Thermodynamic Modeling of Hydroxyapatite Crystallization with Biomimetic Precursor Design Considerations

pubs.acs.org/cm Published on Web 12/02/2009 r 2009 American Chemical Society

36 Chem. Mater. 2010, 22, 36–46DOI:10.1021/cm900183v

Thermodynamic Modeling of Hydroxyapatite Crystallization with

Biomimetic Precursor Design Considerations

Christina Mossaad, Matthew Starr, Swanand Patil, and Richard E. Riman*

Department of Materials Science and Engineering, Rutgers, The State University of New Jersey,607 Taylor Road, Piscataway, New Jersey 08854

Received January 20, 2009. Revised Manuscript Received August 25, 2009

The objective of this paper is to present a hydroxyapatite synthesis biomimetic precursor paradigmand evaluate four precursor systems against its criteria. Thermodynamic modeling was used as amethod of evaluation. The system that best fit the paradigm was experimentally validated bysynthesizing inorganic powders over a range of reaction conditions and characterizing the reactionproducts through X-ray diffraction. The Ca(C2H3O2)2-K3PO4-H2O system was found to best fitthe paradigm and was effective in synthesizing hydroxyapatite as predicted by the yield diagram. Onthe basis of the findings of this diagram and its validation, it appears that thermodynamics largelygoverns the synthesis of hydroxyapatite in all these systems, making these equilibrium diagramsvaluable for future research on hydroxyapatite crystallization.

1. Introduction

Hydroxyapatite (Ca5(PO4)3OH or Ca10(PO4)6(OH)2)is an inorganic material that has a variety of potentialapplications in modern science. It has been used inchromatography,1 corrosion resistant materials,2 drugdelivery,3 and most notably bone implant coatings4 andcomposites5 due to its osteoconductive nature. There hasbeen a shift toward adapting hydroxyapatite synthesis tech-niques to more closely resemble how it is made natu-rally in the human body, or in other words, a biomimeticprocess. However, the majority of synthetic methods in theliterature cannot be classified as biomimetic processes eventhough they may yield biomimetic materials because theprecursors and other reaction conditions for synthesis arenot observed in nature. Many variations of a definition ofbiomimetic materials have been postulated involving the useof simulated body fluid, proteins, or collagen to producehydroxyapatite, but these fail to address the flaws in thestarting chemical precursors used in abiomimetic process.6-8

While these perspectives on biomimesis (the art ofmimic-king natural phenomena) are important, the precursor de-sign is no less crucial in findinga truebiomimetic system thatcan produce hydroxyapatite in high yield. The vision of abiomimetic crystallization paradigm, ormodel, for the crea-tion of a less harsh hydroxyapatite synthesis method neces-

sitates a critical look at the reaction conditions under whichthemineral is formednaturally in thebody.Averagehealthybody conditions are 37 �C, pH of 7.4, which produce10-40 nm hydroxyapatite crystals in hard tissue dependingon mechanical loading, health, and nutrition.9-11 In con-trast, current hydroxyapatite synthesis methods utilize reac-tion conditions from 25 to 1200 �C, pH of 8-12 (with orwithout a pH agent modification), producing particles any-where from 15 nm to 2 μm.12-33

(1) Lawson, M. A.; Xia, Z. Bone 2006, 38(3), S1, 55.(2) Garcia, C.; Cere, S.; et al. J. Non-Cryst. Solids 2004, 348, 218.(3) Ginebra, M. P.; Traykova, T.; et al. J. Controlled Release 2006,

113(2), 102.(4) Kuo, M. C.; Yen, S. K. Mater. Sci. Eng. C 2002, 20(1-2), 153.(5) Wei, G.; Ma, P. X. Biomaterials 2004, 25, 4749.(6) Landi, E.; Tampieri, A.; Celotti, G.; Langenati, R.; Sandri, M.;

Sprio, S. Biomaterials 2005, 26(16), 2835.(7) Kikuchi, M.; Ikoma, T.; Itoh, S; Matsumoto, H. N.; Koyama, Y.;

Takakuda, K.; Shinomiya, K.; Tanaka, J. Compos. Sci. Technol.2004, 64(6), 819.

(8) Ma, P. X. Adv. Drug Delivery Rev. 2008, 60(2), 184.

(9) Ilich, J. Z.; Kerstetter, J. E. J. Am. Coll. Nutr. 2000, 19(6), 715.(10) Silver, F. H.; Freeman, J. W.; et al. J. Biomech. 2003, 36(10), 1529.(11) Katti, K. S. Colloids Surf., B 2004, 39(3), 133.(12) Suchanek, W. L.; Shuk, P.; et al. Biomaterials 2002, 23(3), 699.(13) Shu, C.; Yanwei, W.; Hong, L.; Zhengzheng, P.; Kangde, Y.

Ceram. Int. 2005, 31(1), 135.(14) Liao, Q.; Xu, G.; Yin, G.; Zhou, D.; Zheng, C.; Li, X. J. Funct.

Mater. 2002, 33(3), 38.(15) Cao,M.;Wang, Y.; Guo, C.; Qi, Y.; Hu, C.Langmuir 2004, 20(11),

4784.(16) Yan, L.; Li, Y.; Deng, Z.; Zhuang, J.; Sun, X. Int. J. Inorg. Mater.

2001, 3(7), 633.(17) Liu, J.; Ye, X.; Wang, H.; Zhu, M.; Wang, B.; Yan, H. Ceram. Int.

2003, 29(6), 629.(18) Kim, I.; Kumta, P. N. Mater. Sci. Eng. B 2004, 111(2-3), 232.(19) Kuriakose, T. A.; Kalkura, S. N.; Palanichamy, M.; Arivuoli, D.;

Dierks,K.;Bocelli,G.;Betzel,C.J.Cryst.Growth2004,263(1-4), 517.(20) Ramanan, S. R.; Venkatesh, R. Mater. Lett. 2004, 58(26), 3320.(21) Grandjean-Laquerriere, A.; Laquerriere, P.; et al. Biomaterials

2006, 27(17), 3195.(22) Li-yun, C.; Chuan-bo, Z.; Jian-feng, H. Mater. Lett. 2005, 59

(14-15), 1902.(23) Afshar, A.; Ghorbani, M.; Ehsani, N.; Saeri, M. R.; Sorrell, C. C.

Mater. Des. 2003, 24(3), 197.(24) Zhang, H.; Wang, Y.; Yan, Y.; Li, S. Ceram. Int. 2003, 29, 413.(25) Wang, A.; Liu, D.; et al. Mater. Sci. Eng. C 2007, 27(4), 865.(26) Liu, C.; Huang, Y.; et al. Biomaterials 2001, 22, 301.(27) Saeri, M. R.; Afshar, A.; Ghorbani, M.; Ehsani, N.; Sorrell, C. C.

Mater. Lett. 2003, 57(24-25), 4064.(28) Rend�on-Angeles, J. C.; Yanagisawa, K.; Ishizawa, N.; Oishi, S.

J. Solid State Chem. 2000, 151(1), 65.(29) Yoshimura,M.; Sujaridworakun,P.;Koh,F.;Fujiwara,T.; Pongkao,

D.; Ahniyaz, A.Mater. Sci. Eng. C 2004, 24(4), 521.(30) Jinawath, S.; Polchai, D.; et al. Mater. Sci Eng. C 2002, 22(1), 35.(31) Zhang, L.-J.; Feng, X.-S.; et al. Mater. Lett. 2004, 58(5), 719.(32) Tas, A. C. Biomaterials 2000, 21, 1429.(33) Bayraktar, D.; Tas, A. C. J. Eur. Ceram. Soc. 1999, 19, 2573.

Article Chem. Mater., Vol. 22, No. 1, 2010 37

A paradigm biomimetic precursor hydroxyapatite syn-thesis system would precipitate at or below 37 �C, rela-tively constant neutral pH (7-7.8), in high enough yield(>1 g per 100 mL reaction) for industrial applications inless than 4 h using precursors that do not leave toxic ionsin solution. The ions left in solution should either beabundant in the body or metabolizable so minimumwashing of the powder would be necessary prior to appli-cation. The pH could be controlled through the use ofbenign biological buffers or through the proper precursorchoice. A synthesis time under 4 h is desirable becauselonger than this is not optimal for industrial scale up,could compromise aseptic processing in the manufactureof implant composites, and would not be optimal forsurgical settings. A paradigm biomimetic precursor sys-tem could also mineralize live tissue during operativesurgical procedures, an application that emphasizes theneed for a short time and benign processing conditionwith high enough yield to produce a positive cellularresponse but does not require specialized equipment ormanipulation or washing after crystallization. A chemi-cally robust system is one that does not entirely depend oninput reactant stoichiometry (Ca/P range 0.5-4 and stillproduces HA) . Additionally, having no dependency onthe degree of mixing in both industrial and surgicalvenues would more easily yield consistent results, redu-cing concern of under/over mixing.It is clear that current hydroxyapatite synthesis meth-

ods do not meet all these requirements in various ways.Methods commonly employed include mechanochemi-cal,12,13 hydrothermal,14-17 sol-gel,18-21 precipitation,22-27

phase transition,28-30 and simulated body fluid31-33

crystallizations. The most widely used method is hydro-thermal synthesis, which typically requires a long reactiontime (>24 h), high temperatures (50-300 �C), and highpressures (>4 atm) to produce hydroxyapatite, which arecompletely incompatible processing conditions for bio-logical systems. Hydrothermal synthesis can be adaptedto precipitate at room temperature and atmosphericpressure, but extreme pH is required, also promptingbiological protein denaturation and degradation.Mecha-nochemical synthesis requires the use of high shear ratesthat would tear and break down any soft biologicalmatter present. Sol-gel synthesis requires a solid statethermal treatment at high temperatures (>900 �C) tocrystallize precursor powder, conditions yet again pro-voking breakdown of biological matter. The phase tran-sition approach often requires the use of high pH (>8)and long processing times (>24 h) and high temperatures(>60 �C) to attain complete conversion. The use ofsimulated body fluid has become the most popular meth-od of biomimesis; although the process is biomimetic, ithas a fatal flaw in the innate long reaction time (>7 days)and low yield (, 0.1 g per 100 mL reaction), not con-ducive to application outside of research. Additionally,these methods often employ additional additives such aschelating agents and surfactants, which are necessary toremove prior to any biological exposure because of theirtoxicity.

The empirical, nonpredictive approaches used for hy-droxyapatite synthesis have an opportunity for improve-ment through the use of thermodynamic modeling.Thermodynamic modeling has previously been usedin designing successful synthesis approaches in lead tita-nate34,35 and perovskites.36 By modeling hydroxyapatitesolution synthesis, simple precursor inflows can be de-signed into the system and secondary reaction additivessuch as pH agents, surfactants, and chelating agents andtheir effects can be eliminated. It will also help exploremodifications to systems in the literature so they canbetter fit the biomimetic precursor paradigm. Manycalcium and phosphate precursor combinations havebeen used in solution precipitation of hydroxyapatite;however, many of the individual precursors can be elimi-nated on the basis of chemistry and their failure to fit thebiomimetic precursor paradigm. Ions that are the mostprevalent in human blood plasma are Naþ, Cl-, and Kþ

(142, 103, and 5 mmol/dm3, respectively).37 Precursordesign should take this into account, eliminating theuse of iodides, hydroxides, nitrates, and fluorides, toname a few.Three precursor systems that are widely used in litera-

ture to synthesize hydroxyapatite via solution precipita-tion at room temperature are CaCl2-Na3PO4,

38 CaCl2-K2HPO4,

39-42 and Ca(OH)2-H3PO4.43-45 All have ex-

perimentally used a caustic agent to maintain a pH above9 throughout the reaction, but through the use of thermo-dynamic modeling, it may be possible to find reactionconditions that adhere to the biomimetic precursor para-digm. Following this paradigm further, this analysis willalso omit surfactants, chelating agents, and other addi-tives typically used for the above referencedworks but notfound in biological systems. Excluding these componentsand their potential thermodynamic effects also allowscomparisons between the individual precursor systems.The CaCl2 systems give information about how theprecursor interacts with two different phosphate sourceswhile the CaCl2-Na3PO4 and Ca(OH)2-H3PO4 systemsmay show the effects of using an orthophosphate source.In addition to these previously published chemistries, theCa(C2H3O2)2-K3PO4 system models will be presentedbecause it has not previously been investigated experi-mentally in the literature and will give additional com-parisons for an orthophosphate model (they areanalogous to Na3PO4) and its interactions with a highly

(34) Cho, S.-B.; Noh, J.-S.; et al. J. Eur. Ceram. Soc. 2003, 23(13), 2323.(35) Lencka, M. M.; Riman, R. E. J. Am. Ceram. Soc. 1993, 76(10),

2649.(36) Lencka,M.M.;Riman,R.E.Ferroelectrics 1994, 151(1-4), P1159.(37) Kokubo, T.; Kushitani, H.; Sakka, S.; Kitsugi, T.; Yamamuro, T.

J. Biomed. Mater. Res. 1990, 24, 721.(38) Liu, Y.; Hou, D.; Wang, G. Mater. Chem. Phys. 2004, 86, 69.(39) Wang, Y.; Zhang, S.; et al. Mater. Lett. 2006, 60(12), 1484.(40) Wang, F.; Li, M.-S.; et al. Mater. Chem. Phys. 2006, 95(1), 145.(41) Chou, Y.-F.; Chiou, W.-A.; et al. Biomaterials 2004, 25, 5323.(42) Feng, W.; Mu-Sen, L.; Yu-Peng, L.; Sheng-Song, G. J. Mater. Sci.

2005, 40, 2073.(43) Roeder, R. K.; Converse, G. L.; et al. J. Am. Ceram. Soc. 2006, 89

(7), 2096.(44) Mostafa, N. Y. Mater. Chem. Phys. 2005, 94, 333.(45) Kumar, R.; Prakash, K. H.; Cheang, P.; Khor, K. A. Langmuir

2004, 20(13), 5196.

38 Chem. Mater., Vol. 22, No. 1, 2010 Mossaad et al.

soluble calcium precursor. Acetates are not native tohuman blood plasma; however, they are a natural meta-bolite of alcohol found in the bloodstream, so it isconsidered to fit the biomimetic precursor paradigm aswell. Precursor systems that were modeled are shown inTable 1. The aqueous equilibrium pHs of the precursorsolutions also show the trend in forming hydroxyapatiteover other phases based on chemical selection. At leastone basic component is necessary to form hydroxyapa-tite, as pH typically drops during crystallization due toconsumption of hydroxyl ions in solution. The use of abasic solution does not fit the biomimetic precursorparadigm requirements, but a fast reaction that producesa neutral pH after 4 h could still be promising.Current methods of solution precipitation of hydro-

xyapatite employ harsh precursors and conditions thatlimit their true biomimetic qualities. The objective ofthis paper is to investigate the thermodynamic modelsof these systems and how they may be exploited toimprove hydroxyapatite solution synthesis methods.The biomimetic precursor paradigm has been outlinedto set guidelines for precursor solution design. Fourhydroxyapatite synthesis systems will be thermodynami-cally modeled to evaluate their biomimetic potential. Oneor more these four systems will be chosen for experimen-tal validation based on the results of the thermodynamicmodeling and its fit with the biomimetic precursor para-digm. It is hypothesized that if thermodynamics governthe solution synthesis methods of hydroxyapatite, thenthe proposed simulation approach will prove effective inpromoting biomimetic process design.

2. Thermodynamic Computations

2.1. Design. Thermodynamic modeling methods usedin this work have been previously described in detail.46

The focus of these models, simulated using the OLIStream Analyzer 2.0, will be to describe the relevantsimulations and their relation to the biomimetic precursorparadigm proposed. The thermodynamic databases thatwere specified for these simulations were the public,geothermal, and ceramic databases within the streamanalyzer 2.0 from OLI systems, Inc.47

In the thermodynamic design of a biomimetic system,the main variable parameters that impact the model arerelated to precursor selection, stoichiometry, and tem-perature, all which affect pH. Temperature for modelingwas fixed at 25 �C, while the titrants used in the chemicalstability diagrams were chosen as HNO3 and KOH,which are not biomimetic, but acidic and basic inflowsare necessary to generate this model, but not needed forreal experimentation. Precursor inflows were chosen andfixed (other than where noted) at 0.5m calcium and 0.3mphosphate to model the systems at a relatively concen-trated reaction for high yield, but maintaining stoichio-metric proportions of Ca/P = 1.67. A summary of thecomputations for the various precursor systems is shownin Table 1. Precursor dependent equilibrium reactionsand speciation that take place in the synthesis of hydro-xyapatite are presented in Table 2. All aqueous and solidspecies are labeled with an (aq) or an (s), while the ionicspecies show their corresponding charge.Four types of diagrams were used to evaluate the four

precursor systems and their potential for the biomimeticprecursor paradigm. These diagrams were namely (a)stability (b) yield (c) temperature survey and (d) composi-tion survey diagrams. For diagrams, (a) and (b) theshaded region indicates a 99% yield of hydroxyapatite,and the incipient precipitation boundary of hydroxyapa-tite is shown with a solid line. Any aqueous or ionicspecies present is denoted with a dotted line, whilesecondary phase fields are shown with a dashed ordashed-dotted line.The stability diagram shown in Figures 1-4 in position

(a) correlates equilibrium pH versus log molal concentra-tion ratio (Ca/P=1.67), which is effective in showing howequilibrium variable pH affects phase stability at a fixedinflow precursor ratio. Yield is also shown with a shadedregion on the diagram for a single selected phase, which inthis case is hydroxyapatite.The yield diagram shown in Figures 1-4 in position (b)

shows phase fields produced at different precursor inflowratios of phosphate versus calcium in molal concentra-tions at a fixed temperature (25 �C). The shaded regionindicates a 99% yield of hydroxyapatite precipitate. Thetemperature and composition surveys that are also pre-sented have been mapped on this yield diagram as a point(I) for temperature survey conditions and a line (II) forcomposition survey conditions.The temperature survey shown in Figures 1-4 in

position (c) shows phase output, theoretical output yieldin moles, and equilibrium pH across the specified tem-perature range (25-60 �C), which brackets the range ofthat of the biomimetic precursor paradigm. Thissurvey summarizes the changes in the yield diagram at aset point over a temperature range with additional quan-titative yield and pH information. The precursor concen-trations in all temperature surveys were conducted at0.5/0.3 m (Ca/P = 1.67). The phases that are producedat 25 �C are directly shown within the correspondingphase field on the yield diagram (labeled point (I) inFigures 1-4 in position (b)).

Table 1. Computed Phase Equilibrium Survey for Ca-P Precursor

Systems Considered for the Biomimetic Precursor Paradigma

precursor (pH) K2HPO4 (9.00) H3PO4 (1.32)Na3PO4/K3PO4

(12.13/12.24)

CaCl2 (6.73) N N HCa(OH)2 (12.39) H H HCa(C2H3O2)2 (8.92) N N H

aFixed conditions (Ca/P = 1.67, 0.5 m Ca-precursor and 0.3 mP-precursor solutions used for all computed equilibria (post-mixingconcentrations) and 25 �C). H denotes that hydroxyapatite is thedominant phase in the equilibrium diagram, and N indicates hydro-xyapatite was not dominant. Predicted pH of the respective singlecomponent Ca- or P-precursor is at the same bulk concentration asthe mixed Ca-P precursor solution

(46) Riman, R.; Suchanek, W.; et al. Ann. Chim. Sci. Mater. 200227(6), 15.

(47) http://www.olisystems.com


The composition survey (d) shows the phase output,theoretical yield, and pH with variable phosphate con-centration (0-1 m) at a fixed calcium concentration (0.5 munless otherwise indicated). This is a rerepresentation of theyield diagram, but with quantitative yield and pH at the setcalcium concentration. The showndominant phase changeson the composition survey are directly a result of crossingincipient precipitation boundaries on the yield diagram aslabeled line (II) in Figures 1-4 in position (b).A general way to read each set of figures is to start with

the stability diagram to target the desired phase to be syn-thesized, in this case hydroxyapatite, and the pH/solutionconcentration boundaries of the phase field. For the bio-mimetic precursor paradigm, this would be a pH between6.8 and 7.6. Since the stability diagram requires simulatedtitration of an acid and base, mentioned earlier, to gene-rate the model, the true equilibrium solution pH can bemapped directly to the temperature survey (c) at 25 �C.Depending on precursor selection, this pH range may ormay not fall within the hydroxyapatite incipient precipi-tation boundary on the stability diagram (a). Next,reference the yield diagram and point (I) that correspondsto the conditions modeled in temperature survey (c),

taking note how far inside the phase field the fixed Ca/Pprecursor ratio falls. The span of precursor ratios capableof producing hydroxyapatite is ameasure of process robust-ness; in this case, it will be measured by an estimatedproportion of the yield diagram covered with a shaded99% hydroxyapatite yield. Synthesis process robustnessis important when a process is scaled up to an industrialmagnitude because these processes can have some degreeof experimental error for the selected precursor ratios yet stillproduce phase pure hydroxyapatite. Finally, referencing thecomposition survey (d) at the fixed 0.5m calcium concentra-tion showsandverifies thephases shownon theyielddiagramalong with the pH change seen as stoichiometry crosses thephase boundaries shown in the yield diagrams (Figures 1-4,position (b)) along the composition survey line (II).2.2. Thermodynamic Models. The CaCl2-Na3PO4-

H2O, CaCl2-K2HPO4-H2O, Ca(OH)2-H3PO4-H2O,and Ca(C2H3O2)2-K3PO4-H2O precursor systems areshown in Figures 1-4. The Ca(OH)2-K3PO4-H2O andCa(OH)2-K2HPO4-H2O models are not presented be-cause the final equilibrium pH of these systems wasgreater than 12 in both cases, eliminating them fromconsideration based on the biomimetic precursor paradigm.

Table 2. Relevant Chemical Equilibria for Selected Precursor Systems Used for Synthesizing Hydroxyapatite

H2O(l)=Hþ þ OH- Ca3(PO4)2(s)=3Ca2þ þ 2PO43-

H2O(g)=H2O(l) CaH2(PO4)2 3H2O(s)=Ca2þ þ 2H2PO4- þ H2O

HNO3(g)=HNO3(aq) CaH2(PO4)2(s)=Ca2þ þ 2H2PO4-

HNO3(aq)=Hþ þ NO3- CaH2PO4

þ=Ca2þ þ H2PO4-

HCl(g)=HCl(aq) CaHPO4 3 2H2O(s)=Ca2þ þ HPO42- þ 2H2O

HCl(aq) =Hþ þ Cl- CaHPO4(aq)=Ca2þ þ HPO42-

Na2HPO4 3 12H2O(s)=2Naþ þ HPO42- þ 12H2O CaHPO4(s)=Ca2þ þ HPO4

2-

Na2HPO4 3 2H2O(s)=2Naþ þ HPO42- þ 2H2O Ca(OH)2(s)=Ca2þ þ 2OH-

Na2O(s) þ 2Hþ=2Naþ þ H2O CaOHþ=Ca2þ þ OH-

NaC2H3O2 3 3H2O(s)=Naþ þ C2H3O2- þ 3H2O CaPO4

-=Ca2þ þ PO43-

NaC2H3O2(aq)=Naþ þ C2H3O2- CaO(s) þ2Hþ=Ca2þ þ H2O

NaC2H3O2(s)=Naþ þ C2H3O2- Ca(C2H3O2)2 3H2O(s)=Ca2þ þ 2C2H3O2

- þ H2ONaOH 3H2O(s)=Naþ þ OH- þ H2O Ca(C2H3O2)2 3 2H2O(s)=Ca2þ þ 2C2H3O2

- þ 2H2ONaOH(s)=Naþ þ OH- Ca(C2H3O2)2(aq)=Ca2þ þ 2C2H3O2

-

Na2HPO4 3 7H2O(s)=2Naþ þ HPO42- þ 7H2O Ca(C2H3O2)2(s)=Ca2þ þ 2C2H3O2

-

Na2HPO4(s)=2Naþ þ HPO42- CaC2H3O2

þ=Ca2þ þ C2H3O2-

Na3PO4 3H2O(s)=3Naþ þ PO43- þ H2O Ca2C12O 3 2H2O(s) þ Hþ= 2Ca2þ þ 2Cl- þ OH- þ 2H2O

Na3PO4 3 6H2O(s)=3Naþ þ PO43- þ 6H2O CaCl2 3H2O(s)=Ca2þ þ 2Cl- þ H2O

Na3PO4 3 8H2O(s)=3Naþ þ PO43- þ 8H2O CaCl2 3 2H2O(s)=Ca2þ þ 2Cl- þ 2H2O

Na3PO4(s)=3Naþ þ PO43- CaCl2 3 4H2O(s)=Ca2þ þ2Cl- þ 4H2O

Na4P2O7 3 10H2O(s)=4Naþ þ P2O74- þ 10H2O CaCl2 3 6H2O(s)=Ca2þ þ 2Cl- þ 6H2O

Na5P3O10 3 6H2O(s) þ 2OH-=5Naþ þ 3PO43- þ 6H2Oþ2Hþ CaCl2(aq)=CaClþ þ Cl-

Na5P3O10(s) þ 2H2O(s)=5Naþ þ 3PO43- þ 4Hþ CaCl2(s)=Ca2þ þ 2Cl-

NaH2PO4 3H2O(s)=Naþ þ H2PO4- þ H2O CaClþ=Ca2þ þ Cl-

NaH2PO4 3 2H2O(s)=Naþ þ H2PO4- þ 2H2O Ca(NO3)2 3 3H2O(s)=Ca2þ þ 2NO3

- þ 3H2ONaH2PO4(s)=Naþ þ H2PO4

- Ca(NO3)2 3 4H2O(s)=Ca2þ þ 2NO3- þ 4H2O

Na3PO4 3 0.25NaOH 3 12H2O(s)=3.25Naþ þ PO43- þ 0.25

OH- þ 12H2OCa(NO3)2(s)=Ca2þ þ 2NO3

-

K2HPO4 3 3H2O(s)=2Kþ þ HPO42- þ 3H2O CaNO3

þ(s)=Ca2þ þ NO3-

K2HPO4 3 6H2O(s)=2Kþ þ HPO42- þ 6H2O Ca8H2(PO4)6 3 5H2O(s)=8Ca2þ þ2Hþ þ 6PO4

3- þ 5H2OK2HPO4(s)=2Kþ þ HPO4

2- Ca5(PO4)3OH(s)=5Ca2þ þ 3PO43- þ OH-

K3PO4 3 3H2O(s)=3Kþ þ PO43- þ 3H2O Ca5(PO4)3Cl(s)=5Ca2þ þ 3PO4

3- þ Cl-

K3PO4 3 7H2O(s)=3Kþ þ PO43- þ 7H2O P2O7

4- þ H2O=2PO43- þ 2Hþ

K3PO4(s)=3Kþ þ PO43- H2P2O7

2-=Hþ þ HP2O73-

KH2PO4(s)=Kþ þ H2PO4- H2PO4

-=Hþ þ HPO42-

KC2H3O2(aq)=Kþ þ C2H3O2- H3P2O7

-=Hþ þ H2P2O72-

KOH 3H2O(s)=Kþ þ OH- þ H2O H3PO4(aq)=Hþ þ H2PO4-

KOH 3 2H2O(s)=Kþ þ OH- þ 2H2O H4P2O7(aq)=Hþ þ H3P2O7-

KC2H3O2(s)=Kþ þ C2H3O2- HP2O7

3-=Hþ þ P2O74-

K2O(s) þ 2Hþ=2Kþ þH2O HPO42-=Hþ þ PO4

3-

KCl(aq)=Kþ þ Cl- P4O10(s) þ 6H2O=4H2PO4- þ 4Hþ

KCl(s)=Kþ þ Cl- (C2H3O2)2(aq)=2C2H3O2(aq)KNO3(s)=Kþ þ NO3

- (C2H3O2)2(g)=(C2H3O2)2(aq)KOH(s)=Kþ þ OH- CH3COOH(aq)=Hþ þ C2H3O2

-

CH3COOH(g)=CH3COOH(aq)


The models using Na3PO4 are also not presented as theywere close to identical to the same models presented thatuse K3PO4.Thermodynamic evaluation of each system will be

based on temperature, pH, process synthesis robustness,and yield parameters, and systems notmeeting the desiredparameters will be eliminated from experimental valida-tion. The phase pure precipitation of hydroxyapatiteshould happen between 20 and 40 �C and a pH between7 and 7.8 to be consistent with the biomimetic precursorparadigm. The shaded area and corresponding processsynthesis robustness should cover at least half the yielddiagram indicating that a large variation in compositionwill yield the same phase with no secondary phase con-tamination. Finally, the yield demonstrated in the com-position survey should be at 99% yield of hydroxyapatitewhen the pH is in the targeted range of 7-7.8. The percentyield quoted for each system is calculated by taking theabsolute moles of hydroxyapatite yielded on the compo-sition survey when the pH is between 7 and 8 and dividingby the theoretical full yield of that system. The theoreticalfull yield is calculated by taking the molal quantity ofcalcium used in the composition diagram and dividing byfive calcium ions used in each mole of hydroxyapatite(Ca5(PO4)3OH); for most of the systems explored this is0.5 m/5=0.1 m.

2.2.1. CaCl2-Na3PO4-H2O Equilibria System. In theCaCl2-Na3PO4-H2O system, two solid phases are shown

to precipitate. Figure 1a shows a pH dependent precipita-tion of hydroxyapatite, the majority of this phase fieldoccurring above pH 7. In the 0.5/0.3 m Ca/P ratio, theequilibrium pH at room temperature is about 6.2 (fromFigure 1c), making chloroapatite the dominant phaseunder the desired conditions when temperature is be-tween 20 and 40 �C. The yield diagram in Figure 1b showsthe coprecipitation of chloroapatite and hydroxyapatitein over half the area where calcium concentrations areequal to or higher than the phosphate concentration,corresponding to conditions when the pH is lower than10. The phase pure hydroxyapatite phase field occurs inthe other half, dominantly where the system is in excess ofphosphate (Ca/P < 1.67). Concentrating on these phasepure conditions in light of the biomimetic precursorparadigm, the concentration survey shows that the domi-nant hydroxyapatite precipitation occurs above a phos-phate concentration of 0.35 m (calcium concentration of0.5m). This coincides with an equilibrium pH increase toover 8, which is outside the ideal biomimetic range.Precipitating very close to the single/double phase fieldboundary would target the pH to the proper rangebetween 6.5 and 8; however, it would limit the robustnessof the system to where the phosphate precursor concen-tration could not fluctuate below 0.3 m when the calciumprecursor concentration is kept at 0.5 m. Additionally,when the pH is kept in this range, the yield of hydro-xyapatite is 99% but a slight change in composition will

Figure 1. (a) Stability diagram (0.5 m CaCl2, 0.3 m Na3PO4) and (b) yield diagram with corresponding (I) point and (II) line of conditions surveyed.(c) Temperature survey at 0.5 m CaCl2 and 0.3 m Na3PO4 and (d) composition survey at 25 �C, 0.5 m CaCl2.


precipitate chloroapatite. The calculated yield is 76%which corresponds to a calcium concentration of 0.5 mand a phosphate concentration of 0.35 m which yields apH of 8. These secondary phase precipitation and loweryield limitations indicate this system is not a good fit forthe biomimetic precursor paradigm, eliminating it as acandidate for experimental validation.

2.2.2. CaCl2-K2HPO4-H2O Equilibria System. Whenexamining theCaCl2-K2HPO4-H2Omodels inFigure 2, itis evident that there is a limited hydroxyapatite phasestability field within this precursor system. The stabilitydiagram in Figure 2a shows a pH dependent behavior ofhydroxyapatite precipitation with an additional chloroapa-tite phase field coinciding with monetite between a pH of 1and 7, whereas hydroxyapatite is phase pure above pH 7.The temperature survey (Figure 2c) for the 0.5/0.3 mconcentration yields no hydroxyapatite over the entire20-60 �C range due to the formation of acidic HCl speciesin solution driving the pH far below its stability of 6.5-12,prompting the formation of monetite and chloroapatite.The composition survey in Figure 2d conducted at 0.125 mCaCl2, where hydroxyapatite is stable, shows an increase inequilibriumpH to values close to 7 through the use of excessphosphate above 0.34 m. The lower calcium concentration(0.125m) used in this composition survey was chosen so thesurvey line on the yield diagram would intersect the hydro-xyapatite phase region of interest. The yield diagram inFigure 2b further emphasizes the limited stoichiometry

ranges for the synthesis of phase pure hydroxyapatiteshowing only a small shaded region (∼20% of the totalarea) when the precursor system is kept in extreme excess ofphosphate (Ca/P, 1.67, Ca< 0.2m, PO4> 0.25m). Theresulting robustness of this system is very low, only produ-cing high yield phase pure hydroxyapatite in a small rangeof stoichiometries. At these stoichiometries, the yield is99%, which corresponds to a calcium precursor concentra-tion of 0.125 m and phosphate precursor concentration of0.5 m, producing a pH of 7. Note how far the temperaturesurvey point (I) is away from the hydroxyapatite phasestability region, emphasizing how drastically the stoichiom-etry has to be changed to yield the targeted phase. The lowrobustness of the system alone eliminates this as a candidatefor further experimental investigation.

2.2.3. Ca(OH)2-H3PO4-H2O Equilibria System.Mov-ing away from chloride based chemistries, the investigationof the Ca(OH)2-H3PO4-H2O system (Figure 3) yields adifferent set of thermodynamic issues when looking forprecursor systems to fit the paradigm. The stability diagramin Figure 3a shows the same pH dependent behavior seen inthe previous models indicating this phenomenon is notunique to the precursor systems employed. In each of thethree systems explored so far, the shaded region dimensionsspan from pH of about 6.5 to 14 and log[m(PO4) =m(Ca)/1.6667] of -4.2 to 0. There is a change in the shape of thesecondary monetite phase field, showing an overlap andcoprecipitation with hydroxyapatite around a pH of 5.25.

Figure 2. (a) Stability diagram (0.5 m CaCl2, 0.3 m K2HPO4) and (b) yield diagram with corresponding (I) point and (II) line of conditions surveyed.(c) Temperature survey at 0.5 m CaCl2 and 0.3 m K2HPO4 and (d) composition survey at 25 �C, 0.125 m CaCl2.


The yield diagram in Figure 3b indicates no area wherehydroxyapatite is the only dominant phase at 99% yield,simply because this precursor system has a Ca/P ratio of1.6667 that falls just outside the shaded region adjacentto the monetite field on the stability diagram. Whencalculating percent yield from the composition survey,however, it is found that 99% yield may exist with acalcium precursor inflow of 0.5 m and phosphate pre-cursor inflow of 0.3 m corresponding to a pH of 8.However, any deviation away from this phosphate pre-cursor concentration will coprecipitate an undesiredphase. The small triangular area where temperature sur-vey point (I) is located on the yield diagram in Figure 3byields hydroxyapatite as the only phase precipitated;however, it is obvious this system is very sensitive toprecursor ratio changes. A shift of Ca/P above 1.75 orbelow 1.0 will yield a coprecipitation field, severely limit-ing the robustness of this system and carrying it outsidethe desired pH range. Although hydroxyapatite is thedominant phase at 0.5/0.3 m over the entire temperaturerange of 20-60 �C in Figure 3c, the equilibrium pH, is9.1, which is above the range for the paradigm. However,looking at the composition survey, using a slight excess ofphosphate (Ca/P = 1.55-1.65) may serve to target theinitial pH in the biological range of 7.1-7.6, but thisrange must be maintained throughout synthesis so as notto precipitate secondary phases in this narrow region oftargeted phase stability. On the basis of low yield and lack

of robustness when in accordance to the biomimeticprecursor paradigm, this system was eliminated fromexperimental validation.

2.2.4. Ca(C2H3O2)2-K3PO4-H2O Equilibria System.In the previously unexplored Ca(C2H3O2)2-K3-PO4-H2O system shown in Figure 4, some differentthermodynamic characteristics are observed. The stabi-lity diagram shows the same pH sensitivity as the otherchemistries explored; however, it appears to have aslightly more narrow pH region starting at 7 rather thanaround 6.5, so further analysis was done to see how theshaded 99% yield phase field may move when stoichiom-etry is altered (Ca/P = 1.47-1.6833) in this system. Asshown in Figure 5a, the shaded hydroxyapatite precipita-tion region recedes with a decrease in the phosphateprecursor concentration (increase in Ca/P ratio), thelargest area being with an excess of phosphate at the1.47 Ca/P ratio, or 0.34 m K3PO4. It is likely that thisreceding phenomenon is also seen in the other chemistriesas well. The yield diagram in Figure 4b shows a protrud-ing monetite phase field and its recession with highertemperatures, being completely eliminated by 40 �C. It ispredicted that even inside this phase field hydroxyapatitewill be a dominant coprecipitate, making this the onlydiagram explored that can yield hydroxyapatite over theentire range of precursor ratios (Ca=0.001-1m, PO4=0.001-1m). Using an excess of phosphate or an excess ofcalcium will yield phase pure hydroxyapatite outside this

Figure 3. (a) Stability diagram (0.5 m Ca(OH)2, 0.3 m H3PO4) and (b) yield diagram with corresponding (I) point and (II) line of conditions surveyed.(c) Temperature survey at 0.5 m Ca(OH)2 and 0.3 m H3PO4 and (d) composition survey at 25 �C, 0.5 m Ca(OH)2.


monetite coprecipitation field; however, only an excess ofphosphate will produce 99% yield and reach the targetedbiomimetic pH range (mapped from Figure 4d). Whenthe same model is calculated at 37 �C (Figure 5b), themonetite phase field retracts to a very small area pro-duced at high concentrations of precursors. The 0.5/0.3mtemperature survey indicates that hydroxyapatite is notphase pure at room temperature, but by 31 �C, it is thedominant precipitate. The equilibrium pH is low for theapplication to a biological system, but the compositionalsurvey indicates that the increase of phosphate concen-tration to 0.39 m raises the equilibrium pH to 7.5 and

produces phase pure hydroxyapatite at 99% yield. Therobustness of the system could be considered comparableto that of the CaCl2-Na3PO4-H2O system; however,crossing out of the 99% yield shaded region in a phos-phate-deficient formulation when the temperature isbrought up to 37 �C, hydroxyapatite is still a phase pureprecipitate, giving this system an advantage. Evenwhen compared at 25 �C, the area of the monetitephase field in this system ismuch smaller to design aroundthan the coprecipitating chloroapatite phase field inthe CaCl2-Na3PO4-H2O precursor system. Given theadvantages in robustness and the smaller secondary

Figure 4. (a) Stability diagram (0.5mCa(C2H3O2)2, 0.3mK3PO4) and (b) yield diagramwith corresponding (I) point and (II) line of conditions surveyedand 1-6 points of validation. (c) Temperature survey at 0.5 m Ca(C2H3O2)2 and 0.3 mK3PO4 and (d) composition survey at 25 �C, 0.5 m Ca(C2H3O2)2.

Figure 5. (a) Stability diagram with corresponding adjustments in stoichiometry and its effects on 99% yield area and (b) yield diagram at 37 �C.


coprecipitation phase field, this system was chosen forexperimental model validation.

3. Experimental Validation

3.1. Procedure. Six points were chosen on the diagrampresented in Figure 4b, three within the monetite phasefield, and three at other stoichiometries within the pre-dicted phase pure hydroxyapatite field within and outsidethe region of 99% yield. The synthesis of the powder wascarried out using calcium acetate hydrate (99%, AcrosOrganics, Morris Plains, NJ) and potassium phosphatetribasic monohydrate (96%, Acros Organics, MorrisPlains, NJ) as received with no further purification ortreatment. The concentrations of the precursor solutionstested are presented in Table 3 and are expressed in pre-mixing concentrations. For example, an experimental1.0 m solution of calcium acetate would be 0.5 m aftermixing (hereto referred to as a post-mixing concen-tration) with a corresponding equal volume phosphatesolution. A stock solution (SS) of calcium acetate anda separate stock solution of potassium phosphate tribasicwere made by dissolving the appropriate amount ofprecursor in DI water (deionized water, Milli-Q biocellwith RiOs-S, Millipore Systems, Inc., Billerica, MA) andfiltering through a 0.2 μm PES membrane Nalgene filter(566-0020). An equal volume (50 mL) of each solutionwas measured out, and the phosphate solution was addedto the calcium solution. The resulting viscous liquid wasthen stirred vigorously until the solution formed thixo-tropic gel turned white and reduced in viscosity in about30 s of mixing. This solution was aged for 4 h, centrifuged(Avanti J-26XP, Beckman Coulter, Fullerton, CA),washed (where indicated with 150 mL DI water eachcycle), shell frozen (Labconco, Kansas City,MO), freeze-dried (Dura-Dry μP, FTS Systems, Inc., Stone Ridge,NY), and desiccated (Drykeeper, Sanplatec Corp, Osaka,Japan, humidity 30%) until ready for characterization.Powders that were fired were subjected to a 900 �C heattreatment for 2 h in air to verify complete reaction ofthe precursors and reagent concentrations (2 h ramp to900 �C, 2 h hold, 6 h ramp down).Processing pH was measured versus reaction time for

1-4 h (Table 4) for comparison to predicted equilibriumpH at the various Ca/P stoichiometries. It was notablethat even after 4 h, none of the Ca/P stoichiometries hadpH values that fell as low as the predicted equilibrium pHin Table 4. For example, sample 5 was computed to havean equilibrium pH of 5.8 while the experimental pH aftera 4 h reaction time was 9.0. Thus, these data indicate that

at the end time of synthesis the systems shown in Table 4were still approaching equilibrium.When the precipitate was washed, the pH was noticed

to decrease by 2( 0.5 pHpoints compared to prewashing.On the basis of these results, when it is important toensure phase purity in future studies, pH during post-synthesis processing should be closely monitored or thepowder should bewashed in buffered solutions to preventthe formation of undesired phases. The pH of the residualwashing medium after the powder has been centrifugedout is described as “pH after washing.” The samplenomenclature used is as follows: as-prepared (AP),as-prepared heat treated (APHT), as-prepared washed(APW), and as-prepared washed and heat treated(APWHT).3.2. Characterization. Reaction products for the Ca-

(C2H3O2)2-K3PO4-H2O system were characterized viaX-ray diffraction (XRD) (Kristalloflex D-500, SiemensAnalytical X-ray Instruments, Madison, WI., 30 mA,40 kV, Cu KR, 2-theta= 20-60�, 0.03 step, 2 s dwell).Jade 8.0 Software (MDI, Livermore, CA) was used to aidin phase identification with PDF numbers 01-074-0566(hydroxyapatite), 01-077-0128 (monetite), 97-002-2239(calcium potassium phosphate (V) alpha), 01-070-0364(M tricalcium phosphate, R-TCP), and 01-070-2065(R tricalcium phosphate, β-TCP).3.3. Results and Analysis. The results of the XRD

validation are presented in Figures 6 and 7. Theas-synthesized powders in both the washed and the un-washed conditions (Figures 6a, 7a) yielded hydroxyapa-tite as the dominant phase in most cases. Looking at eachof the six points tested, three were predicted to producephase pure hydroxyapatite (points 2-4) and three werepredicted to produce hydroxyapatite and monetitecoprecipitate (points 1, 5-6). All as-prepared powderproduced hydroxyapatite as predicted but only AP6produced monetite as predicted (2-theta=26.6�). Whatappeared inconsistent with the diagrams were the finalpHs of the as-prepared powders, possibly suggesting thatthere is a kinetic time component that plays an integralpart, as the pHs were significantly higher than predic-ted, but the reaction was stopped at 4 h, long beforeequilibrium was reached. AP2, although with a startingpH of 9.5, higher than the paradigm allows, was the onlystoichiometry experimentally investigated that seemedto fit the biomimetic model post-synthesis (pH = 7.2)

Table 3. Summary of Six Reaction Conditions Chosen from Figure 4d

sampleID

computedCa/P ratio

calciumSS (m)

phosphateSS (m)

post-mixingCa/P concentrations (m)

1 1.67 1.25 0.75 0.625/0.3752 4.00 1.00 0.25 0.500/0.1253 0.50 0.50 1.00 0.250/0.500

4 1.33 1.00 0.75 0.500/0.3755 1.67 1.00 0.60 0.500/0.3006 1.50 1.50 1.50 0.750/0.500

Table 4. Measured pH of Samples 1-6 as a Function of Reaction Time,

Post-Wash pH after 2 Washing Cycles, and Computed Equilibrium pH

Sample ID

time (h) 1 2 3 4 5 6

0 11.0 9.5 14.0 10.0 11.0 12.01 9.5 9.0 14.0 10.0 9.0 9.52 9.5 8.5 14.0 10.0 9.0 9.53 9.0 7.2 14.0 9.5 9.0 9.54 8.5 7.2 14.0 9.5 9.0 9.0

post-wash (2 cycles) 6.9 <6.5 12.0 7.2 6.9 6.9

equilibrium pH 5.9 5.8 12.0 7.2 5.8 6.2


without the use of washing, additives, or buffers.With theuse of an even more calcium-rich ratio, it may have aneven lower starting pH (targeted in the range of 7.4) and ashorter reaction time, although a smaller yield, beingoutside the 99% shaded region.Although the final synthesis pHs are still above the

desired target, it is notable that 1-2 washing cyclesdiluting the synthesis liquid threefoldwas enough to bringthe pH of all but one system down to the targetedpH range of 7-7.8. In comparison, the other chemicalsystems modeled in this study utilize more extreme pHconditions, requiring a minimum of three washing cyclesthat dilute the synthesis liquid volume by up to 10-foldeach time. The lower volumes of severe washing con-ditions used for the acetate system give it a considerableadvantage over other precursor systems for large scaleproduction of hydroxyapatite, since fewer washesand lower volumes of washing media are required. Theonly unexpected result was that after washing, APW2,an extremely phosphate-deficient formulation, wasconverted into monetite. This result was consistent withpH/phase predictions in the stability diagram, associated

with the solution pH dropping below 6.5 in thepost-washing step shown in Table 4. This suggests it isimperative to monitor and control pH within the pHstability range of hydroxyapatite while washing thesepowders to maintain these powders as hydroxyapatite.It is useful to note that the model-predicted synthesis ofhydroxyapatite over the selected regions of the equili-brium diagram did hold true, emphasizing that thischemistry has a much higher degree of robustness thanany previously explored.To determine the presence of incomplete crystallization

and nonstoichiometric hydroxyapatite, heat treatmentsof 900 �C for 2 hwere done on bothwashed and unwashedpowders to observe any high temperature phase develop-ment. Upon heat treating the AP powders, all phasestransformed into dominantly calcium potassium phos-phate (V) alpha (KCP) with a small amount of hydro-xyapatite except AP2, which remained hydroxyapatitewith no detectable secondary phases, indicating thiscalcium rich ratio produces stoichiometric HA, whilethe others are all nonstoichiometric. KCP is phasepredicted to form at high temperatures in traditional

Figure 6. XRD of (a) unwashed powders as-synthesized and (b) unwashed heat treated powders showing a phase development of calcium potassiumphosphate(V) alpha.

Figure 7. XRD of (a) washed powders and (b) washed and heat treated powders demonstrating sensitive secondary phase developments that heavily relyupon washing treatments.


ternary and binary diagrams using a calcium and phos-phate source with a potassium component.48 It can bespeculated that leaving the potassium ions on the surfaceof the particles adjusted the phase stability of hydroxya-patite at high temperatures, prompting the crystal struc-ture to incorporate a significant potassium contingent butremain hexagonal. Hydroxyapatite is known to incorpo-rate large amounts of impurities, so it is expected withhigher rates of diffusion at this high temperature for thisphase to develop. It would be interesting to see if adifferent cation doped calcium phosphate phase developsat high temperature when using an alternative tribasicphosphate source such as sodium orthophosphate orammonium orthophosphate. Moreover, since tricalciumphosphate was not formed, it can be speculated that therewas no amorphous phase left at the end of reaction. Twobrief washing cycles with DI water (Figure 7a-b) purgedthe systemof potassiumpreventing any calciumpotassiumphosphate development upon heat treatment. There wasan exception, however, with the previously mentionedmonetite that developed inAPW2due to a pHdrop, whichcaused a transformation upon firing into dominantlymonoclinic R-TCP. The APW5 did, however, developa small quantity of monetite or amorphous calcium phos-phate during or following the washing cycles, whichappears to have converted into a small amount of hexa-gonal β-TCP with firing (2-theta=30.7�). The other

APWHTs appear to have some small degree of phasetransformation to β-TCP and/or monetite, but a consi-derable portion of these samples remained hydroxy-apatite, even upon heat treatment. The amounts ofmonetite seen in the APWHTs can be likely ascribed tothe second washing cycle causing a drop in pH into themetastable range below 7.

4. Conclusions

Thermodynamic computations were effectively used toevaluate four important precursor systems for the synth-esis of hydroxyapatite using biomimetic precursor designcriteria. Theywere also effective in discovering processingconditions suitable for hydroxyapatite synthesis usingcalcium acetate and potassiumorthophosphate. The calciumacetate-potassium orthophosphate system demonstratedthe most robust processing conditions and has the abilityto synthesize hydroxyapatite over the entire range ofCa/P stoichiometries explored. The pH dependence and99% yield behavior demonstrated in the stability dia-grams for all four precursor systems appears to be uni-versal with respect to precursor choice when singlecalcium and single phosphate precursors are employedat a stoichiometric ratio of Ca/P=1.6667.

Acknowledgment. The authors would like to acknowledgeNASA GSRP Grant NNG04GO44H, the Rutgers-NSFIGERT DGE 0333196, and Osteotech, Inc., for their gener-ous funding of this research.

(48) ACerS-NIST phase Equilibria Diagrams [CD-ROM], version 3.0.1;2004.

Documents

Thermodynamic Modeling of Hydroxyapatite Crystallization with Biomimetic Precursor Design Considerations