Anda di halaman 1dari 15

Corrosion Science 50 (2008) 19391953

Contents lists available at ScienceDirect

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

Inuence of the b-phase morphology on the corrosion of the Mg alloy AZ91


Ming-Chun Zhao a,b,c, Ming Liu a,b, Guangling Song a,d, Andrej Atrens a,b,*
a

Division of Materials, The University of Queensland, St. Lucia, Brisbane, Qld 4072, Australia
Swiss Federal Laboratories for Materials Science and Technology, EMPA, Dept 136, berlandstrasse 129, CH-8600 Dubendorf, Switzerland
School of Material Science and Engineering, Central South University, Changsha 410083, China
d
CAST Cooperative Research Centre, Materials Engineering, University of Queensland, Brisbane, Qld 4072, Australia
b
c

a r t i c l e

i n f o

Article history:
Received 20 December 2007
Accepted 11 April 2008
Available online 26 April 2008
Keywords:
Magnesium
Microstructure
Weight loss
Microgalvanic corrosion
Hydrogen evolution

a b s t r a c t
The inuence of the microstructure, particularly the morphology of the b-phase, on the corrosion of Mg
alloys has been studied using AZ91 as a model Mg alloy. The corrosion behaviour was characterized for
ve different types of microstructure produced by heat treatment of as-cast AZ91. The inuence of microstructure can be understood from the interaction of the following three factors: (i) the surface lms can
be more or less effective in hindering corrosion and more or less effective in controlling the form of corrosion as uniform corrosion or localised corrosion, (ii) the second phase (the b-phase in AZ91) can cause
micro-galvanic acceleration of corrosion and (iii) the second phase can act as a corrosion barrier and hinder corrosion propagation in the matrix, if the second phase is in the form of a continuous network. It is
expected that these factors are important for all multi-phase Mg alloys because all known second phases
have corrosion potentials more positive than that of the a-phase. A particular example of the corrosion
barrier effect is provided by the ne (a + b) lamellar micro-constituent; when a b-phase plate nucleates
this micro-constituent, the b-phase plate acts as a corrosion barrier. In contrast, nano-sized b precipitates,
produced by aging, caused micro-galvanic corrosion acceleration of the adjacent a-phase. However, it is
an important nding that the corrosion rate of the a-phase was decreased by the aging treatments that
caused the precipitation of the nano-sized b particles.
2008 Elsevier Ltd. All rights reserved.

1. Introduction
Mg alloys are of signicant interest to the automobile and aerospace industries due to their low densities and adequate strength/
weight ratios [13]. A signicant limitation, however, is their corrosion performance [411]. It is important to understand the factors that inuence their corrosion and to understand the
conditions necessary to achieve adequate corrosion performance.
Their corrosion behaviour has been investigated [1222] over the
past decade in order to facilitate their use in structural applications. Mg alloys are often multi-phase and their corrosion performance is inuenced by their microstructure, in particular the
amount and distribution of the different phases.
AZ91 is one of the most popular of the cast magnesium alloys,
with nominal composition Mg9 wt% Al1 wt% Zn. The AZ91 ascast microstructure has typically a primary a-phase matrix and
a divorced eutectic distributed along the a-phase grain boundaries
[6,14,16,2327]. The divorced eutectic typically consists of large bphase particles and the eutectic a-phase. The eutectic a-phase is

* Corresponding author. Address: Division of Materials, The University of


Queensland, St. Lucia, Brisbane, Qld 4072, Australia. Tel.: +61 733653748; fax:
+61 733653888.
E-mail address: andrejs.atrens@uq.edu.au (A. Atrens).
0010-938X/$ - see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2008.04.010

super-saturated with Al and can transform, by discontinuous precipitation of the b-phase during cooling from the eutectic temperature, to form a ne lamellar arrangement of a + b [2427]. There
have been several studies [14,16,21,28] of the role of microstructure on the corrosion of AZ91. The corrosion behaviour of the
a-phase and the b-phase are the foundations on which to build
an understanding of the inuence of the microstructure of multi-phase alloys like AZ91. Song et al [13,14] have shown that the
free corrosion potential of the b-phase (1.3 V) is 0.3 V more
positive than the free corrosion potential of the a-phase
(1.6 V) in sodium chloride solutions. The a-phase corrodes
due to its very negative free corrosion potential and there is the
tendency for the corrosion rate of the a-phase to be accelerated
by micro-galvanic coupling between the a-phase and the b-phase.
The b-phase, in contrast, has a relatively lower corrosion rate, is a
more efcient site for the cathodic reaction and may act as a barrier against corrosion propagation [6,13,14,29]. Thus, the b-phase
has two different inuences on the corrosion behaviour: the bphase can act as a galvanic cathode to accelerate corrosion and
the b-phase can act as a corrosion barrier to hinder corrosion.
The micro-galvanic corrosion acceleration is dependent on the anode (a)/cathode (b) area ratio whereas the b-phase acts to hinder
corrosion if it is nely divided and continuous. The corrosion
behaviour of AZ91 can be changed by changes in the amount

1940

M.-C. Zhao et al. / Corrosion Science 50 (2008) 19391953

and distribution of the b-phase. Moreover, there is this tendency


for micro-galvanic corrosion in all multi-phase Mg alloys because
all known second phases have corrosion potentials more positive
than that of the a-phase [5].
The microstructure of AZ91, like other cast Mg alloys, is determined by the casting method, the rate of solidication and subsequent heat treatments used to improve mechanical properties. The
distribution, conguration and size of the b-phase can be changed,
which may result in different corrosion behaviour. Our prior work
[21] showed that the second b-phase in an as-cast AZ91 ingot was
in the form of large b particles and within a ne lamellar (a + b) micro-constituent. Homogenisation annealing heat treatment (HA) at
380 C and 410 C can cause the dissolution of the ne lamellar
(a + b) micro-constituent without signicantly altering the coarse
b particles or can also cause the progressive dissolution of the
coarse b particles. Solution heat treatment can dissolve the b-phase
and the produce a microstructure consisting largely of the a-phase.
The a-phase is super-saturated with Al for AZ91 in the solution
heat treated condition and in the homogenisation annealed condition. Subsequent ageing at a lower temperature, e.g. 200 C, precipitates the Al as ne b precipitates [15,30] in the a-phase, thereby
decreasing the Al concentration in the a-phase to a low level
(2% is the equilibrium Al concentration according to the AlMg
phase diagram [31]).
It is important to understand the corrosion behaviour of these
different microstructures. The present work studied the inuence
of microstructure on corrosion using AZ91 as a model Mg alloy
with the aim to build on the foundations of the prior studies
[14,16,21,28]. Five different types of microstructures were produced as follows: (i) the as-cast microstructure of the AZ91 ingot,
(ii) the as-cast microstructure as modied by homogenization
annealing heat treatments at 380 C and 410 C, (iii) solution heat
treatment to produce a largely a-phase microstructure, (iv) the
solution heat treated, largely a-phase microstructure, aged to precipitate ne b precipitates and (v) homogenization annealed conditions aged to precipitate ne b precipitates.
The aim was to carry out a comprehensive evaluation of the corrosion behaviour of all these microstructures using relatively short
immersion tests of 48 h duration during which the rate of corrosion could be continuously monitored and the area corroding could
be characterised at the completion of the 48 h immersion. Immersion tests of longer duration, 96 h, were carried out to determine if
any new phenomena occurred over the longer time span. These
longer term test took more effort and consequently they were carried out for a smaller pallet of alloys, the most important ones were
chosen to check the trends that were becoming evident from the
shorter term tests of 48 h duration. Detailed observations were carried out on the micro-corrosion morphology of a small selected
number of samples. The corrosion of pure Mg has been included
as a reference standard. Pure Mg has a homogenous single-phase
microstructure consisting only of the a-phase. This comprehensive
investigation into the inuence of the microstructure on the corrosion behaviour of AZ91 has been carried out to provide a better
understanding of the factors that inuence corrosion of two-phase
Mg alloys like AZ91.

2. Experimental procedure
2.1. Materials
The AZ91D
0.14 wt% Mn,
<0.002 wt% Cu
Mg had the
0.008 wt% Mn,

had the composition of 8.26 wt% Al, 0.69 wt% Zn,


<0.002 wt% Fe, 0.002 wt% Ni, <0.001 wt% Cr,
<0.002 wt% Zr and balance Mg. The reference pure
composition of 99.94 wt% Mg, 0.0137 wt% Si,
0.0045 wt% Fe, 0.00092 wt% Ni and 0.0066 wt% Al.

These two Mg alloys are high purity because they both contain
low levels of the impurity elements Fe, Ni and Cu, so the results
of the present research should be directly comparable with the
prior studies [5,6,1316].
The AZ91 specimens were cut from an as-cast ingot and were
subjected to heat treatment as follows. Homogenization annealing
(HA) heat treatments of as-cast specimens were carried out at
380 C and 410 C for 125 h followed in each case by air cooling
(AC); the resulting specimens are designed as HA3805, HA3810,
HA4105, HA4110 and HA4125, where the rst two numbers indicate the temperature of the HA heat treatment (380 C and
410 C) and the second two numbers designate the time in hours
of the HA heat treatment. The solid solution (SS) heat treatment
consisted of heat-treating as-cast specimens at 410 C for 100 h
followed by water quenching (WQ), designed as 4100S (or SS condition). Solid solution treatment plus ageing (SA) consisted of the
solid solution heat treatment followed by an ageing heat treatment
at 200 C for 5, 10, 24 and 48 h; the full designation was
4100S205A, 4100S210A, 4100S224A and 4100S248A, respectively;
the short designation was SA205, SA210, SA224 and SA248. HA and
aging was carried out to produce 4125HA225A (HA for 25 h at
410 C, AC, aging for 25 h at 200 C), 3801S205A (HA for 1 h at
380 C, WC, aging for 5 h at 200 C) and 3810S205A (HA for 10 h
at 380 C, WC, aging for 5 h at 200 C).
The microstructure was examined by optical microscopy and
scanning electron microscopy (SEM) after metallographic preparation by mechanical grinding successively to 1200 grit SiC paper,
polishing successively to 0.5 lm diamond, washing with distilled
water, drying with warm owing air and etching in 3% nital.
X-ray diffraction (XRD) using CuKa radiation was used to characterize the phases present in these samples.
2.2. Corrosion evaluation
The corrosion behaviour was evaluated using separate immersion tests of duration of either 48 h or 96 h, at room temperature,
in 1 N NaCl aqueous solution, which was made with analytical
grade reagent and distilled water. The corrosion rate was evaluated
by measuring (i) the evolved hydrogen during corrosion in the 1 N
NaCl solution and (ii) the weight lost by the specimen. The AZ91
specimens were encapsulated in epoxy resin so that a surface, with
dimension 18 mm  27 mm, was exposed to the solution. The
working surface was mechanically ground to 1200 grit SiC paper,
washed with distilled water, dried with warm owing air, dried
in a desiccator for 1 to 2 days and weighed, to give the specimen
weight before exposure, Wb. The specimen was horizontally immersed in 1500 ml of test solution and the hydrogen evolved during the corrosion experiment was collected in a burette above the
corroding sample. The evolved hydrogen is a direct measure of the
corrosion rate [46,3233], as, in the overall magnesium corrosion
reaction

Mg + H + H2 O = Mg2 + OH + H2

one molecule of hydrogen is evolved for each atom of corroded


magnesium. After the immersion test, the corrosion products were
removed by immersion at room temperature for 510 min in a
chromic acid cleaning solution, the sample was washed with
distilled water, dried with warm owing air, dried in a desiccator
for 12 days and weighted to determine the sample weight
after the immersion test, Wa. The weight loss data is presented as
(weight loss = Wb  Wa [mg])/(specimen area [cm])/(exposure
time [d]). The chromic acid cleaning solution composition was
(200 g CrO3 + 10 g AgNO3)/L; previous work has shown that this
chemical cleaning solution causes almost no weight loss for noncorroded AZ alloy specimens [16]. Separate blank experiments

M.-C. Zhao et al. / Corrosion Science 50 (2008) 19391953

showed that there was negligible weight change of the encapsulating epoxy resin by similar exposures to the 1 N NaCl solution or to
the acid cleaning solution. The hydrogen evolution rate can be related to the weight loss rate, and the weight loss rate can be related
to the corrosion rate, using the following conversions [5,6,20,34
36]:

Weight loss rate mg=cm2 =d 1:085


2

Hydrogen evolution rate ml=cm =d;

2
2

Corrosion rate mm=y 2:10 weight loss rate mg=cm =d: 3


The corrosion macro-morphology was examined using optical
microscopy after the completion of each immersion test; this examination of the corrosion macro-morphology included recording the
macro-morphology using macro-photography.
Corrosion micro-morphology for the as-cast condition and representative HA conditions was examined using optical microscopy
and using another series of metallographic prepared specimens,
which were removed from the 1 N NaCl solution after various
immersion times. After removal from the test solution, these specimens were rinsed with distilled water, dried using owing air and
observed using optical microscopy.
3. Results
3.1. Microstructure-as-cast
The as-cast AZ91 microstructure had some micro-porosity as
illustrated in Fig. 1a for the polished, un-etched surface. The dark
spots in the low-magnication optical micrograph correspond to
micro-pores. The occurrence of micro-porosity in as-cast AZ91 is
consistent with [14]. It is possible to discern the microstructure
as the a-Mg matrix with the second b-phase, because the difference in the hardness between the a-phase and the b-phase causes
a difference in polishing rates.
The detailed microstructure for as-cast AZ91, as revealed by
etching, is presented in the SEM micrographs (Fig. 1a and b). The
microstructure, as illustrated at low magnication in Fig. 1a, contained some dark regions in the vicinity of grain boundaries. These
regions are identied as the eutectic a-phase with an Al content
higher than the primary a-phase [14]; the dark appearance is
attributed to different etching by the nital. The microstructure also
contained large eutectic b-phase particles and an (a + b) microconstituent in a ne lamellar arrangement as illustrated at higher
magnication in Fig. 1b. The ne lamellar (a + b) micro-constituent

1941

has the same outline as the eutectic a-phase and is the result of the
discontinuous precipitation of the b-phase. These ne lamellar
(a+b) micro-constituents and associated large b particles were
interconnected and formed an interconnected network throughout
the microstructure. In addition, the remainder of the microstructure contained large eutectic b-phase particles surrounded by the
ne lamellar (a + b) micro-constituent as isolated entities in the
a-phase matrix.
The microstructure provides clues to the mechanism of formation of the ne lamellar (a + b) micro-constituent in the as-cast
AZ91 microstructure. The ne lamellar (a + b) micro-constituent
had a morphology similar to the typical lamellar pearlite colony
in steel and hence the formation mechanism may be similar [37
39]. There were the cases where it appeared that there was a b
plate at the edge of the ne lamellar (a + b) micro-constituent
and it appeared that this b plate had nucleated the ne lamellar
(a + b) micro-constituent in a mechanism that was similar to the
nucleation of lamellar pearlite in steels as illustrated in Fig. 2
[3739]. The formation of the ne lamellar (a + b) micro-constituent occurs after solidication is complete and it occurs by a solidstate transformation of an a-Mg phase super-saturated in Al. During the transformation, Al is rejected from the a-Mg. If the Al diffusion is not sufciently fast, Al accumulates at the interface and
a b-plate forms when the Al content reaches a critical value. Therefore, the rejection of Al from the super-saturated a-Mg in the ascast AZ91 can lead to the formation of an interface b plate that
can nucleate the ne lamellar (a + b) micro-constituent. After
nucleation, the ne lamellar (a + b) micro-constituent grows due
to the slow diffusion of Al in the Al-rich matrix. This proposed formation mechanism for the ne lamellar (a + b) micro-constituent
is also consistent with instances [16] where the ne lamellar
(a + b) micro-constituent forms when die-cast AZ91 is aged at a
low temperature, e.g. 200 C. In Mg die-casting, molten metal is
rapidly injected into a steel die, solidication is relatively rapid
and the formation of the ne lamellar (a+b) micro-constituent is
prevented. During ageing of die-cast AZ91 at a low temperature,
the ne lamellar (a + b) micro-constituent nucleates from large b
particles. Therefore, it is reasonable that the ne lamellar (a + b)
micro-constituent nucleates from a large b particle and grows
away from the b particle as shown in Fig. 2 for the nucleation
and growth of pearlite in steels. Fig. 2 presents the mechanism of
transformation of austentite (c) to pearlite (the eutectoid microconstituent containing ferrite (a) and cementite in a lamellar
arrangement). The pearlite is nucleated by a cementite plate at
the nucleating grain boundary [3739].

Fig. 1. The microstructure of as-cast AZ91.

1942

M.-C. Zhao et al. / Corrosion Science 50 (2008) 19391953

homogenization annealing temperature and/or a longer time produced a more homogeneous distribution of the b-phase throughout the microstructure.
3.3. Microstructure-solid solution (SS)
The microstructure in the solid solution (SS) condition was
homogenous, largely single a-phase and the b-phase was almost
completely dissolved as illustrated in Fig. 4a.
3.4. Microstructure-solution heat treated and aged (SA)

Fig. 2. Pearlite nucleation at a cementite plate during the transformation of austenite (c) to ferrite (a) in steel. Pearlite is a lamellar arrangement of cementite
(carbide, Fe3C, grey) and ferrite (a, white). As the austenite (c) transforms to ferrite
(a), carbon is rejected from the ferrite and consequently the carbon concentration
builds up at the a/c interface until a carbide forms at the interface. Subsequently
pearlite grows from this carbide [3739].

Furthermore, in the as-cast AZ91 microstructure, there were


different orientations for the packets of the ne lamellar (a + b) micro-constituent, as shown in Fig. 1b, due to multiple nuclei.
3.2. Microstructure-homogenization anneal (HA)
Homogenization annealing (HA) changed the as-cast AZ91
microstructure. The volume fraction of the b-phase decreased
and its distribution became more homogeneous with an increase
of the homogenization annealing temperature and/or homogenization annealing time. Fig. 3a shows that, after HA for 5 h at
380 C (HA3805), most of the ne lamellar (a + b) micro-constituent had dissolved but the large b particles were largely unchanged. The microstructure consisted of the a-Mg matrix and
the isolated large b particles associated with a small amount of
the ne (a + b) lamellar micro-constituent remaining at the
a-Mg grain boundaries. Fig. 3b shows that, after HA for 10 h at
380 C (HA3810), the ne lamellar (a + b) micro-constituent had
completely dissolved with little change of the large b particles;
the microstructure consisted of the a-Mg matrix and isolated
large b particles. Fig. 3c shows that, after HA for 5 h at 410 C
(HA4105), the ne lamellar (a + b) micro-constituent had completely dissolved and the large b particles had partly dissolved,
showing a microstructure similar to that of HA3810, but the large
b particle in HA 4105 were smaller in size due to part dissolution.
Fig. 3d shows that, after HA for 10 h at 410 C (HA4110), the ne
(a + b) lamellar micro-constituent had completely dissolved and
most of the b particles had dissolved, leaving relatively few ne
b particles homogeneously distributed throughout the matrix.
Fig. 3e shows that, after HA for 25 h at 410 C (HA4125) the microstructure was similar to that of HA4110, the ne (a + b) lamellar
micro-constituent had mostly dissolved and most of the b particles had dissolved, leaving relatively few ne b particles homogeneously distributed throughout the matrix and a trace along the
a-phase grain boundaries of the ne (a + b) in the lamellar
arrangement.
HA3805 and HA3810 caused small changes to the large b particles and mainly caused the dissolution of the lamellar (a + b) micro-constituent. In contrast, HA4105 and HA4110 caused some
dissolution of the large b particles and dissolution of the ne
lamellar (a + b) micro-constituent. This indicates that a higher

The solid solution and aged microstructures, Fig. 4be, consisted of the a-Mg matrix with numerous nano-sized b precipitates, which were different in morphology to the b-phase
particles in the as-cast and the HA conditions. The b precipitates
in the solution treated and aged (SA) conditions where needle like,
had an average size of 100 nm in width and had a relatively
homogenous distribution throughout the microstructure. Furthermore, the volume fraction of b precipitates increased noticeably
during aging and their size grew somewhat with an increase of
the ageing time as shown in Fig. 4be. Nevertheless, all the b precipitates where needle like with most of the needles oriented in the
same direction in each grain examined.
3.5. Microstructure-homogenization anneal and aged (SA)
The 4125HA225A microstructure (Fig. 4f, HA for 25 h at 410 C,
AC, aging for 25 h at 200 C) was somewhat similar to SA224
(Fig. 4d). The 4125HA225 microstructure consisted of the a-Mg
matrix with numerous needle-shaped b precipitates, size of
100 nm in width, but in 4125HA225 the needles were oriented
in a number of distinct directions compared with SA224 wherein
the needles were largely oriented in the same direction.
For the 3801S205A microstructure, Fig. 4g, the HA for 1 h at
380 C did not cause a large change to the as-cast microstructure;
the microstructure contained large eutectic b particles and the ne
lamellar (a + b) micro-constituent similar to the as-cast structure.
There were, in addition, numerous needle-shaped b precipitates,
size of 100 nm in width, with an appearance of being oriented
in a number of distinct directions. These were precipitated from
the super-saturated a-phase solid solution during the aging treatment (5 h at 200 C).
For the 3810S205A microstructure, Fig. 4h, it would be expected
that the HA for 10 h at 380 C would have dissolved most of the
ne lamellar (a + b) micro-constituent and would have dissolved
some of the large b particles; however the microstructure, Fig. 4h
did consist of the a-Mg matrix, large b particles and a signicant
amount of the ne lamellar (a + b) micro-constituent; the ne
lamellar (a + b) micro-constituent might have reformed during
the air cooling from the HA treatment temperature or during
the aging treatment. In addition, there were, numerous ne interconnected b precipitates, size of 100 nm in width, with some
appearance of being oriented. These were precipitated from the
super-saturated a-phase solid solution during the aging treatment
(5 h at 200 C).
3.6. XRD
The XRD spectra of specimens in the as-cast condition, representative HA conditions and a representative SA condition, Fig. 5,
indicated that the second b-phase particles were Mg17Al12 because
all of peaks in the XRD spectra corresponded to either Mg or
Mg17Al12, although the second b-phase was reported to sometimes
be present as Mg4Al3 [4041]. This indicates that AZ91 had a matrix of a-Mg grains with the second b-phase consisting of the inter-

M.-C. Zhao et al. / Corrosion Science 50 (2008) 19391953

1943

Fig. 3. The microstructure after homogenization annealing (HA).

metallic Mg17Al12 for the as-cast condition and for the various heat
treatments.
Fig. 5 also indicates that the relative peak heights for the peaks
associated with a-Mg were different for the different heat treated
conditions. The largest difference was that between the XRD spectrum for the as-cast condition and the other spectra. This difference is associated with the changes in the microstructure. In the
as-cast condition, there was three forms of a-Mg: (i) primary a,
(ii) eutectic a-phase and (iii) a-phase in the ne a + b lamellar
arrangement. In contrast, after heat treatment, there was largely
only the primary a-phase left, as the other micro-constituents
had dissolved during the heat treatment.

3.7. Hydrogen evolution


Fig. 6 presents the corrosion rate as measured by hydrogen evolution. The measured values for pure Mg were in good agreement
with prior measurements [5,6,14] as were the measurements for
as-cast AZ91D [5,6,14,15]. This provides condence in the validity
of the data.
Most specimens exhibited an increase in hydrogen evolution
rate with increasing immersion time. In contrast, the hydrogen
evolution volume for pure Mg, also depicted in Fig. 6, increased linearly with exposure time. The increasing hydrogen evolution
meant that the evolved volumes during the 96 h immersions

1944

M.-C. Zhao et al. / Corrosion Science 50 (2008) 19391953

Fig. 4. The microstructures of SS and aged conditions.

1945

M.-C. Zhao et al. / Corrosion Science 50 (2008) 19391953


2000

2000

HA3810

Intensity

Intensity

as-cast

0
2000

20

30

40

50

60

70

80

0
90 2000

Two-Theta

20

30

40

4100S248A

60

70

80

90

70

80

90

Intensity

Intensity

HA4110

50

Two-Theta

0
20

30

40

50

60

70

80

90

20

30

Two-Theta

40

50

60

Two-Theta

Fig. 5. XRD patterns of specimens in the as-cast condition, representative HA conditions and a representative SA condition. The peaks marked with a diamond are from
Mg17Al12, all other peaks are from a-Mg.

were signicantly larger for the 46 h immersion; compare Fig. 6a


with b.
The data of hydrogen evolution volumes are presented in two
groups. Group A includes the as-cast condition, the HA conditions
and the solid solution condition. Group B includes the solid solution condition, the aged conditions and pure Mg. Group A had
higher hydrogen volumes. For the 48 h immersion tests for Group
A, the highest hydrogen evolution volume was for the AZ91D specimen HA for 10 h at 380 C and the lowest hydrogen evolution volume was for the solid solution condition. The hydrogen evolution
volume of all the group A conditions in Fig. 6a can be ranked as a
decreasing series: HA for 10 h at 380 C > HA for 5 h at
380 C > HA for 5 h at 410 C > HA for 10 h at 410 C  HA for
25 h at 410 C > the as-cast condition > the solid solution condition. For the 48 h immersion test for Group B, the highest hydrogen
evolution volume was for the specimen SA for 48 h at 200 C and
the lowest hydrogen evolution volume was for pure Mg. The
hydrogen evolution volume for all the group B conditions in
Fig. 6a can be ranked as a decreasing series: SA for 48 h at
200 C > the solid solution condition > SA for 24 h at 200 C > HA
for 25 h at 410 C + aged for 25 h at 200 C  SA for 10 h at
200 C > SA for 5 h at 200 C > HA for 1 h at 380 C + aged for 5 h
at 200 C  HA for 10 h at 380 C + aged for 5 h at 200 C > pure
Mg. Similar results were obtained for 96 h immersion test for
Groups A and B, that is, the hydrogen evolution volume of all the
group A conditions in Fig. 6b can be ranked as a decreasing series:
HA for 10 h at 380 C > HA for 5 h at 380 C > the solid solution condition > the as-cast condition. The hydrogen evolution volume of
all the group B conditions in Fig. 6b can be ranked as a decreasing

series: the solid solution condition > SA for 48 h at 200 C > SA for
24 h at 200 C > SA for 10 h at 200 C > SA for 5 h at 200 C > pure
Mg. However, there an exception for the solid solution condition:
(i) the hydrogen evolution volume of the solid solution condition
was a little lower than the as-cast condition until about 60 h
immersion and thereafter there was a reverse tendency, and (ii)
the hydrogen evolution volume of the solid solution condition
was a little lower than the SA for 48 h at 200 C until about 80 h
immersion and thereafter there is a reverse tendency.
3.8. Weight loss data
Fig. 7 presents weight loss measurements. Fig. 7a presents the
weight loss data for 48 h immersion and Fig. 7b presents the data
for 96 h immersion. The corrosion rates were signicantly higher
in the 96 h immersions; this is consistent with the accelerating
corrosion rate as shown in Fig. 6. Furthermore, the trends revealed
in the hydrogen evolution data of Fig. 6 are reproduced in the
weight loss measurements presented in Fig. 7.
It appeared that the ne lamellar a + b micro-constituent in ascast AZ91 was somewhat benecial to corrosion. This is proposed
because the as-cast AZ91, with a microstructure containing the ne
lamellar a + b micro-constituent, had a lower hydrogen evolution
rate than all the HA conditions without the lamellar a + b microconstituent in the microstructure, Figs. 6 and 7. Also the HA3805
condition (HA for 5 h at 380 C) had a lower hydrogen evolution
rate than the HA3810 condition (HA for 10 h at 380 C); for these
two conditions, the only microstructure difference was that
HA3805 condition had a small amount of the ne lamellar (a + b)

1946

M.-C. Zhao et al. / Corrosion Science 50 (2008) 19391953


16
as-cast
3805-H
3810-H
4105-H
4110-H
4100-S
4125-H

Group A

10
8
6
4
2

60
50
40
30
20
10

0
20

4100-S
4100S-205
4100S-210
4100S-224
4100S-248
pure Mg
4125HA-225
3801S-205
3810S-205

30

4
0

0
40 0

0
5

Group B

35
2

10

H2 evolution volume, ml/cm

H2 evolution volume, ml/cm

as-cast
3805-H
3810-H
4100-S

70

H2 evolution volume, ml/cm

12

80

Group A

ev

H2 evolution volume, ml/cm

14

30

4100-S20
4100S-205
4100S-210
4100S-224
4100S-248
pure Mg

40

60

80

100

40

60

80

100

40

60

80

100

Group B

25
20
15
10
5
0

0
0

10

20

30

4100-S
pure Mg

4
0

0
5

40

4100-S20
pure Mg

20

H2 evolution volume, ml/cm

H2 evolution volume, ml/cm

35
4

30
25
20
15
10
5
0

0
0

10

20

30

40

50

immersion time, h

(a) 48 h immersion

immersion time, h

(b) 96 h immersion

Fig. 6. Hydrogen evolution for the various microstructures during immersion in 1 N NaCl for the stated period.

micro-constituent in the a-Mg grain boundary whereas the


HA3810 condition did not.
The weight loss rate is directly related to the hydrogen evolution rate as shown in Fig. 8. This direct correlation is expected from
the overall magnesium corrosion reaction (1) that one molecule of
hydrogen is evolved for each atom of corroded magnesium and
from prior research [46,8,32,33,36]. Fig. 8 provides a cross plot
of the independent measurements of weight loss rate and hydrogen evolution rate, for the 48 h immersion tests and for the 96
immersion tests. The line in Fig. 8 is the plot of Eq. (2). Fig. 6 shows
that (i) the weight loss rate data is consistent with the hydrogen
evolution rate data and (ii) that the relationship between these
independent, experimental measurements follows the theoretical
expectation from the corrosion reaction (1) as embodied in Eq.
(2). This provides condence in the experimental measurements,
including condence (i) that the acid cleaning procedure has removed all the corrosion products, (ii) that the acid cleaning proce-

dure has removed negligible Mg metal and (iii) that there was
negligible weight change of the encapsulating epoxy resin due to
exposure to the 1 N NaCl solution or to the acid cleaning solution.
3.9. Macroscopic corrosion morphology
The corrosion of AZ91 initiated as localized corrosion at some
sites on the surface and subsequently expanded over surface. The
advance of the corrosion over the surface was different for different
conditions. At the end of the 48 or 96 h immersion time, the active
corrosion area was covered by a thick layer of corrosion products
and was different for the various conditions as presented by the
macro-photographs (27 mm  18 mm) in Figs. 9 and 10. In most
cases, the area corroding after 96 h immersion was larger than
after 48 h immersion.
In marked contrast to the local corrosion of AZ91, the corrosion
for pure Mg immersed in 1 N NaCl was uniform corrosion; there

M.-C. Zhao et al. / Corrosion Science 50 (2008) 19391953


8
HA3810

48 h immersion

Weight loss rate (mg/cm day)

HA3805

6
5
HA4105

4
HA4110 HA4125

SA248

as-cast
SS

2
SA210

SA224

4125HA225

SA205

3801S205
3810S205

pure Mg

0
1

Weight loss rate (mg/cm day)

20

4HA3810
5

10

11

13 14 15
9612h immersion

15
HA3805
SS

10

SA248
as-cast
SA224

5
SA205

SA210

pure Mg

0
1

Fig. 7. Weight loss data for the various microstructures during immersion in 1 N
NaCl for the stated period.

48 h
96 h
Y=1.085X

Weight loss rate (mg/cm day)

20

15

10

0
0

10

20

30

1947

set of samples in the as-cast and HA conditions exposed to the test


solution. After a few minutes immersion in the test solution, some
small hydrogen bubbles were evolved from all over the whole
working surface. After a longer immersion time (1 h), several sites
suffered corrosion in the form of localized corrosion. The corrosion
in these local areas enlarged in area slowly and a large number of
hydrogen bubbles were evolved from these local areas of corrosion.
Fig. 11a presents the micro-corrosion morphology for the as-cast
condition after immersion in the test solution for 16 h; the samples
had the same surface preparation as the samples for the hydrogen
gas collection. Optical microscope examination led to the conclusion that the ne lamellar a + b micro-constituent and the coarse
b particles provided some action as corrosion barriers; the adjacent
a-Mg matrix had undergone the majority of the corrosion, with
some corrosion in the ne lamellar a + b micro-constituent. The
same surface preparation was used as was used in the corrosion
exposures with hydrogen collection, with the necessary consequence that the quality of the optical micrographs could not be expected to equal that of a metallographic prepared sample. To
obtain good quality micrographs the corrosion exposures were repeated with samples that were metallographic polished but were
not etched; ground successively to 1,200 grit SiC paper, polished
successively to 0.5 lm diamond, washed with distilled water and
dried with warm owing air. The corrosion morphologies for these
metallographic polished samples were essentially the same as for
the samples subjected to only grinding to 1200 SiC paper. Optical
microscope examination of the metallographic polished samples
exposed in 1 N NaCl is presented in Fig. 11b. Fig. 11a and b show
the following three different corrosion cases for the as-cast condition: (i) corrosion of the adjacent a-Mg with no signicant corrosion of the ne lamellar a + b micro-constituent and the coarse b
particles (ii) corrosion of the a-Mg up to the ne lamellar a + b micro-constituent, with the appearance of the corrosion of the a-Mg
stopped at the interface of the ne lamellar a + b micro-constituent
and (iii) corrosion proceeding into the ne lamellar a + b microconstituent, with both phases having been corroded to a similar
extent.
In marked contrast, the HA conditions had no continuous micro-constituents and had only isolated b particles as shown in
Fig. 3. Fig. 12 presents optical microscope photos of the metallographic polished samples exposed to 1 N NaCl for (HA for 5 h at
380 C) and (HA for 10 h at 410 C). Some isolated b particles were
in the forefront of the corrosion advance. However, the distance
between the b particles in these two microstructures was large,
and the corrosion advance was not at all blocked by the b-phase
particles, although the b particles were themselves not signicantly corroded.

4. Discussion

H2 evolution volume, ml/cm day


Fig. 8. Cross plot of the independent measurements of weight loss rate and hydrogen evolution rate, for 48 h and 96 h immersion tests. The line is a plot of Eq (2).

were no preferential sites for corrosion. At the end of the 48 and


96 h immersion periods, the whole exposed surface of pure Mg
was homogenously covered by a layer of corrosion products. The
fact that the whole surface of pure Mg was corroding at the end
of both the 48 and 96 h exposure periods is consistent with the linear rate of hydrogen evolution presented in Fig. 6.
3.10. Micro-corrosion morphology
The inuence of the microstructure on the corrosion behaviour
was studied by observing the micro-corrosion morphologies of a

4.1. Macroscopic corrosion morphology


The corrosion of AZ91 in the various heat treated conditions
was localised corrosion, initiated at some sites on the surface and
subsequently expanded over surface. The advance of the corrosion
over the surface was different for different conditions, as presented
by the macro-photographs (27 mm  18 mm) in Figs. 9 and 10. In
most cases, the area corroding after 96 h immersion was larger
than after 48 h immersion. This provides part of the explanation
for (i) the increasing rate of evolved hydrogen with immersion
time (Fig. 6) (ii) the greater evolved hydrogen volume after 96 h
immersion (Fig. 6) and the greater weight loss after 96 h immersion (Fig. 7). The fact that there was still un-corroded surface after
96 h immersion indicates (i) that the air formed lm provides a
certain resistance to corrosion initiation, (ii) that the air formed

1948

M.-C. Zhao et al. / Corrosion Science 50 (2008) 19391953

Fig. 9. Macro-appearance (27 mm  18 mm) of the corrosion morphology after 48 h immersion for different microstructures.

lm provides a certain resistance to corrosion propagation and (iii)


that steady state corrosion conditions had not been reached. In
marked contrast to the local corrosion of AZ91, the corrosion for
pure Mg immersed in 1 N NaCl was uniform corrosion and there
was a linear rate of hydrogen evolution, Fig. 6.
4.2. Corrosion mechanism homogenization anneal
Figs. 9 and 10 indicate that the corroded area was approximately
the same for all the homogenization annealed conditions after both
48 h and 96 h immersion, although the corroded area after 96 h
immersion was greater than after 48 h immersion. As it is possible
to exclude the inuence of the surface area, surface state and the
initiation of corrosion, it is possible to consider the inuence of
the microstructure on the propagation of corrosion as presented
in Figs. 6 and 7. AZ91 is a two-phase alloy, in which different micro-constituents may each form a micro-cell. The corrosion behaviour and morphology of different microstructure conditions can be
explained on the basis of the microstructure characteristics and the
corresponding corrosion performance of the a-Mg matrix, the ne
lamellar a+b micro-constituent, coarse b particles and ne b particles. When the samples are immersed in the test solution, hydrogen
evolution preferentially starts at some sites and then the corrosion
attack advances. The b-phase can have two roles in corrosion: (i) as
a corrosion barrier and (ii) as a galvanic cathode. The b-phase itself
is more stable in the test solution and it has a lower corrosion rate.
However, the cathodic hydrogen evolution on the b-phase surface is

much faster than that on the surface of the a-phase and thus the bphase is a more effective cathode [14]. A competition between micro-galvanic corrosion and the corrosion barrier effect can explain
the corrosion [14]. If the b-phase is present as a small fraction,
the b-phase serves mainly as a galvanic cathode and accelerates the
overall corrosion of the a matrix. If b-phase fraction is high, the bphase may act mainly as a barrier against the corrosion of the a matrix. In the as-cast condition, Fig. 11 shows that the
a-Mg matrix adjacent to coarse b particles with lack of the surrounding lamellae (a + b) eutectic has undergone the majority of
the corrosion, with there being less corrosion of the b particles
and the ne lamellar a + b micro-constituent. The corrosion is initiated and gradually advanced by the dissolution of the a-Mg matrix
adjacent to the region of the continuous lamellar micro-constituent
plus coarse b particles. When the propagation of the corrosion attack reaches the region of the continuous (a + b) micro-constituent
plus coarse b particles, the corrosion is retarded to a certain extent
as shown in Fig. 11. In this case, even though the a-Mg matrix surrounding the (a + b) plus coarse b particles is corroded, some should
be under the continuous lamellae (a + b) plus coarse b particles as
the a-Mg matrix in the top layer has corroded. In fact, the corrosion
damage as above-mentioned is quite common for AZ91D in a corrosive environment. There is evidence for some b particles being
undermined from corroding areas of AZ91D under an atmospheric
corrosion [28].
Section 3.1 evaluated the formation of the ne lamellar (a + b)
micro-constituent. This evaluation led to the realisation that a

M.-C. Zhao et al. / Corrosion Science 50 (2008) 19391953

1949

Fig. 9 (continued)

plate of the b-phase could be the cause of the nucleation of the ne


(a + b) lamellae micro-constituent. Such b-phase plates could be
the mechanism by which corrosion is stopped by the ne (a + b)
lamellae micro-constituent. It is proposed that, when a b-phase
plate nucleates this micro-constituent, the b-phase plate acts as a
corrosion barrier
In the 380 C HA conditions (5 h and 10 h), isolated b particles
are distributed throughout the microstructure, the distance between the isolated b particles is large and the corrosion attack
is not effectively blocked by this type of isolated b particles. However, the corrosion attack of the a-phase is accelerated by the micro-galvanic coupling with the b-phase. This means that the a-Mg
matrix and the isolated coarse b particles in the 380 C HA conditions (5 h and 10 h) undergo micro-galvanic corrosion in addition
to the spontaneous dissolution of the a-Mg in the aqueous solution (owing to its very negative open circuit potential). The HA
condition at 10 h for 380 C shows marked micro-galvanic corrosion due to complete elimination of lamellar eutectic (a + b) in
the microstructure leaving only the isolated b particles. On the
other hand, the predominant factors determining the rate of galvanic corrosion include the anode-to-cathode area ratios. The
HA treatment for 5 h at 410 C had caused some dissolution of
the large b particles and the size of these b particles was smaller
compared to the HA treatment for 10 h at 380 C, therefore, the
galvanic corrosion effect was decreased. After the HA treatment
for 10 h at 410 C and the HA treatment for 25 h at 410 C, the

microstructure contained only some ne isolated b particles, such


that the smaller cathode area caused less galvanic corrosion to
accelerate the spontaneous corrosion of the a-Mg. Similar corrosions rates are derived by the similar microstructures for the HA
treatment for 10 h at 410 C and the HA treatment for 25 h at
410 C.
The above analysis indicates that the b-phase acts mainly as a
galvanic cathode for the various HA conditions during the initial
period of corrosion, as the b fraction is low. During corrosion, the
a grains dissolve preferentially whereas most of the b-phase particles remain except for isolated b-phase particles which are undermined and which fall out. Hence the b fraction on the sample
surface cannot remain constant during corrosion. It changes from
the initial b surface fraction characteristic of the sample surface
to a steady state surface fraction determined by the preferential
corrosion of the a grains and the undermining of the b-phase particles. As a consequence there is the phenomenon that the hydrogen evolution volume for the as-cast condition with continuous
(a + b) plus b-phase in the microstructure is slightly higher than
that for the solid solution condition with homogenous single a
phase in the microstructure in the beginning after 48 h immersion,
but become slightly lower with the advance of the corrosion attack
at 96 h immersion, as shown in Fig. 6 (compare as-cast and
4100-S); this is attributed to an increased surface coverage by
the b-phase for the as-cast condition and a higher barrier effect
by the b-phase in the area that is corroding. This barrier effect

1950

M.-C. Zhao et al. / Corrosion Science 50 (2008) 19391953

Fig. 10. Macro-appearance (27 mm  18 mm) of the corrosion morphology after 96 h immersion for different microstructures.

Fig. 11. Micro-corrosion morphologies after immersion for as-cast condition.

M.-C. Zhao et al. / Corrosion Science 50 (2008) 19391953

1951

Fig. 12. Typical corrosion morphology for HA3805 and HA4110 conditions for metallographic polished specimens; arrows indicate isolated ne b particles.

explains the change in relative corrosion rates of the as-cast and SS


conditions; the increased area of corrosion explains why their increased absolute corrosion rates after 96 h immersion are both
higher than after 48 h immersion.
There was a similar effect between 4100-S and 4100-248 samples, Fig. 6, and a similar explanation is proposed.
4.3. Corrosion mechanism solid solution
Many studies [12,16,29] have reported that there is a decrease
in corrosion rate for MgAl alloys with an increasing Al content
in the range 211% Al in a-Mg. These studies are consistent. However, comparison of the corrosion rates for MgAl alloys from the
work of Song et al [16] indicate that the corrosion rates for the
MgAl alloys are all greater than the corrosion rate for pure Mg
as measured by Song et al [13,14]. The data in the present work
are consistent with these prior measurements [1214,16,29]. Figs.
6 and 7 shows that the solid solution condition of AZ91 had a corrosion rate higher than that of pure Mg; this comparison is particularly emphasized by Fig. 6. The solid solution condition of AZ91,
with a homogenous single phase, has a higher Al content than pure
Mg with a homogenous single phase and had a higher corrosion
rate. This implies that the corrosion behaviour of pure Mg is characteristically different from that of AZ91. Figs. 9 and 10 shows that
the corrosion was uniform for pure Mg immersed in 1 N NaCl and
that there were no preferential sites for corrosion, i.e. the whole
exposed surface of pure Mg was homogenously covered by a layer
of corrosion lm. This was markedly different from solution treated AZ91 which showed local corrosion. This may explain the
observation that the solid solution condition of AZ91 has a corrosion rate higher than that of pure Mg.
The steady state surface lm on pure Mg does not hinder further corrosion as the corrosion rate is constant with exposure time,
Fig. 6 whereas the prior work indicates that the lm on MgAl alloys is partially protective, because the corrosion rate decreases
with increasing Al content for Mg-Al alloys heat treated to be
homogeneous single a-phase alloys. However, the lm on the solid
solution treated AZ91 (4100-S) allowed localised corrosion, and the
corrosion rate increased with immersion time, Fig. 6.
4.4. Corrosion mechanism aged conditions
The corrosion of the aged samples SA205, SA210, SA224 and
SA248 is consistent with the increased micro-galvanic corrosion
acceleration of an a-phase that has an intrinsic corrosion rate signicantly lower than that of the SS condition and approaching that
of pure Mg. The aged samples SA205, SA210, SA224 and SA248
were aged for 5, 10, 24 and 48 h at 200 C. Increased aging time increased the amount of the b-phase, Fig. 4be which would provide

increased micro-galvanic corrosion acceleration. Furthermore, for


these samples, there was an additional effect due to the fact that
the area corroded increased in the sequence SA205, SA210,
SA224 and SA248 for the 48 h immersions, Figs. 9 and 10 shows
a similar tendency after 96 h immersion. The increased measured
weight loss in the sequence SA205, SA210, SA224 and SA248 can
be attributed to a combination of these two effects: (i) the increased micro-galvanic corrosion directly increased the corrosion
rate and (ii) the greater surface area corroding means that the sample has a greater weight loss. However, it is not entirely clear why
the a-phase would have a corrosion rate signicantly lower than
that of the SS condition and approaching that of pure Mg, particularly as aging would be expected to decrease the Al content of the
a-phase to 2% if equilibrium is approached as predicted by the
phase diagram [31].
The microstructure for 4125HA225A is very similar to that of
SA224, so it is to be expected that they have similar corrosion rates.
Similarly, the microstructures of 3801S205 and 3810S205 are similar as are their corrosion rates. Their low corrosion rates are attributed to the fact that their microstructures contain a continuous
network of the b-phase, which would be expected to act as a barrier to corrosion propagation.
4.5. Longer immersion tests
The longer-term immersion tests for 96 h, Figs. 6, 7 and 10,
exhibited the same trends and corrosion mechanism as shown
by the shorter 48 h immersion tests. This provided condence that
the deductions were indeed valid that were being drawn from the
short tests.
4.6. Corrosion and mechanical properties
A homogenization annealing (HA) heat treatment was proposed
in our prior work [21] for mechanical property enhancement for
AZ91; HA for 10 h at 410 C causes an improvement in hardness,
ultimate tensile strength and ductility without loss of corrosion
properties, i.e. corrosion properties are similar to those of the ascast condition. The improvement of the mechanical properties is
due to the absence of a continuous easy crack path which is present in the as-cast microstructure due to the interconnected network throughout the microstructure of the ne lamellar (a + b)
micro-constituent and associated large b particles. In the as-cast
microstructure, the interconnected network throughout the microstructure of the ne lamellar (a + b) micro-constituent and associated large b particles provides a benet for corrosion by acting as a
corrosion barrier. However, this same microstructure causes inferior mechanical properties due to the continuous easy crack path
through this network.

1952

M.-C. Zhao et al. / Corrosion Science 50 (2008) 19391953

In contrast, the peak strength condition (HA for 10 h at 410 C)


has isolated ne b particles homogeneously distributed through
the a-phase matrix. There is no degradation of the mechanical
properties by a continuous easy crack path, and consequently the
mechanical properties are superior to those of the as-cast condition. As presented above, the isolated ne b particles do not lead
to an obvious loss of corrosion because they act as a small cathode
connected to a large anode (a-Mg matrix) so that the peak strength
condition (HA for 10 h at 410 C) has negligible micro-galvanic
corrosion.
There have been a number of investigations [16,4249] on the
inuence of aging on the mechanical properties of AZ91. These
have investigated the classic two step SA heat treatment. The
two heat treatment steps of SA are (i) rst a solution heat treatment at a higher temperature followed by quick cooling and (ii)
second an ageing heat treatment at a lower temperature. These
studies show that the optimum combination of strength and ductility are obtained for aging 510 h at 200 C. Longer aging times
leads to a decrease in mechanical strength. Fig. 7 show that the
corrosion rate also increases with longer aging time.
Our prior work proposed [21], for mechanical property
enhancement, the homogenization annealing heat treatment: HA
for 10 h at 410 C. Fig. 7 indicates that the corrosion rate for this
proposed HA heat treatment was greater than for SA205 and
SA210, i.e. solution heat treatment at 410 C for 100 h, water
quench and an ageing heat treatment at 200 C for 5 and 10 h.
However, HA only needs reheating up to a temperature for several
hours followed by air cooling rather than the two heat treatment
steps of SA, which has (i) rst a solution heat treatment at a higher
temperature followed by quick cooling and (ii) second an ageing
heat treatment at a lower temperature. As a consequence, HA has
potentially a cost lower than the SA treatment.

5. Conclusions
1. The ne lamellar (a + b) micro-constituent, in as-cast AZ91,
may be nucleated by a b-phase plate. This b-phase plate may
be effective as a corrosion barrier to subsequent corrosion.
2. Most AZ91 specimens exhibited an increase in corrosion rate
with increasing immersion time for immersion times of up to
96 h in 1 N NaCl. Part of the reason for the increasing corrosion
rate is the increase in area corroded with increasing immersion
time. This means that long term immersion tests lasting two
weeks to a month may be needed to measure steady state corrosion rates.
3. Relatively short immersion tests like 48 h are valuable as they
reveal the same trends as longer immersion tests 96 h, and
most probably also correlate with steady state corrosion
conditions.
4. The weight loss rate is directly related to the hydrogen evolution
rate by the relationship: weight loss rate [mg/cm2/d] = 1.085
(Hydrogen evolution rate [ml/cm2/d]), as expected from the
overall magnesium corrosion reaction that one molecule of
hydrogen is evolved for each atom of corroded magnesium.
5. There was considerable difference in corrosion rates and areas
corroded for the various microstructures.
6. AZ91SS (i.e. AZ91 subjected to a solid solution heat treatment of
100 h at 410 C and water quenched to have a microstructure
largely single a-phase) had a corrosion rate higher than that
of pure Mg despite the fact that the corrosion covered part of
the surface for AZ91SS whereas corrosion was uniform for pure
Mg and covered the whole surface of the pure Mg. Thus the
higher corrosion rate for AZ91SS is not an effect of surface area
corroding. An impurity explanation is ruled out by the fact that
both AZ91 and pure Mg were high purity.

7. The microstructural inuence of corrosion can be understood


from the interaction of the following three factors: (i) the surface lm formed on the surface of the a-phase can be more or
less effective in hindering corrosion and more or less effective
in controlling the form of corrosion as uniform corrosion or
localised corrosion, (ii) the second phase (the b-phase in
AZ91) can cause micro-galvanic acceleration of corrosion of
the matrix a-phase and (iii) the second phase (the b-phase in
AZ91) can act as a corrosion barrier and hinder the corrosion
propagation in the matrix a-phase.
8. It is expected that the same factors are important for all multiphase Mg alloys because all known second phases have corrosion potentials more positive than that of the a-phase.

Acknowledgements
This work was supported by the ARC Center of Excellence, Design of Light Alloys. CAST CRC was established under, and is supported in part by, the Australian Governments Cooperative
Research Centres scheme. DH StJohn is thanked for valuable discussion on the formation of the ne lamellar (a + b) microconstituent.
References
[1] K.Y. Sohn, J.A. Yurko, F.C. Chen, J.W. Jones, J.E. Allison, in: Proc. Auto Alloys II,
San Antonio, TX, The Material, Metals and Minerals Society, 1988, p. P81.
[2] R.S. Busk, Magnesium Products Design, 499, Marcel Dekker, New York, 1987.
[3] B.B. Clow, Advanced Materials and Processes 150 (1996) 33.
[4] G. Song, A. Atrens, Advanced Engineering Materials 1 (1999) 1133.
[5] G.L. Song, A. Atrens, Advanced Engineering Materials 5 (2003) 837.
[6] G. Song, Advanced Engineering Materials 7 (2005) 563.
[7] N. Winzer, A. Atrens, G. Song, E. Ghali, W. Dietzel, K.U. Kainer, N. Hort, C.
Blawert, Advanced Engineering Materials 7 (2005) 659693.
[8] G. Song, A. Atrens, Advanced Engineering Materials 9 (2007) 177183.
[9] J.X. Jia, G.L. Song, A. Atrens, Corrosion Science 48 (2006) 21332153.
[10] N. Winzer, A. Atrens, W. Dietzel, G. Song, K.U. Kainer, Materials Science and
Engineering A 472 (2008) 97106.
[11] N. Winzer, A. Atrens, W. Dietzel, V.S. Raja, G. Song, K.U. Kainer, Materials
Science and Engineering A (2008), doi:10.1016/j.msea.2007.11.064.
[12] R. Ambat, N.N. Aung, W. Zhou, Corrosion Science 42 (2000) 1433.
[13] G.L. Song, A. Atrens, X.L. Wu, B. Zhang, Corrosion Science 40 (1998)
1769.
[14] G.L. Song, A. Atrens, M. Dargusch, Corrosion Science 41 (1999) 249.
[15] R.K. Singh Raman, Metallurgical and Materials Transactions A 35A (2004)
2525.
[16] G. Song, A.L. Bowles, D.H. St. John, Materials Science and Engineering A 366
(2004) 74.
[17] W. Zhang, S. Jin, E. Ghali, R. Trmblay, M. Shehata, E. Es-Sadiqi, Advanced
Engineering Materials 8 (2006) 973.
[18] N. Winzer, A. Atrens, W. Dietzel, G. Song, K.U. Kainer, Materials Science and
Engineering A 466 (2007) 18.
[19] G. Ben-Hamu, D. Eliezer, K.S. Shin, Materials Science and Engineering A 447
(2007) 35.
[20] M.C. Zhao, M. Liu, G. Song, A. Atrens, Advanced Engineering Materials 10
(2008) 104111.
[21] M.C. Zhao, M. Liu, G. Song, A. Atrens, Advanced Engineering Materials 10
(2008) 93103.
[22] M . Liu, D. Qiu, M.C. Zhao, G. Song, A. Atrens, Scripta Materialia 58 (2008) 421
424.
[23] C. Suman, SAE Transactions 99 (5) (1990) 849.
[24] A.K. Dahle, Y.C. Lee, M.D. Nave, P.L. Schaffer, D.H. St. John, Journal of Light
Metals 1 (2001) 61.
[25] C.J. Bettles, Materials Science and Engineering A 348 (2003) 280.
[26] M.X. Zhang, P.M. Kelly, Scripta Materialia 48 (2003) 647.
[27] A. Srinivasan, J. Swaminathan, U.T.S. Pillai, K. Guguloth, B.C. Pai, Materials
Science and Engineering A (2007), doi:10.1016/j.msea.2007.09.059.
[28] .Y. Wan, J. Tan, G. Song, C. Yan, Metallurgical and materials Transactions 37A
(7) (2006) 23132316.
[29] O. Lunder, J.E. Lein, T.K. Aune, K. Nisancioglu, Corrosion 45 (1989) 741748.
[30] S. Prem Kumar, S. Kumaran, T. Srinivasa Rao, Materials Science and Technology
20 (2004) 835.
[31] ASM International, ASM Handbook Online, 2003.
[32] G.L. Song, A. Atrens, D.H. St. John, J. Nairn, Y. Lang, Corrosion Science 39 (1997)
855875.
[33] G.L. Song, A. Atrens, X. Wu, B. Zhang, Corrosion Science 40 (1998) 1769
1791.

M.-C. Zhao et al. / Corrosion Science 50 (2008) 19391953


[34]
[35]
[36]

[37]
[38]
[39]
[40]
[41]

ASM International, Metals Handbook, Corrosion, ninth ed., vol. 13, ASM
International, 1987.
H.H. Uhlig, R.W. Revie, Corrosion and Corrosion Control, third ed., John Wiley,
1985.
G. Song, A. Atrens, D. St. John, An hydrogen evolution method for the
estimation of the corrosion rate of magnesium alloys, in: John N. Hryn (Ed.),
Magnesium Technology 2001, TMS, 2001, pp. 255262.
Mingxing Zhang, PhD thesis, Crystallography of phase transformations in
steels, The University of Queensland, 1997.
J.Q. Wang, A. Atrens, D.R. Cousens, C. Nockolds, S. Bulcock, Journal of Materials
Science 34 (1999) 1711.
J.Q. Wang, A. Atrens, D.R. Cousens, N. Kinaev, Journal of Materials Science 34
(1999) 1721.
I.J. Polmear, Light Alloys-Metallurgy of Light Metals, second ed., Edward
Arnold, 1989 (Chapter 5).
E.H. Adolf Beck, The Technology of Magnesium and its Alloys. A translation
from the German by the technical staffs of F.A. Hughes and Co. Ltd and

[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]

1953

Magnesium Elekton Ltd of Magnesium and seine Legierungen. London, 1940


(Chapter 3).
J.P. Zhou, D.S. Zhao, R.H. Wang, Z.F. Sun, J.B. Wang, J.N. Gui, O. Zheng, Materials
Letters 61 (2007) 47074710.
N. Balasubramani, A. Srinivasan, U.T.S. Pillai, B.C. Pai, Materials Science and
Engineering A 457 (2007) 275281.
A. Srinivasan, U.T.S. Pillai, P.C. Pai, Materials Science and Engineering A 452
(2007) 8792.
G.L. Song, A.L. Bowles, D.H. St. John, Materials Science Forum 488489 (2005)
709712.
C. Corby, C.H. Caceres, P. Lukac, Materials Science and Engineering A 387
(2004) 2224.
C.H. Caceres, C.J. Davidson, J.R. Grifths, C.L. Newton, Materials Science and
Engineering A 325 (2002) 344355.
S. Celotto, Acta Materialia 48 (2000) 17751787.
S.P. Kumar, S. Kumaran, T.S. Rao, Materials Science and Technology 20 (2004)
835838.

Anda mungkin juga menyukai