The pre-treated substrates were immersed in 1.5 SBF and the formation of a CaP coating was characterized after 14 days of immersion. Except for Pi-Ti, relatively dense CaP coatings were found by SEM at the surface of all pre-treated substrates (Figure 4). The CaP films comprised globular structures (1 to 0.5 µm) consisting of dense plate-like crystals (Supplementary Figure S4) as a result of a three-dimensional growth process [13,52,53]. These globular structures are typical to biomimetic methods [38,54,55,56,57,58,59,60,61,62] and are observed in the bone cement line matrix interlocking with the surface of the modified titanium oxide at the bone-implant interface [63]. On the surface of Pi-Ti, only small particles of CaP (~50 nm in diameter) were found and large globular structures of CaP could only be observed after 1 month of immersion (Supplementary Figure S5). Accordingly, it can be speculated that the surface of Pi-Ti induces CaP with slower growth kinetics. Such a slow kinetics on Pi-Ti is neither attributed to the reduced thickness of oxide layer on this surface nor to the nature of surface hydroxyl groups. The reason is that when Pi-Ti was compared to EP-Ti, the XPS Ti2p spectra (Supplementary Figure S1) proved a thicker oxide layer on Pi-Ti compared to EP-Ti, and as shown in Figure 3, both had acidic hydroxyl groups. Therefore, it is believed that lack of charged groups on the surface of Pi-Ti is the reason for its slower kinetics in inducing CaP growth as compared to other surfaces.

Comparing the coating on Cl-Ti and EP-Ti, it is observed that the coating on EP-Ti is less dense than that on Cl-Ti and the underlying Ti substrate is exposed in between globular structures on EP-Ti (as indicated by arrows on the magnified inset image of EP-Ti in Figure 4). This could be explained by lower density of surface OH on EP-Ti compared to Cl-Ti as confirmed by XPS spectra in O1s region. Moreover, a sharper Cl2p that is observed on XPS survey spectrum of Cl-Ti compared to EP-Ti shows that there are more Cl⁻ ions contributing to the mineralization process, thus resulting in a denser coating on Cl-Ti compared to EP-Ti.

Further characterization of the film thickness revealed that different growth regimes occur at various Ti substrates, with Cl-Ti and EP-Ti showing the thickest film (~9.5 µm) after 1 month of immersion in 1.5 SBF (Figure 5). It is noteworthy that globular films have formed on all surfaces regardless of the initial surface roughness. This suggests that heterogeneous nucleation of CaP on Ti had happened immediately after immersion in 1.5 SBF and did not depend on the surface topography in agreement with previous reports [57]. However, this statement is only valid for in vitro surface-induced mineralization of CaP on Ti surfaces and it is not applicable to bone bonding onto Ti implant surfaces. It is very well understood that implant surfaces with complex microtopography, and not the smooth surfaces, render surfaces bone bonding [64]. In terms of in vitro surface-induced mineralization, it can be speculated that the surface topography would only influence the mechanical stability of final CaP film and not the initial surface-induced formation of CaP nuclei. This is because a more adherent CaP coating will be formed on rough surfaces due to increased interfacial adhesion strength between the coating and Ti surface [57,65,66].

The crystallinity of the films was investigated by XRD after immersion in 1.5 SBF for 14 days (Figure 6a). All substrates showed peaks at 2θ = 35, 38.2, 40.1 and 52.7° related to the (100), (002), (101) and (102) planes of Ti, respectively [13,18,28,55,67,68]. Once more, except for Pi-Ti, all substrates showed peaks characteristic of a carbonated HA, i.e., at 2θ = 25.9 and 31.8°. The former peak corresponds to the overlapping of the (201) and (002) planes of carbonated HA, while the latter is related to the overlapping of the (211), (112) and (300) reflections [28,55,67,69,70,71,72]. The additional peak at 2θ = 35.5–36° only appearing on the XRD patterns of the Na-Ti and Pi-Ti was attributed to the superposition of the carbonated HA or tricalcium phosphate peak. The broadness of the HA peaks also indicates very small crystal sizes and/or the orientation of crystals at different directions. The absence of these reflexes (and the low intensity of the peaks) is probably due to the thinness of the CaP film [19]. Moreover, the intensity of the peak at 2θ = 25.9° is equal to that of 2θ = 31.8° in the patterns, while it should normally be less than a quarter for the carbonated HA. This could be due to the reflection of other crystalline phases, such as crystalline dicalcium phosphate with an XRD peak at 2θ = 23.2° corresponding to (040) plane [73] and tricalcium phosphate with an XRD peak at 2θ = 25.7° corresponding to the (10 10) plane [74,75]. Both of these peaks may superimpose on the carbonated HA peak at 2θ = 25.9° resulting in a higher intensity peak.

After 1 month of immersion in 1.5 SBF, the intensity of both HA peaks at 25.9 and 31.8° was found to increase considerably on the EP-Ti and Pi-Ti, which confirms the formation of a thicker HA film and/or better crystallinity of the HA phase (Figure 6b). Intensity of the peaks related to HA on other substrates did not show much change. Further determination of the HA crystal size using the Scherrer formula [76] revealed similar crystallite sizes on all substrates expect for EP-Ti (Table 1), where HA crystallite size was found to double from 3.5 after 14 days to 6.4 nm after 30 days of immersion in 1.5 SBF. While the formation of a very thin film on the Pi-Ti compared to other substrates confirms fast kinetics of CaP film growth on the charged Ti surfaces, the formation of thickest film on the Cl-Ti proves enhanced CaP film growth on negatively charged surfaces in agreement with previous observations [23].

Upon EDX analysis, sharp peaks related to Ca and P were observed confirming the formation of CaP phases on all surfaces. Depending on the thickness of the film, peaks of different intensities related to Ti were also observed (Figure 7a). Furthermore, additional peaks corresponding to C and Mg were detected. The C peak was either due to the surface carbon contamination or carbonate ions in the lattice structure of CaP crystals. Similarly, it is believed that Mg²⁺ ions had partially substituted Ca²⁺ ions in the CaP lattice structure. Presence of similar traces of Mg²⁺ in the EDX and XPS spectra of the CaP film formed at alkali treated Ti surfaces also supports such a hypothesis[E1] [40,77].

Further analysis of the EDX results after 14 days of immersion in 1.5 SBF showed that the Ca/P ratio on Pi-Ti is much lower than that of synthetic HA (i.e., ~1.67) and is related to Ca/P of dicalcium phosphate (Figure 7b). This small Ca/P ratio could be either due to the low film thickness along with the low detection limit of the instrument, or an inadequate surface chemistry of Pi-Ti in enabling Ca²⁺ adsorption. However, on Na-Ti, EP-Ti and Cl-Ti, a Ca/P ratio of 1.50 was found and could be related to a Ca-deficient HA phase.

After 30 days of immersion in 1.5 SBF, the Ca/P ratio was found to be close to that of synthetic HA, i.e., 1.67 (Figure 7b). The Ca/P ratio on all substrates was mainly around 1.55, which corresponds to a Ca-deficient HA. This could be due to a partial substitution of Ca²⁺ by Mg²⁺ ions or substitution of PO₄³⁻ by CO₃²⁻ in the crystal lattice structure, as previously discussed. Furthermore, it can be concluded that once the initial growth regime has been completed and an equilibrium state has been achieved, the Ca/P ratio is relatively the same on all substrates regardless of the early stages of the nucleation process. This would indicate similar growth mechanisms on all substrates after the initial nucleation and growth of the apatite.

With respect to the deposition mechanism of CaP, the behaviour of the Na-Ti substrate in SBF has been explained by Kim et al. [15,51,78]. According to their results, the release of Na⁺ ions from the substrate induces a pH increase at the substrate/SBF interface, which leads to the deprotonation of OH_b groups at the Ti surface. Therefore, the adsorption of Ca²⁺ ions, which are initiators of the nucleation and growth of apatite, is enhanced. However, in the case of EP-Ti and Cl-Ti, the nucleation and growth mechanism of apatite may significantly differ, as the additional Cl⁻ at EP-Ti and Cl-Ti may tend to stay on the surface due to the high concentration of Cl⁻ ions in 1.5 SBF. These adsorbed Cl⁻ may facilitate the electrostatic adsorption of Ca²⁺ and thus trigger the heterogeneous precipitation of a CaP phase. This phenomenon could also be enhanced by the deprotonation of OH_a at the surface of EP-Ti, since a pKa ~2.9 would be expected for the OH_a populating the surface of EP-Ti [79]. However, the role of OH_a in comparison to that of Cl⁻ should be minimized, since the initial deposition of CaP phases at the Pi-Ti surface comprising only OH_a surface groups is slow (Figure 3 and Figure 5). The implantation of Cl⁻ ions at the surface of EP-Ti would therefore explain the good CaP induction ability observed after electropolishing in comparison to conventional chemical treatments.

Other ions in SBF, in particular Mg²⁺ and CO₃²⁻, have also influenced the nucleation and growth of CaP, its composition and possibly its kinetics of growth, as evidenced by EDX and XRD analysis. In fact, Mg²⁺ and CO₃²⁻ ions can compete for adsorption at the surface of Ti and form complexes preventing further CaP precipitation, as highlighted in Figure 8 (mechanisms 1, 2 and 3) [80,81]. For example, traces of Mg²⁺ ions have been shown to reduce the overall rate of CaP crystallization and delay the transformation of amorphous CaP to apatite phases [82]. Furthermore, the formation of magnesium phosphates in solution would reduce the amount of phosphate ions available to form CaP phases [80]. As a consequence, high concentrations of Mg²⁺ ions at the substrate/solution interface would favor the heterogeneous nucleation of poorly crystallized CaP phases and hinder the growth of HA crystals. This inhibitory effect of Mg²⁺ and CO₃²⁻ ions on the growth of HA could explain the small crystal size of HA (~4–6 nm) obtained from the XRD patterns in this work. It could also lead to strong binding of HA to the substrate [81].

In order to determine the chemical formula of the carbonated HA formed on the substrates, further characterizations were carried out by FTIR (Figure 9 and Table 2). The obtained spectra, and the symmetric and asymmetric vibrations at 1,015–1,120 and 920 cm⁻¹ confirmed the apatitic nature of the CaP films formed on all substrates. Furthermore, the vibration observed at 872 cm⁻¹, characteristic of a B-type apatite (i.e., substitution of PO₄³⁻ by CO₃²⁻ ions) [83,84], revealed that the apatite obtained on all substrates was partially of type-B. In fact, the frequency observed for C-O asymmetric stretching at 1,415–1,490 cm⁻¹ is too high to correspond to a B-type apatite and is more similar to those usually observed for A-type apatite (i.e., substitution of OH⁻ by CO₃²⁻ ions) [85]. The absence of the apatite OH stretching peaks at 630 cm⁻¹ also partly confirmed the substitution of A-sites by the carbonate ions [70,86,87,88]. However, the characteristic IR peak for the A-type apatite, i.e., at 1,545 cm⁻¹ [89], was not observed. Therefore, an AB-type apatite had most likely formed on all the Ti substrates in agreement with previous investigations showing the formation of an AB-type apatite at neutral pH (7.4 for the 1.5 SBF used in the present study) [83,85].

Based on these results, a chemical formula for the obtained HA film is suggested considering the fact that trivalent anionic phosphate sites can only be occupied by bivalent hydrogen phosphates or carbonate anions, and monovalent hydroxyl sites can only be substituted by carbonate ions. Furthermore, taking into account that calcium sites can be occupied by other cations and can accept a maximum of two vacant sites [90], the AB-type apatite formed may have the following formula: Ca_10−x−yMg_y(HPO₄)_x−z(CO₃)_z(PO₄)_6−x(OH)_2−x−w(CO₃)_w/2, 0 ≤ x,y,z,w ≤ 1.