The role of brushite and octacalcium phosphate in apatite formation.

Critical Reviews http://cro.sagepub.com/ in Oral Biology & Medicine

The Role of Brushite and Octacalcium Phosphate in Apatite Formation Mats S.-A. Johnsson and George H. Nancollas CROBM 1992 3: 61 DOI: 10.1177/10454411920030010601 The online version of this article can be found at: http://cro.sagepub.com/content/3/1/61

Published by: http://www.sagepublications.com

On behalf of: International and American Associations for Dental Research

Additional services and information for Critical Reviews in Oral Biology & Medicine can be found at: Email Alerts: http://cro.sagepub.com/cgi/alerts Subscriptions: http://cro.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav

>> Version of Record - Jan 1, 1992 What is This?

Downloaded from cro.sagepub.com at KANSAS STATE UNIV LIBRARIES on July 9, 2014 For personal use only. No other uses without permission.

Critical Reviews in Oral Biology and Medicine,

3(1/2):61-82 (1992)

The Role of Brushite and Octacalcium Phosphate in Apatite Formation Mats S.-A. Johnsson and George H. Nancollas Chemistry Department, State University of New York at Buffalo, Buffalo, NY 14124 ABSTRACT: Studies of apatite mineral formation are complicated by the possibility of forming several calcium The least soluble, hydroxyapatite (HAP), is preferentially formed under neutral or basic conditions. In more acidic solutions phases such as dicalcium phosphate dihydrate (Brushite, DCPD) and octacalcium phosphate (OCP) are often found. Even under ideal HAP precipitation conditions the precipitates are generally nonstoichiometric, suggesting the formation of calcium-deficient apatites. Both DCPD and OCP havea been implicated as possible precursors to the formation of apatite. This may occur by the initial precipitation of DCPD and/or OCP followed by transformation to a more apatitic phase. Although DCPD and OCP are often detected during in vitro crystallization, in vivo studies of bone formation rarely show the presence of these acidic calcium phosphate phases. In the latter case the situation is more complicated, since a large number of ions and molecules are present that can be incorporated into the crystal lattice or adsorbed at the crystallite surfaces. In biological apatite, DCPD and OCP are usually detected only during pathological calcification where the pH is often relatively low. In normal in vivo calcifications these phases have not been found, suggesting the involvement of other precursors or the formation of an initial amorphous calcium phosphate phase (ACP) followed by transformation to apatite.

phosphate phases.

KEY WORDS:

hydroxyapatite, crystallization, transformation, inhibition, amorphous calcium phosphate.

I. INTRODUCTION Calcium phosphates are the main constituents of the bones and teeth of vertebrates, including most hard tissues of humans. In addition, the formation of calcium phosphates is of importance in the environment because of their participation in areas such as the removal of phosphate from wastewater and the fate of elements, such as aluminum, iron, and other heavy metals, in the formation of lake and ocean sediments. Calcium phosphate precipitation may also be involved in the fixation of phosphate fertilizer in soils. Studies of the uptake of phosphate on calcium carbonate surfaces at low phosphate concentrations, typical of those in soils, reveal that the threshold concentration for the precipitation of the calcium phosphate phases from solution is considerably increased in the pH range 8.5 to 9.0 (Zawack et al., 1986). Earlier works suggested that the car-

bonate ion from calcite inhibits the nucleation of calcium phosphate phases (Kazmierczak, 1978; Koutsoukos, 1984). Although it has been described in terms of hydroxyapatite inhibition, the complete picture is much more complex, since stoichiometric hydroxyapatite is rarely found in vivo. It is now generally recognized that the crystallization of many sparingly soluble salts involves the formation of metastable precursor phases that subsequently dissolve as the precipitation reactions proceed. Thus, in the case of calcium phosphate, the picture is complicated by the relatively large number of phases, summarized in Table 1, that can participate in these reactions. Moreover, the presence in vivo of nucleotides, proteins, and inorganic constituents other than calcium and phosphate has a considerable influence on crystallization, making it difficult to predict possible phases that may form.

1045-4411/92/$.50 © 1992 by CRC Press, Inc.

61 Downloaded from cro.sagepub.com at KANSAS STATE UNIV LIBRARIES on July 9, 2014 For personal use only. No other uses without permission.

TABLE 1 Calcium Phosphate Phases

Empirical

Molar Ca/P ratio

CaHPO,2H20 CaHPO, Ca,H2(PO4)65H20 Ca3(PO,)2 Ca3,Mg(PO4)2 0 < v H3PO4 H+ + HPO2- -> H2PO4 H+ + PO- - HPO2CA2+ + H2PO4 - CaH2PO4 Ca2+ + HPO4- - CaHP04 Ca2+ + PO-- -, CaPO4 H+ + F- -- HF Ca2+ + F- -- CaF-

In Equation 2, Kso and IP are the solubility product and the ionic product, respectively, and v is the number of ions in a formula unit. The solubility products, Ko, of the different calcium phosphate phases are given in Table 3. The values of IP are given by Equations 3 to 6:

IPDCPD

=

[Ca2l ][HPO- ]222

(3)

IPocp

=

[Ca2+]8[H+] 2[PO3- ]6y28y6

(4) (5)

IPTC = [Ca2+]9[PO3-]6Y2Y3 IPHAP

=

]2 [Ca2+ ] [PO3-]6[OHYY2°Yy

(6)

in which the molar concentrations of the ionic species are enclosed in square brackets. The driving force for the crystallization of a particular phase is also often expressed in terms of a conventional relative supersaturation, or, by Equation 7 r

=

(IPI/v

-

Kv)/Ks/

= S - 1

(7)

in which S is the supersaturation ratio. The necessity for carefully taking into account ionic strength effects and for making activity coefficient corrections becomes increasingly apparent as the value of I increases. For instance, at an ionic strength of about 0.1 mol 1-', HAP ionic products may be in error by as much as a factor of 106 if activity coefficients are neglected. Fortunately, it can be seen in Equation 1 that the activity coefficients, at least in terms of this primitive model, depend only upon the ionic strength of the system, and most studies are made at constant ionic strengths close to physiological level

Thermodynamic data K= K= K= K= K= K= K= K=

1641 mol-' 1.52 x 1071mol-1 1.51 x 10121 mol-1 31.91 mol-1 681 mol I-1 3.46 x 1061 mol-' 4.47 x 10-4 1mol8.91 I mol-1

(0.15 mol 1-') so that the activity effects can be assumed constant during the precipitation reactions. Care must be exercised, however, in order to take into account ion-pair and complex formation of both calcium and phosphate species by foreign ions that may be present in the solutions. III. CRYSTAL STRUCTURE OF CALCIUM PHOSPHATES

Amorphous Calcium Phospate It is now generally agreed that, both in vitro and in vivo, precipitation reactions at sufficiently high supersaturation and pH result in the initial formation of an amorphous calcium phosphate with a molar calcium/phosphate ratio of about A.

1.5. The precipitate, with little long-range order, tends to consist of aggregates of primary nuclei, with a composition dependent on the conditions of formation, and hydrolyzes almost instantaneously to more stable phases. As discussed later, this has led to a recent proposal that two discrete forms of amorphous calcium phosphates, ACP1 and ACP2, can be identified. In general, ACP is a highly unstable phase and hydrolyzes almost instantaneously to more stable phases. In the presence of other ions or under in vivo conditions, ACP may persist for appreciable periods due to kinetic stabilization.

B. Dicalcium

Phosphate Dihydrate

The crystal structure of brushite or DCPD consists of chains of CaPO4 arranged parallel to


TABLE 3

Solubilities of the Calcium Phosphates DCPD DCPA TCP OCP HAP FAP

KsoOCPD K.oDCPA Kso-TCP

KwoFAP KaoFAP

1.87 x 10-7 (mol 1-')2 9.2 x 10-7 (mol 1-)2 2.8 x 10-2 (mol I-1)5 2.5 x 10-" (mol I-1)8 5.5 x 10-118 (mol I-1) 5.0 x 10-'23 (mol I- )9

each other (Young and Brown, 1982). Lattice water molecules are interlayered between the calcium and phosphate chains. Anhydrous calcium phosphate dihydrate or monetite (DCPA) is less soluble than DCPD due to the absence of water inclusions. This phase is rarely seen in vivo, although Young and Brown (1982) suggest that the apparent absence of this phase may be due to difficulties of detection as a consequence of its weak X-ray diffraction spectrum.

C. Tricalclum

Phosphate Several phases having a tricalcium phosphate

formula unit [TCP; Ca3 (PO4)2] are known (Koutsoukos, 1984; Nancollas, 1982). The most common, formed under physiological conditions, is whitlockite, which is stabilized by magnesium ions (Nancollas, 1989; LeGeros et al., 1986, 1989a). It has been identified during pathological calcification, such as dental calculus formation and in renal stones, but it has not been observed in enamel, dentin, or bone (LeGeros et al., 1983; Cheng et al., 1988). The ideal structure of PTCP contains calcium ion vacancies that are too small to accommodate a calcium ion but allow for the inclusion of magnesium ions, which thereby stabilize the structures (Dickens et al., 1974). Whitlockite is not formed under physiological conditions unless magnesium ions are

present. D. Octacalcium

Phosphate Octacalcium phosphate (OCP) has a layered structure involving apatitic and hydrated layers (Brown, 1962; Brown et al., 1962, 1987; Taves, 1963). The apatitic layers have calcium and phos-

phate ions distributed in a manner similar to that for HAP, while the hydrated layer contains lattice water and less densely packed calcium and phosphate ions. The water content of OCP crystals is about 1/5 that of DCPD, and this is partly responsible for its lower solubility. The similarity in crystal structure between OCP and HAP is one reason that the epitaxial growth of these phases is often observed (Young and Brown, 1982; Nelson et al., 1986a). Young and Brown (1982) suggest that the hydrated layer of the OCP unit cell may form an interphase between HAP and the surrounding solution. If this occurs, the epitaxial intergrowth of OCP and HAP is favored. E.

Hydroxyapatite

Crystals of HAP may have either monoclinic or hexagonal unit cells. In the more stable, monoclinic form, rows of phosphate ions are found along the a axis, with calcium and hydroxide ions localized between the phosphate groups (Posner et al., 1958; Kay et al., 1964). Calcium ions are located in two triangles surrounding two hydroxide ions and in larger hexagons surrounding these calcium positions. The hydroxide ions are situated either above or below the calcium ion planes. The rows of hydroxide ions are directed, alternately, upward or downward. The hexagonal form of HAP, generally found in biological apatites, has a structure similar to the monoclinic form, but with columns of calcium and hydroxide groups located in parallel channels (Kay et al., 1964). Ion substitution can readily occur in these channels, and this may account for the high degree of substitution found in natural apatites. In hexagonal HAP, the hydroxide ions are more disordered within each row, when compared with the monoclinic form, pointing either upward or downward in the structure. This induces strains that are compensated for by substitution or ion vacancy (Kay et al., 1964). The hexagonal HAP is therefore seldom a stoichiometric phase. IV. CALCIUM PHOSPHATE PHASE STABILITY General considerations

concerning the sta-


v

9

I

T

I

I

2k 3 4 . 5 I-

c0

6

DCPD

0 -J l

7

8.

OCP TCP

9 10

HAP A

3

4

5

6

7

8

9

10

pH FIGURE 1. Solubility isotherms of calcium at 37°C and 0.1 mol I-' ionic strength.

bility of calcium phosphate phases in contact with aqueous solutions can be understood in terms of a typical solubility phase diagram, such as that shown in Figure 1. In this figure, the solubility isotherms are expressed as plots of (log TcTp) as a function of pH. Here, Tca and Tp are the total molar concentrations of calcium and phosphate, respectively. Figure 1 has been constructed on the assumption that the solution contains equal total molar concentrations of calcium and phosphate ions at an ionic strength of 0.1 mol 1-'. The position of the curves and the singular points in Figure 1 will change if the ionic strength of the background electrolyte is varied. It can be seen that at pH above 4.0, HAP is the most stable phase, followed by TCP and OCP.

phosphate phases

At pH values lower than 4.0, DCPD is more stable than HAP. The variations in solubility with pH imply that a phase exposed to acidic conditions may be covered by a surface coating consisting of a more acidic calcium phosphate phase. The apparent solubility behavior will therefore be quite different from that of the original phase, and the measurement of solution concentrations might lead to unusually large estimated solubilities. Moreover, in determining the likelihood of the formation of preferred crystal phases in solutions supersaturated with respect to several different phases, kinetic factors are also important. The formation of HAP is much slower than that of either OCP or DCPD, and during simultaneous phase formation a larger portion of the kinetically 65


favored phase may be observed, even though it has a much smaller thermodynamic driving force. The balance between kinetic and thermodynamic factors is, therefore, very important in discussing the likelihood of precursor formation during calcium phosphate precipitation.

V. PHASE TRANSFORMATION OF CALCIUM PHOSPHATES The large number of calcium phosphate phases that may form (Table 1) and their regions of stability, depending upon lattice ion concentrations and pH, make it possible that apatite formation in vivo involves more acidic calcium phosphate phases that later transform to HAP. In an attempt to understand these events in vivo, numerous spontaneous precipitation studies have been made in vivo. The stoichiometric molar ratio of calcium phosphate calculated from changes in concentration during precipitation from solution is frequently in the range of 1.45 + 0.05, which is considerably lower than the value 1.67 required for the thermodynamically favored HAP. In order to reach the approximate HAP composition, extended solid/solution equilibration is required. At relatively high supersaturations, as stated above, and at pH values higher than 7, an amorphous calcium phosphate phase is formed (LeGeros et al., 1975, 1986; Boskey and Posner, 1976; Tung and Brown, 1983; Tomazic et al., 1987). It has been proposed that subsequent conversion to more stable calcium phosphate phases takes place by an autocatalytic solution-mediate crystallization process (Eanes et al., 1965; Boskey and Posner, 1974; Eanes and Posner, 1965, 1970; Heughebaert and Montel, 1984). Formation of HAP without the intial precipitation of more acidic phases is achieved in solutions supersaturated only with respect to HAP (Figure 1). In spontaneous precipitation studies at high supersaturation and pH, the initial highly hydrated cryptocrystalline ACP was detected by a gradual development of opalescence in the solution following induction periods (Termine and Eanes, 1972) that were dependent upon the concentration, ionic strength, and pH (Termine and Posner, 1970). It was shown that the induction periods for the formation of ACP and the life-

phase were increased at lower supersaturations (Termine et al., 1970). Christoffersen et al., (1989) described the initial precipitation of ACP as a two-step process. An initial transitory phase, ACP1, consisted of aggregates of small spherical particles lacking in long-range order. This phase quickly reordered to an amorphous phase, named ACP2, with a higher degree of coordination. Although amorphous, electron micrographs indicated that ACP2 had a more crystalline structure than ACP1 (Eanes and Meyer, 1977; Christoffersen et al., 1989). In neutral solution, hydrolysis of ACP2 to OCP or poorly crystalline apatite was observed. In more acidic solutions, a precipitate of DCPD was formed, but without involving ACP as an initial phase. Based on the analysis of the measured precipitate induction times and the structure of the developing solid phase, OCP was also proposed by Feenstra and de Bruyn (1979) as an intermediate in the conversion of ACP to apatitic calcium phosphate. Since OCP or apatite crystals were generally found in association with the ACP spherules, it is possible that ACP acts as a template for the growth of these crystal phases. Their formation, however, appears to take place by consuming ions largely supplied from the surrounding solution, rather than from direct hydrolysis of the solid amorphous material. Christoffersen et al. (1989) suggested that the more soluble ACP dissolved during the formation of the more crystalline phase. A study of ACP transformation under more alkaline conditions (pH = 10) was made by Harries et al. (1978a) using extended X-ray absorption fine structure (EXAFS). This method is concerned with the variation in absorption coefficient of an element on the high-energy side of the Xray absorption edge. This is the result of the interference between a primary photoelectron wave emitted by an atom on absorption of an Xray photon and secondary waves backscattered from neighboring atoms. This interference is dependent on the precise geometry of the atomic environment around the emitting atom, thus providing information on the coordination distances of atoms from the excited atoms. EXAFS has the advantage of defining radial distribution functions with a specific atom type defined as the origin, in terms of coordination numbers, atom times of this


types, shell radii, and Debye-Waller factors, provided that spectra of related model compounds of known crystal structure are available. Transformation experiments of ACP at pH 10 showed that the formation of poorly crystalline HAP proceeds without change in the local calcium environment, but with the development of longer range order. For poorly crystalline HAP, it was necessary to invoke a regular structure out to 0.57 nm from the calcium atom, while spectra from ACP could be explained by order extending out to only about 0.31 nm. This corresponds with the first three shells, which have similar radii in ACP to those of poorly and fully crystalline HAP. The lack of order beyond these three shells in ACP may be ascribed to its structure, which is characteristic of an amorphous solid, exhibiting paracrystalline disorder of the second kind (Harries et al., 1987b). The EXAFS results provide an alternative model to the suggestion, based on an analysis of X-ray (XRD) autocorrelation func-

tions, that ACP consists of stereochemical clusters of TCP (Eanes et al., 1965; Meyer and Eanes, 1978a, 1978b; Harries et al., 1987a, 1987b). In the latter study, it was shown that the first two peaks in the reduced radial distribution function of ACP are similar in size and position to the corresponding peaks for HAP arising from spacings up to about 0.3 nm. This corresponds to the three shells that give rise to the features of the EXAFS spectrum, which shows no evidence for order beyond this distance from a calcium atom origin. The striking agreement between the results of these studies indicates that transformation of ACP to poorly crystalline apatite may proceed without changes in the local environment of the calcium ions and involves simply an increase in the long-range order in the structure. Studies where ACPs appear to have a longer range order may possibly have involved a phase similar to that characterized as ACP2 by Christoffersen et al. (1989). The participation of ACP as a precursor phase during apatite formation in vivo has been suggested by several authors (Harper and Posner, 1966; Termine and Posner, 1967; Eanes et al., 1967; Posner et al., 1977). In a comparison of XRD data for synthetic HAP and biological apatite, it was found that the peak intensity of the latter was, in general, lower, suggesting that an-

other phase lacking X-ray fine structure was present. The conclusion that this phase must be ACP was questioned by Glimcher et al. (1981), who compared XRD spectra of stoichiometric HAP with those of nonstoichiometric and carbonatecontaining samples. The lower peak intensities of the latter was suggested as explaining, in some part, the missing fractions, which were attributed to ACP-like inclusions. Moreover, the small size of the apatite crystals, the presence of adsorbed impurities, and intergrown surface layers were also invoked as reasons to explain the differences between the XRD peak intensities of biological and synthetic apatites (Glimcher et al., 1981). In spite of these explanations of the spectroscopic data without invoking ACP, the similarities in short-range order around the calcium ions indicates that the possibility of ACP, as a precursor of in vivo formed apatite, cannot be ruled out. The rapid transformation of this phase to more crystalline calcium phosphate phases makes it unlikely that large amounts of ACP would be maintained, even in the presence of inhibitors found in serum. However, it is important to bear in mind that the rate of this transformation of ACP is normally very rapid in vitro, so that this phase may be transformed before it can be detected experimentally. Consequently, the lack of observed ACP in tissue cannot be used as evidence ruling out ACP as an in vivo precursor. The similarities in the calcium ion environment, as revealed by EXAFS, indicate that local aggregates of ions from hydrolyzing ACP can easily be incorporated into a growing apatite lattice. In contrast to these results at pH = 10, under physiological conditions the picture is quite different. The transformation of ACP to HAP in aqueous medium has been described as an autocatalytic conversion process (Eanes and Posner, 1965; Boskey and Posner, 1974; Eanes et al., 1973), and the results of a number of studies aimed at investigating the importance of solution environment (Blumenthal, et al., 1977, 1981) were usually interpreted as a single event, without involving discrete intermediate phases. However, Tung and Brown (1983) used a titration method to study the conversion of ACP at high slurry concentrations, calculating the thermodynamic driving forces for each calcium phosphate phase to determine if the solution phase was in 67


0 .2

c O0

E0E

1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2

c

mt 0 E

0.05 0.04 0.03 0.02 0.01

Acid Consumed Base Consumed 0

I

I

10

20

I

I

30

50 60 70 Time (min)

40

80

90

100

110

120

FIGURE 2. Changes in calcium and phosphate concentrations during the transformation period (upper) and the amount of acid and base required to maintain pH at 7.4 (lower) during hydrolysis of ACP. (From Tung, M. S. and Brown, W. E., Calcif. Tissue Int., 35, 783, 1983. With permission.)

quasi-equilibrium with precipitated solids. A typical conversion kinetics experiment, shown in Figure 2, clearly indicates two processes: the first, consuming acid, and the second, consuming base, in order to maintain a constant pH of 7.4 (25°C).

In the first process, the calcium concentration increased by about 10%, reaching a maximum when the consumption of acid also reached its maximum. This implied that calcium and hydroxide ions were released simultaneously from the ACP, in accordance with the conversion of ACP to an OCP-like intermediate by reaction (8), Ca9(PO4) ACP

+

7H20

-*

CagH2(PO4)6 5H20

+

Ca2+ + 20H-

(8)

OCP

Moreover, in this first step, the phosphate concentration decreased, ruling out the possibility that the increase in calcium and hydroxide ions was due to the simple dissolution of ACP. In the second process, requiring a consumption of base, the calcium concentration decreased and the phosphate concentration began to increase after about 100 min, indicating that the OCP-like

intermediate,

well as ACP, converted to an apatite. Following 24 h of reaction, the final product was nonstoichiometric, with Ca/P = 1.56, and showed an apatitic X-ray diffraction as

spectrum.

Clear evidence for the formation of OCP as intermediate in the precipitation of HAP was obtained from studies of the transition stage in the spontaneous precipitation of calcium phosphate made under pH-stat conditions by titration with base (Miller et al., 1977). Typical plots of AGocp in Figure 3 show the existence of a secondary inflection during the amorphous to crystalline transformation. Moreover, the ionic activity product of the solution phase in contact with the first-formed crystalline material was invariant and close to the solubility value for OCP, indicating the formation of this phase. The hydrolysis of the OCP to more apatitic phases may occur either by the dissolution of OCP, followed by precipitation of HAP, or by a direct solid-state transformation (Tomazic et al., 1989; LeGeros et al., 1988b). Young and Brown (1982) described the formation of HAP and OCP by hydrolysis of an OCP unit cell to a two-unit cell

an

68


(. o

C

+0.1

+0.2

-

+0.3

-

.0.4 +0.5+0.6

0

60

120

240

180

300

360

min

FIGURE 3. Thermodynamic stability of solutions at pH 7.20, 7.60, 7.80, 8.00, 8.40, and 8.60 compared with solutions in equilibrium with

pure OCP. Values of Gocp are calculated from the measured Tc, and To4, together with the volume of base required to maintain the pH values. (From Meyer, J. L. and Eanes, E. D. Calcif. Tissue Res., 25, 59, 19, 78. With permission.) thick

layer of HAP. In such a model, the HAP crystal would be covered by a hydrated layer consisting of an OCP unit cell (Young and Brown, 1982). This model is supported by the observation of a central defect in many apatite crystals, which has been ascribed to the presence of OCP (Nelson and McLean, 1984; Aoba and Moreno, 1989; LeGeros et al., 1989b; Nelson and Barry, 1989). Upon hydrolysis, OCP forms an apatite-

like phase with a Ca/P ratio lower than that of stoichiometric HAP. Due to the similarities in crystal structure, epitaxial intergrowth can also occur (Brown, 1962; Taves, 1963), and this suggestion is supported by the results of high-reso-

lution electron microscopy (Nelson and McLean, 1984). Either the transformation of OCP or simultaneous growth of both phases would yield an eventual apatitic phase having calcium ion deficiencies. The transformation is probably never complete, leaving unhydrolyzed regions of OCP associated with the precipitated apatite. Moreover, other ions, adsorbed at surfaces having a relatively high specific surface area, may also change the experimental Ca/P ratio. Thus, if a half-unit cell of OCP covered a stoichiometric HAP crystal, the Ca/P ratio would be about 1.64 (Young and Brown, 1982). Owing to the small size and large specific surface area of HAP crys69


tallites, adsorbed ions may constitute a signifi-

cant fraction of the total ionic content of the crystal. Neuman and Neuman (1958) estimated the hydration layer at HAP crystallite surfaces to be about 40 to 80% of the total weight of the crystal. From a more detailed analysis of the experimental data, Brown et al. (1987) suggested that a typical biological apatite is formed by the initial precipitation of OCP, which hydrolyzes to an intermediate phase termed octacalcium phosphate hydrolyzate (OCPH). This phase was thought to be made up of layers of both OCP and HAP-like structures and included impurity ions and ion vacancies. In support of this model, transmission electron microscopy (TEM) studies of hydrolyzed OCP crystals indicate the intergrowth of both HAP and OCP crystallites, together with a mixture of both phases (Nelson and McLean, 1984), which could very well be described in terms of an OCPH phase. Crystal defects and impurity ions would, furthermore, be maintained in the apatitic structure, since the hydrolysis to the thermodynamically most stable phase is irreversible. The final product will, therefore, be a poorly crystalline, calcium-deficient apatite phase. Comparisons of lattice imaging of synthetic OCP and apatite with a central defect showed marked similarity in the images. These central defects are also more sensitve to acid dissolution than the bulk of the crystals, although in synthetic apatites the defects were found only when crystals were prepared at neutral pH in the presence of carbonate ion (Nelson et al., 1986b). This suggests the presence of substituted carbonate ions in the lattice. Daculsi et al. (1989, 1990) compared the dissolution behavior of biological and synthetic apatites containing the central defects in the core. In all cases, dissolution occurred both at the crystal core and at the surface, with the biological apatite showing preferential core dissolution. Although this may be due to the presence of OCP in the crystal core, it could also be attributed to impurities such as carbonate in the biological apatite, with preferential dissolution of carbonated apatite. Although many workers favor the hypothesis that OCP is involved as an in vivo precursor during apatite formation, there appears to be no real evidence for the presence of this phase in non-

pathological mineralization. At one stage, it was suggested that DCPD was formed in the early stages of development of embryonic bones in

(Landis and Glimcher, 1978), but this has finding since been disputed. There is little doubt, however, that OCP occurs as an initial phase in pathological deposits, such as calculus and kidney stones. At lower pH, DCPD may also be involved as a precursor phase, either in place of or together with OCP. Recent studies of DCPD hydrolysis, to be described later, suggest a dissolution and reprecipitation process, rather than a solid-state transformation to OCP. This indicates that an initially formed DCPD serves as a calcium and phosphate ion reservoir for subsequent crystal growth of the apatitic phase (Landis and Glimcher, 1978). Moreover, certain calcium positions in DCPD and HAP are closely aligned, opening the possibility of epitaxial intergrowth and transformation of DCPD to HAP, which has also been demonstrated using constant composition kinetics methods (Koutsoukos and Nancollas, 1981). The results of a typical experiment demonstrated the nucleation, and growth of HAP at DCPD surfaces is shown in Figures 4 and 5. These solutions were supersaturated with respect to HAP and saturated with respect to DCPD. Landis and Glimcher (1978) found that the Ca/P ratio of precipitated calcium-deficient apatite gradually increases with the age of the precipitate, probably due to in situ transformation. In general, the observed sequence of calcium phosphate precipitation has been detected by the relative solubility of the different solid phases at constant pH and temperature. Thus van Kenemade and de Bruyn (1987) showed that the Ostwald rule of stage (Ostwald 1897) was obeyed, regardless of solution conditions under which the relaxation experiments were made. Homogenous formation of HAP at low supersaturation was never observed but was always preceded by the growth of precursors. At pH 6.7, OCP was observed to form at intermediate supersaturation, largely due to the exclusion of other phases. The growth curves and relaxation times obtained under these conditions were analyzed in terms of classic nucleation and growth theories. The kinetics of formation of OCP was best described by a flash-like nucleation step in combination with surface nucleation and growth based on a chicken


Epitaxial growth of HAP on DCPD seed crystals. (From Koutsoukos, Cryst. Growth, 53, 10, 1981. With permission.) FIGURE 4.

mononuclear growth model. At intermediate supersaturations, the surfaces of the grown crystalline particles could be regarded as being relatively smooth, with growth proceeding by a layer mechanism, requiring the formation of a surface nucleus. In attempting to explain the preferential precipitation of one crystalline phase when compared with another, it can be argued that a lower interfacial tension or edge free energy is to be expected for a more soluble phase (Nielsen and Sohnel, 1971; Sohnel, 1982). This could result in a lower free energy for two-dimensional nucleation, in spite of the lower supersaturation. It is quite clear that the Ostwald rule of stages, which states that the least stable, most soluble phase forms preferentially in a sequential precipitation, is likely to be obeyed for the calcium phosphates. However, in these cases, the sequence is also influenced by another important parameter, pH. It would appear, therefore, that where the driving force for HAP is relatively

P. G. and Nancollas, G. H., J.

high, at pH 10, the participation of more acidic phases, such as brushite and OCP, can be ruled out. It is interesting to note that a recent constant composition study of defect apatite crystal growth also pointed to stoichiometries approaching the HAP value at high pH (Heughebaert et al., 1990). -

VI. CRYSTAL LATTICE SUBSTITUTIONS The participation of acidic phases, such as DCPD and OCP, during the precipitation of apatites is often inferred by estimations based upon molar calcium/phosphate ratios of the solid phases. However, these ratios reflect the substitution of foreign ions in the crystal lattice, an event that has been shown to be quite common for calcium phosphates. Young and Brown (1982) suggested that the high degree of substitution in apatites may be due to the diffusion of ions in or out of OH--channels within the structure, facilitating the exchange of OH- ions for water or 71


C

2.0 1.zo0 O

00

10.0

20.0

300 TIm

500

400

600

700

800

(hours )

FIGURE 5. Constant-composition growth of HAP on DCPD seed crystals shown as plots of titrant volume against time for three different concentration conditions: Exp189: Tc. = 2.89 x 10-2 mol I-1; Tp = 4.95 x 10-3 mol I-', pH = 5.0;Exp180: Tc = 1.36 x 10-2 mol I-2; Tp4 = 2.03 x 10-3 mol I-1, pH = 5.6;Exp195: Tc, = 8.70 x 10-2 mol I-1; Tpo4 = 1.21 x 10-3 mol I-', pH = 6.0. The induction period reflects nucleation of HAP on the DCPD seed crystal surfaces.

for other ions. Taves and Reedy (1969) suggested that these channels at the crystal surfaces forms parallel grooves containing adsorbed water, phosphate, OH-, and foreign ions. Ion vacancies in Ca2+ or OH- positions compensate for excess or depleted charges due to substitution. Typical ion substitutions in biological apatites are F- for Cl- or OH-, carbonate for phosphate or OH-, and Sr2 , Mg2 , or Na for Ca2 (Elliott, 1964; LeGeros et al., 1968, 1989a; Dykes and Elliott, 1971; LeGeros, 1974, 1984; Driessen et al., 1987a, 1987b; Bigi et al., 1986, 1988). The ordered array of OH- ions in the OH--channels needed to form the crystal structure increases the possibility of ion vacancies or incorporation of the smaller F- ion. Thus the apparent low molar Ca/P ratio in the apatite crystal can be accounted for by the presence of foreign ions, or even the

adsorbed water molecules in OH--channels with excess phosphate ions in these hydrated monolayers. Moreover, vacancies of calcium and phosphate ions neighboring the OH- channels facilitate the exchange of other ions, such as OH for F-, with less lattice strain (Kay et al., 1964). Such substitution is facilitated by the hydrogenbonding properties of F-, leading to increased crystallinity. In addition, substitution of two OHions for one oxide ion introduces a vacancy in the OH--channels, allowing for charge compensation involving carbonate or hydrogen phosphate. Incorporation of HPO2- may occur by protonation of phosphate sites, while carbonate may substitute both for OH- and phosphate (Elliott, 1964; LeGeros et al., 1968, 1981; Bigi et al., 1986; Brown et al., 1987). LeGeros and coworkers showed that carbonate substitutes mainly

72


for

phosphate (Zapanta-LeGeros, 1965; LeGeros, 1967; LeGeros et al., 1968, 1971). Charge

balance

may be achieved by sodium ion incorporation, and the substitution of carbonate usually results in poorly crystalline structures. In cases where carbonate occupies the phospate sites, the smaller size induces strain in the lattice, leading to a unit cell contraction. An EXAFS spectroscopic study (Harries et al., 1987b) reveals marked structural changes within the HAP unit cell accompanying the substitution of the phosphate ion for carbonate. The structural geometry beyond the nearest neighbor oxygen coordinations to calcium is changed in a manner

consistent with an increase in structural disorder. However, the nearest neighbor coordination to calcium is not detectably influenced by the presence of carbonate. To maintain such coordination in the HAP lattice, it was concluded that the carbonate ion must be placed so that the oxygens coincide with those sites previously occupied by the phosphate oxygens. Consequently, the vacant oxygen sites are probably directed away from the calcium ions, inducing lattice strain and accounting for the observed unit cell contraction as seen by X-ray diffraction. Infrared analysis (LeGeros et al., 1989b) of in vitro hydrolyzed OCP shows the presence of HPO2-, which may indicate missing or substituted calcium ions in the structure and may account for the observed increase in a axis, when compared with stoichiometric, HAP. In contrast to the much more sparingly soluble apatites, both OCP and DCPD dissolve more readily, contain water in their crystal lattices, and do not incorporate impurity ions as frequently (Salimi et al., 1985). Their influence on the rate of formation of apatite from DCPD and OCP depends; in part, on the possibility of incorporation of the foreign ions into the precipitating apatitic phase. In many situations, both in vivo and in vitro, ions such as carbonate will catalyze the formation of carbonated apatite so rapidly that it is usually impossible to detect intermediate phases, such as DCPD and OCP, if they exist, during the reaction. Thus the overall "hydrolysis" of the acidic phases is apparently catalyzed by the presence of such ions. In contrast, some ions, present in the solution phase, are effective inhibitors of selective calcium phosphate phases, leading to the kinetic stabilization of initial tran-

sient phases. Thus the frequently observed amorphous phase, ACP, during in vivo mineralization may be a consequence of the presence of Mg2+ ions, which kinetically stabilize acidic calcium phosphate phases during in vitro precipitation reactions (Boskey and Posner, 1974; LeGeros et al., 1975). It is interesting to note that Vogel (private communication) was able to detect only an apatitic phase in the early fetal bones of magnesium-deficient animals. Normally, the presence of this ion would stabilize ACP (Holt et al.,

1988, 1989).

VII. KINETICS OF CRYSTALLIZATION OF CALCIUM PHOSPHATES It will be apparent from the foregoing that although the likelihood of precipitation of a particular phosphate phase is ultimately determined by the thermodynamic driving force for its formation, kinetic factors may be considerably more important in controlling the nature of the solids formed. There is, therefore, considerable interest in characterizing the kinetics of the growth of individual calcium phosphate phases, as well as their transformation to the thermodynamically most stable phase. Such studies have usually been made by introducing seed crystallites into metastable solutions and following the changes in lattice ion concentrations as a function of time. However, as suggested by Figure 1, different calcium phosphate phases may participate as the concentrations relax to solubility conditions with respect to HAP. One improvement of these freedrift studies was the incorporation of a glass electrode and pH-stat so that alkali could be added in order to maintain the hydrogen ion activity. However, changes in the degree of supersaturation during these experiments still resulted in the transient formation and dissolution of phases during the reactions. Moreover, the relatively small changes in monitored calcium and phosphate concentrations during the reactions made it difficult to use these data in order to calculate the stoichiometry of the solid phases undergoing growth and/or dissolution. It is generally not possible to make quantitative physical chemical measurements on the precipitated solids, since the amount of newly formed phase is very small. The problems associated with these conven73


tional free-drift studies were overcome with the development of the constant composition (CC) and more recently dual constant composition (DCC) methods, in which the chemical potentials of the solution species are maintained constant during the reactions (Tomson and Nancollas, 1978). Following the addition of well-characterized seed material to metastable solutions of calcium phosphate at the desired pH and supersaturation, the concentrations of lattice ions are maintained at their initial values by the simultaneous addition of titrant solutions containing calcium, phosphate, and hydroxide ions from coupled burets. These additions are controlled by suitable potentiometric probes, such as glass and calcium electrodes. In this way, significant amounts of new phase can be formed, even at very low supersaturation, enabling their characterization by physical chemical methods (Koutsoukos et al., 1980). Provided that the effective titrant concentrations are such that their stoichiometry is the same as that of the precipitating phase, the activities of all ionic species in the supersaturated solutions remain constant during the reactions. By using very dilute titrants, it is possible to obtain reliable rate data from the plots of titrant addition as a function of time, even for very slow reactions with a precision normally unattainable in conventional free-drift

experiments. An advantage of the CC method is the ability to select particular points on the solubility isotherms in Figure 1 so that the formation of DCPD and OCP phases during a sequential growth reaction can be readily demonstrated. Thus in the region between DCPD and OCP solubility curves, it was possible to grow, exclusively, OCP in the pH range 6.0 to 7.0 upon adding OCP seed crystals (Heughebaert and Nancollas, 1984). As is true for numerous crystallization reactions involving sparingly soluble salts, the rate of growth, R, can be interpreted in terms of a simple rate equation (9): R = k a"

(9)

in which k is the rate constant and n is the effective order of the reaction. Analysis of the experimental data, together with those for the formation of OCP during the transformation of

DCPD (Heughebaert and Nancollas, 1984; Madsen, 1987), gave a value n = 2, suggesting a spiral growth mechanism at low supersaturation with respect to OCP changing to a polynucleation mechanism, with n 4 at higher driving forces. In these reactions, the formation of highly crystalline OCP platelets (Figure 6) during the reaction led to marked decreases in specific surface area, consistent with the observed morphology changes (Newesley, 1970). Calculations based on the Hartman-Perdok theory (Hartman and Perdok, 1955a, 1955b; Hartman, 1973) also predicted the formation of the observed blade-like morphology of OCP. However, it should be cautioned that phase-formation conclusions based purely on morphological arguments may be entirely unjustified. Thus Driessen et al. (1987a) suggested that the board-like form of the mineral particles in mature bone and dentin indicate that they may have originated from an OCP phase, while the flattened needle-like crystallites of mature enamel indicated DCPD as precursor phase. Such arguments must be made with caution, since they cannot be supported by the results of in vivo mineral characterization. In vitro seeded crystal growth studies clearly demonstrate the importance of pH and a in controlling the phase that forms. Thus in CC studies using OCP seed crystals at pH = 5 in solutions supersaturated with respect to all phases (Figure 1), DCPD was formed on the seed crystallites following an initial induction period, reflecting the nucleation of the new phase (DeRooij et al., 1984). As the supersaturation increased, the induction period for DCPD formation was reduced. A DCC study by Ebrahimpour et al., (1991) at pH 7 has shown that the phase transformation of DCPD to OCP occurs by dissolution of DCPD, followed by precipitation of OCP, in agreement with suggestions by LeGeros et al. (1983) and Young and Brown (1982). At higher pH (7.4 to 8.0), both the hydrolysis of OCP seed crystals and the formation of a more basic phase occurred. The use of titrant solutions designed for the control of a specific phase, such as DCPD growth, resulted in a nonconstancy of composition, since the titrant stoichiometry did not match that of the precipitating phase. However, the measured changes in calcium and phosphate concentration during the experiments could then be used to =

74


FIGURE 6. Plate-like

crystals of OCP from scanning electron microscopy.

calculate the titrant stoichiometry needed to control the concentration and thereby determine the stoichiometry of the precipitating defect apatite. This method was used to characterize the kinetics of crystallization of a series of defect apatites.

Cao_u Hu (PO4)6(OH)2,u

(10)

in which O u < 2 (Heughebaert et al., 1990; Zawacki et al., 1990). The calcium/phosphate molar ratios of phases grown at HAP surfaces over a pH range from 6.0 to 9.0 varied from 1.49 to 1.65. Moreover, sufficient new phase was formed to enable X-ray diffraction confirmation of the solid stoichiometry. At the physiological pH of 7.40, the constancy of calcium and phosphate concentrations during the growth experiments indicated a solid phase stoichiometry, consistent with u = 0.57. In all these studies, it is very important to distinguish between the formation of truly nonstoichiometric apatites and the initial ion-exchange reactions that occur upon the

introduction of seed material into solutions, which, in composition, may be quite different from those in which seed crystals had been aged.

VIII. CRYSTALLIZATION OF CALCIUM PHOSPHATES IN THE PRESENCE OF ADDITIVES

-

It is clear from the foregoing that any discussion of the importance of DCPD and/or OCP during the precipitation of calcium phosphates must take into account the probable influence of extraneous ions or molecules in the solution phase. These range from proteins, such as human serum albumin and statherin, to smaller ions, such as citrate and pyrophosphate, and even to simple ions, such as Mg2+, Zn2+, C02- and F-. In general, at physiological pH, anionic proteins show a considerably greater influence on HAP crystallization than positive or neutral species, suggesting that they bind to calcium ions at the 75


surface. Moreover, their inhibiting influence during extended CC growth periods suggests that they are able to bind to active growth sites that may be formed during crystal growth, as well as those intrinsic to the seed crystallites. These inhibiting properties can be measured either by preadsorbing the molecules to HAP seed crystal surfaces before adding to calcium phosphate metastable solutions or by making crystal growth studies in solutions containing the additives. The general agreement between the results of these studies suggests that many adsorbed molecules are resistant to desorption from the seed crystals. Adsorption, in such cases, probably involves multiple binding sites at the surface. The observation of deposits of ACP in liver mitochondria indicates the presence of natural inhibitors able to stabilize this phase. Blumenthal et al. (1975) suggested that ATP and Mg2+ synergistically delay transformation to HAP by acting as growth poisons and allowing the cells to store ACP as a reservoir of calcium phosphate. Posner et al. (1977) also suggested that the binding of ATP and Mg2+ to HAP nuclei prevented their growth. Regulation of the levels of ATP and Mg2+ might therefore allow the cells to store calcium phosphate as ACP until required for crystallization. Williams and Sallis (1979, 1982) showed that ATP delays the onset of transformation of ACP during in vitro conditions. Both studies, therefore, suggest that the apparent stabilization of ACP may be due to binding of the inhibitors to newly formed nuclei, preventing these from reaching a critical growth size, but leading to redissolution. In addition to macromolecular additives, both Mg2+ and Zn2+ inhibit HAP crystal growth. Clearly the uptake of these small ions is mainly electrostatic, and their adsorption affinities are relatively small. Consequently higher concentrations are required in order to substantially reduce the rate of crystallization. It is interesting to note that the adsorption of Mg2+ continues at HAP seed surfaces, reaching a plateau with little further uptake. In contrast, Zn2+ binding increases monotonically with increased solution concentration beyond the point of monolayer coverage (Chin and Nancollas, unpublished results). This is probably due to the precipitation of zinc phosphate at the apatite surface as the concentration

of zinc ions is increased, and this is an important factor to take into account when attempting to interpret the results of simple ion uptake during

adsorption experiments.

The influence of fluoride ion is especially interesting in view of its importance in controlling caries. The accommodation of F- at OHsites, forming a less soluble fluorohydroxyapatite phase, enhances the crystal growth reaction, reducing the rate of dissolution (Grynpas and Cheng, 1988; Grynpas, 1990). The influence of fluoride ion continues to be a subject of considerable interest, and although statements are made to the effect that fluorapatite catalyzes the formation of HAP, it should be remembered that a new phase is forming, namely, partially or totally substituted fluorhydroxyapatite phases. The degree of substitution may be important, since studies of the solubility of apatites with various degrees of fluoride content have shown that partially fluoridated apatite may have a lower solubility than either HAP or fluorapatite itself (FAP) (Moreno et al., 1976). In contrast to fluoride, carbonate, which is also incorporated readily into the apatite lattice, inhibits HAP crystal growth (Featherstone and Nelson, 1980; Featherstone et al., 1983), probably due to adsorption at surface sites and also the less favorable coordination in phosphate or hydroxide sites in the lattice. Studies of crystallization and dissolution kinetics of DCPD and OCP in the presence of impurities show, in general, a similar trend to that of HAP, although these hydrated phases are less sensitive. The observed differences in the degree of inhibition are important in helping to understand the nature of the calcium phosphate phases observed in vivo, especially for pathological mineralization processes. Thus in a seeded CC crystal growth study (Salimi et al., 1985), it was shown that the magnesium ion markedly inhibits HAP crystal growth in a solution supersaturated

only

with respect to this phase, has a modest influence on the kinetics of OCP growth, and had practically no effect on DCPD crystallization (Figure 7). A possible reason for this wide range of inhibition properties may be attributable to the presence of lattice water in OCP and DCPD, decreasing the adsorption of foreign ions. Moreover, it was shown that these more acidic phases grow at considerably increased rates when com-


R

Ro

DCPD

1 A

0 0

OCP HAP 2

4

6

8

10

[Mg],mM

FIGURE 7. Influence of Mg2+ on the crystal growth of DCPD, OCP, and HAP. (From Salimi, M. H. et al., Langmuir, 1, 119, 1985. With permission.)

pared with HAP. The mineralization of the thermodynamically most stable HAP is, therefore, much more strongly inhibited by the presence of Mg2+. Magnesium is able to mediate in the precipitation process by selectively stabilizing more acidic precursor phases. Similar conclusions have been reached by other authors (Eanes and Rattner, 1981; Cheng, 1989). Extensive studies by LeGeros and co-workers (LeGeros et al., 1984, 1986a, 1989b) have been concerned with the characterization of OCP transformation using scanning and transmission microscopy, infrared spectroscopy, and X-ray diffraction. It was shown that the OCP hydrolysis reaction is influenced by the types of ions in solution, being retarded by Mg2+, citrate, and pyrophosphate, and accelerated by CO-, HPO2-, and Ca2+. The observed incorporation of ions, such as carbonate and hydrogen phosphate, again pointed to a process of OCP dissolution and subsequent precipitation of calciumdeficient apatites during the transformation reactions.

IX. CONCLUSIONS The participation of DCPD and/or OCP during the formation of apatites is closely dependent upon the supersaturation, the pH of the solutions, and, most importantly, the presence of foreign ions and molecules. In vivo, during normal mineralization, there appears to be little evidence for the formation of these discrete acidic phases. The first identifiable phase appears to be ACP, which transforms to a more crystalline apatitic phase by the development of longer range order. Such transformation probably occurs discontinuously with the formation of local, more ordered structures, which may be referred to as different ACP (ACP2?) or OCP-like phases. However, it is important to recognize that since kinetic factors are as important as thermodynamic driving forces in determining the formation of these phases, it is quite possible that they may form and disappear at rates too rapid to follow. Thus, although the earlier identification of brushite in early-forming chicken embryonic bone was questioned, more 77


recent studies by these workers (Lee et al., 1986) using high-voltage electron diffraction methods have again revealed unusual calcium phosphate phases. Although the mineral was largely poorly crystalline apatite, some evidence for locally formed DCPD, as well as some other nonapatitic phases, was presented. The participation of amorphous calcium phosphate phases during in vivo mineralization cannot be disputed. Transformation probably takes place not by a continuous autocatalytic process, as originally supposed (Eanes and Posner, 1965, 1970), but rather by discontinuous steps involving the development of longer range order, as revealed by EXAFS studies. However, it is important not to make generalizations, since traces of ions and molecules may markedly influence the rates of formation of specific calcium phos-

phate phases. Under in vitro conditions and during pathological mineralization induced by trauma, the picture appears to be quite different. Here, there is considerable evidence for the formation of brushite and OCP as precursor phases to the formation of apatite. Moreover, traces of ions present in the solution have been shown to stabilize more acidic phases (e.g., magnesium, citrate, pyrophosphate), whereas in the presence of ions,

such as fluoride and carbonate, a more direct crystallization of phases such as FAP and carbonated apatite occurs. This was well demonstrated in the work of Vogel (1986), who showed that dioleoylphosphatidate liposomes catalyzed the formation of apatite in the presence of carbonate and of OCP when carbonate was absent. Clearly the thermodynamic driving force for the formation of carbonate apatite was sufficiently high so as to mask the formation of OCP, even if it occurred. It is now generally believed that the formation and transformation of precursor phases, such as DCPD and OCP, to apatite occurs by dissolution and reprecipitation processes. If only these two phases were involved, the precipitated apatite would be expected to be relatively stoichiometric. However, even in vitro, OCP-like and HAP-like phases having intermediate stoichiometry are often found. Whether these are formed by intermixtures of pure phases, such as OCP or HAP, or by the gradual development of long-

range order in the solid phases is difficult to determine. In vivo, the apatitic phases are usually nonstoichiometric, and it is unlikely that these form by the same pathway as that induced in vitro

in the laboratory. The presence of a large number of crystallization inhibitors in serum and saliva will greatly influence the overall process, and it is generally believed that less hydrated phases are more inhibited by the presence of these components. The rate of HAP formation would, therefore, be reduced to a greater extent than that of DCPD or OCP. However, although HAP crystallization may be the most inhibited phase in vivo, thermodynamically this phase may be favored in reactions extending over long time periods. Once formed, HAP remains the most stable phase, whereas the precursor phase may dissolve and hydrolyze, depending upon the local concentration conditions. In mineral turnover, such as that occurring in bone, the most stable phase is favored, even if this is inhibited kinetically to a greater extent. In such in vivo systems, the driving force for apatite formation may be so great that it is not possible to detect discrete precursor phases, but rather a gradual increase in long-range order during the transformation of the initially formed ACP. Thus, in the presence of carbonate ions, carbonated apatite forms preferentially, whereas the presence of effective inhibitors, such as magnesium, stabilizes the most amorphous precursors. Important factors in vivo appear to be not only the concentrations of ionic species, but also the presence of proteins and other macromolecular components. The sensitivity to parameters such as pH is striking, since at slightly higher acidity, both in vitro and in mineralization resulting from trauma, considerable evidence has been presented for the participation of DCPD and OCP, which are epitaxially quite closely related to HAP.

ACKNOWLEDGMENTS We thank the National Institute of Dental Research for grants DE07585 and DE03223, under which most of these studies were made. We also thank Dr. J. Zhang and E. Paschalis for their critical comments.


REFERENCES Aoba, T. and E. C. Moreno: Mechanism of Amelogenetic

Mineralization in Minipig Secretory Enamel. In Tooth Enamel V. Proc. 5th Int. Symp. composition, properties and fundamental structure of tooth enamel and related tissues. pp. 163-167. (R. W. Fearnhead, Ed.) Bigi, A., E. Foresti, A. Ripamonti, and N. Roveri: Fluoride and Carbonate Incorporation into Hydroxyapatite Under Conditions of Cyclic pH Variation. J. Inorg. Biochem. 27:31-39 (1986). Bigi, A., M. Gazzano, A. Ripamonti and N. Roveri: Effect of Foreign Ions on the Conversion of Brushite and Octacalcium Phosphate into Hydroxyapatite. J. Inorg. Biochem. 32:251-257 (1988). Blumenthal, N. C., F. Betts and A. S. Posner: Nucleotide Stabilization of Amorphous Calcium Phosphate. Mater. Res. Bull. 10:1055-1062 (1975). Blumenthal, N. C., F. Betts and A. S. Posner: Stabilization of Amorphous Calcium Phosphate by Mg and ATP. Calcif. Tissue Res. 23:245-250 (1977). Blumenthal, N. C., F. Betts and A. S. Posner: Formation and Structure of Ca-Deficient Hydroxyapatite. Calcif. Tissue Int. 33:111-117 (1981). Boskey, A. L. and A. S. Posner: Magnesium Stabilization of Amorphous Calcium Phosphate: A Kinetic Study. Mater. Res. Bull. 9:907-916 (1974). Boskey, A. L. and A. S. Posner: Formation of Hydroxyapatite at Low Supersaturation. J. Phys. Chem. 80:4045 (1976). Brown, W. E.: Octacalcium Phosphate and Hydroxyapatite. Nature. 196:1048-1049 (1962). Brown, W. E., J. P. Smith, J. R. Lehr and A. W. Frazier: Crystallographic and Chemical Relations between Octacalcium Phosphate and Hydroxyapatite. Nature. 196:1049-1055 (1962). Brown, W. E., N. Eidelman and B. B. Tomzaic: Octacalcium Phosphate as a Precursor in Biomineral Formation. Adv. Dent. Res. 1:306-313 (1987). Cheng, P. T., J. Grabher and R. Z. LeGeros: Effects of Magnesium on Calcium Phosphate Formation. Magnesium. 7:123-132 (1988). Cheng, P. T.: Magnesium Inhibits Octacalcium Phosphate and Apatite but Promotes Whitlockite and Brushite Formation. In: Urolithiasis. pp. 225-226. (V. R. Walker, R. A. L. Sutton, E. C. B. Cameron, C. Y. C. Pak and W. G. Robertson, Eds.) Plenum Press, New York, (1989). Christoffersen, J., M. R. Christoffersen, W. Kibalczyc and F. A. Andersen: A Contribution to the Understanding of the Formation of Calcium Phosphates. J. Cryst. Growth. 94:767-777 (1989). Daculsi, G., R. Z. LeGeros and D. Mitre: Crystal Dissolution of Biological and Ceramic Apatite. Calcif. Tissue Int. 45:95-103 (1989). Dalcusi, G., R. Z. LeGeros, M. Heughebaert and I. Barbieux: Formation of Carbonate-Apatite Crystals After Implantation of Calcium Phosphate Ceramics. Calcif. Tissue Int. 46:20-27 (1990).

Davies, C. W.: In Ion Association. Butterworths, London (1962). DeRooij, J. F., J. C. Heughebaert and G. H. Nancollas: A pH Study of Calcium Phosphate Seeded Precipitation. J. Colloid Int. Sci. 100:350-358 (1984). Dickens, G., L. W. Schroeder and W. E. Brown: Crystal-

lographic Studies of the Role of Mg as a Stabilizing Impurity in b-Ca3(PO4)2. I. The Crystal Structure of Pure -Ca3(PO4)2. J. Solid State Chem., 10:232-248 (1974). Driessens, F. C. M., R. A. Terpstra, P. Bennema, J. H. M. Woltgens and R. M. H. Verbeeck: On the Possible Relation between Morphology and Precursors of the Crystallites in Calcified Tissues. Z. Naturforsch. 42c:916920 (1987a). Driessens, F. C. M., J. M. P. M. Borggreven and R. M. H.

Verbeeck: The Dynamics of Biomineral Systems. Bull. Soc. Chim. Belg. 96:173-179 (1987b). Dykes, E. and J. C. Elliott: The Occurence of Chloride Ions in the Apatite Lattice of Holly Springs Apatite and Dental Enamel. Calcif. Tissue Res., 7:24-27 (1971). Eanes, E. D., I. H. Gillessen and A. S. Posner: Intermediate States in the Precipitation of Hydroxyapatite. Nature. 208:365-367 (1965). Eanes, E. D. and A. S. Posner: Kinetics and Mechanism of Conversation of Noncrystalline Calcium Phosphate to Crystalline Hydroxyapatite. Trans. N.Y. Acad. Sci. 28:233-241 (1965). Eanes, E. D., J. D. Termine and A. S. Posner: Amorphous Calcium Phosphate in Skeletal Tissues. Clin. Orthoped. 53:223-235 (1967). Eanes, E. D. and A. S. Posner: A Note on the Crystal Growth of Hydroxyapatite Precipitated from Aqueous Solutions. Mater. Res. Bull. 5:377-384 (1970). Eanes, E. D., J. D. Termine and M. U. Nylen: An Electron Microscopic Study of the Formation of Amorphous Calcium Phosphate and its Transformation to Crystalline Apatite. Calcif. Tissue Res. 12:143-148 (1973). Eanes, E. D. and J. L. Meyer: The Maturation of Crystalline Calcium Phosphates in Aqueous Suspensions at Physiological pH. Calcif. Tissue Res. 23:259-269 (1977). Eanes, E. D. and S. L. Rattner: The Effects of Magnesium on Apatite Formation in Seeded Supersaturated Solution at pH 7.4. J. Dent. Res. 60:1719-1723 (1983). Ebrahimpour, A., J. Zhang, and G. H. Nancollas: Dual Constant Composition Method and its Application to Studies of Phase Transformation and Crystallization of Mixed Phases. J. Cryst. Growth. in press. Elliott, E. C.: The Crystallographic Structure of Dental Enamel and Related Apatites. Ph.D. thesis, University of London, London (1964). Featherstone, J. D. B. and D. G. A. Nelson: The Effect of Fluoride, Zinc, Strontium, Magnesium and Iron on the Crystal-Structural Disorder in Synthetic Carbonated Apatite. Aust. J. Chem. 33:2363-2368 (1980). Featherstone, J. D. B., I. Mayer, F. M. C. Driessens, R. M. H. Verbeeck and H. J. M. Heijligers: Synthetic Apatites Containing Na, Mg and C02- and Their Comparison with Tooth Enamel Mineral. Calcif. Tissue Int. 35:169-171 (1983).


Feenstra, T. P. and P. L. de Bruyn: Formation of Calcium

Phosphates in Moderately Supersaturated Solutions. J. Phys. Chem. 83:475-479 (1979). Glimcher, M. J., L. C. Bonar, M. D. Grynpas, W. J. Landis

and A. H. Roufosse: Recent Studies of Bone Mineral: Is the Amorphous Calcium Phosphate Theory Valid? J. Cryst. Growth. 53:100-119 (1981). Grynpas, M. D. and P. T. Cheng: Fluoride Reduces the Rate of Dissolution ofBone. Bone Mineral. 5:1-9 (1988). Grynpas, M D.: Fluoride Effects on Bone Crystals. J. Bone Mineral Res. 5:S169-175 (1990). Harper, R. A. and A. S. Posner: Measurement of the NonCrystalline Calcium Phosphate in Bone Mineral. Proc. Soc. Exp. Biol. Med. 122:137-142 (1966). Harries, J. E., D. W. L. Hukins, C. Holt and S. S. Hasnain: Conversion of Amorphous Calcium Phosphate into Hydroxyapatite. J. Cryst. Growth. 84:563-570 (1987a). Harries, J. E., S. S. Hasnain and J. S. Shan: EXAFS Study of Structural Disorder in Carbonate-Containing Hydroxyapatites. Calcif. Tissue Int. 41:346-350 (1987b). Hartman, P. and W. G. Perdok: On the Relation between Structure and Morphology of Crystals. I and II. Acta Cryst. 8:49-52 (1955a);Acta Cryst. 8:521-524 (1955b). Hartman, P.: In Crystal Growth: An Introduction. pp. 367402 (P. Hartman, Ed.) North Holland, Amsterdam (1973). Heughebaert, J. C. and G. Montel: Conversion of Amorphous Tricalcium Phosphate into Apatitic Tricalcium Phosphate. Calcif. Tissue Int. 34:104-108 (1984). Heughebaert, J. C. and G. H. Nancollas: Kinetics of Crystallization of Octacalcium Phosphate. J. Phys. Chem.

88:2478-2481 (1984).

Heughebaert, J. C., ,S.

J. Zawacki and G. H. Nancollas: The Growth of Non-Stoichiometric Apatite from Aqueous Solution at 37°C. I. Methodology and Growth at pH = 7.4. J. Colloid Int. Sci. 135:20-32 (1990). Holt, C., M. J. J. M. van Kemenade, J. E. Harries, L. S. Nelson, R. T. Bailey, D. W. L. Hukins, S. S. Hasnaian and P. L. De Bruyn: Preparation of Amorphous CalciumMagnesium Phosphates at pH 7 and Characterization by X-Ray Absorption and Fourier Transform Infrared Spectroscopy. J. Cryst. Growth. 92:239-252 (1988). Holt, C., M. J. J. M. van Kemenade, L. S. Nelson, D. W. L. Hukins, R. T. Bailey, J. E. Harries, S. S. Hasnain and P. L. De Bruyn: Amorphous Calcium Phosphates Prepared at pH 6.5 and 6.0. Mater. Res. Bull. 23:55-62

(1989).

Kay, M. I., R. A. Young and A. S. Posner: Crystal Structure of Hydroxyapatite. Nature. 204:1050-1052 (1964). Kazmierczak, T. F.: Kinetics of Precipitation of Calcium Carbonate. Ph.D. thesis. State University of New York, Buffalo (1978). van Kemenade, M. J. J. M. and P. L. deBruyn: A Kinetic Study of Precipitation from Supersaturated Calcium Phosphate Solutions. J. Colloid Int. Sci. 118:564-585

(1987). Koutsoukos, P. G., Z. Amjad, M. B. Tomson and G. H.

Nancollas: Crystallization of Calcium Phosphates. A Constant Composition Study. J. Am. Chem. Soc. 102:1553-1557 (1980). Koutsoukos, P. G. and G. H. Nancollas: Crystal Growth

of Calcium Phosphates. Epitaxial Considerations. J. Cryst. Growth. 53:10-19 (1981). Koutsoukos, P. G.: Kinetics of Precipitation of Hydroxyapatite from Aqueous Solution. Ph.D. thesis, State University of New York, Buffalo (1984). Landis, W. J. and M. J. Glimcher: Electron Diffraction and Electron Probe Microanalysis of the Mineral Phase of Bone Tissue Prepared by Anhydrous Techniques. J. Ultrastruct. Res. 63:188-223 (1979). Lee, D. D., W. J. Landis and M. J. Glimcher: The Solid Calcium Phosphate Mineral Phases in Embryonic Chick Bone Characterized by High-Voltage Electron Diffraction. J. Bone Mineral Res. 1:425-432 (1986). LeGeros, R. Z.: Crystallographic Studies of the Carbonate Substitution in the Apatite Structure. Ph.D. thesis, New York University, New York (1967). LeGeros, R. Z., O. R. Trautz, J. P. LeGeros and E. Klein: Carbonate Substitution in the Apatite Structure. Bull. Soc. Chim. Fr. 1712-1718 (1968). LeGeros, R. Z., J. P. LeGeros, O. R. Trautz and W. P. Shirra: Conversion of Monetite, CaHPO4, to Apatites: Effect of Carbonate on the Crystallinity and Morphology of Apatite Crystallites. Adv. X-Ray Anal. 14:57-66

(1971). LeGeros, R. Z.: The Unit-Cell Dimensions of Human En-

amel Apatite: Effect of Chloride Incorporation. Arch. Oral Biol. 20:63-71 (1974). LeGeros, R. Z., W. P. Shirra, M. A. Miravite and J. P. LeGeros: Amorphous Calcium Phosphates: Synthetic and Biological. In Physico-chimie et Cristallographie des Apatites d'Interet Biologique. pp. 105-115. CNRS, Paris

(1975). LeGeros, R. Z.: Apatites in Biological Systems. Prog. Crystal. Growth Charact. 4:1-45 (1981). LeGeros, R. Z., D. Lee, G. Quirolgica, W. P. Shirra and

L. Reich: In Vitro formation of Dicalcium Phosphate Dihydrate, CaHPO4'2H2O (DCPD). SEM (Scanning Electron Microscopy). 407-418 (1983). LeGeros, R. Z.: Incorporation of Magnesium in Synthetic and in Biological Apatites. In Tooth Enamel IV. pp. 3236 (R. W. Fearnhead and S. Suga, Eds.) Elsevier, Amsterdam (1984). LeGeros, R. Z., R. Kijkowska and J. P. LeGeros: Formation and Transformation of Octacalcium Phosphate, OCP: A Preliminary Report. SEM (Scanning Electron Microscopy). 4:1771-1777 (1984). LeGeros, R. Z., R. Kijkowska, T. Abergas and J. P. LeGeros: Effect of Magnesium on the Formation/Stability of DCPD, CaHPO,42H20. J. Dent. Res. 65

(Abstr.):293 (1986). LeGeros, R. Z., G. Daculsi, R. Kijkowska and B. Kerebel:

The Effect of Magnesium on the Formation of Apatites and Whitlockites. In Magnesium in Health and Disease. pp. 11-19 (Y. Itokawa and J. Durlach, Eds.) John Libbey & Co., London (1982). LeGeros, R. Z., G. Daculsi, I. Orly, T. Abergas and W. Torres: Solution-mediated transformation of Octacalcium Phosphate (OCP) to Apatite. SEM (Scanning Electron Microscopy). 3:129-138 (1989b). Madsen, H. E. L.: Comment on "The Growth of Dicalcium


Phosphate Dihydrate on Octacalcium Phosphate at 25°C by J. C. Heughebaert, J. F. DeRooij and G. H. Nancollas. J. Cryst. Growth. 80:450-452 (1987). Meyer, J. L. and E. D. Eanes: Thermodynamic Analysis of the Amorphous to Crystalline Calcium Phosphate Transformation. Calcif. Tissue Res. 25:59-68 (1978a). Meyer, J. L. and E. D. Eanes: A Thermodynamic Analysis of the Secondary Transition in the Spontaneous Precipitation of Calcium Phosphates. Calcif. Tissue Res.

25:209-216 (1978b). Miller, J. D., A. D. Randolph and G. W. Drach: Obser-

vations Upon Calcium Oxalate Crystallization Kinetics of Simulated Urine. J. Urol. 117:342-345 (1977). Moreno, E. C., M. Kresak and R. T. Zahradnik: Physiochemical Aspects of Fluoride-Apatite Systems Relevant to the Study of Caries. Caries Res. 11:142-171 (1976). Nancollas, G. H.: Phase Transformation during Precipitation of Calcium Salts. In Biological Mineralization and Demineralization. pp. 79-99. (G. H. Nancollas, Ed.) Springer-Verlag, Berlin (1982). Nancollas, G. H.: In Vitro Studies of Calcium Phosphate Crystallization. In Biomineralization - Chemical and Biochemical Perspective. pp. 157-187. (S. Mann, J. Webb, and R. J. P. Williams, Eds.) VCH Publishers, Dordrecht, The Netherlands (1989). Nelson, D. G. A. and J. D. McLean: High-Resolution Electron Microscopy of Octacalcium Phosphate and its Hydrolysis Products. Calcif. Tissue Int. 36:219-232 (1984). Nelson, D. G. A., H. Salimi and G. H. Nancollas: Octacalcium Phosphate and Apatite Overgrowths: A Crystallographic Study. J. Collagen Int. Sci. 110:32-39

(1986a). Nelson, D. G. A., D. J. Wood, J. C. Barry and J. D. B. Featherston: The Structure of (100) Defects in Carbon-

Apatite Microscope ated

Crystallites: High Resolution Electron Study. Ultramicroscopy. 19:253-266 A

(1986b). Nelson, D. G. A. and J. C. Barry: High Resolution Electron

Microscopy of Nonstoichiometric Apatite Crystals. Anat. Rec. 224:265-276 (1989). Newesley, H.: Factors Controlling Apatite Crystallization, with Particular Reference to the Effect of Fluoride and Accompanying Ions. Adv. Oral Biol. 4:11-42 (1970). Neuman, M. W. and W. F. Newman: In The Chemical Dynamics of Bone Mineral. pp. 1-19. University of

Chicago Press, Chicago (1958). Nielsen, A. E. and 0. Sohnel: Interfacial Tensions Electrolyte Crystal-Aqueous Solution, from Nucleation Data. J. Cryst. Growth. 11:233-242 (1971). Ostwald, W.: Studien iiber die Bildung und Umwandlung Fester K6rper. Z. Phys. Chem. 22:289-330 (1897). Posner, A. S., A. Perloff and A. F. Diorio: Refinement of the Hydroxyapatite Structure. Acta Cryst. 11:308-309 (1958). Posner, A. S., F. Betts and N. C. Blumenthal: Role of ATP and Mg in the Stabilization of Biological and Synthetic Amorphous Calcium Phosphates. Calcif Tissue Res. 22(Suppl.):208-212 (1977).

Salimi, M. H., J. C. Heughebaert and G. H. Nancollas:

Crystal Growth of Calcium Phosphates in the Presence of Magnesium Ions. Langmuir. 1:119-122 (1985). Sohnel, O.: Electrolyte Crystal-Aqueous Solution Interfacial Tensions from Crystallization Data. J. Cryst. Growth.

57:101-108 (1982). Taves, D. R.: Similarity of Octacalcium Phosphate and

Hydroxyapatite

Structures. Nature. 200:1312-1313

(1963). Taves, D. R. and R. C. Reedy: A Structural Basis for the

Transphosphorylation of Nucleotides with Hydroxyapatite. Calcif. Tissue Res. 3:284-292 (1969). Termine, J. D. and A. S. Posner: Amorphous/Crystalline Interrelationships in Bone Mineral. Calcif. Tissue Res.

1:8-23 (1967). Termine, J. D. and A. S. Posner: Calcium Phosphate For-

mation In Vitro. I. Factors Affecting Initial Phase Separation. Arch. Biochem. Biophys. 140:307-317 (1970). Termine, J. D., R. A. Peckauskas and A. S. Posner: Calcium Phosphate Formation In Vitro. II. Effects of Environment on Amorphous-Crystalline Transformation. Arch. Biochem. Biophys. 140:318-325 (1970). Termine, J. D. and E. D. Eanes: Comparative Chemistry of Amorphous and Apatitic Calcium Phosphate Preparations. Calcif. Tissue Res. 10:171-197 (1972). Tomazic, B. B., E. S. Etz and W. E. Brown: Nature and Properties of Cardiovascular Deposits. SEM (Scanning Electron Microscopy). 1:95-105 (1987). Tomazic, B. B., M. S. Tung, T. M. Gregory and W. E. Brown: Mechanism of Hydrolysis of Octacalcium Phosphate. SEM (Scanning Electron Microscopy). 3:119-127

(1989). Tomson, M. B. and G. H. Nancollas: Mineralization Ki-

netics: A Constant Composition Approach. Science. 200:1059-1060 (1978). Tung, M. S. and W. E. Brown: An Intermediary State in Hydrolysis of Amorphous Calcium Phosphate. Calcif. Tissue Int. 35:783-790 (1983). Vogel, J. J.: Calcium Phosphate Solid Phase Induction by Dioleoylphosphatidate Liposomes. J. Collagen Int. Sci. 111:152-159 (1986). Williams, G. and J. D. Sallis: Structure-Activity Relationship of Inhibitors of Hydroxyapatite Formation. Biochem. J. 184:181-184 (1979). Williams, G. and J. D. Sallis: Structural Factors Influencing the Ability of Compounds to Inhibit Hydroxyapatite Formation. Calcif. Tissue Int. 34:169-177 (1982). Young, R. A. and W. E. Brown: Structures of Biological Minerals, In Biological Mineralization and Demineralization. pp. 101-141. (G. H. Nancollas, Ed.) Springer-Verlag, Berlin (1982). Zapanta-LeGeros, R.: Effect of Carbonate on the Lattice Parameters of Apatite. Nature. 206:403-404 (1965). Zawacki, S. J., M. H. Salimi and G. H. Nancollas: The Growth of Calcium Phosphates. In Geochemical Processes at Mineral Surfaces. pp. 650-662. ACS Symposium Series 323 (J. A. Davis and K. F. Hayes, Eds.) American Chemical Society, Washington, D.C. (1986).


Zawacki, S. J., J. C. Heughebaert and G. H. Nancollas:

The Growth of Non-Stoichiometric Apatite from Aqueous Solution at 37°C. II. Effect of pH Upon the Precipitated Phase. J. Colloid Int. Sci. 135:33-44 (1990).


The role of an octacalcium phosphate in the re-formation of infraspinatus tendon insertion.

Phase transformation of octacalcium phosphate in vivo and in vitro.

Importance of nucleation in transformation of octacalcium phosphate to hydroxyapatite.

Thermal conversion of octacalcium phosphate into hydroxyapatite.

Osteoconductive property of a mechanical mixture of octacalcium phosphate and amorphous calcium phosphate.

Visualising phase change in a brushite-based calcium phosphate ceramic.

The increase of apatite layer formation by the poly(3-hydroxybutyrate) surface modification of hydroxyapatite and β-tricalcium phosphate.

Bioceramics composed of octacalcium phosphate demonstrate enhanced biological behavior.

The effects of magnesium and fluoride on the hydrolysis of octacalcium phosphate.

Interactions between acidic matrix macromolecules and calcium phosphate ester crystals: relevance to carbonate apatite formation in biomineralization.

Strontium substituted bioactive glasses for tissue engineered scaffolds: the importance of octacalcium phosphate.

Shear-mediated crystallization from amorphous calcium phosphate to bone apatite.

Zinc induces apatite and scholzite formation during dentin remineralization.

Octacalcium phosphate collagen composite facilitates bone regeneration of large mandibular bone defect in humans.

First clinical application of octacalcium phosphate collagen composite in human bone defect.

Nanostructured Titania Surfaces on Osteoblast Response.

Octacalcium phosphate ceramics combined with gingiva-derived stromal cells for engineered functional bone grafts.

Effect of hydroxyapatite, octacalcium phosphate and calcium phosphate on the auto-flocculation of the microalgae in a high-rate algal pond.

Extraction and characterisation of apatite- and tricalcium phosphate-based materials from cod fish bones.

Apatite formation on the surface of Ceravital-type glass-ceramic in the body.

Influence of Calcium Phosphate and Apatite Containing Products on Enamel Erosion.

The calcium phosphate matrix of FGF-2-apatite composite layers contributes to their biological effects.

Strontium ions substitution in brushite crystals: the role of strontium chloride.

Lead phosphate formation in soils.