Corrosion Science 86 (2014) 231–238
Contents lists available at ScienceDirect
Corrosion Science journal homepage: www.elsevier.com/locate/corsci
Composition of corrosion products formed on Zn–Mg, Zn–Al and Zn–Al–Mg coatings in model atmospheric conditions Tomas Prosek a,⇑, Dan Persson b, Jan Stoulil c, Dominique Thierry a a
Institut de la Corrosion/French Corrosion Institute, 220 rue Pierre Rivoalon, 29200 Brest, France Swerea KIMAB, Isafjordsgatan 28A, 11428 Stockholm, Sweden c Institute of Chemical Technology in Prague, Technická 5, 166 28 Praha 6, Czech Republic b
a r t i c l e
i n f o
Article history: Received 17 April 2014 Accepted 27 May 2014 Available online 6 June 2014 Keywords: A. Metal coatings A. Zinc B. Weight loss C. Atmospheric corrosion
a b s t r a c t Formation of corrosion products on hot-dip galvanised steel (HDG), Zn–5Al, Zn–11Al–3Mg–0.2Si, Zn–16Mg and Zn–1.5Al–1.5Mg with pre-deposited NaCl was followed in humid air at 20 °C. The alloyed coatings showed an improvement in mass loss by a factor of 4–7 to HDG. Corrosion products on the alloyed coatings contained twice as much carbonates than those formed on HDG. Magnesium dissolved preferentially, and aluminium-enriched phases were the most stable. Magnesium buffered the pH at cathodic sites, thus hindering the formation of zinc oxide and inhibiting the oxygen reduction. Magnesium products at the metal/corrosion product interface might also have an inhibiting effect. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Zn–Mg(–Al) coated steel materials have attracted attention since they were shown to be able to provide corrosion stability superior to that of traditional zinc-coated steel. Initially, it was demonstrated on results from the neutral salt spray test (NSST) where 4- to 20-fold improvements to zinc-coated steel in terms of time to red rust were observed [1–4]. Hosking et al. tested coatings consisting of a 2-lm MgZn2 layer on a zinc coating in a cyclic automotive corrosion test . The time to appearance of signiﬁcant red rust was three times longer for the Zn–Mg coating than for zinc-coated references. Comparable improvement was found in open and conﬁned areas of Zn–2Mg–2Al exposed to the same test . It must indeed be noted that newer ﬁeld studies and experiments in less corrosive environments reveal that there are conditions where alloying with Mg and Al leads to only modest improvement in corrosion performance. Current data suggest that the alloy coatings provide a signiﬁcant improvement in comparison to conventional coatings in conditions involving a high salt load . Several research groups have carried out studies aimed at understanding the corrosion mechanisms of the alloyed coatings. Corrosion products formed during NSST [2,3,8] and cyclic corrosion tests [5,9–13] were thoroughly characterised. The studies provided important information and helped to formulate several hypotheses ⇑ Corresponding author. Tel.: +33 298 058 905; fax: +33 298 050 894. E-mail address: [email protected]
(T. Prosek). http://dx.doi.org/10.1016/j.corsci.2014.05.016 0010-938X/Ó 2014 Elsevier Ltd. All rights reserved.
on the role of magnesium in the corrosion process. Their limitations can be seen in complexity and the generally harsh nature of selected corrosion tests. In particular, NSST is known to correlate poorly to any ﬁeld data for zinc-based coatings because of completely different corrosion mechanisms . Newer works have focused on the initial stages of corrosion using surface-sensitive analytical techniques [15,16]. To complete the two approaches, detailed analyses of corrosion products formed under chloride deposits in well-deﬁned and controlled no-rinse conditions were carried out within this study. This paper focuses on the chemical and phase composition of the corrosion products. In a separate paper, their electrochemical properties are characterised .
2. Experimental 2.1. Materials Steel materials with zinc-based coatings used in the study are listed in Table 1. Zn–1.5Al–1.5Mg (ZnAlMg), Zn–11Al–3Mg–0.2Si (ZnAlMgSi), Zn–5Al (Galfan, ZnAl) and Zn–0.2Al (hot-dip galvanised steel, HDG) were produced in continuous galvanising lines. Zn–16Mg (ZnMg) is a laboratory product prepared by deposition of a layer of 0.7 lm Mg on HDG panels with a coating weight of 70 g/m2 by the physical vapour deposition (PVD) process. Then, the panels were annealed at a peak temperature of 320 °C to allow inter-diffusion between Zn and Mg layers and the formation of the intermetallic phase MgZn2 on the top of the zinc coating. The
T. Prosek et al. / Corrosion Science 86 (2014) 231–238
Table 1 Materials.
Al content (wt.%) Mg content (wt.%) Coating weighta (g/m2) Coating thicknessb (lm) Treatment
1.5 1.5 140 11 ± 2 TR
11 3 100 10 ± 1 TR, O
– 16c 70 5±1 A
4.5 – 265 20 ± 3 TR
0.2 – 275 17 ± 3 TR
TR: temper rolling; O: oiling; A: annealing. a Two sides. b Single side, measured by a metallographic microscope at cross-sections at 20 random points. c Mg content in the outer coating layer; average Mg content was 3.4 wt.%.
MgZn2 layer thickness was less than 2 lm. Details on the ZnMg panel preparation are given elsewhere . ZnAlMg is composed of primary Zn dendrites, Zn–MgZn2 binary eutectic and Zn–Al–MgZn2 ternary eutectic [1,2]. ZnAl consists of a zinc phase with approximately 1% aluminium and a lamellar eutectic of zinc containing approximately 5% aluminium . The former phase forms primarily at the steel surface and is usually found in low quantities. ZnAlMgSi consists of a large amount of primary Al-rich phase, MgZn2 crystals and ternary Zn–Al–MgZn2 eutectic. Some Mg–Si particles may be present at the steel surface . 2.2. Test procedure Metal panels were cut to a size of 100 100 mm and 50 50 mm for mass loss measurements and analyses of corrosion products, respectively. The panels were degreased with acetone, gently abraded with 1200/P2500 emery paper, again cleaned with acetone and pre-exposed in air at 50% RH and at 20 °C for 24 h. Individual emery papers were used for each material to avoid surface contamination. This procedure was used in order to remove oil and other organic contaminants and oxide layers formed during material storage and handling and to build natural oxide layers under controlled conditions. After pre-exposure, a solution of 7.21 g/l NaCl in methanol (>99.5%) was applied to the sample surfaces. On each sample of 50 50 mm, 800 ll of the solution was applied in 8 separate doses of 100 ll each using a micropipette; four times more solution was used to contaminate the larger panels. In all cases, the solution on the sample surface was uniformly spread with a micropipette tip and dried out before the application of the next dose. This procedure ensured even distribution of salt deposits over the surface. The ﬁnal chloride surface concentration of 1.4 g/m2 was selected to model conditions in natural coastal atmospheres and on surfaces that are in contact with de-icing salts. Deposition rates from 0.01 to 4 g/m2 day of Cl have been reported in coastal atmospheres [21,22]. The contaminated samples together with non-contaminated reference samples were transferred to a climatic chamber and exposed to 80% relative humidity (RH) and 20 °C for 1, 2, 5, 12 and 28 days. After each exposure period, sets of four 100 100-mm samples and six 50 50-mm samples were removed. In parallel, four other replicates of each alloy were treated identically, except for the contamination with sodium chloride, and they were kept in a desiccator as blanks. Wet mass gain was measured within 15 min of removal of the 100 100 mm samples from the climatic chamber. Dry mass gain was measured after drying for 7 days in a desiccator over silica gel at RH < 15%. Soluble corrosion products were dissolved at laboratory temperatures in 600 ml of deionised water in an ultrasonic bath. Mass loss was measured after removal of corrosion products in 600 ml of saturated water solution of glycine (NH2CH2COOH) in an ultrasonic bath at laboratory temperature following the
ISO 8407 standard. The pickling was carried out in 4-min steps followed by sample drying in warm air and weighing. It was repeated until all corrosion products were dissolved. Complete removal of corrosion products was veriﬁed by FTIR microscopy. The procedure was ﬁrstly tested on non-corroded samples of all compositions. Fresh glycine solution was used for each sample. 2.3. Corrosion products analyses Fourier-transformed infrared spectroscopy (FTIR) was used to assess the phase composition of corrosion products. Samples were analysed using the KBr pellet method. A small mass of 0.70 mg of corrosion products scratched from a sample or a reference mineral was mixed with a spectroscopic grade potassium bromide (KBr) powder to obtain 140 mg of a mixture. The mixture was then grinded in a mortar and pressed to pellets in a die. Spectra were obtained by adding 128 scans at 8 cm1 resolution in the range from 250 to 4000 cm1. The background spectrum was obtained from a pure KBr pellet of the same weight and size. Additional analyses were performed using X-ray diffraction (XRD) with a Bruker AXS D8 Discover. A Cu-anode was used and the detector was a SolX with energy window for Cu Ka. A Göbel mirror was used on the detector side. Measurements were performed both at 5° angle of incidence and a grazing angle geometry at 0.2°. The concentrations of Zn2+, Mg2+, Na+, and Cl ions in water and glycine solutions were measured with ion chromatography (IC) with a Metrohm ion chromatograph with a 732 IC conductivity detector. The detection limit was below 0.2 ppm (ppm), i.e. 12 mg/m2 for each ion. The concentration of aluminium was analysed in solution for selected samples in an external laboratory by inductively coupled plasma optical emission spectrometry (ICP-OES) according to ISO 11885 with the detection limit of 10 parts per billion (ppb), i.e. 0.6 mg/m2. Glow discharge optical emission spectroscopy (GD-OES) was applied to get depth proﬁles of corrosion product ﬁlms. GD-OES Proﬁler 2 of Horiba Jobin Yvon was used. Analysed areas had diameter of 5 mm. The sputtering rate was approximately 1 lm/min. The ﬁnal depth after analyses was veriﬁed by an optical proﬁlometer. The amount of carbonate in corrosion products was determined using a procedure developed by Falk, Svensson and colleagues [23,24]. A sample of 50 50 mm covered with corrosion products was transferred to a glass reactor vessel with 500 ml of 1 M HClO4. Carbonate decomposed quantitatively to carbon dioxide. It was carried by CO2-free air at a ﬂow rate of 500 ml/min to a CO2 analyser WMA4 by PP Systems. Recorded carbon dioxide concentrations were integrated, and the total carbonate content in corrosion products was calculated. The setup was calibrated using known amounts of CaCO3. A schematic drawing of the setup is given in Fig. 1. 3. Results 3.1. Mass loss Mass loss data after 28 days are shown in Fig. 2. The materials can be ranked as follows (from the least to most corroded):
Fig. 1. Schematic drawing of setup for carbonate analyses; 1: pressurised air inlet; 2: air dryer; 3: CO2 absorber; 4: ﬂow meter with reduction valve; 5: reactor vessel with HClO4 solution and a movable sample ﬁxed with a magnet; 6: CO2 detector; 7: exhaust air outlet.
T. Prosek et al. / Corrosion Science 86 (2014) 231–238
Mass loss [g/m²]
15 12 9 14.7
6 3 3.4
Fig. 2. Mass loss of coated steel contaminated with NaCl after 28 days of exposure to humid air.
ZnMg < ZnAl < ZnAlMgSi < ZnAlMg HDG. ZnMg gave 7-fold lower mass loss than HDG. ZnAl, ZnAlMgSi and ZnAlMg were six, ﬁve and four times, respectively, better than the reference material. However, the differences between the alloyed coatings were low considering the experimental error. This is particularly true when ZnAlMgSi and ZnAlMg are compared. Kinetics of mass loss is seen in Fig. 3. The corrosion rate of HDG was rapid in the beginning giving rise to mass loss of 6.1 g/m2 in ﬁrst 24 h of exposure, i.e. 41% of the total mass loss. The later decrease in corrosion rate can be attributed to a decreasing quantity of free sodium and chloride ions on the surface and the protective action of the formed corrosion products. Corrosion kinetics was different for the alloyed materials with mass loss of only 0.5 g/m2 (20% of the total) after 24 h. Their corrosion rate also decreased over time, but the decrease was less steep, which is probably due to a higher amount of non-reacted activators in the surface electrolyte. Consequently, the improvement in mass loss to HDG decreased with time as displayed in Fig. 3. It dropped from 10- to 15-fold after 1 day to 4- to 7-fold after 28 days. The drop was stronger for complex multi-phase coatings ZnAlMg and ZnAlMgSi. The corrosion kinetics of ZnMg and ZnAl were closer to that of HDG. 3.2. Elemental analysis of corrosion products Data on the composition of water soluble and stable corrosion products dissolved in glycine in terms of Cl, Na+, Zn2+, Mg2+ and Al3+ and data on carbonate content, mass gain and mass loss were
combined to obtain a complex picture of the surface chemistry after 28 days of exposure (Fig. 4). Of the total 908 mg/m2 sodium applied prior to the exposure, 69% was bonded in corrosion products on the surface of HDG. This ratio was of only 2–7% for all alloyed coatings. Similarly, 93% of chloride ions was incorporated in stable corrosion products on HDG, whereas most chloride ions were soluble on the alloyed coatings. The data suggest that the build-up of sodium and chloride into stable corrosion products was inhibited most efﬁciently on coatings containing magnesium. Only a small portion of zinc was present in a water-soluble form. On HDG, the percentage was below 1%. The most soluble zinc (7%) was found on ZnAl. Except for the ﬁrst 2 days, the absolute quantity of water-soluble zinc in corrosion products was almost constant. The amount of magnesium in corrosion products increased with the magnesium content in coatings. From 22% to 40% of corroded magnesium was water soluble. No aluminium was detected in water extracts of the corrosion products. Since the detection limit was 0.6 mg/m2, there was less than 1% of water-soluble aluminium on the coating surface. It is interesting to compare ratios of the alloying metals in the coatings and in the corrosion products (Fig. 5). Obtained data clearly prove that magnesium dissolved preferentially, whereas aluminium corroded less readily than zinc. There was 114%, 38% and 21% more magnesium in corrosion products than in the respective ZnAlMg, ZnAlMgSi and ZnMg coatings. Corrosion products on ZnAlMg, ZnAlMgSi and ZnAl were depleted of aluminium by 19%, 79% and 40%. Preferential magnesium dissolution from Mg-rich phases [2,15,25,26] and higher stability of Al-containing phases  in similar structures have been reported previously. The ratio of magnesium in corrosion products on ZnAlMg decreased in time. It contained 7.5% Mg after 2 days of exposure and 3.2% after 28 days. Similarly, the Mg content in corrosion products on ZnMg dropped from 89% after 1 day to 19% after 28 days. This shows on depletion of magnesium in the outer layer of the non-corroded coating. Rather constant Mg content varying from 3.5% to 4.4% was found on ZnAlMgSi. Fig. 6 shows carbonate mass vs. dry mass gain, i.e. mass change due to reaction of applied NaCl and corroded metal with air components. The lowest carbonate to dry mass gain ratio of 44% after 28 days was obtained for HDG. It was 49%, 61%, 62% and 84% for ZnAlMgSi, ZnAlMg, ZnAl and ZnMg, respectively. The ratio tended to increase in time. When compared to mass loss, there were 0.6, 0.5, 0.4, 0.3 and 0.3 g of carbonate per gram of corroded metal on ZnMg, ZnAl, ZnAlMgSi, ZnAlMg and HDG, respectively, after 28 days of exposure. Thus, the relative carbonate content in corrosion products increased with decreasing mass loss. GD-OES depth proﬁling revealed that the approximate thickness of layers of corrosion products on chloride contaminated samples exposed to humid air for 28 days decreased in the following order: HDG (7 lm) > ZnAlMg ZnAl ZnAlMgSi (1 lm) > ZnMg (0.5 lm). Magnesium enrichment in the inner layer of corrosion products adjacent to non-corroded metal was observed on all Mg-containing materials exposed both with and without chloride pre-contamination. An example of a GD-OES proﬁle is seen in Fig. 7. These observations are in good agreement with the GDOES data for model Zn-Mg alloys published earlier  and newer X-ray photoelectron spectroscopy (XPS) results from Diler et al. . Aluminium was depleted in the corrosion products. 3.3. Phase composition of corrosion products
Fig. 3. Development of mass loss of materials contaminated with NaCl and exposed to humid air in time; numbers above points give relative improvement to HDG.
Phase composition of dry corrosion products was analysed by FTIR. Most measurements were performed using the KBr pellet method, but local measurements by infrared microscope were also carried out. In addition, samples exposed for 28 days were
T. Prosek et al. / Corrosion Science 86 (2014) 231–238
Alloying metal [wt. %]
12 8 4 0 Coating
Fig. 5. Ratio of alloying metals in coatings and corrosion products (CP) formed on samples with chloride deposits exposed for 28 days to humid air.
analysed by XRD. Table 2 summarises identiﬁed and suggested phases giving indication on the relative content and changes from day 1 to day 28. On HDG, corrosion products were dominated by simonkolleite, Zn5(OH)8Cl2H2O, in the ﬁrst days of exposure. With time, the simonkolleite relative content decreased due to the formation of an increasing amount of hydrozincite, Zn5(OH)6(CO3)2. Zincite, ZnO, was also detected. It was more easily seen in the shorter exposure times indicating that its relative content decreased in time. This may be connected to a decrease in the corrosion rate. Presence of sodium carbonate, Na2CO3(xH2O) and smithonite, ZnCO3, is more uncertain, probably because of the low amounts compared to simonkolleite and hydrozincite. Sodium carbonate was clearly observable after 1 day of exposure. XRD spectra indicated the presence of Na2[Zn(OH)4]. Zinc and sodium carbonate
Carbonate in dry mass gain [wt. %]
Fig. 4. Distribution of species in corrosion products formed on samples contaminated with NaCl and exposed to humid air for 28 days, in g/m2; percentage of species in stable products (water insoluble) is given; numbers in brackets give ratios of corroded Zn/Mg/Al.
Time [days] (CO2 3 )
Fig. 6. Mass of carbonate in total dry mass gain on samples contaminated with NaCl and exposed to humid air.
and Na2[Zn(OH)4] are water-soluble products that would not be present on materials prone to regular rinsing with rain water. Local analysis using FTIR microscopy showed areas dominated either with simonkolleite or carbonates indicating a strong separation of anodic and cathodic processes. Hydrozincite dominated in corrosion products on ZnAlMg. Simonkolleite was found only on the sample exposed for 1 day. A low amount of layered double hydroxides (LDH) was identiﬁed by FTIR peaks at 1355 and 1110 cm1 and by XRD. The LDH have the general formula [M2+x M3+y (OH)2(x+y)][An]zH2O. According to Persson et al., LDH on ZnAlMg with similar composition contained aluminium as M3+, zinc and magnesium as M2+ and carbonate as An . The presence of zinc-containing LDH on Zn–Al–Mg coatings has been reported previously by other authors [2,3,10].
T. Prosek et al. / Corrosion Science 86 (2014) 231–238
products can be proposed. They are shown in Table 3 for HDG and ZnAlMg. The models were constructed under following assumptions: (1) soluble Mg2+ is present as MgCl2; (2) soluble Zn2+ is in ZnCl2 (ZnAlMg) or split between ZnCl2 and ZnCO3 on an equimolar basis (HDG); (3) the remaining soluble Cl is in NaCl; (4) the remaining soluble Na+ is in Na2CO3H2O (ZnAlMg) or split between Na2CO3H2O and Na2[Zn(OH)4] on an equimolar basis (HDG); (5) the insoluble Na+ is in Na2Zn3(CO3)43H2O (HDG) or Na2ZnCl43H2O (ZnAlMg); (6) the insoluble Cl is in simonkolleite (HDG) or Na2ZnCl43H2O and HZn2Cl52H2O (ZnAlMg); (7) the insoluble Al3+ is split on the equimolar basis between Zn/Al and Mg/Al LDHs with formulas Zn4Al2(OH)12(CO3) and Mg6Al2(OH)12 (CO3)H2O ; (8) the remaining carbonate is in hydrozincite; and (9) the remaining zinc is in ZnO (HDG). According to the model, there is a large amount of Na2Zn3 (CO3)43H2O on HDG. The presence of this compound is speculative as it was detected by neither FTIR nor XRD. However, the data in Fig. 4 evidence that there is an insoluble phase containing sodium ions on HDG and Na2Zn3(CO3)43H2O was identiﬁed as a product under similar conditions . Some insoluble zinc is unattributed on ZnAlMg. It is possible that this zinc is bonded with carbonate and present as hydrozincite and there is less soluble Na+ in sodium carbonate than indicated. Na2Zn3(CO3)43H2O or similar compounds may also be present. No magnesium compound was detected. Still, there was an important amount of magnesium in corrosion products that was not dissolved in water (Fig. 4). Results of the study performed by Persson et al.  indicate that the insoluble magnesium could be partly present in layered double hydroxides (LDH). LDH are good ion exchangers, and the compound Mg0.85Al0.15(OH)2(CO3)0.75 has been reported to form on AZ91 on the Al-rich a phase . The remaining magnesium was probably present at the interface between the metal coating and corrosion products as magnesium hydroxide, carbonate or a mixture or compound of the two. Hydrozincite formed 84 wt.% of the attributed stable phases.
Depth [µm] Fig. 7. GD-OES proﬁle of ZnMg contaminated with NaCl and exposed to humid air for 28 days.
Hydrozincite and other carbonates were major corrosion products on ZnMg. Simonkolleite was detected only on the sample exposed for 1 day. Water content decreased in time. Magnesite MgCO3 and Mg2(CO3)(OH)Cl(H2O)2 were identiﬁed with a low degree of certainty in FTIR and XRD spectra, respectively. Sodium carbonate was found by both techniques. Only traces of simonkolleite were detected on ZnAlMgSi. Hydrozincite dominated in the corrosion products. The peak at 1460 cm1 indicated the presence of sodium carbonate. NaAl23O35 and LDH were also detected. Hydrozincite and other carbonates dominated on ZnAl. The amount of simonkolleite increased somewhat more rapidly than that of hydrozincite. Presence of LDH was revealed. Carbonate compounds dominated on all materials except HDG. The signal in a range from 1590 to 1330 cm1 was complex with several non-attributed peaks belonging to yet-unidentiﬁed carbonate-containing phases.
4. Discussion 3.4. Model for the composition of corrosion products It was reported that the presence of magnesium in zinc-based coatings led to the preferential formation of simonkolleite in cyclic corrosion tests [5,9–11]. The suppression of hydrozincite formation
Based on the chemical and phase composition data, tentative quantitative models on the phase composition of the corrosion
Table 2 Phase composition of corrosion products. Phase
Hydrozincite Zn5(OH)6(CO3)2 Simonkolleite Zn5(OH)8Cl2H2O Zincite ZnO LDH Sodium carbonate Na2CO3(H2O) Other carbonate(s) Sodium chloride NaCl NaAl23O35 Water Smithonite ZnCO3 Na2[Zn(OH)4] Magnesite MgCO3 Mg2(CO3)(OH)Cl2H2O Na2ZnCl43H2O HZn2Cl52H2O Magnesium chloride MgCl2
++++ +c + + ++b ++a +a ++ +a +a
++++ + +b +b ++b ++a ++a ++
++++ +c +a ++ ++b ++a ++ +b +a +a
++++ +++ ++ ++b ++a ++
HDG Dv. " ;
+++ ++++ ++ ++ ++b + +a
" ; ; ;
Final: component present at the end of exposure; ++++: dominant corrosion product; +++: phase detected with high degree of certainty at high relative concentration; ++: medium degree of certainty and/or lower relative amount; +: low degree of certainty; : not detected; Dv.: development in relative quantity in time; ": increase; ;: decrease. a Detected by XRD. b Detected by FTIR. c Observed only after short exposure times.
T. Prosek et al. / Corrosion Science 86 (2014) 231–238
Table 3 Tentative quantitative models of phase composition of dry corrosion products on HDG and ZnAlMg after 28 days of exposure to humid air; in wt.%. Phase
Hydrozincite Zn5(OH)6(CO3)2 Simonkolleite Zn5(OH)8Cl2H2O Zincite ZnO Na2Zn3(CO3)43H2O LDH Zn4Al2(OH)12(CO3) LDH Mg6Al2(OH)12(CO3)H2O MgCO3, Mg(OH)2 HZn2Cl52H2O Na2ZnCl43H2O
18 37 13 27 – – – – –
49 – – – 3 2 1 1 1
Sodium chloride NaCl (sol) Zinc chloride ZnCl2 (sol) Magnesium chloride MgCl2 (sol) Sodium carbonate Na2CO3H2O (sol) Smithonite ZnCO3 (sol) Na2[Zn(OH)4] (sol)
Low Low – 1 Low 2
24 4 1 6 – –
Compounds marked (sol) are readily soluble in water; low: <0.5 wt.%; non-assigned species. a O2, OH, bond H2O. b Mainly insoluble Zn2+.
was observed also when zinc covered with MgCl2 deposits was exposed to humid air . Several authors have attributed the improved performance of Zn–Mg(–Al) to the better protection ability of simonkolleite [5,9]. The current work as well as our previous study on model Zn–Mg alloys  showed that hydrozincite and other carbonate compounds formed more readily in presence of Mg and Al and dominated in the corrosion products on alloyed coatings (Fig. 6). Thus, the superior corrosion performance of Zn–Mg(–Al) under current exposure conditions cannot be explained by any inhibition effect of simonkolleite . The corrosion stability of Zn–2Al–2Mg in NSST was connected to formation of LDH [2,3]. In conditions of this study, LDH were present on the surface of ZnAlMg, ZnAlMgSi and ZnAl with fourto six-times lower mass loss compared to HDG. However, ZnMg containing no aluminium and thus forming no LDH performed even better. The role of LDH in the corrosion protection of Al-containing alloy coatings thus needs to be clariﬁed in further research. Zn–Mg and Zn–Al–Mg coatings were shown to be more protective than traditional zinc coatings in conditions where hydrozincite , simonkolleite [5,9–11] and LDH [2,3] prevailed on corroded surfaces. The composition of bulk corrosion products in terms of these compounds thus does not seem to be of the utmost importance for protection efﬁciency. It is probable that the protection provided by these products is not dramatically different. Their barrier effect and inhibition efﬁciency may depend more on the structure and porosity of corrosion product ﬁlms. The chemical composition and structure of corrosion products is controlled by surface chemistry. For example, LeBozec et al. proposed that the presence of LDH on corroded coatings is a consequence of high pH at cathodic sites, leading to the dissolution of aluminium . The relative content of LDH in corrosion products then correlates to the aggressiveness of exposure conditions, speciﬁcally to a chloride load. Systematic difference between HDG and the alloyed coatings was seen in terms of presence of zincite, ZnO. It was not detected on the surface of ZnAl, ZnAlMg and ZnAlMgSi, and its quantity on ZnMg was negligible. Similar results were obtained in in situ experiments of Persson et al. . Zincite is an undesirable corrosion product. It is a semiconductor with rather good electric conductivity, and the rate of oxygen reduction on surfaces covered with ZnO is relatively high [25,31]. Zincite forms at cathodic locations at high pH. It has been proposed previously that the inhibiting effect of
Fig. 8. SEM image of a cross-section of ZnAlMgSi after 28 days of exposure to humid air with superimposed EDX data; 1: primary Al-rich phase; 2: MgZn2; 3: ternary eutectic Zn–Al–MgZn2; 4: C (carbonate) rich (cathodic) area; 5: Cl (chloride) rich (anodic) area.
magnesium may be due to buffering of cathodic sites at slightly alkali pH favouring formation of low soluble products blocking the oxygen reduction [5,9,32]. Since simonkolleite is stable at an approximately neutral pH, it is supposed to comprise mainly hydrozincite or other carbonate-containing compounds. Similarly, aluminium dissolution at cathodic sites of ZnAl and the other Al-containing materials  can lead to the formation of LDH instead of zincite. An important factor in the mechanism of magnesium protection may be its afﬁnity to CO2 and surface carbonisation . It was found in this study that the relative content of carbonate in dry corrosion products correlated with the coating stability, i.e. there was more carbonate on less corroded materials. It is well known that presence of CO2 is crucial for good corrosion stability of zinc in atmospheric conditions . Recently, it was shown that corrosion rates of Zn–Al–Mg coatings increase dramatically in air depleted of CO2 . Carbon dioxide decreases the pH of the surface electrolyte and allows for formation of protective corrosion products. In atmospheric conditions, magnesium (hydr)oxide reacts very rapidly to carbonate species . Thermodynamic calculations showed that such reactions are preferential to those of zinc [5,10]. It is probable that magnesium ions accelerate the build-up of dissolved carbonate and bicarbonate into the protective ﬁlm. Thus, it is possible to suppose that cathodic sites are rapidly blocked in the presence of magnesium by formation of magnesium (hydroxy) carbonate, and the pH is buffered so that the formation of zinc oxide or even zincates is hindered. It is known that the rate of oxygen reduction controlling the overall corrosion process is elevated on zinc oxide. Indeed, due to the rather high solubility of magnesium carbonate and most other magnesium compounds, such effects would be short term in rinsing conditions. However, it can inhibit the rapid onset of corrosion in harsh exposure conditions and allows for formation of stable zinc-based corrosion products, i.e. hydrozincite and simonkolleite, as observed by Hosking et al. . In the presence of aluminium, zinc oxide can be replaced by LDH with low solubility and expected long-term protection ability. Besides the role of magnesium in the surface chemistry discussed above, Mg-containing products may also have a direct protective effect. Oxidised magnesium accumulates at the interface between the coating and corrosion products (Fig. 7). This was documented previously on model Zn–Mg alloys . It was shown that the electronic properties of Mg-based oxidic layers are superior to those containing zinc with respect to corrosion
T. Prosek et al. / Corrosion Science 86 (2014) 231–238
[31,35,36]. Although the Mg-rich interlayer is probably not formed of MgO but more of MgCO3/Mg(OH)2, it can serve as a barrier for ion and electron transfer, thus limiting the overall corrosion rate . Strong differentiation in composition of corrosion products across the surface was noted on all tested materials suggesting separation of anodic and cathodic areas, see an example in Fig. 8. In order to understand fully the role of magnesium and aluminium in the corrosion process, it seems important to focus on microscale surface chemistry in further studies.
5. Conclusions Alloy zinc coatings containing up to 11 wt.% Al and up to 16 wt.% Mg showed an improvement in mass loss by a factor of 4–7 compared to HDG. The improvement was almost independent of the magnesium and aluminium content. Still, the lowest mass loss was observed for the most alloyed ZnMg with 16 wt.% Mg. Unlike on HDG, sodium chloride tended to stay dissolved in the surface electrolyte on the alloyed materials even after 28 days of exposure. The carbonate content in the corrosion products on the alloyed coatings was up to 2-fold higher than on HDG. Hydrozincite was a dominant corrosion product on the alloy materials, whereas simonkolleite prevailed on HDG. The corrosion products contained small amounts of aluminium and were enriched with magnesium. Magnesium was present preferentially close to the metal/corrosion product interface. The alloying hindered formation of zinc oxide. The following corrosion mechanism for Zn–Mg(–Al) is proposed. Magnesium dissolves preferentially at anodic locations, diffuses and migrates to cathodic sites, and precipitates forming magnesium hydroxide and magnesium carbonate. This process buffers the pH at the cathodic sites hindering formation of zinc oxide and thus inhibiting oxygen reduction. In the presence of aluminium, zinc oxide can be replaced by LDH. Bulk corrosion products formed in later stages contain almost entirely stable zinc-based compounds. Additional inhibition effects of the underlying Mg-based ﬁlm at the metal/corrosion products interface on the corrosion stability remain to be proven.
Acknowledgements The authors wish to thank to Bert Scheffel of FraunhoferInstitut für Elektronenstrahl- und Plasmatechnik for providing ZnMg and Ari Peltola of Rautaruukki for providing ZnAlMg. Anne Le Gac and Cecile Hall of French Corrosion Institute are thanked for help with the experimental work. References  M. Vlot, R. Bleeker, T. Maalman, E. van Perlstein, MagiZincTM, A new generation of hot-dip galvanised products, in: Proc. Galvanized Steel Sheet Forum 2006, Dusseldorf, 2006.  S. Schürz, M. Fleischanderl, G.H. Luckeneder, K. Preis, T. Haunschmied, G. Mori, A.C. Kneissl, Corrosion behaviour of Zn–Al–Mg coated steel sheet in sodium chloride-containing environment, Corros. Sci. 51 (2009) 2355–2363.  S. Schürz, G.H. Luckeneder, M. Fleischanderl, P. Mack, H. Gsaller, A.C. Kneissl, G. Mori, Chemistry of corrosion products on Zn–Al–Mg alloy coated steel, Corros. Sci. 52 (2010) 3271–3279.  T. Koll, K. Ullrich, J. Faderl, J. Hagler, A. Spalek, Properties and potential applications of ZnMg-alloy-coatings on steel sheet by PVD, in: Proc. Galvatech ‘04, 6th International Conference on Zinc and Zinc Alloy Coated Steel Sheet, Chicago, IL, United States, 2004, pp. 803–812.  N.C. Hosking, M.A. Strom, P.H. Shipway, C.D. Rudd, Corrosion resistance of zinc–magnesium coated steel, Corros. Sci. 49 (2007) 3669–3695.  T. Prosek, N. Larché, M. Vlot, F. Goodwin, D. Thierry, Corrosion performance of Zn–Al–Mg coatings in open and conﬁned zones in conditions simulating automotive applications, Mater. Corros. 60 (2010) 412–420.  N. LeBozec, D. Thierry, A. Peltola, L. Luxem, G. Luckeneder, G. Marchiaro, M. Rohwerder, Corrosion performance of Zn–Mg–Al coated steel in accelerated
corrosion tests used in the automotive industry and ﬁeld exposures, Mater. Corros. 64 (2013) 969–978. T.A. Keppert, G.H. Luckeneder, K.H. Stellnberger, G. Mori, Inﬂuence of the pH value on the corrosion of Zn–Al–Mg hot-dip galvanized steel sheets in chloride containing environments, in: NACE Corrosion, Salt Lake City, USA, 2012. P. Volovitch, C. Allely, K. Ogle, Understanding corrosion via corrosion product characterization: I. Case study of the role of Mg alloying in Zn–Mg coating on steel, Corros. Sci. 51 (2009) 1251–1262. P. Volovitch, T.N. Vu, C. Allély, A. Abdel Aal, K. Ogle, Understanding corrosion via corrosion product characterization: II. Role of alloying elements in improving the corrosion resistance of Zn–Al–Mg coatings on steel, Corros. Sci. 53 (2011) 2437–2445. T. Tsujimura, A. Komatsu, A. Andoh, Inﬂuence of Mg content in coating layer and coating structure on corrosion resistance of hot-dip Zn–Al–Mg alloy coated steel sheet, in: Proc. Galvatech ‘01, International Conference on Zinc and Zinc Alloy Coated Steel, Brussels, Belgium, 2001, pp. 145–152. K. Nishimura, H. Shindo, K. Kato, Y. Morimoto, Microstructure and corrosion behaviour of Zn–Mg–Al hot dip galvanized steel sheet, in: Proc. Galvatech ‘98, 4th International Conference on Zinc and Zinc Alloy Coated Steel, Chiba, Japan, 1998, pp. 437–442. N. LeBozec, D. Thierry, M. Rohwerder, D. Persson, G. Luckeneder, L. Luxem, Effect of carbon dioxide on the atmospheric corrosion of Zn–Mg–Al coated steel, Corros. Sci. 74 (2013) 379–386. N. LeBozec, N. Blandin, D. Thierry, Accelerated corrosion tests in the automotive industry: a comparison of the performance towards cosmetic corrosion, Mater. Corros. 59 (2008) 889–894. D. Persson, D. Thierry, N. LeBozec, T. Prosek, In situ infrared reﬂection spectroscopy studies of the initial atmospheric corrosion of Zn–Al–Mg coated steel, Corros. Sci. 72 (2013) 54–63. J. Duchoslav, M. Arndt, R. Steinberger, T. Keppert, G. Luckeneder, K.H. Stellnberger, J. Hagler, C.K. Riener, G. Angeli, D. Stifter, Nanoscopic view on the initial stages of corrosion of hot dip galvanized Zn–Mg–Al coatings, Corros. Sci. 83 (2014) 327–334. J. Stoulil, T. Prosek, A. Nazarov, J. Oswald, P. Krˇízˇ, D. Thierry, Electrochemical properties of corrosion products formed on Zn–Mg, Zn–Al and Zn–Al–Mg coatings in model atmospheric conditions, Corros. Sci., submitted for publication. B. Scheffel, K.-D. Ehlers, B. Schuhmacher, F.-J. Floßdorf, J. Hagler, C. Metzner, R. Steffen, G. Steinbeck, New types of coating systems for steel sheets by highrate evaporation in combination with plasma processes, Steel Res. 73 (2002) 114–122. J. Elvins, J.A. Spittle, J.H. Sullivan, D.A. Worsley, The effect of magnesium additions on the microstructure and cut edge corrosion resistance of zinc aluminium alloy galvanised steel, Corros. Sci. 50 (2008) 1650–1658. E. de Bruycker, Z. Zermout, B.C. De Cooman, Zn–Al–Mg coatings: thermodynamic analysis and microstructure related properties, Mater. Sci. Forum 539–543 (2007) 1276–1281. K. Barton, T.D. Cuc, S. Bartonova, Einﬂuss der Kationenzusammensetzung von Chloridlösungen beim Modellieren von Korrosionseinﬂüssen der Küstenatmosphären, Mater. Corros. 28 (1977) 17–19. I.S. Cole, W.D. Ganther, D.A. Paterson, G.A. King, S.A. Furman, D. Lau, Holistic model for atmospheric corrosion: Part 2 – experimental measurement of deposition of marine salts in a number of long range studies, Corros. Eng. Sci. Technol. 38 (2003) 259–266. T. Falk, J.-E. Svensson, L.-G. Johansson, The inﬂuence of CO2 and NaCl on the atmospheric corrosion of zinc: a laboratory study, J. Electrochem. Soc. 145 (1998) 2993–2999. D.B. Blücher, J-E. Svensson, L.-G. Johansson, Carbon dioxide: the unknown factor in the atmospheric corrosion of aluminium, in: Proc. 15th ICC, Granada, Spain, 2002. T. Prosek, A. Nazarov, U. Bexell, D. Thierry, J. Serak, Corrosion mechanism of model zinc–magnesium alloys in atmospheric conditions, Corros. Sci. 50 (2008) 2216–2231. J. Sullivan, S. Mehraban, J. Elvins, In situ monitoring of the microstructural corrosion mechanisms of zinc–magnesium–aluminium galvanising alloys using time lapse microscopy, in: Galvatech 2011, 8th International Conference on Zinc and Zinc Alloy Coated Steel Sheet, Genova, Italy, 2011. E. Diler, B. Lescop, S. Rioual, G. Nguyen Vien, D. Thierry, B. Rouvellou, Initial formation of corrosion products on pure zinc and MgZn2 examinated by XPS, Corros. Sci. 79 (2014) 83–88. T.E. Gier, X. Bu, S.-L. Wang, G.D. Stucky, Na2Zn3(CO3)43H2O, a microporous sodium zincocarbonate with a diamond-type tetrahedral-triangular topology, J. Am. Chem. Soc. 118 (1996) 3039–3040. S. Mathieu, C. Rapin, J. Steinmetz, P. Steinmetz, A corrosion study of the main constituent phases of AZ91 magnesium alloys, Corros. Sci. 45 (2003) 2741– 2755. T. Ishikawa, K. Matsumoto, A. Yasukawa, K. Kandori, T. Nakayama, T. Tsubota, Inﬂuence of metal ions on the formation of artiﬁcial zinc rusts, Corros. Sci. 46 (2003) 329–342. R. Hausbrand, M. Stratmann, M. Rohwerder, Delamination resistant zinc alloys: simple concept and results on the system zinc–magnesium, Steel Res. Int. 74 (2003) 453–458. T. Prosek, D. Thierry, C. Taxén, J. Maixner, Effect of cations on corrosion of zinc and carbon steel covered with chloride deposits under atmospheric conditions, Corros. Sci. 49 (2007) 2676–2693.
T. Prosek et al. / Corrosion Science 86 (2014) 231–238
 M. Jönsson, D. Persson, D. Thierry, Corrosion product formation during NaCl induced atmospheric corrosion of magnesium alloy AZ91D, Corros. Sci. 49 (2007) 1540–1558.  D. Thierry, T. Prosek, N. Le Bozec, E. Diller, Corrosion protection and corrosion mechanisms of continuous galvanised steel sheet with focus on new coating alloys, in: Proc. Galvatech ‘11, 8th International Conference on Zinc and Zinc Alloy Coated Steel Sheet, Genova, Italy, 2011.
 R. Hausbrand, M. Stratmann, M. Rohrwerder, The physical meaning of electrode potentials at metal surfaces and polymer/metal interfaces: consequences for delamination, J. Electrochem. Soc. 155 (2008) C369–C379.  E. Diler, S. Rioual, B. Lescop, D. Thierry, B. Rouvellou, Stability of ZnMgO oxide in a weak alkaline solution, Thin Solid Films 520 (2012) 2819–2823.