Corrosion protection properties of different inhibitors containing PEO/LDHs composite coating on magnesium alloy AZ31

Corrosion inhibitors 2,5-pyridinedicarboxilate (PDC), sodium metavanadate (SMV) and 5-aminosalicylate (AS) were impregnated into porous PEO coatings respectively via vacuuming process, followed by fast sealing treatment in a Ce containing solution. After that layered double hydroxides (LDHs) based nanocontainers were respectively prepared on them via hydrothermal treatment. In frame of this work it was shown, that sealing effect for the pore was provided by formation of new phase CeO2 on the surface of PEO coatings. And, hydrothermal preparation for preparing LDHs leaded obvious changes in structure and thickness of the coatings. In addition, impregnation of inhibitors was in favor of improving LDHs content in final composite coatings. EIS result indicated that AS/Ce-HT specimen exhibited a best corrosion protection.

www.nature.com/scientificreports/ second/tertiary phase. However, passive film is difficult form on Mg alloys, even highly reactive fluoride ions, since the affinity of Mg to react with species like O 2 , OHor H 2 is high 15,16 ; (III) Precipitation of products from the environment solution and (IV) Prevention of iron or noble metals re-deposition, in this theory, the ability of substances to form stable soluble complexes with Fe 2+ /Fe 3+ was mainly considered. Over the past decades, the emerging technology of smart coatings have been proposed and developed that has some functionality in addition to being a conventional barrier coating. Under this concept, known as selfhealing or self-repair, such a coating has incorporated functionalities to sense damage, process the stimuli signals and to initiate single or multiple healing events or processes [17][18][19][20] . In order to obtain autonomous healing effect, one of strategy is by embedding extrinsic polymerizable healing agents in the coating. Another is by imbedding corrosion inhibitors as the healing agents 21 . When the coating is subjected to stimuli from the environment (such as pH, aggressive species, humidity or moisture), the corrosion inhibitor can leach into the coating defect and inhibit anodic or cathodic reaction, or both of them. The traditional manner of using inhibitors is to directly mix the inhibitor with the binder, but it will result in the poor dispersion and the interaction between the inhibitor and the polymer substrate. Encapsulation strategies to be as an effective solution has been proposed by White et al. 22 and has been widely developed. Examples of encapsulating containers are: (i) inert capsules (e.g. mesoporous silica 23,24 , ceria 25,26 ) which play no role other than to isolate the inhibitors from the polymeric material, and (ii) ion-exchange materials (e.g. clays 27,28 , zeolites 10,29 ) which release the inhibitors in exchange for ions from environment 17,30 .
It had been shown that, the characteristic microstructure on PEO coatings has a great potential to serve as micro/nano reservoirs to store or carry inhibitor 31 . It is worth to mention that a post-sealing treatment is usually necessary step for avoiding loss of the loaded inhibitor from PEO coating. Some researchers [32][33][34][35] have proposed many approaches to seal the pores of PEO coatings, i.e. sol-gel hybrid polymer.
Moreover, layered double hydroxides (LDHs, a kind of clay) as the nano-container have been widely investigated [36][37][38][39][40][41] , especially in combination with plasma electrolytic oxidation (PEO) coatings [42][43][44][45][46] . In our previous works, we reported an easy way to prepare MgAl-LDHs on anodic or PEO layer 47,48 , followed by confirming the two primary internal sources of Mg 2+ and Al 3+ cations for the formation of LDH coatings: degradation of the anodic or PEO layer and dissolution of the AZ31 substrate, and further a two-dimensional layer growth (2D) to three-dimensional (3D) growth model 49 was proposed. Petrova et al. 50 proposed to grow LDHs at ambient pressure on PEO treated magnesium alloy AZ91 in the presence of chelating agents. Recently a synergistic effect for active corrosion protection between Ce species and phosphate loaded LDHs was reported by us 30 . However, most of our study was focusing on loading inhibitors into LDHs, for improving active corrosion protection of coatings. Therefore, based on finding of Lamaka's research 15 , 2,5-pyridinedicarboxilate (PDC), sodium metavanadate (SMV) and 5-aminosalicylate (AS) were chose to impregnate into porous PEO coatings respectively in this work. For avoiding loss of impregnated inhibitors during preparation process of LDHs, a fast sealing treatment was firstly performed in containing Ce solution. The reason for choosing Ce solution was that, presence of Ce ion in coatings was helpful to improve the active corrosion protection of coatings 30 . In such a way, impregnated inhibitors was firstly sealed by Ce conversion coating (Ce ions itself is used as a kind of inhibitors), and then fabrication of LDHs coating could enhance both the sealing effect and overall corrosion resistance of the composite coating. This final obtained composite coatings were characterized using SEM, and EDS, and then the corrosion protection were further investigated via electrochemical tests.

Experimental methods
Materials and reagents. The Mg alloy AZ31 with a nominal composition of Al 2.5-3.5%, Zn 0.6-1.3%, Mn 0.2-1%, Ca 0.04%, Si 0.1%, Cu 0.05% (all in wt%), and balance Mg. Deionized water was used as the solvent in this work. Specimens preparation. AZ31 alloy sheets with dimension of 20 mm × 20 mm × 5 mm were firstly ground on all surfaces up to 2000 grit using SiC papers, and then dried in warm air. The PEO processes were carried out at a constant voltage of 350 V for 600 s with duty ratio 20%, supplied by a pulse DC power source. The electrolyte containing 4 g/L NaAlO 2 and 7.14 g/L NaOH was continuously stirred during the PEO treatment and kept at 20 ± 5 °C by a water cooling system.
Prior to impregnating the inhibitors into porous PEO layer, 0.1 M 2,5-pyridinedicarboxylic acid, sodium metavanadate and 5-aminosalicylic acid aqueous solutions with pH 7 (adjust with NaOH) were prepared respectively. The as-prepared PEO specimens were respectively immersed in the above three aqueous solutions in a well-sealed pressure container. A water pump was employed for constantly extracting the air from this enclosed container through a connected rubber hose. The inhibitors incorporating process lasted for 1 h. After incorporating process, the PEO specimens treated in the three different solutions were taken out of the container, and dried using warm air. The obtained specimens in this stage were designated as PDC, SMV and AS, respectively.
Subsequently, the pores in PEO layer of PDC, SMV and AS specimens were sealed in the aqueous solution containing 5 g/L Ce(NO 3 ) 3 and 0.5 g/L H 2 O 2 at 90 °C for 10 min. After sealing treatment, the specimens were then dried using warm air. For clarity of discussion, the specimens obtained in this stage were called as PDC/ Ce, SMV/Ce and AS/Ce, respectively.
For obtaining MgAl-LDH coating on PEO layer, as-prepared PDC/Ce, SMV/Ce and AS/Ce specimens were hydrothermally treated in 0.1 M NaNO 3 solution in a Teflon-lined stainless steel autoclave at 398 K for 16 h, respectively. Eventually, the specimens were rinsed with distilled water, followed by drying. The final obtained specimens after hydrothermal treatment were designated as PDC/Ce-HT, SMV/Ce-HT and AS/Ce-HT, respectively. detector was employed to investigate the surface and cross-sectional morphologies as well as the chemical compositions of the specimens. X-ray diffraction (XRD; Rigaku D/Max 2500X, Japan) was applied to study the crystal structure and phase composition, using a Cu Kα radiation (40 kV, 40 mA) at a glancing angle of 1.5°, in the range of 2θ from 5° to 80° and at a scanning rate of 0.02° s −1 .
Open circuit potential (OCP) and electrochemical impedance spectra (EIS) were carried out using a CIMPS-2 Zahner system in a three-electrode cell, i.e., a saturated Ag/AgCl reference electrode, a platinum counter electrode and the coated Mg substrate as the working electrode with an exposed area of 1 cm 2 . Before EIS measurements, OCP measurements were performed firstly, and EIS measurements started only when the OCP values kept stable. In addition, the EIS measurements were performed in 3.5 wt% NaCl solution after different immersion times with a sine signal with an amplitude of 10 mV RMS v.s. OCP. The range of measured frequencies extended from 10 MHz to 100 kHz, with a logarithmic sweep of 7 points per decade. The details of characterization steps can be found elsewhere 27,28,30,47 . In order to ensure the reproducibility, there were three parallel samples in each system. Figure 2 shows SEM surface morphologies and corresponding EDS mappings of PDC, SMV and AS specimens. After impregnation of inhibitors, the surface morphologies for all of specimens did not significantly change, and they still revealed a characteristic surface morphology of the PEO layer, composed by pores with different size and little micro-cracks 7,30 . The formation of such a typical morphology was caused by the gas evolution through the molten oxide during the PEO process and the thermal stresses at the sites of the discharge channel 43 . Usually, electrolyte composition 51 , temperature of electrolyte 52 , concentration of electrolyte 53 , applied energy (voltage and current density) 54,55 , additives (such as TiO 2 , CeO 2 ) 56,57 and secondary phases in the alloy 43 play important roles in morphologies of PEO coatings including dimensions and distribution of pores and micro-cracks, and further resulting in different performance of PEO coatings. Because of the surface contamination with entrapped and adsorbed CO 2 and N 2 , it was difficult to determine from EDS mapping if PDC or AS inhibitors had been impregnated successfully into pores of PEO coatings. However, it can be seen clearly from SMV specimens that, there were light spots, which means high element content, in EDS mapping of V. Figure 3 reveals the spectra of point EDS analysis for SMV specimen. From the inserted table in Fig. 3, it shows that the content of V was up to 5.80 at%. All of above results indicate that inhibitors could be successfully impregnated into porous PEO coatings through a vacuuming procedure.

Results and discussion
OCP vs. time measurements in 3.5 wt% NaCl solution after impregnation of inhibitors are shown in Fig. 4. The OCP of PEO specimen was found that the value decreased at the beginning of immersion, which may result from deterioration of PEO coating in such a strong aggressive medium. Subsequently, the OCP value increased gradually (from point 1), due to deposition of corrosion production, likely Mg(OH) 2 . However, it showed a small www.nature.com/scientificreports/ fluctuation in the OCP value during whole immersion process, which was significantly different from others. Although the OCP values for all of impregnated inhibitor specimens were more negative than PEO specimen in the beginning, continuous and fast potential displacements towards positive values were observed and eventually higher than stable OCP of PEO specimen after a longer immersion term (shown in the insert of Fig. 4). When the immersion time was further prolonged, the OCPs dropped suddenly from point 2, point 3 and point 4 respectively (marked in Fig. 4) and became fluctuating, similar to that of PEO specimen. These results indicated the inhibiting effect of PDC, SMV and AS inhibitors for Mg alloys, while it was very limited when impregnated inhibitor specimens were not sealed. As for inorganic inhibitor SMV have been proved that it could reduce the kinetics of oxygen reduction to an extent similar to chromates, and a vanadium-based adsorbed layer may form to block reactive sites on intermetallic particles and discourage the oxygen reduction reaction 47,58 . Actually, as for Mg alloy, water reduction is not the only cathodic process and oxygen reduction is the albeit secondary cathodic process during Mg corrosion. Unusually, as for organic inhibitors, either their ligand anions could form stable complexes with cations (e.g. Mg 2+ in our case) or they could adsorbed physically (electrostatic interaction) between carbonyl functional groups and cations (e.g. Mg 2+ ) on metallic surface 59,60 . In Lamaka's study 15 , as for PDC and AS, it showed almost the same inhibiting efficiency (eg. 83 ± 11% vs. 99 ± 1% for PDC, 86% vs. 85% for AS) for the higher Fe content (CP-Mg-220 ppm) and the lower Fe content (HP-Mg-50 ppm) purity Mg alloys. This result implied that PDC and AS inhibitors mainly interacted with Fe to suppress Mg corrosion. As suggested by their theory 14,61 , at initial cathodic reactions on Fe-rich particles, detachment of Fe particles from the Mg metallic substrate due to anodic    62,63 . Actually, although we briefly discuss the inhibiting mechanism of PDC and AS organic inhibitors, it is more complicated and not yet fully clear so far.
Bode EIS plots of PDC, SMV and AS specimens, shown in Fig. 5, were recorded after 20 min, 1 h and 3 h immersion in 3.5 wt% NaCl solution. It can be seen that two well defined time constants exist in the frequencyphase angle diagram, corresponding PEO coatings and corrosion process. In general, the low-frequency impedance modulus (|Z| 0.01 Hz ) reveals the overall resistance of the protective coating 64,65 . Initially, the corrosion protection of PEO specimen was superior to that of impregnated inhibitor specimens. Nevertheless, its protection performance dropped faster with immersion time, and eventually became the worst in corrosion protection (|Z| 0.01 Hz = 1.4 × 10 3 Ω cm 2 ). SMV was found to be most efficient (|Z| 0.01 Hz = 5.6 × 10 3 Ω cm 2 ), in agreement with OCP vs. time result where final OCP value of SMV was noble with respect to the other two. According the EIS results, in summary, impregnated inhibitor did not obviously improve corrosion resistance of PEO specimen, which may attribute to easy loss of inhibitors from pores without sealing post-treatment.
After sealing treatment in containing Ce solution with short time, SEM surface morphologies of PDC/Ce, SMV/Ce and AS/Ce specimens are presented in Fig. 6. Although the initial morphological characteristics of PEO specimen were still visible (recall Fig. 2), the initial pores of PEO layer were sealed well. But micro-cracks were clearly observed, especially in Fig. 6a, which was obviously not due to PEO treatment and they were new formation during sealing treatment process. Hu et al. 66 suggested that the micro-cracks only appeared at the thick film, not at the thin film, which is not consistent with our results. This contradictory result seems to be because high temperature and short reaction time was employed in our work for avoiding loss of impregnated inhibitors. In our previous study, We have proposed growth mechanism of cerium conversion coating on PEO layer, i.e., the Ce solution could penetrate into the pores of the PEO layer to precipitate of CeO 2 or/and Ce(OH) 3 at the beginning of the treatment, followed by the pores were covered with spherical deposits, and as a result a cracked ''dry mud-like" surface morphology formed 30 . As suggestion, the faster deposition rate caused large stress existed at the PEO coatings, and stress relieve resulted in formation of micro-cracks 67 . A generally accepted opinion, Ce salts precipitated on Mg surface typically showing positive protective performance for Mg alloys 30 . Table 1 lists elemental compositions of PDC/Ce, SMV/Ce and AS/Ce specimens, acquired by the average calculation of the  www.nature.com/scientificreports/ composition of five points on their surface. Higher content of O for PDC/Ce and AS/Ce specimens compared with SMV/Ce specimens and the presence of V for SMV/Ce specimens all indicated that impregnated inhibitors were stored and sealed successfully in porous PEO coatings. Figure 7 shows XRD patterns of PDC/Ce, SMV/Ce and AS/Ce specimens. It was found that Ce mainly exists in the form of CeO 2 . Researchers also detected Ce in the 3-valence state (Ce(OH) 3 ) after treatment in a Ce containing solution 67,68 . There was no obvious diffraction peaks corresponding to Ce(OH) 3 , attributed to the fact that the the Ce(III) hydroxide may present in an amorphous or nano-crystalline form. The transform and formation process can be expressed by following reactions 30 : Figure 8 shows the cross-sectional SEM images and corresponding EDS mappings of PDC/Ce, SMV/Ce and AS/Ce specimens. According our data, the thickness of PEO layer was about 5.0 ± 0.4 μm (not shown here). Thus it can be seen that treatment in containing Ce solution did not significantly change the thickness of PEO coatings (4.0 ± 0.6 μm for PDC/Ce, 4.5 ± 1.0 μm for SMV/Ce and 4.2 ± 0.8 μm for AS/Ce). Furthermore, it was found that all of coatings were relative thin and non-uniform, which is not beneficial to improve of corrosion resistance. From EDS mappings, Mg, Al and O were present throughout the oxide thickness down to the interface between substrate and oxide, while the position of rich Ce closed to outer layer. So far, the inhibiting effect of Ce toward Mg alloys is still in debate. As suggested by Lamaka et al. 15 , CeCl 3 would accelerate corrosion for most alloys, e.g. WE43, ZE41, E21, AZ31, AZ91 and AM50, while a completely opposite result obtained from Ce(NO 3  www.nature.com/scientificreports/ on AZ63, AM50 and AM60 15,69,70 . Hence, some of them proposed that the inhibiting effects of Ce(NO 3 ) 3 were resorted to the presence of nitrate, rather than Ce ions. However, in our published work, it has been proved that presence of Ce ion in coatings was believed to endow the coating with active corrosion protection, but its protective effect was still limited 30 . After fabrication of MgAl-LDHs on PDC/Ce, SMV/Ce and AS/Ce specimens, the SEM surface morphologies are revealed in Fig. 9. The original PEO layer completely fragmented, resulting in the formation of island-like structure covered by tiny nanoflakes, and thus original morphology of PEO layer was still faintly visible. These tiny nanoflakes was generally considered to be the characteristic structures of LDHs in many research cases [36][37][38][39][40][41] . It is worth mention that Mg(OH) 2 reveals a layered structure as well, attributed that MgAl-LDHs unusually originates from the transformation of Mg(OH) 2 or Al(OH) 3 48 . In the strict sense, it's not accurate to confirm presence of LDHs through only SEM image. Thus, XRD measurement was further carried out and the result was shown in Fig. 10. In order to clarify if impregnation of inhibitors has an influence (positive or negative) on subsequent LDHs growth, a reference specimen was produced via PEO specimen with Ce solution sealing treatment, followed by preparation of LDHs on them (without impregnation of inhibitors into porous PEO coatings) and the XRD result of this reference specimen is also shown in Fig. 10. The diffraction peaks associated with impurity phase Mg(OH) 2 were visible. In addition to Mg(OH) 2 , all of specimens appeared the characteristic   71 studied organic chelating agent-assisted in situ LDH growth on AZ31 Mg alloy. As suggested by their theory, the pH range mostly suitable for MgAl-LDH formation was between 9 and 10, while an excess of OHdeveloped in the solution when pH increased to above 9.5, resulting in the formation of Mg(OH) 2 . After addition of chelating agent EDTA, the concentration of Mg-EDTA complex (MgC 10 H 12 N 2 O 8 2− ) reached its maximum at a pH from 7 to 10.5. Thus, the introduction of EDTA delayed the formation of Mg(OH) 2 by 0.8 pH units, i.e., only when the pH exceeds 10.3 could form Mg(OH) 2 (9.5 for that without chelating agent). Similar to EDTA, ligand anions would compete with OHions at a pH from 4.5 to 13.8 when chelating agent NTA was introduced, resulting in formation of Mg-NTA complex (Mg(CH 6 O 6 N) 2 4− ). In a word, they proposed that formation of MgAl-LDH could be promoted via the maintenance of free Mg 2+ or Mg reversibly bound in Mg-ligand complexes at pH exceeding typical precipitation range of Mg(OH) 2 . Back to our work, it has been previously shown that ligand anions of PDC and AS could form stable complexes with metal cations (e.g. Mg 2+ and Fe 3+ ) 62,63 . Therefore, there was reason to believe that a similar effect has taken place in PDC and AS, which also would further promote the formation of LDHs. And the lowest peak intensity of inorganic inhibitor SMV, compared with PDC and AS, also indirectly prove our opinion. However, further experiments are necessary to verify our guess. Figure 11 shows the cross-sectional SEM images and corresponding EDS mappings of PDC/Ce-HT, SMV/ Ce-HT and AS/Ce-HT specimens. Different from Fig. 8, the results of EDS mapping clearly revealed that all composite coatings were consist of typically two-layers structure, which is consistent with our previous result 65 . The evolution of average thickness of the coating obtained by the cross-sectional SEM micrograph is plotted in Fig. 12. It is clearly seen, that the thickness of outer layer of PDC/Ce-HT, SMV/Ce-HT and AS/Ce-HT specimens was almost the same as respective thickness of original specimens without preparation of LDHs. This result implied that severe damage of PEO layer did not take place during twice hydrothermal process. In addition, the thickness of coatings significantly increased after fabrication of LDHs, with the thickest one being 11.7 μm (SMV/Ce-HT). Figure 13 presents Bode EIS plots of PDC/Ce-HT, SMV/Ce-HT and AS/Ce-HT specimens, recorded after 3 h, 24 h and 72 h immersion in 3.5 wt% NaCl solution. According to the low-frequency impedance modulus (|Z| 0.01 Hz ) at initial immersion stage (3 h), the corrosion resistance can be ranked in increasing order as follows: PDC/Ce-HT (|Z| 0.01 Hz = 3.5 × 10 3 Ω cm 2 ) < SMV/Ce-HT (|Z| 0.01 Hz = 3.3 × 10 4 Ω cm 2 ) ≪ AS/Ce-HT (|Z| 0.01 Hz = 1.3 × 10 8 Ω cm 2 ). From frequency-impedance modulus diagram, as for organic inhibitor PDC and AS, the impedance modulus in both higher frequency and lower frequency slightly increased (from 3 h immersion to 6 h immersion) and reached a maximum after ca. 6 h immersion, which may be correlated with impregnated inhibitors resulting in formation of renewed production, resulting in a sealing effect to corrosion defects and active corrosion of LDHs. In contrast, the impedance modulus of SMV/Ce-HT specimens dropped faster from 3 to 6 h immersion. The immersion time was further prolonged to 72 h, all of specimens decrease in the impedance modulus. In summary, in terms of overall corrosion resistance, AS/Ce-HT specimen was best. As a result, equivalent circuits, shown in Fig. 14, were employed to further fit the electrical parameters of AS/Ce-HT specimen. Due to the nonhomogeneity, caused probably by heterogeneity of chemical composition, the degree of crystallinity and stoichiometry, and the presence of the space-charge region, the constant phase elements (CPEs) were used herein to demonstrate the non-ideal resistive and capacitive behavior of the coated specimens 72,73 . In the equivalent circuit, R sol corresponds to the solution resistance; R coating and CPE coating are the resistive and capacitive response of the coating. When the aggressive media (such as Clin our case) diffuses through coating and reaches to the interface between coating and substrate, Mg substrate will be corroded. The corrosion process of   Figure 15 shows nyquist plot of AS/Ce-HT specimen. At initial and medium immersion stage (3 h and 24 h), two capacitive arcs corresponding to two well-defined time constants were observed. Therefore, the equivalent circuit shown in Fig. 14a was used to fit resistive and capacitive of the coating and corrosion process. After immersion of 72 h, there were only one capacitive arc left and its radius dropped fast, indicating the coating barrier effect completely lost at this time. Therefore, only one resistance and capacitance in parallel circuit (Fig. 14b) was used to fit resistive and capacitive of corrosion process. The fitted result is revealed in Fig. 15 as well and the chi-square χ 2 for all fitting lines was reached less than 10 -2 . Table 2 lists the fitting electrical parameters of AS/Ce-HT specimen. Compared with our previous data, this kind of coating own a higher R ct value at initial     Table 2. Fitted parameters for EIS spectra of AS/Ce-HT specimen depicted in Fig. 15.

Conclusion
(1) Corrosion inhibitors PDC, SMV and AS were successfully impregnated into the micro pores of PEO layer through a vacuuming procedure, however, impregnated inhibitor did not obviously improve their corrosion resistance. (2) After sealing treatment with short time in Ce containing solution, the formation of new phase, CeO 2 , sealed inhibitors stored pores of PEO, while a small amount of inhibitor also lost in this process. (3) When LDHs was prepared on PEO layer sealed by Ce conversion coating, resulting in formation of islandlike structure covered by tiny nanoflakes of LDHs. And, the structure of the coatings varied from one layer to two layers and the thickness significantly improved, with the thickest one being 11.7 μm (SMV/Ce-HT). Moreover, it was found that impregnation of inhibitors into porous PEO coatings has played a positive role in improving LDHs content of final coatings. (4) Eventually, EIS result indicated that AS/Ce-HT exhibited a superior corrosion protection, better than other coatings. In addition, further loading inhibitors into LDHs can be considered as a next step.