The Pennsylvania State University. The Graduate School. Intercollege Degree Graduate Program in Materials Science and Engineering

Transcription

1 The Pennsylvania State University The Graduate School Intercollege Degree Graduate Program in Materials Science and Engineering FORMATION OF HYDROXYAPATITE IN VARIOUS AQUEOUS SOLUTIONS A Thesis in Materials Science and Engineering by Jacqueline Lee Sturgeon 2007 Jacqueline Lee Sturgeon Submitted in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy December 2007

2 The thesis of Jacqueline Lee Sturgeon was reviewed and approved* by the following: Paul W. Brown Professor of Ceramic Science and Engineering Thesis Advisor Chair of Committee Della Roy Professor Emerita of Materials Science Erwin Vogler Professor of Materials Science and Engineering Neil Sharkey Professor of Kinesiology, Orthopaedics & Rehabilitation James Runt Professor of Polymer Science Chair, Intercollege Graduate Degree Program in Materials Science and Engineering *Signatures are on file in the Graduate School

3 iii ABSTRACT Hydroxyapatite (HAp), Ca 10 (PO 4 ) 6 (OH) 2, is important in the field of biomaterials as it is the mineral component of bones and teeth. Biological apatites do not maintain an exact composition and are usually calcium-deficient, represented as Ca (10- x)(hpo 4 ) x (PO 4 ) (6-x) (OH) (2-x), where x ranges from 0 to 1, with various ion substitutions. Formation of calcium-deficient hydroxyapatites (CDHAp) from solid calcium phosphate precursor materials was performed at physiologic temperature (37 o C) in a variety of aqueous solutions. Two cement systems were utilized in these experiments: tetralcium phosphate (TetCP) with dicalcium phosphate anhydrous (DCPA) and β- tricalcium phosphate (β-tcp). The kinetics, solution chemistry, phase evolution, and microstructure of the developed apatites were analyzed as appropriate. Reaction of β-tcp in ammonium fluoride solutions formed HAp substituted with fluoride and calculated to be deficient in calcium. A new ratio of TetCP to DCPA was used with solutions of sodium bicarbonate to form a calcium-deficient carbonate hydroxyapatite. The capacity for sodium dihydrogen phosphate to buffer ph increases and enhance reaction kinetics in this system was also explored. Formation of a highly crystalline CDHAp was achieved by hydrolyzing β-tcp in water for extended time periods. Lattice parameters were among the features characterized for this apatite. The hydrolysis of β-tcp in phosphate buffered saline (PBS) and simulated body fluids (SBF) was also investigated; use of SBF was found to completely inhibit formation of HAp in this system while reaction in PBS was slow in comparison to water.

4 iv The effects of filler materials on the mechanical properties of a calcium phosphate cement were examined using the TetCP/DCPA system. Dense aggregates were not found to decrease compressive strength in comparison to the cement alone. The use of aggregates was found to improve the compressive strength of cement formed using NaHCO 3 solution as a setting liquid. The influence of phosphonates on formation of HAp from a TetCP/DCPA system was analyzed. Variations in concentration and phosphonate type were observed to influence the dissolution of precursors and HAp formation in this system. Bisphosphonates were more effective at inhibiting dissolution or growth of HAp in solutions with high liquid to solids ratios. Monophosphonate was found to slightly accelerate the formation of HAp at low concentration.

5 v TABLE OF CONTENTS LIST OF FIGURES...viii LIST OF TABLES...xiii PREFACE...xiv ACKNOWLEDGEMENTS...xv Chapter 1 Introduction...1 Chapter 2 Formation of HAp from β-tcp at Physiologic Temperature Abstract Introduction Materials and Methods Results and Discussion Phase Evolution Microstructure Solution Chemistry HAp Characterization Conclusions...25 Chapter 3 Formation of fluoridated apatite from β-tricalcium phosphate at physiologic temperature Abstract Introduction Materials and Methods Results Discussion Conclusions...54 Chapter 4 Effects of Carbonate on HAp Formed from CaHPO 4 and Ca 4 (PO 4 ) 2 O Abstract Introduction Materials and Methods Precursor Synthesis Solution Chemistry Kinetics Characterization Results and Discussion...62

6 4.4.1 Solution Chemistry Kinetics Characterization Conclusions...78 Chapter 5 Effect of Dense Aggregates on Mechanical Properties of Calcium Phosphate Cements Abstract Introduction Materials and Methods Results Mechanical Properties Variation of Aggregate Quantity Variation of Aggregate Type Effect Aggregates and NaHCO 3 solution Aggregate Reactivity Microstructural Analysis Discussion Conclusions...97 Chapter 6 Effects of Phosphonates on Formation of Hydroxyapatite from Solid Precursors Abstract Introduction Materials and Methods Synthesis of calcium phosphate precursors Phosphonates Kinetics Solution Chemistry Characterization Results Kinetics Bisphosphonates Phosphonates Monophosphonate Solution Chemistry - ph Bisphosphonates Phosphonates Monophosphonate Solution Chemistry Ion Concentrations Bisphosphonates Phosphonates vi

7 Monophosphonate Characterization Bisphosphonates Phosphonates and Monophosphonate Discussion Bisphosphonates Phosphonates Monophosphonate Conclusions Chapter 7 Formation and properties of composites comprised of calciumdeficient hydroxyapatites and ethyl alanate polyphosphazenes Abstract Introduction Materials and Methods Precursor Synthesis Solution Chemistry Characterization Mechanical Testing Results and Discussion Conclusions General Conclusions Bibliography vii

8 viii LIST OF FIGURES Figure 2.1: XRD showing development of HAp phase in β-tcp hydrolyzed in water for up to 32 weeks Figure 2.2: XRD patterns of β-tcp reacted in PBS showing development of HAp...12 Figure 2.3: XRD patterns of β-tcp in solution with SBF-K for up to 32 weeks...13 Figure 2.4: XRD patterns of β-tcp in solution with SBF-T for 1 day and 32 weeks. No HAp phase is observed at any point in the interim...14 Figure 2.5: Microstructure of HAp formed from β-tcp in water after reaction for 32 weeks...15 Figure 2.6: Microstructure of HAp developed after reaction of β-tcp in PBS solution for 32 weeks Figure 2.7: Typical particle observed after reaction in an SBF solution for 32 weeks...17 Figure 2.8: ph values measured after reaction in solution at each time interval analyzed...18 Figure 2.9: Variation in ion concentrations over time for hydrolysis in water and PBS solution...19 Figure 2.10: FT-IR spectra of samples after 32 weeks of reaction in water, PBS, and SBF solutions and sample of unreacted β-tcp Figure 2.11: XRD patterns of HAp formed after reaction for 32 weeks in water and PBS solution scanned using a quartz zero background holder and step scan...23 Figure 3.1: ph variation over the first 24 hours of samples reacted in water or ammonium fluoride solutions Figure 3.2: ph of β-tcp in static solutions of NH 4 F over 8 weeks of reaction. Concentration of NH 4 F decreases from 1 to Figure 3.3: Variations in calcium ion concentrations during the hydrolysis of β- TCP for up to 8 weeks in the NH 4 F solutions and in water Figure 3.4: Phosphate ion concentrations for up to 8 weeks during the hydrolysis of β-tcp in the NH 4 F solutions and in water....36

9 Figure 3.5: Variations in fluoride ion concentrations in solution during the hydrolysis of β-tcp in NH 4 F solutions for up to 8 weeks...37 Figure 3.6: Variations in fluoride, calcium, and phosphate ion concentrations measured over the first 24 hours in a 1:1 NH 4 F:TCP solution Figure 3.7: XRD patterns of β-tcp after 1 month in fluoride solutions with TCP:NH 4 F molar ratios as labeled or in water...39 Figure 3.8: FT-IR spectra after heating samples reacted for one month to 500 o C. Enlarged portion shows change in carbonate peaks around 865cm Figure 3.9: FT-IR spectra regions of interest after heating samples reacted for 1 month to 925 o C Figure 3.10: Fluorapatite formed in 1:1 NH 4 F:TCP solution after 1 month Figure 3.11: FAp formed in 2:1 NH 4 F:TCP solution after 1 month Figure 3.12: β-tcp in 6:1 NH 4 F:TCP solution after 1 month...45 Figure 3.13: Needles of HAp grown from β-tcp in water at 37 o C after 5 months of reaction Figure 4.1: Variation in ph over 24 hours when a TetCP/DCP mixture with Ca/P ratio of 1.8 is reacted in NaHCO 3 solutions with NaHCO 3 -to-cahpo 4 ratios of 0.5, 1.0, and 1.5 and in water...63 Figure 4.2: Variation in ph over 24 hours when a TetCP/DCP mixture with Ca/P ratio of 1.8 is reacted in water or NaHCO 3 solutions containing equimolar amounts of NaH 2 PO 4 added...65 Figure 4.3: XRD patterns of solids obtained after 24 hours of reaction in water and various NaHCO 3 or NaHCO 3 /NaH 2 PO 4 solutions...66 Figure 4.4: Variations in Ca 2+, phosphate ion [P i ], CO 3 2-, and Na + over 168 hours (7 days) for reaction in water Figure 4.5: Variations in Ca 2+, phosphate ion [P i ], CO 3 2-, and Na + over 168 hours (7 days) reaction in solution of NaHCO 3 with NaHCO 3 -to-cahpo 4 ratio of Figure 4.6: Variations in Ca 2+, phosphate ion [P i ], CO 3 2-, and Na + over 168 hours (7 days) for reaction in solution of NaHCO 3 with NaHCO 3 -to-cahpo 4 ratio of 1 with an equimolar amount of NaH 2 PO ix

10 Figure 4.7: Rates of heat evolution during the reaction of TetCP and DCP with various NaHCO 3 solutions and water x Figure 4.8: Total heat evolved during the formation of carbonated apatite solutions of NaHCO 3 or water Figure 4.9: Rates of heat evolution (left axis) and total heat evolved (right axis) for NaHCO 3 solutions in NaHCO 3 -to-cahpo 4 ratios of 0.5 and solution with the corresponding equimolar amount of NaH 2 PO Figure 4.10: Rates of heat evolution (left axis) and total heat evolved (right axis) for NaHCO 3 solutions in NaHCO 3 -to-cahpo 4 ratios of 1.0 and solution with the corresponding equimolar amount of NaH 2 PO Figure 5.1: Average compressive strengths of CDHAp cement with dense HAp aggregate compared to a group with no aggregate addition. Significant differences (p < 0.05, n = 4-6) from the aggregate-free group are indicated by *. Error bars represent the standard deviation for each set of samples Figure 5.2: Average compressive strengths of samples with or without aggregates. A statistically significant difference (p < 0.05, n = 5) from the group without aggregate is indicated by *. Error bars represent standard deviation for each set of samples...87 Figure 5.3: Compressive strengths observed when NaHCO 3 solution is used with and without aggregate in comparison to water without aggregate. Statistically significant differences (p < 0.05, n = 5) are denoted with *. Error bars represent standard deviation for each set of samples Figure 5.4: XRD patterns of CDHAp with CaCO 3 aggregate added at 10 and 25% in comparison to CDHAp with no aggregate...90 Figure 5.5: XRD patterns of CDHAp with dense β-tcp aggregate added at 10 and 25 wt% in comparison to CDHAp with no aggregate...91 Figure 5.6: XRD patterns of CDHAp with dense HAp aggregate added at 10 and 25 wt% in comparison to CDHAp with no aggregate...92 Figure 5.7: Interface between CaCO 3 crystals and CDHAp...93 Figure 5.8: Interface between an aggregate of HAp and surrounding CDHAp Figure 5.9: Microstructure of CDHAp formed with water...94 Figure 5.10: Microstructure of CDHAp formed with carbonate solution...95

11 Figure 6.1: Rate of heat evolution during the reaction of 5wt% phosphonate solution with CDHAp precursors. Inset shows first half hour after injection of liquid Figure 6.2: Rate of heat evolution during reaction of 0.3wt% phosphonate solution with CDHAp precursors. Inset shows first hour Figure 6.3: XRD of solids after reaction for 6 days in 5wt% phosphonate solution Figure 6.4: XRD patterns after 24 hours of reaction in 0.3wt% phosphonate solution at 1:1 powder-to-liquid ratio Figure 6.5: ph behavior of 0.1 wt% phosphonate solutions and water (a) and XRD patterns obtained after reaction in a 0.1 wt% phosphonate solution for 24 hours (b) Figure 6.6: ph behavior of 0.5 wt% phosphonate solutions and water (a) and XRD patterns obtained after reaction in a 0.5 wt% phosphonate solution for 24 hours (b) Figure 6.7: ph behavior of 2.5 wt% phosphonate solutions and water (a) and XRD patterns obtained after reaction in a 2.5 wt% phosphonate solution for 24 hours (b) Figure 6.8: ph behavior of 5 wt% phosphonate solutions and water (a) and XRD patterns obtained after reaction in a 5 wt% phosphonate solution for 24 hours (b) Figure 6.9: Concentration of calcium ions (a) and phosphate ions (b) in solution after reaction for up to 10 days in various 5.0 wt% phosphonate solutions. Inset shows variations in water and phosphonate #1 solution over the first 24 hours Figure 6.10: XRD patterns over the first 18 hours of reaction in water Figure 6.11: XRD patterns after reaction in 5 wt% solution of phosphonate #1 over 24 hours Figure 6.12: XRD patterns after reaction in 5 wt% solution of phosphonate #2 over 10 days Figure 6.13: XRD patterns after reaction in 5 wt% solution of phosphonate #3 over 7 days Figure 6.14: XRD patterns after reaction in 5 wt% solution of phosphonate #4 over 10 days xi

12 Figure 6.15: XRD patterns after reaction in 5 wt% solution of phosphonate #5 over 18 hours Figure 6.16: XRD patterns obtained from pellets reacted for 24 hours in a 50wt% phosphonate solution or water Figure 6.17: Microstructure of shell formed in phosphonate #4 solution Figure 6.18: Portion of shell observed on pellet after reaction in phosphonate #2 solution for 24 hours Figure 6.19: Surface of pellet after reaction in a phosphonate solution Figure 6.20: Surface of pellet after reaction in water Figure 7.1: Structure of poly(phosphazene-ethylalanine), PNEA, (a) and cosubstituted polymer with phenyl phenoxy groups, PNEA 50 PhPh 50, (b) the two polymers used in this study Figure 7.2: The variations in ph with time for the CDHAp composite compositions studied and CDHAp without polymer Figure 7.3: SEM image of fracture surface of sample without polymer Figure 7.4: SEM image of composite with PNEA fracture surface Figure 7.5: SEM image of composite with PNEA 50 PhPh 50 fracture surface Figure 7.6: Fiber showing growth of HAp phase after hydrolysis of a CDH-PNEA composite prepared by mechanical mixing Figure 7.7: FT-IR spectra of CDH-PNEA and CDH-PNEA 50 Ph 50 composites compared to CDHAp without polymer Figure 7.8: Average compressive strengths of CDHAp without polymer and composites containing PNEA and PNEA 50 PhPh 50. Error bars represent the standard deviation for each group of samples (n = 3-5) xii

13 xiii LIST OF TABLES Table 2.1: Types of reagents and amounts used for preparation of simulated body fluid....9 Table 2.2: Calculated Ca/P ratios of the HAp formed in water and PBS solution...20 Table 2.3: Lattice dimensions calculated for HAp formed in water and PBS solution after 32 weeks. All values displayed are in units of angstroms Table 3.1: Percentage of FAp phase observed after reaction. BDL = below detection limit Table 3.2: Calculated ratios of ions in a solid formed in 1:1 TCP:NH 4 F solution calculated normally and with correction for unreacted β-tcp Table 3.3: Calculated ratios of ions present in solid formed calculated normally and corrected for unreacted β-tcp...49 Table 4.1: Calculated ratios of ions in the solids formed after 7 days of reaction in solutions with NaHCO 3 -to-cahpo 4 ratios of 1, with and without NaH 2 PO 4 added as buffer Table 6.1: Names, chemical formulas, and structures of the phosphonates used in this study

14 xiv PREFACE A section of this work, Chapter 7, is excerpted from a multiple author manuscript intended for publication. The data presented in the chapter were obtained by this author and the writing of the manuscript was almost entirely a true collaborative effort between the two lead authors, myself and Dr. Yaser Greish.

15 xv ACKNOWLEDGEMENTS I would like to express thanks to several persons for their support throughout my research. First, I would like to acknowledge my advisor, Dr. Paul Brown, for his help and suggestions in researching and writing this thesis. Dr. Yaser Greish deserves many thanks for his initial help in introducing me to many of the methods used in this research, for his constant support, and hours of discussions about polyphosphazenes and phosphonates. I would also like to acknowledge my committee members, Dr. Della Roy, Dr. Neil Sharkey, and Dr. Erwin Vogler. In particular, I would like to thank Dr. Vogler for his words of encouragement over the past several years. Finally, I would like to thank my family and close friends for their moral support throughout my research.

16 Chapter 1 Introduction Native bone tissue can be considered a composite consisting of a mineralized collagen matrix. 1 The mineral component of bone can be approximated by hydroxyapatite (HAp). Similarly, enamel and dentine can be considered a fluoridated form of hydroxyapatite. Calcium phosphates are also found in pathological calcifications in vivo. Thus, understanding the nature of calcium phosphate formation, dissolution, and interaction with the surrounding environment is of importance for potential biomedical materials applications. Hydroxyapatite is a mineral that can exist over a range of compositions and accept many substitutions. Stoichiometric HAp (SHAp) has the chemical formula Ca 10 (PO 4 ) 6 (OH) 2 (Ca/P = 1.67). A more generalized way of writing the chemical formula for non-substituted HAp is Ca (10-x) (HPO 4 ) x (PO 4 ) (6-x) (OH) (2-x) where x = 0 to 1, with x =1 being a calcium-deficient HAp (CDHAp) (Ca/P = 1.5). Biologically relevant substituent ions for calcium include sodium, magnesium, and potassium. The phosphate can be substituted by carbonate; this substitution decreases the crystal size and perfection of the apatite and increases reactivity. 2 Substitutions for the hydroxyl group include fluoride, chloride, and, to a limited degree at low temperature, carbonate. The fluoride substitution increases crystal size and decreases the solubility of

17 2 the apatite formed. 3 Ion substitutions and calcium to phosphorus ratios vary depending on location in the body, age, and other factors. 4 Many calcium phosphate preparations can generate desirable biological responses when used in vivo. Non-resorbable applications include coatings on metallic implants for load-bearing or dental applications 5, which facilitate osseointegration. 6 Dense SHAp can be synthesized at high temperatures and have high compressive strength, however these solids must be shaped before implantation and are minimally resorbable. Porous carbonated HAp and tricalcium phosphate (TCP) has been formed via hydrothermal reaction of phosphate with CaCO 3 from coral 7 and these materials are osseoconductive. 8 Bioglass, a calcium phosphate enriched glass, can bond with surrounding bone tissue via a surface reaction to form carbonated HAp. 9 Resorbable calcium phosphate biomaterials have also been prepared and include biphasic calcium phosphates, a mixture of HAp and tricalcium phosphate (TCP), and calcium phosphate bone cements. Several methods have been used to prepare synthetic HAp for characterization and scientific studies and for use as a biomaterial. Precipitation from salt solutions at various temperatures and ph ranges is a common synthesis technique to produce HAp having a range of compositions. High temperature formation of HAp is also often used, however only SHAp can be produced by this method. Finally, production of HAp via an acid-base reaction or direct hydrolysis has been studied extensively for use in bone cement applications. Calcium phosphate bone cements have been of interest for many years due to the ability to shape and mold immediately before use. This is of vital importance for filling irregularly shaped defects or in trauma surgery. However, the mechanical properties of

18 3 calcium phosphate cements have limited use to non-load bearing applications. Composites with polymers have been made to decrease the brittleness of the ceramic. 10 A calcium phosphate cement system first introduced by Brown and Chow uses tetracalcium phosphate (TetCP) and dicalcium phosphate (anhydrous or dihydrate) (DCPA or DCPD). 11 The general precept behind this system is an acid-base reaction between two calcium phosphate salts that are more acidic or basic and more soluble than HAp. This reaction can occur at low-temperature, making it relevant for use in vivo. This system has been received much attention and the kinetics, solution chemistry, and phase evolution have been studied. A general reaction equation for the system in water is shown in Eq Similarly, formation of a calcium-deficient HAp (CDHAp) can also be achieved by varying the proportions of the reactants, as shown in Eq CaHPO Ca 4 (PO 4 ) 2 O Ca 10 (PO 4 ) 6 (OH) CaHPO Ca 4 (PO 4 ) 2 O 2 Ca 9 HPO 4 (PO 4 ) 5 OH 1.2 Acid-base reactions have several advantages over other techniques, both for use as a biomaterial, as mentioned above, and understanding how reactions take place. One advantage is producing compositions more closely related to native tissue. Formation of HAp by this reaction is easier to control than precipitation from salt solutions; extraneous ions are not available for substitution into the structure and observation of solution chemistry and phase formation as the reaction occurs is possible. TCP has three different polymorphs: α, α, and β. The α polymorph is not stable below approximately 1500 o C, however α-tcp can be formed by rapid quenching of TCP

19 4 powder from temperatures above approximately 1180 o C. Quenching locks in vacancies, making the form metastable at room temperature. The α form is generally far more reactive than β-tcp, forming HAp in less then 24 hours when immersed in water. The reaction of TCP in water is shown in Eq Ca(PO 4 ) 2 + H 2 O Ca 9 HPO 4 (PO 4 ) 5 OH 1.3 The reaction kinetics, solution chemistry and phase development of α-tcp have been studied extensively. This thesis investigates the formation of calcium-deficient hydroxyapatite in aqueous solutions via reaction of solid calcium phosphate precursors at physiologic temperature. Hydrolysis of β-tcp in water, simulated body fluids, and fluoride solutions was performed and the reaction products characterized. The kinetics and solution chemistry during formation of a calcium-deficient carbonated HAp was studied. The effects of dense aggregates and polyphosphazenes on the mechanical properties of calcium-deficient apatites were observed. Finally, formation of CDHAp in solutions of phosphonates was studied. A general formula for the apatites formed is Ca 9 [ ] a (PO 4 ) 6-b [ ] b (OH,F) a+b where [ ] represents a substitution of Na or NH 4 for a and CO 3, HPO 4, or other divalent anions for b, and the amount of F varies based on the concentration in solution and can be coupled with OH. Chapter 2 presents an overview of the hydrolysis of β-tcp in water and simulated body fluids and characterizes the CDHAp formed. Chapter 3 also involves the hydrolysis of β-tcp, however these reaction were carried out in ammonium fluoride solutions. The nature of the fluoridated apatite that is calcium deficient is characterized. Chapter 4

20 5 follows the formation of a calcium-deficient carbonated apatite using kinetic and solution chemistry studies. The use of buffers in this system was also investigated. The significance of the preceding work is as follows: (1) various apatite compositions are more or less resorbable but the range of compositions which can be produced by cementlike reactions remain undetermined, (2) the relationships between composition and kinetics and mechanisms of formation by reactions, which could occur in vivo, have not been determined for substituted apatites, and (3) the effects of buffer agents on the above have not been established. The effects of fillers remains undetermined, thus Chapter 5 examined the use of dense aggregates and other filler materials on the formation of a calcium phosphate cement. While the role of phosphonates on the biology of bone resorption has been established, their effects, and those of related phosphate-bearing compounds, on the mechanism of apatite formation by cement-like reactions have not been determined. The latter is important to determine if these drugs are to be delivered during procedures to augment hard tissues. Chapter 6 investigates the effects of phosphonate compounds on the formation of HAp from a calcium phosphate cement. Finally, if apatites formed by cement-like reactions are to be used in weight bearing applications, it is likely that they will be constituents of composites. Thus, Chapter 7 reports the effects of a specialized class of biocompatible polymers, polyphosphazenes, in composites with apatite cements.

21 Chapter 2 Formation of HAp from β-tcp at Physiologic Temperature 2.1 Abstract Calcium-deficient hydroxyapatite (CDHAp) was synthesized by hydrolysis of β- tricalcium phosphate (β-tcp) at physiologic temperature. Reaction in water, phosphate buffered saline (PBS) and two simulated body fluids (SBF) were followed over extensive time periods using x-ray diffraction and solution chemistry. Very little reaction was observed in the SBFs even after months of reaction. An almost fully calcium deficient HAp (Ca/P = 1.51) with small amounts of carbonate incorporation was formed by reaction in water. Reaction in PBS also formed a CDHAp with some sodium and carbonate incorporation. Relatively large crystals formed in comparison to those generally seen from reaction at low temperature. Lattice parameters of the CDHAp formed in water and PBS were calculated and are comparable to those expected in 2- CDHAp with HPO 4 and some carbonate incorporation. 2.2 Introduction Tricalcium phosphate (TCP) has been used for many years as a filler phase in calcium-phosphate bone cements or as part of a biphasic HAp/TCP ceramic. Both the

22 7 low-temperature β-tcp and higher temperature α-tcp polymorphs are used in these applications. α-tcp can hydrolyze in water to form a calcium-deficient hydroxyapatite (CDHAp) within 24 hours according to equation Ca 3 (PO 4 ) 2 + H 2 O Ca 9 HPO 4 (PO 4 ) 5 OH 2.1 Because of its ability to act as a self-setting cement, α-tcp hydrolysis has been studied extensively β-tcp is more stable and has lower internal energy, yielding lower reactivity in water. However, in comparison to stoichiometric HAp, β-tcp is more soluble over all ph ranges. Thus biphasic ceramics are frequently used. The β-tcp phase is observed to be resorbable in vivo, with some dissolution of the crystals and precipitation of HAp. 15,16 Some research has been conducted on in vitro formation of HAp from β-tcp using powders. Gbureck et al. found β-tcp could hydrolyze to a poorly crystalline HAp with some β-tcp phase remaining within 24 hours using Na 2 HPO 4 solution. 17 The authors attributed this reactivity to mechanical activation of the particles and deemed this a crystalline to amorphous phase transformation after milling the particles to a poorly crystalline β-tcp. Kaneno et al. observed highly crystalline HAp formation from lightly milled β-tcp particles hydrolyzed in a NH 4 OH solution at 90 o C. 18 Formation of HAp from β-tcp in simulated physiologic fluids has also been researched. Soaking of a β-tcp ceramic in phosphate buffered saline at 20 o C has been observed to form HAp with morphology similar to that observed in α-tcp. 19 Tape-cast sheets of β-tcp have been observed to precipitate apatite on the surface in Hank s

23 8 balanced salt solution, a simulated body fluid, while exhibiting less dissolution in PBS. 20 Other researchers have observed little or no apatite formation on β-tcp surfaces. 21 The aim of this study was to characterize the formation of CDHAp from hydrolysis of β-tcp in water, PBS, and simulated body fluid solutions using crystalline particles. 2.3 Materials and Methods β-tcp was formed by heating monocalcium phosphate monohydrate (MCPM) and CaCO 3 that had been milled together in a molar ratio of 2:1. The precursor solids were fired at 960ºC for 1.5 hours and allowed to slowly cool to room temperature. The resultant product was determined by X-ray diffraction (XRD) analysis to be phase pure β- TCP. The β-tcp was then ground by mortar and pestle and sieved through a 63 µm mesh. The particle size was further decreased by ball and attrition milling for 24 and 8 hours, respectively. Milling was performed in anhydrous heptane. The final average particle size, as measured using a Zetasizer Nano instrument (Malvern, PA), was 0.9 µm. Two solutions of simulated body fluid were synthesized according to published methods of Kokubo and Tas and are referred to SBF-K and SBF-T. 22,23 The most notable difference between the two preparations are the carbonate and chloride contents; the Tas preparation contains higher carbonate concentration and less chloride, making it more similar to concentrations in human plasma. Information relating to the reagents used is listed in Table 2.1; 1L of each solution was made. Pre-packaged phosphate buffered

24 saline concentrate (PBS) (Sigma, St. Louis, MO) was mixed with 1L of distilled, deionized water. 9 Table 2.1: Types of reagents and amounts used for preparation of simulated body fluid. Kokubo Tas Reagent Amount Reagent Amount NaCl 7.996g NaCl 6.457g NaHCO g NaHCO g KCl 0.224g KCl 0.373g KH 2 PO. 4 3H 2 O 0.228g Na 2 HPO g MgCl. 2 6H 2 O 0.305g MgCl. 2 6H 2 O 0.305g CaCl g CaCl. 2 2H 2 O 0.368g Na 2 SO g Na 2 SO g Tris-Buffer 6.057g Tris-Buffer 6.057g 1M HCl 35mL 1M HCl 25mL The solutions were mixed with β-tcp powder using a liquid-to-β-tcp ratio of 10:1 and maintained in 6mL plastic vials in a humidified chamber at 37ºC for up to 32 weeks. Samples were removed at designated intervals for analysis; the liquid supernatant was extracted and passed through 0.2µm filters and the solids were washed with distilled water followed by acetone to stop further reaction. The dried powder samples were analyzed using XRD. The water and PBS solutions were analyzed for calcium, and sodium in the case of PBS, by atomic emission spectroscopy and for phosphate by ion chromatography. The ph values of the liquids were also determined. X-ray diffraction was performed using a Scintag 2 (Scintag, Inc., Sunnyvale, CA) instrument. Generally, powder samples were pressed into glass sample holders and scanned over o 2θ. The patterns were compared to ICDD PDF cards and for HAp and β-tcp, respectively. However, two samples were chosen for lattice parameter calculations. These samples were analyzed over o 2θ using

25 10 steps of 0.02 o and 5 seconds at each step. The samples were placed in quartz zero background holders, and a polycrystalline silicon standard was measured in the same manner to calibrate the sample peaks. Lattice parameters were calculated using MDI Jade 8.0 software, and the reflection planes chosen were (213), (321), (402), (322), (214), and (323). These peaks were chosen to avoid overlap with β-tcp peaks in the PBS sample. Peaks were profile fitted and calculations were performed multiple times for both patterns. Infrared spectroscopy was performed using 1 wt% of sample in 0.3g of KBr. Pellets were pressed for 4.5 minutes under vacuum. A Nexus 670 FT-IR spectrometer (Thermo Nicolet, Corp., MA) was used in DTGS detector mode, and the pellets were scanned over a range from 4000 to 400 cm -1. Scanning electron microscopy was performed on the samples after sputter coating with gold. 2.4 Results and Discussion Phase Evolution Figures 2.1 and 2.2 show the variation in XRD patterns for the solids present when β-tcp is hydrolyzed in water and the PBS solution for 32 weeks. β-tcp converted to HAp completely by 20 weeks of reaction. Significant HAp formation was also observed in solutions of PBS, although conversion remained incomplete at 32 weeks

26 11 O β-tcp HAp 32 weeks 20 weeks O O O 12 weeks Relative Intensity 8 weeks 6 weeks 4 weeks 3 weeks 2 weeks 1 week 1 day Figure 2.1: XRD showing development of HAp phase in β-tcp hydrolyzed in water for up to 32 weeks. 2θ

27 12 O β-tcp HAp O O O O 32 weeks 20 weeks 12 weeks 8 weeks Relative Intensity 6 weeks 4 weeks 3 weeks 2 weeks 1 week 1 day Figure 2.2: XRD patterns of β-tcp reacted in PBS showing development of HAp. 2θ Hydrolysis of β-tcp in simulated body fluid solutions showed virtually no formation of HAp after 32 weeks. Figure 2.3 shows the XRD pattern for reaction in SBF- K at selected time periods while Figure 2.4 shows XRD after reaction for 1 day and 32 weeks in the SBF-T solution. The lack of conversion in the SBF solutions is due to the presence of species in the solution known to suppress HAp formation.

28 13 o β-tcp HAp o o o o o o 32 weeks 20 weeks Relative Intensity 8 weeks 4 weeks 2 weeks 1 day Figure 2.3: XRD patterns of β-tcp in solution with SBF-K for up to 32 weeks. 2θ

29 14 Relative Intensity 32 weeks 1 day Figure 2.4: XRD patterns of β-tcp in solution with SBF-T for 1 day and 32 weeks. No HAp phase is observed at any point in the interim. 2θ Microstructure SEM images of the solids after reaction for 32 weeks show the formation of needle-like crystals of HAp when β-tcp was hydrolyzed in water and PBS solution, as seen in Figure 2.5 and Figure 2.6. The aspect ratio of the HAp formed in the PBS solution is higher than in water; the lengths of HAp were approximately 1.5µm and 0.5 µm from PBS and water, respectively. Despite its continued persistence in the XRD patterns, β- TCP could not be observed by SEM in samples hydrolyzed in the PBS; thus HAp crystallites are surrounding β-tcp particles as expected.

30 15 2.5µm Figure 2.5: Microstructure of HAp formed from β-tcp in water after reaction for 32 weeks.

31 16 2µm Figure 2.6: Microstructure of HAp developed after reaction of β-tcp in PBS solution for 32 weeks. As seen in Figure 2.7 β-tcp immersed in the SBF solution shows minimal evidence of conversion. This is unexpected because SBF solutions generally precipitate a calcium phosphate phase on bioactive surfaces. Lack of HAp formation would be in accord with prior results that demonstrated the ability of magnesium to inhibit HAp formation from mixtures of dicalcium and tetracalcium phosphates. 24,25 Those HAp precursors are capable of producing solutions highly supersaturated with respect to HAp. β-tcp is not as soluble and does not readily form a supersaturated solution with respect to HAp. According to LeGeros, above a 0.03 to 1 Mg-to-Ca molar ratio in solution, a magnesium substituted β-tcp phase will form. 3 The molar ratio of calcium after 1 day of

32 17 reaction in water to the amount of magnesium in the SBF solutions is approximately Thus, the low degree of supersaturation of HAp due to the low solubility of β-tcp is coupled with the effects of magnesium in solution to limit HAp formation. 3µm Figure 2.7: Typical particle observed after reaction in an SBF solution for 32 weeks Solution Chemistry The variations in ph values of the solutions are shown in Figure 2.8. As is generally seen with reactions forming calcium-deficient HAp from calcium phosphate precursors, initially high ph values gradually decrease to below neutral as CDHAp is formed in water and PBS solution. This is typical of TCP hydrolysis, and ph values of solution in contact with α-tcp have been observed to decrease over approximately 16 hours. 13 The ph of the PBS solution remains slightly higher than that of water over the

33 18 extended reaction period, almost certainly due to its buffering capability. However, the increased ph in the SBF solutions is somewhat unexpected, as the SBF solutions alone are at a ph of 7.4 with a high quantity of buffer present. This would suggest the formation of a calcium phosphate with a Ca/P ratio below 1.5 and would be consistent with the surface formation of Mg-containing whitlockite. Thus, ph behavior is in accord with the lack of conversion to HAp. 13 H 2 O 12 PBS SBF-K 11 SBF-T 10 ph Time (days) Figure 2.8: ph values measured after reaction in solution at each time interval analyzed. The calcium and phosphate ion concentrations increase with time as β-tcp hydrolyzes in water and PBS solution, as seen in Figure 2.9. Although the PBS solution is expected to present a higher phosphate ion concentration at all times, a higher

34 concentration develops in water. This is due to the more rapid dissolution of β-tcp in water, as evidenced by its complete conversion to HAp. 19 Ion Concentration (mm) Ca 2+ H 2 O P i H 2 O Ca 2+ PBS P i PBS Time (weeks) Figure 2.9: Variation in ion concentrations over time for hydrolysis in water and PBS solution HAp Characterization The calcium-to-phosphate ratios in the solids formed after reaction were calculated using the solution chemistry data. The Ca/P ratios in the HAp formed in water and PBS are shown in Table 2.2. The HAp formed in water is calcium-deficient, as expected with TCP hydrolysis. When sodium is included with the calculation in PBS, the analysis indicates the incorporation of sodium by the HAp that is formed.

35 20 Table 2.2: Calculated Ca/P ratios of the HAp formed in water and PBS solution. H 2 O: Ca/P PBS: (Ca+Na)/P 1 day week weeks weeks weeks weeks weeks weeks Infrared spectra obtained from samples after 32 weeks of reaction in solution and unreacted β-tcp are shown in Figure All samples, including the unreacted β-tcp, show carbonate peaks around 1460, 1420, and 875 cm -1. Other researchers have noted some degree of carbonate incorporation in β-tcp after milling. 18 The band at 1460cm -1 is almost certainly also related to adsorbed heptane from the milling process; the peak around 715cm -1 and other higher frequency peaks (not shown) seen in all spectra are due to residual heptane. The sample in SBF-T solution is very similar to the unreacted powder, with the exception the band at 3640cm -1. This band is almost certainly related to hydroxyl groups on the surface of the β-tcp. Once the powder is hydrated, the OH groups enter the solution, accounting for the high ph observed initially in water as will be shown in Figure 3.1.

36 21 H 2 O PBS SBF-K Transmittance SBF-T β-tcp Wavenumber (cm -1 ) Figure 2.10: FT-IR spectra of samples after 32 weeks of reaction in water, PBS, and SBF solutions and sample of unreacted β-tcp. The sample hydrolyzed in water has a sharp peak at 3570cm -1 related to the stretching mode of the hydroxyl group, along with a band at 630cm -1 related to OH libration. These peaks are not as strong in the samples in PBS. The broad band at 875cm -1 seen in water is due to HPO 4 2- as well as carbonate. The carbonate bands at higher frequencies are broader in the unreacted powder while that at 875cm -1 is sharp; this reverses in the sample from water. The broader band may be attributed to an overlap of carbonate and acidic phosphate peaks. The band at 875cm -1 is also weaker in the sample 2- from PBS. This may be related to remaining β-tcp in the sample, or less HPO 4 due to the buffering capacity conferred by the solution. Because the relative intensities of the

37 22 carbonate 1420cm -1 to phosphate 600cm -1 bands are very similar in both samples, the reduced intensities noted for the OH and HPO 4 groups in the sample hydrolyzed in PBS cannot simply be attributed to differences in IR sample preparation. The sharp, well-defined peaks seen in the XRD patterns from the samples in water and PBS after complete reaction were suitable for a full scan from 10-65º 2θ to determine the lattice constants. The patterns from this extended measurement can be seen in Figure 2.11, and the lattice parameters calculated using the Jade software are shown in Table 2.3. Lattice parameters for CDHAp formed from α-tcp have been determined by other investigators 26, however those for HAp produced by hydrolysis of β-tcp do not appear to have been established at low temperature. The properties crystallinity, reaction temperature, morphology of α-tcp used for the lattice parameter calculations, however, were not given. Additionally, the morphology observed for the HAp formed from β-tcp in this study varies significantly from the plate-like morphology observed for HAp synthesized from α-tcp hydrolysis. 12,14 For the HAp formed in water, the values were calculated to be a = Å and c = Å. This a parameter value represents an increase over stoichiometric HAp, which has an a-axis lattice dimension of Å 4, and is expected when HPO 4 2- substitutes into the lattice. An increase is expected because HPO 2-4 partially replaces PO 3-4 and this replacement breaks a P-O-P linkage by producing P-OH. Young and Holcomb have correlated the change in a-axis lattice parameter with the amount of HPO 2-4 incorporated. 27 Using this value, the amount of acidic phosphate incorporate is calculated to be very near that of ideal CDHAp. The c- axis lattice parameter is near the same value for a stoichiometric HAp, Å.

38 However, HPO 4 2- is observed to decrease the c-axis lattice dimension. 4 The slightly 23 smaller than expected increase in the a-axis lattice parameter and lack of c-axis decrease may also be attributed to Type-B carbonate incorporation, which is known to decrease the a-axis while increasing the c-axis lengths. PBS Relative Intensity (213) (321) (402) (322) H 2 O (214) (323) Figure 2.11: XRD patterns of HAp formed after reaction for 32 weeks in water and PBS solution scanned using a quartz zero background holder and step scan. 2θ

39 Table 2.3: Lattice dimensions calculated for HAp formed in water and PBS solution after 32 weeks. All values displayed are in units of angstroms. H 2 O PBS a-axis c-axis a-axis c-axis Average ± ± ± ± In PBS, the lattice parameters are calculated to be a = Å and c = Å. According to statistical analyses, the average values of the a-axis parameters measured in this study are significantly different while no difference between the mean of the c-axis parameters can be discerned. These values are slightly higher than expected in comparison to CDHAp formed in water, because sodium incorporation should have no effect on either the a- or c-axis dimensions 3. El Feki et al. have observed that coupled sodium and carbonate incorporation in a calcium-deficient apatite decreases the a-axis length while increasing the c-axis dimension. 28 However, comparison to these lattice parameters is not appropriate as the samples were heated before analysis, removing HPO 4 2- groups from the lattice. The simplest explanation for the increased lattice parameters is that the peaks are broader and less intense than HAp formed in water and exact analysis becomes more difficult when the peaks are less sharp, increasing the uncertainty of correct peak fitting at high angles.

40 Conclusions Formation of crystalline HAp particles from β-tcp at physiologic temperature could be achieved using water and phosphate buffered saline solution. Significant formation of HAp was noted after 1 week of reaction in water. This is less time than was required for other β-tcp samples under the same experimental conditions, as discussed in the following chapter. However, this merely indicates that even subtle differences between batches of prepared β-tcp can have a significant effect on the reactivity of the powder. Solutions with extraneous ions delayed or prevented HAp formation from β-tcp in the conditions studied. In the SBF solutions, this is attributed primarily to magnesium, an inhibitor of HAp formation even in systems known to produce solutions supersaturated with respect to HAp. In the β-tcp system, the amount of magnesium is sufficient to prevent formation of HAp. This occurs in the SBF because calcium will only be present in the solution once the β-tcp dissolves. Lattice parameters for a calcium-deficient HAp could be calculated from the sample reacted in water. These were found to be a = Å and c = Å. Lattice parameters for CDHAp formed from β-tcp have not been calculated previously. Synthesis conditions for crystalline CDHAp with diffraction peaks amenable to lattice parameter calculation are generally limited to relatively high temperatures, around 90 o C, in solution. The only impurity noted in this sample is some carbonate incorporated from the atmosphere, and the Ca/P ratio is Carbonate is ubiquitous and impossible to exclude from substitution. The precise amount of carbonate incorporation is difficult to

41 determine from the experiments conducted, however it is expected to be no more than 0.1 moles per formula unit. 26

42 Chapter 3 Formation of fluoridated apatite from β-tricalcium phosphate at physiologic temperature 3.1 Abstract The formation of calcium-deficient fluoridated apatite from β-tricalcium phosphate (TCP) was observed at physiologic temperature. Reactions in different concentrations of ammonium fluoride solutions were found to influence the rates of β TCP conversion. The variations in solution chemistry were followed for up to 8 weeks of reaction, allowing determination of the molar ratios of calcium, fluoride, and phosphate in the solids formed. All apatites were found to be calcium deficient to some extent while also incorporating small amounts of carbonate. TCP hydrolysis in the solution having the highest fluoride concentration produced an apatite containing a proportion of fluoride in excess of that which could be accommodated in the sites normally occupied by fluoride, regardless of possible CaF 2 formation. When TCP was hydrolyzed in solutions with low fluoride concentrations, early formation of a fluoridated apatite was followed by slow hydrolysis of the remaining β TCP. β TCP hydrolysis in distilled water conversion was about 80% complete and highly crystalline calcium-deficient HAp was produced. Changes in apatite crystallite morphology with varying fluoride incorporation were noted; the aspect ratios of the needles formed increased as the original fluoride concentrations in solution decreased.

43 Lattice parameters were calculated for both fluoride substituted and non-substituted compositions Introduction It is well understood that HAp compositions deficient in calcium can form The compositional range for HAp can be written as Ca (10-x) (PO 4 ) (6-x) (HPO 4 ) x (OH) (2-x) where x ranges from zero to approximately one. While the compositional range for fluoride-containing compositions can be similarly expressed as Ca (10-x) (PO 4 ) (6- x)(hpo 4 ) x F (2-x), whether such a range of compositions exists awaits empirical determination. Human tooth enamel is a fluoridated apatite with a Ca/P ratio less than stoichiometric (1.67) 4. Enamel also contains a small amount of carbonate, and has been found to contain HPO 2-4 ions. A variety of studies have explored fluoride substitution in hydroxyapatite. Substitution of F - for OH - at the surface of hydroxyapatites has been extensively studied for caries prevention. Lin et al. noted increased fluoride uptake with lower Ca/P ratios in hydroxyapatite 32. Alternatively, Leamy et al. observed extents of fluoridation from 17 to 72% when calcium-deficient HAp from hydrolyzed α-tcp was reacted in the presence of NaF. 33 These data indicate the extent of fluoride incorporation into stoichiometric apatite is low while that formed from α-tcp can be high. Other authors have investigated the incorporation of fluoride into apatite during formation. TenHuisen and Brown explored the substitutional chemistry of apatite formed from α-tcp in the presence of NaF. 34 That study indicated paired substitution of Na and F and indicated the (Na+Ca)/PO 4 ratio

44 29 attained in 0.1 M NaF solution was approximately Sakamoto et al. prepared a stoichiometric fluoridated apatite with approximately 86-94% fluoride incorporation from β-tricalcium phosphate at temperatures of 40 o and 70 o C and slightly less uptake 64-81% - using a-tcp. 26,35 High concentrations of fluoride are found to produce CaF 2 in addition to fluorapatite. NMR has been used to detect CaF 2 formation concurrent with fluorapatite at high concentrations. 36 Zeta potential determinations were used by Lin et al. 32 to show a similar trend; however the range of fluorapatite stability was increased at higher ph. Formation of CaF 2 phase is undesirable in applications related to dental materials. The present study investigates the incorporation of fluoride associated with the conversion of β-tcp to hydroxyapatite. Both α-tcp and β-tcp will hydrolyze to HAp; however, β-tcp has a much slower rate of dissolution. Thus, hydrolysis of β-tcp occurs under quasi steady state conditions. In view of the ability of HAp to incorporate Na and K, in the present study hydrolysis was carried out in the presence of ammonium fluoride anticipating minimal uptake of ammonium ion by the HAp being formed. Variations in the solution chemistries of slurries comprised of β-tcp and ammonium fluoride solutions of various concentrations were followed for up to 8 weeks. Four NH 4 F concentrations were analyzed, along with a water control. The molar ratios of TCP-to-NH 4 F employed were 1, 2, 3, and 6. The solid reaction products expected at TCP-to-NH 4 F ratios of 1 and 2 were CaF 2 and fluoride-substituted HAp along with significant accumulation of phosphate in solution. When the mole ratio is 3:1, an ideal reaction would a produce fluoride-substituted HAp as the only solid reaction product according to Eq. 3.1.

45 3Ca 3 (PO 4 ) 2 + NH 4 F(aq) + H 2 O Ca 9 HPO 4 (PO 4 ) 5 F + NH 4 OH(aq) When the mole ratio is 6:1, an ideal reaction would produce the fluoride substituted composition Ca 9 HPO 4 (PO 4 ) 5 F 0.5 OH 0.5 as the only solid. Establishing the compositions formed by determining the variations in solution chemistry is relevant to establishing the stabilities of calcium-deficient, fluoride-substituted apatites. 3.3 Materials and Methods The precursor mixture to β TCP was formed by mixing particulate calcium carbonate and monocalcium phosphate monohydrate at a molar ratio of 2 to 1. The powders were milled in heptane using zirconia media for 16 hours, filtered, and dried. This precursor mixture was then heated at 960 o C for 1.25 hours to form β TCP. After phase purity was confirmed by x-ray diffraction, the β TCP was ground and passed through a 63µm mesh sieve, ball milled for 24 hours, and attrition milled for 8 hours in heptane. This produced β TCP with typical particle size of 1-2µm, with a distribution from 0.5 to 6µm as measured by SEM. The ph variations as β TCP hydrolyzed were followed continuously for 24 hours during its reaction with NH 4 F with stirring under nitrogen at a liquid-to-tcp weight ratio of 50 to 1. This was accomplished by placing a plastic beaker containing the reactants inside a jacketed reaction vessel attached to a water bath held at 37 o C. Four NH 4 F concentrations determined as TCP:NH 4 F mole ratios were used and are referred to as 1:1,

46 31 2:1, 3:1 and 6:1. A control experiment in which β TCP was hydrolyzed in distilled, deionized water was performed as well. Longer-term studies of ph variations were also carried out along with the analyses of variations in the concentrations of calcium, phosphate and fluoride. β TCP powder was mixed with NH 4 F solutions or water in small (6mL) or large (15mL) plastic vials at a liquid-to-tcp ratio of 30:1, capped, and aged at 37 o C. The samples were removed at 1, 2, 5, and 7 days and 2, 3, 4, 6, 8 and 20 weeks depending on the concentration. The sample supernatants were collected, passed through a 0.2µm syringe filter, and analyzed for calcium by atomic absorption spectroscopy and fluoride and phosphate concentrations by ion chromatography. The solids present were flushed with ethanol and allowed to dry in air before undergoing XRD analysis. X-ray diffraction analyses were carried out using a Scintag 2 diffractometer (Scintag, Sunnyvale, CA). All samples were scanned from o 2 theta and compared to ICDD cards 9-169, , and 9-432, corresponding to β TCP, FAp, and HAp, respectively. For analysis of the relative amount of each phase remaining at each time, MDI Jade 8.0 software was also used. All peaks were fitted and non-overlapping peaks (210), (211), (112), (300), (202) and (214), (0,2,10), (220) for HAp and β-tcp, respectively, were selected for analysis. ICDD cards and , FAp and β TCP respectively, were selected. The 4 week sample from solution with ratio 1:1 and 20 week sample in water, along with an external standard of polycrystalline silicon, were scanned over o 2θ using steps of 0.02 o for 5 seconds each. Lattice parameters were calculated using the Jade software. Peaks for the (213), (321), (410), (402), (004) and

47 32 (322) reflection planes were used for analysis of FAp from the 1:1 sample. For HAp, (213), (321), (402), (322), (502), and (323) peaks were chosen for analysis. Infrared spectroscopic analyses were performed using a Nicolet FT-IR spectrometer in MCT and DTGS detector modes. Powders from samples reacted in the fluoride solutions were scanned as prepared and after heating to 500 and 925 o C. Typically 1 wt% of sample was mixed and ground with KBr and pressed at 5000 psi under vacuum. Scanning electron microscopy was also performed on the reacted powders. Samples were sputtered with gold and a Hitachi S-3500N instrument was used at 5kV to collect images. 3.4 Results Figure 3.1 shows the ph changes over the first 24 hours of reaction in water and the fluoride solutions. The initial ph values of the NH 4 F solutions were near 9, and modest decreases in ph were observed during the first 24 hours of hydrolysis. While the variations show some structure, no mechanistically significant changes were observed. Hydrolysis in water rapidly produced ph above 11 and which gradually decreased to near 9 by the end of 24 hours. Over an extended period of time, solutions having TCP:NH 4 F mole ratios of 1:1 and 2:1 continued to maintain a ph around 9.5 as shown in Figure 3.2. After 4 weeks of reaction, the ph of values of solutions having mole ratios of 1:1 and 2:1 plateau at a higher value than do those where the mole ratios are 3 and 6. Over the next 4 weeks, the ph of solutions having mole ratios of 3:1 and 6:1 continue to decrease. Hydrolysis of β TCP in water for 20 weeks produced a ph of near 5.

48 H 2 O 1:1 2:1 3:1 10 ph Time (hours) Figure 3.1: ph variation over the first 24 hours of samples reacted in water or ammonium fluoride solutions.

49 :1 2:1 3:1 6: ph Time (days) Figure 3.2: ph of β-tcp in static solutions of NH 4 F over 8 weeks of reaction. Concentration of NH 4 F decreases from 1 to 6. The concentrations of ions present after reaction are presented in Figures The concentration of calcium in solution having a TCP:NH 4 F mole ratio of 2:1 decreases to below the detection limit by 1 month, but remains measurable, although very low, for solutions having mole ratios of 1:1, 3:1, and 6:1. For the solution having a mole ratio of 1:1, the concentrations of phosphate and fluoride in solution are paired; while the amount of fluoride in solution steadily decreases over the first 24 hours, the amount of phosphate increases as illustrated in Figure 3.6. Both ions reach a plateau after 48 hours, and fluoride remains well above the limit of detection throughout the experiment. Reaction at

50 35 a mole ratio of 2:1 results in fluoride depletion after 3 days, despite containing more fluoride than would be incorporated according to Eq The phosphate concentrations for both solutions remain high. Reaction at a ratio of 6:1 produces a lower concentration of phosphate than of fluoride, likely due to its decreased rate of β TCP dissolution as indicated by minimal HAp formation. [Ca 2+ ] (mm) :1 2:1 3:1 6:1 H 2 O Time (days) Figure 3.3: Variations in calcium ion concentrations during the hydrolysis of β-tcp for up to 8 weeks in the NH 4 F solutions and in water.

51 :1 2:1 3:1 6:1 H 2 O 10 [P i ] (mm) Time (days) Figure 3.4: Phosphate ion concentrations for up to 8 weeks during the hydrolysis of β- TCP in the NH 4 F solutions and in water.

52 :1 2:1 35 3:1 6: [F - ] (mm) Time (days) Figure 3.5: Variations in fluoride ion concentrations in solution during the hydrolysis of β-tcp in NH 4 F solutions for up to 8 weeks Ion Concentration (mm) F - P i Ca Time (hours) Figure 3.6: Variations in fluoride, calcium, and phosphate ion concentrations measured over the first 24 hours in a 1:1 NH 4 F:TCP solution.

53 38 XRD patterns of the solids present after reaction for 4 weeks are shown in Figure 3.7. The calculated percentages of the phases present, based on the peak heights in these patterns, are shown in Table 3.1. After 2 days of reaction at a TCP:NH 4 F mole ratio of 1:1, an apatite has formed that closely matches FAp with no β TCP phase remaining. No further changes were seen over the course of the month with respect to phases present. For reaction at a mole ratio of 2:1, the amount of β TCP declined gradually over the first week, but this decrease tapers off and both FAp and β-tcp were detected for the remainder of the experiment. Reaction at a mole ratio of 3:1 exhibits a similar pattern, with more β TCP present by the end of 2 months than the 2:1 reaction. Finally, reaction at a mole ratio of 6:1 results in an amount of residual β TCP present greater than that of apatite phase even at the end of 2 months. Hydrolysis of β-tcp in water for 5 months resulted in about 80% conversion to HAp; other experiments with β- TCP in water have had complete conversion by this time as evidence by data presented in Section 1.3.

54 39 Apatite β-tcp 1:1 Relative Intensity 2:1 3:1 6: θ (degrees) H 2 O Figure 3.7: XRD patterns of β-tcp after 1 month in fluoride solutions with TCP:NH 4 F molar ratios as labeled or in water. Table 3.1: Percentage of FAp phase observed after reaction. BDL = below detection limit. 1 day 5 days 7 days 2 weeks 4 weeks 6 weeks 8 weeks 1: : : : BDL BDL The lattice parameters of the FAp present after one month of reaction, when the TCP:NH 4 F mole ratio was 1:1, were calculated from XRD scans over o 2θ. The average values are as follows: a = ± Å and c = ± Å.

55 40 Although not all β-tcp had reacted after 20 weeks in water, the lattice parameters of the hydroxyapatite formed could be calculated and are as follows: a = ± Å and c = ± Å. The FT-IR spectra are characteristic of apatites, and indicate incorporation of fluoride and carbonate. Carbonate bands around cm -1 are present in all samples, including the unreacted, milled β TCP powder. In the samples with high fluoride concentration, these carbonate bands are sharper and more resolved. The carbonate band at 865cm -1 has a shoulder at around 872cm -1 in the three highest fluoride concentrations. The carbonate band shifts from 875cm -1 in unreacted β TCP towards 865cm -1 in the two highest concentrations. Figure 3.8 shows the infrared spectra after the samples are heated to 500 o C. All spectra except that from TCP:NH 4 F mole ratio 6:1 have a sharp peak at 3540 cm -1. A peak at 3570cm -1 is readily apparent in the sample from reaction having a mole ratio of 3:1 and a weak shoulder can be seen in the two higher concentrations as well. The carbonate peak at 865cm -1 splits into two distinct bands; the second peak is around 879cm -1. The 865cm -1 peak becomes much stronger and sharper in the sample from the solution with a mole ratio of 1:1. No change in the other carbonate peaks is apparent in any of the samples. Peaks around 2200 cm -1 appear, indicating the presence of nitrogenous species. When the samples are heated to 925 o C, as shown in Figure 3.9, the relative intensities of the carbonate bands change. The band at 3570 cm -1 disappears almost completely, leaving only the 3540cm -1 band. Samples from solutions with mole ratios 2:1 and 3:1 show a shift in the 1455cm -1 band towards higher frequencies and an increased

56 41 intensity in relation to the other carbonate band at 1427cm -1, which has shifted to lower frequencies. The 865cm -1 band is no longer present and has been replaced by bands at 879 and 873cm -1, the former having a higher intensity relative to the latter. Additionally, a small band at 1556cm -1 can be seen for these samples. The carbonate bands in the sample from the 1:1 solution show no such change upon heating. With increasing temperature the band positions remain static while the intensities decrease relative to the phosphate bands. 1:1 Transmittance 1:1 2:1 3:1 6:1 2:1 6:1 3: Wavenumber (cm -1 ) Figure 3.8: FT-IR spectra after heating samples reacted for one month to 500 o C. Enlarged portion shows change in carbonate peaks around 865cm -1.

57 42 1:1 1:1 2:1 2:1 3:1 Transmittance 3:1 6:1 6: Wavenumber (cm -1 ) Figure 3.9: FT-IR spectra regions of interest after heating samples reacted for 1 month to 925 o C. SEM observations indicate changes in the microstructure of the apatite formed with varying concentrations of fluoride. Selected images are shown in Figure All samples formed needle-like crystals in clusters around β TCP particles. As can be seen in Figure 3.10, after 1 month of reaction the needles of FAp grown in the sample with mole ratio of 1:1 have a low aspect ratio. This is in comparison to those in the samples FAp from samples with mole ratios 2:1 or 3:1, as seen in Figure The length of the needles does not change, and is approximately 1µm, but the diameter is approximately double in the sample from high fluoride concentrations. In the image

58 43 shown in Figure 3.12, small crystals, approximately 0.25µm in length, can be seen growing on the surface of the β TCP from solution 6:1 after 1 month. β TCP is not seen in the samples with mole ratio 2:1 or 3:1, despite its continued presence as indicated by XRD. As seen in Figure 3.13, the sample reacted in water had thin, needle-like crystals that were longer than those in any of the fluoride solutions after 20 weeks of reaction. 2µm Figure 3.10: Fluorapatite formed in 1:1 NH 4 F:TCP solution after 1 month.

59 44 3µm Figure 3.11: FAp formed in 2:1 NH 4 F:TCP solution after 1 month.

60 45 3µm Figure 3.12: β-tcp in 6:1 NH 4 F:TCP solution after 1 month. 5µm Figure 3.13: Needles of HAp grown from β-tcp in water at 37 o C after 5 months of reaction.

61 Discussion The ph during the course of reaction in NH 4 F solutions is not high enough to eliminate the possibility of HPO 2-4 incorporation into the apatite formed. The ph plateau in the two highest concentrations is too low to be attributed to NH 4 OH(aq), which would be expected to have a ph of near 11. This, coupled with the amount of phosphate present by solution chemistry measurements, indicates formation of (NH 4 ) 3 PO 4 (aq). The formation of (NH 4 ) 3 PO 4 (aq) rather than NH 4 OH(aq) represents a deviation from the predicted reaction in Eq. 3.1, but would be expected if the product formed had a Ca/P ratio greater than 1.5. The slowly decreasing ph of β TCP in water over the course of the reaction is similar to α-tcp 13 but on a much longer timescale. The initial high ph of β TCP in water may be indicative of a hydrated surface layer. Unlike the samples with mole ratios of 3:1 and 6:1, those with TCP:NH 4 F mole ratios of 1:1 and 2:1 reached equilibrium after one week. Beyond this time, the ph remains steady and the relative amount of each phase present in XRD does not change. In comparison, there is a gradual increase in the amount of apatite phase present in the samples having a mole ratio of 3:1. Between 6 and 8 weeks of reaction the amount of apatite remains constant, which correlates to a nominal ph change between these time periods. Reaction at a mole ratio of 6:1 results in a decrease in ph between weeks 4 and 8, yet the amount of phosphate in solution increases. While this indicates formation of (NH 4 ) 3 PO 4 (aq) in the higher concentration solutions, for this concentration it likely indicates dissolution of the β TCP phase and formation of apatite. A similar increase in

62 47 phosphate in solution is seen in water when at the final time point. XRD of both show increased apatite phase at these points. The solution chemistry data permit calculation of the ratios of each ion present in the solid formed. Samples having TCP:NH 4 F mole ratios of 3:1 and 6:1 match well with their expected reactions, producing solids with Ca/F ratios of 9 and 18, respectively. Samples having mole ratios of 1:1 and 2:1 were expected to retain excess fluoride in solution, or uptake slightly more of this ion. The sample with a mole ratio of 1:1 retains fluoride in solution throughout the reaction, as seen by the plateau in Figure 3.5. The Ca/F ratio in the solid formed, however, holds steady at 3.82 beyond 7 days. Assuming either 9 or 10 moles Ca in the solid, more than 2 moles of fluoride must be incorporated. However, the location of these excess F ions is undetermined. The increased PO 3-4 in solution coupled with decrease of F - may indicate replacement in the phosphate sites or formation of a stoichiometric FAp. Coupled incorporation of F - and CO 3 2- ions into PO 4 sites has been proposed, however only a fraction of the incorporated CO 3 2- ions are associated with excess F Coupled incorporation of F - 2- and CO 3 has more recently been disputed and interstitial sites have been proposed for excess fluoride. 38 Incorporation of 2- an FPO 3 ion has also been observed in apatite Although incorporation of FPO 3 would help maintain charge balance in the apatite formed, there is no evidence that this ion would be formed in the solution. Incorporation of HPO 2-4 would also balance the charge in a calcium-deficient apatite, but would not explain the amount of fluoride without excess fluoride in OH sites. However IR spectra show evidence for some amount OH - remaining even in heavily fluoridated samples, indicating the probability that excess F - occupies these sites to be low. Additionally, no β-tcp is seen in XRD patterns after

63 heating the highly fluoridated sample, indicating no HPO is present. The data for reaction in 2:1 solution can be explained similarly by assuming incorporation of FPO 2-3, however determining a precise stoichiometry is complicated because not all β TCP has reacted. If preferential dissolution has occurred, more phosphate or calcium may be in the solution. Incorporation of excess fluoride into OH sites is possible but, as stated above, these sites retain some amount of OH -. All calculations show a calcium to phosphorus ratio less than 1.67, which, coupled with some amount of CO 3 -for-po 4 substitution, would indicate vacancies in the calcium site. However, if the unreacted β TCP is taken into account and assumed to contribute no ions to the solution, the calculated ratios changed dramatically. Tables 3.2 and 3.3 show the differences observed in Ca/P and Ca/F ratios when this correction is employed for 1:1 solutions over the first 48 hours and the other fluoride solutions over an extended time period, respectively. In this case, reaction in all fluoride solutions, except the solution with a mole ratio of 6:1, produces an HAP in which there is a decrease in Ca/P ratio over time. Reaction in the 6:1 solution and water produce a HAp where there is a increase in Ca/P ratio from 1.5 to 1.69 and 1.58, respectively.

64 Table 3.2: Calculated ratios of ions in a solid formed in 1:1 TCP:NH 4 F solution calculated normally and with correction for unreacted β-tcp. Normal Corrected Hours Ca/P Ca/F Ca/P Ca/F Table 3.3: Calculated ratios of ions present in solid formed calculated normally and corrected for unreacted β-tcp. 2:1 3:1 6:1 Normal Corrected Normal Corrected Normal Corrected Day Ca/P Ca/F Ca/P Ca/F Ca/P Ca/F Ca/P Ca/F Ca/P Ca/F Ca/P Ca/F A decrease in Ca/P ratio over the course of reaction forming HAp has been seen in α-tcp hydrolyzed in water. 40 Formation of a near-stoichiometric FAp followed by hydrolysis of any remaining β-tcp to CDHAp would explain the decrease in Ca/P ratio, particularly for samples reacted in 2:1 and 3:1 solutions, where β-tcp does not fully react. However in this system the Ca/P ratio οf the HAp from β TCP hydrolyzed in water increases over time; this is likely due to carbonate incorporation from exposure to air over the extended time period allotted for reaction. Even a 1 wt % incorporation of

65 50 carbonate, approximately 0.15 moles replacing phosphate, could account for the Ca/P ratio seen in the sample, and the IR spectra indicate carbonate is present in all samples. Variations in Ca/P ratios over time have also been observed in SHAp and CDHAp formed from tetracalcium phosphate and dicalcium phosphate. In the former, a calcium-deficient apatite formed initially and, after the TetCP hydrolyzes, eventually SHAp forms. 41 In a CDHAp system, a SHAp has been theorized to form first followed by reaction with DCPA to form CDHAp. 42 From this we can assume that the expected phase at the end of reaction may not be the first one formed, and hence a stoichiometric apatite may form in this system first, resulting in a decrease in Ca/P ratio over time. The Ca/F ratios also change when the unreacted β-tcp is accounted for and increase over time. Samples having TCP:NH 4 F mole ratios of 3:1 and 6:1 have a final Ca/F ratio of approximately 7.5 if the remaining β TCP is accounted for in the calculations. Samples having a mole ratio of 2:1 have a Ca/F ratio of 5:1, which is expected from a stoichiometric FAp. The Ca/P ratio remains less than 1.67, however, and indicates formation of an apatite with some degree of calcium deficiency. Thus there must be less fluoride in the solid than fully stoichiometric FAp. The amount of fluoride in samples from solution with mole ratio 1:1 at long time periods does not change, however at early times the Ca/F ratio is 0.5. This could indicate initial formation of CaF 2, and although XRD analysis and SEM do not show any evidence for its formation, detection of CaF 2 in the small quantities potentially present is difficult. The possibility of CaF 2 forming and remaining throughout the experiment exists, and is particularly likely in the sample having a TCP:NH 4 F mole ratio of 1:1. As the reaction falls somewhere between producing stoichiometric and calcium-deficient FAp,

66 51 the phosphate in solution would indicate a proportional decrease in calcium in the solid formed. If this calcium reacts with the excess fluoride, CaF 2 would form. Taking this into account, the amount of CaF 2 potentially formed in the solution with mole ratio 1:1 after 1 month is approximately 3.8 wt% and is equivalent to a FAp-to-CaF 2 ratio of 2:1. If this amount of CaF 2 is present, the FAp formed still has excess fluoride because the Ca/P ratio becomes approximately 1.5. The a-axis parameter observed in these experiments is lower for the FAp than the HAp formed in water, as expected with fluoride incorporation. The calculated FAp a-axis value of Å is smaller than the value for hydroxyapatite, Å 3, larger than that of stoichiometric fluorapatite formed at high temperature, Å 43, and slightly less than a synthetic apatite formed in an aqueous system, Å. 3 There are several possibilities for variations in lattice parameters in the fluorapatite. Ammonium and carbonate incorporation are two possible reasons; incorporation of an HPO 4 2- ion is another. Incorporation of a PO 3 F 2- group with Ca site vacancies has been used to account for an increased a-axis parameter in comparison to stoichiometric FAp. 4 The c-axis parameter is slightly larger than that of the HAp formed in water in this study and the generally accepted value of Å and this change is attributed to carbonate incorporation. The a-axis value calculated for CDHAp is very near that reported for a non-stoichiometric calcium-deficient HAp, Å. 4 The FT-IR spectra indicate decreased amounts of F - in the solids with decreasing concentration in solution. Although the two highest concentrations were expected to have more F - than should incorporate based on Eq. 3.1, only the maximum concentration had virtually no OH - present. When the samples are heated, bands at 3540 cm -1 indicate some

67 52 amount of OH- in the solid. The sample from solution with molar ratio 3:1 also has a band at 3570 cm -1, indicating a greater amount of OH- in this sample. The carbonate bands also indicate this lowered F - substitution, as the CO 3 ion can move into the OH - sites upon heating. 4 As evidenced by the appearance of bands at 879 and 1550 cm -1 and increase of the band at 1470 cm -1 relative to the lower frequency band in this region, all associated with Type A carbonate, the carbonate ions replace OH - in the samples from solutions with molar ratios of 2:1 and 3:1. On the other hand, the bands in the sample from the highest concentration do not split or change upon heating. The bands seen at 2177 and 2019 cm -1 in all samples are related to cyanate and cyanamide, respectively, and are due to incorporation of some NH The shoulder at 875 cm -1 on the 865 cm -1 carbonate band seen in the three highest concentration samples is likely due to the shift from higher frequencies, and has been observed in carbonated samples with flouride. 45 Although this is the location expected for P-OH of HPO 2-4 group absorption, this ion is not expected to be incorporated with F -. 3 The IR results indicate the presence of CO 2-3 and OH - in the FAp formed. The carbonate is a contaminant from the atmosphere and is present in low amounts. The presence of hydroxyl groups in the sample from the fluoride solution with the highest concentration is explained by the continued high ph in the solution. The amount of carbonate in the samples is approximately the same for all, including in water. The sample from solution with a TCP/NH 4 F ratio of 6:1 remains β TCP and is very similar to the unreacted powder. This is expected based on the XRD pattern of the sample at 1 month, in which the amount of HAp dropped below the detection limit.

68 53 Larger sized crystals are expected in samples containing fluoride, compared to those grown in the absence of the ion. 3 Increased aspect ratios with decreasing fluoride concentration have been seen by other researchers. 35,46 The morphology of the crystals does not appear to change with increasing incubation periods in the solution with highest fluoride concentration. The growth of apatite in the lowest fluoride concentration solution appears to be less than in water at the same time period. Some researchers have found decreased apatite formation in the presence of low fluoride concentrations during precipitation experiments. 47,48 It has been proposed that fluoride facilitates conversion of β TCP by attacking the particle and forming clusters in which OH - ions can replace F - ions, freeing more F - ions to attack the β TCP. 35 This process explains how Sakamoto et al. found β TCP completely converted to FHAp relatively rapidly in all concentrations. 35 In the current study, however, complete conversion did not occur even over prolonged time periods. This is likely due to differences between stirred and quiescent solutions, as XRD of powders from samples with TCP:NH 4 F mole ratios of 2:1 and 3:1 in the stirred ph experiment show more apatite phase after 24 hours than is present in samples from the static solutions at 3 days. Clearly the presence of fluoride accelerates the formation rate of apatite in a positive manner at appropriate concentrations. The mechanism proposed for the conditions studied is as follows. As β TCP dissolves, a zone of supersaturated solution with respect to HAp is formed. This zone allows the precipitation of a surface layer that ordinarily becomes an amorphous calcium phosphate and prevents, or greatly inhibits,

69 54 further dissolution of the β TCP. This layer converts to HAp over time, as seen for the sample hydrolyzed in water. However, when fluoride is present small crystals of FAp are rapidly formed. These crystals are not effective in blocking the β TCP surfaces and the β TCP continues to dissolve. The solution with the highest fluoride concentration has excess fluoride and can completely react with β TCP. The other solutions lack enough fluoride to completely transform the β TCP, as evidenced by its persistence after long periods of time. The remaining β TCP may slowly convert to HAp, as it does in water, but the time scale of the experiments carried out were too short for this to be observed. This conclusion is related to the behavior in quiescent solutions and is somewhat different than that of Sakamoto et al., 35 wherein the fluoride was assumed to leave the β TCP/Ap cluster initially formed, an assumption that is not consistent with the crystal chemistry of FAp. Rather, the fluoride is substituted into the apatite as it is formed. The low amount of fluoride ions in solution after reaction for more than 3 days in all samples except solution 6 would support this conclusion, while the slow decrease in F - ion seen in solution 6 may indicate inhibition of apatite formation at this concentration. The presence of hydroxyl groups in all samples is due to the high ph of the solution, as in TCP:NH 4 F mole ratio of 1:1, or formation of hydroxyapatite with less F - ion incorporation, as with TCP:NH 4 F mole ratio of 2:1 and 3: Conclusions The formation of a fluoridated apatite from β TCP in ammonium fluoride solutions was found to increase with more F - ion added, as noted by previous researchers.

70 55 However, with the exception of very high amounts of fluoride in solution, the β TCP did not fully convert to apatite after extended reactions at physiologic temperature. This difference was attributed to surface coverage of the reactant in static versus stirred experiment conditions at higher concentrations of fluoride. In lower concentrations, this difference may be due to changes in reaction mechanism. Different crystal morphologies were noted; the shape changed from very fine, long needles in the sample without fluoride to thicker rod-shaped crystals at high fluoride concentrations. Solution chemistry analyses showed formation of calcium-deficient apatites in all samples. The method for charge balance in samples with excess fluoride could not be determined however, as no evidence for HPO 4 2- is seen in IR spectra or XRD upon heating these samples. The possibility of FPO 3 2- ion incorporating for HPO 4 2- is proposed, however its existence is purely speculative. All IR spectra show inclusion of carbonate, even the unreacted β TCP powder. The possibility of paired F -, CO 3 2- incorporation into PO 4 2- sites is possible as well. Lattice parameter calculations for the FAp show increased a-axis and c- axis values in comparison to stoichiometric FAp; this is attributed to the presence of carbonate and ammonium.

71 Chapter 4 Effects of Carbonate on HAp Formed from CaHPO 4 and Ca 4 (PO 4 ) 2 O 4.1 Abstract Carbonated hydroxyapatites were formed by reactions of mixtures of particulate tetracalcium phosphate (TetCP) and anhydrous dicalcium phosphate (DCPA) in NaHCO 3 /NaH 2 PO 4 solutions. Variations in solution chemistry were followed by determinations of ph and ion concentrations. The solids formed were analyzed by XRD and FT-IR. Rates of heat evolution were established by isothermal calorimetry. Reactions in the absence of NaH 2 PO 4 did not reach completion within 24 hours. Constitution of reactants to achieve a DCPA-to-NaHCO 3 ratio of 1, in conjunction with the presence of NaH 2 PO 4 as a buffer, was found to be optimal for formation of apatite with no remaining reactant. The amount of carbonate incorporated in this apatite was 4-5wt%. Calorimetry indicated the reaction mechanism to depend on the bicarbonate concentration in solution. The presence of NaH 2 PO 4 was found to increase the reaction rate but decrease the extent of carbonate uptake. 4.2 Introduction Hydroxyapatite is the mineral component of bones and teeth. Many substitutions can occur within this structure; those most significant physiologically include carbonate, fluoride, magnesium, and sodium. 3 The amount of carbonate present in biological apatite

72 57 varies; enamel typically contains 3.5 wt% while bone typically incorporates approximately 7.4 wt%. 3 The presence of carbonate in HAp increases its solubility. 2 Because this is of consequence for bone remodeling and caries formation, many studies related to carbonate substitution have been performed. Carbonate can substitute in two locations in the HAp structure, namely in the phosphate and hydroxyl group sites. These substitutions are known as type B and type A, respectively. Apatite containing type A substitutions usually forms at high temperatures, while type B apatite forms at low temperatures. However, type A substitution has been observed with low carbonate content in aqueous solutions. 49 Both types of substitutions occur simultaneously in biological apatites 50, although the type B substitution accounts for approximately 90% of the carbonate. 4 Most studies of type B carbonate incorporation into apatites have been carried out by relying on precipitation of particulate apatite from soluble calcium and phosphate sources. Gibson and Bonfield employed a Ca/P ratio of 1.76 during the initial synthesis step of an AB-type carbonated apatite from solutions of calcium hydroxide and phosphoric acid. 51 However it is of interest to establish substitutional chemistries of apatites that can be formed as continuous solids rather than as precipitates. Consequently, the current study focused on apatites formed by cement-type reactions. The precursor phases were tetracalcium phosphate (Ca 4 (PO 4 ) 2 O, TetCP) and dicalcium phosphate anhydrous (CaHPO 4, DCPA) and their reactions at physiologic temperature to form stoichiometric and calcium-deficient HAp have been well studied. 41,42,52,53 While the effects of carbonate additions on apatites formed in this manner have also been examined the simultaneous effects of carbonate substitution and buffering to avoid

73 58 tissue damaging ph values have not received appropriate evaluation. Thus, the objective of this study was to establish the effects of carbonate substitution on the kinetics of HAp formation and on the variations in solution chemistry. Experiments were designed to elucidate the kinetics and solution chemistry pertinent to the formation of calciumdeficient carbonated apatite via a cement-type reaction of particulate calcium-phosphate precursor phases. Depending on the molar ratios of these reactants, a range of HAp compositions can theoretically be formed. In particular if the CaHPO 4 -to-ca 4 (PO 4 ) 2 O ratio is 1-to-1, stoichiometric HAp forms. In the absence of carbonate, if the ratio is 2-to- 1, the calcium deficient composition Ca 9 HPO 4 (PO 4 ) 5 OH forms. In general, Ca (10- x)(hpo 4 ) x (PO 4 ) (6-x) (OH) (2-x) forms and x ranges from 0 to1 where the CaHPO 4 -to- Ca 4 (PO 4 ) 2 O mol ratio is (1+x)-to-1. In the presence of sodium bicarbonate, a broad range of HAp compositions can be formed. This is because Na can partially substitute for Ca, and carbonate can substitute either for phosphate or for hydroxyl. The chemistry is further complicated by the opportunity for the formation of CaCO 3 if its solubility product is significantly exceeded. In the present study, the effects of sodium bicarbonate on the reaction of CaHPO 4 and Ca 4 (PO 4 ) 2 O were examined when the Ca/P ratio was maintained at 1.8. This ratio was selected because 1.8 is near the ratio typical of mature bone apatite which is This ratio can be realized by reacting CaHPO 4 and Ca 4 (PO 4 ) 2 O at a molar ratio of 1:2. However, reacting these constituents at this ratio would not be expected to produce calcium deficient compositions unless other substituents are available. In particular, carbonate for phosphate substitution is required.

74 The full range of compositions that can form in the presence of sodium bicarbonate can be expressed by Eq CaHPO 4 + 2Ca 4 (PO 4 ) 2 O + xnahco 3 (aq) carbonated HAp (+ Na 2 HPO 4 (aq) + NaOH(aq) + CaCO 3 ) 4.1 According to LeGeros, approximately 3 moles of carbonate can be added to the structure before calcite is precipitated. 3 The products formed are anticipated to exhibit a dependence on the proportion of NaHCO 3 present. When x = 1, the limiting compositions that can theoretically form are Ca 9 Na(PO 4 ) 5 (CO 3 )(OH) 2 and Ca 9 CO 3 (PO 4 ) 5 OH + NaOH(aq). Although the formation of compositions at the Na-rich end are unlikely, this compositional range can be expressed as Ca 9 Na (1-x) (PO 4 ) 5 (CO 3 ) x (OH) (2-x) + xnaoh(aq). At values of x < 1, insufficient carbonate is available to permit only the formation of HAp. At values of x > 1, excessive carbonate is present. In these instances, a variety of compositions are possible. On the low end, vacancies in the phosphate sites would be necessary. As this is an unlikely site for a vacancy 56, the low concentration should have the most difficulty forming a carbonated apatite and completely reacting. When x > 1, more possibilities for carbonate incorporation exist, and allow for a variety of substitutions. In the event that a sodium hydroxide solution evolves, the associated ph would likely be aggressive if such a reaction occurred in vivo. Consequently, the effect of a buffer to limit the resultant ph was evaluated. Sodium phosphate monobasic was therefore proportioned to buffer the sodium hydroxide produced. The reaction anticipated is shown in Eq Ca 4 (PO 4 ) 2 O + CaHPO 4 + NaHCO 3 (aq) + NaH 2 PO 4 (aq) Ca 9 (PO 4 ) 5 CO 3 OH + Na 2 HPO 4 (aq) + H 2 O 4.2

75 60 The effects of the presence of this buffer were evaluated by determination of ph, calorimetry and X-ray diffraction analyses. The presence of NaH 2 PO 4 also provides supplemental phosphate, which would be advantageous when x < 1, as discussed above. 4.3 Materials and Methods Precursor Synthesis TetCP was synthesized by milling CaCO 3 and monocalcium phosphate monohydrate (Ca(H 2 PO 4 ) 2 H 2 O, MCPM) for 16 hours and firing the product at 1400 C for 1 hour followed by rapid quenching. TetCP phase purity was confirmed by x-ray diffraction analysis. TetCP was ground in a mortar and pestle and sieved through a 38µm mesh screen, ball milled and attrition milled to a particle size of approximately 1-2µm. Finally, the TetCP was mixed with MCPM in a molar ratio 13:2 and milled for 24 hours. All milling was performed in anhydrous heptane. The powder was stored in a sealed desiccator. Solutions of NaHCO 3 in distilled water, with NaH 2 PO 4 H 2 O as appropriate, were made just prior to their use in experiments. Solution concentrations were calculated with respect to CaHPO 4 such that the amount of NaHCO 3 was added in a molar ratio of 0.5, 1.0, or 1.5 times CaHPO 4 (i.e. x = 0.5, 1.0, or 1.5 in Eq. 4.1).

76 Solution Chemistry Variations in solution chemistry during hydrolysis in water and in sodium bicarbonate solutions, with and without buffer, were followed in two ways. First, the ph was measured using an Orion ph electrode attached to an Orion 920 digital ph meter. The slurries were placed in Nalgene beakers in a glass jacketed reaction vessel attached to a water bath maintained at 37 C. The solutions were stirred and nitrogen gas was bubbled through the slurries for the duration of the experiments. For the second set of experiments, slurries were stored in plastic vials without agitation for up to 7 days at 37 o C. Aliquots were removed from these vials and passed through 0.2µm filters and the liquids analyzed. Ca and Na concentrations were measured using atomic absorption spectroscopy, and PO 4 and C (for CO 3 ) concentrations were measured using ion chromatography. After 24 hours, the solids were filtered from the liquid, and rinsed with distilled water followed by acetone. Both sets of solution chemistry experiments were carried out at a liquid-to-solids ratio of 50-to Kinetics Isothermal calorimetry was performed to determine the kinetics of reaction in the absence and presence of buffers. The powder precursors and liquid solutions were placed in the calorimeter and allowed to equilibrate for 30 minutes before the liquid was injected onto the powder. A liquid to solid ratio of 3-to-1 was used in these samples. A datum point was collected every 3 seconds over 24 hours. The temperature was maintained at 37 o C by an attached water bath.

77 Characterization X-ray diffraction analyses were performed using a Scintag 2 diffractometer (Scintag, Inc., Sunnyvale, CA). The samples were scanned from 20 to 40 o 2θ using a step scan in 0.02 increments and 4 o per minute. Phase formation was assessed by comparing the patterns to ICDD PDF cards , 9-80, and 9-432, or TetCP, DCPA, and HAp, respectively. Infrared spectroscopy was performed using a Nicolet Nexus 670 FT-IR spectrometer using the MCT detector mode. Potassium bromide pellets with 1 wt% sample powder were pressed under vacuum. Changes with heating were assessed by thermogravimetric analysis (SDT 2960, TA Instruments, New Castle, DE) using a heating rate of 10 o C/min to 1000 o C under a nitrogen atmosphere. 4.4 Results and Discussion Solution Chemistry Figure 4.1 shows the ph variations over the first 24 hours of reaction. Reactions in water and in the NaHCO 3 solutions elevate the ph to a value near 12 regardless of NaHCO 3 proportion. There is no practical distinction between the variations in ph when reaction occurred in water, at a NaHCO 3 -to-cahpo 4 molar ratio of 0.5, and at a NaHCO 3 -to-cahpo 4 molar ratio of 1.0. The behavior at the NaHCO 3 -to-cahpo 4 molar ratio of 1.5 suggests that this solution develops some buffer capacity in that a rapid elevation in ph initiating after about 8 hours does not occur. This behavior is consistent

78 63 with the formation of Na x H (3-x) PO 4 solution. Eventually, however, elevation to ph ~12, indicative of the formation of a NaOH solution, does occur. These data indicate limited incorporation of Na into the HAp and also indicate that in vivo formation of these compositions could be problematic with regard to producing locally aggressive conditions ph H 2 O Time (hours) Figure 4.1: Variation in ph over 24 hours when a TetCP/DCP mixture with Ca/P ratio of 1.8 is reacted in NaHCO 3 solutions with NaHCO 3 -to-cahpo 4 ratios of 0.5, 1.0, and 1.5 and in water. To offset the increase in ph due to NaOH formation, hydrolysis reactions were carried out in the presence of NaH 2 PO 4 buffer. Sufficient NaH 2 PO 4 was present to react with the NaOH to produce Na 2 HPO 4. Figure 4.2 shows the variations in ph during the first 24 hours of reaction. With the exception of molar ratio NaHCO 3 -to-cahpo 4 of 0.5-

79 64 to-1, the use of this buffer limited the ph values of the bicarbonate solutions. While behavior similar to that in unbuffered solutions was observed, the changes in ph occurred more rapidly and ph maxima were lower. A ph maximum for NaHCO 3 -to-cahpo 4 of 1- to-1 was reached in 4 hours compared to 12 hours in unbuffered solution. A ph maximum in the solution where the NaHCO 3 -to-cahpo 4 ratio is 1.5-to-1 occurred in 8 hours versus approximately 24 h in unbuffered solution. The behavior when the molar ratio of NaHCO 3 -to-cahpo 4 is 0.5-to-1, is anomalous. Although reaction was accelerated when compared to that in unbuffered solution, there was no reduction in ph. If a portion of the supplementary phosphate added as a buffer is consumed in the formation of HAp, then the ph will be elevated because of NaOH formation according to Eq Ca 4 (PO 4 ) 2 O + CaHPO NaHCO NaH 2 PO 4 Ca 9 (CO 3 ) 0.5 (PO 4 ) 5.5 (OH) NaOH + 0.5H 2 O 4.3

80 ph H 2 O Time (hours) Figure 4.2: Variation in ph over 24 hours when a TetCP/DCP mixture with Ca/P ratio of 1.8 is reacted in water or NaHCO 3 solutions containing equimolar amounts of NaH 2 PO 4 added. As shown in Figure 4.3, XRD analyses of the powders indicated the formation of HAp in all instances. However, residual TetCP was observed in almost all samples that had hydrolyzed for 24 hours. Hydrolysis in water resulted in the greatest proportion of residual TetCP. This accords with prior results in that the reactions will occur in two stages. In the first, the CaHPO 4 will be consumed to produce HAp. Subsequently, TetCP will slowly hydrolyze over a period of weeks to produce HAp and Ca(OH) Residual TetCP was present after hydrolysis for 24 hours in solutions where the NaHCO 3 -to- CaHPO 4 mole ratio was 0.5 or 1.5 regardless of whether the solutions were buffered or non-buffered.

81 66 T = TetCP H = HAp H H H T T T 1.5 Buffered Relative Intensity Buffered Buffered 0.5 H 2 O θ Figure 4.3: XRD patterns of solids obtained after 24 hours of reaction in water and various NaHCO 3 or NaHCO 3 /NaH 2 PO 4 solutions. Although residual TetCP remained after 24 hours of hydrolysis in the unbuffered system when the mole ratio was 1.0, buffering allowed the hydrolysis reaction to approach completion. With buffer present, the lowered ph permitted the consumption of TetCP and formation of carbonated apatite to approach completion as shown in Eq These data, in combination with ph data in Figure 4.2, indicate an optimum proportion of constituents can produce carbonated HAp in the absence of excessively elevated ph values. Alternatively, in buffered solution having a mole ratio of 0.5, NaOH forms because the buffer is consumed.

82 67 Based on these results, solution chemistry analyses were limited to systems where the mole ratio of NaHCO 3 -to-cahpo 4 was 1.0. Both buffered and non-buffered solutions were studied and variations in solution chemistry over 7 days were compared to those in water. XRD analyses were carried out in association with solution chemistry analyses. These showed the intensities of TetCP peaks continuously decreased. When hydrolysis was carried out in water, residual TetCP was still present at the end of the experiments. When hydrolysis was carried out in non-buffered solutions, the TetCP was consumed between 3 and 7 days. TetCP was consumed within 24 hours of hydrolysis in buffered solution. The variations in ion concentrations during this period of reaction are plotted in Figures Ion concentration (mm) 1.1 PO 4 Ca 1.0 Na 0.9 CO 3 as C Time (hours) Figure 4.4: Variations in Ca 2+, phosphate ion [P i ], CO 3 2-, and Na + over 168 hours (7 days) for reaction in water.

83 68 Ion concentration (mm) PO 4 Ca Na CO 3 as C Time (hours) Figure 4.5: Variations in Ca 2+, phosphate ion [P i ], CO 3 2-, and Na + over 168 hours (7 days) reaction in solution of NaHCO 3 with NaHCO 3 -to-cahpo 4 ratio of 1.

84 69 Ion concentration (mm) 60 PO 4 Ca 50 Na CO 3 as C Time (hours) Figure 4.6: Variations in Ca 2+, phosphate ion [P i ], CO 3 2-, and Na + over 168 hours (7 days) for reaction in solution of NaHCO 3 with NaHCO 3 -to-cahpo 4 ratio of 1 with an equimolar amount of NaH 2 PO 4. Figure 4.4 shows a small Na background in the sample hydrolyzed in water. In addition, atmospheric exposure to air resulted in minor carbonate uptake. The calcium concentration remained nominally constant for 24 hours but then increased. This increase is likely associated with the continued hydrolysis of TetCP after the exhaustion of CaHPO 4. The concentration of phosphate ion decreased to below the limit of detection after 16 hours; this also indicates that HAp formation by reaction between TetCP and DCPA approaches completion over this time period. As described above, unreacted TetCP persisted throughout the duration of this experiment.

85 70 Variations in ion concentration during hydrolysis in unbuffered carbonate solution are shown in Figure 4.5. Except for a minor decrease in [P i ], the variations in ion concentrations appear to be complete at 24 hours. In spite of this, a small proportion of residual TetCP persists for 3 days. The presence of NaHCO 3 has a significant effect on [Ca] and [P i ]. [Ca] remains low throughout and is lower than when hydrolysis is carried out in water. The common ion effect of hydroxyl involving NaOH depressed the calcium ion concentration in solution. [P i ] shows a concentration spike of about 24 hours in duration; the duration of the [P i ] spike and maximum concentration reached is higher in the unbuffered carbonate solution than in water. [CO 3 ] remains constant for about 16 hours; between 16 and 24 hours it deceases to a value that remains constant thereafter. Not all the available carbonate was consumed thereby indicating that the (PO 4 )-to-(co 3 ) ratio in the solid is greater than 5-to-1 as it would be if Ca 9 Na (1-x) (PO 4 ) 5 (CO 3 )(OH) (2-x) had formed. [Na] decreased as well indicating uptake during HAp formation. [CO 3 ], [Na] and [P i ] decrease in concert. These data indicate an uptake of Na and carbonate that would be in accord with paired substitution. These data also indicate the formation of HAp of composition Ca (10-a-b) [ ] a Na b (PO 4 ) 6-y (CO 3 ) y (OH) (2-x), where [ ] is a vacancy and 2(10 a b) + b = 3(6-y) + 2y + (2-x) to maintain electroneutrality. The elevated ph condition under which the HAp is forming is inconsistent with the presence of HPO 4 groups. The variations in solution chemistry during hydrolysis of DCPA/TetCP in the solution buffered by NaH 2 PO 4 are shown in Figure 4.6. As during hydrolysis in the nonbuffered solution, [Ca] remained low throughout. Although the ph is lower in the buffered solution, the effect of elevated [P i ] contributed to the solution by the buffer

86 71 would reduce [Ca] via the common ion effect involving HAp. In addition, TetCP will hydrolyze more rapidly in a buffered solution. Excepting a minor reduction in [CO 3 ] at about 16 hours, its concentration remained constant throughout the reaction. This indicates the absence of significant carbonate uptake under this circumstance. Lack of carbonate incorporation would be expected because of high [P i ] concentration conferred by the buffer in that carbonate-substituted HAp is less stable than non-substituted HAp. There is also a reduction in [Na] during the first 12 hours of hydrolysis. This is accompanied by reduction in [P i ] that is somewhat longer in duration. Taken together, these data indicate the formation of Na-substituted, low-carbonate HAp. Mass balance equations based on the solution chemistry data were also used to calculate the compositions of the hydroxyapatites formed in those systems where the solid precursors had been consumed. The results for 7 days are shown in Table 4.1. These calculations indicate the solution without buffer incorporated the most carbonate during reaction for 7 days. A Ca/P ratio of 1.8 is expected for incorporation of 1 mole carbonate. However the phosphate-to-carbonate ratio is 6, indicating incomplete carbonate incorporation according to the idealized product in Eq While the Ca/CO 3 ratio of 11 is also somewhat higher than would be expected, slight incorporation of sodium partially accounts for this. In comparison, the buffered solution has a P/CO 3 ratio of almost 9 indicating incorporation of phosphate from the buffer. The incorporation of less carbonate during reaction in the buffered solution is expected because the elevated phosphate concentration would favor phosphate incorporation. Consistent with maximizing phosphate uptake is the formation of calcium-deficient HAp even though the Ca/P ratio of the solid reactants is 1.8. The ratio of (sodium + calcium)-to-(phosphate +

87 carbonate) is similar between the two solutions, and further indicates formation of a calcium-deficient carbonated apatite. 72 Table 4.1: Calculated ratios of ions in the solids formed after 7 days of reaction in solutions with NaHCO 3 -to-cahpo 4 ratios of 1, with and without NaH 2 PO 4 added as buffer. Ca/P P/CO 3 Ca/Na (Ca+Na)/(PO 4 +CO 3 ) Unbuffered NaHCO 3 :CaHPO 4 1: Buffered NaHCO 3 :CaHPO 4 1: Combining the above-calculated ratios with assumptions regarding the crystal chemical constraints on apatites, the composition of the apatites produced can be calculated. For that formed in non-buffered solution, we can assume that Ca + Na < 10, and PO 4 + CO 3 = 6. Thus the amount of PO 4 incorporated is approximately 5.2 moles with 0.8 mole CO 3. To maintain a Ca/P ratio of 1.8, the calculated amount of calcium is slightly higher as well, with 9.3 moles present. The final product would be Ca 9.3 Na 0.37 (CO 3 ) 0.84 (PO 4 ) 5.16 (OH) 1.8. This is equivalent to 5.3wt% carbonate in the apatite. This apatite composition would likely exhibit substantial bioactivity. For the buffered solution, the amount of phosphate incorporated is increased due to the presence of sodium phosphate. Assuming that all 6 phosphate sites are filled, these will be occupied by approximately 5.4 moles of phosphate and 0.6 moles of carbonate. The proportion of calcium is calculated to be 9.55 moles with 0.27 moles sodium. The overall HAp composition is then Ca 9.55 Na 0.27 (CO 3 ) 0.63 (PO 4 ) 5.37 (OH) 2. This is equivalent to 3.9 wt% carbonate in the apatite. Thus, the buffer increases the rate of HAp formation, reduces the amount of carbonate incorporation, and produces an apatite composition that would likely be bioactive.

88 73 The variations in ph were also measured to determine whether stirring had significant effects. During the first 8 hours, the ph values are highest in the unbuffered solution and lowest in water. The unbuffered solution plateaus around 9.75 for the first 16 hours, then plateaus at 12 for 24 hours and beyond. These variations are similar to those observed in Figure 4.2, except over a longer period of time. The buffered solution reaches a maximum at 16 hours and decreases, variations also similar to those shown in Figure 4.2. Excepting an anomalous data point at day 3, ph variations are similar to those shown in Figure 4.2. Taken together, these data suggest that ph changes in stirred and quiescent slurries in this system are similar thus indicating it is appropriate to generalize these observations to pastes Kinetics Calorimetric analyses established the heat evolution characteristics during the first 24 hours of HAp formation in water or in a NaHCO 3 solution where the NaHCO 3 :CaHPO 4 molar ratio is 0.5, 1.0 or 1.5. Heat evolution curves during the first 12 hours are shown in Figure 4.7. Hydrolysis in water shows an initial peak associated with mixing of the reactants followed by a second peak that initiates after about 3 hours. There is a period of low heat evolution between these distinct peaks. The presence of NaHCO 3 changes the heat evolution characteristics during the first several hours. Heat evolution in the NaHCO 3 solutions occur at descending rates that are proportional to its concentration. When reaction occurs at a NaHCO 3 :CaHPO 4 molar ratio of 1.5, no second peak appears. These data indicate a mechanistic shift that depends on the proportion of NaHCO 3

89 74 present. Figure 4.8 shows the total heat evolution curves for the first 12 hours of reaction. In spite of the mechanistic differences observed in Figure 4.7, there is no systematic trend to the total heat evolution. Previous authors have observed a decrease in crystal size with increasing bicarbonate concentration in solution. 58,59 Thus the lack of a second reaction peak at higher concentrations of carbonate is likely related to this decreased growth W/g H 2 O Time (hours) Figure 4.7: Rates of heat evolution during the reaction of TetCP and DCP with various NaHCO 3 solutions and water.

90 Total Heat (W*h/g) H 2 O Time (hours) Figure 4.8: Total heat evolved during the formation of carbonated apatite solutions of NaHCO 3 or water. Figures 4.9 and 4.10 compares the heat evolution characteristics during the first 12 hours of hydrolysis in buffered and unbuffered solutions where the CaHCO 3 :CaHPO 4 molar ratio is 0.5 or 1.0. Although the time to complete reaction differs substantially, the bases for these differences are not apparent in the heat evolution characteristics during this period. Rather, these data suggest that rates of hydrolysis, during the period when the conversion kinetics are considered to be diffusion controlled, exhibit compositionally dependent differences. In both cases, this may be due to addition of the phosphate group provided by the buffer. The phosphate group may not only lower reaction ph by

91 decreasing NaOH formation, but also permit congruent dissolution of the DCPA increasing reaction rate Rate of Heat Evolution (W/g) buffered Total Heat Evolved (W*h/g) Time (hours) Figure 4.9: Rates of heat evolution (left axis) and total heat evolved (right axis) for NaHCO 3 solutions in NaHCO 3 -to-cahpo 4 ratios of 0.5 and solution with the corresponding equimolar amount of NaH 2 PO 4.

92 Rate of Heat Evolution (W/g) buffered Total Heat Evolved (W*h/g) Time (hours) Figure 4.10: Rates of heat evolution (left axis) and total heat evolved (right axis) for NaHCO 3 solutions in NaHCO 3 -to-cahpo 4 ratios of 1.0 and solution with the corresponding equimolar amount of NaH 2 PO Characterization The infrared spectra were obtained for solid samples formed after hydrolysis. All spectra are characteristic of calcium phosphates, exhibiting a broad band from 1100 to 1000 cm -1. The sample hydrolyzed in water contains carbonate due to reaction in air and incorporation from the milling process, as even unreacted powder shows weak carbonate bands. As expected, HAp formed in carbonate solutions contain more carbonate as indicated by increased intensities of peaks at 873, 1423, and 1473 cm -1 in comparison to powder reacted in water. The locations of these bands indicate formation of

93 78 predominantly type-b carbonated apatite. The high ph of the preparation conditions likely precludes significant type-a incorporation 61. However, Barralet et al. have seen some degree of type-a substitution at low carbonate concentrations even when synthesis was carried out at a high ph 58, and this may explain a weak peak at 1558cm -1 seen in the present study. Thermogravimetric analyses were performed on the aforementioned 7 day samples. The weight loss above 400 o C in each apatite corresponded closely with the calculated carbonate uptake. For apatite formed in unbuffered carbonate solution, the loss was 6.5 wt%. Apatite formed in the buffered carbonate solution lost 4 wt%. Apatite formed in water lost approximately 2 wt%. As these values are higher than the amount calculated from the solution chemistry data, some of the weight loss must be due to other species. Loss of lattice water at temperatures above 400 o C has been proposed to account for this discrepancy Conclusions Formation of essentially phase-pure, calcium-deficient carbonated apatite was realized by combining tetracalcium phosphate and dicalcium phosphate at Ca/P ratio of 1.8 and reacting in sodium bicarbonate solutions. Use of a sodium dihydrogen phosphate buffer to counteract NaOH formation was found to decrease the ph and increase the reaction rate, but to decrease the carbonate content of the apatite formed. Optimum reaction occurred when the CaHPO 4 :NaHCO 3 ratio was 1; at values above and below this, TetCP persisted regardless of whether the system was buffered. Calorimetry

94 79 indicated more rapid reaction with lower CaHPO 4 :NaHCO 3 ratios. Similarly, the presence of phosphate buffer appears to accelerate the reaction by simultaneously decreasing the reaction ph and providing phosphate groups for the formation of a more stable apatite phase.

95 Chapter 5 Effect of Dense Aggregates on Mechanical Properties of Calcium Phosphate Cements 5.1 Abstract The effects of filler materials on the mechanical properties of a calcium-deficient hydroxyapatite (CDHAp) cement were studied. Calcite (CaCO 3 ) and dense aggregates of HAp and β-tricalcium phosphate (β-tcp) were used as filler materials. Inclusion of aggregates did not reduce compressive strength in comparison to the cement alone until approximately 50 wt% aggregate was added. The dense aggregates were not found to decrease mechanical properties of the cement, however use of CaCO 3 resulted in reduction of strength. The use of aggregates with a cement formed in a sodium bicarbonate liquid was also examined. NaHCO 3 decreased the compressive strength of the cement in relation to water, however all aggregates used were found to increase the compressive strength in comparison to the cement alone. The microstructures of the cements formed were observed using scanning electron microscopy, and differences in the interfaces between aggregate particles and the cement were noted. 5.2 Introduction Calcium phosphate cements have been formulated as analogs of natural bone. The cements can be molded and shaped or injected, allowing more precise tailoring to bone

96 81 defects. The bone analogs are resorbed by the body over time. However, unless other solids are present, the size of a defect that can be filled by calcium phosphate-based cements is limited. This is for two reasons. First, heat evolution may result in local temperatures that could be damaging to surrounding tissues. Second, is that apatite formation is accompanied by chemical shrinkage. The volume of HAp formed is smaller than the volume of the solid and liquid reactants. This chemical shrinkage can lead to a weak bond with surrounding tissue or to cracking. The presence of fillers minimizes these effects. However, it is recognized that the incorporation of fillers reduces the strength of cements. The incorporation of dense aggregate, while reducing strength, is the strategy commonly used to control dimensional changes and limit thermal excursions. This study evaluated the presence of CaCO 3, β-tcp, and HAp aggregates on the compressive strengths of HAp matrix composites. Carbonate incorporation into hydroxyapatite occurs in native bone tissue and increases solubility. Thus carbonate incorporation into calcium phosphate cements is expected to improve the resorbability of bone analogs. However, carbonate incorporation has been shown by some investigators to decrease the mechanical strength of the formed HAp. 54,55 Others have found no decrease or some improvement when using carbonate with an accelerant, Na 2 HPO Consequently methods to improve the mechanical properties of carbonated HAp bone cements are desirable. One such technique that has been investigated is the use of dense granules in a calcium phosphate cement matrix. An early study by Ginebra et al. noted increased compressive strength when sintered HAp was added, however no details about particle size were given. 63 Two other studies used dense tricalcium phosphate

97 82 (TCP) granules in calcium phosphate bone cement systems; both noted improved workability at a higher powder to liquid ratio. 64,65 Oh et al. used µm β-tcp granules in a MCPM and calcium sulfate hemihydrate (gypsum) cement system and found drastic improvements in compressive strength. 64 Cho et al. used dense α-tcp particles, sieved to 45-90µm, in a TetCP/DCPA system to form CDHAp. While no improvements in mechanical strength were noted in this system, the addition of the granules was not deleterious at high powder to liquid ratios, and produced a more workable cement. 65 Guo et al. also noted improved mechanical properties due to residual, unreacted TetCP in a TetCP/DCPA cement system. 66 Constantz et al. used 50 to 100 µm calcite aggregate particles and found no change in compressive strength using two calcium phosphate cement systems. 67 Thus the there were two goals of this work. The first was to study the mechanical properties of CDHAp cement formed from TetCP and DCPA with dense aggregate or CaCO 3 additions. The second was to increase compressive strength when CO 3 2- was incorporated using NaHCO 3 solution by use of aggregates. 5.3 Materials and Methods Tetracalcium phosphate (TetCP) was synthesized by reaction of CaCO 3 and Ca(H 2 PO 4 )2. H 2 O (monocalcium phosphate monohydrate, MCPM). The two powders were milled in a 3-to-1 ratio and fired at 1400 o C for 1.5 hours before quenching to room temperature. Phase purity was confirmed by x-ray diffraction (Scintag, Sunnyvale, CA). The TetCP was then ground by hand, passed through a 38µm screen mesh, ball milled,

98 83 and attrition milled. The powder was then mixed with MCPM in a 2-to-1 ratio and ball milled in heptane for 24 hours. This CDHAp precursor powder was stored in a tightly sealed desiccator with new desiccant. Dense aggregate particles were prepared by sintering β-tcp or HAp powders. β- TCP was synthesized using CaCO 3 and Ca 2 P 2 O 7 heated to 900 o C. The calcium pyrophosphate (Ca 2 P 2 O 7 ) was processed by dehydrolysis at 500 o C of CaHPO 4, which had been synthesized from CaO and H 3 PO 4. To increase the stability of β-tcp at temperatures high enough for sintering, 1wt% MgO was added. The HAp powder (calcium phosphate tribasic, Alfa Aesar, MA) was mixed with 5wt% acryloid binder dissolved in acetone. After initial processing, the powders were die pressed uniaxially at psi (175MPa). The compacted discs of β-tcp were heated on platinum foil at 1300 o C for 3 hours and slowly cooled. The discs of HAp followed a slow heating rate to allow binder burnout and were then rapidly heated to 1300 o C for 1 hour. The β-tcp and HAp pellets were calculated to have 94% and 93% theoretical densities, respectively, by measuring the weight and dimensions of six pellets. Diametral compression testing yielded average tensile strengths of 8.7 and 5.9 MPa for the β-tcp and HAp pellets, respectively. Aggregate particles of CaCO 3 were obtained locally. XRD showed the CaCO 3 was in the form of calcite with some quartz and ankerite (Ca(Fe,Mg)(CO 3 ) 2 ) present. All aggregates were ground and then sieved between 63 and 38µm mesh sieves. Mechanical testing samples were prepared by mixing CDHAp precursor powders in a 2.75-to-1 or 2.5-to-1 powder-to-liquid ratio with distilled, de-ionized water. After the paste was formed either 10, 25, or 50wt% aggregate, measured as a fraction of the

99 84 precursor material, was added with the exception of one control set each for water. Thus, a 10wt% aggregate addition would be 0.3g to 3.0g of CDHAp precursor. All mixing was performed on a glass plate using a metal spatula. Once the samples were well combined, the paste was pressed by hand into cylindrical molds measuring 6.35mm in diameter and 12.7mm in height. The molds were then set in a humidified atmosphere at 37 o C for 24 hours. Generally, five samples were made for each set of conditions. Subsequently, another similar test using a 0.75M sodium bicarbonate solution was carried out at a powder-to-liquid ratio of 2.5:1. Two sets were made without aggregate additions, using either water or NaHCO 3 solution, and the other sets used only β-tcp and CaCO 3 aggregates with NaHCO 3 solution. The samples obtained were moderately dry, and aggregate was visible throughout the samples. Mechanical testing was performed on an Instron 4202 testing machine using a cross-head speed of 0.3 mm/min. The samples were loaded until failure through the center of cylinder. The maximum load was recorded, and thus the maximum stress could be calculated. The samples were analyzed for statistical differences using Power Analysis and Sample Size NCSS/GCSS software. One-way ANOVA tests were performed with α = 0.05, and differences were noted using multiple comparison tests. To assess the potential bioactivity of the aggregates, 10 or 25% (0.05g or 0.125g) aggregate was combined with 0.5g CDHAp precursors and reacted in 5 ml water for 1 day and 6 weeks at 37 o C. After the allotted time period, the samples were washed with DI water and flushed with acetone to stop the reaction. The samples were analyzed by XRD and compared.

100 85 Samples from mechanical testing were also viewed using SEM (Hitachi, Japan). The samples were gold coated by sputtering and viewed using a 5kV accelerating voltage. 5.4 Results Mechanical Properties Variation of Aggregate Quantity Figure 5.1 displays the average compressive strengths obtained with varying additions of dense HAp aggregate. The optimum powder to liquid ratio for this CDHAp precursor mixture was determined to be 2.75:1, thus the samples were prepared at this ratio. Both 10% and 50% aggregate additions are statistically significant (p < 0.05) in comparison to the samples with no aggregate. An addition of 25% did not have a major effect on the compressive strength. Because 50% addition negatively affects the observed strength, this amount was excluded from further testing.

101 86 30 * Compressive Strength (MPa) No aggregate 10% HAp 25% HAp 50% HAp * Figure 5.1: Average compressive strengths of CDHAp cement with dense HAp aggregate compared to a group with no aggregate addition. Significant differences (p < 0.05, n = 4-6) from the aggregate-free group are indicated by *. Error bars represent the standard deviation for each set of samples Variation of Aggregate Type Figure 5.2 shows the average compressive strengths using a variety of aggregates. Addition of CaCO 3 significantly (p < 0.05) reduces the strength regardless of proportion. The effects of HAp or β-tcp at either amount are not statistically significant. Contrary to the previous set of experiments, only a 25% HAp addition had a notable increase in strength over the set without additions. Variations between the batches of TetCP

102 87 precursors may account for the changes observed with aggregate addition. The precursor powders used in this experiment were synthesized at different times than those used in the previous study. Thus, this set of experiments was performed at a powder-to-liquid ratio of 2.5:1. Compressive Strength (MPa) * H 2 O 10% CaCO 3 no aggregate * 25% CaCO 3 10% β-tcp 25% β-tcp 10% HAp 25% HAp Figure 5.2: Average compressive strengths of samples with or without aggregates. A statistically significant difference (p < 0.05, n = 5) from the group without aggregate is indicated by *. Error bars represent standard deviation for each set of samples Effect Aggregates and NaHCO 3 solution Figure 5.3 shows the compressive strengths measured when aggregate was used with NaHCO 3 solution as the liquid. When a carbonate solution was used as the liquid

103 88 instead of water, the average compressive strength of the samples without aggregate decreased by almost 2 MPa. Although this difference was not found to be statistically significant, it is notable. All additions of aggregate, except 25% CaCO 3 additions, are significantly stronger (p < 0.05) than the samples without aggregates. * 20 Compressive Strength (MPa) H 2 O NaHCO 3 10% β-tcp25% β-tcp 10% CaCO 3 25% CaCO 3 no aggregate no aggregate Figure 5.3: Compressive strengths observed when NaHCO 3 solution is used with and without aggregate in comparison to water without aggregate. Statistically significant differences (p < 0.05, n = 5) are denoted with *. Error bars represent standard deviation for each set of samples.

104 Aggregate Reactivity The samples reacted in water for an extended time period show some change in the amount of aggregate present in comparison to 1 day. This is most noticeable for the samples containing 10% aggregate, but occurs to some degree in the samples containing 25% aggregate as well. The XRD patterns for these samples are shown in Figures The relative intensities of the aggregate peaks to the HAp peaks in CaCO 3 and β-tcp decrease appreciably with both amounts of aggregate present. The amount of HAp aggregate remaining is difficult to judge due to overlap of the peaks, however the sharpness seen at all times except for 6 weeks with 10% aggregate is an indication of remaining particles. The slow decrease in β-tcp and HAp are expected due to the high density and low reactivity of those phases. In particular, the use of magnesium to stabilize the β-tcp at high temperatures will limit its ability to form apatite. 3 The increased conversion of CaCO 3 is expected, and should form a bioactive carbonated apatite. However, Constantz et al. did not observe reaction of calcite aggregates in vivo 67 ; this may be due to the particle sizes used. Whether the increase in HAp seen over time can be attributed to dissolution of the aggregates with reprecipitation of more apatite phase or simply due to an increase in CDHAp crystal size due to the longer time period for reaction is uncertain.

105 90 T = TetCP C = CaCO 3 D = DCPA H = HAp 10% CaCO 3 H O = Other phase present with CaCO 3 H 25% CaCO H 3 No aggregate H H H 6 weeks H H Relative Intensity T O D O C O T C D O T T T O O D O C O CT T D T O T T 6 weeks 1 day Unreacted θ Figure 5.4: XRD patterns of CDHAp with CaCO 3 aggregate added at 10 and 25% in comparison to CDHAp with no aggregate.

106 91 H = HAp B = β-tcp 10% β-tcp H H H H 25% β-tcp H H H H No aggregate 6 weeks Relative Intensity B B B B B B B B B B B B 6 weeks 1 day Unreacted θ Figure 5.5: XRD patterns of CDHAp with dense β-tcp aggregate added at 10 and 25 wt% in comparison to CDHAp with no aggregate.

107 92 H = HAp 10% HAp H H H H 25% HAp H H H H H No aggregate 6 weeks Relative Intensity 6 weeks 1 day Unreacted θ Figure 5.6: XRD patterns of CDHAp with dense HAp aggregate added at 10 and 25 wt% in comparison to CDHAp with no aggregate Microstructural Analysis Scanning electron microscopy images reveal minor differences between the aggregates and between cements formed in water and bicarbonate solution. Figure 5.7 shows the interface between a CaCO 3 aggregate and the surrounding CDHAp matrix while Figure 5.8 shows the interface between a sintered HAp aggregate and CDHAp. A coherent interface with HAp aggregate appears to develop with the surrounding CDHAp while a gap is present between the CaCO 3 needles and the CDHAp phase. This is one of the reasons for the lower compressive strength observed when aggregates of CaCO 3 are

108 93 used in the system with water. The HAp particles, in addition to being denser, appear to more readily form a bond to the CDHAp phase. These observations are unexpected as CaCO 3 should interact more strongly with the CDHAp phase. However, this may occur over an extended time period, while the HAp provides immediate locations for CDHAp nucleation and growth. CaCO 3 CDHAp 3µm Figure 5.7: Interface between CaCO 3 crystals and CDHAp. CDHAp HAp 2µm Figure 5.8: Interface between an aggregate of HAp and surrounding CDHAp.

109 94 Minor differences are apparent in the CDHAp formed in water versus that formed in sodium bicarbonate solution. Figures 5.9 and 5.10 show the microstructure developed in CDHAp formed in water and NaHCO 3 solution. The two are similar, however HAp formed in carbonate is expected to visibly increase porosity as seen in previous microstructural studies. 54,55 Increasing porosity has been associated with decreased mechanical strength of calcium phosphate cements µm Figure 5.9: Microstructure of CDHAp formed with water.

110 95 10µm Figure 5.10: Microstructure of CDHAp formed with carbonate solution. 5.5 Discussion The use of dense aggregate was generally not found to adversely effect compressive strength. In some regards, this is due to the ability to increase powder-toliquid ratio while retaining workability. However, in this study the powder to liquid ratio of the cement was held constant during mixing of the precursor phase, and added aggregate was not included in this calculation. Addition of larger quantities of aggregate, approximately 50wt%, resulted in a decrease in compressive strength in comparison to the cement alone. Although this amount of aggregate present should not be detrimental to forming links between the

111 96 CDHAp cement, high aggregate content invariably lowered mechanical properties. The total powder to the amount of liquid used in comparison to the other aggregates being larger, thus potentially resulting in friable samples, cannot explain this change. When this amount of aggregate was used in other experiments to determine optimum powder-toliquid ratios, it was also found to decrease compressive strength, even at lower powderto-liquid ratios. The type of dense aggregate added was not observed to have major effects on the compressive strength of the samples. These aggregates have higher mechanical strength than the CDHAp cement material, thus an increase in compressive strength would not be unexpected. Although the calcium carbonate exhibited better phase transformation when reacted with CDHAp for 6 weeks, the lowered mechanical properties observed with this aggregate are undesirable for use as a bone filler phase. Additionally, further lowering of the strength is expected as CaCO 3 dissolves over time, leaving relatively large pores. A decrease in mechanical strength when solutions of carbonate are used as the reaction liquid for calcium phosphate cements has been observed previously. 54,55 The lack of statistically significant change observed between samples formed in NaHCO 3 and water in this experiment can be attributed to the relatively small amount of carbonate incorporated. At 0.75M, approximately 1.9wt% carbonate should be present in the CDHAp formed. Miyamoto found that the mechanical properties of a TetCP/DCP cement did not decrease drastically until approximately 4 wt% of carbonate was added. 54 In the carbonate setting liquid system, CaCO 3 was found to be as effective as a dense aggregate in increasing compressive strength. The aggregates most likely increase

112 97 strength by acting as barriers against crack propagation; this has been observed with smaller TetCP particles in other systems. 66 The period between the set of experiments performed with water and bicarbonate solutions was approximately 4 months. After this time, the compressive strength of CDHAp formed in water decreases by approximately 7 MPa. This raises issues with the storage stability of these precursors. The difference in compressive strength is likely due to reaction of the precursors with ambient moisture, resulting in a thin layer of HAp. This decreases the reaction once the precursor particles are intentionally hydrated. This has no overall effect on the formation of HAp within 24 hours, and generally no large changes are noted in ph or calorimetric measurements, but a decrease in compressive strength is always observed. Whether storage in a desiccator under vacuum or in a low-temperature drying oven would be more likely to prevent this change is uncertain. 5.6 Conclusions Use of a dense aggregate was observed to increase the compressive strength of a CDHAp cement formed in sodium bicarbonate solution by a significant amount. Dense aggregates did not have undesirable effects on the compressive strengths of CDHAp cements formed in water, however CaCO 3 was found to have a negative effect. The dense aggregates may be more bioactive in vivo than most dense ceramics. This conclusion is based on the decreased relative amount of aggregate phase after hydration in water for 6 weeks in comparison to reaction for 1 day.

113 98 In general, storage stability of the CDHAp precursor powder becomes an issue over time, and batch-to-batch variations are problematic. With the difficulty presented in obtaining useable amounts of the dense aggregates, this system is difficult to produce and replicate in quantities sufficient to achieve meaningful comparisons between multiple groups. However, the gains seen using aggregates merit further study of the aggregates or similar materials.

114 Chapter 6 Effects of Phosphonates on Formation of Hydroxyapatite from Solid Precursors 6.1 Abstract Phosphonates are known inhibitors of calcium phosphate formation and resorption, and have many uses throughout industry and medicine. The effects of phosphonates on the formation of hydroxyapatite via a cement-like reaction between two precursor materials were studied. The kinetics of reaction were followed using isothermal calorimetry and the phase development was assessed using x-ray diffraction. Additionally, variations in solution chemistry due to presence of the phosphonates are noted. Bisphosphonates were found to be the most effective compounds for disrupting the reaction of the precursors to form HAp. Short chain length (2 and 3 carbon atoms between phosphonate groups) phosphonates were found to be moderately effective, while a monophosphonate was found to accelerate reaction at low concentrations. These differences are attributed to efficacy of the phosphonates on preventing dissolution of the precursor phases and inhibiting nucleation of HAp by adsorption.

115 Introduction Bisphosphonates are of much interest in medicine due to their ability to inhibit both dissolution and crystallization from aqueous solutions. Bisphosphonates are characterized as having two phosphate ligands attached to a single carbon atom and are formally termed geminal-bisphosphonates. 69 Bisphosphonates were first noted due to their structural similarity to pyrophosphate. Pyrophosphate is known to inhibit dissolution and precipitation of calcium phosphate phases. 70 As the P-O-P bond is readily hydrolyzed at certain locations in vivo and is not an effective inhibitor of resorption, attention was turned to the bisphosphonates, which contain a P-C-P bond that is more stable in biological systems. 71 While the P-C-P bonds in bisphosphonates are considered critical for interaction with bone mineral 72, other effects are associated with the overall structure and composition of the bisphosphonate. Consequently bisphosphonates interfere with resorption of bone and with calcifications in vivo. Studies have shown that the primary mechanism for the inhibition of calcium phosphate formation is phosphonate adsorption on surface sites occupied by calcium However, the influence of bisphosphonates on bone resorption is due to more biological effects and less to physical effects on the bone mineral. 76 The two side groups that can attach to the carbon have a large effect on the inhibitory and anti-resorption properties of the bisphosphonates. Thus, the R 1 side group is a hydroxyl group in all clinically relevant bisphosphonates as it has been observed to increase the affinity of the molecule to bone mineral. 77 The R 2 group can have some effect on the affinity of the bisphosphonate for calcium phosphate phases 78, but is primarily responsible for the biologic effects of the drug. 76 Each successive generation of

116 101 bisphosphonate drugs contain R 2 group variations that enhance potency. 78 Currently, bisphosphonates are used clinically to treat bone resorption problems such as osteoporosis and Paget s disease, and also as markers for bone lesions and calcification inhibitors. 79 Phosphonates in general contain phosphate groups attached to a carbon atom. These can include phosphonates such as diethylenetriaminepenta (methylene phosphonic acid), which has five phosphate groups, to methane phosphonic acid, which has only one phosphate group. When phosphate groups are positioned at either end of a carbon chain, they are most appropriately termed bisphosphonates. 69 The term phosphonate will be used here for simplicity when discussing these non-geminal bisphosphonates. Phosphonates are frequently used in industrial water systems to prevent scaling, as even at concentrations under that required for chelation, the molecules are effective at preventing precipitation of metal salts. 80 Industrial uses of phosphonates also include oil field applications and desalination plants. The phosphonates prevent crystal growth by adsorbing onto the crystal surface. The large amount of charge associated with the phosphonate prevents normal growth of the crystal. Adsorption can also affect dissolution, and soluble or insoluble complex formation can occur as well. The aim of this work was to determine effect of chain length, or distance between phosphate groups, on the efficacy of phosphonates to influence HAp formation. The hypothesis was that formation of 6-member rings with calcium would be more effective at HAp growth inhibition than those requiring 7 or 8 member rings. Phosphonates with P- C-C-P bonds and single P-C bonds have been noted as less or not at all effective at

117 preventing HAp formation or dissolution in in vitro experiments. 72, Longer chain phosphonates have been made biologically relevant by adding keto groups on the carbon atoms adjacent to the phosphate groups; these are known as bisacylphosphonates. 81 The system used in this study is complex, and requires dissolution of calcium phosphate phases and subsequent precipitation of HAp. While this adds a degree of difficulty to interpreting the results, this system is applicable to potential use in bone cements. 6.3 Materials and Methods Synthesis of calcium phosphate precursors Tetracalcium phosphate was synthesized by milling together calcium carbonate (Osram Sylvania, PA) and monocalcium phosphate monohydrate (MCPM, Ca(H 2 PO 4 ). 2 H 2 O, FMC, PA) in a 3 to 1 molar ratio for 16 hours. The slurry was then filtered, dried, heated to 1400 o C for 1 hour, and rapidly quenched to room temperature. Phase purity was determined by x-ray diffraction. The TetCP was then ground, passed through a 63µm sieve, ball milled for 24 hours and attrition milled for 8 hours. All milling was performed in anhydrous heptane (Aldrich). The final particle size averaged 1-2µm as measured by SEM. The TetCP was then mixed with MCPM in a 2:1 molar ratio and milled for an additional 24 hours yielding a CDHAp precursor powder, with a composition of 2 CaHPO 4 (DCPA) to 1 TetCP, that was dried and stored in a desiccator under vacuum.

118 Phosphonates Five different phosphonates were used for this study and are listed in Table 6.1. The phosphonates will hereafter be referred to by the number assigned in Table 6.1. Phosphonates #1-#4 were obtained from Alfa Aesar, while #5 was from Avocado Research Chemicals. Table 6.1: Names, chemical formulas, and structures of the phosphonates used in this study. Number & Abbreviation 1 PDP 2 MDP 3 EDP 4 HEDP 5 MP Name Formula Molecular Weight Propylene [P(O)(OH) 2 CH 2 ] 2 CH g/mol diphosphonic acid Methylene diphosphonic acid 1,2-Ethylene diphosphonic acid Ethane-1- hydroxy-1,1- diphosphonic acid Methane phosphonic acid [P(O)(OH) 2 ] 2 CH 2 [P(O)(OH) 2 CH 2 ] 2 CH 3 C(OH)(PO 3 H 2 ) 2 CH 3 P(O)(OH) g/mol 190 g/mol 206 g/mol 96 g/mol Structure Kinetics Isothermal calorimetry was performed using a system previously described. 82 Briefly, 1.5g of CDHAp precursor powder was placed in a gold plated copper cup along

119 104 with 1.5mL of a phosphonate solution or water. After allowing time for the reaction chamber to re-equilibrate, the liquid was injected onto the powder. These reactions were followed for 24 hours, with a datum point collected every 3 seconds. Two concentrations of phosphonate solution were used; these corresponded to 5% and 0.3% phosphonate to powder precursor by weight Solution Chemistry Variations in solution chemistry were followed using ph and measurement of calcium and phosphate ion concentrations. For ph measurements, solutions of phosphonate were prepared in a 50mL Nalgene beaker using 25mL of distilled water and the desired amount of phosphonate. The beaker was placed inside of a double walled glass reaction beaker attached to a water bath at 37 o C. Nitrogen was bubbled into the liquid throughout the experiment. 0.5g of CDHAP precursor powder was added, yielding a liquid to powder ratio of 50:1. The reaction was followed for 24 hours using an Orion 940 ph meter. After reaction, the solids were filtered, flushed with acetone to stop reaction, and dried for later XRD analysis. Four concentrations of each phosphonate were used in the ph analyses. These concentrations were equivalent to 0.1, 0.5, 2.5 and 5 percent phosphonate to precursor material by weight. For calcium and phosphate ion measurements, experiments were performed using a concentration equivalent to 5 wt% phosphonate-to-precursor ratio. The materials were reacted in plastic vials, using 15mL of phosphonate solution to 0.3g of precursor powder, and stored at 37 o C. At the time of analysis, the liquid was collected and passed through a

120 0.2µm filter and analyzed for Ca 2+ by atomic absorption spectroscopy and PO 4 3- by ion chromatography. The solids were saved for XRD analysis as well Characterization X-ray diffraction analyses were performed using a Scintag 2 diffractometer (Scintag Inc., Sunnyvale, CA). The powder samples were scanned from o 2θ using a scan rate of 4 o per minute and a step size of 0.02 o. The patterns were processed using 5 or 10 point adjacent averaging to smooth. Scanning electron microscopy (Hitachi 3500N, Japan) was used alongside energy dispersive spectroscopy (Princeton Gamma Tech, NJ). The samples were sputter coated with gold for 30 seconds prior to analysis. 6.4 Results Kinetics The heat evolved during reactions in the phosphonate solutions or in water with hydroxyapatite precursors at high and low concentrations is shown in Figure 6.1 and Figure 6.2. The XRD patterns of solids obtained from these reactions are shown in Figure 6.3 and Figure 6.4. At the high concentration, the weak reaction peaks seen in calorimetry curves indicate longer reaction times, hence the XRD patterns are shown for samples reacted for 6 days. At low concentration, the samples were analyzed using XRD after 24 hours.

121 #5 #1, #2, #3 #4 Heat Evolved (W/g) H 2 O #4 # #3 #1 #5 H 2 O Time (hours) Figure 6.1: Rate of heat evolution during the reaction of 5wt% phosphonate solution with CDHAp precursors. Inset shows first half hour after injection of liquid.

122 H 2 O #4 # #3 #5 Heat Evolved (W/g) #5 # H 2 O #3 #4 # Time (hours) Figure 6.2: Rate of heat evolution during reaction of 0.3wt% phosphonate solution with CDHAp precursors. Inset shows first hour.

123 108 T = TetCP D = DCPA H = HAp H D D H T T T D D T T T T #5 Relative Intensity #4 #3 #2 # Figure 6.3: XRD of solids after reaction for 6 days in 5wt% phosphonate solution. 2θ

124 109 T = TetCP D = DCPA H = HAp H H Relative Intensity D T T D T #5 #4 #3 #2 # Figure 6.4: XRD patterns after 24 hours of reaction in 0.3wt% phosphonate solution at 1:1 powder-to-liquid ratio. 2θ Bisphosphonates Initial calorimetric peaks are attributed to mixing reactions wherein initial dissolution and adsorption of species from solution onto solid surfaces occur. When compared to that in water, the mixing peaks of the heat evolution curves for the 5wt% concentration of bisphosphonate solutions are much larger in intensity and in duration. The durations of these peaks are indicative of the occurrence of chemical reactions during this period. Subsequent to the heat evolution associated with these events, no further heat

125 110 is evolved after 30 minutes mixing; after 6 days of reaction TetCP and DCPA precursors remain in the presence of both bisphosphonate solutions as seen in Figure 6.4. When the hydrolysis reactions are carried out in solutions in 0.3wt% phosphonate solutions, differences in calorimetric behavior can be observed. The initial heat evolution behaviors in water and phosphonate #4 solution are similar. Although the bisphosphonates initially appear to be similar to water, having 2 peaks within the first half hour followed by an extended reaction peak to approximately 8 hours, the rate of reaction in phosphonate #2 is much lower. This is confirmed by the analysis of the solids present and the XRD patterns in Figures 6.3 and 6.4, which show that more TetCP and DCPA remain in #2 solution after 24 hours than in #4 solution. Alternatively, the solution chemistry and ph data show phosphonate #4 solution to inhibit more effectively over long time periods. This discrepancy is the result of the low liquid-to-solid ratio utilized in the calorimetry experiments as compared to the higher ratio used in solution chemistry experiments. Even though concentrations are equivalent, the absolute amounts of bisphosphonates present at the higher liquid-to-solids ratios are larger. Thus, phosphonate adsorption onto the surfaces of the precursors and preventing, or inhibiting, dissolution can continue to a greater extent in solutions having a higher liquid-to-solids ratio Phosphonates In comparison to the bisphosphonates, the heat evolved at all times using phosphonate #1 or #3 at a concentration of 5 wt% is negligible. Once again, the initial reaction peaks are much higher than water for both phosphonates and are approximately

126 111 the same value as bisphosphonate #2. In the case of the bisphosphonates, some heat is evolved over the first 6 to 8 hours while phosphonates #1 and #3 show no further heat evolution over the course of 24 hours. XRD patterns of all samples show continued persistence of the TetCP and DCPA phases after 6 days of reaction. At the lower concentration of phosphonate, the differences between the bisphosphonates and phosphonates can be observed. Phosphonates #1 and #3 have extended growth peaks in comparison to water and the bisphosphonates. However the growth peak is complete by 12 and 20 hours for phosphonates #1 and #3, respectively, and the XRD patterns for these samples show complete formation of HAp by 24 hours. Comparison of the behavior suggests that phosphonates act in a generally similar fashion to bisphosphonates, but less effectively Monophosphonate When a solution of phosphonate #5 is used, the reactions do not follow the trends of the bisphosphonates or phosphonates studied. At high concentration the rate of heat evolution curve has a relatively sharp second peak around 2 hours, followed by a broad band over the next 22 hours. XRD patterns obtained after 1 day of reaction indicate formation of HAp with a large amount of DCPA and some TetCP remaining. After 6 days, only HAp and DCPA are visible. At the lower concentration, methanephosphonic acid accelerates the reaction. An initial mixing peak is present, however the secondary peak seen at about 15 minutes in water and the other phosphonates does not occur. A second peak occurs at about 45 minutes after injection of the liquid, and is similar to that

127 112 seen at the high concentration. The main growth peak is larger in intensity and shorter in duration than reaction in water. This indicates more rapid formation of HAp in this phosphonate solution. At the higher concentration, formation of HAp is caused by rapid dissolution of TetCP due to the initial low ph imparted by the phosphonate, with minimally effective adsorption onto this precursor. However, this variation in ph is not observed initially at the low concentration, thus the accelerated formation of HAp is unique. Rapid conversion to HAp using H 3 PO 4 solutions has been observed, but, as will be further shown in section , the monophosphonate system does not require an excessive decrease in ph to accelerate formation Solution Chemistry - ph The ph behavior in 0.1 wt% phosphonate solution and in water and the XRD patterns after 24 hours of reaction at this concentration are shown in Figure 6.5. Reaction in 0.5, 2.5 and 5 wt% solutions are shown in Figures 6.6, 6.7, and 6.8, respectively.

128 a #2 #3 #4 8.5 H 2 O ph #5 6.0 # Time (hours) b H = HAp H H H #5 #4 Relative Intensity #3 #2 #1 H 2 O θ Figure 6.5: ph behavior of 0.1 wt% phosphonate solutions and water (a) and XRD patterns obtained after reaction in a 0.1 wt% phosphonate solution for 24 hours (b).

129 a H 2 O #5 9.0 #3 8.5 # ph 7.0 #4 6.5 # Relative Intensity 5.0 b T = TetCP D = DCPA H = HAp H Time (hours) T D T T D H H #5 #4 #3 #2 #1 H 2 O Figure 6.6: ph behavior of 0.5 wt% phosphonate solutions and water (a) and XRD patterns obtained after reaction in a 0.5 wt% phosphonate solution for 24 hours (b). 2θ

130 a H 2 O #5 # ph #2 #4 # Time (hours) Relative Intensity b T = TetCP D = DCPA H = HAp T H D T T D T H T T H #5 #4 #3 #2 #1 H 2 O Figure 6.7: ph behavior of 2.5 wt% phosphonate solutions and water (a) and XRD patterns obtained after reaction in a 2.5 wt% phosphonate solution for 24 hours (b). 2θ

131 a H 2 O #5 #3 # ph #2 # Time (hours) Relative Intensity b T = TetCP D = DCPA H = HAp T H D T T D T H T T H #5 #4 #3 #2 #1 H 2 O Figure 6.8: ph behavior of 5 wt% phosphonate solutions and water (a) and XRD patterns obtained after reaction in a 5 wt% phosphonate solution for 24 hours (b). 2θ

132 Bisphosphonates At a concentration of 0.1 wt%, phosphonate #4 extends the period of high alkalinity associated with dissolution of the TetCP phase, as seen in Figure 6.5a. This is contrary to expectation because the acidic solutions produced by the phosphonate would be expected to promote TetCP dissolution while retarding that of DCPA, hence decreasing the time required for TetCP dissolution. Phosphonate #2 extends the duration of high ph by a modest amount in comparison to water. However, this may indicate adsorption on the TetCP surface, decreasing dissolution. Conversion to HAp is complete in both bisphosphonate solutions and water by 24 hours of reaction at a concentration of 0.1 wt% as shown in Figure 6.5b. When the reactions are carried out at bisphosphonate concentrations of 0.5 wt%, the lengthy period of high alkalinity is absent, as viewed in Figure 6.6a, and there is only a slight increase in ph before leveling off around a ph of 8. Conversion to HAp is incomplete and residual TetCP and DCPA remain after 24 hours of reaction as observed in Figure 6.6b. This indicates that sufficient bisphosphonate is present to interfere with TetCP dissolution. When the bisphosphonate concentration is increased to 2.5 wt%, the minor increase in ph during the first 6 to 8 hours is not observed and still higher proportions of TetCP and DCPA persist. There are no notable differences in ph variations between concentrations of 2.5 and 5.0 wt%, however more TetCP is observed in the XRD patterns at 5.0 wt%. The liquid supernatant from reaction in 5 wt% phosphonate #4 solution

133 produces a crystalline precipitate solid when dried. This is an indication that some complexing of the phosphonate can occur in these systems Phosphonates Phosphonates #1 and #3 exhibit varying behavior at a concentration of 0.1 wt% in the ph experiment in comparison to each other. As seen in Figure 6.5a, phosphonate #1 solution appears to slightly accelerate reaction in comparison to water while phosphonate #3 solution has similar behavior to bisphosphonate #2 solution, slightly extending the period of high alkalinity and the time of the subsequent drop in ph that corresponds to HAp formation. As with the bisphosphonates, phosphonate #1 and #3 solutions permit complete formation of HAp after 24 hours of reaction at this concentration. At a concentration of 0.5 wt%, both phosphonate #1 and #3 solutions extend the onset and duration of the high alkalinity period, with #3 solution delaying the formation of HAp slightly longer. The inhibition of TetCP dissolution, evidenced by an extended period of intermediate ph during the first 4 hours, is not observed in any of the phosphonate solutions at the lower concentration. This likely signifies adsorption of the phosphonate onto TetCP, and is almost certainly occurring in bisphosphonate #2 and #4 solutions as well. Additionally, the bisphosphonate solutions do not experience a drop in ph to acidic levels as occurs with the other phosphonate solutions. This is due to the continued slow dissolution of the precursors; the increased ph due to TetCP dissolution and lack of DCPA dissolution. Both phosphonate #1 and #3 solutions permit complete

134 119 reaction to HAp within 24 hours at a 0.5 wt% concentration while bisphosphonates #2 and #4 have reactants remaining. When the concentration is 2.5 wt%, phosphonate #1 solution exhibits a longer period before the onset of high alkalinity while phosphonate #3 solution behaves similar to bisphosphonate #2 and #4 solutions. Phosphonate #1 solution allows complete formation of HAp in 24 hours, and phosphonate #3 solution has remaining reactants. At the highest concentration, 5 wt%, the inflection point where the onset of high alkalinity begins does not occur until approximately 15 hours in phosphonate #1 solution. Subsequently, the reaction stops during the period of high alkalinity. The XRD pattern for solids reacted in phosphonate #1 solution shows remaining reactants at this point, predominantly DCPA. Reaction in phosphonate #3 solution exhibits the same behavior as at the lower concentration and, as seen in the bisphosphonate solutions, more TetCP and DCPA remain after 24 hours than at the 2.5 wt% concentration Monophosphonate Formation of HAp with no remaining reactants was observed within 24 hours at all concentrations for phosphonate #5. Phosphonate #5 solutions at concentrations of 0.1 and 0.5 wt%, as shown in Figure 6.5a and Figure 6.6a, accelerate formation of HAp in comparison to water. This is also seen at a 0.3 wt% concentration in the calorimetry data. Even at higher concentrations of 2.5 and 5.0 wt%, the phosphonate only extends the reaction period by a few hours. Thus the monophosphonate is less capable of inhibiting reaction than the other phosphonates. This is despite its higher concentration on a molar

135 120 basis. This is believed to be due to decreased effectiveness of adsorption of the monophosphonate. While the other phosphonates can achieve bidentate or tridentate binding to calcium on the surface of the precursor materials, two monophosphonates are required to coordinate around a cation for this to occur. This may be more difficult due to steric hindrance involving the CH 3 groups attached to the phosphonate group Solution Chemistry Ion Concentrations Experiments to analyze variations in solution chemistry were performed using quiescent 5.0 wt% phosphonate solutions. Based on the results of the ph experiments, solutions with phosphonates 2, 3, and 4 were allowed to react for up to 10 days. The variations in the concentrations of calcium in solution are shown in Figure 6.9a. The calcium concentrations in liquids extracted from phosphonate solutions are higher than those in water at all times. The phosphate concentrations are exhibited in Figure 6.9b. XRD patterns of the solids obtained from these time points are shown in Figures

136 121 [Ca 2+ ] (mm) a H 2 O #1 #2 #3 #4 # Time (hours) [PO 4 3- ] (mm) b 10 H 2 O # # # # # Time (hours) Figure 6.9: Concentration of calcium ions (a) and phosphate ions (b) in solution after reaction for up to 10 days in various 5.0 wt% phosphonate solutions. Inset shows variations in water and phosphonate #1 solution over the first 24 hours.

137 122 T = TetCP D = DCPA H = HAp H H H 18 hours Relative Intensity D T T D 12 hours 6 hours 3 hours Figure 6.10: XRD patterns over the first 18 hours of reaction in water. 2θ

138 123 T = TetCP D = DCPA H = HAp T H D T TD H T 24 hours 18 hours Relative Intensity 12 hours 6 hours 3 hours Figure 6.11: XRD patterns after reaction in 5 wt% solution of phosphonate #1 over 24 hours. 2θ

139 124 T = TetCP D = DCPA H = HAp H H T D T T T T T D 10 days 7 days Relative Intensity 5 days 3 days 1 day Figure 6.12: XRD patterns after reaction in 5 wt% solution of phosphonate #2 over 10 days. 2θ

140 125 Relative Intensity T = TetCP D = DCPA H = HAp H T D T T D T H T T 7 days 5 days 3 days 1 day Figure 6.13: XRD patterns after reaction in 5 wt% solution of phosphonate #3 over 7 days. 2θ

141 126 T = TetCP D = DCPA H = HAp H T D T T D H T 10 days Relative Intensity 7 days 5 days 3 days 1 day Figure 6.14: XRD patterns after reaction in 5 wt% solution of phosphonate #4 over 10 days. 2θ T = TetCP D = DCPA H = HAp H H H H 18 hours Relative Intensity T D T T D T T 12 hours 6 hours 3 hours Figure 6.15: XRD patterns after reaction in 5 wt% solution of phosphonate #5 over 18 hours. 2θ

142 Bisphosphonates The calcium and phosphate ion concentrations in solution after 3 hours of reaction in the bisphosphonate solutions are much higher than for reaction in water, and remain higher for the duration of the experiment, as seen in Figure 6.9. The increased calcium concentration is likely due to the formation of soluble calcium phosphate complexes and is consistent with the effects of phosphonates in precluding nucleation. The calcium and phosphate concentrations gradually declined over 10 days. The bulk of the decrease occurred between days 1 and 3 in phosphonate #2 solution. The decrease was more gradual in phosphonate #4 solution. TetCP and DCPA are present in XRD patterns (Figures 6.12 and 6.14) after 7 days of reaction in both phosphonate solutions. Phosphonate #4 solution permitted slightly less HAp formation than #2 solution as measured by the relative intensities of the TetCP and HAp peaks. However at 10 days conversion to HAp was extensive, with some TetCP remaining present in phosphonate #4 solution. As mentioned previously, this does not accord with the calorimetric data, which shows phosphonate #2 to be more effective, and is believed to be due to the effect of liquid-to-solid ratio Phosphonates The concentrations of calcium in phosphonate #3 solution are higher than #1, #2, and #4 solutions until the last data point, while phosphate concentrations are lower than in both #2 and #4 solutions at all times. The reason for the high calcium concentration coupled with low phosphate concentration, particularly in comparison to bisphosphonate

143 128 #2 and #4 solutions, is unclear. Phosphonate #1 solution also shows high calcium concentrations with phosphate concentrations comparable to those for the reaction in water. This can be attributed to formation of soluble calcium phosphonate complexes, with the calcium present detected by AES. XRD patterns of the solids obtained from solution chemistry experiments with phosphonate #1 solution, displayed in Figure 6.11, show gradual decreases in the amount of reactants present over 24 hours. HAp is the only phase present after 24 hours of reaction. Reaction in phosphonate #3 solution, shown in Figure 6.13, exhibits no HAp formation over the first 5 days followed by a complete conversion to HAp between 5 and 7 days. Similar behavior is observed in the bisphosphonates. Thus, phosphonate #3 solution is almost as effective at inhibition of HAp formation as are the bisphosphonate solutions Monophosphonate The concentrations of calcium in the monophosphonate solution are similar to that in the other phosphonate solutions, and are well above those in water. This may indicate that formation of soluble calcium salts is occurring despite the rapid formation of HAp. The concentration of phosphate is far above those in water or in the other phosphonate solutions. This is unexpected based on the complete formation of HAp and lack of reactants remaining after only 18 hours. If high concentrations of phosphate alone are present in the solution, the dissolution of DCPA is expected to be depressed. However,

144 129 because dissolution is clearly occurring, the phosphate concentration may indicate that a stoichiometric HAp is forming. This solution is supersaturated with respect to HAp. At a low liquid-to-solid ratio, the high concentrations of calcium and phosphate likely depress DCPA dissolution. This accords with the calorimetric data, showing presence of DCPA at the highest concentration although the other phosphonates show no reaction Characterization In an effort to separate the effects of adsorption from the formation of soluble or insoluble calcium-phosphonate salts, an experiment using a pressed pellet of the precursors with a high concentration of each phosphonate solution was performed. Pressed pellets of CDHAp precursors were placed in 50 wt% concentrations of phosphonate solutions for 24 hours. Development of fibrous shells could be seen around the pellets. These shells, along with the pellets, were analyzed using XRD, SEM, and EDS. The XRD patterns after reaction for 24 hours for all phosphonates are presented in Figure 6.16 and SEM micrographs are displayed in Figures

145 130 Relative Intensity T = TetCP D = DCPA H = HAp H D T T D H H H #5 #4 #3 #2 #1 H 2 O Figure 6.16: XRD patterns obtained from pellets reacted for 24 hours in a 50wt% phosphonate solution or water. 2θ Bisphosphonates Pellets reacted in solutions with bisphosphonates #2 and #4 developed the most macroscopically apparent features. The pellet reacted in phosphonate #4 solution was surrounded by a complete shell of fibrous needle-like crystallites. These were separated from the pellet, which had converted to HAp, by a gap. Reaction in phosphonate #2 solution produced a partial shell that remained closer to the pellet, but was nonetheless visible without magnification. SEM images of the shells formed in phosphonate #4 and #2 solutions are shown in Figures 6.17 and The crystallites forming the shell in phosphonate #2 solution are petal-like. The morphology of the crystallites that formed the

146 131 shell in phosphonate #4 solution varied from the inside to the outside of the shell. Inside, crystals were smaller and more rounded while the outer crystals had a large aspect ratio. Long, thin needles of a calcium-phosphonate #4 salt have been previously observed. 83,84 This morphology was reported to be associated with a solid calcium phosphonate precipitate having a ring structure and a calcium-to-phosphonate ratio of 1:1 while spherical precipitates are associated with ratios of 2:1. 84 EDS spectra of the fibers in Figure 6.17 show far more phosphorus and oxygen present than calcium and in weight percentages expected for a precipitate with a 1:1 Ca-to-phosphonate ratio. Thus the solid formed in the present study is anticipated to have a 1:1 calcium-to-phosphonate ratio. However, its diffraction pattern differs from that reported by Browning and Fogler 85 and matches only some peaks of a simulated pattern from the data given by Uchtman. 86 The composition of the precipitate formed in phosphonate #2 solution is unknown, but is also presumed to have a ring structure.

147 µm Figure 6.17: Microstructure of shell formed in phosphonate #4 solution. 100 µm Figure 6.18: Portion of shell observed on pellet after reaction in phosphonate #2 solution for 24 hours.

148 133 The proportions of HAp and reactant remaining after reaction for 24 hours seen in the pellet by XRD are contrary to expectations. At this high concentration, HAp formation would not be expected. However, HAp is completely formed in phosphonate #4 solution and significant amounts of DCPA with no TetCP remain after reaction in phosphonate #2 solution Phosphonates and Monophosphonate The phosphonates did not form visible shells during reaction at high concentration. Reaction in phosphonate #3 solution did result in microscopic surface features on the pellet, similar to those seen in phosphonate #2 solutions. The precipitates causing these were approximately 10 µm in diameter as opposed to over 100 µm. Pellets reacted in phosphonate #1 and phosphonate #5 solutions both lack the notable surface features of those reacted in phosphonates #2-4 solutions when viewed using SEM. The pellet surfaces after reaction in phosphonates #1, #4, and #5 solutions were all similar, as were the surfaces from #2 and #3 solutions if the regions of macroscopically observable surface features were avoided. A representative SEM image of the pellet surface after reaction in a phosphonate solution is shown in Figure 6.19 while the surface after reaction in water is shown in Figure 6.20.

149 µm Figure 6.19: Surface of pellet after reaction in a phosphonate solution. 2 µm Figure 6.20: Surface of pellet after reaction in water.

150 135 The surface of the pellet in water is relatively flat and covered in needles of HAp with lengths of approximately 0.5 µm. The pellet surface after reaction in phosphonate is more jagged, and HAp crystals could not be observed at magnifications similar to that of the sample in water. As XRD indicates formation of HAp occurred in all samples, however, this suggests formation of nanocrystalline HAp Discussion There are two primary methods by which the phosphonates can inhibit growth of HAp via a cementitious reaction between TetCP and DCPA. First, adsorption on the surfaces of precursor particles can interfere with their dissolution, inhibiting the normal route of formation of HAp in this system. Additionally, adsorption onto any HAp nuclei can prevent further growth of HAp. That the calcium and phosphate concentrations indicate supersaturation with respect to HAp indicates the occurrence of this phenomenon. The second mechanism is formation of soluble complexes or precipitated calcium-phosphonates. These can isolate the calcium from reaction with phosphate ions, preventing formation of calcium phosphate phases. Complexation of calcium would also permit an increased phosphate concentration in solution, decreasing the driving force for precursor dissolution. Adsorption onto the precursors is almost certainly occurring, as the presence of TetCP and DCPA at reaction times beyond which HAp formation normally occurs indicates reduced rates of dissolution. Additionally, the heightened value of the initial peak observed in heat evolution curves can be attributed to adsorption. 87 However, some

151 136 dissolution also must be occurring, even when calorimetric peaks are absent, as indicated by the presence of calcium and phosphate ions in solution. Complex formation also appears to occur during these reactions. The increased concentrations of calcium ions in solution may be an indication of complexation. If these ions were free to react with the phosphate groups present in solution, HAp growth could occur. Dissolution and reprecipitation of calcium phosphonate onto the surface of calcite has been observed 83, and a similar reaction may occur in this system but would be difficult to detect due to the thin layers formed. The morphology of the HAp formed in phosphonate is also unusual in comparison to that in water. As seen in Figure 6.20, HAp reacted in water generally forms crystalline needles. No such crystal growth is observed in the pellets reacted in phosphonates. This indicates either formation of nanocrystalline HAp or a total change in morphology when phosphonate is present. Francis 73 observed a gel-like morphology in the presence of phosphonate #4 solution and attributed this to formation of amorphous calcium phosphate, indicating that a change in morphology has precedence Bisphosphonates Several conflicting trends are observed in the bisphosphonate solutions. First, while phosphonate #4 is the most successful at delaying complete formation of HAp in solution chemistry experiments, it shows more HAp formation under the conditions of the calorimetry experiments and complete formation in the pellet experiments. The

152 137 reasons for HAp inhibition in the solution chemistry experiments for both bisphosphonates are those described above: adsorption with some complex formation. Deviations from the results seen in solution chemistry involve more complex explanations. In heat evolution curves, the effects of the bisphosphonates are not as obvious as the effects of phosphonates regardless of concentration. At low concentration, both bisphosphonates have reaction curves similar to that of water. This corresponds to the formation of HAp seen in XRD. However, both contain reactant precursor material at the completion of the experiments. Enough phosphonate is present to inhibit complete conversion by adsorbing onto the surface of precursors. The relatively rapid formation of HAp seen in the bisphosphonate solutions in comparison to the phosphonate solutions in these experiments may be due to a stronger interaction of the bisphosphonates with the precursors. The lower heights of the reaction peaks in comparison to water indicate that precursor particles are not reacting. This might also explain why more precursor remains when phosphonate #2 is used, as phosphonate #4 may interact more strongly with precursor surfaces due to the extra OH group present. 87 The lower liquid-to-solid ratio used in the calorimetry experiment may also affect the efficacy of the bisphosphonates by decreasing the opportunities for calcium phosphonate complex formation. The lower liquid-to-solids ratio also requires less precursor dissolution to achieve a supersaturated solution of HAp. Precipitation of calcium phosphonate shells produces conditions favoring HAp formation. However, it is unclear what those conditions are. Formation of a 2:1 precipitate may permit more water to react with the precursor particles by shielding the hydrophobic -CH 3 group. Additionally, in other precipitates, the extra OH group of

153 138 phosphonate #4 may provide a site for nucleation. Formation of HAp on sites away from the DCPA precursor would lower the local concentration of phosphate, permitting dissolution of the DCPA. Another view of changes with shell is as follows: Adsorption on the surface is associated with decreased dissolution and slowed reaction rates. However, the low ph observed at these phosphonate concentrations may also etch the surface of the pellets by dissolving TetCP. Calcium released can form a precipitate with the phosphonate and form a crystalline solid. This solid can have distinct morphologies, such as fibers, which are not effective at inhibiting dissolution. The crystals that form have a low solubility and thus stay in place, forming a shell Phosphonates In the phosphonate solutions, adsorption onto the surfaces of the precursor particles occurs, but the phosphonates are not as effective at inhibiting dissolution as the bisphosphonates. This is demonstrated in the heat evolution curves at low concentration in the calorimetry experiments. The initial mixing peak is similar to that in bisphosphonate solutions and water, but the peak associated with nucleation is reduced. An extended growth peak is observed, and both phosphonate solutions permit complete formation of HAp within 24 hours. Likewise, in the solution chemistry experiments adsorption is not as effective and HAp is more readily formed. Phosphonate #3 solution is more effective than phosphonate #1 solution; this may be due its shorter chain length and relative similarity to the bisphosphonates, particularly with regard to phosphonate #2. The pellet experiments show complete formation of HAp in phosphonate #1 solution

154 while the pellet in phosphonate #3 solution is more similar to that in phosphonate #2 solution Monophosphonate The monophosphonate solutions behave differently from the other phosphonate solutions in almost all instances. The reaction at low concentration appears to accelerate in comparison to water in both ph and calorimetry experiments at low concentrations. At high concentration in calorimetry, the monophosphonate is the only phosphonate to show any major reaction peaks. The first peak is associated with mixing and adsorption, while the second, which occurs well after that in water, is associated with nucleation and possibly some growth. The third peak seen over a long time interval is associated with growth and possibly diffusion through a surface layer. The accelerated reaction is associated with the relatively poor complex forming ability of this phosphonate. Although bidentate binding has been observed with phosphonate #5 88, at low concentrations this will not occur. The lack of shell formation is also not unexpected, as calcium phosphonate precipitates of this system are more soluble than that in an HEDP (phosphonate #4) system, with log K values of 1.6 and 6.4 (3.3 at 1:1 Ca:Phosphonate), respectively. 89

155 Conclusions Attribution of bisphosphonate effectivity to 6-member ring formation was difficult to establish from the data collected. The results obtained revealed anomalous behavior, as at high concentrations phosphonate with P-C-C-P bonds was similar in effectiveness to the bisphosphonates. However, the general trend anticipated, that increasing chain length would decrease the efficacy of the phosphonate to inhibit HAp formation, was observed. Bisphosphonate #4, known as HEDP and one of the first clinically used bisphosphonates, was found to best inhibit HAp formation at all concentrations in experiments with high liquid-to-solid ratios. Bisphosphonate #2 had similar inhibition properties. However, at low liquid-to-solid ratio the opposite trend was noted. This may be significant if these compounds are to be delivered by an HAp matrix. This is attributed to the effect of liquid-to-powder ratios and affinity of phosphonate #4 for surface adsorption. Phosphonate #3 was similar in effectiveness to the bisphosphonates, however this was not realized until higher concentrations. Phosphonate #1 was only moderately effective for inhibiting HAp formation. This is attributed to the difficulty of forming ring attachments with calcium surface ions on the precursor particles or HAp nuclei. Phosphonate #5 was noted to increase formation of HAp at low concentrations, while preventing DCPA dissolution at high concentrations. The latter is attributed to increasing concentration of phosphate ion in solution, lowering the driving force for DCPA dissolution.

156 141 Much future work can be carried out to elucidate the effects of phosphonates. Addition of more phosphonates with longer chains, possibly along with bisacylphosphonates, would help permit conclusions of the importance of ring formation. Further study of ion concentrations and phase evolution at varying concentrations would refine the conclusions reached from ph experiments. Additional concentrations and liquid-to-powder ratios in calorimetry would also augment the conclusions drawn from aspects of this study.

157 Chapter 7 Formation and properties of composites comprised of calcium-deficient hydroxyapatites and ethyl alanate polyphosphazenes 7.1 Abstract Composites comprised of calcium-deficient hydroxyapatite (CDHAp) and biodegradable polyphosphazenes were formed via cement-type reactions at physiologic temperature. The composite precursors were produced by blending particulate hydroxyapatite precursors with 10 wt% polymer using a solvent/non-solvent technique. HAp precursors having a calcium-to-phosphate ratio of 1.5 (CDHAp) were used. The polymeric constituents were poly[bis(ethyl alanato)phosphazene] (PNEA) and poly[(ethyl alanato) 1 (p-phenylphenoxy) 1 phosphazene] (PNEA 50 PhPh 50 ). The effect of incorporating the phenyl phenoxy group was evaluated as a means of increasing the mechanical properties of the composites without severely retarding the rates of HAp formation. Reaction kinetics and mechanistic paths were characterized by ph determination, X-ray diffraction analyses, scanning electron microscopy, and infrared spectroscopy. The mechanical properties were analyzed by compression testing. These analyses indicated that the presence of the polymers slightly reduced the rate of HAp formation. However, surface hydrolysis of polymer ester groups permitted the formation of calcium salt bridges that provide a mechanism for bonding with the HAp. The compressive strengths of composite containing PNEA 50 PhPh 50 were superior to that

158 containing PNEA, and were comparable to those of HAp produced in the absence of polymer Introduction Bone can be considered a porous composite comprised of collagen mineralized by hydroxyapatite. HAp imparts strength and rigidity to the tissue while collagen serves to direct mineral growth in a manner that confers toughness. 90 Hard tissue analogs based on HAp alone lack toughness and, therefore, tend to fail catastrophically. Consequently, development of synthetic composites that emulate the properties of natural tissue, coupled with an ability to resorb by cell-mediated processes, is desirable. Additionally, an ideal bone analog material should be amenable to introduction into a bone defect as a workable, moldable, and fast-setting putty that hardens in vivo without causing damage to the surrounding tissue. Stoichiometric hydroxyapatite (SHAp, Ca 10 (PO 4 ) 6 (OH) 2 ) is frequently regarded as a model for bone mineral. The Ca/P ratio of this composition is HAp, however, is not a compound of fixed composition and can incorporate many substituents. 4 In natural bone carbonate substitutes for phosphate, and this affects the Ca/P ratio. 3 A general compositional formula for non-substituted HAp is Ca 10-x (HPO 4 ) x (PO 4 ) 6-x (OH) 2-x, where x ranges from 0 to In particular, the ratio of Ca/P in HAp can vary between the compositional limits of about 1.5 and When the Ca/P ratio is below 1.67, the HAp is calcium-deficient. The extent of calcium deficiency affects the stability of HAp, increases its bioactivity, and increases its solubility product. 3

159 144 Polyphosphazenes are characterized by a phosphorus-nitrogen backbone with two functional groups attached to each phosphorus atom. Biodegradable polyphosphazenes are attractive constituents for bone analog composites because their presence confers toughness and they degrade into biologically recognizable byproducts. 91,92 Further, the degradation rates of polyphosphazenes can be varied by changing the lengths and compositions of the side groups. 93 Our strategy has been to tailor the compositions of polyphosphazenes, the Ca/P ratios of hydroxyapatite and the processing methods to establish the rates of composite formation, estimate rates of in vivo resorption, coupled with limited evaluations of mechanical properties. Thus, aspects of the formation and behavior of HAppolyphosphazene composites intended for use as bone analogs have been studied previously Significantly, HAp formation in the presence of a polyphosphazene bearing acidic groups requires less than 24 hours for complete conversion of precursor minerals. 96 In addition, some polymers undergo surface hydrolysis to produce calcium cross-linkages. These salt-bridged regions then provide sites for nucleation of HAp on the polymer surface. 97 The current study specifically investigates the formation and properties of composites comprised of two calcium-deficient HAp compositions and polyphosphazene-ethyl alanine-based polymers. In particular, composites consisting of two polyphosphazene preparations, a biodegradable poly(phosphazene-ethylalanine) homopolymer and a substituted copolymer with phenyl-phenoxy (PhPh) in place of 50% alanine were investigated. The polymers are thus referred to as PNEA and PNEA 50 PhPh 50. The composites currently described exhibit advantages not realized in

160 145 previously examined systems. Calcium-deficient HAp compositions exhibit greater bioactivity than does fully stoichiometric HAp. 99 While previous studies have established that incorporation of polymer tends significantly to decrease the strengths of the composites 100, use of blocky side chains was anticipated to increase mechanical properties. 101 This anticipation is based on incorporation of bulky side groups having been shown to increase the tensile strengths of polyphosphazenes. 102 In addition, polyphosphazenes with ethyl-alanine and ethyl or propyl oxybenzoate side groups were observed to allow cells to attach and proliferate. 103 Finally, the polymers used in this study have been shown to biodegrade over time in vivo, and to invoke only a mild to moderate inflammatory response Materials and Methods Precursor Synthesis The CDH precursor powder was synthesized from mixtures of tetracalcium phosphate (TetCP, Ca 4 (PO 4 ) 2 O) and monocalcium phosphate monohydrate (MCPM, Ca(H 2 PO 4 ) 2 H 2 O) at Ca/P ratios of 1.5 and 1.6, respectively. TetCP was prepared by ball milling CaCO 3 (Osram-Sylvania, PA) and MCPM (FMC Corp., NY) at a 3:1 molar ratio for 16 hours in heptane (Alfa Aesar, Ward Hill, MA). After filtering and drying, the particulate TetCP precursors were fired in air at 1400 o C for 1 hour and quenched rapidly. X-ray diffraction was used to confirm phase purity. The TetCP was ground by hand, passed through a 63µm sieve, ball milled, and attrition milled to reduce particle size.

161 146 TetCP and MCPM were then mixed in a Ca/P ratio of 1.5 and ball milled in heptane. After synthesis, the precursor powders were stored in a desiccator under vacuum. The average particle size was 2.5 µm, as measured by SEM. Figure 7.1 shows the structures of the phosphazenes used in this study. The synthesis and characterization of these phosphazenes have been described in detail by Singh et al. 102 Figure 7.1: Structure of poly(phosphazene-ethylalanine), PNEA, (a) and cosubstituted polymer with phenyl phenoxy groups, PNEA 50 PhPh 50, (b) the two polymers used in this study. Composite precursors containing PNEA or PNEA 50 PhPh 50 and CDHAp precursors were synthesized by an emulsion technique. Briefly, 1.5 g of polymer was dissolved in 30ml of methanol (for PNEA polymer) or dimethylformamide (for PNEA 50 PhPh 50 polymer). The polymer solution was then added dropwise to a vigorously stirred suspension of 15 g of the HAp precursor in 1 liter of heptane (Fisher Scientific, USA) and 50ml of dimethylformamide at room temperature. The suspension was stirred for 10 minutes and the excess solvent was evaporated to dryness using a rotary

162 evaporator. The resultant solid was dried under vacuum at 50 C for 72 hours, and then stored in a desiccator under vacuum Solution Chemistry The variations in ph with time were measured using an Orion 920 ph meter. 0.5g composite precursors were initially mixed with a small amount of water using a mortar and pestle before being placed in a double-walled glass beaker with 35 ml of distilled, de-ionized water. The temperature of the reaction vessel was held constant at 37 o C. The mixture was stirred continuously with nitrogen bubbled through. The reaction ph was followed for 24 hours. At the end of each experiment, the slurry was filtered and the separated solids were flushed with acetone to stop further reaction. After drying, the solids were examined for their phase compositions by x-ray diffraction and infrared spectroscopy Characterization XRD analyses used an automated x-ray diffractometer (Scintag, Inc., Sunnyvale, CA), with a step size of 0.02 o, a scan rate of 2 o per minute, and a scan range from 20 to 40 o 2 theta. Phases present in the pattern were compared to ICDD PDF cards (HAp), (DCPA), and (TetCP). 105 Infrared spectra were obtained using a Nexus 670 FT-IR spectrometer (Thermo Nicolet Corp., MA) with pellets consisting of KBr and 2 wt% sample over a range from 4000 to 650 cm -1.

163 Mechanical Testing For compressive testing, the precursor powders were combined with water at a powder-to-liquid weight ratio of 2.5-to-1 and mixed on a glass plate using a metal spatula. The paste was pressed by hand into cylindrical-shaped molds 12.7 mm in height and 6.35 mm in diameter. Three pellets were made for each composite. The samples were cured in a humidified atmosphere at 37 o C for 24 hours. Compressive testing was performed using an Instron 4202 testing instrument (Instron, MA) using a cross-head speed of 0.3 mm/min. The fractured surfaces were coated with gold, and examined for their microstructure using SEM (Hitachi S-3500N, Japan) with a low accelerating voltage. 7.4 Results and Discussion The kinetics and variations in solution chemistry during CDHAp formation 52 have been previously investigated; those of composites containing the present polymers have not been determined. Thus to establish the kinetics of reaction and to determine whether there should be concern for ph excursions to a cytotoxic range, ph variations during the conversion of precursors to HAp were assessed. The variations in the ph during the conversion reactions of the calcium phosphate in the presence and absence of polymer are shown in Figure 7.2. The variations in ph are attributable to the dissolution behavior of both TetCP and DCPA, which in turn depend on the relative proportions of these constituents in the initial powder mixture. 41 Reaction of the CDHAp precursors reached a value of 8.5 within the first half hour. The period

164 149 over which TetCP dissolution occurs, during which the ph is high, controls the solution chemistry and is short for CDHAp conversion. After reaching a maximum value of about 8.6 after an hour and a half of reaction, ph of the CDHAp solution began to decrease. This decrease is the result of DCPA dissolution. After about 6 hours of reaction the ph attained a value slightly below 7 and subsequent ph changes are small. Once again, variations in the powder manifest during reaction, as the time to ph decrease is approximately half that seen for CDHAp in water in the preceding chapter. This data indicates that the conversion to CDHAp can occur rapidly and presents conditions anticipated to be less aggressive to local cells CDH-PNEA 50 PhPh CDH-PNEA CDHAp 7.5 ph Time (hours) Figure 7.2: The variations in ph with time for the CDHAp composite compositions studied and CDHAp without polymer.

165 150 Reaction of the PNEA and PNEA 50 PhPh 50 composites is inhibited as indicated by the extended period of high ph. This observation indicates that these polymers adsorb on the surfaces CDHAp precursors. Previous studies have shown the proclivity of the ester groups pendant on polyphosphazenes to adsorb onto calcium phosphate precursors, thereby inhibiting precursor hydrolysis. 97 Adsorption then interferes with the dissolution of the rate limiting precursor constituent. Additionally, although PNEA 50 PhPh 50 contains fewer hydrolyzable groups, the slow kinetics may also be attributed to the presence of the bulky, hydrophobic phenyl phenoxy group. The formation of the carboxylic acid groups leads to the formation of calcium salts of these polymers via an acid-base reaction. 96 As will be subsequently described, this phenomenon also affects mechanical properties. XRD patterns of the solids produced from the CDHAp precursors after 24 hours of reaction with and without polymer were compared with the standard patterns of HAp, TetCP and DCPA as described in The major phase observed in all samples was HAp. However, the decreased formation of the composites rate was confirmed by x-ray diffraction analyses. Small amounts of unreacted TetCP and DCPA were observed in the CDH-PNEA 50 PhPh 50 composite and were present to a greater extent than in the composite containing PNEA. The incomplete reaction in the current composites confirms the occurrence of polymer adsorption. Figures show the SEM micrographs of the fracture surfaces of samples from composites with PNEA or PNEA 50 PhPh 50 and sample without polymer. All micrographs show HAp needle-like crystallites, approximately 0.5µm in length, in assemblages as pseudomorphs of the reactant precursors. The HAp crystallites closely resemble bone apatite in shape and size. The presence of unreacted precursors, termed R

166 151 for reactant in the micrographs, can be also observed, with an increased extent in Figure 7.5. Figure 7.4 also shows the presence of flake-like assemblages of apatite crystallites. The presence of this flake-like morphology was not as noticeable in the composites containing PNEA 50 PhPh 50. These morphological differences may be the result of differences in the proportion of the adsorbing ester group in these polymers. Figure 7.6 shows a fiber of PNEA polymer after hydrolysis from a ph experiment where the precursors and polymer were mixed by grinding in liquid nitrogen. Formation of HAp phase on the surface of the polymer is clearly seen in this image. (a) R R R = Reactant 5µm Figure 7.3: SEM image of fracture surface of sample without polymer.

167 152 (b) R R = reactant R 5µm Figure 7.4: SEM image of composite with PNEA fracture surface. (c) R R R = reactant 5µm Figure 7.5: SEM image of composite with PNEA 50 PhPh 50 fracture surface.

168 153 10µm Figure 7.6: Fiber showing growth of HAp phase after hydrolysis of a CDH-PNEA composite prepared by mechanical mixing. The relevant portions of the spectra obtained by FT-IR analyses of composites with PNEA and PNEA 50 PhPh 50 and CDHAp without polymer are shown in Figure 7.7. The broad band from approximately cm -1 noted in all orthophosphates appears in all spectra. 106 Two bands appear in all samples within the range cm -1. These bands are relatively weak in CDHAp spectra, while stronger in the spectra of the composites. Both the carbonyl group (C=O) of CO 2 and the carboxylate (COO - ) group are known to absorb within this range. 106 The weak bands in polymer-free samples suggest minor carbonate incorporation in CDHAp; this is anticipated because of the opportunity to extract CO 2 from air and is observed in all FT-IR spectra of the preceding chapters. The second band that appears at 1463 cm -1 in the spectra of CDHAp was shifted in the spectra