution, indicating the presence of an emergent liquid phase. In the case of this study, we assumed isotropic unbiased diffusion and therefore were concerned only with diffusion
42 rates in the z direction (measured by the magnetic field gradient created in the z direction). The nucleus of interest, in this case the 13 C nucleus of carbonate/bicarbonate, is subjected to a 90 pulse for an encoding step, followed by another 90 pulse to store the magnetization. After an evolution time (diffusion time), a final 90 pulse is applied after the evolution time to refocus the transverse magnetization for acquisition. A fie first and third 90 pulses to apply and remove, respectively, the spatial encoding to the 13 C nuclei. This ensures that the extent of refocusing of the applied 90 encoding pulse is depend ent on the degree of deviation from the initial position that was spatially encoded by the gradient. Translational movement of the dipole during the diffusion time results in an attenuation of the intensity of signal (A/A 0 ) which is related to D (the self diffusion constant) through the following expression: ( 2 2 ) 13 of the equation vs. the right si de with stepwise increases in the gradient strength, G, we generated a line with slope D (negative self diffusion constant), as seen in Figure 2 7 B This figure compares the diffusions of the emergent LCP (obtained by 17 injections of 6 mM CaCl 2 (aq) in to 20 mM bicarbonate buffer, pH 8.5) with the mother bulk solution and a bicarbonate buffer control. This required the deconvolution of NMR spectra in a similar fashion as described for the T 2 determination above. A description of the method and the atte nuation of signal are shown in Figure A 5 of the A ppendix T he
43 results indicate two distinct diffusions exist in the LCP containing solution, a slower one of 5 x 10 6 cm 2 /s which is attributed to bicarbonate biased ions contributing to the emerg ent phase, and a faster one of 9 x 10 6 cm 2 /s which is that of the bica rbonates/carbonates in the bulk solution. The self diffusion of the bulk solution peak is very similar to the diffusion we obtained from our control of carbonate buffer alone (10 x10 6 cm 2 /s), a nd is in close accordance with literature values for the self diffusion of carbonate buffer obtained by simulation 78 and by experiment 79 under similar conditions. It should be noted that t he slower diffusion attributed to the emergence of a LCP is not that of the large LCP droplets seen using the NTA technique which, given their size of > 60 nm, would be orders of magnitudes slower based on Einstein Stokes diffusion (and not detectable with the gradient used in this experiment), but rather are believed to be that of small Ca 2+ bound carbonate / bicarbonate ion pair species that might be constituents of the large metastable droplets. According to the Einstein Stokes equation (using the kinemat ic viscosity of water at 25 C), the bulk solution and buffer control carbonates/bicarbonates have an effective diameter of 0.5 nm The slower diffusing carbonates/bicarbonates have an effective diameter of 1 nm. The favouring of bicarbonate, the apparen t slowing of 13 C based ion rotations, and the roughly doubling of effective diameters suggests that there is a significant amount of calcium bicarbonate monodentate and calcium bicarbonate bide ntate ion pairing phenomenon present which are being detected over relatively long time averages as compared to the bulk mother solution Another interpretation is that the emergent LCP is more viscous than the mother solution. The average effective size of the detected carbonate/bicarbonate may be a
44 singular ion still if the viscosity of the emergent phase were twice that of water, or about 0.018 dyne sec/cm 2 Both interpretations suggest that there is a liquid liquid separation occurring that is forming a LCP. It is important to stress that the effective diamet ers calculated are based on diffusion of the ions, which are expected to be an ensemble average due to the dynamic nature of ion associated waters of hydration. This needs to be considered when comparing thes e TEM 47 and AUC. 46 However, the average diameter of the Ca bicarbonate biased ion pairing species in our 2 phase system of 1 nm is relatively consistent with the diameters of 0.75 nm and 2.0 nm obtained by Cryo TEM and AUC, respectively. The data as a whole s uggests that, at a critical bound Ca 2+ concentration, Ca bicarbonate ion pairs begin to associate due to entropic driving forces (often the liberation of hydration waters) into a sparse, but visible, metastable liquid condensed phase Due to evidence of L CP droplet growth during experiment evolution, the phase is believed to be stable with respect to the solution state. The bound Ca bicarbonate species believed to comprise the LCP appear to have a long enough dynamic lifetime to be detect able using solut ion state NMR, (unlike PNC's ) and they appear to be slowly translating in space and slowly tumbling as compared to the free carbonate/bicarbonate in the bulk solution. It is reasonable to assume that the emergent droplets are comprised of bicarbonates gi ven the data but considering how dispersed the droplets are, the bulk of the detected bicarbonate bias species being detected by the NMR must be in solution possibly exchanging onto and off of the droplets. It is important to point out that the values of T 2 and self diffusion obtained by NMR are on the scale of solutes
45 in solution, supporting the notion that the LCP is a liquid phase distinct from the mother solution The values of T 2 and self diffusion for solids would be many orders of magnitude smal ler. Experiments in the Presence of Polyaspartate In the titration and NMR experiments, the method of generating a super saturated solution with polymer additive differ ed from that without additives. The reason for this was to mimic our lab s convention o f adding carbonate/bicarbonate ion to a solution containing CaCl 2 and polymer additive to initiate the PILP process. Also, the polymer additive has such a strong stabilizing effect on the solution that the concentration of bicarbonate buffer needed to be increased to generate a sufficient super saturation to observe eventual solid nucleation and precipitation A Ca 2+ binding profile in the presence of PAsp (14,000 g/mol and 27,000 g/mol M.W.), as 300 mM bicarbonate buffer, pH 8.5 was titrated into a 10 mM CaCl 2 20 g/m L PAsp solution is shown in Figure 2 8 As is the case with Figure 2 1, there is presumably some kinetic binding effects occurring in this titration and therefore the Ca 2+ binding profile should be treated as qualitative. As a control, a Ca 2+ profile was conducted for a titration in the absence of PAsp for comparison. In this case, one can see an apparent discontinuity in Ca 2+ concentration at an early stage that is analogous to the one seen in Figure 2 1 A As with the discontinuity sh own in Figure 2 1 A this potential discontinuity suggests that a phase transition has occurred. In order to aid in visualization, lines have been added to the inset of the figure to demonstrate discontinuity. In the presence of polymer, the first thing that one notices is how much farther the Ca 2+ profiles extend prior to precipitation relative to the control (bottom curve, with
46 no polymer). In addition to this stabilizing influen ce, the shapes of the titration curves show an interesting trend. The bi nding of Ca 2+ with respect to injections of appears to display a discontinuity (shown in Figure 2 8), suggesting that LCP is forming in the presence of polymer as well as evidenced by the same type of discontinuity as seen in the non additive case (Figure 2 1 A ). The LCP like Ca 2+ binding behavior continues for many more injections beyond where the additive free control would have precipitated (see precipitation indicators in Figure 2 8 ). T he continuation of the same type of binding that forms the LCP ph ase is a strong indication that LCP is being stabilized by the presence of the PAsp polymer additive PILP process may be to stabilize the LCP. ffect of polymer additives, also detected a significant nucleation inhibition and a species after nucleation of CaCO 3 in presence of PAsp (M.W. 6800 g/mol and M.W. 27000 g/mol) that is much more soluble than ACC, which they related to a PILP phase (althoug h they worked at a higher pH of 9.75) 80 A profile of the pH evolution with th e addition of 200 L injections of 300 mM, pH 8.5 bicarbonate buffer into 10 mM CaCl 2 (aq) containing 20 g/ mL of Pasp is shown in Figure 2 9. It is apparent when comparing the pH evolution of bicarbonate buffer into aqueous Pasp solution (diamonds and sq uares) to the pH evolution of bicarbonate buffer into pure water (dotted orange line) that Pasp has a dampening effect on the evolution by maintaining a lower pH. This leads to a higher bicarbonate:carbonate ratio at the early points in the titration. Th is would be a very effective means for directing more bicarbonate to bind with Ca 2+ if it were present. When Ca 2 + is present with the
47 Pasp (triangles and crosses) the pH development suggests this is the case. In comparison to the controls in the absence of Ca 2+ the Ca 2+ containing solutions actually do demonstrate additional bicarbonate binding in terms of a bicarbonate:carbonate ratio, until a certain threshold is met. At this point the pH evolution flattens considerably at a much lower pH than the va rious controls. This supports the assertion that a critical amount of Ca bicarbonate ion pairs accumulates leading to a phase transition (LCP) and a change in Ca 2+ affinities for bicarbonate and carbonate. The preference to maintain a pH in the range of bicarbonate rich phase is participating in directing the binding preference for Ca 2+ Figure 2 9 supports the view that Pasp (and potentially other additives) promote and stabilize the LCP phase and perhaps i s the description for the so called PILP process. To analyze the PILP phase which is suspected to be polymer promoted and stabilized LCP, a 1D NMR spectrum of the system at the condition indicated with a circle in Figure 2 8 was obtained and is s hown in Figures 2 10 and 2 11 The effect of apparent PILP formation severely distorts the spectral peak. As seen in figure 2 10, t here is an emergent peak splitting from the bulk solution peak in the up field direction, indicating that a significant fra ction of the buffer is favoring bicarbonate as compared to the bulk solution over a large multisecond time average. Figure 2 11 shows the time evolution of the two phase bulk solution PILP sample. The process required hours to develop but solid nucleatio n and precipitation did not occur as evidenced by the unchanging area of the peaks. Figure 2 11 shows a 1 D NMR spectral comparison after 18 hours (to attempt to achieve equilibrium) of the one phase, pH 8.5, bicarbonate buffer and Pasp solution versus th e multiphase bicarbonate/Pasp/Ca 2+ solution at the
48 17 th injection. The two phase system emergent peak (bulge of separation from bulk solution) is shifted more up field and is much larger than the NMR evidence of LCP precursor phase in the absence of additi ve, suggesting that polymer is stabilizing and accumulating a bicarbonate biased ionic interaction, which fits the description of the LCP phase, within a second bulk solution (mother solution) phase We used the same parameters and techniques to measure t he T 2 relaxation and diffusion properties of the PILP system as describe d above for the LCP system. The results are shown in Figure 2 12 Raw data for the CPMG (control), PFG STE (control), CPMG (phase separated PILP solution), and PFG STE (phase separat ed PILP solution) are shown in Figures A 6 A 9 of the A ppendix respectively. As with the LCP NMR experiments, the diffusion values obtained for the suspected PILP phase is not being presented as the diffusion of the large droplets. We believe this is t he effective diffusion of the ion solute species that collect into, and are in exchange with, the PILP phase droplets. The diffusion measurements were c onsistent in scale with those obtained for the LCP system. The presence of PAsp in the solution appear ed to reduce the carbonate/bicarbonate ion dif fusion in general, but the ratios are similar to the LCP experiments and controls. The diffusion of phase separated ions (3 x 10 6 cm 2 /s) was a little less than half the diffusion of the carbonate/bicarbonate i n the bulk solution (7 x 10 6 cm 2 /s), and the bulk solution diffused at the same rate as the bicarbonate buffer control (both at 7 x10 6 cm 2 /s). Using the Einstein Stokes equation, the effective diameter of the ionic species was 0.7 nm for the bulk soluti on and control carbonate/bicarbonate ions, and was 1.6 nm for the bicarbonate biased ions. This is consistent with the slowed ions existing disproportionately as calcium monodentate /
49 bidentate over this time average. Note the presence of the Pasp lead t o an overall slowing of the experimental and control diffusions, leading to the calculation of a larger effective diameter. This may be due to increased viscosity caused by the presence of the polymer. Still, the ratio of diffusions (bicarbonate biased i on pairs : mother solution :control) were consistent with the non polymer case and demonstrate the separation of two distinct equilibriums which are comprised of ions d iffusing at different rates. As explained for the non polymer case, this could be due to s ingular ions in a more viscous emergent PILP phase. If this were the case, the emergent PILP phase would have a viscosity of 0.02 dyne sec/cm 2 according to the Einstein Stokes equation. The T 2 relaxation of the up field bulge was 1.1 msec, which sugges ts that the bicarbonate biased ions are slowly tumbling and rotating in solution. It was almost exactly the same as the bulk solution; both were shorter than the relaxation measured for the polymer bicarbonate buffer control (1.5 msec). Although a disti nction between suspected PILP phase and the bulk solution would have been helpful in distinguishing between types of phase environments it is not surprising that they are similar due to the appearance of heavy chemical exchange between the them, as eviden ced by the bridging between the suspected PILP peak and the mother solution peak (see F igure 2 1 0 ). As a side note, the spectrum of the separated PILP phase shown in Figure 2 10 resembles a powder pattern, which is an NMR spectral pattern obtained when t here are solid particles present. However, a true powder pattern due to solids in solution would yield a broad spectrum peak with a width of 100 ppm. The width of the bulk solution/PILP peak is only 0.1 ppm. This could mean that there is a slowing in th e
50 as expe cted for a solute in solution. Discussion The nucleation pathway of calcium carbonate through a multi step process, where the first step is the formation of a metastable liquid and the second is the solid nucleation formation (within or outside the metastable liquid) has been at the forefront in recent literature. 3 50 63 Here, we have presented data that supports the first step of this assertion through a combination of techniques In the absence of polymer addi tives, the ITC, Ca 2+ binding profiles, and pH evolution experiments indicate that there is a phase transitio n. The pH measurements suggest the carbonate/bicarbonate binding becomes bicarbonate biased, as do NMR peak emergences in the bicarbonate direction suggesting that bicarbonate is playing a role in the emergent phase The fluidic character of this nucleated phase is suggested in the sedimentation properties of the species in solution and the apparent co existence of a distinct solution like liquid pha se within the mother solution as measured by AUC and NMR self diffusion and relaxation times respectively Gebauer et al. 46 and Wolf et al. 54 amounts of the bicarbonate ion. Computer simul ations have suggested that DOLLOP and nanoseconds. 81 For these reasons, we believe that the nucleation phenomenon we are seeing cannot be described by PNC theory. The apparent bicarbonate biased constituents we detect accompanying the precursor phase ma nifest their properties for long enough lifetimes to detect, suggesting that the LCP precursor phase is not forming from carbonate
51 biased Ca 2+ interaction which is distinct from the Ca 2+ interaction with carbonates/bicarbonate in the bulk solution over significant time averages. The presence of Ca bicarbonate ion pairing phenomenon or even clustering does not challenge the validity of predominantly Ca wo CO 3 2 ions, with their greater binding affinity to Ca 2+ would dominate. However, at more ely few CO 3 2 ions as compared to HCO 3 ions and weaker Ca bicarbonate ion pairing or clustering phenomenon become significant, leading to accumulation and nucleation of a bicarbonate rich metastable liquid phase like the one that has been proposed in this study. This is significant because it has been proposed that nucleation of solid CaCO 3 can occur through nucleation inside of a metastable liquid phase. 82 The seed nucleus is proposed to form within the droplet and its growth and polymorph is subject to the environment of the metastable droplet, rather than that of the bulk solution. If metastable droplets of bicarbonate biased Ca 2+ interactions phase separate and are stabilized by the presence of polymer (PILP process), as is possible with the droplets observed in this study, then this might explain the incredible amount of nucleation inhibition that polymers (particularly negatively charged, aspartic and glutamic acid rich polymers) give to a solution. An emergent metastable bicarbonate droplet and its smaller constituents would sequester Ca 2+ and carbonate/bicarbonate ions from the solution, thus mitigating the solid phase nucleation through PNC non classical pa thways. However, the resulting concentrated metastable droplet would have difficulty
52 forming a seed nucleation because it is bicarbonate biased and would require the extra step of releasing the H + ions prior to organization into a nucleus. We propose that LCP and PILP (PAsp stabilized LCP) plays a role in the calcium carbonate energetic cascade from supersaturated solution to crystallinity as described in Figure 2 13. Knowing what the PILP phase is may provide insight into various biological nism of control over morphology, phase, and location of the final biominerals product. Although the presence of PNCs and LCP are interesting with respect to theoretical models of crystallization mechanisms, without the polymer, the final products simply r esemble those predicted by classical models. It is the accumulating and stabilizing effect of the polymer that allows for manipulation of the liquid condensed phase into an endless array of non equilibrium morphologies. On the other hand, without this pr opensity of the mineral reactants to form this liquid condensed phase. This work shows the significance of both sides of this interesting mineralization system.
53 Figur e 2 1 The bound [ Ca 2+ ] evolution and solution pH evolution during titration of Ca 2+ (aq) into 20 mM bicarbonate buffer with an initial pH of 8.5. A) The evolution of bound Ca 2+ vs. overall titrated [ Ca 2+ ] Initially, the binding profile is linear which is expected for prenucleation cluster formation. However, a discontinuity in Ca 2+ binding is observed at a value of ~ 0.125 mM [Ca 2+ ] Bound which is evidence of a phase transition. The evolution was done in triplicate and results were averaged. B) Th e pH evolution of the solution due to the injection of CaCl 2 (aq). The pH decreased due to the consumption of CO 3 2 to form ion pair s [Ca 2+ ] Bound of ~ 0.125 mM, suggesting a phase transition at the same bound calcium concentration as seen in part A The change in pH evolution in the upward (basic) direction suggests that a larger fraction of bicarbonates and/or a reduction of carbonates are participating in Ca 2+ binding. Lines were included as an aid. M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
54 Figure 2 2. The enthalpy of reaction during the titration of CaCl 2 (aq) into 20 mM carbonate buffer, pH 8.5. The data shows that the re is an endothermic phase transition occurring, indicating that the phase transition is entropically driven and probably due to liberation of hydration waters. M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
55 Figure 2 3 The emergent phase size distribution, according to nanoparticle tracking analysis (NTA), of the 20 mM bicarbonate, pH 8.5, solution at and after the phase transition (injection 10 and 13, respectively for a 6 mM CaCl 2 (aq) inject ion). Prior to the phase transition no species were detected. At the phase transition (10 th injection), droplets with a distributed diameter averaging ~60 nm emerge. The addition of more Ca 2+ to the solution yields more detectable phase emerging at 60 7 0 nm diameter and they seem to grow larger as well. M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of t he Royal Academy of Chemistry
56 Figure 2 4 Sedimentation coefficients (25 C) for the detected species in solution by means of analytical ultracentrifugation ( AUC ) for the injections indicated in the inset. The emergence of LCP occurs at point C. It is evident that larger species are present for D & E, just after the phase transition. The control consists of just carbonate buffer. PNCs are also seen in the control because they can form with sodium carbonate species as well. M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
57 Figure 2 5 A 1 D 13 C spectrum of the sol ution after the phase transition (17 th injection of 6 mM CaCl 2 (aq) into 20 mM carbonate buffer, pH 8.5). There are no detectable peaks at the solid carbonate or bicarbonate chemical shifts (marked with arrows). The effect of the nucleated liquid precurs or phase is seen as a widening of the peak because it is still behaving as a (non solid) phase M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discus sions 2012. Reproduced by permission of the Royal Academy of Chemistry
58 Figure 2 6 A c omparison between the 13 C 1 D NMR spectrum for bicarbonate solution (control), the bicarbonate solution with CaCl 2 (aq) prior to the phase transition, and the bi carbonate solution with CaCl 2 (aq) after the phase transition A) 1 st slice of T 2 relaxation of 20 mM bicarbonate buffer, pH 8.5 (control). B) 1 st slice of T 2 relaxation of 7 th injection of 6 mM CaCl 2 (aq) into 20 mM bicarbonate buffer, pH 8.5 (prior to phase transition). C) 1st slice of T 2 relaxation of 17 th injection 6 mM CaCl 2 (aq) into 20 mM bicarbonate buffer, pH 8.5 (after phase transition). Both the control ( A ) and the solution prior to phase transition ( B ) are symmetrical Gaussian distributions of signal. After the phase transition ( C ), the evidence of an emergent phase manifests as a n asymmetric bulge shifted up field from the bulk solution peak which requires an additional Gaussian peak to model The data was deconvoluted in order to attribu te T 2 relaxation measurements and 13 C NMR diffusion measurements to each modeled portion of the peak M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
59 Figure 2 7 The results of the CPMG T 2 relaxation measurement and the 13 C PFG STE self diffusion measurement of the deconvoluted NMR peaks. A) The plot compares the reduction in carbonate/bicarbonate peak intensity vs. the tau delay ( Equation 2 1 ). The result is a linear relationship with a slope of 1/T 2 B) The diffusion measurement uses the relationship shown in Equation 2 2 and the measured attenuation of carb onate / bicarbonate signal to determine the self diffusion of the carbonates/bicarbonates, which is the negative slope of the resulting line. The bicarbonate biased ions (bulge) ha ve a shorter T 2 relaxation time and a slower diffusion, suggesting that a fra ction of the carbonate/bicarbonate ions are slowed in rotation and in diffusion, presumably due to interactions with Ca 2+ and the emergent LCP The self diffusion of the bulk solution carbonates/bicarbonates is roughly the same as the self diffusion of ca rbonate buffer alone (~10 x 10 6 cm 2 /s). M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Ac ademy of Chemistry A B
60 Figure 2 8 The [Ca 2 + ] profile of a system where 300 mM bicarbonate buffer, pH 8.5 was injected (40 l initial, 200 l for the rest) into 10 mM CaCl 2 with 20 g/ mL PAsp (no PAsp for control). The results qualitatively show evidence of the phase transition seen in F igure 2 8 and are analogous to the LCP formation in the absence of polymer shown in F igur e 2 1 A After the presumed LCP formation, the solution is stabilized against solid nucleation and precipitation for many more injections suggesting that LCP is stabilized d ue to the interaction of polymer with the LCP phase This may be the basis for t he PILP process. NMR studies to analyze the two phases in solution (bulk solution and phase separated PILP) were conducted at the location indicated by the circle. M. A. Bewe rnitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
61 Fig ure 2 9 The pH evolution of a 10 mM CaCl 2 solution containing 20 g/ mL Pasp due to the punctuated titra tion of 300 mM bicarbonate buffer, pH 8.5. The presence of Pasp has a mitigating effect on the upward evolution of the solution, allowing for more bicarbonates to be present present ( diamond and squares ). When Ca 2+ is present with the Pasp, the pH evolut ion increases in the basic direction faster than the evolution with Pasp alone before level ing off at a preferred pH. This suggests that when enough bicarbonate has bound, an equilibrium is established, presumably due to an emergent phase, resulting in a consistent, flattened pH evolution The typical injection is 200 l, but to characterize the early titration, fractions of 200 l injection were used. M. A. Bewernitz, D. Ge bauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
62 Fig ure 2 10 A comparison between the 1 D NMR spectra of bicarbonate buffer with polyaspartic acid sodium salt and the se parated bicarbonate biased ions The presence of PAsp leads to a broad peak which is shifted up field (bicarbonate weighted direction) distinct from the remaining ions in solution (Bulk Solution). This behavior is similar to the behavior of the liqui d condensed phase except it is enhanced greatly in the presence of polymer, suggesting that the PILP process stabilizes the liquid condensed phase. M. A. Bewernitz, D. Gebauer J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
63 Fig ure 2 11 The time evolution of a 1 D NMR spectrum of the phase separated PILP containing solution shown in Figure 2 10 Initially, it is broader than the control solution (absent of CaCl 2 ) as was observed for the formation of liquid condensed phase. However, in time a more distinct phase separation occurs to yi e ld a large fraction of bicarbonate bias of the ions distinct from the bulk solut io n phase suggesting the possibility of slow ripening and coalescence of LCP phase. The red, bracketed numbers above each spe c tral peak represent the area under the peak, normalized with respect to the 0 min spectrum. The amount of signal remains constant suggesting that all the ions are still behaving as solution state ; therefore, solid nucleation and precipitation has not occurred. M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
64 Fig ure 2 12 The results of the CPMG T 2 relaxation measurement and the 13 C PFG STE self diffusion measurement of t he PILP phase (suspected polymer stabilized LCP), bulk solution, and bicarbonate buffer with polyaspartic acid (control). A) The CPMG plot compares the attenuation in the carbonate/bicarbonate peak intensity to the tau delay (Equation 2 1). The result is a linear relationship with a slope of 1/T 2 B) The diffusion measurement uses the relationship shown in Equation 2 2 and the measured attenuation of carbonate/bicarbonate signal to determine the self diffusion of the carbonates/bicarbonates, which is th e negative slope of the resulting line. The bicarbonate biased ions have the same T 2 relaxation as the bulk solution due to the large amoun t of chemical exchange, but have a shorter T 2 relaxation than the bicarbonate buffer control, suggesting that rotati ons of the bicarbonate biased ions are slowed. The diffusion of the bicarbonate biased ions is slowed with respect to the bulk solution and the bicarbonate buffer control in a way similar to LCP, suggesting that polymer is kinetically stabilizing the LCP phase. M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
65 Figure 2 13 An o verview of the energetics (not to scale) of calcium carbonate precipitation from supersaturated solution, putting LCP into a global context with earlier findings. The diagram has been truncated on the right where the hydrous ACC will ultimately tran sform into a more stable crystalline phase. In solution, calcium bicarbonate and calcium carbonate ion pairs form, the latter of which can associate into larger species, PNC's including dynamically ordered, liquid like oxyanion polymers (DOLLOP). The dif ferent prenucleation species form virtually spontaneously (thermal energy k B T), while formation of LCP is associated to a barrier of nucleation g*(1). (A) At with it, leads to a bicarbonate bias in the LCP and an intrinsic kinetic stabilization, g*(2), (B) At higher pH levels, calcium bicarbonate ion pairing becomes neg ligible, carbonate species dominate the nucleation process, and render LCP transient due to a reduced intrinsic stabilization. The barrier g*(2), may even vanish.(C) Addition of Pasp may lead to a distinct increase of the barrier g*(2) due to a pronounc ed role of bicarbonate species in the LCP in the presence of polymer. The bicarbonate pathway may be preferred in the presence of polymer, and/or the polymer may stabilize bicarbonate species within LCP. M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
67 CHAPTER 3 E FORMATION OF C ALCIUM CARBONATE LCP Introduction It is believed that solid amorphous calcium carbonate (ACC) is a vital predecessor to many of the crystals formed in biomineralization and biomimetic crystallization processes. 83 As demonstrated in C hapter 2, there is a liquid precursor phase (liquid condensed phase, LCP) of calcium carbonate, which arises from liquid liquid phase separation fro m the supersaturated calcium carbonate solution. It is possible that this LCP may be a precursor to the solid ACC, although the later stages of the reaction were not examined here. Although crystallization/solidification pathways through a ltwo step nucle ation pathway are well known in protein and polymer systems, 63 82 it has only recently been proposed that ca lcium carbonate transitions through a similar pathway. Gower et al. were the first to present direct evidence of a liquid precursor phase of calcium carbonate that forms in the presence of negatively charged polyelectrolytes, 52 84 and coined the name polymer induced liquid precursor (PILP) to describe the intermediate. They have since reported evidence that the P ILP process is seen in the barium, and strontium, carbonate systems, 55 56 as well as the calcium phosphat e system. 59 60 62 85 86 Interestingly, Faatz et al. 38 Reiger et al. 35 and Wolf et al. have provided theoretical, indirect, and direct (at non equilibrium conditions) evidence, respectively, that calcium carbonate may transition through a liquid intermediate phase in the absence of additives. However, in my work presented in Chapter 2, I was able to detect the presence of an emergent liquid phase of saturated calcium carbonate at seemingly equilibrium conditions. 87 Furthermore, I provided direct
68 evidence that the liquid precursor phase is rich in bicarbonate species, rather than carbonat e, which has been implicitly assumed by many in the field. I have referred to this phase as a liquid condensed phase (LCP) of calcium carbonate, and it is described as a true liquid phase, as opposed to a liquid like phase because the NMR spectrum obtaine d for the two phase system seen in F igure s 2 6, 2 10, A 3 A 5, A 8, A 9 reveal a phase that contains bicarbonates in equilibrium with carbonates, as would be expected in a liquid phase. If calcium carbonate LCP emerges from a supersaturated calcium carb onate solution phase as droplets of a true ionic solution distinct from the mother solution, this leads to the question what type of liquid is it? As mentioned previously, the concept of a liquid intermediate step during the calcium carbonate solid nucle ation formation process is not a newly proposed concept, 88 but in those days, there was no way to study the process at the molecular level. However, this type of proposed liquid phase is described as an intermediate metastable step in the transition from prenucleation clusters (loose ion pairings or DOLLOPs of Ca 2+ and CO 3 2 ) 46 48 to tightly packed and solid critical nuclei. Because critical ion pairs and/or critical oli gomeric chains of prenucleation clusters must rearrange and condense to become tight packed solid nuclei, there may be an energetic local minima between the two states that is metastable and condensed to some intermediate level. However, if this was the m echanism of LCP formation, one would expect it to be carbonate rich, like ion pairs, calcium carbonate LCP is bicarbonate rich. 87 Therefore, it does not seem likely that the LCP is arising from collections of these carbonate rich species.
69 After discussing the concept of calc ium carbonate LCP with colleagues at conferences, it became apparent that there was a misunderstanding regarding the description of LCP as a liquid phase. The biomineral research field is heavily populated by chemists and microscopists, which tend to view a liquid phase as being a clustering of specific Ca 2+ complexes. However, I am describing a true liquid phase; a phase that is charge neutral with an interface in contact with the mother solution which is best described from a physical chemistry/thermody namic perspective. As a result of this, I am reviewing the description of a liquid liquid phase separation below in these terms for the sake of clarity. The emergent LCP phase may be due to a partial miscibility of solution constituents within each othe r (i.e., a true liquid liquid phase separation). This behavior has been demonstrated and is quite well known in the physical chemistry community for various ionic solutions. 89 93 For example, imagine that a solution B is being added to solution A. Initially all the added B dissolves because B is miscible in A at this dilution concentration. Eventually, as B is added, the Gibbs free energy of further mixing is positive for B into A. Additional increases in B will lead to an emergent phase and a two phase system. An example of this hypothetical two phase system is shown in Figure 3 1. Figure 3 1 plots the phase lines (blue curve) on a plot of temperature v s. mole fraction of B (any state variable can be expressed on the y axis). For mole fractions of B within the curve, phase separation occurs along a horizontal temperature tie line to yield a solution saturated in B within the A major solution phase, and a solution saturated in A species withino the B minor solution phase. The amount of phase is dictated by the inverse lever rule of tie line proportions. The left part of Figure 3 1
70 shows a hypothetical phase diagram with an upper co solute temperature li mit. Above this temperature, the thermal energy of the components is sufficiently high to overcome favorable interactions between like components (B B and A A). It is also possible to have a system with a lower co solute temperature limit, as shown on th e right side of Figure 3 1. Below this temperature, the thermal energy is low enough to allow favorable interactions between A and B, thus increasing the miscibility. In some cases it is possible to have an upper and a lower co solute temperature due to a combination of both justifications, in which case the phase curves become enclosed into a loop. One could imagine that A and B could be carbonate and bicarbonate, respectively. If this were the case, it would suggest that the LCP (and its polymer stabil ized, PILP form) is a fundamental buffer phenomenon, and not necessarily directly due to Ca 2+ binding/complexation with the buffer. This is quite profound because LCP phases are suspected to exist in other biomineral systems such as calcium phosphate 55 57 60 62 an d calcium oxalate. 94 95 It may well be that nature has evolved to use minerals with a buffer as a count erion because of the propensity to form easily controlled, malleable, liquid phases. However, the system is more complex than just carbonate and bicarbonate and there may be many thermodynamic state variables contributing to the phase separation. In fact according to the Gibbs phase rule shown in equation 3 1, the system has at least 5 due to all the ionic species that can partition into both phases. (3 1) Where: F = # of independent state variables rigorously describing the system C = # of components in the system (ions species)
71 P = # of phases present in system In addition, the system may be an irregular system. Irregular systems are systems where the interaction of A and B lead to creation or destruction of entropy. These systems are the rmodynamically complicated to analyze. The supersaturated calcium carbonate system is a good candidate for an irregular mixture due to the myriad of ways for Ca 2+ CO 3 2 and HCO 3 (and their associated hydration waters) to interact and effect changes in entropy. One way to provide evidence regarding the type of liquid condensed phase is to analyze the effect of other components in the system. If the liquid is a type of condensed metastable intermediate between prenucleation clusters and a solid critical to be non participatory in the formation of the emergent liquid phase. If, however, the LCP emerges from a liquid liquid phase separation due to partial miscibilities of the various components, t hen one (equation 3 1). In this study, we have varied the concentration of the Cl ion to provide evidence of which type of liquid the LCP is in the calcium carbonat e system. One of the unexpected results obtained regarding the emergence of calcium carbonate LCP from a saturated mother solution was the Ca 2+ concentration development during and after LCP formation (see Figure 2 1 A ). There appeared to be less Ca 2+ bou nd after the phase transition than before. This is paradoxical. If ever increasing concentration of Ca 2+ binding to anion (carbonate or bicarbonate) yields a critical concentration of Ca 2+ bound complex that spontaneously forms a new phase, than one woul d expect the formation of this relatively energetically favorable phase to
72 promote more Ca 2+ binding, not less. I mentioned this phenomenon in Chapter 2, but did not address it directly because the Ca 2+ binding profile was treated in a qualitative manner to demonstrate that a phase is forming. But it would be desirable to understand this paradoxical behavior. The free Ca 2+ concentration could be truly increasing, but without the use of very convoluted theories, this would seemingly contradict the laws of thermodynamics. I considered that perhaps there is a simple reason why this phenomenon would occur? Two primary reasons came to mind, and will be addressed here. Removal of Counterion from the Mother Solution Perhaps the removal of anionic counterion ( carbonates and bicarbonates) into the emergent LCP phase is so severe that it leads to less Ca 2+ bound complex overall. This principle could be described with use of the expression below based on an equilibrium expression for Ca 2+ binding to counterion: (3 1) In this expression, K 1 and K 2 are calcium ion equilibrium binding constants with carbonate and bicarbonate, respec tively. This assumes that the formation of Ca(HCO 3 ) 2 complexes are insignificant. Equation 3 1 can be rearranged into the following expression : (3 2) The activity coefficient of the free ions needs to be considered resulting in the following expression:
73 (3 3) Ca is the activity coefficient of the Ca 2+ Counter is the effective activity coefficient of the carbonates and bicarbonates ( CO 3 2 and HCO 3 ), respectively. It is very important to stress that combining the activity coefficie nt of the carbonate and bicarbonate ions into a single counterion coefficient in order to pull it out of the bracketed expression is an assumption. A ctivity coefficient s are due to i nhibition of translational diffusion of ions and e lectrostatic shielding due to neighboring ions. Unlike the other ions in solution, carbonate and bicarbonate rapidly interconvert through protonization and deprotonization, meaning that they are not restricted to the same diffusion and shielding restrictions that as other, dist inct ions are. Their diffusion is limited to the kinetics of protonization and deprotonization which is a very rapid process. For this reason, and for ease of calculation, they are assumed to behave as if they have an intermediate, single activity coeffi cient based upon the time average of their ion charge which, at a constant pH of ~8.45 is ~98% 1 and ~0.15% 2 for a time average ion charge of 1.01. For a detailed explanation of ion charge on the activity coefficient, see equation 3.6 below. The ne t result of this assumption is a potential overestimation of the CO 3 2 activity which may be an acceptable estimate considering the small fraction of CO 3 2 present at these conditions. This should be considered when drawing conclusions from this data. F rom equation 3 3 we could see how a reduction of the overall counterion concentration out of the mother solution, into the LCP phase, would be expected to lead to a proportional decrease in the evolution of [Ca 2+ ] Bound The expression can be
74 expressed in terms linear slopes observed in the Ca 2+ profile by dividing both sides of equation 3 3 by the total dosed Ca 2+ concentration. ( ( 3 4) This equation can be expressed in terms of fraction of Ca 2+ bound and free, f. ( ( 3 5) Activity Reduction Due to Ionic Shielding Another means for inadvertently detecting a reduced [Ca 2+ ] Bound evolution would be if the removal of the ions in the mother solution to the LCP phase were to al ter significantly the ionic strength of the solution. Besides maintaining a pH, ion selective electrode standardization requires a matching of ionic strength between the standardizing solution and the experimental solution. Different ionic strengths can lead to different ion activities as shown in the Debye mean activity coefficient with the ionic strength of a solution for dilute ionic solutions: (3 6) Where: z i is the charge number for ion species i A is a solvent dependent constant I is the ionic strength of the solution The solvent constant A has an experimental value of 1.172 mol 1/2 /kg 1/2 .in water. The ionic stre ngth of the solution is a function of the charge and concentration of all the ions in solution:
75 (3 7) Where: c i is the concentration of ion species i It is important to note that even though the assumption was made i n equation 3 3 that carbonate and bicarbonate have a single activity coefficient, the calculation of the ion product remains ideal, treating the carbonate and bicarbonate as individual ions which contribute to the ionic strength. Upon removal of ions from the mother solution into the LCP, the ionic strength of the solution would be reduced, leading to a higher activity of Ca 2+ in the solution and a lower fraction of [Ca 2+ ] Bound being falsely detected. It is likely that both of these phenomena are occurri ng and contributing to the apparent reduction seen in the Ca 2+ bound evolution profile after the emergence of the LCP phase. By assuming that the partitioning of ions between liquid phases is proportional to the carbonate and bicarbonate partitioning, it is possible to express the fraction of counterion in LCP vs. mother solution in terms of binding fraction of Ca 2+ vs. ionic strength to solve for the fraction of ions participating in the LCP phase. By expressing equation 3.5 in terms of after over before the LCP emergence, we yield the following expression: (3 8) Which can be simplified to the following expression : (3 9)
76 Where: (3 1 0) Equation 3 10 is due to the assumption that all the ions partition proportionately between the mother solution and the LCP with respect to each othe r. The subscript 0 indicates the state prior to the phase transition. The left side of equation 3 9 is the ratio of the slopes of a [Ca 2+ ] Bound evolution after vs. before the phase transition. This value will be constant for a given profile if the devel opment before and after the phase transition is linear and intercept the axis at the origin. It is important to note that the mole fractions would usually be difficult to deal with, but the apparent linearity of the Ca 2+ evolution suggests that the mole f ractions are remaining relatively constant in the linear regions. The activity coefficients of equation 3.9 can be expressed in terms of the ionic strength by means of the Debye Huckel equation: (3 11) Which can be rewritten: (3 12) For the assumption that the partitioning of ions between the mother solution and LCP is proportional for all io ns, then ionic strength after the phase emergence can be expressed in terms of the ion partitioning and the ionic strength prior to the phase transition as shown: (3 13)
77 By combining equations 3.12 and 3.13, substituting into equation 3.9, and inserting the appropriate subscript nomenclature, we get the following expression: (3 14) for the carbonate/bicarbonate counterions. Finally, by combining exponential terms, we get an expression that relates our measurable ratio of Ca 2+ bound before and after the LCP emergence vs. the fraction of counterion (carbonate and bicarbonate) that lea ves the mother solution for the LCP phase: (3 15) The mole fractions bound, both prior to and after the phase transition, are obtained experimentally. This equation uses the values obtained from the experiment and back calculates what the mother solution conditions would have to be to change the bound fraction of Ca 2+ from what was measured before the LCP emergence to what was measured after the LCP emergence. Ther efore, we conducted Ca 2+ titrations into a bicarbonate buffer at pH 8.5 with varying amounts of initial Cl ion. Specifically, we compared the system of 0 mM initial Cl (the system described in the polymer free case of C hapter 2), to a system with 10 mM, and 20 mM initial Cl respectively. The free Ca 2+ development was compared for the three cases and NMR was used to verify the formation of LCP. This information allowed for further assertions regarding the nature of the LCP phase.
78 Materials/Methods G eneration of Super Saturated Calcium Carbonate Solution Using a micropipette, 100 mM HCl solution was titrated into a 40 mM sodium carbonate (Fisher) solution to generate a 20 mM, pH 8.5 bicarbonate buffer that was rich in initial chloride concentration ( either 10 or 20 mM initial Cl concentration, depending on the experiment). The solution was counter titrated with nanopure water to adjust volume. Calcium chloride (Fisher, reagent grade) solution with a concentration of 10 mM was titrated into 29 mL of the bicarbonate buffer (containing various concentrations of chloride ion), which was stirred at 100 rpm using a magnetic stir bar. The punctuated volumes of titration were 200 L each, except for the first injection, which was only 40 L to account for infinite dilution phenomenon. Titrations were injected at an approximate rate of 20 L /sec directly over the rotating stir bar to ensure adequate mixing. The titrations were made in a punctuated fashion, 2 minutes of constant [Ca 2+ ] Free measurements were acquired before injecting more CaCl 2 (aq). The solutions were prepared with nanopure water and all were filtered using a 0.22 m Millipore syringe filter prior to any titrations. Each of these titrations was conducted in triplicate and the error express ed in the results is plus or minus two standard deviations. Some solutions were made up to 17 injections of CaCl 2 (aq) and were analyzed using Nuclear Magnetic Resonance (NMR) and Nanoparticle Tracking Analysis (NTA) as described below. Ca 2+ Electrode and pH Electrode Measurements The free Ca 2+ concentration, [Ca 2+ ] Free in the titrated solution was obtained using a Ca 2+ ion selective electrode (Radiometer Analytical, ISE K Ca, E11M006) in conjunction with a reference electrode (Radiometer Analytical E21 M009). A calibration
79 standard curve for calculating free Ca 2+ concentration was generated by 10 mM CaCl 2 (aq) into an aqueous NaCl solution with an appropriate ionic strength (to match 10 or 20 mM Cl initial concentration, depending on the experiment) wh ich had been brought to pH 8.5 by the addition of trace amounts of NaOH (Fisher) (aq). The measured [Ca 2+ ] Free value had to remain constant for at least 2 minutes of mixing before adding another titration injection to verify that the solution equilibrated and that solid nucleation had not yet occurred. It is important to note that we are not quantitatively accounting for CO 2 net diffusion out of the solutions during our experiments. Nanoparticle Tracking Analysis (NTA) of Emergent Phase The hydrodynamic radius of the emergent calcium carbonate LCP droplets was analyzed using the NTA light scattering technique. A bicarbonate buffer, pH 8.5 and with an initial 20 mM Cl ion concentration was injected with 17 injections of 10 mM CaCl 2 (aq) and was analyzed using an LM20 analyzer (Nanosight TM ). The data was processed using an NTA analytical software suite (Nanosight TM ). Samples for analysis 0.3 mL of sample was used in each analysis. NMR, Spin Spin (T 2 ) Relaxation Time Measurement All NMR experiments were conducted on a Bruker Avance DRX 500 MHz vertical bore system using a xyz gradient TXI probe with a 1H and 2H interior coil, 13 C and 15 N exterior coil, and xyz gradient s. All carbonate/bicarbonate buffer solutions were generated as described above except using 100% 13 C enriched sodium carbonate and sodium bicarbonate ingredients (Cambridge Isotopes) to enhance signal/noise. All experiments were conducted at 298 Kelvin. Deuterium oxide was used to obtain a lock at a volume fraction of 2.5% of the total sample. Data was processed using
80 overlapping spectral peaks was required. The T 2 relaxation times o f the various species in solution were obtained using a Carr Purcell Meibloom Gill (CPMG) sequence with msec. 1 H was decoupled from 13 C for the entirety of the pulse seque nce. All processing was zero filled twice and was done with 0.3 Hz line broadening to allow for characterization of NMR spectral features. Results The results of the Ca 2+ titration into bicarbona te buffer, pH 8.5 are shown in F igure 3 2 and were very simi lar to the results obtained previously (Chapter 2) without additional chloride present at the outset. 87 There is a discontinuous change in the Ca 2+ binding development roughly at a bound Ca 2+ concentration of 0.12 mM, which was similar to what was observed in Chapter 2, 87 (see Figure 2 1), suggesting that even in the presence of additional Cl ion, the calcium carbonate LCP phase still emerges from a liquid liquid phase separation. The 20 mM Cl initi al condition was analyzed using the nanoparticle tracking analysis (NTA) light scattering technique. Specifically, a solution with 17 injections of CaCl 2 (aq) was analyzed to verify the existence of a LCP. Object A 3 of the A ppendix is a video of the raw data of the NTA light scattering technique, which shows that there are particles/droplets present which are presumably LCP droplets. Figure 3 3 displays the process results in the form of a particle (droplet) diameter vs. a relative scattering event coun t and show that there are numerous droplets present with a diameter of ~84 nm which is consistent with the LCP droplets previously detected (see Figure 2 3). However, the presence of the additional chloride appears to lead to a portion of the LCP
81 existing as much larger 140 180 nm droplets which were not present in the chloride deficient case (See Figure 2 3). 13 Carbon T 2 NMR relaxation experiments were used to verify the presence of LCP droplets in the 20 mM bicarbonate buffer with an initial chloride con centration of 20 mM at the 17 th injection of CaCl 2 (aq). The same method as described in C hapter 2 and illustrated in A ppendix F igure A 3 was used to deconvolute two separate (assumed Gaussian) overlapping signals for analysis. The results of the T 2 rela xation experiment are shown in Figure 3 4. The bicarbonates/carbonates in the emergent phase have shorter T 2 relaxation times than the bicarbonates/carbonates in the mother solution, yet both have relaxation times consistent with liquids, strongly suggest ing that there are two distinct liquid phases present in the system. This is also consistent with previous experiments regarding the emergent LCP phase. 87 Although the varying initial chloride concentrations did not affect whether or not the LCP phase emerges, it did have an effect on the relative [Ca 2+ ] Bound evolution when comparing before and after the LCP emergence. Figure 3 2 shows that each initial chloride condition during titration development displays a different linear binding Ca 2+ binding affinity prior to than after the LCP emergence (emergence is at the kink). The measured relative change in slope, in conjunction with equation 3 13, can be used to estimate the amount of ions partitioning between the two liquid phases. The generalized solution to Equation 3 13 and the results of this analysis are shown in Figure 3 5. According to this ana lysis, the 0 mM, 10 mM, and 20 mM initial chloride cases yield ~37%, ~20%, and ~30% of the total ions in the LCP phase, respectively. Although the relationship between chloride concentration and amount of LCP formed
82 pear that the higher amounts of chloride ion yield less LCP. The NMR relaxation data was used to further support this trend. Figure 3 6 compares the area attenuation of the LCP Gaussian model peak as compared to the overall (LCP and mother solution) 13 C signal attenuation. Due to possible variability in the deconvolution technique, the NMR peak area was back calculated to the intercept to estimate the ratio of signals, and therefore, the ratio of ions in each phase. The data shows that the LCP for the 0 mM and 20 mM initial chloride conditions contain ~30% and 25% of the ions in the system, respectively. This ratio does not exactly match the ratio calculated from the Ca 2+ titration data, but they are on the same order of magnitude, and support the trend that additional chloride ion leads to less overall LCP phase. Discussion The first thing that should strike a researcher familiar with the concept of precursor phases is the very large fraction of the ions in solution that are participating in the LC P phase. It was proposed that less than 1% of the solution was in the LCP phase prior to this study (Chapter 2). 87 The results reported here suggest that a much larger fraction of the ions are participating in the LCP phase; as much as 20% 40%, depending on the additives in solution. A possible reason for the very low participation previously proposed was our dependence on the NTA technique for a quantitative estimate. NTA has an ideal lower detection threshold of ~20 nm diameter for particles. However, in systems where the particles have a refractive index that is very similar to the bulk solution, t his threshold becomes even higher. In our case, the LCP is proposed to be a ionic solution like the mother solution, and therefore the threshold might be
83 much higher; perhaps 40 nm. If this is the case, then the majority of LCP would not be detected if t hey exist in droplets below this threshold. The results shown in Figure 3 5 and 3 6 suggest that additional chloride results in less formation of LCP. This is also surprising because chloride ion is ubiquitous and in abundance in biological mineralizati on mediums, leading one to conclude that its presence would enhance the amount of LCP (a component of PILP, which is believed to be a common biomineralization mechanism). However, the NTA results, shown in Figure 3 3, indicate much larger droplets of LCP form when chloride is present in 3). This data, combined, suggests that the additional presence of Cl both inhibits overall LCP formation, but increases the average size distribution of the LCP droplet s that do form. This complex behavior of the C l can be explained if the LCP behaves like an emulsion within the system. Reiger et al. 35 and Wolf et al. 54 both proposed that a liquid precursor would behave like an emul sion. Wolf et al. in particular proposed that calcium carbonate LCP is stabilized by a process known as electrostatic stabilization. the surface potential of a droplet in solution in such a way as to effect aggregation/coalescence. 96 This mechanism may explain the observed Cl influence and provide us with a valuable clue regarding the surface potential of calcium carbonate LCP. Although the LCP droplets are charge balanced overall because they are a phase, the surface potential of the droplets appear to be negatively charged. If this is the case, then small amounts of negative chloride ion would inhibit coalescence of the droplets by ai ding in the electrostatic repulsion between droplets. However, as the
84 chloride concentration is increased, coalescence can become an energetically favorable means to reduce the surface area of negative potential exposed to the negative chlorides; thus des tabilizing smaller droplets in favor of larger ones, and reducing the nucleation of new droplets, as observed in Figures 3 3 and 3 5 and 3 6. The T 2 relaxation times for the LCP and the mother solution are longer in the presence of additional chloride ion (compare to T 2 time of chloride deficient case, Figure 2 7 A ). Increased relaxation times can mean that the 13 C nuclei, as part of the bicarbonates and carbonates, are tumbling faster in solution and/or that the average exchange rate of 13 C based ions bet ween the LCP and mother solution phases is increased as the result of the additional chloride. Both of these possibilities (they are not mutually exclusive) are interesting phenomenon if occurring and might lead to further understanding of LCP if investig ated further. Conclusion The evidence provided by these experiments support the view that the LCP phase is a true liquid phase, and not a metastable transition on the way to nucleation, he LCP appears to behave in a manner that is consistent with it being colloidal stabilized droplets with negative surface potential. Additionally, the LCP phase appears to manifest itself much more extensively than previously thought, making it a much mor e reasonable candidate to participate in mineralization than if it were a miniscule fraction of the system. These results may yield additional insight into the PILP mechanism of formation and stabilization. LCP droplets with negative surface potential wo uld be subject to depletion stabilization in the presence of negatively charged polyelectrolytes like the ones employed in the PILP process.
85 Additional work needs to be done to further characterize this system. As mentioned in the introduction section a bove, the liquid liquid phase separation process has at least 5 degrees of freedom. This study only partially analyzed one of them in a somewhat qualitative manner. A fuller understanding of the variables that promote/inhibit/stabilize/destabilize the LC P phase of calcium carbonate may allow for been imagined yet.
86 Figure 3 1. Example of a phase diagram describing the partitioning of species A and B, with an up per consolute temperature limit (left) and a lower consolute temperature limit (right).
87 Figure 3 2. The evolution of [Ca 2+ ] Bound during titration of 10 mM CaCl 2 (aq) into 20 mM bicarbonate buffer, pH 8.5, as measured by a calcium ion selective electro de. A) The evolution of a buffer containing 0, 10, and 20 mM initial Cl concentration. The binding affinity (slope) was measured before and after the liquid liquid phase separation (kink) for solutions with an initial concentration of B ) [Cl ] 0 = 0 mM C) [Cl ] 0 = 10 mM and D) [Cl ] 0 = 2 0 mM
88 Figure 3 3. The NTA obtained size profile of the LCP droplets in a 20 mM bicarbonate solution, pH 8.5, with an initial concentration of Cl of 20 mM after 17 injections of CaCl 2 (aq). The additional chloride at the onset of titration resulted in the formation of larger droplets than in the case where no extra chloride ion was added (see Figure 2 3).
89 Figure 3 4. The results of the NMR T 2 relaxation experiment of the 20 mM bicarbonate solution, pH 8.5, wit h an initial concentration of Cl of 20 mM after 17 injections of CaCl 2 (aq). There are two distinct T 2 relaxations, suggesting two distinct phases are present, verifying the presence of LCP in the mother solution.
90 Figure 3 5. The general solution of Equation 3 13 and its use to measure the partitioning of ions between the LCP phase and the mother solution. The additional chloride ion in the buffer yielded less ions in the LCP phase (more in the mother solution). The generalized solution to the e quation was different for the 0, 10, and 20 mM initial chloride concentrations, but they are nearly overlapping when plotted.
91 Figure 3 6. By expressing the T 2 relaxation data in terms of area of the Gaussian models used to deconvolute the data, the fr action of ions in the LCP phase at the beginning of the experiment ( = 0) is estimated. The extra chloride ions (top) resulted in apparently 25% of the ions participating in the LCP which is reduced from when there is not additional chloride present (bot tom). Lines are present as a guide.
92 CHAPTER 4 BIOMEDICAL APPLICATI ON OF THE PILP PROCE SS: CALCIUM CARBOAN TE COATED EMULSIONS AND LIPOSOMES FOR CONTRO LED RELEASE APPLICATIONS Introduction CaCO 3 based core shell microcapsules are promising candidates for controlled toxicity, biodegradable nature, versatility, and benign synthesis conditions. Currently there is a large focus on controlling the formation of CaCO 3 to yield microparticles for a variety of potential uses, ranging from use as biomedical fillers, to cosmetics and body care lotions, to industrial processing. Potential CaCO 3 core shell particles have been generated by controlling the precipitation of CaCO 3 through the presence of certain block copolymers to direct the polymorph formed 53 97 and morphology 98 99 by using hydrophobic and hydrophilic interactions to direct the CaCO 3 to form capsules through surfactants 100 102 and interfaces, 103 and by the addition of negatively charged electrolytes to form self assembled core shell particles. 104 105 The core shell particles created by these current techniques are often solid solid particles (with a solid coating over a solid core), or are porous for diffusion regulated chemical uptake or release processes. E mulsions and liposomes are promising components for a CaCO 3 coated drug delivery vector due to the ability to tailor their surface properties, and to efficiently store large amounts of release agent in compartments separated from the bulk solution. Oil in water emulsions and liposomes can store hydrophobic and hydrophilic release agents, respectively, without altering the solution properties where the CaCO 3 precipitation is occurring (provided there is no leakage of the sequestered species).
93 Due to these advantages, there is an increasing amount of current research involving the encapsulation of emulsions and liposomes with CaCO 3 such as direct, rapid precipitation onto a favorable emulsion interface, 106 107 or through the pickering process followed by mineral ripening. 108 However, these techniques tend to generate core shell particles that have faceted, rough surfaces, w hich due to the non uniform surface coverage, may lead to complex release profiles. Building on previous work, 109 we show here that the polymer induced liquid precursor (PILP) process is an ideal means of coating the surface of emulsion droplets and liposomes with a smooth and conti nuous CaCO 3 shell, to yield CaCO 3 coated microcapsules containing release agents. The PILP process utilizes negatively charged polyelectrolytes (such as polyaspartic of polyacrylic acid) to direct the mineralization process. However, unlike other polymer additive techniques, which utilize the additive to direct growth by templation, 110 selectiv e orientation, 111 or promotion/inhibition of crystallographic plane growth, 112 113 the PILP process utilizes the polyelectrolyte to transform the conventional crystallization reaction into a two step non classical crystallization process. 52 114 In other words, the negatively charged polyelectrolyte sequesters calcium and carbonate ions, along with their hydration waters, to induce (or stabilize ) 87 liquid liquid phase separation in the crystal lizing media, yielding nanosize ion enriched droplets, which we refer to as polymer induced liquid precursor (PILP) droplets, within the aqueous solution. The droplets deposit on substrates, coalesce i nto a continuous coating, and solidify and crystallize into solid calcium carbonate through the expulsion of water and polyelectrolytes. 114 The presence of Mg 2+ ions during the formation of the CaCO 3 PILP precursor (at levels similar to sea water) can aid the
94 polymer in producing the fluidic character of PILP droplets, while tending to reduce any side products of conventional rhombic calcite crystals. 115 The liquid precursor, when aided by these additional additives (such as Mg 2+ ) can lead to smooth, continuous CaCO 3 films with preferential surface deposition, as has been shown in our prior work on glass substrates. 52 In addition, we have found that these films of CaCO 3 can be patterned using templates of self assembled monolayers, 116 where the PILP droplets were found to preferentially adsorb to negatively charged carboxylate terminated alkane thiols, as opposed to the more hydrophobic methyl terminated alkane thiols. 117 Thus, we hypothesized that the PILP droplets might also deposit on negati vely charged surfactants which are used to stabilize emulsions, thus enabling a way to coat emulsions (and liposomes) to form microcapsules. Of course these colloid systems contain large curvature as compared to the flat substrates we have coated in the p rior work, so this was an important question to address. The PILP process is considered biomimetic because it reproduces many of the non equilibrium crystal morphologies found in biominerals simply by the addition of negatively charged acidic polyelect rolytes which emulate the acidic aspartic acid rich proteins utilized by organisms to direct biomineralization. 3 in the development of CaC O 3 core shell particles comes from the Dinoflagellate cyst. 118 This single celled organism coats itself with a CaCO 3 mineral shell, in effect making it a little living microcapsule. This gave us the idea that fluidic droplets could be coated with a mineral shell, all while under benign processing conditions which could enable a variety of thermally sensitive components (such as proteins for example) to be encapsulated.
95 In this report, CaCO 3 coated microcapsules with fluid interiors were generated using the PILP process to coat oil in water emulsion droplets (for hydrophobic interiors) and phospholipid bilayer liposomes (for hydrophilic interiors). Through the use of polarized light microscopy (PLM), we demonstrate a level of control over the phase of the CaCO 3 (crystalline vs. amorphous). Scanning electron microscopy (SEM) was used to examine the core shell microstructure, as well as qualitatively demonstrate the pH dependent degradation of the core shell microcapsules. The encapsulation capability of the microcapsule interior was demonstrated using confoc al fluorescence microscopy to detect the presence of entrapped hydrophobic and hydrophilic fluorescent model compounds within the interior of the emulsion based and liposome based microcapsules, respectively. Materials/Methods Emulsion P reparation Oil in water emulsion droplets were prepared by blending, in a house hold blender, n dodecane oil (Fluka) containing 1% w/v stearic acid (Aldrich) and deionized water to form an emulsion with a 1:9 oil/water volumetric ratio. The deionized water was adjusted to the desired pH between 7 and 11 using 0.01M NaOH (Fisher Scientific) in deionized water prior to blending. For experiments requiring the entrapment of a hydrophobic model compound, Nile Red dye was added to the oil phase prior to blending. Liposome P rep aration Unilamellar liposomes with diameters of < 2 m were prepared using the solvent mediated dispersion method. 119 100 L of an organic phase, consisting of 20 mg/ mL 1,2 distear oyl sn glycero 3 phosphotidyl choline (DSPC) (Avanti) and 8 mg/ mL
96 cholesterol (Sigma Aldrich) dissolved in chloroform, was injected at a steady rate of 0.25 mL /min into 3 mL of water. The water was at a temperature of 80 C during the injections. The aqu eous solution was then cooled for 30 minutes at 4 C, resulting in the formation of large unilamellar liposomes with diameters up to 2 M. Core Shell Microcapsule Synthesis 1 mL of either the emulsion or the liposomes was pipetted into a 35 mm Falcon pol ystyrene petri dish, followed by 1 mL of 80 mM/400 mM CaCl 2 /MgCl 2 solution (Sigma Aldrich) freshly prepared using deionized water, and filtered through a 0.2 2 m Millex syringe filter. 36 L of a freshly prepared and filtered 1 mg/ mL polyelectrolyte sol ution was transferred to the petri dish by micropipette. In the case of emulsion coating, poly ( L ) D,L aspartic acid sodium salt ( 10,300 g/mol, Mw, Sigma Aldrich) was used as the polyelectrolyte. In the case of liposome coating, polyaspartic acid sodium salt (14,000 g/mol, monodisperse, Alamanda Polymers) was used. The petri dish w as then covered by parafilm through which a small hole was punched into which the outflow end of the tubing from an ultra low flow peristaltic pump (Fisher Scientific) was i nserted. At a rate of approximately 0.0 32 mL /min, 2 mL of a freshly prepared and filtered 300 mM (NH 4 ) 2 CO 3 (Sigma Aldrich) solution was pumped into the petri dish over a period of ~6 0 minutes. A schematic of the process and the desired product is shown in Figure 4 1. The resulting mineral product was collected and centrifuged at 8000 rpm for 10 minutes. The supernatant was discarded and the product pellet was rinsed with anhydrous ethanol (Fisher Scientific) and re centrifuged under the same conditions. T he second supernatant was discarded and the product pellet was either resuspended in 2 mL of anhydrous ethanol or placed on a glass slide and dried under room temperature or vacuum conditions, depending on the characterization technique to be employed.
97 Flu orescence Imaging of Encapsulated Model Compounds To demonstrate the ability to entrap chemicals of interest in the microcapsules, Nile Red (Sigma) and Rhodamine 110 (AnaSpec) fluorescent dyes were used to simulate hydrophobic and hydrophilic model compoun ds, respectively. In the case of the emulsion derived core shell particles, Nile Red was added to the n dodecane to mL synthesis. In the case of the liposome derived core s hell particles, Nile Red was added mL mL concentration prior to liposome synthesis. After synthesis, the mic rocapsules were centrifuged and washed as described in the core shell microcapsule synthesis section and were stored in ethanol for imaging. Fluorescence imaging was conducted with a confocal microscope setup consisting of an Olympus IX 81 inverted micros cope with an Olympus Fluoview 500 confocal scanning system with a tunable excitation laser. The images were taken with a 20x 0.70 NA objective. The Nile Red was excited at 543 nm and the emission was collected using a 560 nm longpass filter after focusin g the image to the highest fluorescence intensity. All images were taken and analyzed using the Fluoview software. Polarized Light Microscopy for Characterization of CaCO 3 Shell Crystallinity Polarized light microscopy was used to detect the presence or ab sence of birefringence from the CaCO 3 shells to determine if the CaCO 3 mineral was amorphous or crystalline. The birefringence of core shells created in a 2.5:1 ratio of Mg 2+ /Ca 2+ environment (described above) were compared to core shells created in a 5:1 ratio of Mg 2+ /Ca 2+ environment (the CaCl 2 concentration was halved). The light microscopy
98 images were obtained using an Olympus BX60 polarized light microscope. The use of a 1 st order red gypsum plate to display birefringence was used as the situation warranted, such as to see both amorphous and crystalline materials. Scanning Electron Microscopy (SEM) for Morphological Analysis To determine the overall particle morphology as well as the thickn ess of the mineral shell, microcapsules were created as described above and resuspended in 2 mL anhydrous ethanol. 200 microscope slide (Fisher) and allowed to air dry. A second microscope slide wa s placed on top of the microcapsules and gently tapped to crush the core shell particles. The crushed particles were then sputter coated with Au/Pd and analyzed using a JEOL 6400 SEM with an accelerating voltage of 15 kV. To demonstrate pH dependent degra dation, core shell microcapsules were created as described in the synthesis section except that, prior to final suspension in ethanol, the microcapsules were suspended in 2 mL of either a 0.1 mN, pH 4 HCl aqueous solution or pH 6.8 nanopure water. In the case of the pH 4 HCl aqueous solution, the sample was resuspended and swirled gently, by hand, for 15 seconds at which point the solution was quenched with 2 mL of a 0.1 mN, pH 10 NaOH aqueous solution. Immediately following the quenching, 36 mL of anhyd rous ethanol was added to halt any further degradation. The whole suspension was then centrifuged at 8000 rpm for 4 minutes, the supernatant was discarded, and the pellet was resuspended in 1 anhydrous ethanol for analysis. In the case of the pH 6.8 nano pure water, the sample was resuspended and swirled gently, by hand, for 2 minutes. Immediately following the swirling, 18 mL of anhydrous ethanol was added. The whole suspension was
99 centrifuged at 8000 rpm for 4 minutes, the supernatant discarded, and th e pellet was resuspended in 1 mL anhydrous ethanol for analysis. 200 L of the degraded core shells suspensions were deposited on a glass slide, dried, and sputter coated with Au/Pd. The visual SEM analysis was conducted using a JEOL 6400 SEM with a 15 k V accelerating voltage. X ray Diffraction (XRD) To verify the crystalline phase of the CaCO 3 coating of the synthesized core shell particles, XRD was conducted. Several batches of core shell p articles stored in ethanol were centrifuged at 8000 rpm for 10 minutes and the supernatant was r emoved. The centrifuge tube was then covered was parafilm with holes punched in it and allowed to dry overnight. The resulting powder was grounded and placed on a glass slide and then scanned with Cu ray radiation from a PANalytical X'Pert P ro Powder Diffractometer at 45 k V and 40 mA, using a step size of 0.02 150. Results A Mg 2+ aided PILP process was used to deposit CaCO 3 shells onto the surfactant mon olayer surface of oil in water emulsion droplets. The oil droplets consisted of biocompatible n dodecane stabilized in solution by a stearic acid monolayer at the oil water interface. The resulting products we re spherical CaCO 3 microcapsules dispersed i n the aqueous solution, with diameter s slightly larger than the original emulsion droplets (typically 2 to 10 microns) due to the addition of the mineral coating. Light microscope images and Polarized Light Microscopy (PLM) images of the core shell microc apsules, created by PILP coating either oil in water emulsion droplets or liposomes, are shown in Figure 4 2A,B and C,D respectively. As evidenced by their
100 birefringent nature ( Figure 4 2B), the mineral shells appear to be crystalline. The presence of birefringence in the center of the spheres, suggests a non crystalline interior (presumably the fluid) with a surrounding shell composed of radially oriented polycrystalline CaCO 3 Core sh ell microcapsules created by coating liposome vesicles using the PILP process appear to be much smaller, with a diameter of less than 2 m ( Figure 4 2C). This size is consistent with large liposomes, which are created by the ether injection technique ( 1 m in diameter), that have been coated with a mineral shell. Similarly, they also display birefringence in the shell and appear to have a non birefringent interior ( 4 2D), suggesting a crystalline core shell structure. However, the light microscopy technique. SEM was used to image the core shell morphology, determine the core shell thickness, and to qualitatively demonstrate pH dependent degradation of CaCO 3 coated emu lsion droplets and liposomes. A collection of both types of microcapsules are shown in Figure 4 3, illustrating that these liquid droplets can be dried down to a powder. The microcapsules appear to be fairly uniform, but with some polydispersity in size, more so with the liposomes. A close up image of a single emulsion microcapsule (Figure 4 4 A) shows how smooth the surfaces of these particles are. Although there is a little bit of debris on the surface, we refer to this as smooth as compared to the alt ernative approach of simply nucleating crystals on a surface, which creates a surface composed of three dimensional aggregates of rhombic calcite crystals (as does the P ickering process). 108 The core shell thickness of the coated emulsion microcapsules
101 was determined b y intentionally fracturing the microcapsules and taking SEM micrographs of the fractured shells (Figure 4 4B). The thickness of the shells was found to be relatively uniform at ~600 nm for several batches of products all synthesized using the identical sy nthesis method described in the experimental methods section. The phase of the CaCO 3 shell appears to depend on the conditions used during precipitation. Figure 4 5 shows that, for the emulsion derived microcapsules, an initial ratio of 2.5:1 Mg 2+ /Ca 2+ yielded crystalline shells, but an initial ratio of 5:1 Mg 2+ /Ca 2+ yielded an apparently amorphous mineral shell, even after gentle heating at 50 C for 24 hours. This suggests an element of control may be achieved over the final phase (crystalline or am orphous) of the core shell product and, in turn, over the degradation properties of the microcapsules, since amorphous calcium carbonate is more soluble than the crystalline form. The amorphous core shell microcapsules were smaller in size (2 3 m) than t he crystalline microcapsules (2 10 m), which is presumably due to the influence of Mg ion on the stability of the emulsion droplets, where the larger droplets may become destabilized with the combination of calcium and magnesium ions. Mineral phase iden tification of the CaCO 3 coating on liposome derived microcapsules created using the 2.5:1 Mg 2+ /Ca 2+ ratio was carried out using powder XRD (Figure 4 6). The pattern contains diffraction peaks that are consistent with the JCPDS standard for calcite, with s ome peak shifts caused by lattice strain from the incorporation of Mg into the lattice which results in slightly smaller characteristic d spacings. The results are consistent with the Mg 2+ incorporation behavior found in calcium carbonate films generated by the PILP process under similar Mg 2+ /Ca 2+ ratios. 115 The absence of the characteristic signal of vaterite at 2
102 aragonite at 45.9 suggest that neither of these phases is present in any significant amount. These characteristic peaks are indicati ve of polymorph presence and are used specifically to quantify the polymorph distribution in calcium carbonate samples. 120 The ability to entrap active agents, and the versatility of these microcapsules is demonstrated with model active agent compou nds. Rhodamine 110 and/or Nile Red fluorescent dye were encapsulated during the synthesis of the core shell particles, and washed particles were subsequently examined by confocal fluorescence microscopy. Confocal microscopy analyzes a specific focal dista nce through the microcapsules, allowing one to discern between fluorescence within the microcapsule and from within the CaCO 3 shell or surrounding solution. Figure 4 7 shows the entrapment of Nile Red fluorescent dye within the emulsion derived microcapsu les. Nile Red has the interesting property of fluorescing only in very hydrophobic environments. Although it is conceivable that a small portion of the model compound could become entrapped within the mineral shell itself, the fluorescence seen in Figure 4 7 indicates that the interior of the microcapsules indeed contains the stored model agent. Fluorescence should not be observed from dye entrapped in the shell or adsorbed to the exterior of the particles due lts also show that the generated microcapsules retain the active agent even under the stress of centrifugation and resuspension in anhydrous ethanol. The liposome derived core shell microcapsules display the ability to entrap both hydrophobic ( Nile Red dy e) and hydrophilic compounds (Rhodamine 110), as is shown in Figure 4 8A and 4 8B, respectively. The overlap of fluorescence ( Figure 8C) demonstrates that the core shell microcapsules have the
103 ability to simultaneously entrap very different compounds in a single liposome based microcapsule. When we first considered this system, there was some concern that the CaCO 3 shells would degrade too slowly given that calcite has a rather low solubility, and of course it is notorious for clogging up industrial pip elines, etc. So we performed a preliminary study of the degradation potential of these particles, which is shown in Figure 4 9. It was found that the emulsion derived microcapsules degrade readily and in a pH dependent manner, enabling a triggered releas e of active agent. The microcapsules exposed to the acidic (pH 4) environment for 15 seconds experienced severe degradation of the CaCO 3 shell around the entire core shell particles. The microcapsules exposed to pure water (pH = 6.8) for 2 minutes experi enced degradation as well, but the dissolution of the mineral was unevenly distributed around the microcapsule, with some hemispheres completely dissolved while others appeared relatively undisturbed. This might be due to channeling of the water through s tacked core shell spheres during the mixing step which, due to the preliminary nature of this study, was conducted by simple swirling by hand for the indicated time period. The volume of shell degraded appears to be similar for both degradation conditions but it is important to note that the microcapsules exposed to neutral pH required 8 times the exposure time to experience similar amounts of degradation as the microcapsules exposed to the acidic environment, and that the degradation in pH 4 conditions a ppears to be a uniform thinning, whereas at pH 6.8 it exhibited a more asymmetric degradation. Discussion In our prior studies, free standing flat films of CaCO 3 were deposited on the anionic head groups of fatty acid monolayers. 117 121 Here we show that analogous
104 templates containing the charged head groups of stearic acid and DSPC phospholipid are effect ive, even when there is significant curvature in the template composed of emulsion droplets and liposomes. The exciting advance provided by this core shell particulate system is that one can take suspensions of liquid droplets and in effect dry them down to a powder, while retaining the fluid interior of the droplet that contains the active agent. which a chemical of interest can be stored/protected at high pH, in organic solv ent or dried down as a powder, until pH or mechanically triggered release. Although not examined here, one should be able to achieve the high encapsulation efficiencies analogous to what can be obtained by the core droplet systems to begin with, but with much easier separation and storage of the microcapsules. The CaCO 3 coating is applied under benign conditions, allowing for the coating and protection of delicate systems as a powder for easier transport and potential long term storage. Storage and tra nsport of a powder provides a major advantage over transport of the large amount of volume required in emulsions and liposome suspensions. The benign processing conditions of this reaction make it highly valuable for encapsulation of sensitive compounds t hat might not be amenable to other synthetic techniques. Of particular relevance could be the encapsulation of sensitive growth factors, or proteins, DNA, etc., and perhaps even live cells. When evaluating a microcapsule for potential use in food and ph armaceutical applications, it is important to consider the tailorability and intrinsic properties of the core shell microcapsule. These properties include the size, the triggering mechanism
105 for release/degradation, the biocompatibility, the encapsulation efficiency and versatility, and the cost, of the microcapsule. Considering the above criteria, the microcapsules generated in this experiment should be ideal candidates for many applications. Our core shell microcapsules consist of a continuous CaCO 3 coating (Figure 4 4), rather than a porous polymer shell, as is found in many other delivery systems. Therefore, even small molecular weight active agents can be stored in the interior of the vesicle without fear of early diffusion release, and without h aving to design a new shell with different pore sizes. On the other hand, these particles could potentially serve as templates for deposition of further materials, such as polymeric coatings that can be tailored for a designed release rate, or have functi onality for attaching ligands for targeting cells. The results also suggest that the generated microcapsules retain the active agent even under the stress of centrifugation, and suspension in anhydrous ethanol. In addition, the liposome derived microcapsu les reported in this study demonstrate versatility in entrapment of either or both hydrophilic and hydrophobic compounds of interest. The PILP generated microcapsule processing allows for the tailoring of microcapsule sizes. The PILP process coats the ve sicles by deposition of precursor droplets. Therefore, the size of the resulting microcapsule is dependent on the size of the vesicles at the time of precipitation; a property that is can be adjustably tailored. This is evident due to the fact that the s ize of the resulting core shell microcapsule is slightly larger than the size of the emulsion droplet or liposome used in the production. However, it is not clear that this method could be used for nanosized droplets because
106 the thickness of the mineral la yer would be so much larger. The reaction might need to be quenched in that case. With regards to the third desirable property of microcapsule release systems, it was demonstrated that the microcapsules are degradable, and that the degradation response c an be enhanced based on a pH triggering mechanism. The fact that these particles are readily degraded, even though calcite has low solubility, is likely due to the non equilibrium morphology of the calcite that is produced by the PILP process because thes e coatings expose a curved mineral surface which is energetically unfavorable Given the spherulitic texture, as well as the magnesium incorporation, there is also a lot of defect texture which could enhance their solubility. The ability to control the amorphous vs. crystalline phase of the core shell may allow for additional tailoring of the solubility/surface properties of the microcapsule. Amorphous calcium carbonate can be induced to form in lieu of the more stable polymorphs such as calcite, aragon ite, and vaterite, due to environmental factors, including the presence of Mg 2+ and negatively charged acidic polymer/protein residues. Some examples of applications that could utilize biodegradable microcapsules include pharmaceutics and industrial applic ations, where a stored powder might dependably stabilize the entrapped compound effectively for long periods of time, but once exposed to neutral or lower pH aqueous environments, release could be triggered. For example, rainwater could release pesticide or fertilizer, or addition of water or pH could trigger release of a catalyst in a reaction solution. Calcium carbonate is more soluble in acidic solution than it is in basic pH solutions, therefore, the microcapsules would be ideal for storage in aqueous solutions that are saturated in calcium carbonate.
107 They should also be highly stable in non aqueous solutions (such as in the anhydrous ethanol used here), or other organic solvents. We anticipate that the release of the entrapped compound would be a r elatively rapid or even a burst type of release in the physiological environment, which could include oral ingestion for various pharmaceutics, as well as dental products for remineralization of enamel. Calcium carbonate is a biocompatible material, and c an even provide a beneficial mineral supplement. These particles might enhance the flow of particulates in inhaler delivery systems, and with rapid release of the drug within the mucousy environment of nasal or lung tissues. They may also be suitable for situations where physical rupturing of the core shell would be a trigger for release. This is particularly popular these days in skin care products, such as cosmetics or lotions, where physical application includes rubbing on a lotion, which would then cr ush the core shells to release entrapped compounds. Similarly, hair care products could release conditioner during the shampoo rinse with water. The microcapsules might also be useful for tissue engineering scaffolds, where they could be readily incorpora ted for encapsulation and release of sensitive growth factors or other compounds. This would be particularly useful for hard tissue biomaterials, where the microcapsules could be incorporated throughout the composite scaffold, both to strengthen the compos ite 122 as well as to release some agent or trigger self healing in the material. 123 124 Dissolution of CaCO 3 can help to buffer aci dic media which is created by the degradation of many polymeric scaffolds, such as PLGA, which cause biocompatibility problems. Lastly, the mineral shell could act as protective coating over existing polymeric core shell delivery
108 systems, or serve as a te mplate for additional surface modification with systems that This work demonstrates a novel application of the biomimetic PILP process, which is based upon microorganisms who biomineralize a protec tive mineral shell around their soft and fluidic interior. Most CaCO 3 based biominerals exhibit non equilibrium morphologies (non faceted crystals), such as elaborate crystal morphologies with smoothly curved surfaces. Our group has proposed that these f eatures could be a result of a PILP type process being involved in the formation of biominerals. 3 This enables the formation of smoothly curved surfaces, as highlighted by the core shell particle system shown here, which emulates the CaCO 3 shell of the dinoflagellate cyst. 118 The smooth and continuous calcite coating that the PILP process generates is quite remarkable considering that calcite usually forms faceted crystals when generated in vitro by the conve ntional crystallization process. One can easily envision that the PILP process can allow for an organism (in vivo ), as well as the materials engineer ( in vitro ), to generate novel CaCO 3 products, such as the core shell microcapsules shown here. Conclusion We have prepared microcapsules for potential use in drug delivery and other industrial applications by coating a mineral precursor onto emulsion droplets and liposomes, thus yielding fluidic droplets coated with a solid mineral shell. The production pr ocess allows for the tailoring of many important drug delivery vector variables. The capacity of the microcapsules to store active agents has been demonstrated by visualizing the presence of hydrophobic (Nile Red) and hydrophilic (Rhodamine 110) fluoresce nt dyes as model entrapment compounds. The
109 microcapsules can be dried down to a powder, and when reconstituted in solution, have a pH sensitive triggering mechanism for shell degradation and release of active agent. We believe these microcapsules are exc ellent candidates for use in the food, pharmaceutical, body care, and chemical industries as encapsulation vectors due to their low cost, biocompatible nature and versatility in storing either hydrophobic and/or hydrophilic compounds of entrapment interest The biomimetic inspiration for this work, the dinoflagellate cysts, has provided a nice platform for the design of applicable core shell particle systems.
1 10 Fig ure 4 1 A schematic describing the synthesis of CaCO 3 core shell microcapsule s Emulsion droplets or liposomes (emulsion shown above) are dispersed within an aqueous solution containing CaCl 2 MgCl 2 and polyaspartic acid. NH 4 CO 3 (aq) is pumped into the solution to initiate the formation of PILP droplets, which then adsorb onto the charged s urface of the emulsion/liposome. The liquid like character of the PILP droplets allows them to coalesce into a smooth and continuous coating, which then solidifies and crystallizes into a continuous shell of calcite. An entrapped agent can be stored in ei ther the aqueous interior of liposomes or the oily interior in emulsions
111 Fig ure 4 2. Light microscopy images of CaCO 3 coated emulsion droplets and liposomes. A) Brightfield light microscopy of coated emulsion droplets. B) Polarized Light Microscopy (PLM) of the same particles shown in A). The indicating the product is crystalline with radial alignment of the polycrystals. C) Brightfield light microscopy of coated liposomes. D) PLM of the product seen in C). Although very small, the clear interiors of the particles can be seen, suggesting a crystalline core shell structure has been achieved.
112 Fig ure 4 3. Scanning Electron Microscopy (SEM) images of the core shell microc apsules. (A) Core shell microcapsules with oil interior created by coating oil in water emulsion droplets with CaCO 3 using the PILP process. Their diameters range from 2 to 4 m. in this batch. (B) Core shell microcapsules with aqueous interior created by coating liposome vesicles with CaCO 3 using the PILP process. Their diameters are distributed below 2 m. Both emulsion and liposome derived microcapsules are slightly larger than the emulsion droplets and liposomes alone due to the ~500 nm thickness of the applied mineral coating.
113 Fig ure 4 4 Scanning Electron Microscopy (SEM) images of intact a nd intentionally fractured CaCO 3 coated emulsion microcapsules. A) The core shell microcapsules consist of a smooth and continuous crystalline CaCO 3 coating. B ) Intentionally f ractured microcapsules show a relatively thick continuous shell with a hollow interior where the liquid resided. The shell thickness was consistently ~500 nm using this procedure. A A B
114 Fig ure 4 5 Potential t ailoring of microcapsule crystallinity as determined by polarized light microscopy. The images on the left are optical micro graphs, while the images on the right are using polarized light. (A & B) Microcapsules generated using a 2.5:1 ratio of Ca/Mg display birefringence, indicating crystallinity Image A is an optical micrograph and image B is a polarized light micrograph. ( C & D ) Microcapsules generated using a 5:1 Mg/Ca ratio lack birefringence indicating that the mineral remains amorphous under these conditions. Image C is an optical micrograph and image D is a polarized light micrograph. A gypsum waveplate was used fo r image B, giving a magenta (1 st order red) background, because the amorphous particles lack birefringence and are therefore difficult to see under crossed polars. If they were birefringent, one would see orange and blue colors.
115 Fig ure 4 6 X ray dif fraction (XRD) analysis of the CaCO 3 shell s of liposome derived microcapsule s The XRD pattern display peaks characteristic of calcite. The theta values are slightly larger than expected for pure calcite due to magnesium ion incorporation (which is small er than calcium ion, and thus reduces the lattice dimensions). The mineral coating does not contain significant amounts of vaterite or aragonite as evidenced by the absence of peaks at their characteristic high intensity locations of = 2 5.0 and = 45.9 respectively.
116 Fig ure 4 7 Confocal fluorescence microscopy of microcapsules from CaCO 3 coated emulsion droplets with Nile Red fluorescent dye entrapped with in the oily interior of the emulsion Confocal fluorescence images are shown in A & C, while B & D show an overlay with the bright field images. The high intensity of the fluorescence suggests that the entrapment of the active agent was very successful, as seen in both a large collection of particles in ( A ) as well as in individual microca psules shown at higher magnification in ( C ). The fluorescence corresponds to the interior of the microcapsules for the cluster ( B ) and the individual microcapsules ( D ) when overlapped with their respective bright field images.
117 Fig ure 4 8 Confocal flu orescence microscopy demonstrat ing the entr apment capability of the CaCO 3 coated liposome s. Rhodamine 110 and Nile Red dyes were used as model hydrophilic and hydrophobic entrapment agents, respectively. A) Rhodamine 110 is a hydrophilic fluorescent dye that would be expected to reside in the water interior of the microcapsule. The bright ring around the particles suggests that some of the dye may also bind to the charged headgroup of the liposome or be incorporated into the shell during the PILP process B) Nile Red is a hydrophobic fluorescent dye that would be expected to reside in the hydrophobic region containing the hydrocarbon tail groups of the phospholipid bilayer. C) An overlay of A and B shows that t he liposome based microcapsule system displa ys versatility in the types of encapsulated agents that can be incorporated
118 Fig ure 4 9 SEM micrographs of water and acid dissolved microcapsules. (A & B) Scanning electron micrographs of emulsion based core shell microcapsules after being exposed t o a pH 6.8 nanopure water for 2 minutes and (C & D) after exposure to a pH 4.0 HCl solution for 15 seconds. The rounded and particulate character of the remaining shells may be showing a memory effect of the PILP droplets that had initially coalesced to f orm the shells.
119 CHAPTER 5 CONCLUSIONS AND OUTLOOK I have presented supporting evidence to the theory that CaCO 3 nucleates by means of a non classical nucleation pathway where it first forms a liquid condensed phase (LCP) through a liquid liquid phase s eparation before solidifying into solid polymorphs of mineral. I have provided evidence that the LCP phase is truly a bicarbonate rich liquid phase (as opposed to an ion complex) and that it is a major fraction of the solution system. I have provided evi dence that supports the notion that the polymer induced liquid precursor (PILP) phase is comprised of a LCP phase which is stabilized/ promoted by the presence of the negatively charged polyelectrolyte. Finally, I have successfully demonstrated an applicat ion of PILP by synthesizing microcapsules comprised of a smooth, continuous CaCO 3 (calcite) shell surrounding a liquid emulsion or liposome core containing a hydrophobic and/or hydrophilic chemical of interest for controlled chemical release applications. All of this is due to the discovery of a novel non classical metastable liquid phase intermediate of CaCO 3 which was inspired by biomineralization processes seen in nature. The NMR spectrum shown in Figure 2 10 was presented as evidence that the PILP p hase consists of the bicarbonate rich LCP phase which is stabilized by the presence of polymer. The data was interpreted to suggest that a portion of the ions in solution are in a liquid phase whose up field chemical shift difference from the mother solut ion is due to the more acidic, bicarbonate rich environment. But it was also briefly spectrum seen in solid NMR when a molecule/complex experiences different chemical s hielding at various orientations with respect to the magnetic field; they have chemical
120 shift anisotropy. Considering that the T 2 relaxation of the emergent bulge seen in Figure 2 10 is the same as the mother solution, a powder pattern interpretation may be more valid than the one presented in C hapter 2 Complexes often have a different chemical shielding in the x, y and z directions with respect to the static applied magnetic field. In solutions, where tumbling is rapid, the differences average out o ver time to yield a tight spectral peak. When tumbling is slow or non existent, as in solids, the differences manifest as a widening of the spectral peak due to the non averaged distribution of orientations. Figure 5 1 shows an example of this phenomenon which yields a powder pattern. In Figure 5 1, the chemical shielding of the various axes were defined as x < y < z for the sake of the example. The resulting powder patterns shown in the figure demonstrate the effect of anisotropic tumbling on the expec ted NMR spectrum. Part C of Figure 5 1 resembles the NMR spectra obtained for the PILP phase shown in Figure 2 10. This suggests that the bicarbonate ions may be spinning rapidly along two axes ( designated x and y in the Figure 5 1 ) but that a third shie lded axis is tumbling relatively slowly. This would occur if the bicarbonate complex comprising the emergent phase is more cylindrical in shape accentuating the chemical shift anisotropy in one of the three dimensions When no polymer is present, the re sulting bulge of the NMR spectrum shown in Figure 2 6 resembles part B of figure 5 1. This would be due to a slower rotation of the shielded axis (perhaps all axes) but to a much lesser degree than in the PILP case resulting in accentuation of the chemic al shift anisotropy in all 3 dimensions The emergent bicarbonate species is still behaving as a small solute/complex in solution in both cases because the breadth of the powder peak is much smaller in the presence of emergent
121 phase (~0.05 ppm seen in F i gure 2 10) than expected for small solid particulates which can be as large as several hundred ppm. In order to demonstrate whether the interpretation of the PILP phase presented in C hapter 2 or as a powder pattern is more accurate, further NMR experiment s involving spin angles are required. A magic angle spin NMR experiment can be conducted to verify whether the up field bulge in the presence of LCP (in the presence and absence of polymer) is due to chemical shift anisotropy. The sample tube is held a respect to the static magnetic field and is rotated on this axis at a rate of several kHz to simulate molecular tumbling in solution. At the magic angle and at a rapid rotation, the anisotropic shielding component can be av eraged to zero. If the widening of the NMR spectrum due to LCP is an anisotropic tumbling phenomenon, the width of the spectrum should tighten considerably. The up field bulge should reincorporate with the down field tight peak as the rate of rotation of the sample tube approaches that of the fast rotation axes of the bicarbonate complex. If the bicarbonate ions are present as nanoscopic solid particulates, the spectrum should tighten into a series of small peaklets rather than a single peak to represent Further detailed information can be obtained about the behavior of the carbonate/bicarbonate ions in the LCP phase by conducting NMR spin experiments at angle other than the magic angle (see Fi gure 5 2). Each axis (assuming x, y, and z shown in Figure 5 1) could have different magnitudes of chemical shielding. By measuring the frequency of spin at various tube angles with respect to the bulk magnetization, information regarding the tumbling ra te and magnitude of chemical shielding around each axis of the complex could be obtained. This information in
122 conjunction with computer simulations could allow for a more complete understanding of the structure and behavior of the carbonate/bicarbonate st ructures forming during LCP emergence. Although novel in the field of biomineralization, other research fields are familiar with the concept of non classical nucleation through a LCP. It was over 35 years ago when the protein community began reporting th e unusual formation of a liquid condensed phase of crystallins 125 and lysozyme. 126 That community has since described the thermodynamics of protein condensed phase formation quite extensively for several more proteins, and have established a non classical pathway through which they nucl eate. 127 128 However, it is not inherently obvious th at ionic solutions would behave similarly to protein solutions. In that respect, the physical chemistry field is quite familiar with ionic solution liquid liquid phase separation phenomenon as well. Theoretical models have been successfully used to inves tigate solute induced LCP nucleation of different salts caused by selective solvation 129 and by the close proximity of charged surfaces in solution. 130 The later driving force is particularly pertinent to to the concept of biomineralization where calcium carbonate LCP may be selectively deposited or formed near surfaces whose surface properties are carefully controlled by the organism. non classical nucleation through a metastable liquid precursor, it would be advantageous to consult established wo rk in analogous systems within these other fields. The physical chemistry research community has established a thermodynamic understanding of the principles that drive ionic solutions to undergo liquid liquid phase
123 separation, 92 131 to interact with proteins/polymers, 91 and to nucleate in a non classical manner, 132 which is analogously applicable to CaCO 3 LCP, CaCO 3 PILP, and CaCO 3 non classical nucleation, respectively. Lutsko et al. 132 suggests that, from modeling idealized solute behavior, a critical nucleus formation event is less driven by overall thermodynamic favorability than it is by kinetic favorability. What is required for nucleation is a kinetically favora ble environment that contains "spatially extended, low amplitude density fluctuations" 132 a condition that might not be found in the bulk mother solution, but may be found in a separate, viscous liquid phase; similar to the LCP presented in this dissertation. Solute induced liquid liquid phase separation is well known and of particular interest in the field of salt melts (molten salts at very high temperatures which h ave unusual conductivity properties). Low temperature salt melts are studied at ambient temperatures due to their practicality in applications and to more easily verify high temperature melt behavior models. 131 They are known to be subject to a liquid liquid coexistence curve. 89 Simula tions even suggest that NaCl(aq) solutions undergo liquid liquid phase separations when supercooled. 133 Although the salt is still soluble in the solvent medium (often, but not limited to, water), the liquid condensed phase of the salt ase, leading to a biphasic system. The parallels between ionic liquid phase separation and the calcium carbonate LCP phase discovered in this project may be more than superficial. The LCP of calcium carbonate formed in the presence of Ca 2+ ions is act ually bicarbonate rich. This is puzzling because the eventual product is a form of solid calcium carbonate, not calcium
124 bicarbonate. As far as this author is aware, solid calcium bicarbonate, Ca(HCO 3 ) 2 has never been isolated, even though there is not c lear reason why it would prohibited from forming. Perhaps calcium bicarbonate does form, but it does not exist in a solid phase at biological conditions. Perhaps calcium bicarbonate salt is a low temperature ionic salt melt, which phase separates from th e mother solution into a solvated liquid phase. In conjunction with the alternate interpretation of the NMR data presented in the Conclusions/Outlook section above, the ionic liquid phase separation principles may be exactly what leads to the LCP of calciu m carbonate. A hypothetical Ca(HCO 3 ) 2 salt would have many of the properties that other low temperature ionic melts possess: (1) a branched structure that is complex enough to inhibit solid formation (2) a small molecular weight to be subject to entropic thermal fluctuations (3) and a large degree of conformational freedom of the branches due to rotations and (de)protonization. The alternate interpretation of the NMR data suggests the presence of rod like structures of a molecular size (the proposed pow der pattern is only 0.05 ppm wide as compared to anisotropy. Calcium bicarbonate salt is of this size range which could yield this the observed anisotropy and has a la rge degree of conformational freedom which is necessary for a low temperature salt melt as demonstrated in Figure 5 3. This hypothesis is profound because researchers over the last 100 years have experienced great difficulty in reproducing calcium carbona te biomineral structures. Perhaps this was due to a flaw in the fundamental assumption that a precursor entirely comprised of Ca 2+ and CO 3 2 is responsible for coalesced, colloidal calcium carbonate biomineral formation. Perhaps the precursor responsible is a nucleated, liquid calcium bicarbonate
125 phase which has been difficult to detect due to its expected early stage rapid conversion to solid calcium carbonate and the need to stabilize large droplets of it with polymer additive, as in the PILP process. T he occurrence of a non classical nucleation route via liquid precursor formation seems to be characteristic of many carbonate minerals and biominerals. Our observations suggest that, although no additives are needed to induce calcium carbonate LCP, the pr esence of negatively charged polyelectrolytes to initiate the PILP process appears to stabilize the droplets to aid in their accumulation, coalescence and solidification into mesocrystal structures. Evidence that the PILP phase undergoes this type of soli dification can be seen in Figure 4 9 C D This figure shows that mineral dissolution reveals a colloidal type of structure as a result of the coating by the PILP process. Further developing and refining our level of understanding of calcium carbonate LC P and PILP are vital if this process is to be extensively used in engineering applications. Perhaps the information obtained about calcium carbonate LCP and PILP can be used to stabilize the PILP droplets for better coating of materials, better infiltrati on of scaffoldings, or long term stabilization for mineral based cements. This information may be translated to biomedical applications involving other biomineral systems that also appear to utilize a PILP phase, including calcium phosphate in bone format ion and, possibly, the calcium oxalate/phosphate system involved in kidney stone formation.
126 Figure 5 1. Chemical shift anisotropy can affect the width and shape of the NMR spectrum. The dipole is assigned Cartesian coordinates x, y, and z with dif ferent amounts of chemical shielding. For the sake of this example, the relative strength of shielding is defined as x
127 Figure 5 2. NMR experiments at various spin rates and varying (the angle between the sample and the bulk magneti c field, B 0 ) other than the magic angle of 54.7 could lead to valuable information regarding the tumbling rate and complex structure of the carbonate/bicarbonate solutes that form during LCP emergence. Varying would be a valuable means for collecting i nformation regarding powdered solid granules, but might not be valuable in the LCP/PILP situation considering the rapid rotation they appear to experience in two of the three dimensions
128 Figure 5 3. A calcium bicarbonate complex would have a large amo unt of rotational and protonization freedom making it entropically unfavorable to solidify. This makes it a candidate for a low temperature of melting. As the structure is drawn, the chemical s hift anisotropy is minimal in the x and z directions due to r apid rotation around those axes. Chemical shift anisotropy is expected to be much larger in the y direction due to the additional electron shielding expected from the double bonded oxygen. This NMR spectrum of this rod like structure would be expected to reflect this anisotropy into a minor powder pattern similar to what was reported in C hapter 2
129 A PPENDIX SUPPLEMENTARY MATERI AL Figure A 1. NMR pulse sequence used to acquire the 13 C diffusion data. It is a modified version of stegs1s, which is Bruke equidistant varying z gradient (Gz) strengths (2% 95%) to generate the data. The following are the relevant variables used for our acquisition : Gz (100%) = 50 g/cm Recovery time (Tr) = 25 s ) = 1.6 s Z spoil = 17.13% Z spoil gradient time (d31) = 1.1 ms Acquisition time (AQ) = 4.9 s M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
130 Figure A 2. A comparison between the same titrations conducted with injection nozzle (micropipette tip) in the solution (as conducted in this paper) vs. with the nozzle just above the solution. We still see the same qualitative character to both curves. The nozzle out method (blue, diamonds) kinetically binds more Ca 2+ under the less ideal mixing conditions than the nozzle in method (red, squares). Even with the presence of some kinetic binding effects, the qualitative character of the curve is maintained, including a close agreement of bound Ca 2+ at which the phase transition occurs, and at which solid nucleation occurs. Since the nozzle is out of the solution and the qualitative character of the profile is maintained, the behavior of the profile is not due to nucleation at the nozzle tip and therefore can be discussed in a qualitative fashion. M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
131 Figure A 3 An example of our modeling technique. Raw data of an attenuat ing signal was collected (upper left). Gaussian distributions were used to model the bulging asymmetry observed in the NMR spectra due to the fact that there is exchange between the carbonate and bicarbonate species in solution (bottom left). The overlap ping Gaussian model was then subtracted from the raw data to yield a residual (right). The residual suggests that some of the signal at the raw data peaks having some Lorenzian ch aracter that a Gaussian would clip at the edges. The edge data is not as important as the overlap and exchange between the two phases present and therefore was not modeled. Once overlapping Gaussian distributions are modeled representing each of the phas es, they can be used to solve for the T 2 relaxation of each phase. A similar technique was used to deconvolute the Pulse Field Gradient Stimulated Echo (PFG STE) data and determine its attenuation for a diffusion measurement of the two phases. M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
132 Figure A 4 induced attenuation of the 13 C peak during the Carr Purcell Meiboom Gill (CPMG) T 2 relaxation experiment of the bicarbonate peak of the 20 mM, pH 8.5, bicarbonate control solution titrated with injections of nanopure water. Using the attenuation of the intensity of this one phase sequence, the T 2 relaxation of the 13 C (carbonates/bicarbonates) in this phase can be determined by using Equation 2 1 M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
133 Figure A 5 The real data and the modeling of the 13 C diffusion during the 17 th injection without polyaspartic acid additive. Only slices 3, 6, 9, 12, and 15 are shown for clarity. Left) The raw data of the 3 rd slice of the widened 13 C NMR peak after the first order phase transition modeled with t hree Gaussian peaks. A single large peak (center model) with similar width as the buffer and 7 th injection system is not sufficient to model the morphology of the spectral peak. Additional peaks ( up field down field ) were modeled to account for the morp hology. Upper right) The real attenuation data for the system during a diffusion NMR experiment. Bottom right) the sum of the Gaussian modeling of the attenuation using the 3 Gaussian peaks. Due to the asymmetric attenuation, and the inability to mode l the system with a singular Gaussian distribution, there is justification for use of the additional Gaussian peaks and additional information about the properties of the up field widening (bicarbonate direction) can be obtained. In the case of the submit ted publication, we found that the up field widening diffused slower than the other two Gaussian models, suggesting the presence of a distinct phase. M. A. Bewernitz, D. Gebau er, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
134 Figure A 6 induced attenuation of the 13 C peak during the Carr Purcell Meiboom Gill (CPMG) T 2 relaxation experimen t of the bicarbonate peak of the 20 g/ mL M.W. 14,000 g/mol Pasp control solution titrated with 30.15 mM bicarbonate buffer, pH 8.5. Using the attenuation of the intensity of this one nce, the T 2 relaxation can of the 13 C (carbonates/bicarbonates) in this phase can be determined by using Equation 2 1 (shown in the main paper) M. A. Bewernitz, D. Gebauer, J Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
135 Figure A 7 The gradient induced attenuation during the Pulse Field Gradient Stimulated Echo (PFG STE) T 2 relaxation experiment of the bicarbonate peak of the 20 g/ mL polyaspartic acid sodium salt (M.W. 14,000 g/mol) control solution titrated with 300 mM bicarbonate buffer, pH 8.5, yielding a final concentration of 18 g/ mL pasp and 30.15 mM bicarbonate. Using the attenuation of the intensity of this one phase system and the known increases in gradient strength for each experimental slice, the self diffusion of the 13 C (and therefore the carbonates/bicarbonates) in this phase can be determined by using Equation 2 2 (shown in the main paper). M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
136 Figu re A 8 induced attenuation of the 13 C peak during the Carr Purcell Meiboom Gill (CPMG) T 2 relaxation experiment of the bicarbonate peak of the 10 mM CaCl 2 mL polyaspartic acid sodium salt (M.W. 14000 g/mol) solution titrated with 17 injecti ons of 300 mM bicarbonate buffer, pH 8.5. The final concentration of the resulting solution is 9 mM CaCl 2 18 g/ mL pasp, 30.15 mM bicarbonate buffer, pH 8.5. An additional phase emerges (PILP phase). Using the observed attenuation of the intensities of both phases due 2 relaxation of the 13 C (carbonates/bicarbonates) in both distinct phases can be determined by using Equation 2 1 (shown in the main paper). M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
137 Figure A 9 The gradient induced attenuation duri ng the Pulse Field Gradient Stimulated Echo (PFG STE) diffusion measurement experiment of the bicarbonate peak of the 20 g/ mL polyaspartic acid sodium salt (14,000 g/mol M.W.) solution titrated with 300 mM bicarbonate buffer, pH 8.5, yielding a final conc entration of 18 g/ mL pasp and 30.15 mM bicarbonate buffer. Using the attenuation of the intensity of this two phase system and the known increases in gradient strength for each attenuated experimental slice, the self diffusion of the 13 C (and therefore t he carbonates/bicarbonates) in these phases can be determined by using Equation 2 2 (shown in the main paper). The carbonates/bicarbonates in the suspected PILP phase are diffusing at less than half the rate of the carbonates/bicarbonates in the bulk solu tion. M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. Reproduced by permission of the Royal Academy of Chemistry
138 Object A 1. First emergent phase experiment, 6mM CaCl 2 into 20 mM bicarbonate buffer, 10 th injection ( 1.29 mb) Object A 2. Second emergent phase experiment, 6mM CaCl 2 into 20 mM bicarbonate buffer, 10 th injection (4.23 mb) Object A 3. 10 mM CaCl 2 into 20 mM bicarbon ate buffer with 20 mM Cl initial concentration, 17 th injection (11.31 mb)
139 LIST OF REFERENCES 1. H. A. Lowenstam and S. Weiner, On biomineralization 1989. 2. S. Mann, J. Webb and R. J. P. Williams, Biomineralization: chemical and bio chemical processes 1989. 3. L. B. Gower, Chemical Reviews 2008, 108 4551 4627. 4. F. C. Meldrum and H. Coelfen, Chemical Reviews 2008, 108 4332 4432. 5. S. W. Lee, S. B. Park, S. K. Jeong, K. S. Lim, S. H. Lee and M. C. Trachtenberg, Micron 2010, 41 273 282. 6. Y. Zhao, Y. Lu, Y. Hu, J. P. Li, L. Dong, L. N. Lin and S. H. Yu, Small 2010, 6 2436 2442. 7. S. Biradar, P. Ravichandran, R. Gopikrishnan, V. Goornavar, J. C. Hall, V. Ramesh, S. Baluchamy, R. B. Jeffers and G. T. Ramesh, Journal of Nanosci ence and Nanotechnology 2011, 11 6868 6874. 8. J. Seto, Y. Ma, S. A. Davis, F. Meldrum, A. Gourrier, Y. Y. Kim, U. Schilde, M. Sztucki, M. Burghammer, S. Maltsev, C. Jger and H. Clfen, Proceedings of the National Academy of Sciences 2012, 109 3699 37 04. 9. C. M. Zaremba, A. M. Belcher, M. Fritz, Y. L. Li, S. Mann, P. K. Hansma, D. E. Morse, J. S. Speck and G. D. Stucky, Chemistry of Materials 1996, 8 679 690. 10. I. Sethmann and G. Woerheide, Micron 2008, 39 209 228. 11. D. Ren, Z. Li, Y. Gao and Q. Feng, Biomedical Materials 2010, 5 12. H. Mutvei, Biomineralization Res Rep 1972, 4 80 86,illust. 13. C. Gregoire, Journal of Biophysical and Biochemical Cytology 1961, 9 395 &. 14. B. A. Gotliv, N. Kessler, J. L. Sumerel, D. E. Morse, N. Tuross, L. Addadi and S. Weiner, Chembiochem 2005, 6 304 314. 15. E. DiMasi and M. Sarikaya, Journal of Materials Research 2004, 19 1471 1476. 16. F. Nudelman, H. H. Chen, H. A. Goldberg, S. Weiner and L. Addadi, Faraday Discussions 2007, 136 9 25. 17. H. Mu tvei, The nacreous layer in molluscan shells 1980. 18. S. W. Wise, Science 1970, 167 1486 &.
140 19. M. A. Meyers, P. Y. Chen, A. Y. M. Lin and Y. Seki, Progress in Materials Science 2008, 53 1 206. 20. F. Nudelman, B. A. Gotliv, L. Addadi and S. Weiner, Journal of Structural Biology 2006, 153 176 187. 21. K. Okazaki, R. M. Dillaman and K. M. Wilbur, Biological Bulletin 1981, 161 402 415. 22. G. Donnay and D. L. Pawson, Science 1969, 166 1147 &. 23. H. U. Nissen, Science 1969, 166 1150 &. 24. D. F. Blake and D. R. Peacor, Scanning Electron Microscopy 1981, 321 &. 25. G. B. Cai, S. F. Chen, L. Liu, J. Jiang, H. B. Yao, A. W. Xu and S. H. Yu, Crystengcomm 2010, 12 234 241. 26. G. T. Zhou, Y. B. Guan, Q. Z. Yao and S. Q. Fu, Chemical Geology 2010, 279 63 72. 27. W. W. Schmahl, E. Griesshaber, K. Kelm, A. Ball, A. Goetz, D. Xu, L. Kreitmeier and G. Jordan, Zeitschrift Fur Kristallographie 2012, 227 604 611. 28. J. Seto, Y. Ma, S. A. Davis, F. Meldrum, A. Gourrier, Y. Y. Kim, U. Schilde, M. Sztucki M. Burghammer, S. Maltsev, C. Jaeger and H. Coelfen, Proc. Natl. Acad. Sci. U. S. A. 2012, 109 3699 3704. 29. M. Niederberger and H. Coelfen, Physical Chemistry Chemical Physics 2006, 8 3271 3287. 30. I. Sethmann, R. Hinrichs, G. Worheide and A. Putn is, Journal of Inorganic Biochemistry 2006, 100 88 96. 31. I. Sethmann, U. Helbig and G. Woerheide, Crystengcomm 2007, 9 1262 1268. 32. E. Beniash, L. Addadi and S. Weiner, Journal of Structural Biology 1999, 125 50 62. 33. N. Gehrke, N. Nassif, N. P inna, M. Antonietti, H. S. Gupta and H. Colfen, Chemistry of Materials 2005, 17 6514 6516. 34. F. C. Meldrum, International Materials Reviews 2003, 48 187 224. 35. J. Rieger, T. Frechen, G. Cox, W. Heckmann, C. Schmidt and J. Thieme, Faraday Discussion s 2007, 136 265 277. 36. J. Aizenberg, Bell Labs Tech. J. 2005, 10 129 141.
141 37. P. Fratzl, F. D. Fischer, J. Svoboda and J. Aizenberg, Acta Biomater. 2010, 6 1001 1005. 38. M. Faatz, F. Grohn and G. Wegner, Advanced Materials 2004, 16 996 +. 39. Y. Ma, S. Weiner and L. Addadi, Advanced Functional Materials 2007, 17 2693 2700. 40. Y. Politi, T. Arad, E. Klein, S. Weiner and L. Addadi, Science 2004, 306 1161 1164. 41. I. M. Weiss, N. Tuross, L. Addadi and S. Weiner, Journal of Experimental Zoology 2002, 293 478 491. 42. E. Beniash, J. Aizenberg, L. Addadi and S. Weiner, Proceedings of the Royal Society of London Series B Biological Sciences 1997, 264 461 465. 43. N. Nassif, N. Pinna, N. Gehrke, M. Antonietti, C. Jager and H. Colfen, Proc. Natl. Acad. Sci. U. S. A. 2005, 102 12653 12655. 44. W. Ostwald, Z. Phys. Chem 1897, 22 289 330. 45. R. Becker and W. Doring, Annalen Der Physik 1935, 24 719 752. 46. D. Gebauer, A. Volkel and H. Colfen, Science 2008, 322 1819 1822. 47. E. M. Pouget, P. H. H. Bomans, J. Goos, P. M. Frederik, G. de With and N. Sommerdijk, Science 2009, 323 1455 1458. 48. R. Demichelis, P. Raiteri, J. D. Gale, D. Quigley and D. Gebauer, Nature Communications 2011, 2 49. J. D. Gale, P. Raiteri and A. C. T. van Duin, Phy sical Chemistry Chemical Physics 2011, 13 16666 16679. 50. D. Gebauer and H. Coelfen, Nano Today 2011, 6 564 584. 51. D. Gebauer, P. N. Gunawidjaja, J. Y. P. Ko, Z. Bacsik, B. Aziz, L. Liu, Y. Hu, L. Bergstrm, C. W. Tai, T. K. Sham, M. Edn and N. Hed in, Angewandte Chemie International Edition 2010, 49 8889 8891. 52. L. B. Gower and D. J. Odom, Journal of Crystal Growth 2000, 210 719 734. 53. B. Guillemet, M. Faatz, F. Grohn, G. Wegner and Y. Gnanou, Langmuir 2006, 22 1875 1879. 54. S. E. Wolf, L Mueller, R. Barrea, C. J. Kampf, J. Leiterer, U. Panne, T. Hoffmann, F. Emmerling and W. Tremel, Nanoscale 2011, 3 1158 1165.
142 55. S. J. Homeijer, R. A. Barrett and L. B. Gower, Crystal Growth & Design 2010, 10 1040 1052. 56. S. J. Homeijer, M. J. Ols zta, R. A. Barrett and L. B. Gower, Journal of Crystal Growth 2008, 310 2938 2945. 57. F. F. Amos, L. Dai, R. Kumar, S. R. Khan and L. B. Gower, Urological Research 2009, 37 11 17. 58. A. Dey, P. H. H. Bomans, F. A. Mueller, J. Will, P. M. Frederik, G. de With and N. A. J. M. Sommerdijk, Nature Materials 2010, 9 1010 1014. 59. S. S. Jee, R. K. Kasinath, E. DiMasi, Y. Y. Kim and L. Gower, Crystengcomm 2011, 13 2077 2083. 60. S. S. Jee, T. T. Thula and L. B. Gower, Acta Biomater. 2010, 6 3676 3686. 61. T. T. Thula, D. E. Rodriguez, M. H. Lee, L. Pendi, J. Podschun and L. B. Gower, Acta Biomater. 2011, 7 3158 3169. 62. S. S. Jee, L. Culver, Y. Li, E. P. Douglas and L. B. Gower, Journal of Crystal Growth 2010, 312 1249 1256. 63. P. G. Vekilov, Crys tal Growth & Design 2010, 10 5007 5019. 64. O. Galkin, W. Pan, L. Filobelo, R. E. Hirsch, R. L. Nagel and P. G. Vekilov, Biophysical journal 2007, 93 902 913. 65. Y. Jiang, H. Gong, D. Volkmer, L. Gower and H. Coelfen, Advanced Materials 2011, 23 354 8 +. 66. Y. Jiang, L. Gower, D. Volkmer and H. Colfen, Physical Chemistry Chemical Physics 2012, 14 914 919. 67. Crystal Growth & Design 2011, 11 3243 3249. 68. S. Wohlrab, H. Colfen and M. Antonietti, Ang ewandte Chemie International Edition 2005, 44 4087 4092. 69. Y. Ma, G. Mehltretter, C. Plueg, N. Rademacher, M. U. Schmidt and H. Coelfen, Advanced Functional Materials 2009, 19 2095 2101. 70. D. A. W. Thompson, On growth and form / by D'Arcy Wentworth Thompson University Press ;, Cambridge :, 1945. 71. A. V. Radha, T. Z. Forbes, C. E. Killian, P. U. P. A. Gilbert and A. Navrotsky, Proceedings of the National Academy of Sciences 2010.
143 72. D. Gebauer, A. Verch, H. G. Boerner and H. Coelfen, Crystal Gro wth & Design 2009, 9 2398 2403. 73. P. Schuck, Biophysical Journal 2000, 78 1606 1619. 74. P. Schuck, Biophysical Journal 1998, 75 1503 1512. 75. K. Strenge and A. Seifert, Progress in Colloid and Polymer Science 1991, 86 76 83. 76. H. Nebel, M. Ne umann, C. Mayer and M. Epple, Inorganic Chemistry 2008, 47 7874 7879. 77. F. M. Michel, J. MacDonald, J. Feng, B. L. Phillips, L. Ehm, C. Tarabrella, J. B. Parise and R. J. Reeder, Chemistry of Materials 2008, 20 4720 4728. 78. R. E. Zeebe, Geochimica Et Cosmochimica Acta 2011, 75 2483 2498. 79. K. Kigoshi and T. Hashitani, Bulletin of the Chemical Society of Japan 1963, 36 1372 1372. 80. A. Verch, D. Gebauer, M. Antonietti and H. Colfen, Physical Chemistry Chemical Physics 2011, 13 16811 16820. 8 1. R. Demichelis, P. Raiteri, J. D. Gale, D. Quigley and D. Gebauer, Nature communications 2011, 2 590. 82. P. G. Vekilov, Crystal Growth & Design 2004, 4 671 685. 83. L. Addadi, S. Raz and S. Weiner, Advanced Materials 2003, 15 959 970. 84. L. A. Go wer and D. A. Tirrell, Journal of Crystal Growth 1998, 191 153 160. 85. Y. Li, T. T. Thula, S. Jee, S. L. Perkins, C. Aparicio, E. P. Douglas and L. B. Gower, Biomacromolecules 2012, 13 49 59. 86. M. J. Olszta, X. Cheng, S. S. Jee, R. Kumar, Y. Y. Kim, M. J. Kaufman, E. P. Douglas and L. B. Gower, Materials Science & Engineering R Reports 2007, 58 77 116. 87. M. A. Bewernitz, D. Gebauer, J. Long, H. Colfen and L. B. Gower, Faraday Discussions 2012. 88. W. Ostwald, Z. Chem. Phys. 1897, 22 289. 89. H Weingartner, Pure and Applied Chemistry 2001, 73 1733 1748. 90. Y. Jiang, H. Nadolny, S. Kaeshammer, S. Weibels, W. Schroeer and H. Weingaertner, Faraday Discussions 2012, 154 391 407.
144 91. H. Weingaertner, C. Cabrele and C. Herrmann, Physical Chemist ry Chemical Physics 2012, 14 415 426. 92. W. Schroeer, Contributions to Plasma Physics 2012, 52 78 88. 93. M. Kleemeier, S. Wiegand, W. Schroer and H. Weingartner, Journal of Chemical Physics 1999, 110 3085 3099. 94. H. El Shall, J. H. Jeon, E. A. Ab del Aal, S. Khan, L. Gower and Y. Rabinovich, Crystal Research and Technology 2004, 39 577 585. 95. H. El Shall, J. H. Jeon, E. A. Abdel Aal, S. Khan, L. Gower and Y. Rabinovich, Crystal Research and Technology 2004, 39 214 221. 96. J. Verwey E. and T. Overbeek J., Theory of the Stability of Lyophobic Colloids. The Interaction of Sol Particles Having a Electrical Double Layer. Elsevier Pub. Comp., Amsterdam, New York, 1948. 97. C. X. Luo, H. Z. Xie, J. K. Liu and G. M. Li, Journal of Composite Materi als 2012, 46 91 97. 98. H. Colfen and M. Antonietti, Langmuir 1998, 14 582 589. 99. M. F. Butler, W. J. Frith, C. Rawlins, A. C. Weaver and M. Heppenstall Butler, Crystal Growth & Design 2009, 9 534 545. 100. D. Rautaray, K. Sinha, S. S. Shankar, S. D. Adyanthaya and M. Sastry, Chemistry of Materials 2004, 16 1356 1361. 101. Q. Shen, H. Wei, L. C. Wang, Y. Zhou, Y. Zhao, Z. Q. Zhang, D. J. Wang, G. Y. Xu and D. F. Xu, Journal of Physical Chemistry B 2005, 109 18342 18347. 102. B. P. Bastakoti, S. Guragain, Y. Yokoyama, S. i. Yusa and K. Nakashima, Langmuir 2011, 27 379 384. 103. Y. Han, M. Fuji, D. Shehukin, H. Moehwald and M. Takahashi, Crystal Growth & Design 2009, 9 3771 3775. 104. M. Yang, X. Q. Jin and Q. A. Huang, Colloid Surf. A Physicoc hem. Eng. Asp. 2011, 374 102 107. 105. N. Loges, K. Graf, L. Nasdala and W. Tremel, Langmuir 2006, 22 3073 3080. 106. J. A. Thomas, L. Seton, R. J. Davey and C. E. DeWolf, Chemical Communications 2002, 1072 1073. 107. M. Fujiwara, K. Shiokawa, K. Mori gaki, Y. Zhu and Y. Nakahara, Chemical Engineering Journal 2008, 137 14 22.
145 108. X. Wang, W. Zhou, J. Cao, W. Liu and S. Zhu, J. Colloid Interface Sci. 2012, 372 24 31. 109. S. P. Patel VM, Kurz A, Ossenbeck M, Shah DO, Gower LB, in Concentrated Disper sions: Theory, Experiments, and Applications, 2002 2004 ed. S. P. Markovic B, American Chemical Society, Washington, DC, 2002, pp. 15 25. 110. J. Aizenberg, A. J. Black, G. M. Whitesides and M. Whitesides, Abstracts of Papers of the American Chemical Soci ety 1999, 218 U604 U604. 111. Y. J. Han and J. Aizenberg, Abstracts of Papers of the American Chemical Society 2003, 226 U766 U766. 112. J. Aizenberg, J. Hanson, T. F. Koetzle, S. Weiner and L. Addadi, J. Am. Chem. Soc. 1997, 119 881 886. 113. A. Ber man, J. Hanson, L. Leiserowitz, T. F. Koetzle, S. Weiner and L. Addadi, Journal of Physical Chemistry 1993, 97 5162 5170. 114. L. J. Dai, E. P. Douglas and L. B. Gower, J. Non Cryst. Solids 2008, 354 1845 1854. 115. X. Cheng, P. L. Varona, M. J. Olszta and L. B. Gower, Journal of Crystal Growth 2007, 307 395 404. 116. Y. Y. Kim, E. P. Douglas and L. B. Gower, Langmuir 2007, 23 4862 4870. 117. F. F. Amos, D. M. Sharbaugh, D. R. Talham, L. B. Gower, M. Fricke and D. Volkmer, Langmuir 2007, 23 1988 1 994. 118. D. Janofske, in Biomineralization 93: 7th International Symposium on Biomineralization eds. D. Allemand and J. P. Cuif, Muse ocanographique, Monaco, 1993, vol. 17 20, pp. 295 303. 119. D. Deamer and A. D. Bangham, Biochimica et Biophysica Acta (BBA) Biomembranes 1976, 443 629 634. 120. S. R. Dickinson and K. M. McGrath, Analyst 2001, 126 1118 1121. 121. E. DiMasi, V. M. Patel, M. Sivakumar, M. J. Olszta, Y. P. Yang and L. B. Gower, Langmuir 2002, 18 8902 8909. 122. S. Mishra, A. Chatter jee and R. Singh, Polymers for Advanced Technologies 2011, 22 2571 2582. 123. N. Wang, Q. She, H. Xu, Y. Yao, L. Zhang, X. Qu and L. Zhang, Journal of Applied Polymer Science 2010, 115 1336 1346.
146 124. S. Hikasa, K. Nagata, K. Miyahara, T. Izumi, T. Suda, A. Toyohara, A. Kato and Y. Nakamura, Journal of Applied Polymer Science 2009, 114 919 927. 125. J. A. Thomson, P. Schurtenberger, G. M. Thurston and G. B. Benedek, Proc. Natl. Acad. Sci. U. S. A 1987, 84 7079 7083. 126. T. Tanaka, C. Ishimoto and L. T. Chylack, Science 1977, 197 1010 1012. 127. P. G. Vekilov, Journal of Physics Condensed Matter 2012, 24 128. A. C. Dumetz, A. M. Chockla, E. W. Kaler and A. M. Lenhoff, Biophysical journal 2 008, 94 570 583. 129. A. Onuki, R. Okamoto and T. Araki, Bulletin of the Chemical Society of Japan 2011, 84 569 587. 130. S. Samin and Y. Tsori, Epl 2011, 95 131. W. Schroeer, A. Wagner and O. Stanga, Journal of Molecular Liquids 2006, 127 2 9. 132. J. F. Lutsko, Journal of Chemical Physics 2012, 136 133. D. Corradini, P. Gallo and M. Rovere, Journal of Physics Condensed Matter 2010, 22
147 BIOGRAPHICAL SKETCH Mark was born and raised in southeastern Michigan. He attended Michigan State Universi ty from fall of 1995 to spring of 2000 where he graduated with a Bachelor of Science in chemical engineering and biochemistry/molecular biology. He attended the University of Florida from fall of 2001 to fall of 2012. During that time he obtained a Maste r of Science in b iomedical e ngineering and a Doctorate in b iomedical e ngineering.