Analytical grades Ca(NO₃)₂∙4H₂O, Mg(NO₃)₂∙6H₂O, (NH₄)₂HPO₄ and NH₄F were used as starting materials. Appropriate amounts according to the stoichiometric formulas of Ca₁₀(PO₄)₆(OH)F and Ca₉Mg(PO₄)₆(OH)_2-yF_y with y = 0, 0.5, 1, 1.5 and 2, were weighed, respectively and dissolved into 5 cm³ of deionised water under vigorous stirring. The pH of the mixed solution was adjusted to 9 by adding a concentrated ammonia solution. After that, the mixed solution was transferred to a Teflon vessel (model 4749 Parr Instrument) and sealed tightly. The autoclave was oven-heated at 180˚C for 6 h, and then cooled to room temperature naturally. The collected precipitates were washed with deionised water and dried at 70˚C overnight. After drying, the powders were calcined under argon flow at 500˚C for 1 h with a heating rate of 10˚C∙min⁻¹.

In the following sections, the compositions Ca₁₀(PO₄)₆(OH)F, Ca₉Mg(PO₄)₆(OH)₂, Ca₉Mg(PO₄)₆(OH)_1.5F_0.5, Ca₉Mg(PO₄)₆(OH)F, Ca₉Mg(PO₄)₆(OH)_0.5F_1.5 and Ca₉Mg(PO₄)₆F₂ will be named as HFA, MHA, MHF_0.5A, MHF₁A, MHF_1.5A and MFA, respectively.

The (Ca + Mg)/P molar ratios in the as-prepared powders were evaluated by a chemical analysis [23,24]. The fluoride content was measured using a fluoride selective electrode (Ingold, PF4-L).

The XRD patterns of the as-prepared and calcined powders were collected on a Philips X-pert diffractometer operating with Cu-Ka radiation (l = 1.5406 Ǻ) for a 2θ range from 20˚ to 55˚. The scan step was 0.02˚ and the integration time was 1 s per step. The crystalline phases were identified by comparing the experimental XRD patterns to the standards compiled by the Joint Committee on Powder Diffraction and Standards (JCPDS cards).

The ³¹P magic angle spinning nuclear magnetic resonance (³¹P MAS NMR) spectra were performed on a Brucker 300 WB spectrometer. The ³¹P observational frequency was 121.49 MHz with a spin speed 8 kHz. The ³¹P shift is given in parts per million (ppm) referenced to an aqueous solution of 85 wt% H₃PO₄.

The specific surface area (SSA) of the as-synthesized and calcined powders was measured with a Belsorp 28 SP apparatus using the BET method, while nitrogen was utilized as an adsorbed gas.

To carry out sintering experiments, the calcined powders were uniaxially pressed under 45 MPa into pellets in a 13 mm diameter steel die, then the pellets were sintered under argon flow in a temperature ranging from 950˚C to 1250˚C with 50˚C in interval for various times.

The sintered samples were characterized based on relative density, X-ray diffraction analysis and microstructural analysis using a scanning electron microscope (PHILIPS XL 30).

Both green and sintered densities (d_ex) of compacts were determined through dimension and weight, and the relative density was calculated using the formula:

The theoretical density for each composition (Ca₉Mg(PO₄)₆(OH)_2-yF_y) was calculated taking in account its molecular weight (W), the number of units per unit cell (1) and the volume of the unit cell, according to the following equation:

where A is Avogadro’s number, and a and c are the lattice parameters.

The mechanical properties were investigated on pellets of 13 mm in diameter sintered at different temperatures for 1 h. The sintered samples were polished to mirror finish prior to mechanical investigation using various grade silicon carbide papers (grade 800 - 1200) and a 0.2 mm diamond paste.

The hardness was checked with Vickers’ indentation technique using a Matsuzawa Seiki digital micro-hardness Tester (Japan). Five samples were used for each hardness data point, and for each sample, ten indentations were performed at an applied load of 200 g for 15 s. Thus, the reported hardness H_v is the average of the fifty values calculated according to the equation [25]:

where P is the indentation load and d is the length of the diagonal of the indentation.

The Elastic modulus (Young’s modulus) was estimated from an empirical relationship reported by Marshall et al. [26] using Vickers’ microhardness:

where H_v is Vickers’ indentation, a is the length of the shorter diagonal, b is the length of the longer diagonal determined using a Knoop’s indenter and b' is the crack length.

In Equation (4)

with

The comparison of Vickers and Knoop’ hardness for ceramics material reported by several authors [27,28] showed that the average value of the ratio H_K/H_V is 1.105.

Under these conditions, the value of a according to the Knoop indentation is determined according to the following equation:

where l is the length of Vickers’ indenter diagonal.

So, Equation (4) may be written as follows:

The fracture toughness (K_IC) was determined using the indentation technique, and following the relationship given below [26]:

where E is Young’s modulus; H_v, Vickers’ hardness; P, the applied load; c, the crack length indentation and a, the length of Vickers’ indenter diagonals.

The quantitative chemical analyses of the samples are listed in Table 1. As has been observed, the amount of Mg in the powders was close to those introduced in the solutions, indicating that all Mg was incorporated in the synthesized materials. The (Ca+Mg)/P molar ratios are very close to the theoretical value of 1.67 for the stoichiometric apatite. On the other hand, fluorine contents were not affected by the presence of Mg, and they are consistent with the nominal composition (Table 1).

The X-ray diffraction patterns of the as-prepared powders are shown in Figure 1. At the available resolution, there was no evidence of any crystalline phase other than the apatite, which is consistent with the JCPDS #01-074-0566 or #00-071-0880 file data for HA and FA, respectively (space group P6₃/m). As expected, the insertion of Mg into the apatite structure was accompanied by the decrease of both a and c parameters, compared to those of the original HFA (Table 2). This decrease of the lattice parameters, which is consistent with the radius of the Mg²⁺ ion (= 0.72 Å), that is smaller than that of Ca²⁺ (= 1.00 Ǻ) [29] confirms that the Mg²⁺ has entered the apatite structure. On the other hand, the substitution of F⁻ for OH⁻ can be also demonstrated by monitoring the variation of the lattice parameters as a function of the incorporated amount of fluoride into the apatite structure. As has been reported in the literature for hydroxyfluorapatite [30-33], with increasing content of F⁻, a decreased progressively and continuously compared to that of MHA, while c did not vary significantly (Table 2).

The ³¹P MAS NMR spectra of the as-prepared samples are shown in Figure 2. All the spectra exhibited a single resonance peak, which is a main feature of phosphorus in an apatite environment. For MHA, the chemical shift is of 2.62 ppm. This value is similar to that reported for this

Table 1. Chemical analysis data of as-prepared powders.

Table 2. Lattice parameters of nonand Mg/F co-substituted hydroxyapatites.

kind of compounds [34]. For the substituted samples, a slight chemical shift towards higher values occurred as the F⁻ content rose. In agreement with the XRD analysis, Figure 2 confirms that the powders consisted of a single apatite phase.

As can be seen from Table 3, with the increasing of the content of F⁻, the SSA increased up to y = 0.5 and then decreased. This variation of the SSA as a function of the fluoridation degree is similar to that observed for pure hydroxyfluorapatite [35,36]. The increase of the SSA when fluoride content increased is attributed to the difficulty of the grains to grow due to the interactions between hydroxyl and fluoride ions, which limited the kinetics of the crystallites’ growth [35,37]. It is worth noting that the presence of Mg into the apatite structure, as is well known, inhibits the crystal growth and thus contributes to the increase of the SSA of the substituted powders with respect to the pure powder [38].

After calcination at 500˚C, the XRD patterns of Mg/Fco-substituted hydroxyapatites showed only the apatite reflections, and neither a decomposition sign nor an appearance of a new crystalline phase, resulting from the crystallization of an amorphous phase in the as-prepared powders was detected by XRD (Figure 3). However, the pattern of MHA exhibited reflexions other than those of the apatite. These reflexions whose intensities were very low belong to the b-Mg-substituted tricalcium phosphate (Ca_2.81Mg_0.19(PO₄)₂, b-MTCP) (JCPDS #01-070-0682). As the XRD and ³¹P MAS NMR analyses had shown that

Table 3. Specific surface area of the powders.

the as-prepared powder was single-phased, b-MTCP would result from the decomposition of MHA. Thus, these findings show the destabilizing effect of Mg on the hydroxyapatite, and the stabilizing role of the fluorine on the Mg-substituted hydroxyapatite.

The values obtained for the SSA of the powders calcined at 500˚C are summarized in Table 3. As expected, the SSA was significantly reduced with respect to that of the as-prepared powders. It decreased from 49 to 12.7 m²∙g⁻¹ and from 51.7 to 11.6 m²∙g⁻¹, for MHA and MFA, respectively. Furthermore, The SSA followed versus the fluorine degree the same evolution as that of the as-prepared powder, the highest SSA value was also observed for y = 0.5.

Figure 4 illustrates the relative density of the samples with different fluoridation degrees sintered at various temperatures. As shown in Figure 4(a), two composition ranges can be distinguished. For MHA and MHF_0.5A, the relative density continuously increased as the sintering temperature increased. However, the density of MHF_0.5A between 1000˚C and 1200˚C was much lower than that of MHA. For MHF₁A, MHF_1.5A and MFA, the relative density rose with the increase in temperature, reached a

T ˚C

(a)

0.0        0.5       1.0       1.5       2.0 y(F)

(b)

Figure 4. Relative density of the sintered samples as a function of: (a) sintering temperatures and (b) fluorine contents.

maximum of about 96% at around 1050˚C - 1100˚C, and then decreased. For MFA, the maximum density was observed at 1050˚C, while for the two other samples it was attained at 1100˚C, indicating that MFA densified better than the partially fluoridate samples, and obviously hydroxyapatite. Above 1200˚C, the sintered samples of the latter group were distorted, making impossible the determination of their density. The difference in the curve shape of the densities suggests that the mechanisms responsible for the densification are different for the two groups of composition.

According to Figure 4(b), showing the relative density against the fluoride content, two kinds of curves can be also distinguished. At 1200˚C, the relative density, whose highest value (0.94) was observed for MHA, decreased continuously and uniformly with the fluoridation degree to 0.82 for MFA. Below 1200˚C, and apart from a temperature reaching 950˚C, all the curves have roughly the same shape with a large decrease in density for y = 0.5. Such a shape of the curves has been previously observed for the sintering of the hydroxyfluorapatites [37]. However, Senamaud observed the decrease in density for y = 1 [35], while for Gross [37], the decrease occurred for y = 1.2. This low sinterability was attributed to the interactions between OH⁻ and F⁻, which reduce the species mobility [35,37]. If, for the samples with y = 0 and 0.5, the relative density increased with the increasing of temperature, it did not follow the temperatures’ increase for y > 0.5. For example, the density of the y =1 sample was higher at 1100˚C than at 1150˚C. Similarly, for the y = 2 sample, the density was higher at 1050˚C than at 1100˚C or 1150˚C.

The effect of time on the relative density of the sintered samples was examined by fixing the temperature at 1050˚C (Figure 5). This figure highlights the low sinterability of MHF_0.5A, and the high densification of the other compositions. Indeed, the samples with y = 1, 1.5 and 2 sintered to a relative density of ~0.97 after a heat treatment of 0.5 h. This density was slightly higher than that of MHA. We note that, for the latter compositions, the effect of the fluorine substitution on the densification of MHA was not noticeable at 1050˚C since there was no significant difference between the sintered densities of the fluoridate compositions. However, the y = 0.5 sample sintered only to 66.6%, and 75.4% of the theoretical density after 0.5 and 1 h, respectively. After that, the density decreased for 2 h and increased again to reach its maximum (81%) for 4 h.

Figure 6 shows the XRD patterns of the samples sintered at 1050˚C. The patterns of MHA and MHF_0.5A showed the presence of b-MTCP, indicating the decomposition of MHA, as has been reported in the literature [19,20,39,40]. This phase remained stable up to 1300˚C [18]. Indeed, it is well-known that the Mg substitution