Crystal growth of HAp on plate-like ZnO particles using APTES as surface treatment agents

ABSTRACT We successfully prepared Hydroxyapatite/Zinc oxide (HAp/ZnO) composite particles with hexagonal plate-like shapes. The surface of the ZnO particles was treated with (3-aminopropyl)triethoxysilane (APTES) as a scaffold for the crystallization of HAp from the precursor solution. From the results of XRD measurements and SEM images of composite particles, formation of HAp on the ZnO particles was revealed. In addition, diffraction peaks associated with the (0 0 2) and (0 0 4) planes of HAp were clearly observed. They also revealed that the HAp/ZnO composite particles had a c-face orientation. Furthermore, element mapping analysis by EPMA showed that the elemental distributions of Ca, P and Zn of the composites were almost coincident. These results suggested that the formation of HAp occurred on the ZnO particles. In contrast, the ZnO particles without APTES treatment readily dissolved in the precursor solution and the diffused Zn2+ ions reacted with PO4 3- ions and Ca2+ ions to form CaZn2(PO4)2. APTES treatment of the ZnO surfaces appeared to prevent the dissolution of ZnO in the solution and induce the adsorption of anions such as phosphate on ZnO.


Introduction
Hydroxyapatite (HAp, Ca 5 (PO 4 ) 3 OH) is a well-known biocompatible ceramic material that has high biocompatibility and bioactivity [1][2][3][4]. HAp has been applied as a biomaterial in the fields of medicine and tissue engineering as a bone replacement agent [5,6], coating agent on implants [7,8] and adsorbent for the separation of proteins and viruses [9,10]. These functions change depending on the shape of the HAp particles. Therefore, controlling the particle shape is important. Much effort has gone into particle shape control, and plate [11,12], rod [13,14], sphere [15,16] and other particle shapes have been reported. HAp belongs to a hexagonal crystal system and exhibits polarity along the c axis. Also, the ab plane of HAp is rich in Ca 2+ ions and is positively charged, whereas the c plane is rich in PO 4 3ions and OH − ions and is negatively charged [17]. Since the surface properties change depending on the orientation of the crystal plane, the selective synthesis of particles with specific orientations of crystal planes is important for developing functional materials. In contrast, photocatalytic materials such as titanium dioxide (TiO 2 ) and zinc oxide (ZnO) are expected to be applied as functional inorganic materials in areas such as environmental purification [18,19], cancer therapy [20,21], and the decomposition of organic compounds [22,23]. A combination of HAp and such photocatalysts permits the simultaneous adsorption and decomposition of viruses, bacteria and other toxic agents [24][25][26]. As mentioned above, since the adsorption properties of HAp depend on the crystal plane, it is necessary to control the shape of composite particles according to the target substance. However, there are few reports on shape control of composite particles focusing on the orientation of crystal planes. We considered that it important to limit the reaction field as well as to control the direction and reaction rate of crystal growth. There are several reports on the synthesis of HAp using chelating agents such as EDTA to control the reaction rate [27][28][29]. In addition, one method of limiting the reaction field is the epitaxial growth of particles on substrates used for thin films [30][31][32]. If the reaction field can be limited to particles rather than planar substrates, composite particles can be synthesized that are oriented along specific crystal planes.
In our recent work, we have succeeded in preparing hexagonal plate-like ZnO particles with crystal growth control [33][34][35]. We focused on surface treatment by (3-aminopropyl)triethoxysilane (APTES) as a method to enable the use of these hexagonal plate-like ZnO particles as substrates. APTES is known to induce the adsorption of anionic species such as phosphate [36,37]. Therefore, treating the surface of the ZnO particles with APTES promotes the adsorption of phosphate ions and the formation of HAp on the ZnO. We hypothesized that crystal growth of HAp could be controlled on the ZnO particles. In this report, we propose a novel synthetic method that incorporates a combination of these techniques.

Materials
Hexagonal plate-like ZnO particles prepared by our previously reported method [33][34][35] were used. APTES (Sigma Aldrich) was used for the surface treatment of ZnO particles. Ca(NO 3 ) 2 · 4H 2 O (Wako Pure Chemical Co., Ltd., Japan) and (NH 4 ) 2 HPO 4 (Wako Pure Chemical Co., Ltd., Japan) were used as the calcium and phosphorus sources for HAp, respectively. Ethylenediamine-N,N,N',N'-tetraacetic acid (EDTA; Dojindo Laboratories, Japan) was used as a chelating agent for Ca ions for the preparation of HAp/ZnO composite particles.

Surface treatment of hexagonal plate-like ZnO using APTES
First, the specific surface area of the hexagonal platelike ZnO particles was measured using nitrogen adsorption/desorption measurements by the BET method. From the specific surface area values, the APTES concentration (n) for surface treatment of ZnO particles was calculated as follows.
Here, M is the molecular weight of APTES, S is the amount of ZnO added, and A ZnO and A APTES are the specific surface area of hexagonal plate-like ZnO particles and the area occupied by APTES molecules, respectively. A APTES was calculated as the area of a sphere having a radius the length of the Si-O bond. APTES coverage corresponding to a monolayer film on the total surface area of hexagonal plate-like ZnO particles was set to n(coverage) = 1.0, and surface treatments were performed at n = 0.5, 1.0, 3.0, and 5.0. For comparison, ZnO particles without surface treatment were also prepared, with n = 0.
Second, the surface treatment was performed as follows. A mixture of ethanol and APTES with a volume ratio of 1:1 was prepared for each concentration. A total of 0.2 g of ZnO particles was added to the solution and then left to stand for 12 hours at room temperature. In addition, samples were held at 70°C for 12 h to obtain surface-treated ZnO(ZnO/APTES) particles with various coverage levels (n = 0.5-5.0). The ZnO/APTES particles were collected by centrifugation and washed three times with ethanol and deionized water. These samples were characterized by Fourier transform infrared spectroscopy (FT-IR; FT/IR4200, JASCO, Japan). The morphology of the particles was observed by scanning electron microscopy (SEM; VE-7800, Keyence, Japan).

Preparation of HAp/ZnO composite particles with hexagonal plate-like shapes
A 10 mM calcium nitrate solution and a 10 mM EDTA solution were mixed. Subsequently, a 6 mM diammonium hydrogen phosphate solution was added to the mixed solution to prepare a precursor solution of HAp. The Ca/P molar ratio in the precursor solution was 1.67. Then, 0.2 g of various ZnO/APTES particles were added to the solution and allowed to stand at room temperature for 6-48 hours to synthesize HAp/ ZnO composite particles. All obtained composites were collected by centrifugation and washed three times with deionized water, then dried at 70°C for 24 hours. The sample was characterized by X-ray diffraction (XRD; MiniFlex600, Rigaku, Japan Cu-Kα radiation) and FT-IR (FT/IR4200, JASCO, Japan). The morphology of the particles was observed by SEM (VE-7800, Keyence, Japan). An elemental mapping analysis of composites was performed by an Electron Probe Micro Analyzer (EPMA; EPMA-1720, Shimazu, Japan). Figure 1 shows the FT-IR spectra and SEM images of ZnO/APTES(n = 1.0), respectively. For comparison, the FT-IR spectrum and SEM image of hexagonal platelike ZnO without APTES treatment (ZnO/APTES (n = 0)) is also shown. In the spectrum of ZnO/ APTES(n = 1.0) in Figure 1(b), the Si-O-Si stretching vibration at 1020 cm −1 [38][39][40] and C-N stretching vibration at 1477 cm −1 [38][39][40] attributed to APTES were observed. Furthermore, C-H stretching vibrations at 2875 cm −1 and 2931 cm −1 [38][39][40] attributed to the alkyl chain in APTES were observed. The peaks of the asymmetric stretching mode of NH 2 at 3300 cm −1 [38][39][40] and bending vibration of NH 2 at 1620 cm −1 [38][39][40] were also observed. The peak appearing at 900 cm −1 could be assigned to Zn-O-Si bonds [39,40]. These absorption peaks were not observed for the ZnO particles without APTES (Figure 1(a)). In addition, SEM observation was performed to investigate the effect of surface treatment on particle shape and morphology. There was no difference observed between the SEM images of particles before and after APTES treatment. APTES modifies the surface of ZnO particles by dehydrationcondensation but does not seem to cause associated particle aggregation. These results suggest that the surface of the ZnO particles was successfully treated by APTES.

Preparation of HAp/ZnO composite particles
In the XRD patterns of the composites prepared with a reaction time of 6 hours, diffraction peaks attributed to ZnO were observed in the samples with ZnO/APTES (n = 0.5-5.0) (Supporting information Figure 1). There were no differences compared with the patterns of single crystal ZnO particles. In contrast, not only the diffraction peaks of ZnO, but also a diffraction peak attributed to CaZn 2 (PO 4 ) 2 [41,42] was observed in the sample with ZnO/APTES(n = 0). This peak was not observed in the other patterns. Under this reaction condition, we found that only ZnO/APTES(n = 0) formed CaZn 2 (PO 4 ) 2 , while the others did not.
In the XRD patterns of composites prepared with reaction times of 12 and 24 hours, not only diffraction peaks attributed to ZnO, but also diffraction peaks attributed to other products were observed (Supporting information Figures 2 and 3). Composites prepared using ZnO/APTES(n = 0) showed formation of CaZn 2 (PO 4 ) 2 at both 12 hours and 24 hours of reaction. In contrast, composites prepared using ZnO/APTES (n = 0.5-5.0) showed a new peak at 2θ = 25.94 °. This peak was not consistent with either product. This 2θ value is almost consistent with the diffraction peak attributed to the (0 0 2) plane of HAp; thus, these results suggested the possibility of HAp formation. Figure 2 indicates the XRD patterns of composites prepared using ZnO/APTES(n = 0-5.0) with a reaction time of 48 hours. In the case of the composite prepared  using ZnO/APTES(n = 0), CaZn 2 (PO 4 ) 2 was formed as well as in some other systems. However, composites prepared using ZnO/APTES(n = 0.5-5.0) showed new peaks at 2θ = 25.94 ° and 2θ = 53.5 °. These peaks are nearly identical to the peak positions of the (0 0 2) and (0 0 4) planes of HAp, respectively [43][44][45]. On comparing these 2θ values with the standard values for HAp reported by Rodriguez-Lorenzo et al. [45], each peak was found to shift to a higher angle. Zinc-substituted HAp is known to show an increase in the "a" lattice parameter and decrease in the "c" lattice parameter [46,47]. This is due to the replacement of Ca ions with Zn ions, which have a smaller ionic radius. Therefore, the composites prepared using ZnO/APTES(n = 0.5-5.0) are likely to contain zinc-doped HAp. Focusing on the diffraction peak of the (0 0 2) plane, this began to appear at a reaction time of 12 hours. The diffraction peak intensity increased with increasing reaction time. It appeared that the formation of HAp starts at around 12 hours of reaction time, and that crystal growth proceeds with the passage of time. These results suggested that the HAp/ ZnO/APTES composite was formed under these reaction conditions.
The crystallite size was calculated using Scherrer's equation from the diffraction peaks attributed to the (0 0 2) and (0 0 4) crystal planes of HAp (Figure 3). The crystallite size tended to increase at surface treatment concentrations up to n = 1.0. In contrast, crystallite size decreased at the surface treatment concentrations with n > 1.0. When the composite particles were prepared using ZnO/APTES (n = 1.0), produced HAp particles would have most crystalline.  [48][49][50]. Bands observed at about 1046 cm −1 and 1096 cm −1 were assigned to the asymmetric stretching vibration mode of PO 4 3-(ν 3 ) [48][49][50]. In addition, bands attributed to CO 3 were observed at around 1440-1470 cm −1 [51,52]. This suggests that the composite contains some B-type carbonated HAp [51,52]. These peaks were observed for all composite particles prepared with ZnO/APTES(n = 0.5-5.0) regardless of reaction time; whereas, in the case of reaction times from 6 to 24 hours, C-H stretching vibrations attributed to the alkyl chain of APTES were observed. It is possible that for the composite particles obtained under these reaction conditions, the products do not completely coat the surface of the ZnO particles, and the APTES-coated surface is exposed. However, these C-H peaks were not observed for the reaction time of 48 hours. These results suggested increasing coverage of the HAp occurs on the ZnO particle surface due to crystal growth of HAp with increasing reaction time. Figure 5 shows SEM images of HAp/ZnO composite particles prepared under various conditions. As shown in the SEM image in Figure 1(a), the hexagonal platelike ZnO single-crystal particles have very smooth c faces, although the composite particles prepared using ZnO/APTES(n = 0) have hexagonal plates with rounded corners and microscopic pores on the c faces. Furthermore, as the reaction time was increased to 12 and 24 hours, the ZnO particles were observed to gradually dissolve, and when the reaction time reached 48 hours, the hexagonal plate-like shape of  the ZnO particles could not be maintained. This indicates the dissolution of ZnO particles in the precursor solution. In contrast, no micropores were observed on the c face of the ZnO particles, and their surfaces were observed to be coated by different products for the composite particles prepared using ZnO/APTES (n = 0.5-5.0). As the reaction time was increased, the product formation on the surface became more pronounced.
Elemental mapping analysis was performed on the composite particles prepared with a reaction time of 48 hours ( Figure 6). In the case of ZnO/APTES(n = 0), the elemental distribution positions of Ca and P do not match those of Zn. It is clear that CaZn 2 (PO 4 ) 2 , which is formed when ZnO without APTES treatment is used, is not formed on the ZnO surface but in the bulk solution. In contrast, the elemental distribution positions of Ca, P and Zn are nearly coincident when using ZnO/APTES (n = 0.5-1.0). However, distribution of Ca and P was observed in the vicinity of ZnO but mostly on the side (ab plane) of ZnO particles and almost no distribution on the c plane was observed for ZnO/APTES(n = 0.5-3.0). In particular, the composite particles prepared using ZnO/APTES(n = 5.0) resulted in significantly lower Ca and P elemental distributions.
First, the effect of the presence or absence of APTES treatment on the products was considered. ZnO without APTES treatment is expected to easily dissolve in a pH 4 precursor solution containing EDTA. Therefore, we considered that high dissolution of Zn 2+ ions from inside the crystal occurs with diffusion into the solution, and that CaZn 2 (PO 4 ) 2 is formed in regions other than the surface of ZnO particles. In contrast, APTES treatment of ZnO is expected to prevent the dissolution of Zn 2+ ions from the ZnO crystal. In addition, APTES has an amino group, so the surface of the ZnO particles coated by APTES will have a positive charge. Anionic species such as PO 4 3ions adsorb on the ZnO/ APTES surface by electrostatic interaction; thus, ZnO/ APTES limited the reaction field of HAp formation. Second, the effect of the APTES treatment concentration on the product was considered. In the treatment concentrations with n ≤ 1.0, the ZnO/APTES particle surfaces were smooth due to APTES forming a monolayer. In contrast, the ZnO/APTES surface is expected to become irregular at APTES concentrations with n > 1.0 because APTES may oligomerize locally [53]. As a result, the smoother surfaces are more likely to form HAp than the irregular ones.
Finally, to elucidate the formation mechanism of HAp/ZnO composite particles and CaZn 2 (PO 4 ) 2 particles in our system, the change in the solution pH from 0 hours (immediately after the addition of various ZnO/ APTES particles to the precursor solution) to 48 hours of reaction was investigated (Figure 7). The increase in pH with reaction time was found to be significantly different between ZnO/APTES(n = 0) and ZnO/APTES (n = 0.5-5.0). When ZnO/APTES(n = 0) was used, the pH increased rapidly from 0.5-1.0 hours, reaching a pH of about 6.3 at 2 hours, and then remained constant. In contrast, when ZnO/APTES(n = 0.5-5.0) was used, the pH gradually increased with reaction time independent of the APTES concentration. The pH reached 6.5 at 10 hours and then did not change further up to 48 hours. This difference is thought to be due to the dissolution rate of Zn 2+ ions and affects the composition of the product.
In this experiment, we added EDTA to inhibit the reaction between Ca 2+ ions and PO 4 3ions. The Ca-EDTA chelate is formed in the precursor solution, but this chelate structure changes when Zn 2+ ions dissolve from the ZnO particles. The chelate formation reaction between completely dissociated EDTA (E −4 ) and metal ions (M n+ ) is given as follows: and its equilibrium constant k is given as follows: The value of k changes depending on conditions such as pH, but it is known that under the same conditions, a chelate composed of a metal ion having a low k value will be replaced with a metal ion having a higher k value. Zn-EDTA has a larger k value than Ca-EDTA [54,55]. Therefore, the Zn 2+ ions undergo a substitution reaction with the Ca-EDTA chelate, and Ca 2+ ions are released into the solution with Zn-EDTA chelate formation. It is considered that the released Ca 2+ ions react with PO 4 3ions and OH − ions adsorb on the ZnO/APTES(n = 0.5-5.0) particles to obtain plate-like HAp/ZnO composite particles (Figure 8(a)). Wang et al. reported that -NH 3 + groups of amino acid forms hydrogen bonds with -PO 4 3on the HAp faces [56]. Therefore, the crystal plane of HAp in contact with the ZnO/APTES particles is expected to be the c plane. Since the c plane of HAp is fixed on the ZnO/APTES particles, HAp grows epitaxially along the c axis. HAp/ZnO composite particles are considered to have a large proportion of the c plane of HAp.
When ZnO/APTES(n = 0) is used, PO 4 3ions and OH − ions cannot be adsorbed on the surface, so a product cannot be formed on the surface (Figure 8(b)). In addition, CaZn 2 (PO 4 ) 2 was produced when the dissolution rate of Zn 2+ ions exceeded the amount of Ca 2+ ions released by the substitution reaction. On the other hand, when ZnO/APTES (n = 0.5-5.0) was used, the dissolution rate of Zn 2+ ions was slowed, so the release rate of Ca 2+ ions due to the substitution reaction also slowed down, and HAp was generated on the surface of the ZnO particles.
For comparison, experiments were performed under the same conditions using spherical silica and plate-like mica as examples of metal oxide particles different from ZnO particles. There was no change in the pH of the precursor solution, and there was no change in the composition of the product before versus after the treatment (Supporting information Figure 7-9). This revealed that Si 2+ , Mg 2+ , and Al 3+ ions did not replace Ca 2+ in the Ca-EDTA chelate, and HAp and other calcium phosphate compounds were not produced. The above results show that the ZnO particles treated with APTES not only induce the adsorption of PO 4 3-ions and OH − ions on the surface, but also change the release rate of Zn 2+ ions from ZnO and Ca 2+ ions from the Ca-EDTA chelate. By slowing down the release of ions in solution, it became possible to limit the HAp formation reaction to the particle surface.

Conclusion
In this study, we succeeded in preparing plate-like HAp/ ZnO composite particles by controlling the crystal growth of HAp on ZnO particles using APTES as a surface treatment agent. When ZnO/APTES(n = 0) that had not been surface-treated with APTES was used, CaZn 2 (PO 4 ) 2 was newly generated in the solution instead of on the ZnO particle surface. In contrast, when ZnO/APTES(n = 0.5-5.0) with APTES treatment was used, HAp was generated on the surface of ZnO particles for reaction times of 12 hours or more regardless of the APTES concentration. It was also found that the thickness of HAp generated on the ZnO particles increased as the reaction time increased. The formation of HAp and CaZn 2 (PO 4 ) 2 occurred when Zn 2+ ions dissolved from ZnO particles replaced Ca 2+ ions in the Ca-EDTA chelate to release Ca 2+ ions into the solution. The surface treatment of APTES has two effects on the formation of HAp. One is that the amino group of APTES imparts a positive charge to the surface of the ZnO particles, so PO 4 3ions and OH − ions are adsorbed on the surface to limit the production field of HAp. The other is to reduce the release rate of Zn 2+ ions in the precursor solution and slow down their substitution reaction with Ca 2+ ions in the Ca-EDTA chelate. When ZnO/APTES(n = 0) was used, the dissolution rate of Zn 2+ ions and the substitution reaction rate of the Ca-EDTA chelate were relatively high, and CaZn 2 (PO 4 ) 2 was produced, which did not lead to shape control for the particles. In contrast, when ZnO/APTES(n = 0.5-5.0) was used, the dissolution rate of Zn 2+ ions and the substitution reaction rate of Ca-EDTA chelate with Ca 2+ ions were relatively slow, and the formation reaction rate of HAp was slow, so the shape of the particles was controlled.

Author contributions
The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

Disclosure statement
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.