ect of Surface Pre-Treatments on the Formation and Degradation Behaviour of a Calcium Phosphate Coating on Pure Magnesium

Calcium phosphate (CaPh) coatings are considered promising surface treatments for Mg-based implants. Normally, the phase conversion process of CaPh compounds occurs during immersion in simulated body fluid (SBF) and allows the easy penetration of a corrosive medium. To solve the issue, pre-treatment is often performed, creating an effective barrier that further improves the corrosion resistance of the underlying Mg. In the present work three pre-treatments including hydrothermal treatment, anodization, and plasma electrolytic oxidation (PEO) were performed on pure Mg prior to CaPh deposition. Results indicated that the composition, morphology, and thickness of the CaPh coatings were strongly influenced by the pre-treatments. Dicalcium phosphate dihydrate (DCPD) was formed on PEO surface, whilst DCPD and hydroxyapatite (HA) were deposited on hydrothermally prepared and anodized surfaces. HA could be deposited on the studied samples during immersion in SBF. The electrochemical impedance spectrum indicated that CaPh coating combined with PEO pre-treatment had the highest corrosion resistance at 120 h due to the superior barrier properties conferred by the PEO layer.


Introduction
Mg and its alloys have gained increasing attention as biomaterials considering their superior biocompatibility, high specific strength and ideal Young's modulus (41)(42)(43)(44)(45) [1][2][3].They are capable of minimising the risk of the stress-shielding effect since they possess a much closer elastic modulus to bone tissue compared to other traditional implants [4,5].Specifically, as bone-fixation devices and stents, Mg-based materials can obviate the need for a second surgery for device removal because of their degradation ability [6].However, the fast degradation rate remains a critical issue that limits their applications in clinic [7][8][9].To solve this issue, protective coatings have been widely utilized to enhance the corrosion performance of Mg and its alloys [10][11][12].Considering the biocompatibility and the osteogenicity, calcium phosphate (CaPh) compounds are promising candidates for bioactive coatings on Mg-based implants [13][14][15].
CaPh compounds can efficiently improve the biocompatibility of Mg and its alloys on account of the similar composition to the inorganic component of bone [16][17][18].Thus surface modifications and treatments, including chemical conversion [19], sol-gel [20], vapour deposition [21], plasma electrolytic oxidation [22], hydrothermal treatment [23] and cathodic electrodeposition [24,25], have been developed to deposit on the surface of Mg-based materials.However, the corrosion properties of a single CaPh coating prepared by these methods are still controversial for long-term implants considering the solubility and structural characteristics [26][27][28].Moreover, defects, cracks and poor adhesion of the CaPh coatings are often inevitable, which allow the penetration of the corrosive medium (simulated body fluids) and, accordingly, corrosion of Mg substrates [29,30].
Reinforcing the corrosion performance by introduction of a pre-treatment layer prior to the deposition of CaPh layer on Mg-based materials has proven efficient in previous studies [31][32][33].The pre-treatments can dramatically increase the corrosion resistance of the substrate and the full-coated samples, and can also provide long-term biodegradability for the implant [34][35][36].For instance, Liu et al. [37] prepared a CaPh coating on a PEO-coated Mg substrate by the chemical conversion method.The results showed that the PEO layer significantly improved the corrosion resistance of the system, and revealed a self-healing ability in simulated body fluid.In addition, the pre-treatments are closely related with the precipitation process of the CaPh coating.Hiromoto et al. [38] reported that the precipitation of CaPh coating was influenced by the morphologies of an anodizing pre-treatment.The absence of local corrosion on the CaPh/anodizing-coated sample during a 40-h immersion in SBF indicated its higher corrosion resistance.Furthermore, the pre-treatments can be applied to improve the antibacterial activity and cytocompatibility together with the CaPh coating [39].In a recent study [40], PEO/hydroxyapatite composite coatings were prepared on a Mg-Ca alloy via PEO coupled with electrophoretic deposition.Such a duplex coating with high corrosion resistance, antibacterial activity and cytocompatibility was suggested for orthopaedic surgical applications.For these reasons, surface pre-treatment of Mg and its alloys is an appropriate solution for the application of CaPh coatings.Nevertheless, although some studies have been reported, there are still open questions.Most of these studies focus on the corrosion resistance, degradable property and functionality of the CaPh coatings [34,41,42].Research into the influence of pre-treatment layer on the formation mechanism of CaPh coating is scarce, and the corrosion behaviour of the hybrid CaPh/pre-treatment-coated samples in SBF still needs to be investigated.
In this study, three pre-treatments were carried out on the surface of pure Mg prior to cathodic electrodeposition (CED) of CaPh.The formation mechanism of the CaPh coating was studied, the phase evolution and corrosion behaviour of the CaPh-coated samples in simulated body fluid were investigated, and the influence of the pre-treatments on the morphology, composition and performance of the duplex coating was discussed.

Materials and Pre-Treatments
Commercial pure magnesium stripes with dimensions of 20 × 20 × 3 mm 3 were used as substrates in this study.The chemical composition of Mg was determined by Spark OES (Spark analyser M9, SPECTRO Ametek, Harmbug, Germany), as shown in Table 1.Prior to the pre-treatment, the specimens were abraded using SiC paper from 800 to 2500 grade, then washed in distilled water and dried in air.Three pre-treatments were prepared on pure magnesium.The electrolyte composition and parameters of the pre-treatments are shown in Table 2.
Hydrothermal treatment was the first method.The Mg samples were placed in a climate chamber under conditions of 90 • C and 90% humidity.The samples were treated for a period of 24 h.Anodization treatment was carried out with a constant current density of 0.1 A/cm 2 for 60 min at 40 • C by a DC power supply.The specimen served as the anode and a graphite sheet was employed as the cathode.An aqueous electrolyte containing 10 g/L sodium carbonate (Na 2 CO 3 ), 40 g/L sodium hydroxide (NaOH) and 50 ml/L ethanediol (C 2 H 6 O 2 ) was used.
Finally, the PEO process was carried out with a pulsed DC power supply (t on :t off = 0.5 ms:4.5 ms and 100 Hz) under a constant voltage of 450 V for 10 min.A stainless-steel tube was used as the cathode, while the specimens were the anode.The electrolyte contained 20 g/L Na 3 PO 4 and 1 g/L KOH; the temperature of the PEO electrolyte was controlled at 20 ± 1 • C using a cooling system.To achieve a uniform surface state, both AT and PEO treatments were performed under continuous agitation with a magnetic stirrer rotating at 150 r/min.

Deposition of the CaPh Coating
The CaPh coatings were deposited on the pre-treated samples by direct current (DC) cathodic electroassisted deposition method under 47 ± 0.5 • C. A two-electrode cell was used for the preparation; a graphite sheet was selected as the anode while the pre-treated samples were utilized as the cathode.A 600 mL aqueous electrolyte containing 19.7 g/L Ca(NO 3 ) 2 and 6.8 g/L KH 2 PO 4 with a adjusted pH value of 4.6 was used as the deposition solution.CaPh is soluble at pH 4.6 and exists in a solution that is not soluble at higher pH during cathodic deposition.The deposition duration was 90 min.The current-time curves during the electroassisted deposition period were monitored by Sigma soft 6000 software.
The pre-treated samples are named HT (hydrothermal treatment), AT (anodization treatment) and PEO (plasma electrolytic oxidation), respectively, throughout the paper.The samples are referred to as HT-C, AT-C and PEO-C after CaPh coating, respectively.

Characterization
The top-view and cross-sectional morphologies of the pre-treated and CaPh-coated samples were characterized using a Tescan Vega 3 SB scanning electron microscope (SEM, LabWrench, Waukesha, WI, USA).The phase composition was measured by an X-ray diffraction (XRD with Cu Kα radiation, Bruker, Karlsruhe, Germany) with a scanning rate of 0.01 • /s and scanning range of 10 • -50 • .Grazing incidence XRD was used under 3 • of incidence angle at 40 kV and 40 mA.

Electrochemical Tests
The electrochemical measurements were carried out with a Gamry Interface 3000 potentiostat (Gamry, Shanghai, China) in a simulated body fluid (SBF) at 37 ± 0.5 • C. The composition of SBF was 8.035 g/L NaCl, 0.355 g/L NaHCO 3 , 0.225 g/L KCl, 0.231 g/L K 2 HPO 4 •3H 2 O, 0.311 g/L MgCl•6H 2 O, 39 mL/L 1 mol/L HCl, 0.292 g/L CaCl 2 , 0.072 g/L Na 2 SO 4 and 6.118 g/L Tris (trimethylolamine); the pH was adjusted to 7.4 using 1 mol/L HCl.A typical three-electrode system was used in this study with a saturated Ag/AgCl reference and a platinum counter electrode.The pre-treated samples served as the working electrodes, with an exposed area of 1 cm 2 .Potentiodynamic polarization curves were recorded from -150 mV relative to open circuit potential (OCP) to a terminal current density of 2 mA/cm 2 with a scan rate of 0.5 mV/s after a 30-min stabilisation in SBF.Electrochemical impedance spectroscopy (EIS) measurements were carried out after OCP measurement at regular intervals; the scan amplitude ranged from 100 kHz to 10 mHz for five days by applying 10 mV RMS relative to OCP.The tests were performed under continuous agitation with a magnetic stirrer rotating at 150 r/min.EIS results were analysed by fitting the data with Zsimpwin software 3.20.Each test was carried out three times to ensure the reproducibility.

Immersion Test
Immersion tests were performed in SBF at 37 ± 0.5 • C with an immersion period of 120 h.The immersion tests were also done under the same stirring conditions described above.The SBF was refreshed every two days.The morphologies and phase compositions after 120 h immersion were then analysed.

Phase Composition of the Pre-Treatments and CaPh-Coated Samples
The XRD patterns of the pre-treated samples are shown in Figure 1a.The diffraction peaks of Mg from the substrate can be found in all the cases.Mg(OH) 2 was the main component of the HT and AT samples, while MgO and Mg 3 (PO 4 ) 2 were detected on the surface of the PEO sample.The lower temperature and moderate reaction on the surface of the HT and AT samples were responsible for the formation of Mg(OH) 2 .For PEO treatment, MgO always formed as a result of a series of complex interface reactions as well as electrical, mechanical and thermal theories [43].In addition, the corresponding intensities of Mg(OH) 2 on the surface of AT sample were significantly higher than those of the HT sample, suggesting that the HT pre-treated layer was probably thinner.The corresponding intensities of the Mg substrate were lower than those of the rest, suggesting that the PEO pre-treated layer was probably thicker.samples.Correspondingly, the intensities of the HA peaks were lower than that of DCPD, illustrating that the contents of HA were possibly lower than that of DCPD in these coatings.This indicated that the pre-treatment layers strongly influenced the phase composition of CaPh coatings.

Morphologies of the Pre-Treatments and CaPh-Coated Samples
Diverse types of surface morphologies are observable on the pre-treated and CaPh-coated Mg surfaces, as shown in Figure 2. Paralleled scratches introduced during the grinding process remained on the surface of the HT sample, as seen in images depicted at lower and higher magnification (Figure 2a).This indicated that a thin pre-treated layer exists on the substrate surface.In addition, a uniform, interlacing feather-like structure appeared in the high-magnification morphology of the HT sample.This is a typical appearance of hydroxides on magnesium and its alloy surfaces that can be formed in a humid ambient environment [44].Charged by a DC power supply, a cloudy surface morphology featuring an island-like structure was identified on AT-treated Mg (see Figure 2b).However, a completely different surface morphology was seen on the surface of PEO-treated Mg (Figure 2c), featuring discontinuous cracks and pores on the nano or micron scale.Such a surface resulted from the consecutive breakdown of the dielectric oxide layer by the discharge; the pores and cracks were introduced due to the quenching effect by the electrolyte [12].As a result, a sintered and porous surface morphology was achieved during PEO treatment.
The surface morphologies of the pre-treated surfaces were modified by the CaPh coating, as shown in Figure 2d-f.Plate-like and flower-like microstructures could be identified for HT-C (Figure 2d) and AT-C (Figure 2e).However, a uniform, large plate-like structure (about 100 µm) appeared on the PEO-C surface (Figure 2f), suggesting a single CaPh phase was deposited.The size difference of the plate-like structures also indicated that the nucleation and growth conditions were different on the three pre-treated samples; our previous study also found that the CaPh phase tended to grow along a fixed direction on a PEO-treated Mg substrate [37].The cross section morphologies of the pre-treated and CaPh-coated samples are shown in Figure 3.The thinnest pre-treatment layer was formed on the HT sample surface (about 1-3 µm).The average thicknesses of the AT and PEO layer were about 25 µm (Figure 3b) and 30 µm (Figure 3c), respectively.For the PEO coating, two layers could be clearly observed and assigned to the porous layer (outer layer) and compact layer (inner layer).After CaPh deposition, new coatings were formed on the pre-treated layers.For the HT-C sample, the average thickness of the CaPh coating was about 5 µm, which was thinner than that on the AT-C (10-12 µm) and PEO-C samples (10-15 µm).In addition, the porous structure could hardly be seen, implying that the porous layer of PEO was filled with the CaPh phase after deposition, as clearly seen in Figure 3f.This proved that the CaPh phase could nucleate at the open pores and seal the surface cracks of the PEO layer.Moreover, the thickness of AT and PEO layers decreased slightly (to about 20 µm and 18-25 µm, respectively) after CaPh deposition, indicating that the pre-treatment layers could be slightly dissolved in acidic electrolyte.As seen in Figures 1 and 2, the pre-treatment has a significant influence on the morphologies and phase composition of the CaPh coatings.Various pre-treatments are performed on the Mg substrate, which leads to differences in the chemical environment prior to the CaPh deposition.The thickness, morphologies and composition of the pre-treatment layers play vital roles in the formation of CaPh coatings.
As the start of the electrodeposition, abundant ions of Ca 2+ and H 2 PO 4 − exist in the solution.
For HT and AT samples, Mg(OH) 2 could easily hydrolyse in an electrolyte with a pH value less than 10.5; the local increase of pH occurs partially to compensate for the acidity at the interface and releases OH − (Equation ( 1)).
As the deposition process goes on, the influent electrons would cause a reduction in H 2 O.More OH − is released and hydrogen bubbles appear (Equation ( 5)), resulting in an increase in pH of the deposition solution.The elevated pH and influent electrons make HPO 4 2− ions convert into PO 4 3− ions (Equations ( 6) and ( 7)).As reported, HA tends to be formed in an alkaline environment [45].
HA is deposited on the surface of HT and AT samples according to Equation ( 8) following DCPD.
The balance between DCPD and HA is most probably controlled by the local pH at the interface.
For the PEO sample, the pre-treatment layer is mainly composed of MgO.Thus a lower amount of OH − is produced since the hydrolysis of MgO is more difficult than Mg(OH) 2 [46] ions (Equation ( 4)).The reaction in Equation ( 8) is restrained because of the absence of enough OH − and the difficult transformation of HPO 4 2− ions to PO 4 3− ions.
With the quick growth of CaPh phases, the surface resistance should increase and the current density should decrease at the same time.However, a different trend is observed (Figure 3).The thin Mg(OH) 2 coating of HT sample dissolves quickly in acidic electrolyte, and the electrolyte in contact with the bare metal substrate surface causes an increase in current density.For the AT and PEO samples, the thicker pre-treatment layers prevent the free permeation of the electrolyte, and the current density decreases in the initial period, which is related to the formation of CaPh phases in defects sealing the pre-treatment layers.In the following deposition period, for all samples (after 15 min), the pre-treatment layers are completely covered with CaPh, and the further dissolution of pre-treatment layer is prevented.The nucleation and growth rate of CaPh coatings should slow down, and the current density decreases slightly.

Potentiodynamic Polarization Curves
Potentiodynamic polarization curves of the samples with and without CaPh coatings are presented in Figure 6.The Mg substrate was also measured as a reference.The corresponding corrosion potential (E corr ) and corrosion current density (i corr ), determined by extrapolation of the polarization curves, are listed in Table 3.After pre-treatment, E corr shifted to positive values and i corr decreased compared to the Mg substrate, suggesting that the corrosion reactions on the surfaces were suppressed by the pre-treatment layers, as shown in Figure 6a.The HT sample exhibited the highest i corr value (9.85 × 10 −5 A/cm 2 ), indicating that the corrosion resistance of the HT treated sample was the lowest.The AT sample possessed higher i corr (4.95 × 10 −6 A/cm 2 ) than that of the PEO sample (1.25 × 10 −6 A/cm 2 ), though the thickness of the pre-treated layers was almost the same.The cracks and loose structure of the AT sample (Figure 3b) might be responsible for the poorer corrosion resistance.More positive E corr and lower i corr values of the coated samples indicated that the corrosion resistance of the Mg substrate was further enhanced (see Figure 6a and Table 3).The HT-C sample also possessed the worst corrosion resistance compared to the rest of the coated samples due to the lowest E corr and the highest i corr .The AT-C and PEO-C samples showed similar E corr (−1.51 V) and i corr values (9.65 × 10 −7 A/cm 2 and 5.09 × 10 −7 A/cm 2 ); however, a more extended region with retarded dissolution existed in the anodic branch of PEO-C sample, indicating better corrosion resistance.
Figure 6b shows the potentiodynamic polarization curves of the CaPh-coated samples after a 120-h immersion.Comparing the as-deposited samples, the increasing i corr and decreasing E corr indicated a decrease in the corrosion resistance.The HT-C sample showed the most negative corrosion potential and the largest corrosion current density, with values of −1.734 V and 1.58 × 10 −5 A/cm 2 , respectively.The corrosion performance of AT-C sample dropped significantly after a 120-h immersion since the i corr of the AT-C sample increased to 6.2 × 10 −6 A/cm 2 , revealing poor resistance during long-term immersion.The PEO-C sample exhibited minimum corrosion current density of 6.92 × 10 −7 A/cm 2 , which demonstrates the overall best corrosion resistance.Moreover, extended regions could be observed in the anodic branch of all the samples indicated the retarded anodic dissolution during the immersion.

OCP
OCP was measured as a function of immersion time during immersion in SBF, as shown in Figure 7.The OCP of the pre-treated samples are presented in Figure 7a.Various OCP values at 5 min indicated the different physical barriers of pre-treatment layers in the initial period.The OCP values of the HT sample increased over 6 h, while a declining trend was exhibited for the AT and PEO samples.For the HT sample, the thinnest pre-treated layer existed on its surface (Figure 3a).In this case, the corrosion of the Mg substrate should lead to OCP decreasing.However, the formation of corrosion products on the surface resulting from the quick corrosion of substrate would lead to OCP increasing.For the AT and PEO samples, the decrease of OCP values might be related to the infiltration of the SBF into the pre-treated layers due to cracks and pores on their surfaces (Figure 3b,c).The potential of all pre-treated samples fluctuated but had an increasing tendency during the following immersion period, and finally approached −1.690 V.After CaPh deposition, as shown in Figure 7b, the OCPs shifted to more positive values compared to the pre-treated samples.OCPs of HT-C and AT-C samples decreased first, and then increased continuously during the following immersion, approaching final values of −1.66 and −1.63 V, respectively.For the PEO-C sample, the OCP decreased slightly in the first 6 h and quite stable OCP values (around −1.60 V) were recorded from 6 to 120 h, indicating relatively stable properties during the immersion.

EIS Spectra of the Pre-Treated Samples
The EIS spectra of the pre-treated samples were consecutively recorded in SBF, as shown in Figure 8. Two well-defined time constants in medium-(10 1 -10 3 Hz) and low-frequency (10 −1 -10 0 Hz) ranges were observable on the HT and AT samples (Figure 8a,b).Time constants at medium and low frequency were related to the pre-treatment layers and the electrochemical double layer at the exposed metal/electrolyte interface, respectively.The HT sample showed a low impedance modulus after 5 min of immersion.Subsequently, it decreased after 6 h, indicating that SBF penetrated into the pre-treated layer.Afterwards, the modulus increased gradually with the increase in immersion time.This is related to the blocking effect due to the formation of corrosion products on the surface.For the AT sample, a relatively high modulus value appeared in the whole frequency range after 5 min of immersion.This indicates that the AT layer has a relatively high barrier property.The impedance reduced quickly after 6 h and gradually increased later, because of the deposition of corrosion products.However, for the PEO samples, the asymmetry of the mid-frequency phase angle peak suggests the appearance of another time constant at higher frequencies in the period before 24 h of immersion.Three time constants distributed at high, medium, and low frequency were observed in the plots, which reflected the electrochemical properties of the outer layer, inner layer and the double electric layer, as shown in Figure 8c.During the immersion process, the impedance modulus showed a relatively slight change for the whole frequency range.This indicates a positive stabilization effect of the PEO layer on the barrier properties of the oxide layer at the surface.
The physical models combined with equivalent circuits (EQC) for EIS of the pre-treatments are shown in Figure 9.In this work, we use the constant phase element (Q) instead of a capacitor in order to account for the dispersion effects.In the circuits, R s represented the solution resistance, Q p and R p represented the capacitance and resistance of the HT and AT layers, Q o , R o , Q i and R i stand for the capacitance and resistance of the outer and inner layer of PEO coating, respectively.Q dl and R ct represented the electric double layer capacitance and charge transfer resistance, respectively.

EIS Spectra of the CaPh-Coated Samples
With the additional CaPh coatings deposited on the pre-treated surfaces, the EIS spectra exhibited distinctly different evolution behaviour, as shown in Figure 10.The EIS plots of the HT-C and AT-C samples showed three time constants originating from high, medium and low frequency during the whole immersion period.Time constants in high and medium frequency could be assigned to the CaPh coatings and the pre-treatment layers, respectively.Another time constant representing the electrochemical double layer at the metal/electrolyte interface appeared in low frequency (10 −1 -10 0 Hz).For the PEO-C sample, the spectra displayed three time constants during the whole immersion period as well.The time constant at high frequency (10 3 -10 5 Hz) can be assigned to the CaPh coating, the one at middle frequencies (10 1 -10 2 Hz) originated from the sealed outer layer of PEO, and the relaxation process at low frequency (10 −1 -10 1 Hz) was related to the electrochemical activities at the electrolyte/inner layer of PEO.
The EQC model in Figure 11 was used to fit the EIS plots of the full coating systems.Figure 11a represents the HT-C and AT-C samples from 0 to 120 h, and the EQC of PEO-C from 0 to 120 h is shown in Figure 11b.Q c and R c are the capacitance and resistance of the CaPh coating.Q cp and R cp represent the capacitance and resistance of the pre-treatments.Q dl is the capacitance of the electrochemical double layer and R ct is the charge transfer resistance of the HT-C and AT-C samples.Q so , R so , Q i and R i stand for the capacitance and resistance of the CaPh sealed outer layer and the compact inner layer of PEO-C sample, respectively.The electrochemical processes at the interface should be manifested at low frequencies as a double-layer capacitance and charge transfer resistance.However, considering the relatively low values of R ct and high Q dl , their response is not visible in the respective spectra in the used frequency range.Therefore, the fitting of impedance data was performed using modified EQC (Figure 11b), which excluded the electrochemical response components.With the presence of the CaPh coatings, the R cp and R sum values for all the pre-treated samples show an increase, as demonstrated in Figure 12d,e.For the PEO-C sample, the R sum values increased by more than one order of magnitude, proving that the corrosion resistance of the systems was further improved.Furthermore, the highest R c and R cp values appeared in the PEO-C sample during the whole immersion process, indicating that it was the best combination of PEO and CaPh coatings (see Figure 12c).The R c , R p and R sum values stayed almost stable after 24 h, indicating a better stability of the corrosion protection for the substrate.After 120 h of immersion, the R c and R cp of the HT-C and AT-C samples reached almost the same value.In addition, R sum values of the HT-C and AT-C samples stayed the same after 24 h of immersion.

Immersion Tests
XRD patterns of all the coated samples in SBF after 120 h of immersion are depicted in Figure 13.For the HT-C and AT-C surfaces, the diffraction peaks of DCPD disappeared and distinct HA diffraction peaks appeared on the surfaces.The diffraction peaks of Mg(OH) 2 that appeared on the HT-C and AT-C surfaces might have originated with the corrosion products.For PEO-C sample, DCPD peaks could still be detected after a 120-h immersion, and yet the corresponding intensities of DCPD peaks were low compared with Figure 1, while HA diffraction peaks were also observed.This means that the dissolution of DCPD improved the precipitation of HA.The morphologies of the CaPh-coated samples after a 120-h immersion in SBF are shown in Figure 14.The large plate-like structure of DCPD (in Figure 2) disappeared, and new structures were found on the surfaces of each sample.Loose and porous morphologies with cracks appeared on the HT-C (Figure 14a) and AT-C (Figure 14b) surfaces, which might be related to the mineralization process of the CaPh phases in SBF.In addition, the entire dissolution of DCPD on the surfaces is responsible for the newly formed structures.A compact surface morphology could be found on the PEO-C sample, as shown in Figure 14c.The dense CaPh coating could provide long-term protection for the substrate.

Corrosion Behaviour of Pre-Treated and CaPh-Coated Samples in SBF
The top CaPh coatings offer the resistance of the initial immersion period [47].However, as mentioned above, the corrosion resistance of a single CaPh coating is inadequate for long-term immersion in SBF due to the loose structure and the dissolution of CaPh compounds.The SBF penetrates into the coating and the barrier effect of the coating will be weakened.Pre-treatments can effectively improve the corrosion resistance of the Mg substrate.An appropriate pre-treatment on the surface of magnesium prior to the application of the CaPh coatings can also stabilize the magnesium substrate/CaPh coating interface.Our previous study mentions that surface pre-treatment prior to a hydrophobic calcium stearate coating affects the thickness and surface morphology but does not alter the phase composition of the coating [48].However, in this study, we find that the pre-treatments not only influence the thickness and morphology of the CaPh coatings, but also the phase composition along with the adhesion between the CaPh coating and the magnesium substrate, thus playing an important role in the corrosion resistance of the coated samples.
Although the R c values are low for all the coated samples, the R cp values have been significantly improved by the presence of the CaPh coatings.The relatively low R c values of the HT-C and AT-C samples indicate the poor combination of the CaPh coating and pre-treatment layer.For the HT sample, the R cp presents the lowest value during the whole immersion process.Both the CaPh coating and the pre-treatment layer are penetrated easily by the medium due to the thickness of the surface (Figure 3d).Thus, the corrosion resistance of the sample is the lowest.As to the AT sample, the thicker CaPh coating and pre-treatment layer result in higher R c and R cp at the initial stage of the immersion process.However, the dissolution of DCPD leads to a quick decrease of R c , and the coating does not provide much protection for the pre-treatment layer.In addition, the presence of cracks on the pre-treatment layer is responsible for the penetration of the SBF into substrate.The corrosion resistance of the AT-C sample is also low.
However, the PEO-C sample exhibits the highest resistance during the whole immersion process, showing the best corrosion protection properties.The highest R c values indicate the best combination of PEO/CaPh coating.The thickest DCPD coating inhibits the penetration of the corrosive medium at first.With the increase in immersion time, DCPD would dissolve continuously.However, the sealed PEO layer exhibits excellent barrier properties and allows sufficient time for the precipitation of HA, so a dense CaPh coating is formed on the surface (Figure 14c).Meanwhile, the adhesion strength between the CaPh and PEO coatings is enhanced because of the interlocking effect since the porous structures of PEO layer provides more sites for the deposition of the CaPh coating [49,50].However, R c is still one order of magnitude lower than that of the R cp value, so the main corrosion resistance of the PEO-C sample is provided by the pre-treatment layer.In a previous study, CaPh coating coupled with PEO treatment efficiently inhibited the local corrosion of Mg substrate, especially the pitting corrosion [51,52].In this study, the CaPh phase nucleates at the open pores and defects of the outer PEO layer, as seen in Figure 3f.The sealed outer layer further reinforces the corrosion resistance of the system.Moreover, the inner (compact) layer of PEO efficiently inhibits the penetration of aggressive ions and the release of Mg, so the highest R cp is presented.Owing to the overall factors, a PEO/CaPh hybrid coating exhibits the best corrosion resistance.

Conclusions
In this study, three pre-treatments were performed on a pure Mg surface prior to the deposition of a calcium phosphate (CaPh) coating, and the following conclusions could be drawn:

•
When a surface treatment is performed on pure magnesium, different surface treatment leads to different structures on the magnesium surface.After anodizing and hydrothermal treatment, Mg(OH) 2 is formed on the magnesium surface, and MgO and Mg 3 (PO 4 ) 2 are the phase compositions of the PEO layer.

•
The surface pre-treatment of pure magnesium prior to the CaPh coating influences the thickness, morphologies and phase composition as well as the formation mechanism of CaPh coatings due to the different morphologies and composition of the pre-treated layers.

•
DCPD and HA are together deposited on the HT-C and AT-C samples, while only DCPD is detected on the PEO-C sample.During the SBF immersion, DCPD on the HT-C and AT-C samples disappears, whereas DCPD is present on the PEO-C sample all along.In any case, the coating surface can improve HA's ability to precipitate in SBF.The pre-treatments play an important role in the corrosion resistance of the duplex coating systems.The best corrosion resistance is obtained for the PEO/CaPh combination because of the superior barrier properties of the PEO layer.

Figure
Figure 1b displays the phase composition of the CaPh-coated samples.Apart from the diffraction peaks originated from the pre-treated substrates (Mg and MgO), the only DCPD peaks were detected on the PEO-C sample.DCPD and HA were deposited together on the surfaces of the HT-C and AT-C

3. 3 .
Figure4shows the current density-time curves of the pre-treated samples monitored during the electrodeposition process.The current density of HT sample rapidly increased from 5.2 to 8.5 mA/cm 2 in the initial period (0 to 14 min), then gradually decreased to 7 mA/cm 2 within 50 min of deposition and stayed almost constant in the following period.The curves of the AT and PEO samples exhibited a different trend.The current densities quickly decreased from 4.8 to 3.0 (AT sample) and from 3.6 to 2.3 mA/cm 2 (PEO sample) in the first 20 min, finally approaching a constant value with values of 2.4 (AT sample) and 1.8 mA/cm 2 (PEO sample), respectively.It should be noted that the constant values of the current densities were different for the CaPh-coated samples, suggesting a difference in the cathode resistance to the electrolyte of the hybrid coatings.

Figure 4 .
Figure 4. Current density-t curves recorded during the electrodeposition process of different pre-treated substrates in an aqueous mixed solution (0.0835 mol/L Ca(NO 3 ) 2 and 0.05 mol/L KH 2 PO 4 ).A DC voltage of 3.5 V was applied, while different pre-treatments ((a) HT, (b) AT, (c) PEO) were used as cathode and a graphite plate was used as anode.
H 2 PO 4 − , combining with OH − , formed HPO 4 2− (Equation (2)).Meanwhile, H 2 PO 4 − could also convert to HPO 4 2− under a DC applied voltage, according to Equation (3).DCPD is formed preferentially as a result of HPO 4 2− combining with Ca 2+ , according to Equation (4).To prove the preferential deposition of DCPD, XRD patterns of the HT-C and AT-C samples under a short deposition period (10 min) are recorded, as shown in Figure 5. XRD patterns indicate that, besides the peaks originating from the Mg and pre-treatment layers, only DCPD diffraction peaks appear during a 10-min deposition.

Figure 8 .
Figure 8. Bode plot of the pre-treated samples in SBF at 37 ± 0.5 • C: (a) HT sample; (b) AT sample; (c) PEO sample.Symbols represent experimental data and solid lines are fitted data.

Figure 9 .
Figure 9. Equivalent circuit models of the pre-treated samples: (a) HT and AT samples; (b) 5 min to 24 h for PEO sample; (c) 48 to 72 h for PEO sample.

Figure 10 .
Figure 10.Bode plot of the CaPh-coated samples in SBF at 37 ± 0.5 • C: (a) HT-C sample; (b) AT-C sample; (c) PEO-C sample.Symbols represent experimental data and solid lines are fitted data.

Figure 11 .
Figure 11.Equivalent circuit models of the CaPh-coated samples: (a) 5 min to 120 h f for HT-C and AT-C samples; (b) 5 min to 120 h for PEO-C sample.

3. 4 . 5 .
Evolution of the EIS Fitting Parameters R p , R c , R cp and the sum of the resistances R sum for each sample obtained from the EIS results are shown in Figure12.R p represented the barrier property of pre-treatments to the system.Herein, R p values of PEO and PEO-C samples are the sum of R o (or R so ) and R i .The R p and R sum of the pre-treated samples are shown in Figure12a,b, respectively.The R p and R sum values of HT and AT samples exhibited a similar trend, that is, decreasing for a short period (6 h) and increasing gradually in the following period (6-72 h).The penetration of the corrosive medium resulted in the decreasing R p and R sum values in the first 6 h and the deposition of corrosion products was responsible for the increasing trend.It can be noted that the highest R p and R sum values appeared in the PEO sample, indicating superior corrosion resistance.

Figure 12 .
Figure 12.Fitting results of the resistance for pre-treated and CaPh-coated samples.(a,b) represent for the R p and R sum of the pretreated samples, respectively; (c-e) represent for the R c , R cp and R sum of the CaPh coated samples.

Figure 13 .
Figure 13.X-ray diffraction patterns of the CaPh-coated samples after 120 h immersion in SBF at 37 ± 0.5 • C.

Table 1 .
Chemical composition of Mg determined by spark OES.

Table 2 .
Details of the pre-treatments on pure Mg.
. In this case, less H 2 PO 4 − is converted into HPO 4 3− because the reaction of Equation (2) is slow, though Equation (3) occurs.Only DCPD is formed on the PEO surface by combining Ca 2+ ions and HPO 4

Table 3 .
Electrochemical data of the samples as fitted by extrapolation from potentiodynamic curves.