Use of synergistic mixture of chelating agents for in situ LDH growth on the surface of PEO-treated AZ91

The principal possibility to grow layered double hydroxide (LDH) at ambient pressure on plasma electrolytic oxidation (PEO) treated magnesium alloy AZ91 in the presence of chelating agents is demonstrated for the first time. It avoids hydrothermal autoclave conditions, which strongly limit wide industrial application of such coating systems, and the presence of carbonate ions in the electrolyte, which lead to the formation of “passive” non-functionalizable LDH. A combination of chelating agents (sodium diethylenetriamine-pentaacetate (DTPA) and salicylate) were introduced to the treatment solution. The role of each additive and the influence of treatment bath composition on the LDH formation processes are discussed. A synergistic effect of DTPA and salicylate during LDH formation is discovered and its possible explanation is proposed.

Plasma electrolytic oxidation (PEO) treatment is an effective way to improve corrosion and wear resistance of Mg alloys due to the formation of ceramic oxide layers [1][2][3][4][5][6] . The main advantages of PEO treatment as compared to classical anodization are related to the higher current and voltage applied, introducing micro-discharges, thus changing phase composition, increasing layer thickness and improving properties. Moreover, PEO treatments are normally performed in diluted alkaline electrolytes 7,8 . Thus, the process is significantly more environmentally friendly in comparison with conventional chromic acid anodizing, previously widely used for corrosion protection and forbidden now due to the high toxicity for humans and for environment 9 .
In previous publications it was shown, that corrosion behavior and wear resistance of PEO-coated magnesium alloys depend strongly on the treatment conditions, such as voltage, current density, treatment time and electrolyte composition 2,10-12 . However, since micro-discharges are responsible for coating formation, the produced PEO coatings always contain a number of discharge channels, pores and micro-cracks. These defects strongly compromise the corrosion resistance of PEO layers, because the corrosive solution can easily penetrate through them directly to the metallic interface. Thus, sealing the pores is an effective way to improve long-term corrosion resistance of PEO-treated magnesium alloys. Conventional sealing approaches traditionally include application of organic polymers 13,14 and compounds 15 and/or different post-treatment in electrolytes containing lanthanum 16 , silane 17 or cerium 14,18,19 .
In the last years a lot of attention has been paid to self-healing coatings, which do not only act as barrier layer between the metal and corrosive media, but also release corrosion inhibitors in the case of mechanical damage of the coating. For this purpose, layered double hydroxides are promising materials as they can be effectively intercalated with suitable corrosion inhibiting anions [20][21][22][23] . It was shown, that LDH films can be successfully loaded for example with vanadate 24-26 , molybdate 27 or mercaptobenzothiazolate anions 28,29 via anionic exchange processes. Up to now, a number of studies demonstrated the possibility of sealing PEO pores on aluminum alloys by in situ hydrothermally grown LDH nanocontainers under ambient pressure conditions. In these studies LDHs are produced via one-step processes at 95 °C in an aqueous electrolyte with a substrate acting as a source of Al 3+ ions

Materials
The following materials were used for PEO synthesis: potassium hydroxide (KOH, ≥ 85%, ChemSolute, Germany), sodium phosphate (Na 3 PO 4 , ≥96.0%, pure, anhydrous, ACROS Organics, Germany), sodium aluminate (NaAlO 2, (Al 2 O 3 50-56%, Na 2 O 40-45%, Fe 2 O 3 ≤ 0.05%), anhydrous, techn., Sigma-Aldrich Laborchemikalien GmbH, Germany). Magnesium alloy AZ91 was used as a substrate for PEO processing followed by LDH growth. Methods. PEO preparation. First, AZ91 alloy coupons with a size of 15 × 15 × 4 mm were ground with SiC paper up to 1200 grit, rinsed with deionised water and dried with a stream of cold air. PEO processing of the specimen was conducted for 11 min with a bipolar power supply pe861-UA 500-10-24-S (plating electronic, Sexau, Germany) under the following conditions: current control mode with positive pulses of 3 A, pulse ratio of t on : t off = 1 ms: 9 ms. The resulting voltage at the end of the PEO treatment was ~460 V. The electrolyte contained 1 g/L KOH, 8 g/L Na 3 PO 4 and 12 g/L NaAlO 2 dissolved in deionized water. The specimen was put into the electrolyte bath with constant stirring and kept at 20 ± 2 °C by a water cooling system. The counter-electrode was made of stainless steel. Finally, the specimens were rinsed in deionized water and dried in warm air.  Characterization. Crystallographic structure of the coatings was studied using Bruker D8 Advance diffractometer (Karlsruhe, Germany) with Cu Kα radiation. The scans were performed in the range of 2 Theta from 5 to 80° (exposure time 1 s, step 0.02°). High resolution two dimensional XRD maps were acquired at the nanofocus end station of the beamline P03 at PETRAIII storage ring (DESY, Hamburg, Germany) 51 . An Eiger 9 M (pixel size 75 µm by 75 µm) and beam size of 1.5 by 1.5 µm was used. The scanning step was 2 µm in vertical direction and 4 µm in horizontal direction, with an acquisition time of 0.5 sec for each point for a XRD pattern. The measured area was 80×80 µm. The energy of the X-rays was 19.7 keV. The horizontal axis of the diffraction patterns were recalculated to Cu K-alpha radiation.
Scanning electron microscope (Vega 3, Tescan, Brno, Czech Republic) equipped with energy dispersive X-ray (EDS) spectrometer (EDS, Heidenrod, Germany) was used to evaluate the surface morphology, cross-sections and composition of the coatings. The cross-sections were prepared using standard metallographic techniques. Glow discharge optical emission spectroscopy (GDOES) was used to investigate the concentration depth profile of the coatings (GD-Profiler 2, HORIBA, Longjumeau, France) with an anode of 4 mm in diameter, at an Ar pressure of 650 Pa and 30 W power.

Results and Discussion
The acceleration effect of different chelating agents on LDH formation on bare AZ91 alloy is related to the increase of soluble Mg 2+ concentration in the pH range favorable for LDH growth 45 . On the other hand, PEO coatings containing metal oxides can also serve as a potential source of soluble Mg 2+ species. Previously NTA was found to be the most promising chelating agent for bare AZ91 alloy 45 . In this study, for PEO treated AZ91 DTPA was chosen instead of NTA as an environmentally friendly alternative with similar chelating properties. The selection was done based on thermodynamic calculations performed with Hydra-Medusa software ( Fig. 1) 49 . The stability constant of Mg-DTPA complex (7.90-10.70) depends on the ionic strength of the electrolyte 52,53 . In the frame of current work, a value of 9.03 was used as stability constant 54 , which is close to that of NTA (5.50 and 10.20 for ML and ML 2 complexes, respectively).
Moreover, at pH range 9-11 suitable for LDH formation 28 , following could be observed: 1. Aluminium is present in the same form as it was predicted for successful LDH formation. Its solubility is not differ from one in the presence of NTA 45 . 2. ca. 50% of free magnesium cations are present in the range of possible LDH formation (pH range from 9 till 11 indicated in Fig. 1) and available for the reaction. 3. Phosphates containing in PEO do not show a significant effect on the concentration of Al 3+ and Mg 2+ in the pH range suitable for LDH formation (indicated in Fig. 1).
As a result, the conditions for Mg,Al-LDH growth are satisfied: the required concentration of aluminum and magnesium cations are reached in the preferential pH range of LDH formation (pH 9-10) 26 .
However, the direct replication of LDH synthesis method on PEO treated AZ91 in the presence of DTPA as chelating agent was not successful -PEO layer was fully etched (Fig. 2b). This observation is confirmed by XRD results (Fig. 2a) -MgAl 2 O 4 spinel, forming PEO layer on magnesium alloy, could not be detected anymore after the hydrothermal treatment in the solution, containing DTPA. One can also see, that typical pores PEO structure can not be identified via SEM imaging. Instead of it, the grains of bare magnesium alloy are visible.
Different approaches were used in order to influence the kinetics of PEO dissolution and LDH formation:  It was noticed that in the case of PEO-treated AZ91 and in the absence of external Mg source, a complete etching of the PEO layer in the DTPA-containing solution took place in any case (regardless of temperature, DTPA concentration and treatment time). The typical XRD patterns of initial PEO sample before and after treatment with DTPA solution containing only additional Al(III) species are shown in Fig. 2a.
According to XRD results, the PEO coating consists mainly of MgAl 2 O 4 spinel. This result is in agreement with previously published data for Mg-PEO systems formed in aluminate electrolytes [55][56][57] . However, the respective XRD peaks of PEO layer, which correspond to MgAl 2 O 4 , disappear completely after 3 h of treatment in a DTPA containing bath, while no LDH is formed yet. No respective diffraction peak is visible in the range of 9.5-11.5°, typical for (003) LDH peak. One can conclude that, on the one hand, the concentration of Mg 2+ required for LDH formation is not high enough (since the additional source of magnesium helped to overcome the problem) and, on the other hand, that the PEO layer needs to be protected against etching in the presence of such complexants. On the first view, these requirements are contradicting, but a new concept was developed to overcome this problem: 1) 0.5 g of AZ91 flakes were added to the reactive electrolyte as an additional source of magnesium; 2) Salicylate was used as it forms adsorption layer on PEO surface significantly slowing down etching effect characteristic for DTPA 58 . 3) Use of mixture of salicylic acid and DTPA as additives, with an assumption, that DTPA acts as strong Al 3+ and Mg 2+ complexing agent hence promoting LDH formation, while SA readily adsorbs on PEO surface and weakens its etching by DTPA. It is known that the dissolution rate of metal (hydr)oxides can be reduced in presence of several complexing agents, even though each of them individually promotes the dissolution due to the formation of surface metal-chelate complexes and surface screening due to adsorption [59][60][61][62] .
The SEM micrographs (top and cross-section view) of the PEO coatings before and after suggested treatments with chelating agents are shown in Fig. 3. It can be seen, that the concept of synergistic DTPA and SA action is the most effective for LDH formation. For instance, in the case of SA (Fig. 3c,d), only a few LDH-like flakes are formed on the surface, while the pores remain LDH-free. In the presence of both DTPA and AZ91 flakes, the sealing of PEO pores with Mg,Al-LDH occurs. However, there is a considerable difference in the morphology of PEO-D and PEO-DS samples. In the presence of additional salicylate, the topography of LDH-covered sample resembles initial PEO coating, while DTPA without SA leads to deterioration of the top PEO layer (partial etching). Without salicylate present, PEO layer is strongly deteriorated by DTPA, however, some of it still remains on the surface of the alloy. This can be explained by the presence of AZ91 flakes, which act as additional source of Mg(II). In the case when no AZ91 flakes were present in the treatment bath, complete etching of PEO layer took place after 3 h of treatment (Fig. 2). In presence of AZ91 flakes and DTPA, relatively separated LDH flakes are formed on the PEO layer (Fig. 3e,f).
Finally, in the presence of both DTPA and SA the topography of LDH-covered sample resembles initial PEO coating with homogeneous and continuous flake-film (Fig. 3g,h). Therefore SA, although it does not cause a substantial LDH formation, has a significant effect on the structure and morphology of the resulting coatings.
XRD patterns of PEO samples which underwent the LDH treatment in the presence of different chelating agents and AZ91 flakes are shown in Fig. 4. It can be seen, that the spinel part (characteristic peak of MgAl 2 O 4 is  45 and observed during SEM analysis (Fig. 3), confirming that salicylate is unlikely to promote LDH formation on AZ91-based systems (Fig. 4, PEO-S sample). However, simultaneous introduction of both DTPA and SA to the treatment bath positively influences LDH formation. In the case of PEO-DS sample, second LDH peak (006 reflection) at 23.3° is clearly visible and the relative intensity of both LDH peaks ((003) and (006) reflections) against spinel PEO peak increases as compared to PEO-D sample. These peaks are related to hydroxide-loaded LDH, which is quite typical for LDH grown on various magnesium alloys 45,63,64 . The broad peak around 14° at PEO and PEO-S patterns corresponds to thin nanocrystalline surface film on PEO layer and is not related to LDH phase.
The hybrid PEO/LDH coating system, formed as a result, was studied with GDOES in order to confirm that LDH is formed in the pores of PEO coating and not on the surface only. Depth profile analyses for the main elements across the coating for all the investigate samples are shown in Fig. 5 Overall, two main regions (zones I and II) were found on depth profiles of PEO coatings. The first zone with a relatively high content of aluminum, magnesium, oxygen and phosphorus corresponds to the PEO layer. The second zone is characterized by a higher level of aluminum and magnesium and corresponds to the AZ91 substrate. www.nature.com/scientificreports www.nature.com/scientificreports/ The thickness of PEO layer decreases after treatment with DTPA-containing solutions, which confirms partial dissolution of the PEO coating due to the etching by DTPA chelating agent.
In contrast to the initial PEO, PEO-D and PEO-DS samples are characterized by relative increased hydrogen signal throughout the PEO layer. This can be associated with LDH formation (source of hydrogen) not only on the surface, but also inside the PEO pores and their sealing. In the case of PEO-DS sample, the relative intensity of hydrogen signal is even higher. This corresponds well with the XRD data and proves that addition of SA increases the amount of formed LDH. In contrast, for the PEO-S sample almost no increase in the hydrogen signal is observed, confirming that there is no LDH formed inside of PEO pores. Unfortunately, based on GDOES results it is not possible to judge, which compounds (organic chelating agents, inorganic LDH species, etc.) are exactly  www.nature.com/scientificreports www.nature.com/scientificreports/ responsible for this hydrogen signal. In order to understand the mechanism of LDH formation on PEO treated magnesium alloys in detail, additional investigations will be required.
Another characteristic feature of PEO-D and PEO-DS samples in comparison with parental PEO is a significant decrease of phosphorus and magnesium signals within PEO layer, while Al signal remains at the same level. This can be explained by preferential interaction of DTPA with phosphorus-rich part of PEO layer (most likely amorphous Mg 3 (PO 4 ) 2 , which is not visible in XRD patterns, but was detected in similar coatings 65,66 ) rather than with spinel phase. However, the reaction between DTPA and spinel part of PEO takes place as well, which can be seen from the aluminum distribution profile of the PEO-D sample. The intensity of Al signal decreases at the beginning and at end of zone I which represents PEO layer. Therefore, a conclusion can be made, that DTPA penetrates the pores and promotes the dissolution of the inner (barrier) layer of the coating as well as the surface. This suggestion is in a good agreement with our first results, when no PEO layer was seen on the magnesium surface after LDH synthesis without addition of AZ91 flakes and salicylic acid in the solution: the flaking off of the coating was strongly supported by destruction of inner PEO layer.
This process, however, might be inhibited by protective effect of salicylate. It can be seen via the absence of such or much weaker decrease of aluminum signal in presence of SA (PEO-S and PEO-DS samples) in the GDOES elemental distribution profiles. In other words, the introduction of salicylate to the reaction bath changes the mechanism of PEO layer etching and/or changes the preferentially soluble components from the layer. This is likely due to the adsorption of salicylate on the bottom of PEO pores preventing or reducing the destruction of barrier layer by DTPA.
Overall, the mechanism of PEO-chelating agent interaction and subsequent LDH formation is yet not fully understood. According to GDOES data, the content of phosphorus in the PEO layer decreases after treatment with DTPA containing solutions. This was additionally confirmed by EDX mapping to determine the elemental distribution in the coatings (Fig. 6). It can be seen, that spinel part of the coating, which contains magnesium-aluminum oxide, remains almost intact after DTPA and SA treatment. This result is in good agreement with XRD measurements (Fig. 4, spinel peak at 18.9°). As demonstrated by GDOES data, aluminum containing part of PEO layer tends to dissolve from the bottom of pores as a result of interaction with DTPA, which does not happen in presence of salicylate. This can also be seen from EDX mapping: in the case of parental PEO www.nature.com/scientificreports www.nature.com/scientificreports/ and the samples treated in the presence of SA, thin layer with high concentration of aluminum is located at the bottom of a pore. In contrast, for PEO-D sample the pore bottom is depleted of aluminum. This might be due to adsorption of SA on the surface of oxide layer. In the presence of SA, the concentration of Al-DTPA complexes on the surface of PEO layer should decrease, and the dissolution of spinel slows down.
However, the phosphorus containing part of the coating almost disappears as a result of interaction with DTPA. The introduction of salicylate changes this behavior and therefore keeps PEO layer almost intact. Most likely it also preserves its mechanical and barrier properties after sealing with LDH, but this will be investigated additionally in future studies.
Synchrotron XRD measurements were conducted to investigate the LDH distribution in pores of PEO coatings. For this a two-dimensional maps of the sample at the air/PEO/substrate interface were created for each sample (Fig. 7). From the diffraction patterns, the intensity of the (003) LDH peak was extracted and a two-dimensional maps showing the spatial intensity variation were composed (Fig. 8). The maps of LDH distribution correspond well with XRD and GDOES data. No LDH phase is present in the case of initial PEO and only trace amounts of it can be seen at the map of PEO-S sample (Fig. 8A,B, respectively). In contrast, for PEO-D and PEO-DS samples the formation of LDH phase throughout the PEO coating is confirmed (Fig. 8C,D). In the case of the PEO-D sample LDH is formed within the PEO layer, and the intensity of the LDH signal increases closer to AZ91 interface. This is confirmed by GDOES data on partial dissolution of inner layer of PEO coating (Fig. 5c), which is likely to make it less stable with a tendency to flake off. The relative intensity of LDH signal for PEO-DS sample is higher, which proves that more LDH is formed in the presence of both DTPA and SA.
Overall, as the next step, a deeper investigation of the mechanisms of PEO interaction with different chelating agents and the optimization of LDH sealing conditions will be conducted. Furthermore, the functionalization of formed LDH nanocontainers with different species (e.g. corrosion inhibitors) and testing of formed hybrid systems for possible industrial applications are necessary. Beyond the frame of this work, we got a preliminary confirmation that this approach can also be extended for other magnesium-based substrates, such as PEO-treated AZ31 alloy.

conclusion
The principal possibility of LDH growth on PEO-treated magnesium alloy AZ91 under ambient pressure is demonstrated in this work. In order to avoid the hydrothermal autoclave conditions, organic additives (DTPA and SA) were successfully added to the treatment bath. It was shown that sealing of PEO pores with LDH nanocontainers is possible. The suggested mechanism includes formation of Me-ligand complexes during DTPA-assisted partial dissolution of PEO layer. A synergistic effect of SA addition on the surface morphology of  www.nature.com/scientificreports www.nature.com/scientificreports/ complex PEO-LDH coating was discovered. It was explained via the formation of salicylate adsorption layer on PEO surface that hampers access of DTPA to PEO surface and weakens its dissolution. The results of this work are promising for further development of LDH-sealed porous coatings on magnesium alloys, their further functionalization and industrial application.