**2. Experimental**

#### *2.1. Materials and Specimen Preparation*

The steel substrate, produced by Hyundai Steel Corp.(Dang Jin, Korea), was classified as 22MnB5 steel with 0.27–0.3 wt % C, 1.3–1.5 wt % Mn, 0.2–0.25 wt % Cr, 0.18–0.2 wt % Si, 0.0025–0.0035 wt % B, 0.02–0.035 wt % Ti. The S and P contents were kept as low as possible to avoid high-temperature cracking during the subsequent stamping process. Two types of metallic coated steel were used: AlSi and AlSi with Zn and Mg (AlSiZnMg) coated steels, which were manufactured using a hot-dip simulator. For the coating process, two types of molten bath were prepared: Al-Si (7 wt %) and Al-Si (7 wt %)-Zn (10 wt %)–Mg (0.5 wt %). The substrates were then dipped in each molten bath, and two types of metallic coating were produced: AlSi-MC and AlSiZnMg-MC, at 50/50 g/m2. For the following stamping process, the two metallic coated steels were heated to 930 ◦C for 300 s and cooled to room temperature by die quenching.

Some of the two types of metallic coated steel sample were also coated with electrodeposited (ED) paint. For this ED coating, the coated steel samples underwent the following processes: degreasing, surface conditioning, and phosphating in a pretreatment simulator. The ED coating was then conducted at 300 V for 180 s while the voltage was increased uniformly for 30 s. The ED coated samples were dried in a circulating oven at 170 ◦C for 20 min. Detailed information on the ED coating process can be found elsewhere [21].

#### *2.2. Microstructure Examination*

The microstructure and composition of the coated steel samples before and after the corrosion test (potentiostatic (PS) polarization measurement) were examined by FE-SEM and energy dispersive spectroscopy (EDS). Before the corrosion test, the coated steel samples were cleaned ultrasonically in ethanol and dried in air. In particular, for the cross-sectional observations, the samples were mounted with the cut-edge sides facing the surface, and they were polished with a final polishing step of 0.04 μm. After the corrosion test, the sample surface was washed with distilled water and dried in air. The sample was then stored in a vacuum chamber. The various phases in the coating layers were also characterized by XRD.

#### *2.3. Depth Profile Analysis*

The compositional distributions of the coating layers as a function of depth were examined by glow discharge spectroscopy (GDS) analysis with a Leco (St. Joseph, MI, USA) GDS-850A spectrometer using argon plasma equipped with an RF lamp. The diameter of the analysis area in a sampling area of 20 × 20 mm<sup>2</sup> was 4 mm.

#### *2.4. X-ray Photoelectron Spectroscopy Analysis*

XPS was used to analyze the chemical state of the corrosion products formed on the metallic-coated layers after the corrosion test (PS polarization measurement). Prior to analysis, the two types of coated steel sheets that had been corroded by the polarization test were cut into 7 × 8 mm<sup>2</sup> samples and cleaned using ethanol. XPS (VG Scientific Escalab 250 (Waltham, MA, USA)) was conducted using mono chromatic Al K α radiation (1486.7 eV) with a 500 μm diameter spot size. A constant analyzer energy mode with 200 and 50 eV for the survey and high-resolution spectra, respectively, were used. Data processing of the spectra was performed using a spectral data processor (SDP) v 3.0 software. The adventitious C 1s peak at 284.8 eV was used as a reference for charge correction [22].

#### *2.5. Electrochemical Measurements*

Potentiodynamic and potentiostatic polarization tests were performed after the open circuit potential (OCP) evolution measurements for 1 h in a 3.5% NaCl using a GAMRY (Philadelphia, PA, USA) reference 600. A three-electrode system was used, which consisted of the metallic-coated sample as the working electrode (tested area of 1 cm2), a platinum grid as the counter electrode, and a saturated calomel electrode (SCE) as the reference electrode.

For the potentiodynamic polarization scan, the working electrode was polarized from −0.5 V to 0.25 V vs. OCP at a scan rate of 0.16 mV s<sup>−</sup>1. The corrosion current densities were determined by curve-fitting using the Wagner-Traud equation (Equation (1)), described below, to the potentiodynamic polarization curves.

$$i = i\_{corr} \left[ \exp\left(\frac{2.303(E - E\_{corr})}{\beta\_a}\right) - \exp\left(\frac{-2.303(E - E\_{corr})}{\beta\_c}\right) \right] \tag{1}$$

where *i*corr is the corrosion current density (A/cm2), *Ecorr* and *E* represent the corrosion potential (V) and the measured potential (V), respectively. β*a* and β*c* are Tafel slopes (V/decade), and *i* is the total corrosion density (A/cm2).

Based on the polarization curves, the applied potentials for the anodic and cathodic polarization during the potentiostatic polarization measurements were determined to be 50 mV above the OCP and 100 mV below the OCP, respectively. A preliminary test showed that under an applied potential of 100 mV above the OCP in the potentiostatic polarization, the coating dissolution was too severe to analyze the corrosion behaviors of the coating surface covered with some corrosion products. Therefore, the anodic potential was set to 50 mV lower than the cathodic potential in the polarization measurements.

Three repetitive tests were conducted for each experiment to ensure reproducibility.

#### *2.6. Cyclic Corrosion Test for ED Coated Samples*

To evaluate the corrosion resistance of the ED coated AlSi and AlSiZnMg samples, a VDA (Berlin, Germany) 233-102 cyclic corrosion test was conducted according to VDA 233-102 [12], which is a standard for the accelerated corrosion testing of coatings/paintings within the automotive industry. Before the corrosion test, the samples were cut to a size of 150 × 70 mm, and damaged artificially using the following two methods. One set of samples was scratched with an "×" incision in reference to ASTM D-1654 [23], and the other set of samples abraded with a spray of fine stone chips, as reported elsewhere [24]. Salt spraying with a 1% NaCl solution (pH 6.5) at a rate of 2.0 to 4.0 mL/h was then conducted on the samples positioned 65◦ to the horizontal in an enclosed chamber. A full test cycle lasted for seven days, and the multi-step cycles were composed of varying temperatures ranging from −15 ◦C to 50 ◦C and humidity between 50% and 95%. Further information on test conditions can be found in VDA 233-102 [12].

#### **3. Results and Discussion**

#### *3.1. Microstructure Characterizations of Metallic Coatings*

Figure 1 presents the surface and cross-sectional morphologies and EDS mapping of AlSi-MC (Figure 1a,b) and AlSiZnMg-MC (Figure 1c,d). EDS mapping showed that the two types of coating were composed mainly of Al matrix with two (Al,Fe,Si)-rich intermetallic phases. One of the intermetallic phases formed at the interface between the coating layer and Fe substrate during the hot-dip aluminizing process. The other phases formed in the coating layer during the austenitizing process. The formation of the former phase was attributed to the inter-diffusion of Fe atoms from the steel substrate into the coating and of Al atoms from the coating into the steel substrate [25]. At the same time, Si, known to be concentrated at the outer surface of the steel substrate [26], was also enriched at the inter-diffusion layer. According to previous studies [27,28], the composition of the intermetallic phase is close to τ5 or τ6 shown in the ternary phase diagram of Al-Fe-Si [27,28]. The latter phase was formed by the diffusion of Fe and Si atoms from the steel substrate into the coating during austenitizing for the hot-stamping process, as reported elsewhere [29]. A noticeable feature in the coating morphologies was the presence of micro-cracks and pores. These may be closely associated with the very high hardness (900~1100 HV0.05) of Al-rich Al-Fe intermetallic compounds, such as Al13Fe4 and Al5Fe2 [30,31], and the difference in thermal shrinkage among the intermetallic phases during cooling in the hot-stamping process [32].

**Figure 1.** Surface and cross-sectional observations with eds mappings of (**<sup>a</sup>**,**b**) AlSi-MC and (**<sup>c</sup>**,**d**) AlSiZnMg-MC.

Although Figure 1 indicated that there were more pores and micro-cracks on the AlSi-MC, the microscopic observations could not be representative of the entire coating layers. On the other hand, the composition features were different apparently in that the outer-surface of AlSiZnMg-MC was covered with a thin (Mg, Zn)—based phase. GDS analysis was conducted for further clarification; Figure 2 presents the depth profile results. In contrast to the surface of AlSi-MC, consisting only of a thin Al-based oxide, the Mg and Zn-based oxide may cover the coating layer of AlSiZnMg-MC. In addition, based on the AlSiZnMg quaternary system [33], Zn-Mg intermetallic phases may exist on the coating surface of AlSiZnMg. For closer analysis, phase characterization was carried out using XRD; the results are shown in Figure 3. As expected, one of the major phases in both coating systems was an Al-Fe intermetallic compound.

**Figure 2.** Glow discharge spectroscopy (GDS) depth profile of the coating layers: (**<sup>a</sup>**,**b**) AlSi-MC, (**<sup>c</sup>**,**d**) AlSiZnMg-MC.

**Figure 3.** X-ray diffraction patterns of (**a**) AlSi-MC and (**b**) AlSiZnMg-MC.

Al-Fe, Fe-Si, and Fe-Al-Si intermetallic phases are generally difficult to distinguish by XRD due to overlap of the XRD peaks [12]. The phase existing only in AlSiZnMg-MC was MgZn2. The other possible phases in the sample may be Mg2Si and MgO. However, the initial fraction of Mg (0.5 wt %) was too low to give prominent peaks of Mg-containing phases in the XRD measurements. Even if some Mg-containing phases were formed, minor fractions that are below the detection limit of the XRD instrument, might be present. Instead of Mg-containing phases, some peaks for the Al-Fe intermetallic phase and Si were detected with low intensity. Considering that the phases, such as MgZn2 and Mg2Si, have different electrochemical potentials of their own, they can affect the corrosion kinetics and resulting coating life. From an electrochemistry point of view, the phases of MgZn2 and Mg2Si can act as an anode from the coating matrix because of their lower corrosion potential (−1538 and −1029 mVSCE, respectively) [18]. On the other hand, unlike the case of MgZn2, which provides sacrificial protection at an early stage of corrosion, Mg2Si can form Si clusters after the selective dissolution of Mg, which will

be cathodic and cause the localized corrosion attack of an Al-based coating layer. [6,20]. This will be discussed in more detail in the following section.

Another possible phase in AlSiZnMg-MC is MgO, which may be present at the outermost part of the coating surface. Although MgO with low electrical conductivity [34] could contribute in part to the suppression of the cathodic reduction reaction during the early corrosion stage, it is generally non-uniform and, in most cases, increases local corrosion attack. Under this coating composition with a minor fraction of Mg (0.5 wt %), it is extremely di fficult to form a stable and uniform MgO film over the coating surface. These discussions were based primarily on the theoretical studies reported elsewhere, and they should be clarified experimentally, which will be discussed in the following section.

#### *3.2. Electrochemical Characterizations*

The evolution of the OCP (Figure 4) was measured for 1 h before the electrochemical polarization tests. The coating potentials of AlSi-MC and AlSiZnMg-MC after 1 h of immersion were approximately −0.59 VSCE, and −0.62 VSCE, respectively. In contrast to the Al coating covered by a thin Al oxide film (Al2O3), Zn and Mg, which were enriched at the coating surface, rarely form an oxide/hydroxide film with a high passivity coe fficient in a neutral solution, which could result in a lower corrosion potential. As mentioned previously, the presence of active phases, such as Mg2Si and MgZn2, can contribute to the decrease in corrosion potential. On the other hand, the preferential dissolution of Mg from the active phases can provide sacrificial protection. Previous studies [16,35] have shown that Mg and Zn cations, which are supplied by the selective dissolution of MgZn2 (reaction (2)), can also lead to rapid coverage of the coating surface by the precipitation of corrosion products with a protective nature, such as simonkolleite (Zn5(OH)8Cl2·H2O), which can prevent further corrosion.

$$\text{MgZn}\_2\text{(s)} \rightarrow \text{Mg}^{2+}\text{(aq)} + 2\text{Zn}^{2+}\text{(aq)} + 6\text{e}^-\tag{2}$$

**Figure 4.** Open circuit potential (OCP) evolution measurement of AlSi-MC and AlSiZnMg-MC for 1 h in a 3.5% NaCl solution [20].

According to a previous study, the selective dissolution of Mg from the Mg2Si phase can provide cathodic protection. They proposed that the formation of less conductive SiO2 can decrease the galvanic current between Al/SiO2 according to reaction (3) [18,36]:

$$\rm Mg\_2Si(s) + 2H\_2O(l) \to 2Mg^{2+}(aq) + SiO\_2(s) + 4H^+(aq) + 8e^- \tag{3}$$

On the other hand, the formation of a stable SiO2 film uniformly formed on the coating surface cannot be guaranteed. Under this condition, an unstable film may create active local galvanic coupling, increasing the dissolution of the coating layer.

These beneficial and harmful e ffects on the corrosion resistance can be examined by polarization measurements, which are shown in Figure 5; Figure 6. Potentiodynamic and potentiostatic polarizations

showed that the anodic and cathodic reaction rates of AlSiZnMg-MC were much smaller than those of AlSi-MC. The corrosion current densities (*icorr*) determined by curve-fitting with the Wagner–Traud equation to the polarization curves of AlSiZnMg-MC and AlSi-MC were 3.5 and 8.5 μA cm<sup>−</sup>2, respectively. The lower anodic and cathodic current densities of AlSiZnMg-MC sugges<sup>t</sup> that the liberated Mg<sup>2</sup>+ and Zn2<sup>+</sup> ions, supplied by the selective dissolution of Mg and Zn, provided an environment for the formation of corrosion products contributing to the e ffective reduction of the anodic and cathodic current densities. On the other hand, the polarization curve of AlSiZnMg-MC has many fluctuations, which may be closely associated with localized corrosion caused by the non-uniform distribution of MgO or the selective dissolution of Mg from Mg2Si phases in the coating. This can be supported in part by several local corrosion attacks after the corrosion test, as shown in Figure 7. From morphological observations before (Figure 1) and after (Figure 7) the corrosion test (PS polarization), there was little di fference except that the outermost surfaces were less uniform after the corrosion test. Compared to the case of AlSi-MC, the surface morphology of AlSiZnMg-MC after the corrosion test appeared to be denser (Figure 7c). On the other hand, AlSiZnMg-MC exhibited a more uneven cross-sectional view, which may be associated with the formation of corrosion products containing Mg, Zn, and Si, and localized corrosion attacks mentioned previously. Owing to the low resolution of EDS or GDS, XPS was performed to characterize the corrosion products formed on the two types of coating after the corrosion test.

**Figure 5.** Potentiodynamic polarization curves of AlSi-MC and AlSiZnMg-MC, measured after 1 h immersion in a 3.5% NaCl solution [20].

**Figure 6.** Potentiostatic polarization curves of AlSi-MC and AlSiZnMg-MC, measured after 1 h immersion in a 3.5% NaCl solution under (**a**) anodic (50 mV above OCP), and (**b**) cathodic (100 mV below OCP) potentials [20].

**Figure 7.** Surface and cross-sectional observations with EDS mappings of (**<sup>a</sup>**,**b**) AlSi-MC and (**<sup>c</sup>**,**d**) AlSiZnMg-MC after the potentiostatic (PS) test.

#### *3.3. Characterization of the Corrosion Products by XPS*

Figure 8 presents the XPS survey spectrum of the surface of AlSi-MC. The spectrum of the coating surface revealed C 1s, Al 2p, O 1s, and Si 2p primarily. The high-resolution spectra, shown in Figure 8c–e, indicated that the corrosion products formed on the surface were composed mainly of Al2O3 with a smaller portion of SiO2. On the other hand, the survey spectrum of the surface on AlSiZnMg-MC (Figure 9) showed that, in addition to C 1s, Al 2p, O 1s, and Si 2p, Mg 2p and Zn 2p were observed on the coating surface, and the oxygen intensity was higher. This suggests that Zn and Mg-based oxide/hydroxide are major components in the coating surface. From the high-resolution spectra of the surface on AlSiZnMg-MC, as shown in Figure 9c–g, the coating surface was composed of a wider variety of corrosion products including Al2O3, Mg(OH)2, SiO2, Zn(OH)2, Zn5(OH)8Cl2·H2O, and Mg-Al layered double hydroxide (LDH) with the general formula, MxAly(A)m(OH)n.zH2O (where A and M represent an anion and a di-valent cation, respectively) [16,18,37]. As mentioned previously, the addition of Zn and Mg to the coating can help increase the corrosion resistance in a neutral aqueous solution, in such a way that several corrosion products with an inhibiting nature are formed preferentially on the surface at the early stages of corrosion. Among the products, Mg-Al LDH and Zn5(OH)8Cl2·H2O can act as effective barriers for oxygen diffusion [38,39] and provide superior corrosion resistance in neutral aqueous solutions [38,39]. On the other hand, the formation or dissolution of other products can also help increase the corrosion resistance by stabilizing Mg-Al LDH and Zn5(OH)8Cl2·H2O. First, the formation of Mg-Al LDH, as shown in Figure 9d,f involves reactions (2), (4), (5), and (6).

**Figure 8.** (**a**) XPS survey spectrum and (**b**–**<sup>e</sup>**) high-resolution spectra of the surface on AlSi-MC after the PS test.

Oxygen reduction:

$$\rm O\_2(g) + 2H\_2O(l) + 4e^- \rightarrow 4OH^-(aq) \tag{4}$$

Dissolution of alumina:

$$\text{Al}\_2\text{O}\_3(\text{s}) + 3\text{H}\_2\text{O}(\text{l}) + 2\text{OH}^-(\text{aq}) \rightarrow 2\text{Al(OH)}^{4-}(\text{aq})\tag{5}$$

Dissolution of metallic Al:

$$\mathrm{Al(s)} + 4\mathrm{OH}^- \mathrm{(aq)} \to \mathrm{Al(OH)}^{4-} \mathrm{(aq)} + 3\mathrm{e}^- \tag{6}$$

Reactions (5) and (6) can occur locally at the cathode area because of the instability of Al/Al2O3 at alkaline pH [16]. As a result, the formation reaction of LDH can proceed, as described below:

$$2\text{Al(OH)}^{4-}\text{(aq)} + 6\text{Mg}^{2+}\text{(aq)} + 8\text{OH}^{-}\text{(aq)} + \text{CO}\_{3}^{2-}\text{(aq)} \rightarrow \text{Mg}\_{6}\text{Al}\_{2}\text{(OH)}\_{16}\text{CO}\_{3}\text{(s)}\tag{7}$$

The stability of LDH is dependent on the pH [40,41], and LDH can dissolve under weakly alkaline condition. On the other hand, the dissolution of LDH can be suppressed by the formation of a thin Mg(OH)2 layer (reaction (8)), called the skin effect [15,37]:

$$\text{Mg}^+\text{(aq)} + 2\text{OH}^-\text{(aq)} \rightarrow \text{Mg(OH)}\_2\text{(s)}\tag{8}$$

Second, the formation of Zn5(OH)8Cl2·H2O, as shown in Figure 9g, can be described below:

$$2\text{ZnO(s)} + \text{Zn}^{2+}(\text{aq}) + 5\text{H}\_2\text{O(l)} + 2\text{Cl}^-(\text{aq}) \rightarrow \text{Zn}\_5\text{(OH)}\_8\text{Cl}\_2\cdot\text{H}\_2\text{O(s)}\tag{9}$$

The presence of Mg<sup>2</sup>+ supplied by reaction (2) can also delay the following reaction, which leads to slower dissolution kinetics of Zn5(OH)8Cl2·H2O. The formations of Mg-Al LDH and Zn5(OH)8Cl2·H2O with high stability are the major mechanistic reasons for the higher corrosion resistance of AlSiZnMg in a neutral environment containing Cl<sup>−</sup>. Nevertheless, the increased possibility of local corrosion of the AlSiZnMg sample cannot be excluded. In addition, its superiority and utility as a coated steel should also be evaluated after the electropainting process, which will be discussed in the following section.

**Figure 9.** (**a**) XPS survey spectrum and (**b**–**g**) high-resolution spectra of the surface on AlSiZnMg-MC after PS test.

#### *3.4. Surface Characteristics after Electropainting*

Figure 10 presents the GDS depth profiles of AlSi-MC and AlSiZnMg-MC after the ED coating. In contrast to AlSi-MC, Zn(Mg)—based oxides were present on the outer surface of the ED-coated AlSiZnMg-MC. This suggests that the oxides remained on the outer surface, even after surface cleaning before the ED coating. As mentioned in the previous section, a stable thick MgO film cannot be formed in a minor fraction of Mg (0.5 wt %) in this coating system. An unstable oxide film on the surface may have weakened the adhesion between the inner metallic coating/outer ED coating and reduced the corrosion resistance.

**Figure 10.** (**a**) Cross-sectional observation and (**b**) GDS depth profile of ED-treated AlSiZnMg-MC.

#### *3.5. Corrosion Resistance Evaluations after Electropainting*

Figure 11 shows the surface appearance of the ED-treated AlSi-MC (Figure 11a) and AlSiZnMg-MC (Figure 11b), after the accelerated corrosion test in reference to VDA 233-102 [12]. Considering the many test results presented above, it was expected that ED-treated AlSiZnMg-MC showed higher corrosion resistance when evaluated by an accelerated corrosion test. After the ED coating, however, its higher corrosion resistance was not observed clearly in that the damaged areas and extent of red rust formation on the scribed regions with an X between the two coated steels sheets became similar as the number of cycles increased. In addition, the surface morphologies of the two ED-coated steel samples that had been subjected to stone chipping and a subsequent corrosion test revealed even more severe surface degradation of ED-coated AlSiZnMg-MC. As shown in Figure 12, the size and number of surface blisters were larger on the ED-coated AlSiZnMg-MC. These blisters resulted from volumetric expansion caused by the formation of corrosion products on the steel substrate covered with a metallic coating layer. This suggests that the corrosive species are more able to permeate through the ED coating on AlSiZnMg-MC, and the electrochemical corrosion reactions occurred rapidly. The higher degradation by the corrosion reactions of ED coated AlSiZnMg-MC may be associated closely with the lower adhesion between the inner metallic coating/outer ED coating. As mentioned previously, the (Zn, Mg)-based oxide formed on AlSiZnMg can weaken the adhesion between the inner/outer coating layers, leading to delamination of the outer coating layer when attacked mechanically and chemically. This is supported by the lower coefficient of friction for the ED coating on AlSiZnMg-MC, which was measured by surface and interfacial cutting analysis, and was reported elsewhere [21]. Hence, the addition of Zn and Mg to the AlSi-based metallic coating does not always have beneficial effects on the corrosion resistance when the metallic-coated steels are coated additionally with ED paint. These results provide useful insights into the development of coated steel sheets used in automotive industries. Nevertheless, further technical research on optimizing the compositions of Zn and/or Mg alloying that provides sacrificial corrosion protection to the AlSi coated steel while reducing the possibility of local corrosion and securing superior corrosion resistance even after additional ED coating will be needed.

**Figure 11.** Surface appearance of electrodeposited (ED)-treated (**a**) AlSi-MC and (**b**) AlSiZnMg-MC, which had been scratched with an "×" incision, after the accelerated corrosion test.

**Figure 12.** Surface appearance of ED-treated (**a**) AlSi-MC and (**b**) AlSiZnMg-MC, which had been collided with fine stone chips, after the accelerated corrosion test.
