Investigation on Inﬂuencing Mechanism of Processing Parameters on Corrosion Resistance and Zinc Content of Anodic Coatings Developed on Magnesium Alloys in Near-Neutral Solutions

: In near-neutral solutions, the effects of NH 4 HF 2 , H 3 PO 4 , phytic acid (IP6), and EDTA-ZnNa 2 concentration on corrosion resistance and the Zn amount of micro-arc oxidation (MAO) coatings were revealed by an orthogonal experiment. The inﬂuencing order of four factors on coating corrosion resistance is EDTA-ZnNa 2 > NH 4 HF 2 > IP6 > H 3 PO 4 , while the sequence on the Zn amount is ranked as EDTA-ZnNa 2 > NH 4 HF 2 > H 3 PO 4 > IP6. The fabricated Zn-containing coatings exhibit excellent corrosion resistance, and their i corr values are two orders of magnitude lower than that of the WE43 substrate, while the highest Zn amount achieves 4.12 wt.%. P and F compete to take part in coating formation, and Zn ions enter into anodic coatings by diffusion. Coating corrosion resistance is jointly determined by surface characteristics, which will provide the important theoretical foundation for fabricating Zn-containing coating with high corrosion resistance.


Introduction
Compared with traditional bone-repair materials such as titanium alloys, stainless steel, and cobalt-based alloys [1], magnesium (Mg) alloys have been thought to be excellent materials for metallic implants due to their similar density to natural bone and good biodegradability in bioenvironment [2]. However, Mg alloys belong to very active metals with a too-fast corrosion rate in vivo, which may cause surgery failure [3]. Therefore, it is necessary to improve corrosion resistance of Mg alloys to meet clinical applications [4].
As an important trace element, Zn distributes in all tissues and organs and participates in many enzymatic reactions [17]. It is reported that Zn plays an important role in bone metabolism, cell metabolism and apoptosis, gene expression, and neurotransmission [18]. Zn is also closely related with human intelligence, memory, and neurological disorders such as cerebral ischemia and Alzheimer's disease. Moreover, zinc has a positive effect on improving immunity [19]. In addition to directly promote the development and function of immune organs, Zn can also directly act on certain bacteria and viruses and possesses strong antibacterial properties [20]. Therefore, it is important to fabricate Zn-containing coatings on medical Mg alloys.
For Mg alloys, MAO coatings are generally fabricated in strong alkaline solutions containing silicate [14], phosphate [21], and natural organic phytic acid (C 6 H 6 (H 2 PO 4 ) 6 , abbreviated as IP6 [22]. At present, Zn-doped MAO coatings on Mg alloys are rarely reported using a one-step method [23], and the reported coatings were developed in strong alkaline solutions [14][15][16]. Compared with alkaline solutions, near-neutral solutions cannot only achieve a higher final voltage and therefore better surface property, but also prolong solution service lifetime [22]. In this study, ammonium bifluoride (NH 4 HF 2 ), phosphoric acid (H 3 PO 4 ), IP6, and ethylene diamine tetraacetic acid zinc disodium salt ((NaOOCCH 2 ) 2 -N=CH 2 CH 2 =N(CH 2 COO) 2 Zn, abbreviated as EDTA-ZnNa 2 ) were selected as MAO electrolytes to develop ceramic coatings on WE43 substrate in a near-neutral solution composed of 360 g/L hexamethylenetetramine (HMTA). The influencing regularity of electrolyte composition and concentration on coating property were clarified by using an orthogonal experiment. More specifically, NH 4 HF 2 was used as a corrosive inhibitor of Mg alloys, while EDTA-ZnNa 2 was used as a Zn-containing electrolyte. H 3 PO 4 and IP6 were selected as inorganic and organic P-containing electrolytes, respectively. The prepared anodic coatings were measured by SEM (ΣIGMA, Zeiss Sigma, Oberkochen, Germany), XRD (Shimadzu XRD-6100, Kyoto, Japan), XPS (ESCALAB250, Thermo VG, Waltham, MA, USA), and potentiodynamic polarization curves to clarify the affecting regularity of electrolyte composition and concentration on coating corrosion resistance and the Zn content.

Materials and Coating Preparation
Extruded WE43 provided from Suzhou Chuan Mao Metal Materials Co., LTD (Suzhou, China), contained 4.01 wt.% Y, 0.47 wt.% Zr, 1.72 wt.% Gd, 2.35 wt.% Nd, 0.003 wt.% Cu, 0.004 wt.% Ni, 0.0003 wt.% Fe, and balanced Mg. Before MAO treatment, all samples were first ground by using SiC waterproof abrasive paper and then cleaned successively with distilled water and ethyl alcohol. Finally, all samples were dried by using a hairdryer. The selected base solution in this study was composed of 360 g/L HMTA (Ph = 6.30). According to Table 1, the effects of NH 4 HF 2 , H 3 PO 4 , IP6, and EDTA-ZnNa 2 concentration on corrosion resistance and the Zn amount were investigated via an orthogonal experiment with four factors and three levels. According to the processing conditions given in Table 1, nine MAO samples, namely No. 1 to No. 9, were prepared by using a homemade MAO-5D power supply (Chengdu Tongchuang New Material Surface Engineering and Technology Center, Chengdu, China). A WE43 sample was used as the anode, while a stainless-steel barrel with the MAO solution was connected with the cathode. By adopting a constant current control mode, the selected current density, duty cycle, frequency, and treating time were 60 mA/cm 2 , 35%, 2000 Hz, and 3 min, respectively. The solution pH was determined by using a PHS-3C digital pH meter (Shanghai Yoke Instrument Co., Ltd., Shanghai, China). The MAO solution was continually stirred to ensure its temperature under 40 • C.

Microstructural Characterization
Surface and cross-sectional morphologies were measured by a field emission scanning electron microscope (FE-SEM). Coating compositions were detected by an energydispersive X-ray spectrometer (EDS, OxfordINCA Energy, Oxford, UK) attached to an FE-SEM. In the 2θ range of 10 • to 80 • , at a scanning speed of 4 • /min, coating phase structures were analyzed by an X-ray diffractometer, using a Cu Kα radiation. Chemical states of anodic coatings were detected by an X-ray photoelectron spectroscopy (XPS), with an Al Kα anode (λ = 1486.6 eV). The porosity of MAO coatings was detected by Image-Pro Plus 6.0 imaging software.

Electrochemical Test
In order to evaluate the corrosion resistance of the WE43 substrate and MAO-treated samples, potentiodynamic polarization curves were measured by an electrochemical workstation (Reference 600+, Gamry Instruments, Lafayette, LA, USA). All measurements were performed at 37 • C in Hanks' solutions, as reported by others [22]. A typical tree-electrode cell was used with a sample as the working electrode, while a saturated calomel electrode (SCE) (Tianjin Aida Hengsheng Technology Development Co., Ltd, Tianjin, China) and a platinum electrode were separately served as the reference electrode and the counter electrode, respectively. The initial delay and scanning rate were set as 300 s and 1 mV/s, respectively. Potentiodynamic polarization curves were measured from −0.25 to 0.5 V with respect to the open circuit potential (OCP), and the Tafel extrapolation method was used to derive the corresponding electrochemical parameters, including corrosion potential (E corr ), corrosion current density (i corr ), and anodic/cathodic Tafel slopes [24,25]. In order to ensure result repeatability, five parallel samples were measured to obtain the final results.

Surface Morphology and Composition
As shown in Figure 1, it was evident that MAO coatings fabricated by nine processes displayed typically porous characteristics due to spark discharge (Figure 1a-i). Anodic coatings developed on samples No. 4, No. 7, and No. 8 exhibited uniform characteristics, with pore sizes in the range of 1 to 4 µm (Figure 1d,g,h), while most micropores on samples No. 5 and No. 9 (especially No. 9) were larger than 5 µm in diameter (Figure 1i). According to EDS spectra, anodic coatings developed on nine samples were mainly composed of C, O, F, Na, Mg, P, and Zn elements ( Figure 2). It was clear that the coating compositions were closely related with processing parameters. Sample No. 3 achieved the highest Zn amount, i.e., 4.12 wt.%, while sample No. 9 had the lowest Zn amount, i.e., 1.42 wt.% (Figure 2i).

Cross-Sectional Morphology
According to the cross-sectional morphologies shown in Figure 3 (Figure 1d,g,h), while most micropores on samples No. 5 and No. 9 (especially No. 9) were larger than 5 μm in diameter (Figure 1i). According to EDS spectra, anodic coatings developed on nine samples were mainly composed of C, O, F, Na, Mg, P, and Zn elements ( Figure 2). It was clear that the coating compositions were closely related with processing parameters. Sample No. 3 achieved the highest Zn amount, i.e., 4.12 wt.%, while sample No. 9 had the lowest Zn amount, i.e., 1.42 wt.% (Figure 2i).   with pore sizes in the range of 1 to 4 μm (Figure 1d,g,h), while most micropores on samples No. 5 and No. 9 (especially No. 9) were larger than 5 μm in diameter ( Figure 1i). According to EDS spectra, anodic coatings developed on nine samples were mainly composed of C, O, F, Na, Mg, P, and Zn elements ( Figure 2). It was clear that the coating compositions were closely related with processing parameters. Sample No. 3 achieved the highest Zn amount, i.e., 4.12 wt.%, while sample No. 9 had the lowest Zn amount, i.e., 1.42 wt.% (Figure 2i).    Figure 4. Anodic coatings mainly contained Mg phase and MgO, while P-containing or F-containing crystalline components were not found, indicating that these substances existed mainly in amorphous state. The diffraction peaks of the WE43 substrate were detected, and they could be attributed to thin and porous MAO coatings.

XPS Analysis
An XPS analysis was used to clarify elemental chemical states of sample No. 4. The developed anodic coatings were mainly composed of Mg, Na, P, C, O, F, and Zn elements (Figure 5a). The C 1s spectrum exhibited two peaks at 284.6 and 285.4 eV (Figure 5b), corresponding to C-C(H) and C-O, respectively [26]. As shown in Figure 5c, the P 2p spectrum was fitted into two peaks, at 133.6 and 134.4 eV, assigned to PO 4 3− and HPO 4 2− , respectively [27]. The Zn 2p spectrum was fitted into two peaks located at 1023.5 and 1046.8 eV (Figure 5d), assigned to Zn 3 (PO 4 ) 2 [14,28]. The F 1s spectrum exhibited only one peak at 686.3 eV (Figure 5e), which was ascribed to MgF 2 [29]. Mg 1s spectrum was decomposed into four peaks (Figure 5f), which were separately assigned to MgO, Mg 3 (PO 4 ) 2 , MgHPO 4 , and MgF 2 .  Figure 4. Anodic coatings mainly contained Mg phase and MgO, while P-containing or F-containing crystalline components were not found, indicating that these substances existed mainly in amorphous state. The diffraction peaks of the WE43 substrate were detected, and they could be attributed to thin and porous MAO coatings.

XPS Analysis
An XPS analysis was used to clarify elemental chemical states of sample No. 4. The developed anodic coatings were mainly composed of Mg, Na, P, C, O, F, and Zn elements (Figure 5a). The C 1s spectrum exhibited two peaks at 284.6 and 285.4 eV (Figure 5b), corresponding to C-C(H) and C-O, respectively [26]. As shown in Figure 5c, the P 2p spectrum was fitted into two peaks, at 133.6 and 134.4 eV, assigned to PO4 3-and HPO4 2-, respectively [27]. The Zn 2p spectrum was fitted into two peaks located at 1023.5 and 1046.8 eV (Figure 5d), assigned to Zn3(PO4)2 [14,28]. The F 1s spectrum exhibited only one peak at 686.3 eV (Figure 5e), which was ascribed to MgF2 [29]. Mg 1s spectrum was decomposed into four peaks (Figure 5f), which were separately assigned to MgO, Mg3(PO4)2, MgHPO4, and MgF2.   Figure 4. Anodic coatings mainly contained Mg phase and MgO, while P-containing or F-containing crystalline components were not found, indicating that these substances existed mainly in amorphous state. The diffraction peaks of the WE43 substrate were detected, and they could be attributed to thin and porous MAO coatings.

XPS Analysis
An XPS analysis was used to clarify elemental chemical states of sample No. 4. The developed anodic coatings were mainly composed of Mg, Na, P, C, O, F, and Zn elements (Figure 5a). The C 1s spectrum exhibited two peaks at 284.6 and 285.4 eV (Figure 5b), corresponding to C-C(H) and C-O, respectively [26]. As shown in Figure 5c, the P 2p spectrum was fitted into two peaks, at 133.6 and 134.4 eV, assigned to PO4 3-and HPO4 2-, respectively [27]. The Zn 2p spectrum was fitted into two peaks located at 1023.5 and 1046.8 eV (Figure 5d), assigned to Zn3(PO4)2 [14,28]. The F 1s spectrum exhibited only one peak at 686.3 eV (Figure 5e), which was ascribed to MgF2 [29]. Mg 1s spectrum was decomposed into four peaks (Figure 5f), which were separately assigned to MgO, Mg3(PO4)2, MgHPO4, and MgF2.

Coating Corrosion Resistance
Potentiodyanmic polarization curves of WE43 substrate and nine MAO samples are shown in Figure 6. The fitted i corr values using Gamry's software are listed in Table 2. Surface-treated samples achieved lower i corr values and, therefore, have better corrosion resistance than the WE43 substrate. As listed in Table 2, sample No. 7 achieved the lowest i corr value (1.09 × 10 −8 A·cm −2 ), which was three orders of magnitude lower than that of the WE43 substrate (2.11 × 10 −5 A·cm −2 ). The i corr values of samples No. 1 and No. 4 were separately 1.12 × 10 −7 and 1.09 × 10 −7 A·cm −2 , two orders of magnitude lower than that of the WE43 substrate.

The Orthogonal Results
In this study, the measured i corr and Zn amounts were selected as the objective parameters, and the experimental results were obtained by using the intuitionistic analysis. Based on the values of the differences listed in Table 1, the influencing order of four factors on i corr of MAO samples was EDTA-ZnNa 2 > NH 4 HF 2 > IP6 > H 3 PO 4 . In detail, EDTA-ZnNa 2 was the first important factor influencing coating corrosion resistance. With the increasing EDTA-ZnNa 2 concentration, i corr values decreased from Level 1 to Level 2 but increased from Level 2 to Level 3, showing that a proper EDTA-ZnNa 2 concentration was helpful for improving corrosion resistance. For increasing NH 4 HF 2 concentration, i corr values slightly increased from Level 1 to Level 2 but significantly decreased from Level 2 to Level 3. The IP6 concentration was the third important factor influencing the i corr value, and with the increase of IP6 concentration, i corr values gradually decreased (Table 1).

Coating Corrosion Resistance
Potentiodyanmic polarization curves of WE43 substrate and nine MAO samples are shown in Figure 6. The fitted icorr values using Gamry's software are listed in Table 2. Surface-treated samples achieved lower icorr values and, therefore, have better corrosion resistance than the WE43 substrate. As listed in Table 2, sample No. 7 achieved the lowest icorr value (1.09 × 10 -8 A•cm -2 ), which was three orders of magnitude lower than that of the WE43 substrate (2.11 × 10 -5 A•cm -2

E (V vs SCE)
Log i (Aꞏcm -2 )

Coating Corrosion Resistance
Potentiodyanmic polarization curves of WE43 substrate and nine MAO samples are shown in Figure 6. The fitted icorr values using Gamry's software are listed in Table 2. Surface-treated samples achieved lower icorr values and, therefore, have better corrosion resistance than the WE43 substrate. As listed in Table 2, sample No. 7 achieved the lowest icorr value (1.09 × 10 -8 A·cm -2 ), which was three orders of magnitude lower than that of the WE43 substrate (2.11 × 10 -5 A·cm -2 ). The icorr values of samples No. 1 and No. 4 were separately 1.12 × 10 -7 and 1.09 × 10 -7 A·cm -2 , two orders of magnitude lower than that of the WE43 substrate.  For the Zn amount in MAO coatings, the influencing order of four factors was ranked as EDTA-ZnNa 2 > NH 4 HF 2 > H 3 PO 4 > IP6. EDTA-ZnNa 2 played the most important role on affecting the Zn amount, and the increased EDTA-ZnNa 2 concentration continually improved the Zn amount in anodic coatings, indicating that Zn ions entered into anodic coatings by diffusion. NH 4 HF 2 concentration was the second important processing factor affecting the Zn amount. With the increase of NH 4 HF 2 concentration, the Zn amount gradually decreased. Meanwhile, the Zn amount slightly increased with increasing H 3 PO 4 concentration, which was the third important factor.  Figure 7 shows the i corr values, thickness; porosity; and chemical compositions, including F, P, Zn, and Mg amounts of anodic coatings fabricated on MAO samples. The i corr values somewhat depend on, but not very monotonically, the thickness, porosity, or chemical composition of the MAO coatings, indicating that coating thickness, porosity, or composition was not the only important factor governing coating corrosion resistance. In other words, coating corrosion resistance was jointly determined by surface characteristics. It was clear that the P amount was negatively correlated with the F amount, indicating that P and F competed to take part in coating formation (Figure 7b). Except for samples No. 5 and No. 9, the Mg amount was also negatively correlated with the Zn amount (Figure 7b), which could be attributed to their competition to combine with PO 4 3− into phosphate salts.

Influence of Treatment Time on Coating Property
Under constant current oxidation mode, the treatment time cannot only determine the final voltage of MAO process but also the surface characteristics and coating corrosion resistance. The prepared MAO sample in solution No. 2 achieved good corrosion resistance and a high Zn amount (3.20 wt.%). In order to clarify the influences of treatment time on the coating property, MAO samples were separately fabricated in solution No. 2 with treatment times of 2.5, 3.0, 3.5, and 4.0 min. In addition, the corresponding final voltage values under different treatment times are recorded in Table 3.

Influence of Treatment Time on Coating Property
Under constant current oxidation mode, the treatment time cannot only determine the final voltage of MAO process but also the surface characteristics and coating corrosion resistance. The prepared MAO sample in solution No. 2 achieved good corrosion resistance and a high Zn amount (3.20 wt.%). In order to clarify the influences of treatment time on the coating property, MAO samples were separately fabricated in solution No. 2 with treatment times of 2.5, 3.0, 3.5, and 4.0 min. In addition, the corresponding final voltage values under different treatment times are recorded in Table 3.  Figure 8 shows surface characteristics of the anodic coatings prepared in solution No. 2 under different treatment times. According to the surface morphologies, all MAO coatings were rough with porous characteristics, and there were not any obvious differences in pore size or porosity. However, with the prolonged treatment time, the coating surface became rougher (Figure 8(a1-d1)). As shown in Figure 8(a2-d2), the prepared coatings mainly contained C, Na, F, Mg, O, Zn, and P elements. The Zn amounts in prepared anodic coatings with treatment times of 2.5, 3.0, 3.5, and 4.0 min were 3.23, 3.20, 3.08, and 2.61 wt.%, respectively, exhibiting a slightly decreased trend. According to cross-sectional morphologies shown in Figure 8(a3-d3), the thicknesses of MAO coatings prepared with treatment times of 2.5, 3.0, 3.5, and 4.0 min were 11.02 ± 0.78, 16.06 ± 1.31, 16.45 ± 0.45, and 16.58 ± 0.61 µm, respectively, showing that extended treatment time could gradually increase coating thickness.   Figure 9. According to the obtained corresponding electrochemical parameters listed in Table 4, with the prolongation of treatment time, the i corr values slightly increased, indicating that the prolonged treatment time was unfavorable for coating corrosion resistance.

Formation Mechanism of Zn-Containing Coatings
The coating formation by MAO treatment is a complex process, which may involve some chemical reactions, such as anodization, thermal oxidation, and plasma oxidation [30]. Based on coating surface morphology, chemical composition, and phase structure, it is possible to deduce the formation mechanism of Zn-containing coatings by applying a two-stage process.

Formation Mechanism of Zn-Containing Coatings
The coating formation by MAO treatment is a complex process, which may involve some chemical reactions, such as anodization, thermal oxidation, and plasma oxidation [30]. Based on coating surface morphology, chemical composition, and phase structure, it is possible to deduce the formation mechanism of Zn-containing coatings by applying a two-stage process.
In the first stage, the used electrolytes in MAO solutions begin to ionize into corresponding ions. EDTA-ZnNa 2 can be ionized into EDTA-Zn 2− , Na + , EDTA 4− , and Zn 2+ ions. Meanwhile, NH 4 + , H + , and F − ions are present from NH 4 HF 2 . Meanwhile, IP6 and EDTA 4− compete to combine with Zn 2+ ions. In the second stage, sparks are present on the anode, the Mg sample is oxidized, and Mg 2+ ions are developed: At the same time, anions such as OH − , EDTA-Zn 2− , PO 4 3− , F − , IP6 radicals, and negatively charged zinc phytic acid complexes move toward the sample surface under the electric field between the anode and the cathode [11,31]. Meanwhile, the instantaneous temperature on the surface of the oxidized sample is very high due to spark discharge, which can hydrolyze IP6 molecules into inorganic phosphate and lower inositol phosphate (IP1-IP5) [32,33]. When these anions reach the anode surface, they combine with cations and form a stable ceramic coating under transient high temperatures and pressures. The main reaction equations on the Mg alloy surface are as follows [23]: Mg 2+ + 2F − = MgF 2 (7) In this study, Zn 2+ ions in MAO solutions exist as EDTA-Zn 2− or zinc phytic acid complexes, and these negatively charged Zn-containing complexes enter into MAO coatings under the electric field. EDTA-ZnNa 2 played the most important role in determining the Zn amount (Table 1). With the increase of the EDTA-ZnNa 2 concentration, the Zn amount continually increased, indicating that Zn 2+ ions enter into anodic coatings by diffusion.
Besides OH − ions, PO 4 3− ions and F − ions can also take part in coating formation. PO 4 3− ions can react with both Mg 2+ and Zn 2+ into Mg 3 (PO 4 ) 2 and Zn 3 (PO 4 ) 2 , while F − ions combine with Mg 2+ into MgF 2 . Because of low Zn amount in anodic coatings, Mg 3 (PO 4 ) 2 may be the main phosphate. Therefore, P competes with F to take part in coating formation (Figure 7b), which could clearly explain the results that the increasing NH 4 HF 2 concentration gradually decrease the Zn amount (Table 1). In addition, Mg 2+ ions in anodic coatings can exist as MgO, Mg 3 (PO 4 ) 2 , or MgF 2 , while Zn 2+ ions exist as Zn 3 (PO 4 ) 2 . Under some conditions, Mg 2+ ions compete with Zn 2+ ions to combine with PO 4 3− , and, therefore, the Mg amount was also negatively correlated with the Zn amount except for samples No. 5 and No. 9 (Figure 7b).

Effects of Coating Characteristics on Corrosion Resistance
For biomedical Mg alloys, the corrosion resistance of the MAO-treated sample is very important in order to achieve clinical applications. Recent results show that corrosion resistance of MAO coatings is mainly determined by chemical composition [34], micropore size or porosity [10,12], and coating thickness [12,23], while these characteristics are influenced by electrolyte composition and concentration, and electrical parameters [11].
Coating composition is an important factor in determining the corrosion resistance. It is believed that the fabricated anodic coatings composed of stable components can exhibit good corrosion resistance. In this study, Zn in MAO coatings existed as Zn 3 (PO 4 ) 2 , which is a very stable substance in water solution, and its solubility product constant (K sp ) at 298.15 K is 9.1 × 10 −33 [35]. Therefore, the Zn-containing coatings achieve good corrosion resistance. However, among the nine MAO-treated samples, there are no evident differences on their Zn amounts, and, therefore, coating compositions is not an important factor in regard to corrosion resistance.
The coating thickness plays an important role on influencing the corrosion resistance. Under the same condition, thicker coatings can inhibit corrosive ions, such as chloride ions, entering into the substrate and therefore achieve better corrosion resistance [12,23]. According to Figure 7a, the i corr values did not vary monotonically with coating thickness, thus indicating that coating thickness is not an important factor governing coating corrosion resistance.
Besides the coating composition and thickness, the coating micropore size or porosity can also influence the corrosion resistance of anodic coatings. Because of spark discharge, the developed MAO coatings display typically porous characteristics with some microcracks. When MAO samples are served in a corrosive environment, the defective areas provide a path for chloride ions to easily reach the substrate surface and finally result in the initiation of corrosion [36,37]. How to fabricate a uniform MAO coating with a small pore size is an important research work. According to the experimental results, the corrosion resistance is also not closely related with chemical composition, thickness, or coating porosity, thus suggesting that coating corrosion resistance is jointly determined by the coating-surface characteristics [9].
The EDTA-ZnNa 2 concentration is the first important factor in regard to coating corrosion resistance. As listed in Table 1, EDTA-ZnNa 2 at Level 2 achieved the best corrosion resistance, which can be explained based on the following two aspects. On the one hand, as our previous experimental results obtained in alkaline solutions, EDTA-ZnNa 2 was a corrosive agent of Mg alloys and therefore harmful for coating formation [15]. On the other hand, EDTA-ZnNa 2 can increase the Zn 3 (PO 4 ) 2 amount and may be helpful for improving the coating corrosion resistance attributed to the highly stable characteristics of Zn 3 (PO 4 ) 2 . According to the influencing regularity of EDTA-ZnNa 2 concentration on coating property, EDTA-ZnNa 2 at Level 2 achieved the best corrosion resistance. NH 4 HF 2 concentration is the second most important factor in determining the corrosion resistance of MAO coatings, and the i corr values increased from Level 1 to Level 2 but decreased from Level 2 to Level 3 (Table 1). NH 4 HF 2 is a widely used corrosive inhibitor of Mg alloys and is helpful for coating formation. However, with the increase of NH 4 HF 2 concentration, the F amount in MAO coatings increased, and, therefore, the MgF 2 amount increased, which resulted in the decrease of the amount of Zn 3 (PO 4 ) 2 . Moreover, the K sp of MgF 2 is 6.5 × 10 −9 [38], indicating that MgF 2 is less stable compared with Zn 3 (PO 4 ) 2 . Therefore, the coating corrosion resistance did not exhibit a simple positive or negative relationship with the NH 4 HF 2 concentration. Although the fabricated Zn-containing MAO coatings in this study exhibit excellent corrosion resistance, the Zn amount should be further increased in the future.

Conclusions
In near-neutral solutions, EDTA-ZnNa 2 was used as Zn source and Zn-containing MAO coatings were developed on WE43 alloys. The influences of the regularity of electrolyte compositions on corrosion resistance and the Zn content of anodic coatings were clarified by the orthogonal experiment. The order of four factors on coating corrosion resistance is EDTA-ZnNa 2 > NH 4 HF 2 > IP6 > H 3 PO 4 , while the sequence on the Zn amount is ranked as EDTA-ZnNa 2 > NH 4 HF 2 > H 3 PO 4 > IP6. The fabricated Zn-containing coatings achieve excellent corrosion resistance, and the lowest i corr is three orders of magnitude lower than that of the WE43 substrate. The highest Zn amount of MAO coatings prepared in near-neutral solutions achieves 4.12 wt.%, and Zn ions enter into anodic coatings by diffusion. The prolonged treating time can slightly increase the coating thickness but insignificantly decrease the Zn content and corrosion resistance. In the future, the biocompatibility and antibacterial ability will be further evaluated in order to clarify in vitro biological property.