Influence of the Treatment Time on the Surface Chemistry and Corrosion Behavior of Cerium-Based Conversion Coatings on the AZ91D Magnesium Alloy

The aim of the present work was to investigate the effect of the treatment time on the surface chemistry and corrosion behavior of cerium-based chemical conversion coatings on the AZ91D magnesium alloy. The conversion coating was prepared by the immersion technique from a bath consisting of 0.05 mol.L -1 Ce(NO 3 ) 3 .6H 2 O and 0.254 mol.L -1 H 2 O 2 (30 wt.%) for times ranging from 20 s to 120 s. The surface chemistry was examined by X-ray photoelectron spectroscopy (XPS). The corrosion behavior was assessed by electrochemical impedance spectroscopy and potentiodynamic polarization. XPS analysis detected the presence of cerium oxides (Ce 2 O 3 and CeO 2 ) and cerium/ magnesium hydroxides. The best corrosion behavior was observed for the treatment conducted for 60 s. The results are discussed with respect to coating morphology and composition.

The corrosion resistance of magnesium alloys with cerium-based conversion coatings has been evaluated by several authors [17][18][19][20] . Protective coatings have been developed in order to overcome this limiting design feature of magnesium alloys. Different deposition methods may be employed such as electrodeposition, physical vapor deposition, anodization and chemical conversion coatings 1 . The conversion process is usually conducted by immersion at temperatures below 50°C in the presence of an oxidant, resulting in the formation of hydrated species or mixed oxides 21 .
Chromates based conversion coating has good efficiency, but its toxicity has led to legislation restrictions, driving the development of non-aggressive treatments 5 . Cerium-based conversion coatings have gained growing interest in the corrosion control of magnesium alloys. Rare-earth elements act as cathodic inhibitors. Cerium is the most active of the rare-earth elements. The corrosion resistance follows the order Ce>Nd>Pr>Y>La 21 . The main action mechanism of rare earth elements is by the precipitation of oxides/ hidroxides on substrate cathodic regions. Coating integrity and thickness determine its protection. Some cracks appear in the rare earth-based conversion coatings [21][22][23] .
It is generally agreed that coating morphology has a marked influence on the overall protection ability of the converted surface film. Moreover, surface chemistry is also an important issue. X-ray photoelectron spectroscopy (XPS) is frequently used to examine the chemical composition of cerium conversion films 15,18,[24][25][26][27][28] . It provides unique information regarding the chemical states of different species that constitute the converted layer. Cerium compounds can be found in two different oxidation states, namely Ce 4+ and Ce 3+ 24, 25 . The treatment time can be varied to tailor the atomic fraction of each one of these components in the conversion layer. The corrosion behavior of the treated substrate is, therefore, affected by this parameter.
The aim of this work was to investigate the effect of the treatment time on the corrosion behavior of cerium-based conversion coatings developed on the AZ91D magnesium alloy. The correlation between the surface chemistry of the converted layers, their morphology and corrosion resistance is examined.  Table 1. Specimens were cut from the ingot using silicon carbide blades, thus obtaining pieces with approximately 7 x 6 x 5 mm. Next, the specimens were connected to a copper wire using colloidal silver paste, followed by embedding in cold curing epoxy resin. After curing, the specimens were mechanically ground using silicon carbide emery papers up to 2400 grit. The final surface finishing step was polishing by diamond paste (6 µm), being subsequently, degreased with ethanol, rinsed in deionized water and dried under a warm air stream provided by a conventional heat gun.

Materials, surface preparation and conversion coating procedure
The conversion coatings were prepared by immersing the specimens in a solution consisting of 0.05 mol.L -1 Ce(NO 3 ) 3 .6H 2 O and 0.254 mol.L -1 H 2 O 2 . The treatment was carried out at room temperature for different times: 20 s, 60 s and 120 s. After immersion, the specimens were rinsed in deionized water and let dry at open air.

Characterization
Coating morphology was examined by confocal laser scanning microscopy (CLSM -Olympus LEXT OLS4100 instrument) and scanning electron microscopy (SEM -FEI Quanta 250). The surface chemistry of the converted films was analyzed by energy dispersive spectroscopy (EDS -JEOL JSM-6010LA) and X-ray photoelectron spectroscopy (XPS) using a ThermoFisher Scientific K-alpha + spectrometer operating with Al-kα radiation source. The chemical states for Ce3d and O1s were identified by analyzing the binding energies calibrated to that of C1s peak (adventitious carbon) at 284.8 eV. The spectra were curve-fitted using the Avantage TM software and background subtraction by the Smart algorithm. The coating thickness was analysed by surface profilometry (KLA -Tencor model P7).
The corrosion behavior was assessed by electrochemical impedance spectroscopy (EIS) and potentiodynamic polarization. All tests were carried out using a potentiostat/ galvanostat Autolab M101. A conventional three-electrode cell configuration was employed with a platinum wire as the counter-electrode, Ag/AgCl as reference and the AZ91D samples as the working electrodes. Initially, the open circuit potential was monitored for 1 h. Next, EIS measurements were performed at the open circuit potential in the frequency range from 100 kHz to 10 mHz with an amplitude of the perturbation signal of ±10 mV (rms) and an acquisition rate of 10 points per decade. Then, the samples were subjected to potentiodynamic polarization from -300 mV with respect to the open circuit potential up to +0.5 V Ag/AgCl at a sweep rate of 1 mV.s -1 . The tests were conducted in triplicate. Figure 1 shows CLSM micrographs of the uncoated and cerium-coated AZ91D alloy. After the conversion treatment, the samples had a good coating coverage and presented a yellow-colored surface, whose intensity was dependent on the treatment time, ranging from soft yellow to a more intense gold-colored aspect for longer periods of immersion. The yellow-colored surface is typical of the presence of Ce 4+ on the conversion layers. It scales up with the concentration of this species 21,[29][30] . SEM micrographs show the presence of several cracks and exfoliated regions on the surface film. It is possible to observe the formation of two layers as shown in Figure 2. The outer layer presents several cracks and exfoliated regions. There are also small nodular clusters distributed within it, suggesting the initial stage of a third layer formation. Microcracks and nodular clusters increase with the immersion time, but exfoliation doesn't follow the same process ( Figure 3).

Coating morphology
Literature presents many possibilities to explain microcracks formation 31 . Wang et al. 29 observed microcracks on the cerium conversion layer formed over the AZ91D alloy. The cracks were mainly concentrated on the α-Mg phase, which promoted a thicker layer growth. This increase of the α-Mg phase thickness can occur because of its lower electrochemical potential with respect to the β phase that induces a preferential deposition region. The different coating growth rates can cause stress generation and microcracks formation 29,32 . Maddela et al. 33,34 related the microcracks incidence, density, morphology and size with the dimensions of cerium particles and coating thickness. According to these authors, coating homogeneity is due to the cerium oxide particles distribution. Smaller particles result in better homogeneity. Bagalà et al. 35 related microcracks formation with stress generation during drying process, which act as stress relieve mechanism. Besides all presented possibilities for this phenomenon, the better hypothesis, in accordance with other authors, is that the lower electrochemical potential of α-Mg phase, when exposed to the conversion bath, promotes hydrogen evolution around itself, during coating formation. Thus, hydrogen blisters are formed and, when they burst, microcracks and exfoliations can be formed 32,34,36 . It was possible to see hydrogen evolution during the conversion coating formation on the surface of our samples, as showed in Figure 4.
Lee et al. 36 reported H 2 evolution is a common fact during cerium-based conversion coating treatment on magnesium alloys in the presence of H 2 O 2 . Blisters are formed due to     H 2 evolution during metal/coating interface anodic and cathodic reactions. When blisters evolution is more severe some cracks and exfoliation may appear in the top surface.
Blisters are mainly formed around α-Mg phase because of the electrochemical potential difference between this phase and the Al-rich β-phase on the AZ91D alloy. Blisters pressure increase during the process and can promote cracks and exfoliations over the α-Mg phase. EDS maps were acquired to examine the distribution of Ce, Mg and O over the coated surfaces. The results are shown in Figure 5. Oxygen and cerium are the major elements in the coating layer. Magnesium major incidence is at the cracked regions. This signal may arise from both the inner layer of the converted film and from the substrate. Sun et al. 30 reported EDS analysis of the inner layer of cerium conversion coatings on the AZ91D alloy. They have also found Mg and O at this part of of the film. Sun et al. 32 reported that the inner layer has MgO in the composition and the outer layer has Ce 2 O 3 and CeO 2 . The formation of nodular distributed over the outer layer of the coating was also observed by the same authors 30,32 , who showed that they are enriched with Ce.

Surface chemistry
XPS survey spectra of the cerium-coated AZ91D alloy are shown in Figure 6. Ce3d was detected in all the converted layers, independently of the treatment time, revealing that cerium incorporation was successful. Magnesium and oxygen peaks were also detected. Oxygen is typically associated with the presence of cerium oxides on the conversion coating,  but can be related to other oxides, like magnesium and/ or aluminum oxides. Magnesium can be either part of the converted layer or the signal can also be due to the metallic substrate. The C1s peak was ascribed to surface contamination by hydrocarbons (adventitious carbon).
XPS core levels for the Ce3d and O1s regions were obtained for the cerium-coated AZ91D alloy. A representative spectrum for each one of these regions is shown in Figure 7 and Figure  8 for the Ce3d and O1s regions of the treated samples. The simultaneous presence of two oxidation states, covalence hybridization effects and spin-orbit splitting make the Ce3d spectrum complex. The covalency hybridization effects imply in the presence of three doublets for the Ce 4+ oxidation states, designated as v-u, v''-u'' and v'''-u''', and two doublets for the Ce 3+ oxidation state, designated as v 0 -u 0 and v'-u'. A strong spinorbit splitting is observed for the 3d 5/2 and 3d 3/2 components 24-26 .  The presence of Ce 4+ in the conversion coating can be unequivocally indicated by the u''' peak at approximately 916 eV which is normally considered as a fingerprint for this oxidation state, being well isolated from the other peaks and exclusive of Ce 4+ species [24][25][26] . Peak fitting allowed calculating the relative area of Ce 3+ and Ce 4+ peaks, thus enabling to quantify the ratio of these oxidation states in the conversion layers. The results are shown in Table 2. Ce 4+ is the major species independently of the treatment time which is associated with the effect of hydrogen peroxide in the conversion bath that promotes oxidation of Ce 3+ to Ce 4+ . Ce 2 O 3 and CeO 2 presence in the layer indicates that Ce 4+ and Ce 3+ contribute to coating formation. The formation of Ce 4+ is reported to be beneficial to the corrosion resistance of the conversion film due to the passivating effect of CeO 2 25,26,[37][38][39] . Considering, though, the intrinsic uncertainty of XPS quantitative analysis 40 , it is possible to infer that the Ce 3+ /Ce 4+ ratios are very similar for all coatings and the treatment time did not affect it significantly.
The O1s core level is characterized by three components, independently of the immersion time, as shown in Figure 8. The low binding energy component is assigned to CeO 2 whereas the intermediate component is assigned to Ce 2 O 3 . The highest binding energy component can be due to hydroxide species or adsorbed water 18,27,28 .
Based on the results obtained by XPS and SEM/EDS analyses and supported by the literature 36 , it is possible to suggest a coating formation mechanism for cerium-based conversion coating produced in the present work. It has a layer by layer formation. The composition of the coating in the metal interface mainly consists of MgO as indicated in the EDS maps. The inner layer is followed by layers rich in cerium oxides, as illustrated in Figure 9.
The reactions involved in the conversion coating formation process would be:

Electrochemical behavior
Nyquist plots of the cerium-coated samples are shown in Figure 10. The data were obtained in 3.5 wt.% NaCl solution at room temperature. Magnesium and its alloys are commonly corroded in NaCl solutions, caused by the presence of Clions and its corrosion susceptibility (reactivity) 3 . The shape of the plots is similar irrespective of the immersion time, being characterized by a capacitive loop in the medium to high frequency domain whose diameter depends on the treatment time. At the lowest frequencies an inductive loop appears which is typical of adsorbed species in magnesium alloys, arising from dissolution of a metallic species, releasing its cation to solution with the formation of an intermediate species such as Mg(OH) + ads or Mg + ads 41,42,43 . This effect is associated with corrosion processes of the underlying metal and possible initiation of pitting corrosion through defects in the coating layer [44][45][46] . However, corrosion pits were not detected in the samples subjected to the electrochemical tests. The diameter of the capacitive loop is associated with the corrosion resistance of the electrode, being related to the charge transfer resistance 47,48 . In this respect, the film obtained at an immersion time of 60 s presented the highest corrosion resistance. The diameter of the capacitive loop presents an increasing trend with the immersion time up to 60 s. This trend is reversed at 120 s, that is, the corrosion resistance was found to decrease when the treatment time increased from 60 s to 120 s. This behavior seems to be related to the surface chemistry of the cerium-based films, as the Ce 3+ /Ce 4+ ratios shown in Table 2 correlate with the corrosion resistance as obtained from the EIS measurements. Potentiodynamic polarization curves of the AZ91D alloy in the as-polished and cerium-coated conditions are shown in Figure 11. The curves were obtained in 3.5 wt.% NaCl solution at room temperature. The values of corrosion potential (E corr ) and corrosion current density (i corr ) were determined from these curves by the Tafel extrapolation method and are displayed in Table 3. The lowest icorr was obtained for the 60 s condition. This result is in agreement with the EIS measurements. The sample immersed for 60 s presents the highest corrosion potential and the lowest anodic current densities. Coating thickness may affect the corrosion resistance of conversion films 41 . Surface profilometry analyses allowed determining the thickness of the converted layers as 533 nm, 846 nm and 1900 nm for the coatings obtained at 20 s, 60 s, and 120 s, respectively. There is a clear trend of increasing coating thickness as the treatment time increases which is consistent with the literature 41 . Notwithstanding, the corrosion resistance of the thickest film (120 s) was lower than that of the coating obtained at 60 s which, in turn, presented an intermediate thickness. This result suggests that coating thickness may not guarantee a more protective character of the coating against corrosion. Coating integrity can play a dominant role. The presence of more defective surfaces is reported to increase with the treatment time for different types of coatings such as hydroxyapatite layers obtained by dip coating 49 and sputtered cerium oxide films 50 This effect lead to a defective barrier against electrolyte penetration, thus leading to higher corrosion rates when the thicker, defective coating is compared with a more compact and thinner surface. In this respect, coating morphology would be of prime importance to the corrosion behavior of the conversion coatings obtained in the present work.
Additionally, our results pointed to a similar Ce 3+ /Ce 4+ ratio for all coatings ( Table 2). The influence of the treatment time was, therefore, more significant on the coating morphology and thickness than its surface chemistry. The corrosion behavior was correspondingly more dependent on the morphological features of the converted layer than on its chemical composition. Table 3. Electrochemical parameters determined from the potentiodynamic polarization curves shown in Figure 11.

Conclusions
The surface chemistry was studied by XPS. The results indicated that Ce 4+ was the predominant oxidation state in the converted films irrespective of the treatment time. In addition, the coating is formed by two types of layers: an inner layer, for which the composition of the coating at the metal interface mainly consists of MgO, followed by outer layers rich in cerium oxides. The electrochemical behavior of the conversion films was dependent on the treatment time. The best corrosion resistance was observed for the samples that remained 60 s in the conversion bath. It was not possible to identify a clear correlation between the surface chemistry and corrosion resistance. The results suggest that the treatment time affects the electrochemical behavior by increasing the thickness of the converted layer. Notwithstanding, there is an optimum condition at 60 s. For longer immersion times the corrosion resistance decreases which is likely to be associated with the formation of a more defective film.

Supplementary material
The following online material is available for this article: Figure A1 -Representative XPS core level for the Ce3d region of the conversion film obtained after 20 s of immersion. Figure A2 -Representative XPS core level for the O1s region of the conversion film obtained after 20 s of immersion. Figure A3 -Representative XPS core level for the Ce3d region of the conversion film obtained after 120 s of immersion. Figure A4 -Representative XPS core level for the O1s region of the conversion film obtained after 120 s of immersion.