Varying Synthesis Conditions and Comprehensive Characterization of Fluorine-Doped Hydroxyapatite Nanocrystals in a Simulated Body Fluid

: Bone supports animal bodies, is the place where blood is produced, and is essential for the immune system, among other important functions. The dominant inorganic component in bone is hydroxyapatite (Hap), the structure and dynamics of which still pose many unsolved puzzles. An updated understanding of HAp is of great signiﬁcance to osteology, dentistry, and the development of artiﬁcial bone and other biomaterials. In this work, HAp nanoparticles were synthesized with the wet chemical precipitation method and their structure and morphologies were controlled by varying pH and adding ﬂuoride ions by two different routes: (1) ﬂuoride ions were added during synthesis, and (2) ﬂuoride ions were introduced after the samples were synthesized by soaking the samples in solutions with ﬂuoride ions. XRD and HRTEM were employed to conﬁrm the composition and structure, while various multinuclear ( 1 H, 19 F, 31 P) solid-state nuclear magnetic resonance (NMR) methods including 1D single pulse, cross-polarization under magic-angle spinning (CPMAS), and 2D heteronuclear correlation (HETCOR) were used to characterize the structure, morphology, and dynamics, validating the general core-shell morphology in these F-HAp samples. It was found that all hydroxide ions were substituted when the ﬂuoride ion concentration was above 0.005 M. An NMR peak corresponding to water structure emerged and the bulk water peak was shifted upﬁeld, indicating that ﬂuoride substitution modiﬁes both the crystalline core and the amorphous shell of F-HAp nanoparticles. With the second route of ﬂuoride substitution, increases in soaking time or ﬂuoride ion concentration could increase ﬂuoride substitution in HAp, but could not achieve complete substitution. Finally, with 1 H- 31 P CPMAS and HETCOR, it was established that there are two types of phosphorous, one in the crystalline core (PO 43 − ) and the other in the amorphous shell (HPO 42 − ). These results are valuable for clarifying the ﬂuoride substitution mechanism in HAp in biomaterials or in organisms, and provide insights for developing next generation replacement materials for bone, tooth, or coating ﬁlms, drug delivery systems, etc.


Introduction
An obvious and essential feature of vertebrate animals is that their bodies contains various types of bone. Their bodies are supported by a sophisticated skeletal system that consists of hundreds of bones of differing sizes, shapes, and biophysical, biochemical, and mechanical properties. Bones are also places where blood is made and some bones such as vertebrae also provide neurological signal pathways for the central nerve system. The importance of bones for vertebrates can never be overemphasized. Similar to other organs, bones may be subject to many diseases and injuries, which are becoming increasingly worse with the growth of the aged population. Research on bones has a long history since ancient times and a huge body of data has been accumulated. The most remarkable achievements are those made in the past century. Significant advances have been achieved in understanding their anatomic structure [1][2][3], physiological functions, and biochemical mechanisms of synthesis, repair, and aging [4,5]. These achievements have revolutionized the medical treatment of various disorders, diseases, and injuries related to bones [4][5][6][7]. Tremendous progress has also been made in the development of artificial bones [8,9] and bone regeneration [10]. Based on our current understanding [11][12][13][14], the main components of bone have been identified as a protein collagen (mostly of type I) and an inorganic compound hydroxyapatite (HAp). The growth and maintenance of a bone are highly complex and involve many cells, proteins, hormones, ions, and, of course, water. Blood vessels and lymphatic systems are also an active participant.
Biophysical chemists and materials scientists are interested in bone for a number of reasons. Firstly, it is a system responsible for a variety of functions from mechanical support, anti-wear, anti-tear, and hormone and ion balancing, to blood-making and neuro signaling. Secondly, it is a good example of a hierarchical multiscale complex system ranging from the nanometer to meter scales, providing good opportunities for demonstrating all kinds of interesting physical chemistry phenomena. Lastly, but not least, it is a typical organicinorganic hybrid material, so bone studies can offer valuable insights for the understanding of organic-inorganic hybrid materials and the development of better biomimetic, bionic, or biogenic materials. Organic-inorganic hybrid materials have been an important subject of research over the past decades and play a more and more important role in photonics, gas sensing, biomedical detection, replacement biomaterials, etc. [15][16][17][18][19][20][21]. They are also very common in living things [22][23][24][25][26][27][28]. To build a good understanding of certain biological functions, pathological mechanisms, or the effect of a drug or a therapy, or to develop next generation materials, a better understanding of the physicochemical principles underlying this type of material is indispensable.
Hydroxyapatite nanocrystals play an important role in the bone, teeth, joints, shells, and other organs of animals and belong to a class of representative systema that inspires the study and development of biogenic, biomimetic, and bionic materials, and new organicinorganic hybrid materials [22,23]. In the hierarchical structure, mineralized collagen is believed to be the most important building block and has been intensively studied, although its mass accounts for only 20 wt% to 30 wt%, while the mineral Hap, which accounts for 60 wt% to 70 wt% of the total mass, plays an auxiliary, though essential, role [11][12][13][14]. As the dominant inorganic component, the structure, morphology and dynamics of HAp still hold many unsolved puzzles. An updated understanding of HAp is of great significance to osteology, dentistry, and the development of artificial bone and other biomaterials.
To gain a better understanding of HAp in real animal bones and for the development of next generation HAp-based ceramic materials, as well as bionic materials for bone/tooth replacement, drug delivery, anti-microbial use, and other purposes, it is pertinent to study the structure and morphology of HAp synthesized in conditions similar to the physiological environment (ions and small and large molecules). Therefore, in this work, our HAp samples are synthesized in a simulated body fluid (SBF) [29][30][31][32][33] to mimic the in vivo physiological conditions. Fluoride ions are one of the most common ions in bone, tooth, and other tissues, and SBF used in this work contains all the important ions in real blood and body fluids. The motivation of this work can be summarized as follows: (1) to explore the changes of structure and morphology of HAp synthesized in SBF by varying pH slightly and comparing the outcome with previously published results; (2) to investigate the structure and morphology of fluorine-substituted HAp synthesized with wet precipitation method in SBF; (3) to characterize the structure and morphology of the HAp systems with XRD, HRTEM, and most importantly, solid-state NMR, focusing on the core-shell structure. To achieve these goals, HAp nanoparticles were synthesized with the wet chemical precipitation method and their structure and morphologies were controlled by varying pH and adding fluoride ions by two different routes: (1) fluoride ions were added during synthesis, and (2) fluoride ions were introduced after the samples are synthesized by soaking the samples in solutions with fluoride ions. XRD and HRTEM were employed to confirm the composition and structure, while various solid-state NMR methods including 1D single pulse under high speed -MAS-, 1 H-31 P CPMAS, 2D 1 H-31 P, 19 F-31 P HETCOR were used to characterize the structure, morphology, and dynamics, validating the general coreshell morphology in these F-HAp samples. These results are well supported by each other and are valuable for clarifying the fluoride substitution mechanism in HAp in biomaterials or in organisms, and provide insights for developing next generation replacement materials for bone, tooth, etc.
For pH measurement, a G&B Instruments (Lymington, UK) PL-700 multifunction pH meter for laboratory use was used; buffer solution with pH = 7.00, pH = 4.01, pH = 10.00 produced by Rocker was selected as the standard calibration solution of the pH meter; the pH values of all samples were calibrated prior to measurement with a calibration percentage higher than 90%.

Preparation of Simulation Body Fluid
The SBF used in this work was prepared using the formula published by Koboku [29][30][31][32][33]. Each batch was 1000 mL and all ingredients were accurately weighed to the third decimal place. The procedure of preparation is summarized as follows: (1) Clean a 1000 mL plastic container and fill it with 900 mL deionized water to control the water temperature to 36.5 ± 1.5 • C; (2) Add the following ingredients in order: NaCl 8.035 g, NaHCO 3 0.355 g, KCl 0.225 g, K 2 HPO 4 ·3H 2 O 0.231 g, MgCl 2 ·6H 2 O 0.331 g, 1.0 M HCl 39 mL, Na 2 SO 4 0.072 g; (3) Slowly add 0.292 g of CaCl 2 to ensure that all of it is dissolved without precipitation; (4) Verify that the pH value of the solution, which should be 2.0 ± 1.0 at this point; (5) Slowly add Tris (trishydroxymethylaminomethane), and continue to add it after the pH has risen and stabilized. Do not change the pH too much; (6) Continue until 6.118 g Tris is added to make the pH of the solution reach 7.45 ± 0.01; (7) Use a syringe to suck out 1.0 M HCl, and slowly drop it so that the final pH value reaches 7.40 ± 0.01; (8) Rinse the remaining solution on the pH meter into the solution; (9) Fill up to 1000 mL of water; (10) Seal and store it in the refrigerator below 10 • C. The prepared SBF solution must be used within one month to avoid deterioration and compromised results. Prior to use, it must be heated to 37.5 • C in water.

Synthesis of Fluorinated HAp
In this work, the chemical precipitation method of wet synthesis was used as the synthesis route for calcium hydroxyphosphate. In order to avoid the precipitation reaction of SBF solution in glassware, all the containers used in the work were high-density polyethylene containers.

Synthesis of General Calcium Hydroxyphosphate
(1) Weigh 2.5 g of Na 2 HPO 4 and 7 g of Ca(NO 3 ) 2 , with calcium ions controlled as the limiting reagent. Add each to a beaker of 350 mL containing SBF solution to complete the preparation. Keep them at temperature at 37 • C and stir at 400 rpm for 1 h to ensure that the solutions are completely dissolved; (2) Mix the two beakers of solutions together, stir the mixture for 10 min, take out the magnet stirrer, seal it with paraffin film, and let it settle for 24 h. The temperature of the solution is kept at 37 • C throughout the process; (3) After 24 h, open the sealing film and verify that the precipitation is complete; then carefully pour out the supernatant liquid, add 250 mL of deionized water, put in a magnetic stirrer and rotate it at 400 rpm for 10 min, then take the stirrer out and let it stand for 1 h until it settles again; (4) Repeat Step 3 several times. After confirming that it is clean, pour out the supernatant liquid by decantation method, then place it in an oven and bake it at 60 • C for 24 h so that the water evaporates completely and the final sample of calcium hydroxyphosphate is obtained.

Synthesis of Calcium Hydroxyl Phosphate with Changing pH
To prepare SBF with different pH values, in Step 5 of Section 2.1.1, Tris was added slowly to achieve pHs of 7.0 and 7.2 to obtain SBF with pH = 7.0 and pH = 7.2, respectively; then calcium hydroxyphosphate with different pH values was synthesized according to the method described in Section 2.1.2.1.

Synthesis of Calcium Hydroxyphosphate Substituted with Fluoride Ion
After the SBF with pH = 7.4 was prepared, 250 mL of SBF was taken out and 0.02625 g, 0.0525 g, and 0.105 g of sodium fluoride were added to make the SBF bionic solution so that solutions with 0.0025 M, 0.005 M, 0.01 M fluoride ion concentrations were prepared, respectively. Then, following the method of Section 2.1.2.1, fluoride ion-added calcium hydroxyphosphate samples were synthesized and numbered as SBF + 0.0025 M, SBF + 0.005 M, and SBF + 0.01 M, respectively.

Synthesis of Hydroxy Calcium Phosphate Soaked in Fluoride Ion Solution
Hydroxycalcium phosphate was synthesized with pH = 7.4 according to the method described in Section 2.1.2.1, ground it into fine powder, and prepared in 0.0025 M, 0.005 M, 0.01 M NaF aqueous solutions. Then, 1 g of calcium hydroxyphosphate powder was put into each of the NaF solutions of different concentrations and soaked for 1 day, 3 days, and 7 days, respectively. Then, the samples were taken out, cleaned, and dried according to

Instrumental Characterization: XRD, HRTEM and Solid-State NMR
The XRD patterns were obtained on a powder diffractometer Bruker D2-Phaser using a copper target; the characteristic X-ray wavelength formed was 0.154056 nm (Cu-Kα = 0.154056); the 2θ scanning range was 20 •~8 0 • ; the κ (shape factor) of the calcium hydroxyphosphate sample in the Scherrer Equation was set to about 0.9; the JCPDS phase identification of Ca 2+ , PO 4 3− , and F − was performed following general diffraction analysis method with MDI JADE.
High resolution TEM images for the samples were obtained on a JEOL TEM-3010 Analytical Scanning Transmission Electron Microscope with a 250 Mesh copper mesh dedicated to TEM and a high-energy electron beam acceleration voltage of 300 kV.
Various 1D and 2D solid-state NMR spectra were obtained on a Varian-Agilent Inova 500 MHz spectrometer and a Bruker AVANCE II 600 MHz solid-state NMR spectrometer. For 19 F, 31 P and 1 H single pulse MAS spectra, 90 • pulse and 35 kHz sample spinning speed was used. For 1 H-31 P CPMAS spectra, 10 kHz sample spinning speed and contact times of 0.2 ms, 0.5 ms, 2 ms, 5 ms, and 10 ms were used. For 1 H-31 P HETCOR spectra, the sample spinning speed was 35 kHz; the frequency-switched Lee-Goldberg (FSLG) pulse sequence was used to remove 1 H-31 P dipolar interaction during the t1 evolution period; the contact time for 1 H-31 P CP was 1 ms.

Results and Discussion
The characterization results with XRD, HRTEM, and different solid-state NMR methods are presented as follows and their significance from the perspective of biogenic or biomimetic materials, as well as fluoration, will be discussed. Figure 1 shows the powder XRD spectra of 12 HAp and F-HAp samples synthesized at different pH values, at different NaF concentrations, or soaked in different NaF solutions for different numbers of days. The influence of the pH value is minor, although the crystallinity increases slightly at higher pH values, in agreement with previous observations [34][35][36][37]. Compared to HAp nanoparticles synthesized in ordinary solvents, HAP nanoparticles synthesized in SBF show less crystallinity, implying that HAp nanoparticles grown in vivo may have less crystallinity. This is supported by studies of animal bones [38][39][40][41]. It is interesting to notice that in F-HAp synthesized in SBF and NaF, the crystallinity increased. This is consistent with the previous observation of F-HAp synthesized in ordinary solutions and the fact that fluoride substitution of hydroxy reduces solubility and improves stability [42]. Furthermore, as shown at the bottom of Figure 1, simply soaking HAp (synthesized in SBF) in NaF can improve crystallinity, although it is far less effective than HAp synthesized in NaF. The higher the NaF concentration and the longer the soaking time, the better the crystallinity. However, after about three days of soaking, there is little room for further improvement. These observations are supported by other methods, as shown below.

HRTEM
To see how synthesis conditions affect the nanoscale structure and morphology of F-HAp nanoparticles, HRTEM was employed to generate images from the atomic scale to about 100 nm. Figure 2 shows the representative images of F-HAp nanoparticles synthesized in SBF at different NaF concentrations (2.5 mM, 5 mM and 10 mM). They share some common features: (1) all have the general core-shell morphology with a crystalline core and largely amorphous shell regions and (2) the crystalline region for all samples is rod-like. It was found that the crystallinity was improved by increasing the pH value ( Figure S1) (see Supplementary Materials), by increasing the NaF concentration ( Figure S2), or by increasing the soaking time ( Figure S3). These results are consistent with the XRD results above and the solid-state NMR results in the following section.

HRTEM
To see how synthesis conditions affect the nanoscale structure and morphology of F-HAp nanoparticles, HRTEM was employed to generate images from the atomic scale to about 100 nm. Figure 2 shows the representative images of F-HAp nanoparticles synthesized in SBF at different NaF concentrations (2.5 mM, 5 mM and 10 mM). They share some common features: (1) all have the general core-shell morphology with a crystalline core and largely amorphous shell regions and (2) the crystalline region for all samples is rodlike. It was found that the crystallinity was improved by increasing the pH value ( Figure  S1) (see Supplementary Materials), by increasing the NaF concentration ( Figure S2), or by increasing the soaking time ( Figure S3). These results are consistent with the XRD results above and the solid-state NMR results in the following section.

Solid-State NMR
To further investigate how synthesis conditions affect HAp and in particular to see straightforward evidence of fluoride substitution in F-HAp, both 1D and 2D multinuclear ( 1 H, 19 F, 31 P) solid-state NMR spectra were acquired. Figure 3 is the 1 H 35 kHz MAS spectra of various HAp and F-HAP samples. Based on the assignment of 1 H spectrum [43], the hydroxyl group corresponds to a peak at around 0.00 ppm; peaks between~0.5-1.5 ppm are from surface-structured water; peaks above 2 ppm are from HPO4 2− (bulk, distorted, or surface), except for the large peak at around 5 ppm, which arises from adsorbed water. It must be pointed out that the last region (peaks above~2 ppm) varies greatly depending on random changes in synthesis conditions such as the rate and order of adding reactants, stirring or vibration modes, humidity, etc. Therefore, huge variations were observed in reported results [43][44][45][46][47][48][49] as well as in the spectra shown in Figure 3 (top left in particular). It is advisable not to make quantitative claims based on the peaks in this region (please refer to the 2D HETCOR spectra discussed, below which offer far more consistent results). In contrast, we can draw far more quantitative conclusions from the region between 0 and 2 ppm, which is more reliable and much less variable. The top left of Figure 3 indicates that changes in pH have little influence on hydroxide and structured water, indicating that the crystalline region is hardly changed, but that the amorphous region may be changed even with a slight change of pH. Clear evidence of fluoride substitution of hydroxyl is displayed in Figure 3b, where the gradual vanishing of the hydroxide peak at~0.0 ppm is shown. It is found that all hydroxide ions were substituted when the fluoride concentration was above 50 mM (Figure 3b). In the meantime, a peak corresponding to structural water emerged and the bulk water peak was shifted upfield, indicating that fluoride substitution

Solid-State NMR
To further investigate how synthesis conditions affect HAp and in particu straightforward evidence of fluoride substitution in F-HAp, both 1D and 2D mul ( 1 H, 19 F, 31 P) solid-state NMR spectra were acquired. Figure 3 is the 1 H 35 kHz MA of various HAp and F-HAP samples. Based on the assignment of 1 H spectrum hydroxyl group corresponds to a peak at around 0.00 ppm; peaks between ~0.5 are from surface-structured water; peaks above 2 ppm are from HPO4 2-(bulk, d or surface), except for the large peak at around 5 ppm, which arises from adsorb It must be pointed out that the last region (peaks above ~2 ppm) varies greatly de on random changes in synthesis conditions such as the rate and order of adding r stirring or vibration modes, humidity, etc. Therefore, huge variations were obs reported results [43][44][45][46][47][48][49] as well as in the spectra shown in Figure 3 (top left in pa It is advisable not to make quantitative claims based on the peaks in this regio refer to the 2D HETCOR spectra discussed, below which offer far more consisten The 1D 31 P 35 kHz MAS spectra of the samples ( Figure S3) are much less informative because the difference between the chemical shifts of phosphate and hydrogen phosphate is small. However, some consistent trends are noteworthy. For instance, the 31 P peak shifted upfield as the concentration of NaF or the soaking time increased. This means fluoride substitution brings more shielding onto phosphorus than hydroxide. This does not seem to make sense, since fluoride is more electronegative and 31 P nuclei should be less shielded after substitution. However, if we notice that the unit cell size of F-HAp (a = 937 pm) is smaller than that of pristine HAp (a = 941 pm), then we can solve this problem. There are two competing facts here: the increased electronegativity and the increased local electronic density; the higher local electronic density overwhelms the effect of increased electronegativity. It is noteworthy that our results and interpretation on F-HAp are in agreement with the enhanced stability of fluorine-substituted organic molecules and proteins, as discussed in recent years [50][51][52][53].
ion concentration could increase fluoride substitution in HAp, but could not achieve com plete substitution (Figure 3, bottom), as the hydroxide peak was still strong after soakin in a 50 mM NaF solution for 7 days. This phenomenon implies that fluoride substitutio of hydroxide in deep regions of bone or tooth is inefficient or unlikely. The 1D 31 P 35 kHz MAS spectra of the samples ( Figure S3) are much less informativ because the difference between the chemical shifts of phosphate and hydrogen phosphat is small. However, some consistent trends are noteworthy. For instance, the 31 P pea shifted upfield as the concentration of NaF or the soaking time increased. This means flu oride substitution brings more shielding onto phosphorus than hydroxide. This does no seem to make sense, since fluoride is more electronegative and 31 P nuclei should be les shielded after substitution. However, if we notice that the unit cell size of F-HAp (a = 93 pm) is smaller than that of pristine HAp (a = 941 pm), then we can solve this problem There are two competing facts here: the increased electronegativity and the increased loca electronic density; the higher local electronic density overwhelms the effect of increase electronegativity. It is noteworthy that our results and interpretation on F-HAp are i agreement with the enhanced stability of fluorine-substituted organic molecules and pro teins, as discussed in recent years [50][51][52][53].
The variations of 19 F 35 kHz MAS spectra over samples (Figure 4, Figures S4 and S5 are more appreciable than that of its 31 P counterparts. Common to all 19 F MAS spectra the fact that substitution of hydroxide with fluoride leads to an increase of the 19 F peak a ~−104 ppm, which should be assigned to crystalline fluoride. The peaks between ~−11 ppm and ~−115 ppm should be assigned to fluoride in the amorphous region. As show The variations of 19 F 35 kHz MAS spectra over samples (Figure 4, Figures S4 and S5) are more appreciable than that of its 31 P counterparts. Common to all 19 F MAS spectra is the fact that substitution of hydroxide with fluoride leads to an increase of the 19 F peak at~−104 ppm, which should be assigned to crystalline fluoride. The peaks betweeñ −110 ppm and~−115 ppm should be assigned to fluoride in the amorphous region. As shown in the top left spectra of Figure 4, increasing the NaF concentration leads to a great increase of crystalline fluoride. A remarkable feature of the 19 F MAS NMR spectra shown in Figure 4 is that the spectra show significant change with different recycle delays. This can be explained by the presence of fluoride species in vastly different micro-environments; the fluoride ions in crystalline regions have much shorter longitudinal relaxation T1, whereas the fluoride ions in the amorphous regions have much longer longitudinal relaxation times. The difference in relaxation time in different regions is not surprising; what is surprising here is the magnitude of the difference. The relaxation rate in the amorphous region is more like that of free liquids than that of soft solids. We mention that this phenomenon deserves more extensive investigation, but this is beyond the scope of this work. explained by the presence of fluoride species in vastly different micro-environments; the fluoride ions in crystalline regions have much shorter longitudinal relaxation T1, whereas the fluoride ions in the amorphous regions have much longer longitudinal relaxation times. The difference in relaxation time in different regions is not surprising; what is surprising here is the magnitude of the difference. The relaxation rate in the amorphous region is more like that of free liquids than that of soft solids. We mention that this phenomenon deserves more extensive investigation, but this is beyond the scope of this work. While 1D multinuclear solid-state NMR spectra can offer valuable structural, morphological, and dynamic information on F-HAp, as pointed out above, there are some disadvantages. For instance, for 1 H MAS spectra, the peaks above ~2 ppm vary so much that they hardly offer quantitative conclusions; for 31 P MAS spectra, the broad peaks have low resolution. Therefore, we also performed a series of 2D heteronuclear correlation solid-state NMR experiments and 1 H-31 P and 19 F-31 P HETCOR spectra were obtained, as shown in Figures 5 and 6, Figures S6 and S7, respectively. The most important feature of these 2D 1 H-31 P and 19 F-31 P HETCOR spectra is that different 31 P sites corresponding to (PO4 3-) and (HPO4 2− ) in various micro-environments are clearly resolved. In particular, be- While 1D multinuclear solid-state NMR spectra can offer valuable structural, morphological, and dynamic information on F-HAp, as pointed out above, there are some disadvantages. For instance, for 1 H MAS spectra, the peaks above~2 ppm vary so much that they hardly offer quantitative conclusions; for 31 P MAS spectra, the broad peaks have low resolution. Therefore, we also performed a series of 2D heteronuclear correlation solidstate NMR experiments and 1 H-31 P and 19 F-31 P HETCOR spectra were obtained, as shown in Figures 5 and 6, Figures S6 and S7, respectively. The most important feature of these 2D 1 H-31 P and 19 F-31 P HETCOR spectra is that different 31 P sites corresponding to (PO 4 3− ) and (HPO 4 2− ) in various micro-environments are clearly resolved. In particular, because of the long transverse relaxation time in the mobile regions, 1 H-31 P and 19 F-31 P correlation peaks in these regions are not only observable, but also resolvable. The gradual vanishing of the hydroxide peak with increasing NaF concentration correlates well with the increase of hydrogen phosphate, offering straightforward evidence that full fluoride substitution of hydroxide could be achieved at 10 mM concentration of NaF. The remaining strong 1 H-31 P cross peaks clearly show that these peaks arise from hydrogen phosphate. On the hand, as shown by the conspicuous OH − -31 P cross peaks in Figure S6, complete fluoride substitution could not be achieved simply by soaking, although higher NaF concentration and longer soaking time help fluoride substitution.
substitution of hydroxide could be achieved at 10 mM concentration of NaF. The remaining strong 1 H-31 P cross peaks clearly show that these peaks arise from hydrogen phosphate. On the hand, as shown by the conspicuous OH − -31 P cross peaks in Figure S6, complete fluoride substitution could not be achieved simply by soaking, although higher NaF concentration and longer soaking time help fluoride substitution.  Finally, to obtain quantitative dynamics information in different regions, 1 H-31 P CPMAS spectra with different contact times for each sample were acquired. Figure 7 shows the CP curves for different samples from which the 1 H-31 P cross relaxation time TCP and 1 H transverse relaxation time during the RF pulse, T1ρ, can be obtained by fitting the CP curves with the following formula: substitution of hydroxide could be achieved at 10 mM concentration of NaF. The remaining strong 1 H-31 P cross peaks clearly show that these peaks arise from hydrogen phosphate. On the hand, as shown by the conspicuous OH − -31 P cross peaks in Figure S6, complete fluoride substitution could not be achieved simply by soaking, although higher NaF concentration and longer soaking time help fluoride substitution.  Finally, to obtain quantitative dynamics information in different regions, 1 H-31 P CPMAS spectra with different contact times for each sample were acquired. Figure 7 shows the CP curves for different samples from which the 1 H-31 P cross relaxation time TCP and 1 H transverse relaxation time during the RF pulse, T1ρ, can be obtained by fitting the CP curves with the following formula: Finally, to obtain quantitative dynamics information in different regions, 1 H-31 P CPMAS spectra with different contact times for each sample were acquired. Figure 7 shows the CP curves for different samples from which the 1 H-31 P cross relaxation time T CP and 1 H transverse relaxation time during the RF pulse, T 1ρ , can be obtained by fitting the CP curves with the following formula: It is found that best fitting is reached for all samples when a single component model is used. This seems a surprising result since there are at least two types of 31 P microenvironments. However, considering the fluoride substitution and the rigid hydroxide in the crystalline region (which means very short T 1ρ ), it is reasonable to assume that the 1 H-31 P CPMAS spectra from the signal of HPO 4 2− is dominant. From this result, we may have a more straightforward interpretation of the 1 H-31 P HETCOR S spectra ( Figure 5, Figures S5 and S6), namely, the 1 H-31 P cross peaks corresponding to a 1 H chemical shift larger than~2.0 ppm arise predominantly from HPO 4 2− , since in the HETCOR pulse sequence, there is a CP block [54].
ronments. However, considering the fluoride substitution and the rigid hydroxide in the crystalline region (which means very short T1ρ), it is reasonable to assume that the 1 H-31 P CPMAS spectra from the signal of HPO4 2− is dominant. From this result, we may have a more straightforward interpretation of the 1 H-31 P HETCOR S spectra ( Figure 5, Figures S5  and S6), namely, the 1 H-31 P cross peaks corresponding to a 1 H chemical shift larger than ~2.0 ppm arise predominantly from HPO4 2− , since in the HETCOR pulse sequence, there is a CP block [54]. Research in the past few decades has achieved great progress in the structure, morphology, dynamics, and functions of the HAp system, which enables us to have a deeper understanding of bone structure, bone physiology, and bone pathology, and to develop a large number of related bionic materials, coatings, or drug delivery systems [42,[55][56][57][58]. However, there are still many unanswered questions. The effect of ions on the composition, structure, and morphology of the HAp system is an example. This work is a further effort in this direction. It is anticipated that in the future, there will be more research on ionic behavior in the HAp system, such as how interactions between ions and other large and small molecules (hormones, collagen, regulatory proteins, etc.) affect the structure, morphology, and function of HAp. There is also significant room for improvement in experimental methods (including sample preparation, instrument technology, and applications) and theoretical computation and simulation. Research in the past few decades has achieved great progress in the structure, morphology, dynamics, and functions of the HAp system, which enables us to have a deeper understanding of bone structure, bone physiology, and bone pathology, and to develop a large number of related bionic materials, coatings, or drug delivery systems [42,[55][56][57][58]. However, there are still many unanswered questions. The effect of ions on the composition, structure, and morphology of the HAp system is an example. This work is a further effort in this direction. It is anticipated that in the future, there will be more research on ionic behavior in the HAp system, such as how interactions between ions and other large and small molecules (hormones, collagen, regulatory proteins, etc.) affect the structure, morphology, and function of HAp. There is also significant room for improvement in experimental methods (including sample preparation, instrument technology, and applications) and theoretical computation and simulation.

Conclusions
In summary, some important discoveries based on various characterization methods employed on fluorinated HAp samples have been achieved, particularly from multinuclear solid-state NMR. (1) Morphology changes with pH (7.0~7.4) are minor. Crystalline (core)amorphous (shell) structure is universal, with the core/shell ratio varying with synthetic condition. (2) Differences between our samples and Kobayashi's work [32] are significant, indicating the importance of synthetic conditions in controlling morphology. Their method of adding a nucleation center helps rapid crystallization (in~1 h), in contrast to the wet precipitation method, which requires 24 h before the formation of precipitate. (3) 1D 1 H and 19 F spectra show that full substitution of hydroxide by fluoride can be reached when the concentration of F − reaches 5 mM. When the concentration of NaF approaches 10 mM, a new peak at~3.5 ppm arises, supposedly from bulk HPO 4 2− , probably because of the hydrogen bonding between F − and HPO 4 2− . (4) OH − peak diminishing is accompanied with the appearance of peaks at 1~2 ppm corresponding to structural water; the water peak is upfield shifted with the increase of NaF. (5) The 1D 19 F MAS NMR spectra show significant change with different recycle delays, indicating the presence of different fluorine environments, which was also confirmed with 2D 19 F-31 P HETCOR, showing larger CSA for samples synthesized in higher NaF concentrations. (6) Based on the experimental data, single component fitting produced better accuracy; the CP signal is stronger for the brushite region.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/cryst12020139/s1; Figure S1. HRTEM images of HAp synthesized in SBF at different pH values pH = 7.0 (a); pH = 7.2 (b) and pH = 7.4 (c). Figure S2. HRTEM images of HAp synthesized in SBF followed by soaking in NaF solution at 0.0025 M for 1 day (a); 3 days (b) and 7 days (c). Figure S3.  Acknowledgments: The XRD and HRTEM experiments were carried out on the NSYSU high valued instrument center. NSYSU and National Taiwan University (NTU) solid-state NMR spectrometers are acknowledged for the 1D and 2D experiments. In particular, we gratefully acknowledge Shing-Jong Huang of NTU for his professional assistance. The National Center for High-Performance Computing of Taiwan is gratefully acknowledged for providing the computing software and hardware to support this project.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.