Corrosion Science 51 (2009) 1251–1262

Contents lists available at ScienceDirect

Corrosion Science
journal homepage: www.elsevier.com/locate/corsci

Understanding corrosion via corrosion product characterization: I. Case study of

the role of Mg alloying in Zn–Mg coating on steel
P. Volovitch a,*, C. Allely b, K. Ogle a
a
LPCS, CNRS UMR7045, Ecole Nationale Supérieure de Chimie de Paris, 11 rue Pierre et Marie Curie, 75231 Paris Cedex 05, France
b
ArcelorMittal Research SA, 57283 Maizières-lès-Metz, France

a r t i c l e i n f o a b s t r a c t

Article history: In the present work the corrosion inhibitive role of Mg in Zn–Mg coatings is considered for different
Received 27 November 2008 stages of corrosion. Corrosion product characterization was carried out using XRD, IRRAS, MEB–FEG–
Accepted 5 March 2009 EDS on technical Zn–Mg coatings after various exposure times in a standardized cyclic corrosion test.
Available online 19 March 2009
The results are compared with artificial corrosion products obtained by chemical and electrochemical
synthesis. The importance of the ageing and the role of the atmospheric CO2 on the nature and morphol-
Keywords: ogy of the corrosion products are discussed. The corrosion resistance of Zn–Mg alloy is correlated with
A. Metal coating
the stabilization of simonkolleite against its transformation into smithsonite, hydrozincite, and zincite
A. Zinc
A. Magnesium
during ageing cycles in presence of CO2. The stabilization appears to be due to the preferential formation
B. Corrosion product characterization of magnesium carbonates. Thermodynamic modeling and titrometric analysis demonstrate that Mg2+
C. Corrosion mechanism enhances simonkolleite during dry–wet cycling by (1) removing carbonate from the environment and
thereby limiting of the transformation of simonkolleite into zincite, smithsonite, and hydrozincite and
by (2) buffering the pH of the electrolyte around 10.2 due to the precipitation of Mg(OH)2 preventing
the dissolution of zinc based corrosion products into soluble hydroxide complexes.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction sion products formed on zinc panels after long-term exposure

(13–16 years) in various types of atmospheres [14] has shown that
Magnesium containing zinc alloy coatings on steel have been formation of basic zinc salts (simonkolleite and hydrozincite)
shown to have markedly superior corrosion resistance than con- improves significantly the long-term corrosion resistance.
ventional hot dip and electrogalvanized coatings [1–3] although The presence of magnesium in the alloy coating will probably
the mechanism has not been fully explained. affect the anticorrosion mechanism in each stage of corrosion.
The progression of corrosion on galvanized steel may be divided Under most conditions, magnesium is more electroactive than zinc
into several distinct periods [4,5]. At first, only the coating is ex- and therefore the galvanic properties will be changed. The
posed and it corrodes directly by exposure to the atmosphere. electrochemical properties of Zn–Mg coatings have been previously
The rate of corrosion in this stage is determined by the barrier investigated [15,16]. An enhanced self-healing effect for Zn–Mg and
properties of the resulting insoluble corrosion product films. At la- Zn–Mg–Al coatings was also previously reported [3,8,13,15–18].
ter stages, the underlying steel surface will be exposed in areas Nevertheless, short time atmospheric corrosion tests used in the
where the corrosion has penetrated the thickness of the coating automotive and building industries mainly test for the first period
and rust forms due to the atmospheric corrosion of steel. With re- of corrosion resistance and this is determined exclusively by the
spect to uncoated steel, however, the rate of steel corrosion is sig- intrinsic corrosion rate of the coating alloy. The breeching of the
nificantly reduced by the galvanic properties of the sacrificial coating is determined visually by the appearance of red rust and
coating and by the precipitation of inhibiting zinc based corrosion often corrosion results are presented as a percentage of the surface
products on the exposed steel cathode [11–13]. In general zinc covered by red rust determined by image analysis.
based coatings corrode on the order of 100 times slower than cold It is therefore of great interest to produce metal coatings that
rolled steel due to the inhibiting or barrier effect of zinc-based cor- favor inhibiting corrosion products. Such corrosion products
rosion products [4–14]. At much later stages the coating will be should be electrically insulating and compact. For zinc, the insulat-
completely removed and yet even in these stages the corrosion rate ing nature of the corrosion products depends upon the chemical
may be reduced by residual corrosion products. The study of corro- composition since most products are insulating except for ZnO
which is an n-type semiconductor. The compactness depends upon
* Corresponding author. Tel.: +33 1 44 27 67 36; fax: +33 1 46 34 07 53. specific kinetic features of its formation and is very sensitive to
E-mail address: [email protected] (P. Volovitch). environment and history. For example, zinc hydroxide is formed

0010-938X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2009.03.005
1252 P. Volovitch et al. / Corrosion Science 51 (2009) 1251–1262

at room temperature by precipitation of Zn2+ by OH despite the In the present article, we combine the two approaches (charac-
fact that ZnO is the thermodynamically favored specie [19,20]. At terization of corrosion products on technical samples of Zn–Mg
34 °C and above, a transition to ZnO is observed after a delay time coatings and the synthesis of artificial corrosion products) with
attributed to a slow dehydration reaction leading to the formation thermodynamic modeling in order to clarify the protective mecha-
of zincite (ZnO) seeds, permitting the nucleation process [21]. At nism of zinc–magnesium alloy coatings. In the next article [part 2],
34 °C, where nucleation is slow, the films obtained are porous we will apply the proposed mechanism to understand the influ-
and seed orientation depends on size since a shorter delay time fa- ence of corrosion testing conditions on the product formation on
vors the texturation of the film with the c-axis perpendicular to the the example of VDA cycling tests for open surface and in confined
surface. In presence of Cl and carbon dioxide, zinc corrosion prod- zone and extend the study to include Zn–Al–Mg coatings.
ucts can also contain [5] Zn5Cl2(OH)8H2O (simonkolleite, zinc
hydroxychloride ZHC), Zn4CO3(OH)6 (hydrozincite, HZ), ZnCO3 2. Experimental
(smithsonite, S), etc. The basic zinc salts can also be transformed
into zinc oxide by thermal decomposition. It was shown [21–25] 2.1. Materials
that the nature of the precursor and the precipitation time changes
the precipitation pH and the growth mechanism, involving either The steel samples were prepared and supplied by ArcelorMittal.
ramified branches and stratified tips, or hexagonal rods with flat The reference samples without Mg used in this work was electro-
ends (zinc hydroxide precursor) or even porous spherical aggre- galvanized (EZ) sheet steel with total sheet thickness between 0.7
gates (HZ precursor). and 0.8 mm and the coating thickness 7.5 lm with a composition
The promotion of a specific corrosion product can be achieved of nearly pure zinc typical of use in the automotive industry. The
by the shift of thermodynamic equilibria. This may be induced Zn–Mg coatings with the final composition of the coating Mg
by varying the alloy composition and/or the electrolyte composi- 16.0 wt%, Zn 84.0 wt% composed of intermetallic phase Zn2Mg
tion. The formation of artificial corrosion products in the pres- were 5 lm thick and were obtained by the Physical Vapor Deposi-
ence of cations such as Na+, Ca2+, Mg2+, Al3+, Fe3+, and Fe2+ tion (PVD) process on the substrate similar to that used for the
was previously studied by two complementary methods. Prefer- electrogalvanization step. The vacuum evaporation process was
ential simonkolleite formation in the presence of Mg2+ was chosen because of its higher deposition rate, which permits to de-
observed for zinc coatings on carbon steel [26]. In contrast, posit thick layers.
Al(III) and Fe(II) were found to promote the formation of basic
zinc chloride in detriment of zinc oxide, whereas Fe(III) and 2.2. Exposure conditions
Mg (II) did not promote simonkolleite formation under identical
conditions (pH adjusted at 9.2) in experiments involving artificial The technical samples were exposed in the ECC1 cyclic corro-
precipitation from aerated zinc salt solutions containing different sion test referenced as D172028 in CC1000 Ascott chamber (see
cations [27,28]. This apparent disaccord illustrates the complex [33] for a detailed description of the chamber and one cycle condi-
mechanism of magnesium action. tions). This test uses in a salt spray test (SST) phase a 1% NaCl spray
Several recent attempts have been made to clarify the effect of Mg solution with pH adjusted to 4. The samples were exposed in a
in Zn–Mg alloy coatings by investigating the corrosion products 120 lm confined zone configuration prepared with a metal/glass
obtained on ‘‘model” coatings. Prosek et al. [29] proposed a protec- assembly as described in Fig. 1.
tive magnesium containing layer with reduced electron transfer In an initial study, the samples were corroded for a time, tmax,
kinetics because of the low band gap of magnesium oxides. This until red rust was observed on approximately 5% of the exposed
hypothesis correlates well with Kelvin probe measurements and surface as determined by visual inspection. The tmax was 27 cycles
the element distribution into the native oxide layer enriched in for electrogalvanized steel and 70 cycles for Zn–Mg coatings. Addi-
Mg. Although the nature of the Mg containing film was not clearly tional samples were prepared for a number of cycles roughly cor-
defined, this model seems interesting for an understanding of the responding to 0.4 tmax and 0.1 tmax.
initiation of corrosion, but does not seem to explain either the better
corrosion resistance once the protective layer is locally dissolved nor 2.3. Characterization methods
the observed resistance at near neutral pH, corresponding to the
washing conditions in NaCl solutions at which the MgO film should After visual inspection and optical micrograph observations, the
be rapidly removed. Moreover, the morphology of Mg–hydroxides corroded samples were cut into 2 cm  2 cm squares and depos-
and/or carbonates is nearly always porous and does not seem to offer ited onto a carbon ribbon. Some samples were metallized with car-
a good corrosion protection [30–32]. The exception is basic hydroxy- bon using thermal vaporization in a Cressington 208 apparatus.
carbonate–hydromagnesite, Mg5(CO3)4(OH)24H2O which can be The majority of work was performed using a scanning electron
formed during ageing at certain experimental conditions [31]. microscopy (FEG-SEM) and energy dispersive spectroscopy (EDS)
The model of Hosking et al. [3] predicts the beneficial action of using LEICA S 440 microscope with W filament coupled with PGT
Mg by the ability of magnesium corrosion products to neutralize Spirit type EDS system with Si(Li) diode. Additional work was per-
the hydroxide ions at cathodes creating conditions for forming pro- formed using other SEM instruments.
tective simonkolleite layer over the corroding surface. This model Polarization modulation infrared reflection–absorption spec-
accounts for the enrichment of simonkolleite in the corrosion troscopy (PM IRRAS) was performed directly on the corroded sam-
products observed on technical Zn–Mg coated steel samples at ad- ples using a Nicolet spectrometer with MCTA detector. For some
vanced stages of corrosion. As simonkolleite is considered to have a samples corrosion product was step by step partially removed
much better corrosion resistance this mechanism does explain the from the surface by a surgeon’s scalpel. The results were compared
effect. However, the magnesium alloying also promotes the forma- with reference spectra taken from the literature [26,29,31,34–36]
tion of hydrozincite which is considered to be less protective than and from the internet mineral database Mindat [37].
simonkolleite. Moreover, it is in contradiction with some experi- For X-ray diffraction studies, the corrosion products were re-
mental observations at early stages of corrosion in which the for- moved from the surface using a surgeon’s scalpel, collected and
mation of simonkolleite was retarded in presence of Mg (Fig. 4 in triturated. The obtained powders were deposited on the glass sup-
[17]) and with previously cited data from artificial corrosion prod- port with a thin VaselineTM layer for fixation and analyzed by X-ray
uct synthesis in presence of different ions [27]. diffraction using the Cu(Ka) radiation in a Siemens D5000
 P. Volovitch et al. / Corrosion Science 51 (2009) 1251–1262 1253

metal sheet 20°

Upper part
= area to collect the liquid

mm
160
80 to
Overlap area

40 mm
(gap width 120 µm)

60 mm
120
µm
Adhesive
Studied area Glass

90 to 100 mm Sample in the corrosion chamber

Fig. 1. Schematic representation of glass–metal assembly.

diffractometer. The XRD were collected with the angular resolution Zn(NO3)2 in 1 M KNO3 and either 0 M, 0.1 M or 0.5 M Mg(NO3)2.
0.2°. The phase identification has been carried out by referencing The copper/gold electrode was obtained by the vapor deposition
the software PCPDFWIN version 2.02 containing the JCPDS (ICDD) of a 25 nm gold film on a mechanically polished copper discs using
database files. a Balzers Union SCD040 deposition apparatus. In general, deaera-
tion was performed using a continuous flow of nitrogen to prevent
2.4. Artificial corrosion product synthesis CO2 uptake however, the effect of low amounts of carbonate was
investigated by stopping the nitrogen flow for 10 min just before
Two methods of artificial corrosion product synthesis were the experiment. Under all conditions, the resulting oxide films
used: precipitation and electrodeposition. were visually uniform on the electrode surface. XRD showed only
the presence of crystalline form of ZnO. The Mg:Zn atomic ratio
2.4.1. Precipitation in the electrodeposited ZnO films determined by ICP-OES (after
Artificial corrosion products were precipitated from solution by dissolution in 1 M HCl) was less than 1:100 for all samples.
the addition of NaOH to Zn2+ and/or Mg2+ containing electrolytes. A
Metrohm automatic burette was used for the stepwise addition of 3. Results
0.1 M NaOH and the pH was continuously monitored with a com-
bined Ag/AgCl glass electrode (Metrohm) calibrated with dispos- 3.1. Visual comparison
able standard buffer solutions (Centipur, Merck). The complete
pH vs. volume of NaOH was recorded using home-made software. Fig. 2 illustrates the effect of Mg on the visual aspect of the coat-
The tested solution (see Table 1 for compositions) was under con- ing after different number of cycles of the ECC1 test. Image analysis
tinuous flow of nitrogen to prevent CO2 uptake. The high purity shows the surface is covered with approximately 5% of red rust for-
(99.999%) Mg(NO3)26H2O, Zn(NO3)26H2O, and NaCl compounds mation for pure zinc coatings (Fig. 2a), but no red rust is observed
(Merck) were used for solution preparation. The solution tempera- at this stage for Zn–Mg coatings (Fig. 2b). At 7 ECC1 cycles (total
ture during the experiment was stable (25 ± 2 °C). The added vol- test time equal to 40% of the time necessary for of the appearance
umes of base are indicated in the table with a 360 s interval of 5% red rust on the surface of electrogalvanized steel) Zn (Fig. 2c)
between additions to give a total titration time of 48 h. and Zn–Mg (Fig. 2d) samples, are covered by the white and gray
corrosion product of zinc. It can be seen (Fig. 2e) that on Zn–Mg
2.4.2. Electrodeposition coatings the red rust is not visible even after 70 ECC1 cycles.
Synthetic zinc oxide was obtained by cathodic electrodeposition These results demonstrate a clear amelioration in the corrosion
on the copper/gold electrode at an applied potential of 0.9 V vs. resistance of 5.0 lm Zn–Mg coating vs. a 7.5 lm Zn coatings. This
Ag/AgCl for 30 min at 70 °C in stagnant solution using a 0.001 M amelioration reflects the first stage of corrosion that is the intrinsic
corrosion rate of the coating, as the experiments have not been
conducted for sufficient time for the galvanic coupling stage to
Table 1 have been reached. As Mg is known to be electrochemically more
Experimental conditions used for the titration experiments.
active than Zn, it is reasonable to attribute this increased corrosion
N Solution for titration Solvent Initial Volume of 0.1 M NaOH resistance to the nature of the corrosion products. Indeed, during
volume (ml) at each addition (ml) sample preparation, the light gray layer that formed close to the
6 Mg(NO3)2 (0.005 M) NaCl (5%) aq 50 0.05 metal surface for the Zn–Mg coatings was found to be more adher-
7 Zn(NO3)2 (0.010 M) NaCl (5%) aq 50 0.05 ent and more compact than the white corrosion products formed
9 Zn(NO3)2 (0.010 M) NaCl (5%) aq 22.5 0.02
on pure zinc coatings, consistent with the improved barrier prop-
+ Mg(NO3)2 (0.05 M)
10 Zn(NO3)2 (0.010 M) NaCl (5%) aq 50 0.05 erties of this layer.
+ Mg(NO3)2 (0.05 M)
11 Zn(NO3)2 (0.010 M) NaCl (5%) aq 50 0.15 3.2. X-ray diffraction and IR spectroscopy
+ Mg(NO3)2 (0.05 M)
12 Zn(NO3)2 (0.010 M) Pure water 50 0.05 X-ray diffraction and IR spectroscopy have been extensively
+ Mg(NO3)2 (0.05 M)
used to characterize zinc-based corrosion products because of their
1254 P. Volovitch et al. / Corrosion Science 51 (2009) 1251–1262

Fig. 2. General view of the corroded surface at different number of ECC1 cycles, CCD camera. (a) Reference zinc coating on steel, 7 cycles; (b) Zn2Mg coating, 7 cycles; (c)
reference zinc coating on steel, 27 cycles; (d) Zn2Mg coating, 27 cycles; (e) Zn2Mg coating, 70 cycles.

complementarities for identification crystalline and amorphous 3.3. Corrosion product morphology
products. Fig. 3 shows X-ray diffraction patterns for all samples.
The crystalline phase of ZnO is present in the corrosion products The previous results suggest that the major species of the corro-
of all samples. Simonkolleite was detected in all Zn–Mg samples sion reaction are simonkolleite, zincite, and various carbonates. As
at all stages of corrosion and in Zn samples only at the early stage, presented in Section 1, these products may serve as barrier layers
in this case at 7 cycles. No other crystalline products were deter- to aggressive species in an environment, however, the efficiency of
mined in any significant quantity. the barrier properties will depend upon the morphology. In this
Fig. 4 gives the IR spectra of the different samples. The convo- section, we investigate the morphology of the corrosion products
lution of peaks characteristic for different Mg and Zn compounds, and compare them with synthetic ZnO and carbonate samples.
as well as the presence of large peaks near 1460 cm1 (carbonates
ion in Na2CO3) [35,38] and near 1620 cm1 (water bending) [38],
make a quantitative analysis extremely difficult. In particular, ZHC reference ZnO reference
SEM–EDS analysis showed large areas of the surface covered
exclusively by sodium carbonate on the Zn–Mg coated surface NaCl
1500
after 70 cycles of ECC1 as indicated by a ratio of Na:O = 1:1.53, EZ coat, after 27 cycles
with the presence of a large C peak. It is in good accordance with
the large peak in Fig. 4 on the corresponding curve. The attribu- EZ coat, after 7 cycles
tion of different peaks observed for EZ and Zn–Mg coatings are
1000
presented in Fig. 4.
I,cps

These results demonstrate a clear difference between the Zn– Zn2Mg coat, after 7 cycles
Mg and Zn coatings. Simonkolleite is formed as the initial product
on both Zn and Zn–Mg coatings during the early stages of their 500 Zn2Mg coat, after 27 cycles
exposure, however, the crystalline simonkolleite disappears after
Zn2Mg coat, after 70 cycles
continued exposure in absence of Mg. Interpretation of the IR spec-
troscopy is limited due to interference, it was clear that amorphous
carbonate mixture was a major product of the reaction especially 0
at longer times. However, some exclusion could be made. The pres- 10 15 20 25 30 35 40 45 50 55 60

ence of smithsonite (ZnCO3) can be considered only in EZ coatings 2θ,° (vs. λ - Cu (Kα))
(narrow peaks at 840, 1090, and 1435 cm1 indicated by narrow Fig. 3. Typical X-ray diffraction spectra of corrosion products of Mg-free (EZ) and
bars in Fig. 4), but this compound is not detected in Zn–Mg Zn–Mg (Zn2Mg) coatings in ECC1 test. The reference peak positions for simonkolle-
samples. ite (zinc hydroxichloride, ZHC) and zincite (ZnO) are also shown.
 P. Volovitch et al. / Corrosion Science 51 (2009) 1251–1262 1255

EZ, 7 cycles Zn2Mg, 7 cycles 3.3.1. Corroded EZ Coatings

1.2
Fig. 5 shows the typical microstructural elements observed on
1.1
the corroded surfaces of the Zn coating after 27 cycles. Different
1.0
types of structure are observed. A needle or hexagonal rod mor-
0.9 phology (Fig. 5a) with a near 1:1 Zn:O ratio is typical for ZnO. A
0.8 spherical aggregate morphology of what appears to be an amor-
Transmittance,a.u.

0.7 phous phase with an atomic ratio Zn:O close to 1:3 which is con-
0.6 Zn2Mg, 70 cycles sistent with zinc carbonate (smithsonite). The morphology and
0.5 higher than 2 ratio O:Zn for Fig. 5d would suggest hydrozincite
Zn2Mg, 27 cycles
0.4 pending further confirmation.
0.3 EZ, 27 cycles
0.2 3.3.2. Electrodeposited ZnO
0.1 It is interesting to compare these images with those of electro-
0.0
deposited ZnO in Fig. 6. Fig. 6a and b shows the SEM micrographs
2000 1800 1600 1400 1200 1000 800 of ZnO deposited in nitrogen aerated solutions in absence (a) and
frequency, cm
-1 in presence (b) of 0.5 M Mg2+ in the electrolyte. Although Mg2+ is
only slightly incorporated into the ZnO film, there is a clear effect
on the size and shape of ZnO particles. This effect is in coherence
HZ HM NH M ZHC SM
with previously reported capacity of the Mg2+ cation to promote
0.0
certain crystalline forms of calcareous deposits on freely corroding
-0.1
steel [39]. It is interesting to note that ZnO observed on Zn–Mg
-0.2
coatings after 70 cycles of ECC1 test shows the similar change in
-0.3
morphology (Fig. 6c).
Transmittance,a.u.

-0.4
Fig. 7 illustrates the effect of CO2 traces on the deposit in the
-0.5 presence of 0.1 M Mg2+ in the electrolyte. By the comparison of
-0.6 the top view (Fig. 7a) and the angular view (Fig. 7b), a new phase
-0.7 in the form of thin leafs that are somewhat transparent to the elec-
-0.8 tron beam is observed. We attribute this secondary phase to the
-0.9 formation of zinc carbonate. The identification was confirmed by
-1.0 the presence of small C peaks in the EDS spectra, absent for the
-1.1 neighboring oxide.
-1.2
2000 1800 1600 1400 1200 1000 800 3.3.3. Zn–Mg coatings
-1
frequency, cm Fig. 8 shows typical corrosion product morphologies observed
on Zn–Mg coatings after 70 cycles. The platelet-like phase of
Fig. 4. Typical IR spectra of corrosion products of Mg-free (EZ) and Zn–Mg coatings Fig. 8a gave a Zn:Mg:O:Cl ratio of near 5:0:8:2 consistent with pure
after ECC1 test (a) and the peaks attribution (b). The positions of peaks are
simonkolleite. The platelet-like structure and monolith blocks in
attributed for the following compounds: ZHC, simonkolleite; HZ, hydrozincite; S,
smithsonite; M, magnetite; NH, neishonite; HM, hydromagnesite. Fig. 8b and c give a Zn:Mg:O:Cl near 4:1:8:2 suggesting a Mg mod-
ified simonkolleite as previously reported by Thierry et al. [26]

Fig. 5. Some typical microstructures (SEM–FEG) observed in EZ samples after 7 cycles of ECC1 test. The crystalline phase is quantified by EDS as ZnO, the supposed
amorphous phases indicated on the images should be carbonates because of presence carbon peak in EDS spectra and consequently could not be precized by EDS. The
proposition of phases is made by the Zn:O ratio values and by the morphological analysis of mineral database for zinc minerals.
1256 P. Volovitch et al. / Corrosion Science 51 (2009) 1251–1262

Fig. 6. Typical morphology (SEM–FEG) of ZnO electrodeposited from the solution in absence (a) and in presence (b) Mg2+ in comparison with the morphology of ZnO formed
by the corrosion of Zn–Mg coating after 70 cycles of ECC1 test (c).

Fig. 7. Zinc oxide electrodeposited in presence of minor quantities of CO2, the top view (a) and the angular view (b). A new phase in the form of thin leafs is transparent to the
electron beam and contains an important peak of carbon in the EDS spectra.

Some additional arguments in the favor of possibility of the the crystalline zinc-rich phase closer to the metal surface and
co-precipitation of Zn–Mg product are discussed in Section 4. Some monolith or amorphous phase containing magnesium with high-
Mg-rich zones were detected, but the observation was hampered er oxygen fraction on the outside. The Mg content of this phase
because of their modification under the electron beam (Fig. 8f). appears to be lower on the outer surface of this phase than in
Fig. 9 shows SEM micrographs of simonkolleite in a variety of the middle. Fig. 10f gives a higher resolution micrograph of the
forms observed on these samples showing a very dense and com- external layer, the arrow direction corresponding to the arrow
pact structure. Simonkolleite crystals are often observed as hexag- direction in Fig. 10e.
onal plates of different width (Fig. 9a) or as monolith blocks The SEM results demonstrate that the corrosion products on the
(Fig. 9b). The high resolution image of Fig. 9c indicates that the Zn–Mg coating are more compact. This is consistent with an im-
monolith block is composed of a 3–5 nm thick layer. The layered proved diffusion barrier properties and therefore better corrosion
microstructure of the monolith is consistent with the formation resistance. The two layer structure might suggest that ZnO precip-
of a layered double hydroxide (LDH) species [40] which are known itation occurs during the early wet part of the cycle while the
to be stabilized in presence of some cations including Al3+ and external layer is precipitated under final drying. Two factors are
Mg2+. in the favor of this hypothesis: (1) the similarity of oxide morphol-
Fig. 10 compares the microstructure of the corrosion products ogy with that of ZnO electrodeposited from the solution containing
on Zn–Mg obtained at 7 cycles (0.1 tmax, Fig. 10a–c) and 70 cycles an excess of Mg2+ (Figs. 5 and 6), and (2) the decrease of Mg2+ con-
(tmax, Fig. 10d–f). The apparent porosity of this layer would suggest tent with the distance from the surface. The later may be due to the
limited barrier properties. By the comparison with the magnesium decrease in Mg2+ concentration associated with the precipitation
mineral databases [37], the morphology observed in Fig. 10b gives reaction.
the impression of being amorphous hydromagnesite. The structure
showed in Fig. 10c is similar to porous magnesium hydroxide 3.4. Titration experiments
formed on the cathodic areas of high-Mg alloys [41].
After 70 cycles a more compact structure is observed The results of the previous sections have demonstrated that an
(Fig. 10d–f). The EDS spectra of corresponding zones reveals enhanced stability of the simonkolieite on the Zn–Mg surface is
 P. Volovitch et al. / Corrosion Science 51 (2009) 1251–1262 1257

Fig. 8. Morphological structures presented in Mg containing samples, the powder obtained after scratch from the metal surface. (a–c) Platelet-like morphology of
simonkolleite (a) and probably Mg-modified simonkolleite (b and c), atomic ratios obtained by EDS are Zn:Mg:O:Cl  19:5:53:11 for (b) and Zn:Mg:O:Cl  19:8:39:17 for (c);
(d) probable carbonates containing product (Zn:Mg:O:C  10:10:56:5);(e and f) Mg-reach product modified under the beam (Zn:Mg:O  1:30:64).

Fig. 9. Dense and compact morphology of simonkolleite presented on the corroded surface, direct observation. Hexagonal crystals (a) and monolith blocks (a–c). The
monolith high resolution image (c) shows the microstructure which was previously described for layered double hydroxides (LDH) [40] which are known to be stabilized in
presence of some cations.

probably responsible for the better corrosion resistance. In this sec- Fig. 11 shows titration curves (pH vs. added OH) for Zn2+ and/
tion, we undertake some classical titration experiments to isolate or Mg2+ containing solutions. Note that the x-axis is presented as
the role of the Mg2+ ion in stabilizing simonkolleite. the ratio of the hydroxide added to the metal ion initially present.
1258 P. Volovitch et al. / Corrosion Science 51 (2009) 1251–1262

Fig. 10. Some morphologies observed in Mg-containing corrosion product after ECC1 tests on Zn–Mg coatings: (a–c) after 7 cycles of ECC1; (d–f) after 70 ECC1 cycles. The
flash direction in (e) indicates the decrease of Mg content in the phase on the outer surface than in the middle.

In the case of curves 1 and 4, however, two metal ions are present
and only the ratio is made with only the Zn2+.
Curve 1 shows a mixture of Zn and Mg in the absence of Cl. The
3
12 4
first endpoint occurs at a stoichiometry of 2:1 OH/Zn2+ consistent
with the formation of Zn(OH)2 as the predominant reaction pro- 11
duction. The reaction is
10 1
Zn2þ þ 2OH ¼ ZnðOHÞ2 ð1Þ
pH

9
A second plateau is observed at pH of 10.5 due to the formation 2
of Mg(OH)2. This attribution is confirmed by curve 4 obtained for a 8
Mg2+ solution alone. The reaction is
7
Mg2þ þ 2OH ¼ MgðOHÞ2 ð2Þ
0 1 2 3 4
Curve 3 gives the titration curve of Zn2+ in 5% NaCl. The solubility - 2+
plateau is approximately the same as in deionized water, however,
[OH ]/[Me ]
the precipitation stoichiometry is different and occurs at 1.6–1.7, Fig. 11. Typical results of titration experiments, titrant – NaOH 0.1 M, added
very close to the theoretical value of simonkoeillite. volume 0.05 ml. Solutions: 1, 0.010 M Zn2+ + 0.005 M Mg2+ solution w/o NaCl; 2,
 0.010 M Zn2+ + 0.005 M Mg2+ solution in 5% NaCl; 3, 0.010 M Zn2+ solution in 5%
5Zn2þ þ 8OH þ 2Cl ¼ Zn5 ½ðOHÞ8 Cl2  OH : Zn ¼ 8 : 5 ¼ 1:6 ð3Þ NaCl; 4, 0.005 M Mg2+ solutions in 5% NaCl.
 P. Volovitch et al. / Corrosion Science 51 (2009) 1251–1262 1259

Curve 2 shows mixed Zn2+ and Mg2+ in a 5% NaCl solution. Again the at the first cycles of dry–wet cycling, when carbonate concentra-
results are consistent with the formation of simonkoelite for Zn2+ tion is very low.
and Mg(OH)2 for Mg2+. The difference between curves 2 and 3 is We propose that the instability of the simonkolleite is due to an
also interesting. Two supplemental inflection points are observed increase in the carbonate fraction during dry–wet cycling. With
in presence of Mg2+ ions. This might be attributed to co-precipita- growing carbonate-ion concentration simonkolleite can transform
tion of Zn and Mg product and the kinetic sluggishness related to into hydrozincite or smithsonite by the reactions (4) and (5)
the presence of magnesium. It should be noted that when the rate
Zn5 ðOHÞ8 Cl2 þ 5HCO3 ¼ 5ZnCO3 þ 5H2 O þ 3OH
of titration was changed (conditions 9 and 11 in Table 1), the deflec- 
tion in supplemental inflection points of curve 2 was more or less þ 2Cl ðacid pHÞ ð4Þ
pronounced demonstrating the kinetic origin of these effects. In-
deed, it can be seen from curve 4, that the kinetic effect is important Zn5 ðOHÞ8 Cl2 þ 2HCO3 ¼ Zn5 ðCO3 Þ2 ðOHÞ6 þ 2H2 O
for Mg2+ base titration. 
þ 2Cl ðneutral and basic pHÞ ð5Þ
The results of this section demonstrate that simonkolleite is the
product formed spontaneously when Zn2+ reacts with OH in a 5% The increase of zinc oxide observed experimentally on Zn coating
NaCl solution, whether or not Mg2+ is present. In fact, in the 5% samples probably occurs by the transformation of simonkolleite
NaCl solution, Mg2+ does lead to a very slight modification of the into hydrozincite under carbonate action (Eq. (5)) followed by the
titration curve which may be due to the formation of a Mg2+ mod- transformation of hydrozincite into zinc oxide by common action
ified form of simonkolleite. of temperature and/or increase of pH in the wet stage of corrosion
(Eqs. (6) and (7)).
4. Discussion Zn5 ðCO3 Þ2 ðOHÞ6 ¼ 5ZnO þ 2CO2 þ 3H2 O ð6Þ

The results of this work allow us to propose the simonkolleite

Zn5 ðCO3 Þ2 ðOHÞ6 þ 2OH ¼ 5ZnO þ 2HCO2
3 þ 3H2 O ð7Þ
species as the phase primarily responsible for the corrosion resis-
tance for the coatings and conditions investigated in this work. In addition to these reactions, the solid-state transformation of
The platelet-like morphology of simonkolleite crystals seems to simonkolleite into ZnO can probably be also considered at very high
be the most compact between the flower-like ZnO and amorphous pH and temperature. Such a mechanism is in agreement with the
porous zincite. Moreover, the benefit of basic salts of zinc for cor- XRD results presented in this work and with previous work in the
rosion protection and the absence of protection by zincite were al- literature [17] – simonkolleite is detected after the first dry–wet cy-
ready noted in the literature (see for instance [3–5,14]). In cles and its fraction decreases with cycling for Zn coatings. As a
combination with the knowledge that corrosion products of unal- function of pH and carbonate content, the transition between car-
loyed Mg do not provide significant corrosion protection [30–32] bonate and hydroxycarbonate of zinc is reversible:
it seems, that the better corrosion resistance of Zn–Mg coatings
5ZnCO3 þ 6OH ¼ Zn5 ðCO3 Þ2 ðOHÞ6 þ 3CO2
3 ð8Þ
is related to stabilization of simonkolleite and hydrozincite trans-
formation into zincite and smithsonite during cycling. We attribute the protective nature of the Zn–Mg corrosion products
Zn–Mg coatings show an enhanced stability of simonkolleite not to a protective Mg2+ product, but to an indirect influence of
after longer cycling periods. For Zn coatings, the main corrosion Mg2+. In the presence of carbonate, the formation of magnesite
products are ZnO (crystalline) and smithsonite or hydrozincite (MgCO3) is actually preferred over the formation of zinc carbonates.
(amorphous). The fraction of ZnO increases at advanced stages of Thus Mg2+ will eliminate the carbonate from the electrolyte render-
corrosion probably through the decomposition of hydrozincite ing reactions (4) and (5) impossible.
(see reactions (7) and (8) below in Section 4) and crystalline simo- This result is clearly illustrated by the equilibrium calculations
nkolleite is present only at the early stages of corrosion. in Fig. 12. The calculation is made for an aqueous solutions of Zn2+,
In contrast, corrosion products of Zn–Mg coatings do not con- Mg2+, Cl, and CO32 at 25 °C using the Hydra-MedusaÓ software
tain smithsonite at all. The crystalline product consists mainly of [42]. This software includes a dedicated database of equilibrium
the crystalline form of simonkolleite regardless of the number of constants included most known zinc corrosion products. The re-
cycles. The ZnO fraction decreases with corrosion advance. Hydro- sults clearly demonstrate that Mg2+ ions will preferentially precip-
zincite (amorphous) as well as amorphous magnesium carbonates itate with carbonate (Fig. 12a), thereby lowering the carbonate ion
and hydrocarbonates were also found. The presence of Mg(OH)2 concentration and stabilizing the protective simonkolleite phase
could not be verified for technical samples as the spectral region (Fig. 12b and c). Further, the critical concentration of carbonate-
above 3000 cm1 in the IR spectra was too noisy to evaluate the ion formation of Zn carbonates and hydrocarbonates increases
presence of the most significant peak of brucite. significantly in presence of Mg because of the formation of more
The comprehension of Mg action requires the decomposition of stable Mg carbonates (Fig. 13). It can be also noted that in the
corrosion mechanism by different stages. Initially the metal dis- absence of Mg2+, zinc carbonate is favorable even in the range of
solves to produce dissolved cations (in our case Zn2+ and Mg2+) neutral and low-basic pH (up to pH 8, Fig. 14a) while in presence
which in turn precipitate once the critical concentration is of Mg this product can be formed only in the acid region (pH < 7,
achieved. The solubility is decreased by the increase of pH associ- Fig. 14b). That means that the basic zinc salts should stay stable
ated with the cathodic reduction of oxygen or hydrogen ions. The for more dry–wet cycles for Zn–Mg coatings than for Zn coatings.
chemical composition and the critical concentration for precipita- This indirect protective action of Mg is important for long-time
tion depend on the chemistry and the concentration of different dry–wet corrosion resistance when the important quantity of car-
ions in liquid phase which varies from one cycle to another be- bonate-ions can be accumulated.
cause of the dissolution–precipitation process as well as ageing Of course, our results do not completely rule out a direct protec-
of product during dry–wet cycles. tion mechanism as proposed by Hosking et al. [3] due for example
The titration results of Section 3.4 demonstrate that simo- to the formation of Mg(OH)2 at very high pH which has a signifi-
nkolleite is immediately formed in NaCl solutions for both Mg-con- cantly improved alkaline resistance with respect to Zn.
taining and Mg-free solutions. This conclusion differs with the However, as mentioned before, Mg(OH)2 is probably not a sig-
reaction sequence proposed in [3] to describe the initial stages of nificant barrier to oxygen diffusion due to its porous microstruc-
Mg action. These results are significant only for the precipitation ture. In fact, other experimental work in our laboratory has
1260 P. Volovitch et al. / Corrosion Science 51 (2009) 1251–1262

[Z n 2 + ] T O T = 5 0 0 .0 0 m M [C l − ] T O T = 1 .0 0 M [ Z n 2 + ] T O T = 5 0 0 .0 0 m M
[M g 2 + ] T O T = 5 0 0 .0 0 m M [C O 3 2 − ] T O T = 4 0 0 .0 0 m M [ C l −] T O T = 1 .0 0 M [N a+ ]T O T = 1 .00 M
pH = 8.00 [M g 2+]T O T = 0 .0 0
M g2+
1.0 Z n 5(O H )6(C O 3)2(c)
1.0 Z n 5(O H )8C l2(c) Z n C O 3:H 2O (c )

M g C O 3 (c )
0.8
0.8

0.6
Fraction

0.6

Fraction
M g (O H ) 2 (c )

0.4 0.4

0.2 0.2

0.0 0.0
2 4 6 8 10 12 0.0 0.2 0.4 0.6 0.8 1.0

a) pH b) [ C O 3 2 −] T O T M

[Z n 2 + ] T O T = 5 0 0 .0 0 m M
[C l − ] T O T = 1 .0 0 M [N a + ]T O T = 1 .0 0 M
pH = 8.00 [ M g 2 + ] T O T = 5 0 0 .0 0 m M
Z n 5 ( O H ) 8 C l 2 (c ) Z n 5 (O H ) 6 (C O 3 ) 2 (c )
1.0 Z n C O 3 : H 2 O (c )

0.8

0.6
Fraction

0.4

0.2

0.0
0.0 0.2 0.4 0.6 0.8 1.0

c) [ C O 3 2 −] T O T M

Fig. 12. Example of evolution of the precipitation product composition, simulation: (a) magnesium product as function of pH; (b) zinc product in function of total carbonate
concentration in absence of Mg2+; (c) zinc product in function of total carbonate concentration in presence of Mg2+ in the same concentration as Zn2+.

[C O 3 2 − ] T O T = 1 .0 0 M species might be to limit the pH increase in the cathodic area to

[C l − ] T O T = 1 .0 0 M [N a +]T O T = 1 .0 0 M around 10.2 due to the Mg(OH)2 precipitation. The lower pH would
pH = 8.00 [Z n 2 +]T O T = 5 0 0 .0 0 m M
prevent the dissolution of already formed Zn-containing product
1.0 Z n C O 3 : H 2 O (c )
Z n 5 ( O H ) 6 (C O 3 ) 2 (c )
through formation of soluble zincate – complexes [Zn(OH)4]2
Z n 5 ( O H )8 C l 2 ( c )
again enhancing the stability of the zinc based corrosion products.
In summary, the progression of corrosion on Zn and Zn–Mg
0.8
coatings may be described as follows:

0.6
Fraction

1. At first the coatings experience an identical environment as

the local pH is not very elevated and the most probable precipita-
0.4 tion products should be simonkolleite and in case of presence of
Mg a slightly soluble magnesium carbonate.
2. Some quantity of zinc oxide (or magnesium doped zinc oxide)
0.2
can be immediately formed on the cathodic areas due to local pH
increase.
0.0
0.0 0.2 0.4 0.6 0.8 1.0
3. In absence of magnesium the increase in total carbonate-
[M g 2 + ] TOT M ion during cycling provokes the transformation of simonkolleite
into hydrozincite and following transformation into smithsonite
Fig. 13. Example of evolution of the zinc precipitation product composition in (for low pH) or zincite (for high pH). The increased ZnO fraction
function of total Mg2+ concentration, simulation. and the presence of smithsonite are experimentally observed.
Smithsonite and zincite are known to be ineffective for corro-
demonstrated that the cathodic induced precipitation of Mg(OH)2 sion protection because of their high solubility or porous
on steel in 5% Cl-solutions, in the absence of Zn2+, not only does morphology.
not significantly decrease the cathodic current and in fact seems 4. In presence of Mg2+ the effect of increasing carbonate-ion
to increase it slightly [43]. Nevertheless, a possible effect of this concentration is balanced by formation of magnesium carbonates
 P. Volovitch et al. / Corrosion Science 51 (2009) 1251–1262 1261

[Z n2+ ]T O T = 5 0 0.00 m M [ Z n 2 + ] T O T = 5 0 0 .0 0 m M
[ C O 3 2 − ] T O T = 5 0 0 .0 0 m M [N a+ ]T O T = 1.00 M [ C O 3 2 − ] T O T = 5 0 0 .0 0 m M [N a+]T O T = 1.00 M
[ C l −] T O T = 1.00 M [M g2+]TO T = 0 .0 0 [ C l −] T O T = 1 .0 0 M [ M g 2 + ] T O T = 5 0 0 .0 0 m M
Z n C O 3:H 2O (c) Z n 5(O H )6(C O 3)2(c) Z n 5(O H )6(C O 3)2(c) ZnO(cr)
1.0 ZnO(cr) 1.0 Z n5(O H ) 8C l2(c )
Z n C O 3:H 2O (c)

0.8 0.8

0.6 0.6
Fraction

Fraction
0.4 0.4

0.2 0.2

0.0 0.0
6 8 10 12 14 6 8 10 12 14
a) pH b) pH

Fig. 14. Examples of pH areas for precipitation of different zinc compounds (a) in absence of Mg2+; (b) in presence of Mg2+ in the same concentration as Zn2+.

until the quantity of Mg2+ is sufficient. The transport of carbonate- Acknowledgements

ion is limited and the transformation of reactions (4–7) is stopped.
The formation of zincite is slow in comparison with formation of Authors are grateful for Dr. Grégory Léfévre for help in the titra-
simonkolleite and its fraction diminishes. This explains the exper- tion experiments and discussions and for Dr. Pièrre Dubot for help
imentally observed better corrosion resistance of Zn–Mg alloys for in PM-IRRAS experiments. We would like to thank the RFCS (Re-
cyclic salt spray tests independently of solution pH. search Fund for Coal and Steel) for partial support of this work.
5. In absence of Mg2+, the pH increases continuously and at very
high alkalinity the dissolution of oxide layer can be achieved due to References
the zincate formation.
6. In presence of Mg2+ the pH increase is limited to 10.2 by mag- [1] T. Koll, K. Ullrich, J. Faderl, J. Hagler, A. Spalek, Properties and potential
applications of ZnMg-alloy-coatings on steel sheet by PVD, in: Proceedings of
nesium hydroxide formation. This is translated into better corro- Galvatech ‘04, Association for Iron & Steel Technology, Chicago, IL, United
sion resistance in high-alkaline media (and hence in confined States, 2004, p. 803.
zones). [2] C. Schwerdt, M. Riemer, S. Koehler, B. Schuhmacher, M. Steinhorst, A. Zwick,
Properties and potential applications of novel ZnMg alloy coatings on steel
7. The absence of brucite in corrosion product suggests that dur- sheet, in: Proceeding of Galvatech ‘04, Association for Iron & Steel Technology,
ing cycling magnesium hydroxide Mg(OH)2 transforms to magne- Chicago, IL, United States, 2004, p. 783.
sium carbonates. [3] C. Hosking, M.A. Ström, P.H. Shipway, C.D. Rudd, Corrosion resistance of zinc–
magnesium coated steel, Corros. Sci. 49 (2007) 3669–3695.
[4] I. Odnevall, C. Leygraf, Reaction sequences in atmospheric corrosion of zinc, in:
5. Conclusions W.W. Kirk, Herbert H. Lawson (Eds.), Atmospheric Corrosion, ASTM STP 1239,
American Society for Testing and Materials, Philadelphia, 1994.
[5] I. Odnevall, Atmospheric Corrosion of Field Exposed Zinc. A Multianalytical
The results presented in this work demonstrate a clear bene- Characterization from Initial Films to Fully Developed Layers, Doctoral Thesis,
ficial anti-corrosion effect of Zn–Mg alloys as compared to Zn. Royal Institute of Technology, Stockholm, Sweden, 1994, ISBN 91-7170-866-9.
We attribute this effect to the enhanced stability of simonkolle- [6] Y. Miyoshi, Y. Ito, Corrosion protection functions of zinc coating on steel sheet,
as revealed by the investigation of field cars and laboratory test, Galvatech ’92,
ite on the Zn–Mg coating as compared to the Zn coating. Quali- Verlag Stahleisen, Amsterdam, The Netherlands, 1992, p.528.
tative analysis of the corrosion products indicate that [7] S. Kurokawa, M. Uchida, K. Mochizuki, Perforation corrosion mechanisms and life
simonkolleite is stabilized on the Zn–Mg coating throughout of coated steel sheets for automotive bodies, Kawasaki Steel Ges. 27 (1995) 33.
[8] Y. Morimoto, K. Honda, K. Nishimura, S. Tanaka, A. Takahashi, H. Shindo, M.
the atmospheric test exposure while it is observed only during
Kurosaki, Nippon Steel Technical Report No. 87, Excellent Corrosion-Resistant
the early stages on the Zn coating. Microscopic analysis demon- Zn–Al–Mg–Si Alloy Hot-dip Galvanized Steel Sheet ‘‘SUPER DYMA”, 2003, p.
strates that simonkolleite is more compact than other forms of 24.
[9] Shimizu Takeshi, Yoshizaki Fukio, Miyoshi Yasushi, Andoh Atsushi, Corrosion
Zn corrosion products observed here and thus probably serves
products of hot-dip Zn–6%Al–3%Mg coated steel sheet subjected to
as a stronger barrier to oxygen diffusion. Titrometric results have atmospheric exposure, Tetsu to Hagane 89 (1) (2003) 166–173.
demonstrated that simonkolleite is the most stable species [10] Q. Qing, C. Yan, Y. Wan, C. Cao, Effects of NaCl and SO2 on the initial
formed when hydroxide is added to a deaerated 5% NaCl solu- atmospheric corrosion of zinc, Corros. Sci. 44 (2002) 2789–2803.
[11] K. Ogle, V. Baudu, L. Garrigues, X. Philippe, Localized electrochemical methods
tion containing Zn2+ ions, regardless of whether or not Mg2+ is applied to cut edge corrosion, J. Electrochem. Soc. 147 (2000) 3654.
present. [12] K. Ogle, S. Morel, D. Jacquet, Observation of self healing functions on the cut
The protective influence of the Mg2+ ions formed by the anodic edge of galvanized steel using SVET and pH microscopy, J. Electrochem. Soc. B1
153 (2006).
dissolution of the Zn–Mg coating may contribute to this enhanced [13] F. Thébault, B. Vuillemin, R. Oltra, C. Allely, L. Dosdat, K. Ogle, Cathodic
stability in two unique mechanisms: reduction inhibition on galvanized steel cut-edges, Electrochim. Acta 53
(2008) 5226–5234.
[14] D. de la Fuente, J.G. Castano, M. Morcillo, Long-term atmospheric corrosion of
(1) Buffering of the pH around 10.2 due to the precipitation of zinc, Corros. Sci. 49 (2007) 1420.
Mg(OH)2 as demonstrated by the titration experiments. [15] M. Stratmann, R. Hausbrand, M. Rohwerder, Delamination resistant zinc
(2) The preferential reaction of Mg2+ with carbonate to form alloys: Simple concept and results on the system zinc-magnesium, Steel Res.
Int. 74 (2003) 453.
MgCO3, thereby removing carbonate ions from the system
[16] M. Stratmann, R. Hausbrand, M. Rohwerder, Proceedings of 13th Asian-Pacific
and stabilizing the simonkolleite and other zinc based corro- Corrosion Control Conference Osaka University, Japan, November, Japan
sion products. Society of Corrosion Engineering Paper No. PL-5, 2003.
1262 P. Volovitch et al. / Corrosion Science 51 (2009) 1251–1262

[17] K. Ishizuka, K. Nishimura, M. Kurosaki, K. Hayashi, H. Shindo, Nippon Steel [30] R. Lindstrom, L.G. Johansson, G.E. Thompson, P. Skeldon, J.-E. Svensson,
Technical Report No. 88, Chromate Free Electro-Galvanized Steel Sheets for Corrosion of magnesium in humid air, Corros. Sci. 46 (2004) 1141–1158.
Automobile Use, Zinkote-MZ, 2003, p. 44. [31] M. Jonsson, D. Persson, D. Thierry, Corrosion product formation during NaCl
[18] T. Tsuru, K.-I. Tamiya, A. Nishikata, Formation and growth of micro-droplets induced atmospheric corrosion of magnesium alloy AZ91D, Corros. Sci. 49
during the initial stage of atmospheric corrosion, Electrochim. Acta 49 (2004) (2007) 1540–1558.
2709–2715. [32] A. Froats, T.Kr. Aune, D. Hawke, W. Unsworth, J. Hillis, Corrosion of magnesium
[19] G. Deroubaix, P. Marcus, X-ray photoelectron spectroscopy analysis of copper and magnesium alloys, in: Metals Handbook, Corrosion, ninth ed., vol. 13,
and zinc oxides and sulphides, Surf. Interf. Anal. 18 (1992) 39. 1987.
[20] S. Peulon, D. Lincot, Mechanistic study of cathodic electrodeposition of zinc [33] Available from: <http://www.ascott-analytical.com/renault_bulletin.pdf>.
oxide and zinc hydroxy-chloride from oxygenated aqueous zinc chloride [34] A. Botha, C.A. Strydom, DTA and FT-IR analysis of the rehydration of basic
solutions, J. Electrochem. Soc. 145 (1998) 864. magnesium carbonate, J. Therm. Anal. Cal. 71 (2003) 987–995.
[21] A. Goux, T. Pauport, J. Chivot, D. Lincot, Temperature effects on ZnO [35] B. Lefebre, Corrosion des aciers revetus de zinc dans les zones confines des
electrodeposition, Electrochim. Acta 50 (2005) 2239–2248. carrosseries automobiles, Ph.D. Thesis, 2004, 221p. (in French).
[22] F.A. Sigoli, M.R. Davolos Jr., M. Jafelicci, Morphological evolution of zinc oxide [36] H.S. Hosseini, E. Forssberg, XPS & FTIR study of adsorption characteristics
originating from zinc hydroxide carbonate, J. Alloys Compd. 262–263 (1997) using cationic and anionic collectors on smithsonite, J. Miner. Mater.
292–295. Characterization Eng. 5 (1) (2006) 21–45.
[23] N. Kanari, D. Mishra, I. Gaballah, B. Dupré, Thermal decomposition of zinc [37] Available from: <http://www.mindat.org>.
carbonate hydroxide, Thermochim. Acta 410 (2004) 83–199. [38] R.L. Frost, J.Th. Kloprogge, Infrared emission spectroscopic study of brucite,
[24] J.-F. Hochepied, A.P. Almeida de Oliveira, V. Guyot-Ferreol, J.-F. Tranchant, Zinc Spectrochim. Acta A 55 (1999) 2195–2205.
oxide pompom-like particles from temperature-driven ammonia decomplexation, [39] H. Moller, The influence of Mg2+on the formation of calcareous deposits on a
J. Crystal Growth 283 (1–2) (2005) 156–162. freely corroding low carbon steel in seawater, Corros. Sci. 49 (2007) 1992–
[25] Ming Yang et al., Hydrothermal synthesis of one-dimensional zinc oxides with 2001.
different precursors, Nanotechnology 17 (2006) 206. ZnHO. [40] Sylvia Britto, A.V. Radha, N. Ravishankar, P. Vishnu Kamath, Solution
[26] T. Prosek, D. Thierry, C. Taxen, J. Maixner, Effect of cations on corrosion of zinc decomposition of the layered double hydroxide (LDH) of Zn with Al, Solid
and carbon steel covered with chloride deposits under atmospheric State Sci. 9 (2007) 279–286.
conditions, Corros. Sci. 49 (2007) 2676–2693. [41] G. Zou, R. Liu, W. Chen, Highly textural lamellar mesostructured magnesium
[27] Tatsuo Ishikawa, Kumi Matsumoto, Akemi Yasukawa, Kazuhiko Kandori, hydroxide via cathodic electrodeposition process, Mater. Lett. 6 (2007) 1990–
Takenori Nakayama, Takayuki Tsubota, Influence of metal ions on the 1993.
formation of artificial zinc rusts, Corros. Sci. 46 (2004) 329–342. [42] Available from: <http://www.kemi.kth.se/medusa/>.
[28] Tatsuo Ishikawa, Makoto Murai, Kazuhiko Kandori, Takenori Nakayama, [43] K. Ogle, H. Bouazaze, M.N. Yattara, Direct measurement of the chemical
Structure and composition of artificially synthesized rusts of Zn–Fe and Zn– stability of the Zn/ZnO/polymer interface, in: The 59th Annual Meeting of the
Ti alloys, Corros. Sci. 48 (2006) 3172–3185. International Society of Electrochemistry, 7–12 September 2008, Seville, Spain,
[29] T. Prosek, A. Nazarov, U. Bexell, D. Thierry, J. Serak, Corrosion mechanism of s06-P-037.
model zinc–magnesium alloys in atmospheric conditions, Corros. Sci. 1 (2008) 1.