Samples of CCaApOH were synthesized from readily soluble sources of calcium, phosphate, and carbonate in a pH-buffered aqueous solution. All water used in the synthesis was double-deionized (Milli-Q) or deionzied and passed through a hydroxylapatite column. Nitrogen gas was bubbled through the water for ca. six hours to remove carbon dioxide. The NaOH solutions were freshly prepared after each 5–6 syntheses, and the solution bottles were sealed with Parafilm to prevent contamination by carbon dioxide. Magnetic stirrers were used for mixing.

Approximately 250 mL of water was heated to the desired temperature (typically 60 °C or 80 °C) in a large beaker. According to the desired final carbonate to phosphate ratio, a specific amount of NaHCO₃ (Acros-Organics, 99.5%) was dissolved in the water. The pH was elevated to and maintained at 9.0 ± 0.3 (Hanna Instruments pH 209 meter with a Thermo Ross Sure-Flow combination reference/pH electrode) through addition of 0.1M NaOH (Acros, 97%). To the heated carbonate solution, 25 mL of a 0.15M Ca(NO₃)₂·4H₂O (99%, Sigma-Aldrich, St. Louis, MO, USA) solution and 25 mL of a 0.09 M NaH₂PO₄·H₂O (Aldrich, 98%) solution were added at a rate of 1 mL/min from separate burettes. After addition of the calcium and phosphate solutions, the mixture was stirred and kept at 60 °C (some syntheses were also performed at 80 °C) for 2 hours (some for 1 h). After this equilibration time, stirring was discontinued, and the mixture was allowed to cool to room temperature and then was filtered through a glass filter crucible of medium porosity. The product powder was washed three times with warm water, and then dried at 120 °C in a vacuum pistol or dried in an evacuated vacuum dessicator for a minimum of 24 h.

The synthesis of CCaApF followed the same procedure used for CCaApOH, except that a 0.6 M NaF solution was added simultaneously with and at the same rate as the addition of the calcium and phosphate solutions.

Carbonate analysis. The weight percent carbonate in each sample was determined via C, H, N combustion elemental analysis or from the weight loss between 600 °C and 800 °C in TGA analyses. A standard curve for TGA weight loss converted to carbonate wt% was constructed using the results of combustion analyses performed by Schwarzkopf Microanalytical Laboratories (Woodside, NY, USA) or Galbraith Laboratories (Knoxville, TN, USA).

X-ray diffraction. To confirm the identity of the precipitate and to reveal any contaminating phases (particularly CaCO₃), samples were subjected to X-ray diffraction analysis with either a Phillips 3520 X-ray diffractometer producing Cu-Kα radiation in conjunction with a monochromator or a PANalytical X’Pert PRO MPD (Multi-Purpose Diffractometer) Theta-Theta System with Cu-Kα radiation (λ = 1.54060 Å). Instrumental settings included a step size of 0.02° 2-theta and a dwell time of 1 s over the analysis range of 2 to 60° 2-theta. The scans were analyzed using the program PANalytical X’Pert HighScore Plus (Version 2.2e). Sample diffraction patterns were compared with the ICDD powder X-ray diffraction pattern database (PDF Release 2) using the HighScore Plus search-match algorithm. X-ray diffractograms were also used to detect other phases present in the synthetic apatites, particularly CaCO₃, and to look for patterns of changing crystallinity of the samples.

Cell parameters were determined using the PANalytical X’Pert PRO MPD diffractometer with samples mounted in 16 mm cavity slides with a step size of 0.02 2-theta over the range 15 to 70° 2-theta. The entire XRD pattern was profile-fitted by determining the profile for each individual peak and ultimately fitting the entire scan. The indexing method used was usually McMaille, but in some circumstances the Treor method was employed. The unit cell candidate was chosen by its high figure of merit relative to other candidates, its symmetry, and the absence of un-indexed peaks. Although the uncertainties in the cell parameters were generally within ±0.0007 Å for a single measurement, repeated measurements in some cases produced larger errors, more of the order of ±0.002 Å.

Raman microprobe analysis. Raman spectroscopy was used to further characterize the phosphate phase in the precipitate and to reveal the presence of contaminating phases more sensitively than by powder X-ray diffraction. The integrated, fiber-optically coupled microscope-spectrometer-detector confocal Raman system used in this study is from Kaiser Optical (Ann Arbor, Michigan). The HoloLab Series Research Raman Spectrometer was configured for 532-nm laser excitation, and it covered the spectral range of 0 to 4300 Δcm⁻¹ with a resolution of 2.5 cm⁻¹ using an Andor high-resolution, thermoelectrically cooled CCD array detector. Analyses of powders were made using an Olympus 80× objective with an N.A. of 0.75 or an Olympus 50x objective with an N.A. of 0.55. Each Raman analysis represents the average of 32 acquisitions of 4 seconds each. Every sample underwent 3 to 5 analyses. The laser power was 10 mW at the sample surface, and the diameter of the beam was one to several micrometers at the sample surface. The focus was optimized so as to obtain the maximum signal:noise ratio.

The program GRAMS 32 (Galactic, Salem, NH, USA) was used to deconvolve appropriate Raman peaks into their component spectral bands. The bands were modeled as mixed Gaussian and Lorentzian in form, in order to derive the wavenumber position, width at half-maximum intensity, and area of the bands.

The amount of structural and adsorbed water in the samples was determined by thermogravimetric analysis (TGA) using the TA Instruments TGA Q500. Unless otherwise stated, thermograms were obtained using a balance flow of 40 mL/min and a sample flow of 60 mL/min of dry nitrogen gas. Samples of 7–20 mg were heated to 1,000 °C at a rate of 15 °C/min. The relative uncertainty for the measurement of weight loss in multiple scans is estimated to be about 8%. Linear regression lines are shown for most plots involving weight loss, but correlation coefficients are not given because of their generally low values and limited utility.

Prior to rehydration experiments, the purity of the phase was checked by XRD, and after vacuum drying for 24 h, TGA was used to measure the amount of water in the apatite. The weight percent lost between room temperature and 200 °C was taken to be adsorbed water, while the weight percent lost between 200 °C and 550 °C indicated the loss of structural water.

For each rehydration experiment four 50 mg samples of an apatite were dehydrated by heating in four ceramic crucibles in a muffle furnace for 2 h at 550 °C. These samples were then either immersed in water for 2 h at room temperature (22 °C), immersed in boiling water for 2 h, immersed in water for 4 days at room temperature, or immersed in water for 4 days at 100 °C. The rehydrations were performed in 100 mL round-bottom flasks with magnetic stirring using double-deionized (Milli-Q) water previously saturated with nitrogen gas. The samples were filtered by suction with Buchner funnels and dried in vacuo (4 torr) for at least 24 hours. The dried samples were analyzed by TGA and XRD.

Raman spectroscopic analysis confirmed that the only phosphate phase present is CCaApOH or CCaApF, as appropriate to the given experiment. No other carbonate-bearing phase was detected; for instance, no calcite (CaCO₃) was found. The overwhelming dominance of B-type substitution of carbonate for phosphate was confirmed by the detection of only the peak at ~1,073 Δcm⁻¹. No peak was detected at ~1,108 Δcm⁻¹, as would be expected for A-type substitution of carbonate for the monovalent anion [20]. The exact detection limits for the latter spectral peak are not known. The presence of 10–15% of the total carbonate in A-type substitution, as typically reported for low-temperature (including biological) apatite [19], cannot be ruled out. The higher the carbonate concentration of the precipitated apatite, the broader the ν₁ symmetric P-O stretch band at about 960 Δcm⁻¹, indicating that the internal atomic ordering of the phase decreased as carbonate concentration increased. The intensity of the O-H stretching band for hydroxyl at ~3,570 Δcm⁻¹ decreased as carbonate concentration increased as expected, based on the dominant B-type substitution as discussed below.

The presence of the correct apatite phase and the absence of CaCO₃ at greater than the detection limit of 0.3% was confirmed by powder X-ray diffraction. Cell parameters were determined for each apatite with a different carbonate content.

The decrease in the a-axis value of approximately 0.1 Å over a carbonate concentration range of 14% and the increase in the c-axis of 0.06 Å over the same range strongly suggest that carbonate has substituted for phosphate (i.e., B-type substitution) in our hydroxylapatites [6,17,18,21]. For the fluorapatites the a-axis decreased by about 0.09 Å over a carbonate concentration range of 12%, while the c-axis increased by about 0.03 Å [22]. The maximum observed substitution value of ca. 16 wt.% carbonate in our syntheses is also consistent with B-type substitution; A-type substitution can accommodate a maximum of only 6 wt.% carbonate [21]. Based on both Raman and X-ray diffraction data, we have adopted the B-site formula Ca_10−x(PO₄)_6−x(CO₃)_x(X)_2−x, where X = F or OH, as an appropriate stoichiometric formulation of our apatites.

We have used thermogravimetric analysis to determine the weight percent structural water as a function of the carbonate composition in both the hydroxyl and fluor series of apatites. Weight loss from 200 °C to 550 °C was selected as the temperature range most indicative of water tightly constrained within the apatite lattice. LeGeros et al. [5] used the range of 200 °C to 400 °C, while Ivanova et al. [6] chose the range of 150 °C to 500 °C for the loss of structural water. The release of water from room temperature to 200 °C can be attributed to adsorbed or surface water, although most studies use a value closer to 150 °C as the upper limit of adsorbed water [6]. Elliot’s 1994 summary [18] of previous thermal decomposition studies demonstrated release of lattice water from 120 °C to 400–500 °C. Holcomb and Young have suggested that structural water continues to be released above 550 °C [23]. Füredi-Milhofer et al. identified the release of water by mass spectrometry in two major intervals, one centered at 88 °C, the other at 225 °C [24]. Because carbonate decomposition begins in the 600 °C to 650 °C region, 550 °C is a pragmatic upper limit for the identification of structural water by TGA. In fact, our TGA scans show that most of the weight loss between 200 °C and 550 °C occurs below 400 °C.

Figure 1 displays the weight percent structural water in 20 hydroxylapatites prepared using a two-hour digestion time. The weight percent values are clustered about three, consistent with values found by LeGeros et al. [5] for both enamel and synthetic CCaApOH, and are not dependent on the carbonate concentration of the apatite. Weight losses between 100 °C and 450 °C of about 3 wt.% for synthetic non-stoichiometric hydroxylapatites have also been observed by Reyes-Gasga et al. [25].

The number of water molecules per unit cell (mpuc) was obtained by assuming that carbonate substitutes for phosphate and that this substitution is accompanied by vacancies in both calcium and hydroxyl sites, as depicted in the B-type formula Ca_10−x(PO₄)_6−x(CO₃)_x(OH)_2−x. Conversion of weight percent carbonate, as obtained from either TGA or combustion analysis, to a B-type formula, followed by use of weight percent structural water to produce a hydrate formula Ca_10−x(PO₄)_6−x(CO₃)_x(OH)_2−x(H₂O)_y provides the number of moles (molecules) of structural water per formula unit (cell). Three interesting conclusions can be derived from Figure 1. The first is that there is no obvious dependence of the structural water content on the carbonate content of the apatite. The second is that the number of molecules per unit cell shows a greater regularity than does the weight percent water. The third conclusion is that the average mpuc is 1.5 for the hydroxylapatites.

Weight percent adsorbed water as determined by weight loss from room temperature to 200 °C is shown in Figure 2, from which it is clear that there is considerably more scatter than that observed for structural water.

The greater scatter is not surprising given the many variables that can affect the surface area of the apatite crystallites and consequently their ability to adsorb water. These variables include: (a) method of synthesis, generally the more rapid the rate of addition of reagents in a precipitation reaction the smaller the particle size; (b) digestion time and temperature, longer times and higher temperatures giving rise to more mature, larger particles; (c) method and duration of drying, including mechanical details such as the amount of sample dried and the extent of exposed surface area.

The effect of digestion time on adsorbed water is illustrated by the weight loss for compounds prepared with a one-hour digestion (rather than the two-hour digestion period shown in Figure 2) in Figure 3. For one-hour digestion the scatter in weight percent of adsorbed water is greater than for a two-hour period, and an upward trajectory with increasing carbonate concentration is noticeable.

For the fluorapatites, made by a two-hour digestion process, the weight percent structural water and the adsorbed water are presented in Figure 4 and Figure 5, respectively. The adsorbed water relationship (Figure 5) shows less scatter than that of the hydroxyl analog (Figure 3), and also shows a more distinct correlation with weight percent carbonate. By comparison with the hydroxylapatites, the effect of carbonate in restricting crystallite size [26] is more obvious in the synthesis using the two-hour digestion period for the fluorapatites. It is important to notice that the average mpuc of 1.6 is essentially the same as that for the hydroxyl analogs.

We also investigated the ease with which structural water can be replaced in the apatite lattice after dehydration. These experiments consisted of the dehydration of small amounts of apatite samples with different carbonate compositions in a muffle furnace at 550 °C for two hours, followed by immersion (with magnetic stirring) in: (a) water for 2 hours at 22 °C, (b) water for 4 days at 22 °C, (c) water at 100 °C for 2 hours, and (d) water at 100 °C for 4 days. Samples were filtered by suction, dried in vacuo for 24 h and then analyzed by XRD and TGA. Figure 6 shows the TGA plots of a hydroxylapatite containing 13.4 wt.% carbonate that had been rehydrated under different conditions. The difference in the weight loss at the beginning and end of the 200 °C to 550 °C interval (essentially the slope of the line in this area) indicates that the effectiveness of rehydration is a stronger function of rehydration temperature than duration. Thus, scans E and F of samples rehydrated at 100 °C show considerably more reincorporation of water than scans C and D on samples rehydrated at 22 °C.

The same is true for the rehydration of the CCaApF sample shown in Figure 7. The “double drop” seen in scan F after the rehydration experiment at 100 °C for four days appears in several of the TGA scans obtained under these conditions for both fluor- and hydroxylapatites and suggests the possible release of water molecules at two distinct temperatures.

A total of six apatites—three hydroxyl- and three fluorapatites—with different carbonate concentrations were subjected to these rehydration experiments. The results show no discernible relationship between the extent of rehydration of the apatites and their concentration of carbonate. Consequently, we have summarized the effect of rehydration temperature and time by averaging the wt% water released in each of the three carbonated CCaApOH samples as well as the wt% water released in each of the three CCaApF samples. As shown in Figure 8 and Figure 9, on average, samples dehydrated at 550 °C are not completely rehydrated even after four days or immersion at 100 °C. Additionally in these experiments, the temperature of rehydration seems to be a more important factor than the duration of the rehydration.

The XRD patterns of CCaApOH and CCaApF samples are shown in Figure 10 and Figure 11, respectively. For the CCaApOH sample there is a considerable decrease in full width at half maximum (FWHM) after dehydration; the decrease in full width at half maximum for CCaApF is much less. For both apatites, there is little change in peak positions after dehydration and rehydration, and the rehydrated samples have FWHM values similar to those of the dehydrated samples.

Dehydration at 300 °C did not have a significant effect on the XRD peak widths of three CCaApF samples, and rehydration of these samples for 2 h at 22 °C resulted in peak widths about the same as those of the original and dehydrated samples. The same behavior was observed for the one CCaApOH sample examined. Thus, although 300 °C is sufficient to remove a significant amount of water from the lattice, the temperature may not be high enough to enlarge the crystallites. Consistent XRD peak positions document that removal of water from the apatite lattice does not change the structure of the lattice, whereas peak narrowing indicates an increase in crystallite size. Because our method of water removal involves heating, we cannot separate the effects of temperature from those of water loss as causes for the increase in crystallite size. Interestingly, replacement of at least some of the water after dehydration produces crystallites with about the same (or in same cases larger) size.

Subtle changes in the structure of the apatites were monitored by measuring unit cell parameters from a full pattern analysis of the XRD patterns. The bar graphs in Figure 12 and Figure 13 portray the a- and c-axis lengths as a function of the treatment of the precipitated CCaApX. In most cases, the cell parameters of the treated samples are within about 0.02 Å of those of the originally precipitated apatite. Note, however, the larger changes in the a-axis after dehydration of the CCaApF samples. In all cases, dehydration produced an increase in the a-axis and a decrease in the c-axis. Upon dehydration our samples showed a larger change in the a-axis for the fluoraptites, but a larger change in the c-axis for the hydroxylapatites.