You are on page 1of 12

Available online at www.sciencedirect.

com

Corrosion Science 50 (2008) 823–834


www.elsevier.com/locate/corsci

Corrosion behaviour of magnesium/aluminium alloys


in 3.5 wt.% NaCl
A. Pardo a,*, M.C. Merino a, A.E. Coy a, R. Arrabal b, F. Viejo b, E. Matykina b
a
Departamento de Ciencia de Materiales, Facultad de Quı́micas, Universidad Complutense, 28040 Madrid, Spain
b
Corrosion and Protection Centre, School of Materials, The University of Manchester, P.O. Box 88, Sackville Street,
Manchester M60 1QD, United Kingdom

Received 13 July 2007; accepted 7 November 2007


Available online 24 November 2007

Abstract

Corrosion behaviour of commercial magnesium/aluminium alloys (AZ31, AZ80 and AZ91D) was investigated by electrochemical and
gravimetric tests in 3.5 wt.% NaCl at 25 °C. Corrosion products were analysed by scanning electron microscopy, energy dispersive X-ray
analysis and low-angle X-ray diffraction. Corrosion damage was mainly caused by formation of a Mg(OH)2 corrosion layer. AZ80 and
AZ91D alloys revealed the highest corrosion resistance. The relatively fine b-phase (Mg17Al12) network and the aluminium enrichment
produced on the corroded surface were the key factors limiting progression of the corrosion attack. Preferential attack was located at the
matrix/b-phase and matrix/MnAl intermetallic compounds interfaces.
Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: A. Magnesium; B. Polarization; B. Weight loss; C. Corrosion

1. Introduction knowledge of corrosion mechanisms has lead to an increase


of the real and potential applications of magnesium alloys
AZ (Mg–Al–Zn) system, containing 2–10% Al with [3].
minor additions of Zn and Mn, is the most widely used Corrosion resistance of Mg alloys depends on many
among Mg–Al alloys. They are characterised by low cost factors: (i) environment, (ii) alloy composition and micro-
of production and also by relatively good corrosion resis- structure, and (iii) properties of the film developed in the
tance and satisfactory mechanical properties from 95 to medium to which they are exposed. Concerning the
120 °C [1]. environment, corrosion resistance of magnesium alloys in
However, corrosion performance of magnesium alloys chloride containing solutions greatly depends on pH and
has been a major obstacle to their growth in structural Cl concentration, with no significant influence of oxygen
applications, despite their high stiffness/weight ratio, ease concentration [4,5]. In general, magnesium and its alloys
of machinability, high damping capacity, castability, wel- dissolve at a very low rate in alkaline or poorly buffered
dability and recyclability [2]. These properties make them sodium chloride solutions, where the pH can increase,
suitable for aerospace and automotive applications, where due to the formation of a partially protective Mg(OH)2
light metals are mandatory in order to reduce weight and layer [4]. On the other hand, chloride ions promote rapid
greenhouse gas emissions. Over the past 30 years, alloy attack in neutral aqueous solutions and even higher in
design development, new surface treatments and improved acidic solutions [4,6]. It is also common to find higher cor-
rosion rates with increasing Cl ion concentration at all pH
levels [6]. The corrosion of magnesium alloys in non-oxidiz-
*
Corresponding author. Tel.: +34 1 3944348; fax: +34 1 3944357. ing neutral or alkaline chloride solutions at free corrosion
E-mail address: anpardo@quim.ucm.es (A. Pardo). potential typically initiates as irregular pits, which spread

0010-938X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2007.11.005
824 A. Pardo et al. / Corrosion Science 50 (2008) 823–834

laterally and cover the whole surface [7,8]. However, the sion behaviour in chloride media [13–15], but the specific
mechanism is different [9,10] from the auto-catalytic pitting mechanism and influence of Al is still not well understood.
experienced by stainless steels [11], since there does not For instance, Lunder [16] found that increasing concentra-
seem to be much tendency for deep pitting, possibly as a tions of 2–8 wt.% Al in die cast AS, AM and AE alloys
result of pH increase and magnesium hydroxide film for- decrease the corrosion rate in 5% NaCl, however, the
mation. However, this is not always true, since there is a decrease in corrosion rate appears to be related to
significant influence of the microstructure on the corrosion a decrease in the impurity level with increasing Al content
mechanism, especially in two-phase magnesium alloys. (it is known that Al reduces the iron tolerance limit from
On the topic of alloy composition and microstructure it 170 wt.-ppm to 20 wt.-ppm [17]). More recent data com-
is known that alloying elements not only modify the pared the corrosion rates of high purity alloys in 3% NaCl
mechanical properties of magnesium, but also impart a [18]. Results showed that HP–Mg5Al alloy had a corrosion
significant impact on the corrosion behaviour. Alloying ele- rate significantly higher than that of HP–Mg due to the
ments can form secondary particles, which are noble with micro-galvanic corrosion acceleration of the corrosion of
respect to the magnesium matrix, thereby facilitating the a-phase by the adjacent b-phase. On the other hand,
corrosion, or enrich the corrosion product thereby possibly the high purity two phase commercial Mg alloys (MEZ,
inhibiting the corrosion rate [12]. AM60 and AZ91D) had similar corrosion rates to that of
In general, it is reported that increasing Al concentra- HP–Mg, despite the fact that these commercial alloys each
tions in Mg–Al alloys have a beneficial effect on the corro- had a two phase microstructure which would cause a

Table 1
Nominal composition of the materials tested
Material Chemical composition (wt.%)
Al Zn Mn Si Cu Fe Ni Ca Zr Others
Mg (99%) 0.006 0.014 0.03 0.019 0.001 0.004 <0.001
AZ31 3.1 0.73 0.25 0.02 <0.001 0.005 <0.001 <0.01 <0.001 <0.30
AZ80 8.2 0.46 0.13 0.01 <0.001 0.004 <0.30
AZ91D 8.8 0.68 0.30 0.01 <0.001 0.004 <0.008 <0.30

Fig. 1. SEM micrographs of: (a) unalloyed Mg; (b) AZ31 alloy; (c) AZ80 alloy and (d) AZ91D alloy.
A. Pardo et al. / Corrosion Science 50 (2008) 823–834 825

galvanic acceleration of the corrosion of the a-phase


[19,20]. Therefore, apart from impurities and Al concentra-
tion it is also necessary to study the effect of phase
distribution.
Normally, in Mg–Al alloys the aluminium is partly in
solid solution and partly precipitated in the form of
Mg17Al12 along grain boundaries as a continuous phase
as well as part of a lamellar structure. It is known that
Mg17Al12 exhibits a passive behaviour over a wider pH
range than either of its component aluminium and magne-
sium [21], and it was found that the distribution of the
Mg17Al12 phase determines the corrosion resistance of the
Mg–Al alloys [22]. Song et al. [17] suggested that the b-
phase mainly served as a galvanic cathode and accelerated
the corrosion process of the a-matrix if the volume fraction
of b-phase was small; however for a high volume fraction,
the b-phase might act as an anodic barrier to inhibit the Fig. 2. Mass loss versus time for the materials immersed in 3.5 wt.% NaCl
overall corrosion of the alloy. It has also been reported that solution.
the ratio of the b-phase to the surrounding Al-rich-a or
eutectic-a, which can be about 12 wt.% compared to Table 2
1.5 wt.% in the grain centre [23], plays an important role Kinetic laws of the materials immersed in 3.5 wt.% NaCl solution
in the formation of galvanic cells. Thus, Raman [24] found Material Kinetic law: y = b  t; [y (mg/cm2), t (h)]
that an increase in the relative size of the b-phase at the Mg (99%) y = 13.03t 0 6 t 6 16 (r2 = 0.99)
expense of the Al-rich-a area, increasing the cathode to AZ31 y = 2.31  101t 0 6 t 6 240 (r2 = 0.99)
anode area ratio, results in an increase in the localized cor- AZ80 y = 3.10  103t 0 6 t 6 240 (r2 = 0.98)
rosion. However, the existence of a surrounding Al-rich-a AZ91D y = 3.70  103t 0 6 t 6 240 (r2 = 0.98)
area with independent electrochemical identity appears to
be greatly dependant on the thermal history and composi-
a -1.1
tion of the alloy. There are also contradictory results AZ31
depending on aluminium concentration in the a-phase. -1.2
For instance, Lunder et al. [21] proposed that aluminium
-1.3
accelerate anodic dissolution below 8% whereas it
decreases corrosion above 10%. However, Song [25,26]
E SSE (V)

-1.4
found that an increase of Al in the a-phase improves the
-1.5
corrosion resistance. Moreover, results obtained for rapid
solidified alloys with different additions of Al and no pre- -1.6
cipitation of b-phase revealed a positive effect of Al on cor-
-1.7
rosion resistance [27]. In summary, it seems that Al
improves the corrosion resistance of Mg–Al alloys in chlo- -1.8
-7 -6 -5 -4 -3 -2
ride environments but its effect is strongly influenced by the 10 10 10 10 10 10
2
b-phase morphology and distribution. I (A/cm )
The aim of this work was to study the corrosion behav- b -1.2
iour of three commercial magnesium/aluminium alloys in AZ91D
marine environment. The effect of immersion time and Al -1.3
concentration on corrosion resistance was monitored by
electrochemical and gravimetric tests, scanning and elec- -1.4
ESSE (V)

tron microscopy (SEM), energy dispersive X-ray analysis


-1.5
(EDX) and low-angle X-ray diffraction (XRD).
-1.6
2. Experimental procedure
-1.7
2.1. Test materials
-1.8
-7 -6 -5 -4 -3 -2
10 10 10 10 10 10
Chemical compositions of the tested magnesium alloys, 2
namely AZ31, AZ80 and AZ91D, are listed in Table 1. I (A /cm )
Unalloyed Mg was used as the reference material. Pure Fig. 3. Double cyclic polarization curves after the immersion in 3.5 wt.%
Mg and AZ31 alloy were fabricated in wrought condition, NaCl solution during 1 h for the alloys: (a) AZ31 and (b) AZ91D.
826 A. Pardo et al. / Corrosion Science 50 (2008) 823–834

whereas AZ80 (chill casting) and AZ91D alloys were cases, the tests were performed in duplicate to guarantee
manufactured by ingot casting process. All the materials the reliability of the results.
were supplied by Magnesium Elektron Ltd.
2.3. Gravimetric measurements
2.2. Preparation and surface characterization
Gravimetric measurements were performed using rect-
For metallographic characterization, samples were wet angular (Mg and AZ31) and cylindrical (AZ80 and
ground through successive grades of silicon carbide abra- AZ91D) specimens of working area 16 cm2 immersed in
sive papers from P120 to P2000 followed by diamond fin- 3.5 wt.% NaCl solution at room temperature (pH 5.6,
ishing to 0.1 lm. Two etching reagents were used: (a) 25 °C). Prior to the tests, specimens were measured and
Nital, 5 ml HNO3 + 95 ml ethanol, to reveal the constitu- weighed. Once the test was finished for each immersion
ents and general microstructure of Mg, AZ80 and time, the samples were extracted, rinsed with isopropyl
AZ91D materials and (b) Vilella reagent, 0.6 g picric alcohol, dried in hot air, and then weighed again in order
acid + 10 ml ethanol + 90 ml H2O, to reveal grain bound- to calculate the mass gain per unit surface area. Corrosion
aries of AZ31 alloy. The constituents were examined by rate was calculated from the mass losses per unit of surface
SEM using a JEOL JSM-6400 microscope equipped with area, calculated from the expression (Mi  Mf)/A, where
Oxford Link EDX microanalysis hardware. For low-angle Mi is the initial mass, Mf the final mass and A the exposed
XRD studies a Philips X́Pert diffractometer Ka Cu = surface area. In order to obtain further information, pH
1.54056 Å) was used. changes of the test solution during the gravimetric experi-
Prior to corrosion tests, specimens were wet ground to ments were recorded with a standard pH-meter. For the
grade P1200 grit, followed by rinsing with isopropyl alco- same reason, hydrogen evolution was measured using a
hol in an ultrasonic bath and drying in warm air. In all simple procedure described elsewhere [9].

-1.0 -1.0
Mg AZ31
-1.1 1h -1.1 1h
1d 1d
-1.2 -1.2 3d
7d
-1.3 -1.3
ESSE (V)

-1.4 -1.4

-1.5 -1.5

-1.6 -1.6

-1.7 -1.7

-1.8 -1.8
-8 -7 -6 -5 -4 -3 -2 -8 -7 -6 -5 -4 -3 -2
10 10 10 10 10 10 10 10 10 10 10 10 10 10
2 2
I (A /cm ) I (A /cm )
-1.0 -1.0
AZ80 AZ91
-1.1 1h -1.1 1h
1d 1d
-1.2 3d -1.2 3d
7d 7d
-1.3 -1.3
ESSE (V)

-1.4 -1.4

-1.5 -1.5

-1.6 -1.6

-1.7 -1.7

-1.8 -1.8
-8 -7 -6 -5 -4 -3 -2 -8 -7 -6 -5 -4 -3 -2
10 10 10 10 10 10 10 10 10 10 10 10 10 10
2 2
I (A /cm ) I (A /cm )
Fig. 4. Influence of the immersion time on the anodic behaviour for the materials tested in 3.5 wt.% NaCl solution.
A. Pardo et al. / Corrosion Science 50 (2008) 823–834 827

2.4. Electrochemical measurements ited linear kinetics of mass loss associated with magnesium
matrix dissolution. Unalloyed Mg presented the highest
DC electrochemical measurements were performed mass loss and it was completely dissolved after 16 h of
using specimens with 2 cm2 of working area exposed to immersion with a mass loss of 196 mg/cm2. On the other
3.5 wt.% NaCl at room temperature (25 °C). An AUTO- hand, the addition of aluminium increased notably the cor-
LAB model PGSTAT 30 potentiostat connected to a rosion resistance. Thus, 3 wt.% Al in AZ31 alloy reduced
three-electrode cell was used for the electrochemical mea- the mass loss to 55.2 mg/cm2 after 10 days of immersion,
surements; the working electrode was the test material, and 8–9 wt.% Al in AZ80 and AZ91D alloys diminished
whereas the counter and reference electrodes were graphite the mass loss to 0.75 and 1.05 mg/cm2 at the end of the test,
and Ag/AgCl (SSE), respectively. Solution concentration respectively.
inside the reference electrode compartment was KCl 3 M, Table 2 shows the kinetic laws, calculated from the
with a potential of 0.197 V with respect to hydrogen. Ano- experimental data, corresponding to the growth of corro-
dic, cyclic and double cyclic polarization measurements sion products layer generated during the gravimetric test.
were carried out at a scan rate of 0.3 mV/s, from In each case, data have been approximated by a linear
100 mV to +400 mV with respect to the corrosion poten- equation y = b  t, where ‘‘y” coordinate represents the
tial (Ecorr). For cyclic polarization, the scan direction was mass gain in units of mg/cm2 and ‘‘t” is the immersion time
reversed when the samples reached the anodic corrosion in hours. Aluminium reduced the magnesium reactivity, i.e.
current of 5 mA and potential was scanned back to the 3 wt.% Al reduced corrosion rate by a factor of 56 and 8–
starting potential. 9 wt.% Al did it by a factor of 300. In summary, Mg and
AZ31 alloy manifested very low corrosion resistance in sea-
2.5. Characterization of corrosion products water, whereas AZ80 and AZ91D alloys presented a mod-
erate resistance after 10 days of immersion.
After the tests, magnesium alloys were observed by
SEM in order to study the morphology and evolution of
corrosion products formed on the material surface. Also,
the composition of the corrosion layer was analysed by
low-angle XRD.
a -1.0
Mg
-1.1 AZ31
3. Results AZ80
AZ91
-1.2
3.1. Microstructural characterization
Ecorr (VSSE)

-1.3
Fig. 1 shows the SEM microstructure of tested materi-
als. Unalloyed Mg only reveals equiaxial grains with aver- -1.4
age size of 45 lm, whereas the presence of aluminium
(3.1 wt.%) and manganese (0.25 wt.%) in AZ31 alloy -1.5
favours the formation of intermetallic phases, mainly in
-1.6
form of MnAl2 inclusions (Figs. 1a, b). On the other hand, 0 1 2 3 4 5 6 7
AZ80 and AZ91D ingot casting alloys disclose two differ- Time (d)
ent types of solidification microstructures. AZ80 alloy
shows a biphasic microstructure with a a-Mg solid solution b 10
5

Mg
and a discontinuous precipitation in lamellar form of b-
AZ31
phase (Mg17Al12), which starts at the grain boundaries 4 AZ80
from solid solution (Fig. 1c). AZ91D alloy reveals a-Mg 10
AZ91
primary dendrites and eutectic a-Mg/Mg17Al12 in the inter-
Rp (Ω·cm )
2

dendritic region, which appears completely in divorced 10


3

form with respect to solid solution (Fig. 1d). Indeed, the


rapid solidification obtained by chill casting process used
2
for AZ80 alloy promotes a refinement of the microstruc- 10
ture compare to AZ91D alloy. Finally, MnAl2 intermetallic
inclusions are also found for AZ80 and AZ91D alloys. 1
10
0 1 2 3 4 5 6 7
3.2. Gravimetric results Time (d)
Fig. 5. Influence of the immersion time for materials tested in 3.5 wt.%
Fig. 2 shows the mass loss versus time of tested materials NaCl solution on: (a) polarization resistance (Rp) and (b) corrosion
immersed in 3.5 wt.% NaCl solution. All materials exhib- potential (Ecorr).
828 A. Pardo et al. / Corrosion Science 50 (2008) 823–834

Fig. 6. SEM micrographs of alloys immersed in 3.5 wt.% NaCl solution for 10 days: (a) AZ31; (b) AZ80 and (c) AZ91D.

Fig. 7. SEM micrographs of alloys immersed in 3.5 wt.% NaCl solution for 2 h: (a) AZ31; (b) AZ80 and (c) AZ91D.
A. Pardo et al. / Corrosion Science 50 (2008) 823–834 829

3.3. Electrochemical results one order of magnitude lower. AZ31 presented the most
significant change on the Ecorr values, possibly due to the
In order to obtain further information, the gravimetric high reactivity of this alloy which promotes formation of
study was supplemented with potentiodynamic polariza- a thick corrosion layer, limiting the attack progression
tion measurements for different times of immersion in and increasing Ecorr values (Fig. 5a). Concerning polariza-
3.5 wt.% NaCl solution. Double cycle polarization curves tion resistance (Rp) values, they remained low after 7 days
for AZ31 and AZ91D alloys after immersion for 1 h reveal of immersion (between 102 and 104 X cm2) (Fig. 5b). How-
a different behaviour compared to materials with a stable ever, an increase of Rp values and, therefore, a better cor-
passive layer (Fig. 3). The onset of pitting is not visible in rosion behaviour, it is observed for Mg and AZ31
the forward scan of both cycles, since pitting potential materials, possibly related to the nucleation and growth
(Epit) is very close to Ecorr and, consequently, it should be of a thick and semi-protective layer of corrosion products.
expected that these alloys suffer pitting attack immediately
after their immersion in the aggressive media at the open 3.4. Characterization of corrosion products
circuit potential. However, the current densities of the
cathodic branch and, therefore, the growth of corrosion The SEM micrographs of the surface of the tested mate-
products on the material surface are quite high, suggesting rials after 10 days of immersion in 3.5 wt.% NaCl solution
general corrosion attack as the main mechanism of are shown in Fig. 6. Unalloyed magnesium is not presented
degradation. due to its complete dissolution after 24 h of exposure.
Fig. 4 discloses the evolution of the electrochemical Although AZ31 alloy revealed higher corrosion resistance
behaviour of each material with increasing immersion time. than unalloyed magnesium, its surface was completely cov-
The high dissolution rate observed for unalloyed magne- ered by a thick and uneven film of corrosion products. An
sium limits this study, since it was dissolved after 1 day increase of aluminium content up to 8–9 wt.% reduced the
of immersion according to gravimetric results. Neverthe- reactivity of the magnesium alloys and, thereby, AZ80 and
less, longer immersion times shifted its curves to greater AZ91D alloys presented a lower degree of corrosion.
current densities, possibly associated with the formation Initial stages of corrosion for AZ31, AZ80 and AZ91D
of a low-protective oxide layer, which does not impede alloys immersed in 3.5 wt.% NaCl revealed localised corro-
the corrosion attack progression. AZ80 and AZ91 alloys sion around MnAl2 inclusions and b-phase interfaces,
revealed a slight shift of Ecorr towards more noble values, which form a galvanic couple with the surrounding Mg
and corrosion current densities (icorr) were approximately matrix (Fig. 7). Polarization tests did not show a clearly

Fig. 8. Cross-section BSE image and corresponding X-ray maps of Mg, Al and O for the AZ31 alloy immersed in 3.5 wt.% NaCl for 10 days.
830 A. Pardo et al. / Corrosion Science 50 (2008) 823–834

defined pitting potential, so the corrosion process should tance exhibited by this alloy. Additionally, it can be
be associated with a localised attack, but not to a typical observed that the front of the corrosion attack was stopped
pitting attack reported for materials with a stable passive when it reached the lamellar aggregate of b-phase
layer, such as aluminium alloys, stainless steels, etc. (Mg17Al12). Thus, the higher corrosion resistance of
X-ray map analysis of the cross section of AZ31 alloy AZ80 alloy may be associated with the presence of alumin-
after immersion for 10 days in the corrosive medium con- ium and its direct or indirect intervention in corrosion
firmed the formation of a thick corrosion layer with an mechanism, consisting of two steps: (a) formation of a
irregular thickness between 200 and 400 lm and mainly semi-protective Al-rich oxide layer and (b) reduction of
constituted by magnesium oxides and/or hydroxides the corrosion progression near to the lamellar b-phase
(Fig. 8). AZ80 alloy revealed a discontinuous and cracked aggregate.
corrosion film of lower thickness than AZ31 alloy Finally, Figs. 10 and 11 show the study of the corrosion
(<200 lm) (Fig. 9). Furthermore, aluminium enrichment layer formed for AZ91D alloy after immersion for 10 days.
was found at the metal/corrosion layer interface, possibly Like for the AZ80 alloy, corrosion layer was irregular,
produced by preferential Mg dissolution during the immer- cracked, and 200 lm thick (Fig. 10). However, the posi-
sion test. The formation of Al oxide species is likely to be tive effect of Al on the corrosion performance cannot be
one of the main reasons for the improved corrosion resis- explained in the same way as above, since aluminium

Fig. 9. Cross-section BSE image, profile line and corresponding X-ray maps of Mg, Al and O for the AZ80 alloy immersed in 3.5 wt.% NaCl for 10 days.
A. Pardo et al. / Corrosion Science 50 (2008) 823–834 831

rosion mechanism of magnesium needs further investiga-


tion, it is generally reported that the following anodic
and cathodic reactions take place in marine environments:
MgðsÞ ! Mg2þ ðaqÞ þ 2e ðanodic reactionÞ ð1Þ
2H2 O þ 2e ! H2 þ 2OH ðaqÞ ðcathodic reactionÞ ð2Þ
2+
Firstly, magnesium dissolves and Mg (aq) cations are
produced (Eq. (1)), possibly through intermediate steps
involving monovalent magnesium ion [14,30]. Secondly,
magnesium dissolution is accompanied by hydrogen evolu-
tion (Eq. (2)), since magnesium in neutral and low pH
aqueous solutions is well below the region of water stabil-
ity. Finally, pH raises along with the cathodic reaction due
to the formation of OH–, which favours the formation of
Mg(OH)2(s) according to Pourbaix diagram (Fig. 13a)
[31]. Thus, the overall reaction could be expressed as
Mg2þ ðaqÞ þ 2OH ðaqÞ ! MgðOHÞ2 ðsÞðcorrosion productÞ
ð3Þ
Since pH changes and H2 evolution can be easily measured,
it is feasible to study the corrosion evolution of magnesium
with these parameters [19]. Fig. 13b shows the variation of
pH for all tested materials immersed in 3.5 wt.% NaCl (pH
5.6). For commercially pure Mg, test solution revealed the
highest pH raise, pH >10 after immersion for 1 h. On the
Fig. 10. Cross-section BSE image and corresponding X-ray map of Al for other hand, even though aluminium reduces the magne-
the AZ91D alloy immersed in 3.5 wt.% NaCl for 10 days. sium reactivity, pH values were registered between 8 and
9.5 after 1 h, the pH increasing up to 10 for longer immer-
enriched eutectic aggregate solidifies at the interdendritic sion times. Therefore, these technique may be convenient
spaces resulting in an a-primary phase with lower Al con- and reliable for measurement of initial corrosion of magne-
tent than the predicted by the equilibrium phase diagram. sium and its alloys, however it becomes gradually insensi-
Therefore, (a) there was not an appreciable Al enrichment tive for long times, since the solution is saturated with
observed on the surface after the Mg dissolution during the Mg(OH)2, and concentrations of Mg2+ and OH cannot
corrosion attack. Although, as mentioned above, (b) the further increase while corrosion is still proceeding [19].
presence of a relatively fine b-phase network, which for this Concerning H2 evolution Fig. 13c compares mass loss
particular alloy is extended all along the surface, partially obtained by gravimetric measurements with H2 evolution,
impeded the corrosion attack (Fig. 11). assuming that the dissolution of one atom of magnesium
Low-angle XRD study (incident angle of 1°) of the cor- generates one molecule of hydrogen (Eqs. (1) and (2)).
rosion layer produced after immersion in 3.5 wt.% NaCl Results confirmed the higher corrosion resistance of
for 10 days revealed brucite (Mg(OH)2) as the main corro- AZ80 and AZ91D alloys and the high dissolution rate of
sion product, and its peaks exhibited higher intensity for pure Mg in the test solution. In all cases, corrosion rate
unalloyed Mg and AZ31 alloy due to the formation of a estimated by H2 evolution was lower than the one obtained
thicker corrosion layer during the severe attack that both by gravimetric tests, possibly due to small particles falling
materials suffered (Fig. 12). Unlike other works, evidences out of the surface after having been undermined by the cor-
of magnesite (MgCO3) formation, due to CO2 presence in rosion process [9]. Therefore, although pH and H2 evolu-
the atmosphere, were not found [28]. tion provide information regarding the corrosion of
magnesium alloys, it is still suggested to perform gravimet-
ric measurements, since they are the most basic and reliable
4. Discussion method.
In general, it is reported that the presence of aluminium
When unalloyed magnesium is exposed to the atmo- in single-phase Mg–Al alloys has a beneficial effect on the
sphere or aqueous solutions, a grey oxide (mainly magne- corrosion behaviour in chloride media [13–15]. In the pres-
sium hydroxide, brucite) forms on its surface, which is ent work, this behaviour has been clearly revealed for the
stable for the basic range of pH values [29]. Nevertheless, AZ31 alloy, which exhibited higher corrosion resistance
in presence of chloride anions, this surface film breaks than commercially pure Mg. However, aluminium influ-
down and magnesium appears unprotected. Although cor- ence in two-phase Mg–Al alloys is still not well understood
832 A. Pardo et al. / Corrosion Science 50 (2008) 823–834

Fig. 11. Cross-section BSE image of detail of the corrosion layer, profile line and corresponding X-ray maps of Mg, Al and O for the AZ91D alloy
immersed in 3.5 wt.% NaCl for 10 days.

since depends on many factors, such as impurities, b-phase


morphology and distribution, aluminium content and size
of the a-Mg primary dendrites and the eutectic-a, alumin-
ium enrichment on the corrosion product layer [12], etc.
Magnesium and magnesium alloys present an active
position in both the electromotive force (EMF) and the gal-
vanic series for seawater. As a result, galvanic interactions
between magnesium and other metals are a serious con-
cern. Thus, the corrosion resistance of Mg–Al alloys will
depend on the presence of alloy elements acting as active
cathode, such as manganese, iron, etc. The most detrimen-
tal potential cathodes in Mg–Al alloys are iron-rich and
MnAl second phase particles [22]. The former were not
found in any of the studied alloys, whereas MnAl2 interme-
Fig. 12. Low-angle XRD (incidence angle: 1°) of tested materials tallic compounds were identified as preferential sites for
immersed in 3.5 wt.% NaCl solution for 10 days. localised corrosion in AZ31, AZ80 and AZ91D alloys.
A. Pardo et al. / Corrosion Science 50 (2008) 823–834 833

b 12
Mg
AZ31
11 AZ80
AZ91

pH
10

8
0 1 2 3 4 5 6 7
Time (d)

Fig. 13. (a) Pourbaix Diagram for magnesium; (b) pH variation and (c) mass loss and H2 evolution rates for all materials tested in 3.5 wt.% NaCl for 10
days.

Previous investigations reported that distribution of the AZ80 and AZ91D showing the highest corrosion resis-
b-phase (Mg17Al12) determines the corrosion resistance of tance; (ii) an independent electrochemical Al-rich-a area
the Mg–Al alloys [17,19–22,24,32]. According to this, for surrounding the b-phase was not observed by metallo-
a low volume fraction, the b-phase serves as a galvanic graphic characterization; (iii) AZ91D alloy exhibits a
cathode and accelerates the corrosion process of the a- slightly worse corrosion behaviour than AZ80 alloy,
phase, whereas large volume fractions act as an anodic bar- although the aluminium content in its nominal composi-
rier and the overall corrosion diminishes [26,33]. Moreover, tion is slightly higher and both alloys reveal the presence
it has also been reported that the ratio of the b-phase to the of relatively fine b-phase network segregated in the matrix
surrounding Al-rich-a or eutectic-a plays an important role alloy, which acts as an effective barrier against progression
in the formation of galvanic cells. Apart from the micro- of corrosion attack. As a result, the main reason for this
structure, the composition of the corrosion layer may different corrosion behaviour observed for the AZ80 and
improve the corrosion resistance. For instance, Nordlien AZ91D alloys could be explained in terms of aluminium
et al. [34] observed that alumina components forms a con- enrichment on the metallic surface during the magnesium
tinuous skeletal structure in an oxide layer of the magne- matrix dissolution. According to this, aluminium content
sium alloy, which passivating properties are much better for both alloys in the a-Mg primary dendrites was esti-
than Mg(OH)2 and MgO layers. mated by EDX. Chill casting employed for AZ80 alloy pro-
Therefore, keeping in mind these observations and the motes high aluminium contents in the solid solution
results obtained in the present work, the following consid- (13.3 at.%), meanwhile ingot casting, with lower solidifica-
erations about the corrosion mechanism of studied Mg–Al tion rate, promotes a coarse dendritic microstructure for
alloys can be summarized: (i) an increase of the aluminium the AZ91D alloy and less aluminium content in the solid
concentration in the nominal composition of the alloys solution (8.4 at.%). Hence, AZ80 alloy revealed aluminium
reduced the activity of the pure Mg in 3.5 wt.% NaCl, with enrichment on the metallic surface after immersion in
834 A. Pardo et al. / Corrosion Science 50 (2008) 823–834

3.5 wt.% NaCl for 10 days, whereas this aluminium-rich [4] W.S. Loose, Corrosion and Protection of Magnesium, in: L.M.
oxide layer was not observed for the AZ91D alloy (Figs. Pidgeon, J.C. Mathes, N.E. Woldmen (Eds.), ASM Int, Materials
Park, OH, 1946, p. 173.
9 and 11). [5] R. Ambat, N.N. Aung, W. Zhou, J. Appl. Electrochem. 30 (2000)
865.
5. Conclusions [6] R. Ambat, N.N. Aung, W. Zhou, Corros. Sci. 42 (2000) 1433.
[7] G. Song, A. Atrens, Adv. Eng. Mater. 9 (2007) 177.
1. Corrosion attack of all tested magnesium alloys occurs [8] R. Tunold, H. Holtan, M.-B.H. Berge, A. Lasson, R. Steen-Hansen,
Corros. Sci. 17 (1977) 353.
at the a-magnesium matrix/Al–Mn and Mg17Al12 inter- [9] G. Song, A. Atrens, Adv. Eng. Mater. 5 (2003) 837.
metallic compounds interfaces, by means of the forma- [10] G. Song, A. Atrens, D. St John, J. Nairn, Y. Li, Corros. Sci. 39 (1997)
tion of galvanic couples. Later, the nucleation and 855.
growth of an irregular and less protective corrosion [11] A. Atrens, R. Coade, J. Allison, H. Kohl, G. Hochoertler, G. Krist,
layer consisted mainly of Mg(OH)2 is produced from Mater. Forum 17 (1993) 263.
[12] B.E. Carlson, J.W. Jones, The Metallurgical Aspects of the Corrosion
a-Mg matrix. Behavior of Cast Mg–Al Alloys, in: Light Metals Processing and
2. An increase of the aluminium concentration in the Applications, METSOC Conference, Quebec, 1993.
nominal composition of the alloys reduced the activity [13] K. Nisancioglu, O. Lunder, T.K. Aune, Corrosion Mechanism of
of the pure Mg in 3.5 wt.% NaCl. However, the AZ31 AZ91 Magnesium Alloy, in: Proceedings of 47th World Magnesium
alloy (3.1 wt.% Al) still presented high corrosion rates. Association, Mcleen, Virginia, 1990.
[14] G.L. Makar, J. Kruger, J. Electrochem. Soc. 137 (1990) 414.
3. The principal cause of higher corrosion resistance of [15] C.B. Baliga, P. Tsakiropoulos, Mater. Sci. Technol. 9 (1993) 513.
AZ80 alloy is associated with a dual mechanism. Firstly, [16] O. Lunder, K. Nisancioglu, R.S. Hansen, Corrosion of Die Cast
the magnesium matrix dissolution, during the corrosion Magnesium–Aluminum Alloy, in: Congress and Exposition, Detroit,
attack, favours aluminium enrichment on the metallic Michigan, 1993, p. 117.
surface and allows the formation of a semi-protective [17] G.L. Song, A. Atrens, Adv. Eng. Mater. 1 (1999) 11.
[18] Z. Shi, G. Song, A. Atrens, Surf. Coat. Technol. 201 (2006) 492.
Al-rich oxide layer which improves the corrosion resis- [19] G. Song, Adv. Eng. Mater. 7 (2005) 563.
tance of the alloy. Additionally, the lamellar aggregate [20] G. Song, D.H. StJohn, J. Light Met. 2 (2002) 1.
network of b-phase acts as a barrier to the progression [21] O. Lunder, J.E. Lein, T.K. Aune, K. Nisancioglu, Corrosion 45
of the corrosion attack. (1989) 741.
4. AZ91D presented similar corrosion behaviour as AZ80 [22] R.C. Zeng, J. Zhang, W.J. Huang, W. Dietzel, K.U. Kainer, C.
Blawert, W. Ke, Trans. Nonf. Metals Soc. China 16 (2006) s763.
alloy, but the different solidification microstructure [23] L. Wei, H. Westengen, T.K. Aune, D. Albright, Magnesium
changes the mechanism of the corrosion attack for this Technology, Warendale, 2000, pp. 153.
alloy. In this case, the corrosion resistance is exclusively [24] R.K.S. Raman, Metall. Mater. Trans. A 35 (2004) 2527.
attributed to the presence of a network of eutectic aggre- [25] G. Song, A.L. Bowles, D.H. StJohn, Mater. Sci. Eng. A 366 (2004)
gate with higher Al content, which limits the advance of 74.
[26] G. Song, A. Atrens, X. Wu, Z. Bo, B. Zhang, Corros. Sci. 40 (1998)
the corrosion attack. 1769.
[27] F. Hehmann, F. Sommer, H. Jonets, R. Edyvean, J. Mater. Sci. 24
(1989) 2369.
Acknowledgements [28] M. Jönsson, D. Persson, D. Thierry, Corros. Sci. 49 (2007) 1540.
[29] B.A. Shaw, Corrosion Resistance of Magnesium Alloys, in: ASM
Handbook. Corrosion Fundamentals, Testing and Protection, vol.
The authors wish to thank the MCYT for the financial 13A, 2003.
support given to this work (Project MAT2006-13179- [30] G.R. Hoey, M. Cohen, J. Electrochem. Soc. 105 (1958) 245.
C02-01-02). [31] J. Van Muylder, M. Pourbaix, Magnesium, in: M. Pourbaix (Ed.),
Atlas of Electrochemical Equilibria in Aqueous Solution, Pergamon
References Press, Oxford, 1966, pp. 139–145.
[32] G. Ben-Hamu, A. Eliezer, E.M. Gutman, Electrochim. Acta 52 (2006)
304.
[1] B.L. Mordike, T. Ebert, Mater. Sci. Eng. A 302 (2001) 37.
[33] H. Alves, U. Koster, E. Aghion, D. Eliezer, Mater. Technol. 16 (2001)
[2] E. Ghali, Magnesium and magnesium alloys, in: R.W. Revie (Ed.),
110.
Uhlig’s Corrosion Handbook, John Wiley & Sons, New York, 2000,
[34] J.H. Nordlien, S. Ono, N. Masuko, K. Nisancioglu, Corros. Sci. 39
p. 793.
(1997) 1397.
[3] D. Eliezer, E. Aghion, F.H. Froes, Adv. Perform. Mater. 5 (1998) 201.

You might also like