Will Fluoride Toughen or Weaken Our Teeth ... - ACS Publications

Nov 24, 2009 - In this regard, much effort has been focused on the mineralization of HAP as it is the major component of teeth and bones.(5-12) It is ...
0 downloads 0 Views 4MB Size
Download PDF
J. Phys. Chem. B 2009, 113, 16393–16399

16393

Will Fluoride Toughen or Weaken Our Teeth? Understandings Based on Nucleation, Morphology, and Structural Assembly Zhiqiang Wang,†,‡ Guobin Ma,†,‡ and Xiang Yang Liu*,† Department of Physics, National UniVersity of Singapore, 2 Science DriVe 3, Singapore 117542, and National Laboratory of Solid State Microstructures and Department of Physics, Nanjing UniVersity, Nanjing 210093, China ReceiVed: June 22, 2009; ReVised Manuscript ReceiVed: October 20, 2009

Fluoride exerts a substantial impact on the biominearlization of hydroxyapatite (HAP). In this study, the effect of fluoride on the nucleation kinetics of HAP was examined quantitatively in a simulated body fluid. It was found for the first time that fluoride promotes the nucleation of HAP. Furthermore, both the concentration of fluoride and the supersaturation of HAP play a key role in the formation of an ordered assembly of rodlike HAP crystallites. At relatively low concentrations of fluoride, an ordered assembly of rod-like crystallites similar to the structure of enamel could be obtained in the simulated body fluid, which could be explained by the self-epitaxial nucleation-mediated assembly. At high concentrations, fluoride may lead to the deterioration of the ordered assembled structure of HAP crystallites. That result is caused by the fast nucleation and growth of HAP crystallites when the supply of nucleation seeds is limited, resulting in the formation of spherulites. Our results illustrate how and why fluoride influences the self-assembled structures of HAP crystallites at low and high concentrations of F-. Thus, they will substantially advance our understanding on the impact of fluoride ions on the mineralization of HAP and consequently on the dental and bone fluorosis. Introduction Biological mineralization and demineralization of hydroxyapatite (HAP, Ca5(PO4)3OH) and CaCO3 play a vital role in an organism.1-4 In nature, the highly ordered mineral structures performing different functionalities, such as dentin enamel and mollusc shell, exhibit superior mechanical properties. Such amazing structures have been attracting the interest of biologists, physicists, chemists, and materials scientists. However, the understanding of the mechanism of biomineralization/demineralization and the mimicking of the highly ordered assembly of biomaterials remain a challenge. In this regard, much effort has been focused on the mineralization of HAP as it is the major component of teeth and bones.5-12 It is found that fluoride not only plays a key role in prevention and control of dental caries but also leads to dental and bone fluorosis.13 In the past few decades, many papers reported the effect of fluoride on the morphology of HAP under different conditions.14-19 However, despite recent research efforts, the way that fluoride affects the nucleation and assembly of HAP crystallites remains unknown. Notice that the mineralization process includes nucleation and growth of the mineral crystallites. The nature of nucleation will determine the structure and morphology of the crystallites. To establish a comprehensive understanding of the effect of fluoride on nucleation, the morphology and orientation of HAP can help us control and mimic the mineralization process of HAP in the presence of fluoride. Our goal is to obtain a fundamental understanding of the effect of fluoride on the mineralization of HAP and, in particular, on the ordering of the HAP crystallites’ * To whom correspondence should be addressed. Address: MIT-Singapore Alliance and Department of Physics, National University of Singapore, 2 Science Drive 3, Singapore 117542. Tel.: 65-68742812. Fax: 65-67776126. E-mail: [email protected]. † National University of Singapore. ‡ Nanjing University.

TABLE 1: Nominal Ion Concentrations of the cSBF in Comparison with Those of Human Blood Plasma and the Ion Concentrations of Calcium, Phosphate, and cSBF Solutions Used in the Nucleation Experimentsa

ion

blood plasma (mM)

cSBF (mM)

Na+ K+ Mg2+ Ca2+ ClHCO3HPO42SO42-

142.0 5.0 1.5 2.5 103.0 27.0 1.0 0.5

142.0 5.0 1.5 2.5 147.8 4.2 1.0 0.5

calcium solution (mM)

phosphate s olution (mM)

284.0 10.0 3.0 11-21.0 295.6 8.4 4.4-8.4 1.0

cSBF in the experiments (mM) 142.0 5.0 1.5 5.5-10.5 147.8 4.2 2.2-4.2 0.5

a The calcium and phosphate concentrations used in each nucleation experiment are 5.5, 6, 6.5, ..., 10.5 and 2.2, 2.4, 2.8, ..., 4.2 mM, respectively.

assembly, which is directly related to the mechanical properties of dental enamel.20 As found recently,21-24 the assembly of HAP crystallites is brought about by nucleation (or self-epitaxial nucleation)mediated self-assembly. In this connection, we will examine the nucleation kinetics and the structural match/mismatch growth of HAP in a simulated body fluid with fluoride included as an additive. Experimental Section To investigate the nucleation kinetics of biomineralization and to mimic that process in the human body, the nucleation of HAP was carried out in a conventional simulated body fluid (cSBF).25 The ion concentrations of the cSBF, shown in Table 1, are similar to the physiological concentrations in the human blood plasma. To study the effect of fluoride on the mineraliza-

10.1021/jp905846p  2009 American Chemical Society Published on Web 11/24/2009

16394

J. Phys. Chem. B, Vol. 113, No. 51, 2009

Wang et al.

tion of HAP, F- was introduced by adding NaF into the cSBF; three concentrations of F-, 0, 1, and 5 mM, were examined. In order to facilitate the accurate measurement of induction time, the cSBF solutions were prepared by mixing calcium and phosphate solutions together. Before preparing the cSBF solution, all of the apparatuses were carefully washed with 1.0 M HCl, neutral detergent, and deionized water. For the preparation of calcium solution, at first, approximately 700 mL of deionized water was poured into a 1000 mL glass bottle, and this was stirred using a magnetic bar at 37 °C. According to the ion concentrations listed in Table 1, the reagents were dissolved in the deionized water in the sequence of NaCl (99.5%, Sigma), NaHCO3 (99.5%, Sigma), KCl (99%, Sigma), MgCl2 · 6H2O (99%, Merck), CaCl2 (99%, Sigma), and Na2SO4 (99.5%, Sigma). After that, the solution was buffered at pH ) 7.40 at 37 °C with tris(hydroxylmethyl)aminomethane (TRIS; 99%, Sigma) and 1.0 M HCl aqueous solution. Finally, the solution was transferred to a 1000 mL volumetric glass flask and cooled to approximately 20 °C, and the total volume was adjusted to 1000 mL by adding deionized water. For the preparation of phosphate solution, the procedures are similar to those for preparing calcium solution (the reagent is K2HPO4 · 3H2O (99%, Sigma)). For the cSBF with fluoride as an additive, F- was introduced by dissolving NaF into the phosphate solution, and three concentrations of F-, 0, 1, and 5 mM, were prepared. The nucleation kinetics experiments of HAP were performed on a dynamic light scattering (DLS) system (BI-200SM; Brookhaven Instruments Corporation) with a He-Ne laser (632.8 nm) source and a photomultiplier tube (PMT) detector.22,24,26-28 The system could detect particles of a size down to several nanometers, which allowed an in situ measurement of the nucleation process and of the size increase of the nuclei. To measure the induction time of HAP within a range of supersaturations, a series of supersaturations of HAP were obtained by adjusting the concentrations of calcium and phosphate ions in the cSBF solution according to that listed in Table 1. The ratio of calcium to phosphate ions was fixed at 5:2, the same as that in the cSBF. Before carrying out the experiments, the nucleation vessels were cleaned properly, and the calcium and phosphate solutions were filtered three times by filters with a pore size of 0.2 µm. Then, the calcium and phosphate solutions were mixed and introduced into a quartz cell via a stopped-flow system. This is a device which can facilitate the instant mixing of two different solutions. At the moment of injection, there appeared a sudden increase in the intensity of the scattering light, and this point is defined as time “0” for the measurement of induction time. The experimental temperature was kept at 25 °C by a water bath. When the nucleation measurement via DLS was finished, the quartz cell with cSBF solution was kept in the water bath for 3-5 h until white product deposited at the bottom of the quartz cell. After that, the precipitates were centrifuged and washed three times with deionized water and dried in air at 25 °C. X-ray diffraction (XRD; PANalytical’s X’Pert PRO MRD) with Cu KR radiation and scanning electron microscopy (SEM; JEOL JSM6700F) were carried out to analyze the composition and morphology of the precipitates under different conditions.

3 [a(Ca2+)]5[a(PO34 )] [a(OH )] ∆µ ) kT ln Ksp(HAP)

Kinetic Model for HAP Nucleation. The supersaturation σ is the source of the thermodynamic driving force for biomineralization and can be defined by29

()

a ∆µ ) ln kT ae

(1)

where ∆µ denotes the chemical potential difference between the actual state and the equilibrium state of the crystallizing

(2)

where Ksp is the solubility product at a given temperature (in our case, T ) 298 K). To obtain the activities of the ions in the solution (e.g., a(Ca2+)), first, the ionic strength I is calculated by31

I)

1 2

∑ cizi2

(3)

i

where ci and zi represent the concentration and charge of species i, respectively. The calculation of the activity coefficient γ by the Debye-Huckel law is subject to some limitations because of the point charge assumption made in that law. An estimation using this law is reliable only for extremely dilute solutions, and therefore, it can only be applied at low ionic strength. However, in the extended Debye-Huckel theory, not only the ion size is taken into account by incorporating the mean effective diameter of the hydrated ions, but also, the temperature and solvent effect are taken into account, thereby improving the accuracy of the prediction. Thus, it is appropriate to calculate the activity coefficient and the activity of the ions in solutions with much higher ionic strengths, especially in multi-ionic complex systems, by the following equations31

log γi )

-0.51zi2√I

( )

Ri√I 1+ 305 ai ) γi × ci

(4)

(5)

where γi, Ri, and ci are the activity coefficient, the hydrated ionic radius, and the activity of species i, respectively. In general, in the process of nucleation, the constituent molecules or ions in the solution may, upon collision, join into groups of two or more particles to form dimers, trimers, tetramers, and so forth. Before they can reach a critical radius rc, the embryos are unstable, even when a positive thermodynamic driving force ∆µ is applied. To reach rc, an energy barrier, the so-called nucleation barrier ∆G*, needs to be overcome. At this stage, the main issue is how the embryos can reach the critical radius. When the embryos become stable after they overcome the nucleation barrier, the second stage of the phase transition begins, growth. In the case of homogeneous nucleation, the effect of the substrate is negligible. Then, the nucleation barrier for a spherical nucleus is given by29,32

∆G*homo )

Results and Discussion

ln(1 + σ) )

species, k is the Boltzmann constant, T is the absolute temperature, and a and ae are the actual and equilibrium activities of a stoichiometric crystallizing unit, respectively. As biomineralization occurs in electrolyte solutions, the thermodynamic driving force ∆µ is given in the case of HAP by30

16πγ3cfΩ2 3[∆µ]2

)

16πγ3cfΩ2 3k2T2[ln(1 + σ)]2

(6)

where ∆G*homo is the nucleation barrier for homogeneous nucleation, γcf is the specific interfacial free energy between the crystals and the fluid, and Ω is the volume of the growth unit. In the case of heterogeneous nucleation, the substrate has a significant influence on the nucleation of the crystal. In the presence of a substrate, the nucleation barrier is reduced to

Will Fluoride Toughen or Weaken Our Teeth?

∆G*hetero ) ∆G*homo f

J. Phys. Chem. B, Vol. 113, No. 51, 2009 16395

(7)

[0 e f e 1]

where f is a factor describing the lowering of the nucleation barrier ∆G* resulting from the occurrence of foreign bodies. On the one hand, the presence of a substrate facilitates the nucleation of the embryos and speeds up the nucleation kinetics. On the other hand, the occurrence of a foreign body will affect the transport of the growth unit to the surface of the crystalline clusters. In the case of homogeneous nucleation, the growth units can be incorporated into the nucleus through collisions from all directions. However, nucleation on a foreign particle is accompanied by a reduction in the “effective surface” of the nucleus where the growth units are incorporated into the nucleus. This effect, similar to the effect of a shadow, tends to slow down the nucleation kinetics, which counteracts the effect of the substrate in lowering the nucleation barrier. Taking into account the effect of the substrate on both the nucleation barrier and the transport process, the nucleation rate J is given according to the model23 as

[

J ) (Rs)2N0f''(m)[f(m)]1/2B exp -

]

∆G*homo f(m) kT

(8)

with

1 f''(m) ) (1 - m) 2

(9)

1 f(m) ) (2 - 3m + m3) 4

(10)

and

where Rs and N0 are the radius and density of the substrates (or “seeds”) respectively and m (-1 e m e 1) depends on the interaction and interfacial structural match between the crystalline phase and the substrates. Both f(m) and f ′′(m) are functions of m. When m ) -1, f(m) and f ′′(m) ) 1, corresponding to weak interaction and poor structural match between the substrate and nucleating phase. In this case, the substrate exerts almost no influence on the nucleation. This is equivalent to the case of homogeneous nucleation (∆G* ) ∆G*homo). In the case where m f 1, one has f(m) and f ′′(m) ) 0. This implies that the nucleating phase has strong interaction and excellent structural match with the substrate (∆G* ) 0). The crystallites are well-oriented and ordered with respect to the structure of the substrate. Normally, heterogeneous nucleation occurs in the range of m between 1 and -1, or f(m) between 0 and 1, depending on the nature of the substrate surface and supersaturation. The factor f(m) in the exponential part of eq 8 describes the reduction of the nucleation barrier from a genuine homogeneous nucleation ∆G*homo to the actual heterogeneous nucleation ∆G*heter in the presence of a substrate. The factors f(m) and f′′(m) appearing in the pre-exponential part of eq 8 describe the negative shadow effect of the substrate on the nucleation kinetics. The two contradictory effects play different roles in different regimes. At low supersaturations where the nucleation barrier is very high (cf. eq 6), heterogeneous nucleation with an optimal structural match between the crystalline phase and the substrate (f(m) f 0) will be kinetically favored. In this case, the nucleation of crystalline materials will be best templated by substrates. However, at higher supersaturations, the nucleation barrier becomes less important. Then, instead of the nucleation barrier, it is the “effective surface” described by the factors f(m) and f′′(m) that dominates in controlling the nucleation kinetics. Thus, nucleation on substrates with larger f(m) and f′′(m) will be favored and lead to a mismatch structure.

To describe the kinetics of nucleation, the induction time ts of nucleation at different supersaturations will be measured. By definition, the nucleation rate J can be expressed as

J)

1 (tsV)

(11)

where V is the volume of the system. It then follows from eq 8 that21-24,26-28

ln ts )

κf(m) - ln{V(RS)2N0f''(m)[f(m)]1/2B} 2 [ln(1 + σ)] (12)

with

κ)

16πγ3cfΩ2 3(kT)3

(13)

where B is the kinetic constant. In general, the interfacial structure match between the crystalline phase and the substrate changes from a completely correlated and ordered match state to a completely uncorrelated and disordered mismatch state as m varies from 1 to -1. For instance, an excellent structural match (m f 1) implies that ∆G*heter vanishes almost completely. This occurs only when the growing crystals are well-oriented and ordered with respect to the structure of the substrate. In contrast, in the case of m f -1, the substrate exerts almost no influence on the nucleation, and the nucleation is controlled by the kinetics of homogeneous nucleation, which results in the emergence of disordered nuclei. Due to the anisotropy of the crystalline phase, the factors m and f(m) take discrete values corresponding to some crystallographic orientations {hkl} of the optimal structural match position with respect to the secondary and tertiary optimal structural matches. Therefore, we expect to observe in the ln ts versus 1/[ln(1 + σ)]2 plot of eq 12 a set of intersecting straight lines with different slopes. These slopes depend on the value of f(m) (κ will remain constant in a given system) and indicate that the nucleation is dominated by a sequence of progressive heterogeneous nucleation processes. It follows from the above analysis that if σ progressively increases from low supersaturations to high supersaturations, nucleation will be governed by a sequence of heterogeneous nucleation processes associated with progressively increasing discrete values of f(m). This principle shows that the quantity f(m) describing the interfacial correlation between crystallites and substrates will increase with supersaturation, as illustrated in Figure 1. This implies that the increase in supersaturation will drive the substrates/crystallites from an interfacial structural match state (a lower f(m)) to a state of higher mismatch (a higher f(m)). This phenomenon is referred to as supersaturation-driven interfacial structural mismatch.24 As mentioned above, the changes from one state to the other occur abruptly at certain supersaturations (such as A, B in Figure 1) due to the anisotropy of the crystalline phase. Therefore, it is predicted that the space in the plot of f(m) versus σ would be divided into several regimes by a set of ladder-like lines, as shown in Figure 1. Verification of the Kinetic Model. The concrete evidence supporting the above analysis of the nucleation kinetics can be identified from the ln ts versus 1/[ln(1 + σ)]2 plot within a range of supersaturations. The induction time ts was measured by a dynamic light scattering system,24 and the factor ln(1 + σ) can be calculated from eqs 1-5. As shown in Figure 2a, the curve plotted by a solid line, presenting the nucleation of HAP in the cSBF system, can be, in principle, fitted by three pairwise

16396

J. Phys. Chem. B, Vol. 113, No. 51, 2009

Wang et al.

Figure 3. The XRD patterns of the precipitates from the system with/ without fluoride. The patterns 1, 2, and 3 correspond to the XRD pattern of the precipitates in cSBF with 0, 1, and 5 mM fluoride as additives, respectively. Figure 1. Illustration for supersaturation-driven interfacial structural mismatch. With the increase of supersaturation, the interfacial correlation factor f(m) will increase abruptly at certain supersaturations, such as A, B, ..., corresponding to the transition from an ordered and structural matched to a less ordered and structural mismatched crystal/ substrate interface; m1 > m2 > m3 > m4.

Figure 2. (a) Plot of ln ts versus 1/[ln(1 + σ)]2 for HAP nucleation in the presence/absence of fluoride. The solid, dashed, and dotted curves describe the nucleation of HAP in cSBF with 0, 1, and 5 mM fluoride as additives, respectively. (b) Plot of κf(m) versus σ. The step increase of κf(m) with the increase of supersaturation corresponds to the supersaturation-driven interfacial structure mismatch.

intersecting straight lines with different slopes, partitioning the space into three regimes. The supersaturation increases progressively from regime III to regime II and finally to regime I. If

we let f(m3), f(m2) and f(m1) denote f(m) in these three regimes, the slope increases sharply from κf(m3) to κf(m2) and κf(m1) at the critical points. This confirms that the nucleation of HAP in cSBF is dominated by a sequence of progressive heterogeneous nucleation processes. Similar results were also obtained from the ln ts versus 1/[ln(1 + σ)]2 plot for the nucleation of HAP in the cSBF system using 1 mM fluoride as an additive (the curve plotted by the dashed line). Effect of Fluoride on the Nucleation of HAP. In this section, we will focus on the effect of fluoride on the nucleation kinetics of HAP. In Figure 2a, the solid curve describes the nucleation kinetics of HAP in the cSBF system, whereas the dashed and the dotted curves describe the nucleation kinetics under the influence of fluoride (the fluoride concentrations are 1 and 5 mM, respectively). The kinetics described by the solid curve is very different from that described by the dashed and the dotted curves in the presence of fluoride. These curves are shifted further downward with respect to the solid curve. At the same supersaturation, the induction time of HAP in the presence of fluoride is much shorter than that in its absence. This implies that fluoride promotes the nucleation of HAP. Comparison of the curves with different fluoride concentrations shows that a higher fluoride concentration will increase the rate of HAP nucleation. Due to the very rapid nucleation of HAP in cSBF with 5 mM fluoride, the induction time of HAP in regime I is too short to be measured (ts , 1 s). Thus, only one straight line (the dotted curve) is obtained in such a system within a certain range of supersaturations. For the cSBF with 0, 1, and 5 mM fluoride, the plots of κf(m) versus σ are shown in Figure 2b. It is explicitly seen that, in the systems with 0 and 1 mM fluoride, κf(m) sharply increases at some critical points with the progressive increase of σ, and the space is divided into three regimes by a set of ladder-like lines. This result is in very good agreement with the predicted changes shown in Figure 1 and confirms the supersaturation-driven interfacial structural mismatch from the point of view of nucleation kinetics. Effect of Fluoride on Composition of HAP. The composition of the precipitates obtained in the systems with/without fluoride was examined by X-ray diffraction (XRD). As shown in Figure 3, the XRD pattern of the precipitates from the system without fluoride (pattern 1) accords well with hexagonal hydroxyapatite (HAP) with lattice constants a ) 9.416 and c ) 6.874 Å (JCPDS 84-1998). The energy-dispersive X-ray spectroscopy (EDX) results (see Supporting Information) illustrate that the precipitates from cSBF are comprised of Ca, P, and O elements, and the ratio of Ca and P is ∼1.65. This is

Will Fluoride Toughen or Weaken Our Teeth? consistent with that in HAP (Ca/P ) 1.67). The XRD and EDX results imply that the precipitates obtained in cSBF are hexagonal HAP. Both patterns 2 and 3, obtained after introducing fluoride into the cSBF and corresponding to the XRD patterns of the precipitates from the system with 1 and 5 mM fluoride, respectively, are very similar to pattern 1. This suggests that the precipitates from the system with fluoride as additives are of the hexagonal HAP structure. On the other hand, the XRD peaks in patterns 2 and 3 are rather strong, indicating that fluoride improves the crystallinity of HAP in cSBF.18 At higher concentrations of fluoride (1 and 5 mM), no other calcium phases (such as octacalcium phosphate (OCP; Ca8H2(PO4)6 · 5H2O), tricalcium phosphate (TCP; Ca3(PO4)2), brushite (DCPD; CaHPO4 · 2H2O), and CaF2) besides the hexagonal HAP structure, were found in the precipitates. The EDX results of the precipitates from the system with fluoride (see Supporting Information) indicate that there is a certain quantity of fluoride in the precipitates, which suggests that a fraction of the hydroxyl groups in HAP was substituted by fluoride during the mineralization process (in the case of 1 mM fluoride, ∼5.8%, and in the case of 5 mM fluoride, ∼10%). The XRD and EDX results are proposed to explain why the nucleation of HAP in the system with fluoride was much faster than that without fluoride. The introduction of fluoride during the nucleation of HAP in cSBF will cause the substitution of a few of the hydroxyl groups in HAP by fluoride and result in the formation of Ca5(PO4)3(OH)1-xFx (in our study, x e 0.1). For the fluoridated HAP (Ca5(PO4)3(OH)1-xFx), the solubility product constant decreases with the increase of fluoride (when x < 0.57).33 Therefore, at the same ion concentrations, the nucleation of HAP with partial substitution of fluoride may become easier and faster than the nucleation of HAP alone. On the other hand, introducing fluoride in the cSBF system will cause the drastic increase in the supersaturation of fluoridated HAP, resulting in the nucleation of fluoridated HAP being greatly favored. From the point of view of nucleation kinetics, the increase of supersaturation will give rise to the fact that the nucleation of fluoridated HAP in the system with fluoride becomes faster. Effect of Fluoride on Morphology of HAP. The morphology and the assembly/aggregation state of the precipitates from the system with/without fluoride were examined by scanning electron microscopy (SEM). A typical SEM image of the precipitates (in regime II) from the system without fluoride is shown in Figure 4a. The precipitates consist of plate-like HAP crystallites, which are assembled into spherulitic microstructures with a diameter of ∼2 µm. Such assemblies of HAP crystallites are incompact and disordered. The morphology of the precipitates in regimes I and II is similar to that in regime II (see Supporting Information). It should be highlighted here that adding fluoride into the cSBF system modifies the morphology of the precipitates and exerts a strong impact on the structure of the HAP crystallite assembly. As shown in Figure 4b-f, when the system contained 1 mM fluoride, the precipitates were composed of rod-like crystallites. Single rod-like crystallites are obtained in regime III (see Figure 4b). This is a very low supersaturation regime, in which self-epitaxial nucleation can not take place; the reason is that self-epitaxial nucleation at the side face of the long rodlike HAP crystallites requires a critical supersaturation.34 In this regime, the HAP crystallites randomly aggregate due to the interfacial tension (see Figure 5a). The ordered assemblies of rod-like crystallites are obtained at a relatively low supersaturation in regime II (see Figure 4c and d). There, the supersatu-

J. Phys. Chem. B, Vol. 113, No. 51, 2009 16397

Figure 4. (a) The SEM image of the precipitates from regime II in the system without fluoride. (b)-(f) The SEM images of the precipitates in the system with 1 mM fluoride as an additive; (b) single rod-like crystallites obtained in regime III; (c) and (d) compact and ordered assembly of rod-like crystallites obtained at relatively low supersaturations in regime II; (e) small mismatch rod-like crystallites obtained at relatively high supersaturations in the regime between II and I as shown in Figure 1; (f) incompact and disordered assembly of rod-like crystallites obtained in regime I. (g) and (h) The SEM images of the precipitates in the system with 5 mM fluoride as an additive; (g) lowmagnification SEM image of the sphere-like crystallites; (h) highmagnification SEM image of the sphere-like crystallites; the spherelike structures are composed of small particles. Scale bars: 200 nm.

Figure 5. Schematic illustration of the formation process of HAP assembly in cSBF with 1 mM F- at different supersaturations. (a)-(d) The mineralization processes of the assembly of HAP crystallites in cSBF with 1 mM F-; (a) in regime III, (b) in regime II, (c) in the regime between II and I, and (d) in regime I, as shown in Figure 2, respectively.

ration is high enough to trigger self-epitaxial nucleation at the side face of long rod-like HAP crystallites. In this process, the daughter HAP crystallites share the same orientation with the parent crystals through self-epitaxial nucleation and growth (see Figure 5b). Because at low supersaturations a good structural match between daughter and parent crystals (f(m) f 0) gives rise to a much lower nucleation barrier, the assembled crystallites exhibit an excellent structural synergy. Figure 4d shows the ordered assembly of the rod-like crystallites. This is

16398

J. Phys. Chem. B, Vol. 113, No. 51, 2009

Figure 6. The change of the nucleation barrier ∆G* with supersaturation (cf. eqs 6 and 7).

similar to the microstructures of enamels, which are compact and highly ordered. As the supersaturation increases, assemblies of rod-like crystallites exhibiting small structural mismatch are obtained between regimes II and I, as shown in Figure 2 (see Figure 4e). The disordered assemblies of rod-like crystallites (see Figure 4f) are obtained while further increasing the supersaturation to regime I. The orientations of the crystallites are random, in which case the assemblies of the crystallites are open and porous. The formation of the structural match/mismatch assembly can be understood as follows. As shown in Figure 6, due to the high nucleation barrier at low supersaturations, it is very difficult for the mismatch epitaxial nucleation to occur. Such conditions are only suited for matched epitaxial nucleation and growth, explaining the formation of the ordered HAP assembly observed in Figure 5b. As the supersaturation increases, the nucleation barrier for mismatch epitaxial nucleation drops rapidly (cf. eqs 6 and 7). At relatively high supersaturations, the requirement of the structural match between the daughter crystals and the substrates becomes less strict, in view of the decrease of the nucleation barrier. This facilitates self-epitaxial nucleation leading to formation of the assembly of HAP crystallites with small mismatch (see Figure 5c). As the supersaturation increases further, the nucleation barrier for mismatch epitaxial nucleation totally collapses. Self-epitaxial nucleation occurs much more easily, resulting in a severe interfacial structural mismatch. In this case, the HAP assembly will often be randomly and highly branched (see Figure 5d). Here, the morphological evidence also confirms the existence of a supersaturation-driven interfacial structural mismatch. When the concentration of fluoride in cSBF reaches 5 mM, the precipitates in regime II adopt a spherulitic pattern. The diameter of the spherulites ranges from 100 to 200 nm (see Figure 4g). Some spherulites are assembled together to form a cluster. As shown in Figure 4h, the spherulites are comprised of small crystallites with size ∼ 20 nm. Such spherulitic structures are porous, and the mechanical property is incomparable to those of the compact ones. We notice that the graphs given in Figure 2 show that once the fluoride concentration in cSBF reaches 5 mM, regimes I, II, and III become indistinguishable. As an outcome, the morphology and the state of assembly/aggregation of crystallites in regimes I, II, and III are all very similar to that shown in Figure 4g and h. The cause of the emergence of spherulites as opposed to normal crystals has been studied extensively.35-40 Spherulites follow a pattern where the radial crystalline arms initiate from a common center (the core).38-40 This happens when the initial nucleation from nucleation seeds (normally dust particles) is

Wang et al.

Figure 7. Schematic illustration of the crystallization process of (a) HAP rods and (b) HAP spherulites in cSBF with F-. In the nucleation process, the adsorption of F- on the surface of the nucleation seeds may inactivate some of them. Therefore, the rise of F- concentration will lead to “the decrease of the nucleation seeds” and trigger the formation of spherulites due to the fact that the fast nucleation and growth of HAP crystallites occurs in limited nucleation seeds.

Figure 8. Schematic illustration of the influence of F- on the HAP assembly in cSBF.

hindered at very low supersaturations. The fact that the fast nucleation and growth of HAP crystallites occurs in a limited amount of nucleation seeds gives rise to the formation of spherulites.38-40 In this process, the adsorption of F- on the surface of the seeds may inactivate some of them (see Figure 7). Therefore, the rise of the F- concentration will lead to an effective “decrease of nucleation seeds” and trigger the formation of spherulites. The aforementioned results indicate that the introduction of fluoride will lead to the formation of a compact and highly ordered assembly of rod-like HAP crystallites (see Figure 8). Since the toughness of human dental enamel will increase exponentially as the ordering of the crystallite assembly is enhanced,20 it is proposed that such a structural change may lead to the toughening of dental enamel. On the other hand, as the concentration of F- increases, the spherulitic growth of HAP is promoted (see Figure 8). As the spherulitic pattern is a sort of localized nonordered crystal network, which will weaken the overall mechanical property, the occurrence of such a crystallite assembly may lead to the weakening of the HAP assembly structure. This may be regarded as the reason for the deterioration of the teeth structure during the formation of teeth in the environment with high concentration of fluoride. Conclusions In this paper, at first, we quantitatively investigated the effect of fluoride on the nucleation kinetics of HAP in cSBF by the DLS system. The results show that the introduction of fluoride

Will Fluoride Toughen or Weaken Our Teeth? promotes the nucleation of HAP in cSBF. Second, we studied the effect of fluoride on the morphology and the assembly structure of HAP crystallites during the biomineralization in cSBF. It is found that the fluoride is able not only to promote the ordered assembly of HAP at relatively low fluoride concentrations but also to induce the spherulitic growth of HAP at high fluoride concentrations. This explains why a low concentration of fluoride will help toughen the assembly of HAP crystallites while, on the other hand, a high concentration of fluoride may lead to the deterioration of the HAP assembly. Acknowledgment. The research is supported by Singapore ARC MOE funding (Project No. T206B1114). Supporting Information Available: The calculation of ln(1 + σ) for Ca5(PO4)3(OH)1-xFx (x e 0.1), experimental details, EDX results, and SEM images of the precipitates from cSBF in regime I and regime III. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Berman, A.; Addadi, L.; Weiner, S. Nature 1988, 331, 546. (2) Mann, S. Nature 1993, 365, 499. (3) Berman, A.; Hanson, J.; Leiserowitz, L.; Keotzle, T. F.; Weiner, S.; Addadi, L. Science 1993, 259, 776. (4) Stupp, S. I.; Braun, P. V. Science 1997, 277, 1242. (5) Wen, H. B.; Wolke, J. G. C.; deWijn, J. R.; Liu, Q.; Cui, F. Z.; deGroot, K. Biomaterials 1997, 18, 1471. (6) Long, J. R.; Dindot, J. L.; Zebroski, H.; Kiihne, S.; Clark, R. H.; Campbell, A. A.; Stayton, P. S.; Drobny, G. P. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 12083. (7) Wen, H. B.; Moradian-Oldak, J.; Fincham, A. G. Biomaterials 1999, 20, 1717. (8) Murphy, W. L.; Mooney, D. J. J. Am. Chem. Soc. 2002, 124, 1910. (9) Song, J.; Saiz, E.; Bertozzi, C. R. J. Am. Chem. Soc. 2003, 125, 1236. (10) He, G.; Dahl, T.; Veis, A.; George, A. Nat. Mater. 2003, 2, 552. (11) Yamagishi, K.; Onuma, K.; Suzuki, T.; Okada, F.; Tagami, J.; Otsuki, M.; Senawangse, P. Nature 2005, 433, 819. (12) Tlatlik, H.; Simon, P.; Kawska, A.; Zahn, D.; Kniep, R. Angew. Chem., Int. Ed. 2006, 45, 1905. (13) Aoba, T.; Fejerskov, O. Crit. ReV. Oral Biol. Med. 2002, 13, 155. (14) Christoffersen, M. R.; Christoffersen, J. J. Cryst. Growth 1984, 67, 107. (15) Margolis, H. C.; Moreno, E. C.; Murphy, B. J. J. Dent. Res. 1986, 65, 23.

J. Phys. Chem. B, Vol. 113, No. 51, 2009 16399 (16) Gobel, C.; Simon, P.; Buder, J.; Tlatlik, H.; Kniep, R. J. Mater. Chem. 2004, 14, 2225. (17) Zhang, H. G.; Zhu, Q. S.; Wang, Y. Chem. Mater. 2005, 17, 5824. (18) Iijima, M.; Moradian-Oldak, J. Biomaterials 2005, 26, 1595. (19) Chen, H. F.; Sun, K.; Tang, Z. Y.; Law, R. V.; Mansfield, J. F.; Czajka-Jakubowska, A.; Clarkson, B. H. Cryst. Growth Des. 2006, 6, 1504. (20) Jiang, H. D.; Liu, X. Y.; Lim, C. T.; Hsu, C. Y. Appl. Phys. Lett. 2005, 86, 163901. (21) Liu, X. Y. J. Chem. Phys. 1999, 111, 1628. (22) Liu, X. Y. Langmuir 2000, 16, 7337. (23) (a) Liu, X. Y. J. Chem. Phys. 2000, 112, 9949. (b) Liu, X. Y. In AdVances in Crystal Growth Research; Sato, K., Nakajima, K., Furukawa, Y., Eds.; Elsevier Science: Amsterdam, The Netherlands, 2001; pp 4261. (24) Liu, X. Y.; Lim, S. W. J. Am. Chem. Soc. 2003, 125, 888. (25) Oyane, A.; Kim, H.; Furuya, T.; Kokubo, T.; Miyazaki, T.; Nakamura, T. J. Biomed. Mater. Res. A 2003, 65A, 188. (26) Jiang, H. D.; Liu, X. Y. J. Biol. Chem. 2004, 279, 41286. (27) Jiang, H. D.; Liu, X. Y.; Zhang, G.; Li, Y. J. Biol. Chem. 2005, 280, 42061. (28) Liu, J. F.; Jiang, H. D.; Liu, X. Y. J. Phys. Chem. B 2006, 110, 9085. (29) Chernov, A. A. In Modern Crystallography III - Crystal Growth; Springer-Verlag: New York, 1984. (30) Zettlemoyer, A. C. In Nucleation; Marcel Dekker, Inc.: New York, 1969. (31) Harris, D. C. QuantitatiVe Chemical Analysis, 5th ed.; W. H. Freeman: New York, 1999; pp 175-190. (32) (a) Nielsen, A. E.; Christofferesen, J. In Biological Mineralization and Demineralization; Nancollas, G. H., Ed.; Springer-Verlag: Berlin, Germany,; pp 37-77; (b) Skripov, V. P. In Current Topics in Materials Science, 2nd ed.; Kaldis, E., Scheel, H. J., Eds.; North-Holland: Amsterdam, The Netherlands, 1977; pp 327-378. (33) Moreno, E. C.; Kresak, M.; Zahradnik, R. T. Nature 1974, 247, 64. (34) Liu, X. Y.; Strom, C. S. J. Chem. Phys. 2000, 113, 4408. (35) Wang, R. Y.; Liu, X. Y.; Xiong, J. Y.; Li, J. L. J. Phys. Chem. B 2006, 110, 7275. (36) Li, J. L.; Liu, X. Y.; Strom, C. S.; Xiong, J. Y. AdV. Mater. 2006, 18, 2574. (37) Li, J. L.; Liu, X. Y. Appl. Phys. Lett. 2005, 87, 113103. (38) Liu, X. Y. In Nanoscale Structure and Assembly at Solid-Fluid Interfaces; Liu, X. Y., Yoreo, J. J. D., Eds.; Springer: London, 2004; Vol. 1, pp109-175. (39) Liu, X. Y. In Low Molecular Mass Gelators: Design, Self-Assembly, Function; Frederic, F., Eds.; In Series Topics in Current Chemistry; Springer: Berlin, Germany, 2005; Chapter 1, pp 1-37. (40) Wang, R. Y.; Liu, X. Y.; Narayanan, J.; Xiong, J. Y.; Li, J. L. J. Phys. Chem. B 2006, 110, 25797.

JP905846P