E ﬀ ect of V 2 O 5 Additive on Micro-Arc Oxidation Coatings Fabricated on Magnesium Alloys with Di ﬀ erent Loading Voltages

: E ﬀ ect of V 2 O 5 additive in silicate-containing electrolyte on AZ91D magnesium alloys treated by micro-arc oxidation (MAO) technology under di ﬀ erent loading voltages was investigated. The results showed that vanadium was well up-taken into the coating chemically. Moreover, a new phase of MgV 2 O 4 with spinel structure was obtained in MAO coatings due to V 2 O 5 added into the electrolyte. The MgV 2 O 4 phase was responsible for the coatings exhibiting brown color and also was beneﬁcial to improving the anti-corrosion property. In spotting tests, the corrosion resistances of coatings prepared under the high voltage are about 6–9 times higher than those of the low voltage because of the thicker coatings of the former. In potentiodynamic polarization tests, the coatings’ corrosion resistances were improved with the addition of V 2 O 5 , which was more signiﬁcant under the low voltage than that under the high voltage. When the concentration of V 2 O 5 was 0.2 g / L, the corrosion current density of the coating was the lowest, which means that the coating’s corrosion resistance under the low voltage is the best. Hence, it is necessary to carry out targeted design of the coating’s microstructure according to the di ﬀ erent applications.


Introduction
Magnesium and magnesium alloys exhibit superior advantages in various applications such as aviation, transportation, and 3C products, due to their outstanding properties like low density, high specific stiffness, and strength to weight ratio [1]. Unfortunately, Mg is the most negative metallic structural material, and its standard electrode potential is only −2.37 V. Therefore, the corrodibility of Mg alloys is a remarkable barrier to wide applications. Micro-arc oxidation (MAO), as a technique developed in the last two decades based on traditional anodic oxidation, is carried out generally in alkaline electrolytes to develop oxide layers on the surface of valve metal. During the MAO process, the localized instantaneous high temperature sintering of the oxide promotes the formation of the ceramic-like coatings with compact structure, good adhesion with substrate, superior corrosion resistance, and excellent comprehensive performance [2][3][4][5]. A relatively thick coating with modified microstructure and phase composition can be produced by optimizing the electrolyte composition to further improve the performance of the coating [6][7][8][9].
To date, studies on the electrolyte formulas of the MAO process mainly focus on silicate, phosphate, sodium hydroxide, and aluminate system, etc.; the combination of the above several kinds of systems; or joining different additives into the electrolyte system to obtain good performance coatings [6,[8][9][10][11][12][13][14][15]. Up to now, reports about soluble oxides, which also have a relatively low cost attribute compared to those nanoparticles [16][17][18][19][20], as additives in MAO electrolyte are still very rare. Vanadium is

Coating Fabrication
The treatment was implemented using a home-made, pulsed DC power device up to the two limiting voltages, U L = 380 V or 480 V for 30 min. The frequency and the duty cycle were 700 Hz and 20%, respectively. The specimen was used as the anode, and a piece of stainless steel sheet was employed as the cathode. The MAO electrolytes were composed of sodium silicate (Na 2 SiO 3 , 15 g/L) and potassium fluoride (KF, 13 g/L), with V 2 O 5 additive of 0 g/L, 0.2 g/L, 0.5 g/L, and 0.8 g/L, respectively. V 2 O 5 was purchased from Tianjin Sitong Chemical Plant (Tianjin, China). All the other chemicals, such as Na 2 SiO 3 , KF, and KMnO 4 , were supplied by Sinopharm Chemical Reagent Co., Ltd., (Shanghai, China). All the reagents, unless otherwise noted, were of analytical grade. The electrolyte was controlled at a temperature range of 20-40 • C by a stirring and water cooling device. At this point, all MAO specimens were entirely rinsed in deionized water, dried with the blower, and then stored in a drying cabinet before characterization.

Coating Characterization
An eddy meter (TT260, Time Company, Beijing, China) was used to assess the thickness of MAO coatings. Eight parallel samples were selected to be examined and 10 points were taken on each sample surface.
The coatings' surface morphologies and cross-sectional elemental analyses were performed with a scanning electron microscopy (SEM, JSM-6700F, Jeol, Tokyo, Japan) coupled with energy dispersive spectroscopy (EDS). Quantitative SEM image analysis was done with the software Image J to obtain the distribution of the micropore and the surface porosity. The phase composition was identified by X-ray diffraction (XRD, D/Max-2400. Rigaku, Tokyo, Japan) with a Cu Kα target in a scan range from 20 • to 100 • (in 2θ) and a step size of 0.02 • .
For the purpose of appraising the coating corrosion resistance in an acidic condition, the spotting test was conducted based on the standard HB5061-77, but the concentration of HNO 3 was increased to 2.5 times [33]. Therefore, the recipe of the spotting solution was as follows: 0.05 g KMnO 4 , 12.5 mL HNO 3 , and 87.5 mL H 2 O. The device is shown in Figure 1 below. The sample was installed on the device and the screws were tightened to avoid the liquid missing the test hole. Titration was taken with a 1 mL dropper held vertically and approximately 5 mm above the test hole (note: there should be no gas in the dropper). Stopwatch timing started after continuous dropping of 2 drops, and the timing ended when the liquid completely changed from burgundy to colorless. When the spotting solution was dropped on the surface of the sample, HNO 3 reacted preferentially with MgO (the main composition of the coating) and further corroded the substrate by the following reactions: At the same time, KMnO 4 reacts with reducing substance (Mg substrate) in a strong acidic environment (Equations (2) and (3)), and the resulting Mn 2+ is colorless.
Metals 2020, 10, x FOR PEER REVIEW 3 of 14 For the purpose of appraising the coating corrosion resistance in an acidic condition, the spotting test was conducted based on the standard HB5061-77, but the concentration of HNO3 was increased to 2.5 times [33]. Therefore, the recipe of the spotting solution was as follows: 0.05 g KMnO4, 12.5 mL HNO3, and 87.5 mL H2O. The device is shown in Figure 1 below. The sample was installed on the device and the screws were tightened to avoid the liquid missing the test hole. Titration was taken with a 1 mL dropper held vertically and approximately 5mm above the test hole (note: there should be no gas in the dropper). Stopwatch timing started after continuous dropping of 2 drops, and the timing ended when the liquid completely changed from burgundy to colorless. When the spotting solution was dropped on the surface of the sample, HNO3 reacted preferentially with MgO (the main composition of the coating) and further corroded the substrate by the following reactions: (1) At the same time, KMnO4 reacts with reducing substance (Mg substrate) in a strong acidic environment (Equations (2) and (3)), and the resulting Mn 2+ is colorless. For electrochemical testing, the potentiodynamic polarization (PDP) curves were obtained by an electrochemical workstation (CHI660C, Chenhua, Shanghai, China) controlled by a computer connecting to a conventional three-electrode system assisted by a platinum sheet auxiliary electrode, a Hg/Hg2Cl2/sat KCl reference electrode, and the testing sample for a working electrode. The test was operated at a sweep rate of 10 mV·s −1 and a scan range of −1.8 V to −0.8V after 24 h immersion in a 3.5 wt% sodium chloride aqueous solution (1cm 2 exposed surface area). To ensure the repeatability of the above tests, 8 groups of parallel experiments coming from two sides of 4 samples were carried out.
It should be pointed out that the thickness and anti-corrosion values were final average values after removing exceptional data by Grubbs test. Figure 2 depicts the macro photographs of MAO coatings with various V2O5 concentrations under different loading voltages. As shown in Figure 2, the substrates were coated with MAO layer and the color depended on both the UL and V2O5 contents. The MAO of tantalum also marked For electrochemical testing, the potentiodynamic polarization (PDP) curves were obtained by an electrochemical workstation (CHI660C, Chenhua, Shanghai, China) controlled by a computer connecting to a conventional three-electrode system assisted by a platinum sheet auxiliary electrode, a Hg/Hg 2 Cl 2 /sat KCl reference electrode, and the testing sample for a working electrode. The test was operated at a sweep rate of 10 mV·s −1 and a scan range of −1.8 V to −0.8 V after 24 h immersion in a 3.5 wt% sodium chloride aqueous solution (1 cm 2 exposed surface area). To ensure the repeatability of the above tests, 8 groups of parallel experiments coming from two sides of 4 samples were carried out.

Macroscopic Morphologies
It should be pointed out that the thickness and anti-corrosion values were final average values after removing exceptional data by Grubbs test.  Figure 2, the substrates were coated with MAO layer and the color depended on both the U L and V 2 O 5 contents. The MAO of tantalum also marked similar observations [34]. The color of the coatings gradually darkened from the initial grey to the final brown with the increase of the U L and V 2 O 5 contents. When V 2 O 5 was 0.2 g/L, the color of the coatings was similar to their counterpart coatings without the V 2 O 5 additive, as shown in Figure 2a,b,e,f, and the coatings exhibited an obvious brown color under both two loading voltages when V 2 O 5 was 0.8 g/L. Interestingly, the uneven color on the coating surface was visible to the naked eye with concentrations of V 2 O 5 up to 0.8 g/L, as evident in Figure 2d, which indicated the concentration should not be too high.

Macroscopic Morphologies
Metals 2020, 10, x FOR PEER REVIEW 4 of 14 similar observations [34]. The color of the coatings gradually darkened from the initial grey to the final brown with the increase of the UL and V2O5 contents. When V2O5 was 0.2 g/L, the color of the coatings was similar to their counterpart coatings without the V2O5 additive, as shown in Figure  2a,b,e,f, and the coatings exhibited an obvious brown color under both two loading voltages when V2O5 was 0.8 g/L. Interestingly, the uneven color on the coating surface was visible to the naked eye with concentrations of V2O5 up to 0.8 g/L, as evident in Figure 2d, which indicated the concentration should not be too high.

Microscopy Morphologies
The surface SEM images and the distribution of micro pores of MAO coatings prepared under various conditions are displayed in Figures 3 and 4. Quantitative statistical results of the surface porosity of the coatings are shown in Figure 5. The surface of the coating possesses numerous irregular micro-pores with different sizes, which were formed during the breakdown of the coating under the strong electric field. These micro-pores were also the discharge channels within the coating, releasing the melts and gas outward to the coating surface. Comparing the surface morphology of the coatings under different voltages, the amount of micro-pores on the surfaces prepared under 480 V is lower than that under 380 V, but the pores' size and the corresponding surface porosity is larger, as illustrated in Figures 4 and 5. This was mainly because the sample obtained more energy during MAO treatment under the high voltage, and the breakdown occurred easily so the pores size on the surfaces is larger and micro-cracks also appear (Figure 4a-d). On the other hand, it can be seen from Figure 5 that, with the addition of V2O5 to the electrolyte, the surface porosity of the coatings under both voltages increased. As the concentration of V2O5 increases from 0.2 g/L to 0.8 g/L, the surface porosity of the coatings under 480 V first increased slightly and then decreased, whereas under 380 V, it increased gradually. The coating with minimal V2O5 concentration presents the preferable compactness. Furthermore, large protrusions and valleys were locally distributed on the coatings' surfaces ( Figure 3c,d). This phenomenon is consistent with the uneven color appearance of the macro morphology, especially in Figure 2d.
According to the literatures [35,36], V2O5 reacts in water to give vanadic and metavanadic acid after adding it into the solution (Equations (5) and (6)).

Microscopy Morphologies
The surface SEM images and the distribution of micro pores of MAO coatings prepared under various conditions are displayed in Figures 3 and 4. Quantitative statistical results of the surface porosity of the coatings are shown in Figure 5. The surface of the coating possesses numerous irregular micro-pores with different sizes, which were formed during the breakdown of the coating under the strong electric field. These micro-pores were also the discharge channels within the coating, releasing the melts and gas outward to the coating surface. Comparing the surface morphology of the coatings under different voltages, the amount of micro-pores on the surfaces prepared under 480 V is lower than that under 380 V, but the pores' size and the corresponding surface porosity is larger, as illustrated in Figures 4 and 5. This was mainly because the sample obtained more energy during MAO treatment under the high voltage, and the breakdown occurred easily so the pores size on the surfaces is larger and micro-cracks also appear (Figure 4a-d). On the other hand, it can be seen from Figure 5 that, with the addition of V 2 O 5 to the electrolyte, the surface porosity of the coatings under both voltages increased. As the concentration of V 2 O 5 increases from 0.2 g/L to 0.8 g/L, the surface porosity of the coatings under 480 V first increased slightly and then decreased, whereas under 380 V, it increased gradually. The coating with minimal V 2 O 5 concentration presents the preferable compactness. Furthermore, large protrusions and valleys were locally distributed on the coatings' surfaces ( Figure 3c,d). This phenomenon is consistent with the uneven color appearance of the macro morphology, especially in Figure 2d.
According to the literatures [35,36], V 2 O 5 reacts in water to give vanadic and metavanadic acid after adding it into the solution (Equations (5) and (6)).

of 14
The detailed pH and conductivity of the electrolytes are presented in Table 1. The electrolytes were abbreviated as 0 g/L V 2 O 5 (base electrolyte: 15 g/L Na 2 SiO 3 + 13 g/L KF), 0.2 g/L V 2 O 5 (base electrolyte + 0.2g/L V 2 O 5 ), 0.5 g/L V 2 O 5 (base electrolyte + 0.5 g/L V 2 O 5 ), and 0.8 g/L V 2 O 5 (base electrolyte + 0.8 g/L V 2 O 5 ). It can be seen that the alkalinity of the electrolyte composed of Na 2 SiO 3 and KF decreases after the addition of V 2 O 5 , and the acidity of the electrolyte increases with the increasing of the V 2 O 5 concentration. In addition, with an increase of the amount of V 2 O 5 added, the conductivity of the electrolyte also decreases. It is generally believed that the MAO process involves the dissolution of old coatings and the formation of new coatings, resulting in a competition in both processes [27,37]. The decrease of pH and conductivity changes the properties of the electrolytes, which affects the formation of passivation film in the initial stage of MAO and the discharge characteristics of the coatings. Therefore, the addition of V 2 O 5 to the solution accelerates the dissolution process due to the reduced alkalinity and conductivity, and may result in the decreased growth rate and a weak state in a certain local area on the coating surface where the continual breakdown occurs, inducing the emergence of the large-size micropores and microcracks, which in turn become the access for corrosive media entering into the coating, and thus affects the corrosion resistance of the coating. The detailed pH and conductivity of the electrolytes are presented in Table 1. The electrolytes were abbreviated as 0g/L V2O5 (base electrolyte: 15g/L Na2SiO3+13g/L KF), 0.2g/L V2O5 (base electrolyte+0.2g/L V2O5), 0.5g/L V2O5 (base electrolyte +0.5g/L V2O5), and 0.8g/L V2O5 (base electrolyte +0.8g/L V2O5). It can be seen that the alkalinity of the electrolyte composed of Na2SiO3 and KF decreases after the addition of V2O5, and the acidity of the electrolyte increases with the increasing of the V2O5 concentration. In addition, with an increase of the amount of V2O5 added, the conductivity of the electrolyte also decreases. It is generally believed that the MAO process involves the dissolution of old coatings and the formation of new coatings, resulting in a competition in both processes [27,37].The decrease of pH and conductivity changes the properties of the electrolytes, which affects the formation of passivation film in the initial stage of MAO and the discharge characteristics of the coatings. Therefore, the addition of V2O5 to the solution accelerates the dissolution process due to the reduced alkalinity and conductivity, and may result in the decreased growth rate and a weak state in a certain local area on the coating surface where the continual breakdown occurs, inducing the emergence of the large-size micropores and microcracks, which in turn become the access for corrosive media entering into the coating, and thus affects the corrosion resistance of the coating.

Chemical Composition of MAO Coating
The cross-section elemental analysis of one typical coating prepared with 0.8 g/L V 2 O 5 under 480 V is shown in Figure 6.Theelement content of the coating is summarized in Table 2. Generally speaking, the type and the content of the elements are closely related to the substrate and electrolyte composition. As seen in Figure 6 and Table 2, the coating mainly consists of magnesium, aluminum, zinc, oxygen, silicon, fluorine, and a trace of vanadium. Magnesium, aluminum, and zinc are the major constituents of AZ91D Mg alloy substrate, while fluorine, silicon, and vanadium are attributed to the composition of the electrolyte, KF, Na 2 SiO 3 , and V 2 O 5 , respectively. Oxygen may be derived from the electrolyte or the air. The contents of V element distributed in the coating were low, which may result from the fact that the amount of V added to the solution was low, as shown in Figure 6h and Table 2. The contents of F element near to the coating-substrate interface were relatively high and present a tendency of diffusion towards the substrate. In addition, O and Si elements in the coating were uniformly spread across the transverse section. These results were also reported by other scientists [38,39].

Chemical Composition of MAO Coating
The cross-section elemental analysis of one typical coating prepared with 0.8 g/L V2O5 under 480 V is shown in Figure 6.Theelement content of the coating is summarized in Table 2. Generally speaking, the type and the content of the elements are closely related to the substrate and electrolyte composition. As seen in Figure 6 and Table 2, the coating mainly consists of magnesium, aluminum, zinc, oxygen, silicon, fluorine, and a trace of vanadium. Magnesium, aluminum, and zinc are the major constituents of AZ91D Mg alloy substrate, while fluorine, silicon, and vanadium are attributed to the composition of the electrolyte, KF, Na2SiO3, and V2O5, respectively. Oxygen may be derived from the electrolyte or the air. The contents of V element distributed in the coating were low, which may result from the fact that the amount of V added to the solution was low, as shown in Figure 6h and Table 2. The contents of F element near to the coating-substrate interface were relatively high and present a tendency of diffusion towards the substrate. In addition, O and Si elements in the coating were uniformly spread across the transverse section. These results were also reported by other scientists [38,39].  Figure7 demonstrates the XRD pattern of the coatings prepared under 480 V without V2O5 and with 0.8 g/L V2O5. It was observed that the coatings without V2O5 added into the electrolyte were mainly composed of MgO with a cubic structure, Mg2SiO4 spinel phase and MgF2. These phases were gray or white. After adding V2O5 into the electrolyte, a new phase of MgV2O4was observed in the coating, as shown in Figure 7a, and this coincides with the result of map scanning in Figure 6. It can be concluded that MgV2O4 should be the substance which affects the coatings' coloring. MgV2O4 is also a spinel structure and possesses plenty of advantages, such as high temperature resistance, corrosion and wear resistance, high strength, and high hardness [40,41], so V2O5 added into the electrolyte can improve the coating's corrosion resistance.   Figure 7 demonstrates the XRD pattern of the coatings prepared under 480 V without V 2 O 5 and with 0.8 g/L V 2 O 5 . It was observed that the coatings without V 2 O 5 added into the electrolyte were mainly composed of MgO with a cubic structure, Mg 2 SiO 4 spinel phase and MgF 2 . These phases were gray or white. After adding V 2 O 5 into the electrolyte, a new phase of MgV 2 O 4 was observed in the coating, as shown in Figure 7a, and this coincides with the result of map scanning in Figure 6. It can be concluded that MgV 2 O 4 should be the substance which affects the coatings' coloring. MgV 2 O 4 is also a spinel structure and possesses plenty of advantages, such as high temperature resistance, corrosion and wear resistance, high strength, and high hardness [40,41], so V 2 O 5 added into the electrolyte can improve the coating's corrosion resistance. electrolyte, the amount of MgV2O4increasesand the coating's color deepens gradually to finally be brown ( Figure 2). Furthermore, the color of the coatings obtained under 480 V are darker than those under 380 V (Figure 2),since the driving force of VO3 − and VO4 3− in the electrolyte under the high voltage is greater, which is beneficial for the ions entering into the discharging channels of the coatings and participating in the reactions. Therefore, the more MgV2O4phase in the coatings, the darker the color of the coatings.  Figure 8 illustrates the thickness and the corrosion resistance corresponding to the spotting test results of the coatings prepared under various conditions. The loading voltage is one of the primary parameters to determine the MAO coating's thickness. Consequently, it can be seen from Figure 8 that the effect of voltages on the thickness is noticeable with the constant V2O5 concentration, and the thickness of the coatings prepared under 480 V are about 2-3 times higher than those under 380 V. Additionally, as the concentration of V2O5 increases, coating thickness obtained under 480 V first increases and then decreases, and an inflection point appears at the minimum concentration of V2O5. In the case of the low voltage, the thickness of the coatings decreased slightly with increase in the V2O5 dosage. This may be ascribed to the vanadic and metavanadic acid generated by V2O5 added into the electrolyte, which changed the nature of the solution and dissolved the coating, decreasing the coating's growth rate. Nevertheless, the side effect of vanadic and metavanadic acid was weakened under the high voltage due to the faster growth rate.

Corrosion Studies
Obviously, according to Figure 8, the corrosion resistance in the spotting test of the coatings formed under two kinds of voltages exhibits a similar variation trend with the coatings' thicknesses when changing the mass concentration of V2O5. In the spotting corrosive solution, the coating thickness plays a leading role on the corrosion resistance (Equation (1)), and the coatings' corrosion resistance under the high voltage was much better than that formed under the low voltage, which may be attributed to the thicker coating of the former. Besides the thickness, the formation of chemically stable Mg2SiO4and MgF2phase during the MAO process could also have contributed to Equations (5) and (6) Figure 8 illustrates the thickness and the corrosion resistance corresponding to the spotting test results of the coatings prepared under various conditions. The loading voltage is one of the primary parameters to determine the MAO coating's thickness. Consequently, it can be seen from Figure 8 that the effect of voltages on the thickness is noticeable with the constant V 2 O 5 concentration, and the thickness of the coatings prepared under 480 V are about 2-3 times higher than those under 380 V. Additionally, as the concentration of V 2 O 5 increases, coating thickness obtained under 480 V first increases and then decreases, and an inflection point appears at the minimum concentration of V 2 O 5 . In the case of the low voltage, the thickness of the coatings decreased slightly with increase in the V 2 O 5 dosage. This may be ascribed to the vanadic and metavanadic acid generated by V 2 O 5 added into the electrolyte, which changed the nature of the solution and dissolved the coating, decreasing the coating's growth rate. Nevertheless, the side effect of vanadic and metavanadic acid was weakened under the high voltage due to the faster growth rate.

Corrosion Studies
Obviously, according to Figure 8, the corrosion resistance in the spotting test of the coatings formed under two kinds of voltages exhibits a similar variation trend with the coatings' thicknesses when changing the mass concentration of V 2 O 5 . In the spotting corrosive solution, the coating thickness plays a leading role on the corrosion resistance (Equation (1)), and the coatings' corrosion resistance under the high voltage was much better than that formed under the low voltage, which may be attributed to the thicker coating of the former. Besides the thickness, the formation of chemically stable Mg 2 SiO 4 and MgF 2 phase during the MAO process could also have contributed to the enhancement of the coating's corrosion resistance and it requires long time to react with the HNO 3 etchant. Meanwhile, the anti-corrosion property of the coating obtained under the high voltage in the electrolyte with the minimum amount of V 2 O 5 (0.2 g/L) was the best. The coating's corrosion resistance obtained under 380 V decreases somewhat with the increase of V 2 O 5 , owing to the slightly decreasing thickness.  The PDP plots of the coatings prepared with different V2O5concentrations under different voltages are presented in Figure 9. The corrosion potential (Ecorr), corrosion current density (Jcorr), and linear polarization resistance (Rp) derived from the PDP curves are summarized in Table 3. The PDP curve of the untreated AZ91D sample is also included in Figure 9 as a background. It can be directly seen that the Jcorr values of specimens fabricated by MAO under two kinds of voltages both reduce by two orders of magnitude, compared to that of the uncoated, and it can be certified that MAO can remarkably enhance the corrosion resistance of magnesium alloys. The Jcorr values of the coatings after adding V2O5 into the electrolyte decrease by about one order of magnitude, compared to that without V2O5 additive, which indicates that the tiny concentration of V2O5 can significantly enhance the coating corrosion resistance, and the amplitude of improvement is large under the condition of low voltage.  The PDP plots of the coatings prepared with different V 2 O 5 concentrations under different voltages are presented in Figure 9. The corrosion potential (E corr ), corrosion current density (J corr ), and linear polarization resistance (R p ) derived from the PDP curves are summarized in Table 3. The PDP curve of the untreated AZ91D sample is also included in Figure 9 as a background. It can be directly seen that the J corr values of specimens fabricated by MAO under two kinds of voltages both reduce by two orders of magnitude, compared to that of the uncoated, and it can be certified that MAO can remarkably enhance the corrosion resistance of magnesium alloys. The J corr values of the coatings after adding V 2 O 5 into the electrolyte decrease by about one order of magnitude, compared to that without V 2 O 5 additive, which indicates that the tiny concentration of V 2 O 5 can significantly enhance the coating corrosion resistance, and the amplitude of improvement is large under the condition of low voltage.  The PDP plots of the coatings prepared with different V2O5concentrations under different voltages are presented in Figure 9. The corrosion potential (Ecorr), corrosion current density (Jcorr), and linear polarization resistance (Rp) derived from the PDP curves are summarized in Table 3. The PDP curve of the untreated AZ91D sample is also included in Figure 9 as a background. It can be directly seen that the Jcorr values of specimens fabricated by MAO under two kinds of voltages both reduce by two orders of magnitude, compared to that of the uncoated, and it can be certified that MAO can remarkably enhance the corrosion resistance of magnesium alloys. The Jcorr values of the coatings after adding V2O5 into the electrolyte decrease by about one order of magnitude, compared to that without V2O5 additive, which indicates that the tiny concentration of V2O5 can significantly enhance the coating corrosion resistance, and the amplitude of improvement is large under the condition of low voltage.   Table 3. Derived results from polarization curves shown in Figure 9.  Figure 10 shows a scatter plot of the surface porosity and J corr of the corresponding coating. It can be seen that except for the two coatings without V 2 O 5 additive (indicated by the arrows in the Figure 10), the other data confirm a linear relationship between the surface porosity of the coating with V 2 O 5 and J corr . In NaCl corrosive media, the corruption of corrosion resistance of the specimen resulted from the hydration of MgO, occurring preferentially inside the micropores of the oxide film, corroding the substrate by the following reaction [32,42]:

Parameter
The coating is easy to be damaged due to the presence of chloride ion. Firstly, Mg(OH) 2 produced by the hydration of MgO is not dense and is quite porous and would not provide a protective barrier to the corrosion of AZ91D substrate [32]. On the other hand, compared with other anions, chlorine ions with high activity are more easily adsorbed, and the radius of a chloride ion is smaller, so it is able to more easily penetrate into the coating through defects [43,44]. Thus, the coating corrosion resistance in NaCl solution is not only related to the thickness and phase composition, but also the tightness of the coating [26]. A MAO coating with less micro-defects, lower porosity, and compact structure is better at suppressing the penetration of the corrosive liquid and provides better corrosion protection.  Figure 10 shows a scatter plot of the surface porosity and Jcorr of the corresponding coating. It can be seen that except for the two coatings without V2O5 additive (indicated by the arrows in the Figure 10), the other data confirm a linear relationship between the surface porosity of the coating with V2O5 and Jcorr. In NaCl corrosive media, the corruption of corrosion resistance of the specimen resulted from the hydration of MgO, occurring preferentially inside the micropores of the oxide film, corroding the substrate by the following reaction [32,42]: Mg + 2H2O → Mg(OH)2 + H2(↑) The coating is easy to be damaged due to the presence of chloride ion. Firstly, Mg(OH)2 produced by the hydration of MgO is not dense and is quite porous and would not provide a protective barrier to the corrosion of AZ91D substrate [32]. On the other hand, compared with other anions, chlorine ions with high activity are more easily adsorbed, and the radius of a chloride ion is smaller, so it is able to more easily penetrate into the coating through defects [43,44]. Thus, the coating corrosion resistance in NaCl solution is not only related to the thickness and phase composition, but also the tightness of the coating [26]. A MAO coating with less micro-defects, lower porosity, and compact structure is better at suppressing the penetration of the corrosive liquid and provides better corrosion protection. In the case of the low voltage, the micro pores are small and the coatings are dense (Figures3 and 5), so NaCl corrosive medium has difficulty permeating the coating to corrode the substrate, and the coatings exhibit the better corrosion resistance with respect to the ones obtained under the In the case of the low voltage, the micro pores are small and the coatings are dense (Figures 3  and 5), so NaCl corrosive medium has difficulty permeating the coating to corrode the substrate, and the coatings exhibit the better corrosion resistance with respect to the ones obtained under the high voltage. With an increase in V 2 O 5 concentration from 0.2 g/L to 0.8 g/L, J corr of the coatings under the high voltage initially increases slightly, then decreases; however, J corr under the low voltage increases gradually, and the coating with lower V 2 O 5 concentration presents the minimum corrosion current density. This result corresponds strictly to the surface porosity of the coating. For the coatings without V 2 O 5 , the surface porosity is very low, but the corrosion resistance is still poor. This is mainly due to the MgV 2 O 4 phase in the coatings with V 2 O 5 , which is beneficial to the improvement of the corrosion resistances of the coatings. When the added amount of V 2 O 5 is 0.2 g/L, the coating's corrosion resistance formed under the low voltage is more obviously better than that prepared under the high voltage, which is mainly because the larger-size micropores and microcracks in the latter provide channels for NaCl solution to infiltrate into the coating and further corrode the matrix, as shown in Figure 4b. As a result, the J corr of the coating with the minimum V 2 O 5 added under the low voltage is the lowest, and the coating corrosion resistance is the best. At this point, the energy consumption is low, and to some extent the cost is also relatively low.
In conclusion, the coating's corrosion resistance is influenced by many microstructure characteristic parameters, including the coating thickness, phase composition, compactness, the defects, and so on. The corrosion mechanisms of different corrosive media are diverse, and the influence of each parameter on the coating's corrosion resistance will be dissimilar. In HNO 3 medium, the coating's corrosion resistance is greatly affected by the coating thickness and phase composition. In NaCl corrosive medium, the compactness and defects seem to be the primary reasons for the coating corrosion resistance. The voltage and a small amount of V 2 O 5 additive can affect the coating's corrosion resistance significantly by influencing the abovementioned parameters. Therefore, the targeted coating design should be performed for different cases of application: In the acidic environment, a thicker coating under the high voltage is needed to improve the anti-corrosive property; In the neutral service condition, a dense coating with the low voltage seems to be insusceptible to corrosion media. Moreover, in both cases, the best performance is obtained when the minimum V 2 O 5 was added. Meanwhile, it could also be concluded that there is an interaction between the voltage and V 2 O 5 additive.

Conclusions
A brown coating was obtained on the surface of AZ91D magnesium alloy due to V 2 O 5 added into the electrolyte, and the higher the voltage was, the darker the color of the coating turned out to be. The substance affecting the coatings' coloring was the MgV 2 O 4 phase with spinel structure, newly generated in the coatings, which was beneficial to improving the coatings' corrosion resistance.
In the spotting tests of the HNO 3 medium, the corrosion resistance of the coating mainly depended on the thickness. The thick coating's spotting corrosion resistance, which was fabricated under the high voltage, was about 6-9 times better than that under the low voltage. In the electrochemical tests of the NaCl corrosive medium, the corrosion resistance of the coating mainly depends on the microstructure. The best electrochemical corrosion resistance is achieved when the added amount of V 2 O 5 is minimal under the low voltage, which means that the corrosion current density is the lowest and is about five times lower than that under the high voltage, one order of magnitude lower than that without V 2 O 5 , and two orders of magnitude lower than that of uncoated.
Although the coating's corrosion resistance is different as a result of the different corrosion mechanisms in the different corrosive media, one thing in common is that the coating's corrosion resistance is influenced interactively by many microcosmic characteristic parameters, including the thickness, phase composition, compactness, and defects of the coating. Hence, it is necessary to carry out targeted designing of the coating's microstructure according to the different applications.

Conflicts of Interest:
The authors declare no conflict of interest.