Corrosion Mechanism of Press-Hardened Steel with Aluminum-Silicon Coating in Controlled Atmospheric Conditions

Abstract

The effect of various atmospheric parameters on the corrosion mechanism of press-hardened steel (PHS) coated with Al-Si (AS) was studied. Quantitative models of the composition of soluble and stable corrosion products were developed. A high chloride concentration led to a localized corrosion due to the presence of cracks in the coating. Increased corrosion resistance of silicon-rich Al₈Fe₂Si and AlFe at the expense of the Al₅Fe₂ phase with low silicon content was shown. Under low-chloride-deposition conditions, the coating exhibited good corrosion resistance and provided sufficient protection to the underlying steel. The formation of more local anodes and cathodes under conditions of lower relative humidity led to a reduction in the depth of corrosion pits in the steel substrate. Constant high relative humidity and sulphate deposits on the surface were critical for the acceleration of steel corrosion in coating cracks.

1. Introduction

Over the last two decades, one of the goals of the automotive industry was to reduce vehicle weight in order to lower fuel consumption and CO₂ emissions. At the same time, passenger safety must be maintained. Press-hardened steels (PHSs) with tensile strength from 1500 to 2000 MPa and good formability meet both requirements. The excellent mechanical properties allow for the manufacture of passive safety components such as A/B-pillars or crash bars [1,2,3,4,5]. They are made using the hot-stamping process. Sheets of hardenable boron steel with an initial ferritic–pearlitic structure are austenitized at temperatures around 900 °C. This is followed by simultaneous rapid quenching and forming to the desired final shape. High mechanical stress together with rapid cooling leads to the formation of the martensitic structure [6,7]. The corrosion resistance of these steels is typically provided by metallic coatings, either Zn [8,9,10]- or Al-Si [11,12,13]-based.

Al-Si coating is applied at the steel surface before hot stamping by hot-dip aluminizing. Due to the high temperature during the hot-dipping and austenitization processes, diffusion occurs at the coating/steel interface. Iron from steel diffuses towards the coating, while aluminum and silicon from the coating diffuse towards the steel. This leads to the formation of different Al-Fe-Si intermetallic phases and an increase in the coating thickness, typically ranging from 35 to 45 µm [14,15,16]. Hot-dip aluminizing usually proceeds at temperatures around 650 °C. The microstructure of hot-dip Al-Si coating consists typically of a top layer of an Al matrix with Si needles and a layer of an Al₈Fe₂Si phase at the interface with steel [12,17]. During subsequent austenitization, diffusion processes are accelerated due to the higher temperature. In the first minutes, the diffusion of Fe from steel dominates and Al-rich intermetallics Al₅Fe₂ and Al₁₃Fe₄ form. The later diffusion of Al from the coating causes the partial transformation of these Al-rich intermetallics into Fe-rich intermetallics, such as AlFe. At the same time, an α-Fe layer with a ferritic structure is formed at the interface [18,19,20,21]. This results in a multi-layered and multiphase coating structure containing various defects, such as microcracks caused by differences in the thermal expansion of coating components or Kirkendall voids formed due to the different diffusion rates of Al and Fe [4,22].

The corrosion behaviour of press-hardened steel with aluminum-silicon coating (PHS AS) is influenced by the complex microstructure of the coating. Unlike zinc-based coatings, Al-Si is not able to provide a sufficient sacrificial protection to steel because of a small corrosion potential difference lower than 30 mV [11]. Due to the elevated thickness and the ability of aluminum to spontaneously passivate with a thin layer of Al₂O₃, Al-Si provides mainly barrier protection. Therefore, it is assumed that the most critical features are cracks in the coating [11,23].

Several studies have investigated the behaviour of PHS AS under atmospheric corrosion conditions. Allély et al. [11] exposed it to the cyclic accelerated corrosion test VDA 233-102 and the continuous neutral salt spray test. The simultaneous active corrosion of Al-Si and steel was shown. The main corrosion products identified were hydroxides and aluminosilicates after VDA 233-102 and oxides of Al, Fe, and Si after the neutral salt spray test. The influence of Zn and Mg alloying on the corrosion performance of Al-Si coating was studied by Nicard et al. [12]. After exposure of an unalloyed coating to VDA 233-102, Al(OH)₃ and SiO₂ were detected. The presence of Zn and Mg led to an improvement in the coating sacrificial protection of steel. However, there was a significant increase in mass loss due to the acceleration of coating corrosion. Dosdat et al. [15] compared the corrosion performance of bare and Zn-, Zn-Fe-, and Al-Si-coated PHS after VDA 233-102, analyzing mass loss, cut-edge corrosion, and pit depth. Except for cut-edge corrosion, PHS AS exhibited similar results to steel with Zn-based coatings. All tested materials significantly outperformed bare steel.

The mentioned studies used the cyclic test VDA 233-102 to assess the corrosion mechanism of PHS AS. It is a complex procedure with several parameters changing simultaneously. In this study, a series of five simplified corrosion tests were carried out to describe the influence of selected atmospheric parameters on the corrosion of PHS AS separately.

2. Materials and Methods

2.1. Material

PHS sheets with Al-Si coating and ultimate tensile strength of approximately 1500 MPa were provided by ArcelorMittal (Maizières-lès-Metz, France. The steel substrate thickness was 1.5 mm. The elemental composition in wt.% determined by optical emission spectroscope (OES) Q4 Tasman (Bruker AXS GmbH, Karlsruhe, Germany) is listed in

Table 1. Elemental composition of bare PHS in wt.%.

C	Si	Mn	P	S	Cr	Ni	Mo	Al	Cu	B	Fe
0.22	0.24	1.06	0.01	<0.005	0.18	0.02	0.01	0.04	0.01	0.0022	bal.

. The 150 g∙m⁻² coating was applied to steel in an industrial hot-dip process with the nominal bath composition Al-9%Si. Coated sheets were austenitized at 920 °C for 6 min and quenched rapidly under simultaneous forming into the final shape. The direct hot-stamping process resulted in a fully martensitic steel microstructure and final coating thickness of 35 ± 3 µm.

The characterization of the coating microstructure was performed by a ZEISS EVO15 scanning electron microscope (Carl Zeiss NTS GmbH, Oberkochen, Germany) with an energy dispersive spectrometer (SEM-EDS). The micrograph in Figure 1 shows a thin layer of Al₂O₃ on the outer surface of the coating, which was identified by EDS measurement on the top surface. An EDS spot analysis conducted at PHS AS cross-sections in different parts of the coating revealed the presence of multiple Al-Fe-Si intermetallic phases; see

Table 2. Chemical composition of Al-Si coating in spots marked in Figure 1 determined by EDS analysis in at.% and corresponding phases identified based on [18,19,24].

Spot	Al	Fe	Si	Phase
1	55.0	36.5	8.5	AlFe
2	68.4	28.9	2.7	Al5Fe2
3	66.4	20.5	13.1	Al8Fe2Si
4	42.8	44.5	12.6	AlFe
5	66.8	30.3	2.9	Al5Fe2
6	25.8	64.7	9.5	α-Fe
7	10.1	85.7	4.1	α-Fe

. The phases were identified according to previously published data [18,19,24]. In the outer part of the coating, isolated islands of a Si-rich AlFe phase are present. Below, a layer of Al₅Fe₂ was detected. The solubility of silicon in this phase is very low, typically below 5 at.% [25], which corresponds to performed EDS measurement, where the Si content did not exceed 3 at.%. In the middle part of the coating, a layer composed of Al₈Fe₂Si and AlFe phases with the highest Si content is present. The solubility of silicon in binary intermetallic AlFe at 900 °C corresponding to the PHS AS heat treatment conditions is around 16 at.% [26]. The inner part of the coating is again formed by the Al₅Fe₂ phase. At the steel/coating interface, there was a layer of an α-Fe solid solution with high Si content. The cross-section view also revealed a significant number of micro-defects in the coating, particularly cracks and Kirkendall voids.

2.2. Sample Preparation

Two types of samples were used: rectangular panels of 150 × 100 mm for a mass loss analysis and square panels of 50 × 50 mm for corrosion product identification. In total, four mass loss and three corrosion product panels were exposed in every test. Samples were rid of manufacturing contaminants by rinsing with ethanol and dried with a stream of oil-free air. Back sides and cut-edges of samples were protected by adhesive tape throughout exposures. This resulted in a total exposed area of 126 cm² and 16 cm² for mass loss and corrosion product coupons, respectively.

2.3. Corrosion Tests

Five different corrosion tests, see

Table 3. Parameters of corrosion tests; initial conditions and differences from them are highlighted.

Parameters	RH Wet Phase [%]	RH Dry Phase [%]	Chloride Concentration [g∙m−2]	Contamination Salt	Wet/Dry Cycle
Initial conditions	95	50	0.9	NaCl	ISO 16701:2015
Sulphates	95	50	0.9	NaCl+Na2SO4	ISO 16701:2015
Low chlorides	95	50	0.09	NaCl	ISO 16701:2015
Low RH	80	50	0.9	NaCl	ISO 16701:2015
Constant RH	95	-	0.9	NaCl	Only wet phase

, were performed to study the influence of different environmental parameters on corrosion behaviour of PHS AS. The conditions were consistent with those used in our previous work [9]. The first exposure, denominated as initial, follows ISO 16701:2015 [27] at a constant temperature of 35 °C. Each cycle consists of four hours of a wet phase at 95% RH, two hours of drying, four hours of a dry phase at 50% RH, and two hours of wetting. All samples were contaminated with a solution of NaCl in methanol (6.49 g∙L⁻¹) in order to reach a uniform surface chloride concentration, 0.9 g∙m⁻². In the other tests, a single environmental parameter was modified to study the influence of the presence of sulphates (contamination with a mixture of NaCl and Na₂SO₄), the lower surface chloride concentration (0.09 g∙m⁻²), the lower relative humidity in the wet phase (80% RH), and the constant high relative humidity (95% RH; no dry phase). The duration of each corrosion test was four weeks.

2.4. Test Procedure and Performed Analyses

Before each test, the first contamination of all samples was performed. The whole contamination procedure was, in detail, described elsewhere [9]. In short, in all tests except Sulphates and Low chlorides, a solution of NaCl dissolved in methanol (6.49 g∙L⁻¹) was used. For the Sulphates exposure, a solution of 2 g∙L⁻¹ Na₂SO₄ in a mixture of methanol and demineralized water (3:2 vol.) in combination with the NaCl solution was used to achieve the surface concentration of sulphates and chlorides, 0.28 and 0.9 g∙m⁻², respectively. The goal of this test was to simulate a marine environment polluted with anthropogenic sulphates [28]. For the Low chlorides test, a ten times diluted solution of NaCl in methanol (0.649 g∙L⁻¹) was used. To ensure a homogeneous spreading of the contaminants, the solutions were applied by a micropipette in several doses.

The samples were placed into a climatic chamber, WEISS C600/70/3 (Weiss Technik, Reiskirchen-Lindenstruth, Germany), horizontally. After each week, they were removed from the chamber and rinsed with demineralized water and ethanol and dried with a stream of oil-free air. The water-based solutions rinsed from the mass loss panels were collected and the concentration of chlorides was measured by an ion-selective electrode (ISE). Then, the solutions were acidified with diluted HNO₃ to achieve pH < 4. The concentration of Al³⁺, Si⁴⁺, Fe²⁺/Fe³⁺, and Na⁺ ions was determined by inductively coupled plasma optical emission spectroscopy (ICP-OES) by Agilent 5100 (Agilent Technologies, Santa Clara, CA, USA). For corrosion product characterization, one square sample was stored after the end of the 1st, 2nd, and 4th week without rinsing. Before returning to the chamber, all remaining samples were contaminated in the same way as at the beginning of the exposure. In total, 3.6 g∙m⁻² of chlorides was applied during every 4-week exposure except the Low chlorides test (0.36 g∙m⁻²). At the end of the test, the same rinsing and drying procedure was repeated.

Dry mass gain values were evaluated by weighting the rectangular panels after complete drying in a desiccator for 24 h. Then, corrosion products were removed by chemical pickling using a procedure described in the ISO 8407 standard [29]. A solution of CrO₃ (20 g∙L⁻¹) and H₃PO₄ (50 mL∙L⁻¹) in demineralized water was heated up to 80 °C. Each sample was pickled with a fresh solution in an ultrasonic bath with pickling intervals of 5 min. The total pickling time varied from 30 to 40 min. ICP-OES analyses of the pickling solutions were performed in order to quantify dissolved metal ions in stable corrosion products.

The evolution of the corrosion product composition was followed by several methods. Photographs of corroded surfaces were taken after each exposure period. SEM-EDS analyses of top surfaces and cross-cuts were performed on selected samples. Cross-cuts were prepared by cutting 15 × 15 mm samples from the corrosion product panels. They were embedded in epoxy resin, ground up to P2500 SiC emery paper, polished with a 1 µm diamond paste using an oil-based lubricant, rinsed with ethanol, and dried with oil-free air.

The phase composition of corrosion products was studied by Fourier transform infrared spectroscopy (FTIR) in a range from 4000 to 400 cm^–1 and 128 scans at a resolution of 4 cm^–1 using a Nicolet iS50 spectroscope (Nicolet CZ, Prague, Czech Republic). All FTIR spectra were corrected by applying a linear ramp. Additionally, an X-ray diffraction (XRD) analysis was performed using an AERIS XRD diffractometer (Malvern Pananalytical, Malvern, UK).

3. Results

3.1. Corrosion Properties of PHS AS

Figure 2 shows the progress of the corrosion of PHS AS during the corrosion tests. In all tests, except Low chlorides, red corrosion products were observed already after the first week. The corrosion attack was localized. Due to a very high iron content in the intermetallic phases of the Al-Si coating, the presence of red corrosion products cannot be attributed solely to the corrosion of the underlying steel substrate. In areas where no red corrosion products were present, the coating remained largely intact. The visually most severe corrosion occurred in the Constant RH and Sulphates exposures.

Figure 3 shows mass loss after four weeks of the corrosion tests. PHS AS showed the highest mass loss of 29.1 g∙m⁻² after the Constant RH exposure, followed by the Sulphates exposure with a mass loss of 21.1 g∙m⁻². These values align with the surface appearance in Figure 2. There is no significant difference between the Initial conditions and Low RH exposures in terms of mass loss with values of 12.8 and 12.0 g∙m⁻², respectively. In the Low chlorides exposure, the mass loss of 2.0 g∙m⁻² was almost 15 times lower compared to the Constant RH test.

After corrosion product removal, it was evident that pitting through the coating was the dominant type of corrosion degradation with no significant coating damage outside the pits. Figure 4 shows the mean and maximal values of pit depth in steel obtained by a 3D analysis with a digital microscope. A total of 30 pits on two samples randomly selected after each test were captured and their depths were evaluated. The coating thickness of 35 µm was subtracted from the measured values. The results show a trend similar to the mass loss data in Figure 3 with the most severe attack in steel after the Constant RH and Sulphates tests. The deepest pit of 62 µm was identified after the former test. For Initial conditions, the average pit depth was twice as deep as for Low RH, 27 and 14 µm, respectively. Given the almost identical values of mass loss and comparing the sample appearances in Figure 2, it can be inferred that the lower relative humidity of the wet phase (80%) in the Low RH test led to the initiation of shallower pits as opposed to 95% RH in the Initial conditions test. In the Low chlorides test, no pits were found in steel, indicating that the mass loss value can be attributed exclusively to the corrosion of the Al-Si coating.

SEM images of PHS AS cross-sections after various durations of the Initial conditions test are shown in Figure 5. The corrosion of the substrate occurred already after the first week of exposure. As time progressed, a stronger degradation of the coating can be observed in the pitting areas. The intermetallic phase Al₅Fe₂ with approximately 3 at.% of Si, which predominates in the coating, transformed into bulk corrosion products. The AlFe phase with Si content of 9 at.% originally located near the surface also could not be found. On the other hand, the Al₈Fe₂Si and AlFe phases located in the middle of the coating were still observable as an unevenly cracked layer even after the fourth week of the exposure. This higher corrosion resistance may be linked to the highest Si content, which in both phases is approximately 13 at.%.

Figure 6 shows a SEM-EDS analysis of the top surface of PHS AS after the end of the Initial conditions test. The separation of cathodic and anodic areas is evidenced by the distribution of sodium and chlorides, respectively [30]. In anodic areas, an increased signal of iron is visible, while aluminum is located primarily in cathodic areas. No distinct area with silicon-based corrosion products was observed on the surface.

3.2. Balance of Substances Involved in the Corrosion Process

One of the objectives of this study was to describe in detail what happens under different atmospheric conditions in terms of all the components involved in the corrosion process of PHS AS. To achieve this, various analyses were performed both during and after each exposure, as described in detail in our previous work [9], and graphically shown for this particular material in Figure 7. At the end of each exposure, mass loss was determined after removal of corrosion products in a chromic acid solution. This parameter represents all metallic ions (Al³⁺, Fe²⁺/Fe³⁺, Si⁴⁺) oxidized and incorporated into corrosion products, which can be separated into water-soluble and stable ones. The amount of water-soluble metal ions was determined by ICP-OES analyses of the rinsed water solutions collected after each week of the exposures. These analyses, together with the ISE measurements, were also used to determine the amount of Na⁺ and Cl^– present in the water-soluble corrosion products.

The amount of water-insoluble metal ions bound in stable corrosion products was determined by ICP-OES analyses of the pickling solutions. Dry mass gain consists of all components added to the system except free water. It includes Na⁺ and Cl⁻ in stable corrosion products. Their amount was determined as the difference between the total mass of deposited NaCl and the mass of Na⁺ and Cl^– in water-soluble corrosion products. The rest of the dry mass gain consists of atmospheric species bound in stable corrosion products (O₂/O^2–, CO₂/CO₃^2–, and H₂O/OH^–). For the Sulphates exposure, the balance was extended for the additional contaminant, Na₂SO₄. In total, the amount of stable corrosion products formed throughout the tests consists of the dry mass gain and mass loss, excluding the water-soluble metal ions.

Table 4. Summary of data on mass loss, dry mass gain, and analyses of rinsed and pickling solutions.

	Initial Conditions	Sulphates	Low Chlorides	Low RH	Constant RH
Mass loss [g∙m−2]	12.8	21.1	2.0	12.0	29.1
Water-soluble metal ions [g∙m−2]	2.9	0.4	0.16	2.9	1.1
Al:Fe:Si [%]	33:59:8	28:56:16	39:48:13	49:44:7	14:84:2
Water-insoluble metal ions [g∙m−2]	9.9	20.7	1.84	9.1	28.0
% of mass loss	77	98	92	76	96
Al:Fe:Si [%]	31:66:3	27:70:3	27:69:4	36:61:3	19:78:3
Dry mass gain [g∙m−2]	15.7	19.4	5.5	14.8	23.2
Water-insoluble contaminants [g∙m−2]	4.3	4.0	0.5	4.4	4.1
Cl– [%]	79	79	98	83	78
Na+ [%]	58	43	78	62	55
Atmospheric species [g∙m−2]	11.4	15.4	5.0	10.4	19.1

summarizes obtained data. Diagrams describing the composition of stable corrosion products on PHS AS after each type of exposure are shown in Figure 8.

In terms of the presence of metal ions in soluble or insoluble form, performed corrosion tests can be divided into two groups. A significant proportion of water-soluble metal ions in the total mass loss is observed for Initial conditions and Low RH, where they account for 23 and 24 wt.%, respectively. For the other tests, a preferential formation of stable corrosion products was observed, with water-soluble ions contributing a maximum of 8 wt.% to the total mass loss for Low chlorides. For the Sulphates and Constant RH exposures, their contribution was negligible. The ratio of individual water-soluble metal ions varied considerably. For Initial conditions and Low RH, a noticeable difference appears particularly in the Al:Fe ratio. The decrease in RH during the wet phase from 95 to 80% resulted in a more pronounced representation of Al³⁺ soluble ions at the expense of Fe²⁺/Fe³⁺. In the case of water-insoluble metal ions, a significantly different behaviour is only observed for the Constant RH exposure, where iron ions dominated at 79%. This can be explained by the highest measured mass loss corresponding to the strongest dissolution of steel. For the other tests, the ratio ranged from 61 to 70%.

Except for the Low chlorides exposure, the ratio of chlorides bound in stable corrosion products was relatively stable, ranging from 78 to 83 wt.%. When a tenfold smaller amount of NaCl was applied, almost all chlorides were consumed to form stable corrosion products. Similar results were observed when zinc-coated PHS was subjected to the same series of simplified corrosion tests [9]. The amount of water-insoluble sodium varied from 43 to 78 wt.%.

Figure 9 and Figure 10 show the kinetics of the incorporation of chloride and sodium ions, respectively, into stable corrosion products during each corrosion test. These data were obtained from the ICP-OES and ISE analyses of the rinsed water solutions. The measured amount of water-soluble contaminants was then subtracted from the amount of applied contaminants. Especially in the case of chlorides, the evolution of the material’s corrosion rate over time can be inferred from their consumption. Figure 9 shows that the amount of incorporated chlorides did not change significantly over time. A slight gradual decrease can only be observed for the Initial conditions and Sulphates exposures. As mentioned above, during the Low chlorides exposure, almost all chlorides were consistently consumed in the corrosion reaction. It suggests that the corrosion rate did not change much during the tests, and the chloride-based corrosion products did not provide any significant protection. Dependencies of the incorporation of sodium ions in Figure 10 show a different pattern. They tended to form stable corrosion products the most in the first week of exposure. A small proportion of stable corrosion products containing sodium formed during the Sulphates exposure. However, it should be considered that the total amount of available Na⁺ ions was higher in this test due to additional contamination with Na₂SO₄.

The time dependence of the amount of water-soluble metal ions released from PHS AS obtained from the ICP-OES analyses of the rinsed solutions is shown in Figure 11. Significant differences between individual exposures are evident. The Initial conditions and Low RH exposures show an increase in the amount of water-soluble corrosion products over time. The effect of the lower relative humidity is limited. The Constant RH exposure shows the opposite trend, as the amount of metal ions in soluble corrosion products decreased significantly over time. A comparison of these three tests reveals that the wet–dry cycling had a significant effect on the formation of soluble corrosion products. This effect is strongly suppressed in the presence of sulphates, when the amount of released metal ions was sometimes up to eight times lower compared to the Initial conditions exposure. In the case of Low chlorides, the low amount of soluble corrosion products corresponds to the lowest mass loss value.

A relationship between measured mass loss and the total amount of individual metal ions formed during the corrosion tests is shown in Figure 12. It is obvious that the total amount of corroded iron increased linearly with mass loss. Aluminum and silicon exhibit only a minor increase within the mass loss range from 12.0 g·m⁻² for Low RH to 29.1 g·m⁻² for Constant RH. Data in Table 2 show that the only phase in which iron is the dominating element is the α-Fe solid solution at the coating/steel interface. The significant increase in the amount of corroded iron can therefore be attributed mainly to the corrosion of the steel substrate. This is in line with the observation exemplified in Figure 2 and Figure 5 showing preferential localized corrosion with deep pits in the base steel, while the surrounding AS coating remained relatively unaffected. The ratio between the total amount of released Si and Al was in all cases around the initial Al-9%Si coating composition. Thus, no preferential dissolution can be inferred for either element.

3.3. Corrosion Product Composition

The phase composition of the corrosion products was analyzed using a combination of FTIR and XRD. Cross-section SEM-EDS analyses were used as a complement. Samples were analyzed after the first, second, and fourth week for each exposure. An example of the FTIR spectra evolution over time for Initial conditions is shown in Figure 13. It can be observed that there was no significant change in the corrosion product composition during 4 weeks of the exposure.

Table 5. Corrosion products identified by FTIR and XRD on PHS AS.

Water-soluble compounds and corrosion products	**■ *■ **° *■	Aluminum chloride, AlCl3 Cadwaladerite, AlCl(OH)2·4H2O Sodium chloride, NaCl Sodium carbonate, Na2CO3
Stable corrosion products	*■ **■ *° **■ **■ *° *S■	Analcime, Na(AlSi2O6)∙H2O Dawsonite, NaAlCO3(OH)2 Boehmite, γ-AlOOH Bayerite, Al(OH)3 Akaganeite, β-FeOOH Magnetite, Fe3O4 Basaluminite, Al4SO4(OH)10∙5H2O

* Compounds identified with a lower certainty; ** compounds clearly identified; ^S compounds identified only after the Sulphates test; ■ compounds detected by FTIR; ° compounds detected by XRD.

lists all identified corrosion products. With the exception of the Sulphates and Low chlorides tests, the same phases were identified after all tests.

The FTIR analysis provided the most comprehensive information on the corrosion product composition. The broad peak at 3260 cm⁻¹ is typical for O-H stretching. Different types of aluminum-containing corrosion products were found. Peaks at 1018, 955, 790, and 538 cm⁻¹ correspond to bayerite, Al(OH)₃. Soluble aluminum chloride, AlCl₃, was detected by peaks at 1647 and 848 cm⁻¹, as well as an increased absorbance around 600 cm⁻¹ [31]. The soluble intermediate of Al(OH)₃ to AlCl₃ transformation, cadwaladerite, AlCl(OH)₂·4H₂O, was identified by peaks at 1647, 1094, 955, 790, and 665 cm⁻¹. Dawsonite, NaAlCO₃(OH)₂, was identified as another stable aluminum corrosion product with characteristic peaks at 1558, 1384, 1094, 955, 848, 665, and 538 cm^–1 [32]. The only silicon-containing corrosion product detected was analcime, Na(AlSi₂O₆)·H₂O. The presence of this phase is indicated by a shoulder at 1018 cm⁻¹, corresponding to the asymmetric stretching of the Al–O and Si–O belonging to the AlO₄ and SiO₄ tetrahedra, a peak at 890 cm⁻¹, and an increased absorbance in the region 600–700 cm⁻¹, and at 426 cm⁻¹ related to the Si-O-Al bond [33]. Allély et al. [11] and Nicard et al. [12] detected this type of corrosion product after the VDA 233-102 accelerated corrosion test of similar Al-Si-based coatings. Akaganeite, β-FeOOH, the main stable corrosion product of iron, was identified by peaks at 1647, 848, and 665 cm⁻¹ [34]. It is a typical iron corrosion product forming under elevated chloride contamination [35]. The presence of sodium carbonate, Na₂CO₃, was indicated by strong carbonate bonding in the region around 1400 cm⁻¹, and a peak at 890 cm⁻¹ [36]. Basaluminite, Al₄SO₄(OH)₁₀∙5H₂O, was identified as the only sulphate-based corrosion product in the Sulphates exposure. It has characteristic peaks at 1653, 1100, 940, and 610 cm⁻¹ [37,38]. As seen in Table 5, the XRD analysis allowed for the identification of only a few phases with less certainty, mainly due to the low level of crystallinity. It confirmed the presence of unreacted NaCl and possible occurrence of boehmite, γ-AlOOH, and magnetite, Fe₃O₄. Magnetite was not identified on the Low chlorides samples.

Figure 14 and Figure 15 show cross-sections of PHS AS samples in the pit locations after four weeks of the Initial conditions and Sulphates tests, respectively. The elemental maps allow us to infer locations of different corrosion products. Chloride-based corrosion products are mainly found on the outer surface. These are mainly AlCl₃ and akaganeite with chemical formula FeO_0.833(OH)_1.167Cl_0.167 [39]. Inside the pit, Fe and O prevails, suggesting that magnetite dominates there. Limited oxygen access in the pit creates conditions favouring the formation of magnetite over iron oxohydroxides [35]. Co-occurrence of Na, Al, Si, and O in the central part of the film of corrosion products in Figure 14 indicates the presence of analcime, Na(AlSi₂O₆)∙H₂O. Figure 15 shows the preferential occurrence of sulphur in aluminum corrosion products located at the outer surface.

3.4. Corrosion Product Composition Models

Based on the performed analyses, models of the composition of soluble and stable corrosion products after four weeks of corrosion exposures were built. Intermediate phases were not included. None of the phase composition analyses found a soluble corrosion product containing silicon. This is likely due to the very low content of soluble Si⁴⁺ ions detected in the rinsed water solution by ICP-OES; see Table 4. Based on the presence of insoluble aluminosilicate basaluminite, it was concluded that the likely soluble silicon phase is sodium silicate, Na₂SiO₃ [11]. Soluble Fe²⁺/Fe³⁺ and Al³⁺ are most probably present in unspecified intermediate hydroxides. In summary, the following assumptions were used to build the models:

Model of water-soluble compounds and corrosion products:

• Soluble Si⁴⁺ is present in Na₂SiO₃;
• Soluble Fe²⁺/Fe³⁺ is present in Fe(OH)_x;
• Soluble Al³⁺ is present in Al(OH)_x;
• Soluble Cl^– is present in NaCl;
• Remaining soluble Na⁺ is present in Na₂CO₃.

Model of stable corrosion products:

• Water-insoluble Si⁴⁺ in present in analcime, NaAlSi₂O₆∙H₂O;
• Remaining water-insoluble Na⁺ is present in dawsonite, NaAlCO₃(OH)₂;
• Water-insoluble Fe²⁺/Fe³⁺ is equimolarly split between akaganeite, β-FeOOH, and magnetite, Fe₃O₄, except the Low chlorides test where Fe³⁺ is present only in akaganeite;
• Remaining water-insoluble Cl^– is present in AlCl₃;
• Remaining water-insoluble Al³⁺ is present in bayerite, Al(OH)₃.

Model of stable corrosion products in the presence of sulphates:

• All applied SO₄^2– is present in basaluminite, Al₄SO₄(OH)₁₀∙5H₂O;
• Water-insoluble Si⁴⁺ is present in analcime, NaAlSi₂O₆∙H₂O;
• Remaining water-insoluble Na⁺ is present in dawsonite, NaAlCO₃(OH)₂;
• Water-insoluble Fe²⁺/Fe³⁺ is equimolarly split between akaganeite, β-FeOOH, and magnetite, Fe₃O₄;
• Remaining water-insoluble Cl^– is present in AlCl₃;
• Remaining water-insoluble Al³⁺ is present in bayerite, Al(OH)₃.

Quantitative models of phase compositions of corrosion products on PHS AS after each test are shown in

Table 6. Quantitative models of soluble and stable corrosion product composition on PHS AS after four weeks of exposure.

Phase	Initial Conditions	Sulphates	Low Chlorides	Low RH	Constant RH
Soluble compounds and corrosion products [wt.%.]
Al(OH)x Fe(OH)x Na2SiO3 NaCl Na2CO3	25 36 15 19 5	8 12 12 53 15	30 30 28 3 9	37 27 14 16 6	8 37 3 42 10
Total amount of soluble compounds and corrosion products [g∙m−2]	6.3	2.3	0.3	6.1	3.2
Stable corrosion products [wt.%.]
Analcime Dawsonite Akaganeite Magnetite AlCl3 Bayerite Basaluminite	5 31 23 18 13 10 -	5 12 27 22 6 16 12	4 24 51 - 6 15 -	5 34 20 16 14 11 -	6 11 34 27 4 18 -
Total amount of stable corrosion products [g∙m−2]	25.6	40.1	7.3	23.9	51.2

. The equimolar split of insoluble Fe²⁺/Fe³⁺ ions between akaganeite and magnetite was applied based on the observations reported above. Except for the Low chlorides exposure, both corrosion products were detected. Images in Figure 2 show strongly localized corrosion and the occurrence of red rust only at the sites of pitting. The SEM-EDS distributions of Fe, O, and Cl suggest a preferential occurrence of akaganeite in the upper layer of the corrosion products, while magnetite was located inside pits. After the Low chlorides test, no pits were detected; see Figure 4.

AlCl₃ was included in the model of stable corrosion products although it is generally considered as a water-soluble compound. With the exception of akaganeite, no other chlorine-containing corrosion product was detected. Even assuming that all insoluble Fe²⁺/Fe³⁺ ions bonded in akaganeite, a large amount of chlorine was still present in the stable corrosion products. It can be inferred that the sample rinsing did not fully dissolve AlCl₃ or other intermediate phases, such as cadwaladerite. Figure 14 shows that there is a significant amount of these corrosion products especially in the vicinity of pits. This clustering of multiple phases can be expected to significantly slow down the dissolution of AlCl₃ compared to a situation where this phase was present on the surface separately. The full dissolution of AlCl₃ might occur under prolonged leaching in water, which does not correspond to realistic atmospheric corrosion conditions. Therefore, AlCl₃ is expected to be an important phase of corrosion products of PHS AS.

The model describing the Sulphates test assumes that all SO₄^2– ions were bond in stable corrosion products. This is supported by the data presented in Table 4 and in Figure 11 for the Initial conditions and Sulphates tests, which differed only in the addition of Na₂SO₄. Table 4 shows that notwithstanding significantly higher mass loss, more than seven times less soluble metal ions were found on samples after the Sulphates test. Figure 11 further indicates a consistently low amount of soluble metal ions formed over time in this test whereas it increased in the Initial conditions test. It suggests that sulphates promote the preferential formation of stable corrosion products.

4. Discussion

4.1. Corrosion Mechanism of PHS AS in Atmospheric Conditions

Already after the first week of the Initial conditions exposure, the localized corrosion of PHS AS was observable. The high incorporation of Na⁺ and Cl^– ions of 75 and 84% of the total applied mass, respectively, indicates a rapid formation of stable corrosion products. A thin layer of Al₂O₃ was present initially on the coating surface; see Figure 1. Chlorides can disrupt this layer, producing Al³⁺ ions. Considering the high density of cracks in the AS coating, these acted as corrosion initiation points. While most of the Cl^– ions were transported inside the cracks, Na⁺ ions preferentially remained at the coating surface around these defects. Al³⁺ ions on the coating surface underwent hydrolysis; see Equation (1). Subsequently, dawsonite formed according to Equation (2). It is typically observed on aluminum in cathodic regions when the pH exceeds 7 [32,40].

Al³⁺ + 2H₂O ↔ Al(OH)₂⁺ + 2H⁺

(1)

Al(OH)₂⁺ + HCO₃^– + Na⁺ ↔ NaAlCO₃(OH)₂ + H⁺

(2)

Over time, the ratio of Na⁺ ions incorporated into stable corrosion products decreased, suggesting the formation of Na₂CO₃ at cathodic sites. The alkaline conditions at these sites were mainly caused by the cathodic oxygen reduction reaction (ORR; Equation (3)). The hydrogen evolution reaction (HER; Equation (4)) occurred in parallel in acidified and oxygen-depleted defects [41]. Under atmospheric corrosion conditions, HER was shown to constitute up to 27% of the total cathodic process on PHS AS [42].

O₂ + 2H₂O + 4e^– → 4OH^–

(3)

2H⁺ + 2e^– → H₂

(4)

The cracks and their vicinity were anodic as evidenced by the elevated chloride concentration; see Figure 6. Here, the corrosion of the AS coating and steel substrate occurs simultaneously. Al-Si coatings do not provide strong galvanic protection to steel unlike Zn-based coatings. The potential difference between non-heat-treated Al-Si coating and steel is approximately 100 mV, which is almost 300 mV less compared to that between Zn coating and steel. Due to the diffusion of Fe from steel into the coating during the heat treatment of PHS AS, the potential difference decreases by approximately 50–70 mV [8,11]. As shown in Figure 5, Figure 14 and Figure 15, the dominant coating phase Al₅Fe₂ corroded preferentially. Remnants of Si-rich phases AlFe and Al₈Fe₂Si in the middle of the coating were present even in strongly corroded areas after four weeks of exposure. This can be caused by microgalvanic coupling between the intermetallic phases. The main difference between Al₅Fe₂ and Al₈Fe₂Si is the silicon content, which is approximately 3 and 13 at.%, respectively; see Table 2. The open corrosion potential (OCP) of pure Al and Si in 0.1 M NaCl is −823 and −441 mV/SCE [43]. Thus, the higher Si content makes the intermetallic phase nobler. OCP of Al_xFe_ySi in 0.1M NaCl is −470 mV/SCE [12]. In the same solution, OCP of Al₃Fe with elemental composition very similar to Al₅Fe₂ is −539 mV/SCE [43]. The AlFe phase has the same content of Si as Al₈Fe₂Si and thus supposedly more positive OCP than Al₅Fe₂. Based on local electrochemical measurements at various coating depths on PHS AS, it was determined that the OCP of the intermetallic phases decreases in the following order: AlFe/Al₈Fe₂Si > α-Fe > Al₅Fe₂ [44]. The increased corrosion resistance of Si-rich phases may also be due to the greater amount of strong covalent bonds in the intermetallic structure [45,46]. Therefore, the Al₅Fe₂ phase served as the main source of Al³⁺, Fe²⁺/Fe³⁺, and Si⁴⁺ ions from the coating.

In the beginning of the corrosion process, boehmite formed. However, the main aluminum corrosion products were bayerite and aluminum chloride; see Equations (5) and (6).

Al³⁺ + 3OH^– → Al(OH)₃

(5)

Al³⁺ + 3Cl^– → AlCl₃

(6)

By a sequence of chemical reactions, these corrosion products can be mutually converted based on local conditions; see Equations (7)–(9). As an intermediate compound, cadwaladerite formed [47].

Al(OH)₃ + Cl^– ↔ Al(OH)₂Cl + OH^–

(7)

Al(OH)₂Cl + Cl^– ↔ Al(OH)Cl₂ + OH^–

(8)

Al(OH)Cl₂ + Cl^– ↔ AlCl₃ + OH^–

(9)

Analcime was identified as the only silicon-based stable corrosion product. Its formation is described by Equation (10). This compound typically forms at elevated pH [48]. The source of OH^– can be the chlorination of Al(OH)₃, see Equations (7)–(9), or ORR (Equation (3)). Due to the low amount of Si⁴⁺ ions generated by the corrosion process, and therefore its minimal representation in the stable corrosion products, the influence of this phase on the corrosion process is expected to be rather negligible. This is confirmed by its relatively stable ratio in corrosion products after different corrosion tests from 4 to 6 wt.%; see Table 6.

Na⁺ + Al³⁺ + 2SiO₃^2– ↔ NaAlSi₂O₆

(10)

Two principal corrosion products of iron were identified. Akaganeite was mainly located in the outer layer of the corrosion products. Akaganeite requires chlorides and a pH in the range of 4–6 to form, Equation (11) [35]. Hydrolysis of Al³⁺ and Fe²⁺/Fe³⁺ ions as well as the formation of dawsonite served as the source of H⁺ ions, Equations (1) and (2).

FeCl₂ → β-Fe₂(OH)₃Cl → Fe₃FeCl₂(OH)₇ → β-FeOOH

(11)

Magnetite was located mainly inside corrosion pits. Therefore, it can be assumed that the main source of Fe²⁺ ions for its formation was the corrosion of steel. A pit environment with a limited oxygen access promotes the magnetite formation by the reduction of iron oxohydroxides according to Equation (12) [49,50].

Fe²⁺ + 8FeOOH + 2e^– → 3Fe₃O₄ + 4H₂O

(12)

Figure 9 shows that there was no significant decrease in the consumption of Cl^– for the formation of stable corrosion products during the exposure, which was around 80%. Simultaneously, the amount of soluble metal ions increased over time. Thus, it can be assumed that the corrosion rate of PHS AS was about constant. With the exception of dawsonite, the corrosion products mentioned above do not have significant protective capabilities. On the contrary, the transformation of AlCl₃ to bayerite releases previously bound chloride ions, which may further accelerate the corrosion of steel.

Figure 16 shows a schematic of the corrosion mechanism of PHS AS under atmospheric conditions in the presence of chlorides, summarizing the above reactions. The schematic also highlights the distribution of various corrosion products within the pit and in its vicinity. It should be noted that the corrosion exposure conditions were harsh, leading to an intense degradation of the material. It also needs to be pointed out that PHS AS is in practice usually protected by organic coatings. Performed corrosion exposures in this work lasted for four weeks. Regarding the development of corrosion products, it can be assumed that in the longer term under natural atmospheric conditions, further transformations may occur.

4.2. Influence of Environmental Conditions on Corrosion Behaviour of PHS AS

The addition of 0.28 g∙m⁻² sulphates to 0.9 g∙m⁻² chlorides to the PHS AS surface led to an increase in mass loss of almost 65%. Basaluminite was identified as the only sulphate-based corrosion product. It is an amorphous, insoluble corrosion product whose formation is favoured over the crystalline form of aluminum sulphates under outdoor conditions. Its formation is described by Equation (13) [37,51,52].

4Al³⁺ + SO₄^2– + 10OH^– → Al₄SO₄(OH)₁₀

(13)

It is well known that the presence of sulphates leads to a decrease in the pH and a significant acceleration of the corrosion of pure aluminum [53,54]. On PHS AS, it can be assumed that the aforementioned local acidification promoted the corrosion of the underlying steel substrate. This is supported by the observed higher depth of corrosion pits in steel after the Sulphates test (Figure 4) and a link between the iron content in stable corrosion products and mass loss (Figure 12). Table 6 shows that despite the presence of basaluminite with high molar mass, the content of Fe-based corrosion products was higher compared to the Initial conditions exposure. The lower pH may also have suppressed the formation of dawsonite, which is stable under alkaline conditions [32]. The proportion of dawsonite in the Sulphates corrosion products was 62% lower.

It the Low chlorides and Low RH tests, the amount of chlorides and the level of relative humidity in the wet phase were reduced, respectively. The tenfold reduction in the amount of chlorides had a significant effect on the overall behaviour of PHS AS as no corrosion attack in steel was detected and mass loss dropped by a factor of 6. Reduction in RH from 95 to 80% did not cause a decrease in mass loss but more pits in steel with a lower depth formed. It is likely that a thinner electrolyte presence at 80% RH led to the formation of more local anodes and cathodes. Overall, the lower RH reduced the corrosion of steel and pronounced the corrosion of the coating. This is supported by the data in Table 4, which show higher Al:Fe ratios in both water-soluble and stable corrosion products for the Low RH test.

When the dry phase was left out in the Constant RH exposure, the corrosion rate increased by a factor of 2 compared to the Initial conditions. The continuous presence of the surface electrolyte prevented any slowdown in the corrosion process in cracks and pits, which would be expected during the dry phase at 50% RH. It led to an increased corrosion depth in steel (Figure 4) and the highest akaganeite and magnetite contents among the tests where pits were observed (Table 6). Also, the lowest amount of dawsonite formed under these conditions. Only dawsonite can be considered as a protective corrosion product as it helps to decrease the effective cathodic area [32]. At constant 95% RH, a thick layer of the surface electrolyte formed, limiting the transport of CO₂ to the metal and carbonate formation.

With the exception of the reduced dawsonite content in the tests with the highest mass loss, no other correlation was observed between the composition of the stable corrosion products and the extent of corrosion. It can thus be concluded that the compounds formed during the atmospheric corrosion of PHS AS had a low protective ability.

5. Conclusions

The following conclusions can be drawn:

• Press-hardened steel coated with Al-Si was susceptible to the localized corrosion of steel in coating cracks under chloride deposits.
• The Al-Si coating acted as a barrier between the corrosive environment and steel. It did not provide galvanic protection to the substrate. High iron content of the coating participated in the formation of red corrosion products immediately at the beginning of the exposure.
• The Al₅Fe₂ phase corroded preferentially in the coating, whereas the Si-rich phases Al₈Fe₂Si and AlFe were more corrosion-resistant.
• Analcime, bayerite, AlCl₃, dawsonite, and akaganeite were identified as the main stable corrosion products. Magnetite formed preferentially in oxygen-depleted pits. The corrosion products had a low protective ability.
• Wet–dry cycling and a high level of chloride contamination led to a higher ratio of soluble corrosion products.
• The addition of sulphates to surface deposits increased the corrosion rate by 65%, probably due to the acidification of the surface electrolyte.
• Lower relative humidity caused a drop in the pit depth in steel at the expense of a more intense corrosion of the coating.
• Constant high relative humidity and the addition of sulphate to the surface deposits were critical for the acceleration of steel corrosion in coating cracks.