Sie sind auf Seite 1von 14

Corrosion Science 51 (2009) 1950–1963

Contents lists available at ScienceDirect

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

Stress corrosion cracking of gas-tungsten arc welds in continuous-cast AZ31 Mg


alloy sheet
N. Winzer a,*, P. Xu a, S. Bender b, T. Gross a, W.E.S. Unger a, C.E. Cross a
a
Federal Institute for Materials Research and Testing, Unter den Eichen 87, 12200 Berlin, Germany
b
Otto-von-Guericke University Magdeburg, Universitätsplatz 2, 39106 Magdeburg, Germany

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

Article history: AZ31 Mg alloy sheet was welded using a gas-tungsten arc (GTA) process over inserts containing 2.3–
Received 17 February 2009 9.3 wt.% Al. The welded specimens were susceptible to SCC in distilled water, with susceptibility increas-
Accepted 14 May 2009 ing with decreasing weld metal Al (or b particle) concentration. Primary stress corrosion cracks initiated
Available online 6 June 2009
at the weld metal–HAZ interface by stress-assisted localised dissolution and propagated through the
weld and base metals by transgranular and intergranular H-assisted fracture (TG-HAF and IG-HAF)
Keywords: respectively. The IG fracture mode may be intrinsic to the texture imparted upon the base metal by roll-
C. Stress corrosion
ing. The increase in SCC susceptibility with decreasing weld metal Al concentration is contrary to the pur-
C. Hydrogen embrittlement
A. Magnesium
ported roles of b particles in promoting localised corrosion and as crack nucleation sites, but corresponds
C. Welding with increases in weld – base metal galvanic current density and weld metal localised corrosion
susceptibility.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction form) on SCC susceptibility has thus far been related to their pro-
pensity to act as crack nucleation sites [3,4,7–10] and/or drive
The increasing demand for lighter, more fuel-efficient automo- localised corrosion [11–14]. Intergranular stress corrosion cracking
biles has spurred renewed interest in the use of Mg alloys for auto- (IGSCC) is also possible in Mg–Al alloys, with the generally ac-
mobile components. However, a major obstacle to the widespread cepted mechanism being galvanically-driven localised dissolution
use of Mg alloys in such applications is their susceptibility to stress (tunnelling) in the a-matrix adjacent to grain boundary precipi-
corrosion cracking (SCC), which can cause failure at stresses as low tates [15–17]. The nature of this mechanism implies that it is spe-
as 50% rY in ordinary environments [1–4]. The renewed interest in cific to alloys with continuous networks of the b phase (i.e. those
Mg alloys also brings with it the need to develop suitable joining with high concentrations of Al) in aggressive environments (e.g.
processes. One method of achieving high quality joints in Mg alloys those containing Cl).
is gas-tungsten arc (GTA) welding [5,6]. To date there have been
few publications on the susceptibility of welds, particularly GTA
welds, in Mg alloys to SCC. Thus, it is necessary to relate our cur- 2. Experimental method
rent understanding of SCC of Mg alloys to GTA weld and weld-af-
fected microstructures. Recent works have provided new insights Sheets of 3 mm thick, continuous-cast, hot-rolled AZ31 from
into the mechanisms for SCC of Mg alloys; however, a unified MgF Magnesium Flachprodukte GmbH were welded using a vari-
understanding of these mechanisms remains elusive. There is a able polarity, gas-tungsten arc, cold wire feed (VP–GTA–CWF) pro-
general consensus that transgranular stress corrosion cracking cess, with the weld direction perpendicular to the rolling direction.
(TGSCC) is the inherent mode of SCC of Mg alloys. The mechanism The welding parameters are given in Table 1. Welds were made
for TGSCC is some form of hydrogen embrittlement (HE) with the H using 3  4 or 3  5 mm inserts (Fig. 1) machined from as-cast in-
coming from the cathodic partial reaction (H generation) of the Mg gots of the Mg alloys AM20, AM50, AM60 and AZ91. Welds were
corrosion reaction [1]. There is also a growing consensus that SCC also made without inserts using AZ61 filler wire. The bulk compo-
susceptibility of Mg–Al alloys is dependent on the concentration sitions of the base metal, inserts and filler wire are given in Table 2.
and morphology of Mg17(Al,Zn)12 b particles, with an increase in The concentrations of elements not shown were below measure-
b particle concentration associated with an increase in SCC suscep- ment tolerance or 10 ppm. The welded sheets were machined into
tibility. The influence of b particles (at least in their discontinuous un-notched tensile specimens, with their tensile axis collinear with
the rolling direction (Fig. 2). Prior to testing, the weld-reinforce-
* Corresponding author. Tel.: +49 3081044602; fax: +49 3081041557. ment was removed and the gauge sections polished with up to
E-mail address: nickwinz@graduate.uwa.edu.au (N. Winzer). 1200 grade emery paper and degreased using ethanol.

0010-938X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2009.05.037
N. Winzer et al. / Corrosion Science 51 (2009) 1950–1963 1951

Table 1
Welding parameters.

Weld type: GTAW, 5 mm/s, Ar shielding


Preparation: Square but joint, 3  4 mm square insert
Insert materials: AM20, AM50, AM60 and AZ91
Polarity: EN 135 A for 16 ms, EP 80 A for 4 ms (50 Hz)
Filler wire: 1.2 mm diameter AZ61, 42 mm/s feed rate

Fig. 2. Schematic showing layout of specimen tensile axis and welding direction
relative to the rolling direction.

Polarisation and electrochemical noise (EN) measurements


were performed by Göllner et al. at Otto-von-Guericke University
Magdeburg. Polarisation measurements were performed using
the rotating disc electrode (RDE) technique, which involves rotat-
Fig. 1. Schematic of weld preparation showing machined insert. ing the working electrode at high speed (2000 rpm in this case)
in order to separate excess corrosion products and H bubbles from
the electrode surface. A slightly-aggressive solution of 0.05 M
The specimens were tested under constant extension rate test Na2SO4 at pH 9 was used, with Ar gas bubbled through the solution
(CERT) conditions in double-distilled water, with control tests car- during the test to prevent CO2 absorption. Potential sweeps were
ried out in laboratory air. Double-distilled water was previously performed at 1 mV/s using a Pt counter electrode and a saturated
shown to induce TGSCC in Mg–Al alloys in the absence of severe Ag/AgCl reference electrode. EN measurements were performed
localised corrosion [3,4]. The mechanical load was applied using using two identical working electrodes (machined from the base
an InstronTM 4505 electro-hydraulic testing machine. The applied metal, HAZ or weld metal of the specimens) and a saturated Ag/
crosshead speed was 5 lm/hr, which corresponded to an actual AgCl reference electrode in a corrosion-inhibiting solution of NaOH
strain rate in the gauge section of 6  10 9 s 1 as measured dur- at pH 12 and open circuit conditions. The working electrodes were
ing elastic deformation in laboratory air. The gauge surface was polished to 1000 emery paper immediately before testing. Electro-
completely immersed in distilled water at room temperature and chemical current noise (ECN) was measured between the two
open circuit conditions in the environment cell. The water was working electrodes and electrochemical potential noise (EPN) be-
slowly circulated between the environment cell and a larger reser- tween one working electrode and the reference electrode. The
voir. The constant extension rate was maintained until there was a sampling rate was 20 Hz. The measurements were shielded from
significant reduction in load. The specimen was then removed from external noise sources using a Faraday Cage. A low pass filter with
the environment and the remaining cross-section fractured under 1 Hz cut-off frequency was applied to the mean components of the
very high strain rate in laboratory air in order to preserve the frac- current and potential. A band pass filter with 0.1–10 Hz range was
ture surface. The SCC fracture surfaces were rinsed in ethanol applied to the current and potential noise.
immediately following the test, and cleaned using 180 g/L CrO3 X-ray photoelectron spectroscopy (XPS) measurements were
immediately prior to SEM examination. performed on 5  5  3 mm specimens cut from the weld metals.
SEM fractography was performed using a CamScanTM 2 machine. The specimens were polished on one side to 2400 grade emery pa-
HR-SEM, EBSD and EDS/X microstructural composition analyses per, degreased using ethanol and immersed in double-distilled H20
were performed using a LEOTM Gemini 1530 VP machine. Grain for 72 h. The XPS measurements were performed on the polished
boundary characterisation was performed using a JEOLTM JEM4000 surfaces using a Kratos AnalyticalTM AXIS Ultra DLD machine. Sur-
FX type TEM, with the TEM lamellae prepared using a FEITM Strata vey spectra were recorded using monochromated Al Ka excitation
200 xP type focused ion beam (FIB) and transferred to the TEM grid at a pass energy of 80 eV. The electron emission angle was 0° and
with the aid of a light microscope and micro-manipulator [18]. the source-to-analyzer angle was 54°. The binding energy scale of
Microprobe analysis was performed using a JEOLTM JXA8900RL. the instrument was calibrated according to a Kratos AnalyticalTM
Metallographic specimens were etched using a solution containing procedure which uses ISO 15472 [20] binding energy data. Spectra
10 mL acetic acid, 4.2 g picric acid, 10 mL H2O and 70 mL ethanol. were taken by setting the instrument to the hybrid lens and slot
Specimens were prepared for EBSD by mechanically polishing the modes providing an analysis area of approximately 300 
specimens using colloidal silica slurry and then etching in a solu- 700 lm. The samples were fixed by double sided adhesive tape
tion containing 10 mL HNO3, 30 mL acetic acid, 40 mL H2O and on the sample holder. The charge neutralizer was used. Quantifica-
120 mL ethanol for 10 s as per Keshavarz and Barnett [19]. tion of the survey spectra was carried out with the help of the

Table 2
Bulk compositions of base metal, weld inserts and filler wire.

Mg (wt.%) Al (wt.%) Zn (wt.%) Mn (wt.%) Si (ppm) Fe (ppm) Sn (ppm) Cu (ppm) Th (ppm)


AZ31 (base metal) 96.1 2.70 0.81 0.30 231 40.1 <20 11.7 239.1
AM20 (weld insert) 97.2 2.35 <0.0025 0.37 31.2 63.6 <20 <5 241.5
AM50 (weld insert) 95.1 4.47 <0.0025 0.35 43.6 <12 <20 <5 <150
AM60 (weld insert) 93.8 5.85 <0.0025 0.28 53.3 <12 23.1 <5 <150
AZ61 (filler wire) 94.7 5.94 0.90 0.30 <10 <10 – 8.4 –
AZ91 (weld insert) 89.8 9.21 0.76 0.18 81.9 22.6 54.5 <5 <150
1952 N. Winzer et al. / Corrosion Science 51 (2009) 1950–1963

was characterised by a slight increase in grain size (to 40–


80 lm) relative to the base metal. The weld metal for AZ61 and
AZ91 was characterised by a relatively uniform 50–100 lm grain
size throughout, whereas for AM20, AM50 and AM60 the grains in-
creased in size (up to 1000 lm) with distance from the arc con-
tact point and were elongated close to the HAZ interface. Fig. 4
shows the inverse pole figures (IPFs) for a region traversing the
base metal, HAZ and weld metal for a specimen welded using
AZ61 filler wire. Grains in the base metal and HAZ were oriented
with their (0001) basal plane parallel to the rolling plane, which
is horizontal in this case (i.e. with their c-axis parallel with the
Y-direction in the IPF coordinate system). Grains in the weld metal
appear randomly oriented.
Phase characterisation was performed using EDS/X and micro-
probe. The base metal (Fig. 5) was comprised of an a-matrix
(97 at.% Mg–2.5 at.% Al–0.5 at.% Zn) with 1–2 lm Mg–Al–Mn parti-
cles (too small to accurately characterise). The Mg–Al–Mn particles
were sometimes, but not invariably, aligned collinearly in the roll-
ing direction. Examination of the FIB-prepared base metal samples
Fig. 3. Etched optical micrograph of the base metal–HAZ–weld metal interface in a
specimen welded using an AM60 insert.
using TEM revealed no preferential precipitation at grain bound-
aries (Fig. 7). For the weld metals, the average a-Al concentrations
were 1.5 at.% for AM20 WM, 2.0 at.% for AM50 WM and AZ61 WM,
CasaXPSTM Version 2.3.12 software by using relative sensitivity 2.7 at.% for AM60 WM and 3.7 at.% for AZ91 WM. All weld metals
factors (RSFs) taken from the KratosTM element library. By doing this contained 5–10 lm interdendritic particles with the composition
RSFs of 0.193 and 0.168 were used for Al 2p and Mg 2p, 62 at.% Mg–38 at.% Al (for AMxx weld metals) or 62 at.%
respectively. Mg–31 at.% Al–7 at.% Zn (for AZxx weld metals) (Fig. 6). Mn was
preferentially precipitated in the form of 1 lm almost-spherical
3. Results Mg–Al–Mn particles. The small Mg–Al–Mn particles often occurred
within the larger Mg–Al–Zn particles. The size and density of the
3.1. Microstructural characterisation Mg–Al–Zn particles increased with the Al-content of the welding
inserts; however, the proportion of particles in the matrix was very
Welding resulted in a bead of 10–12 mm width at the face side low (<1 vol.%) even for AZ91 WM. The size and density of the Mn-
and 8–10 mm width at the root side. The apparent width of the rich particles were relatively consistent in all cases. The Mg–Al–Zn
HAZ varied from 200 lm at the face side to 500 lm at the root particles occurred within grains and at grain boundaries, with an
side. Fig. 3 shows an optical micrograph of the base metal–HAZ– increasing tendency (particularly for larger particles) to occur at
weld metal interface for a typical specimen. The base metal con- grain boundaries with increasing Al concentration. The Mg–Al–Zn
sisted primarily of 20–60 lm equiaxed grains, with some larger particles in AZ91 WM were surrounded by an Al-rich (up to
grains (up to 140 lm) elongated in the rolling direction. The HAZ 8 at.% measured) zone. The Al-rich zones were not apparent for

Fig. 4. Inverse pole figures for the base metal, HAZ and weld metal showing normal crystallographic directions for a specimen welded using an AZ61 insert (note that [0 0 1]
denotes [0 0 0 1] in HCP notation).
N. Winzer et al. / Corrosion Science 51 (2009) 1950–1963 1953

weld metals with lower Al concentrations. The Mg–Al–Zn particles


had a composition and morphology similar to those for the b phase,
which has the nominal composition Mg17(Al,Zn)12 [15–17,21], and
are thus referred to as b particles herein. The Al and Zn concentra-
tions of the particles are inconsistent with those of the s and U
phases, which have the nominal compositions Mg32(Al,Zn)49 and
Mg5Zn2Al2, respectively, [22].

3.2. SCC characterisation

Fig. 8 shows the mechanical response of welded and non-


welded specimens in distilled water and laboratory air. Note that
the horizontal axis has been designated ‘‘apparent strain”, having
been calculated from the crosshead displacement rather than the
actual strain at the specimen gauge section. For welded specimens
in laboratory air, the reduction in weld metal Al concentration was
characterised by an increase in Young’s Modulus and decreases in
Fig. 5. Electron backscatter image of base metal microstructure showing Mg–Al– rp0.2, UTS and elongation-to-failure (except for AZ91 WM, which
Mn particles, which were sometimes aligned collinearly in the rolling direction. had lower-than-expected UTS and elongation-to-failure). For spec-
imens in distilled water, the apparent plastic deformation is largely
attributed to SCC as per Winzer et al. [2–4]. Thus, the threshold
stress for SCC, rSCC, was taken as that corresponding to the onset
of apparent plastic deformation. All specimens suffered a reduction
in rSCC, UTS and elongation-to-failure in distilled water relative to
tests in laboratory air. Welded specimens had a lower rSCC and UTS
and higher elongation-to-failure relative to non-welded speci-
mens, with rSCC and UTS decreasing and elongation-to-failure
increasing with decreasing weld metal Al concentration (i.e. from
AZ91 WM to AM20 WM). The decrease in rSCC and increase in elon-
gation-to-failure with decreasing weld metal Al concentration
were also apparent when normalised against the equivalent
parameters for tests in laboratory air (Table 3). The normalised
UTS’s were relatively consistent in all cases except AZ91 WM.

3.3. Crack morphology

Primary stress corrosion cracks in welded specimens generally


Fig. 6. Electron backscatter image of AZ91 WM microstructure showing: occurred at the interface between the weld metal and HAZ
(A) interdendritic b particles; (B) spherical Al–Mn particles; and (C) Al-rich zones. (Fig. 9). Secondary cracks occurred in the base metal, HAZ and weld

Fig. 7. TEM images of grain boundaries in AZ31 base metal.


1954 N. Winzer et al. / Corrosion Science 51 (2009) 1950–1963

Fig. 8. Results of CERT tests for welded and non-welded specimens in distilled water and laboratory air at a crosshead speed of 5 lm/h.

metal (Figs. 11 and 12). The cracks appeared to emanate from the associated with the deviations in the (0 0 0 1) basal plane from
corners of the specimens, presumably due to the higher stress con- the rolling plane or changes in grain size. It was also possible for
centration at these points, and often spanned the entire width of the cracks to propagate across the weld metal–HAZ-base metal
the specimen. SCC initiation, whether in the base metal, HAZ and interface, and thus change morphology from transgranular to
weld metal, was manifested by transgranular ‘‘finger-like” tunnels intergranular.
(Fig. 12, Detail B). Such features have previously been attributed to
a mechanism involving alternating stages of crack tip film rupture
and dissolution, possibly triggered by film rupture and dissolution
at surface slip steps [3,4]. It is unclear from the micrographs
whether the initiation points correspond to second phase particles.
Secondary cracks in non-welded specimens occurred across the en-
tire length of the gauge section (Fig. 10). These cracks also ema-
nated from the corners of the specimens, but did not span their
entire width. The morphology of secondary cracks in non-welded
specimens was similar to that for welded specimens (i.e. consistent
with the film rupture mechanism). Crack propagation through the
weld metal was transgranular (Fig. 11, Detail C), although it is un-
clear whether this involved fracture of b particles, which is a
critical process in the mechanism for TGSCC in cast AZ91 alloy
[3,4,7–10]. Cracks in the weld metal generally propagated perpen-
dicular to the loading direction, but sometimes along the interface
between the base metal and HAZ. Crack propagation through the
base metal and HAZ was macroscopically perpendicular to the
loading direction, with the crack tip characterised by an array of
sharp intergranular branches (Fig. 11, Detail A; Fig. 12, Details Fig. 9. Stress corrosion cracks on the surface of a specimen welded using an AM20
C–E). The changes in the direction of crack branches might be insert after testing in distilled water.

Table 3
Mechanical and SCC performance indicators for welded and non-welded specimens in distilled water and laboratory air. DlAIR and DlSCC are the elongations-to-failure (or
elongation corresponding to a 2% reduction in load in the case of distilled water) in laboratory air and distilled water, respectively.

rAIR,p0.2 (MPa) UTSAIR (MPa) DlAIR (%) rSCC (MPa) UTSSCC (MPa) DlSCC (%) rAIR,p0.2/rSCC UTSSCC/UTSAIR DlAIR/DlSCC
Non-welded 140 255 25.9 120 135 3.02 0.86 0.53 0.12
Welded (AZ91) 115 190 5.97 105 125 3.15 0.95 0.66 0.53
Welded (AZ61) 105 245 17.5 95 120 3.45 0.90 0.49 0.20
Welded (AM60) 105 230 13.6 75 115 3.50 0.70 0.50 0.26
Welded (AM50) 100 205 8.74 75 115 3.65 0.75 0.56 0.42
Welded (AM20) 80 190 8.44 50 105 3.73 0.65 0.55 0.44
N. Winzer et al. / Corrosion Science 51 (2009) 1950–1963 1955

granular and transgranular features, with the proportion of each


morphology depending on whether the crack propagated through
the weld metal, HAZ or base metal (Fig. 16). Transgranular features
were generally characterised by quasi-cleavage-like markings
(Fig. 16, Detail C), similar to those observed for TGSCC of cast
AZ91 and attributed to HE [2–4]. The intergranular facets con-
tained extremely fine parallel markings (Fig. 17). It has been pro-
posed [23] that these markings may be artefacts of the
intersection of slip bands with the crack surface in the wake of
the crack tip plastic zone. It is also possible that the appearance
of these slip bands is exacerbated by dissolution at the film-free
slip step. For all specimens, the fracture surface regions corre-
sponding to fast rupture were oriented 45° to the loading axis
and consisted of coarse, dimpled features as per the fracture sur-
faces resulting from tests in laboratory air (Fig. 13).
For specimens tested in distilled water there was an increase in
damage to the fracture surface by corrosion with crack length (i.e.
with time of exposure to environment). The corrosion damage was
Fig. 10. Stress corrosion cracks on the surface of an un-welded specimen after
testing in distilled water.
manifested by widening of fissures along grain boundaries (for
intergranular fracture surfaces) or dissolution adjacent to b parti-
cles (for transgranular fracture surfaces, Fig. 16, Detail A). Due to
The fracture surfaces for welded and non-welded specimens the corrosion damage, the fine parallel markings within intergran-
tested in laboratory air were oriented 45° to the loading axis and ular facets were only visible on grains close to brittle-ductile tran-
characterised by coarse, dimple-like features (Fig. 13). The fracture sition (i.e. with least exposure to environment).
surfaces for welded and non-welded specimens tested in distilled
water were comprised of zones corresponding to: (i) crack initia- 3.4. Characterisation of corrosion
tion by localised corrosion (tunnelling); (ii) crack propagation by
SCC/HE; and (iii) final rupture. Crack initiation was manifested Fig. 18 shows the electrochemical polarisation curves for the
by small (extending <100 lm from the surface) thumbnail-shaped base and weld metals, and for the HAZ adjacent to AZ91 WM.
zones (Fig. 14). These zones sometimes featured fine parallel mark- The base metal had a considerably higher free corrosion potential,
ings, which may be an artefact of the final stages of crack tip Ecorr, and slightly higher corrosion current density, Icorr, relative to
repassivation, mechanical film rupture and bare metal dissolution. the weld metals. The polarisation curve for the HAZ appeared sim-
For non-welded specimens, the fracture surface regions corre- ilar to that for the base metal, although this result may be inaccu-
sponding to crack propagation were comprised of intergranular rate due to practical difficulties associated with measuring such a
features (Fig. 15). For welded specimens, the fracture surface re- narrow region. For AZ91 WM, AM60 WM, AM50 WM and AM20
gions corresponding to crack propagation were comprised of inter- WM, Icorr was relatively consistent but there was a decrease in Ecorr

Fig. 11. Secondary stress corrosion cracks in a specimen welded using an AM50 insert showing intergranular crack propagation in the HAZ (Detail A) and transgranular crack
propagation through the weld metal (Detail C).
1956 N. Winzer et al. / Corrosion Science 51 (2009) 1950–1963

Fig. 12. Secondary stress corrosion cracks in aspecimen welded using an AM20 insert showing crack initiation by transgranular localised dissolution (Detail B) and
intergranular crack propagation through the base metal (Detail E).

with decreasing Al concentration. AZ61 WM did not conform to the AM20 WM) and slightly higher icorr relative to the other weld
this trend, having a considerably lower Ecorr (similar to that for metals. The decrease in weld metal Al concentration also corre-
sponded with a general increase in the current densities at which
the anodic branches of the weld metal polarisation curves inter-
sected with the cathodic branches of the base metal/HAZ polarisa-
tion curves (galvanic current densities). This indicated that the
potential galvanic current between the weld and base metals in-
creased with decreasing weld metal Al concentration. It should
be remembered that these measurements were performed with
the working electrode rotating at 2000 rpm, such as to separate ex-
cess corrosion products and H bubbles from the electrode surface.
Without rotation, Ecorr was considerably more positive and icorr
considerably lower than with rotation.
Typical examples of the electrochemical current noise (ECN)
and electrochemical potential noise (EPN) signals are shown in
Fig. 19. These signals were measured over a period of 5 min begin-
ning immediately after immersion in the solution. During the ini-
tial 50 s, the mean measured potential underwent an exponential
change, eventually converging on steady value. This initial tran-
sient may correspond to a change in surface film composition to in-
clude thick Mg(OH)2 layers upon immersion in the aqueous
environment, as shown by Nordlien et al. [24] for pure Mg in dis-
Fig. 13. Ductile fracture surface morphology for non-welded specimen tested in tilled water. The EN signals were characterised by a high fre-
laboratory air. quency, low amplitude ‘‘background noise” and low frequency,
N. Winzer et al. / Corrosion Science 51 (2009) 1950–1963 1957

Fig. 14. HR-SEM images depicting zones corresponding to stress corrosion crack initiation by slip-induced localised corrosion (non-welded specimen in distilled water).

high amplitude transients (Fig. 20). It is unclear whether the back- siderably with reducing weld metal Al concentration for AZ61 WM,
ground noise corresponds to processes involved in general corro- AM60 WM and AM50 WM samples (the EN signals for AM20 WM
sion or external interference (e.g. from the measuring could not be measured due to excessive noise). A reduction in the
equipment). The low-frequency transients were characterised by amplitude of the low-frequency transients over time was also
a sudden deviation from, followed by exponential regression to, apparent, particularly for AM50 WM specimens. EN parameters
the mean current and potential. Such transients are often associ- for all measured samples, calculated over the time intervals 50–
ated with metastable pit formation and repassivation [25,26]. 150 s, 250–350 s and 450–550 s, are given in Table 4. The standard
The low-frequency transients were barely apparent for AZ31 BM, deviations for the ECN and EPN signals, rI and rV, respectively,
AZ91 HAZ and AZ91 WM samples, but increased in magnitude con- were relatively similar for AZ31 BM, AZ61 HAZ, AZ91 WM and
AZ61 WM, but considerably higher for AM60 WM and AM50
WM. The increases in rI and rV for AM60 WM and AM50 WM
are in line with the increase in the magnitudes of the low-fre-
quency transients.
The EN signals were converted to power spectral density (PSD)
plots using the Fast Fourier Transform technique in the frequency
range 0.1–10 Hz. The PSD plots (Fig. 21) were characterised by a
log-linear gradient in the high frequency range (>0.5 Hz), tending
towards a plateau (‘‘white noise”) in the low-frequency range.
The key PSD parameters are given in Table 4. There was a general
decrease in the log-linear gradients, SI and SV, with decreasing weld
metal Al concentration. It has been proposed [27] that this gradient
is indicative of the mode of corrosion; however, as there is little
consensus in the literature regarding the physical meaning of the
gradient, it is sufficient to say here that the decrease in gradient
may be indicative of a change in the corrosion mechanism. The
low-frequency limits of the PSD curves, WI,0.1Hz and WV,0.1Hz, were
considerably higher for AM60 WM and AM50 WM than for the
other samples, with those for AM50 WM being highest. For WM
specimens with lower Al concentrations, there was also a decrease
in WI,0.1Hz and WV,0.1Hz and increase in SI and SV over time, consis-
tent with the reduction in the amplitudes of the low-frequency
transients. Cottis and co-workers [28–30] proposed that for noise
containing discrete packets of charge (i.e. metastable pitting) the
corrosion process may be characterised by the average corrosion
current, Icorr, and the characteristic frequency, fn. Icorr is indicative
of the corrosion rate and is proportional to rI/rV. fn is indicative
of the mode of corrosion (high fn is associated with general corro-
sion, low fn with localised corrosion) and is inversely proportional
to WV,0.1Hz. In the present case, rI/rV was generally similar for most
specimens, except AM50 WM for which it was considerably higher
(Table 4). Given that Icorr and fn are independent, this suggests that
the corrosion for AM50 WM is more localised, with a higher overall
rate of corrosion, than for the other samples.
The anodic surfaces of the EN specimens were examined using
SEM. The surfaces contained deep, hemispherical or shallow, circu-
lar pits with diameters varying from 30 to 300 lm (Fig. 22). The
pits contained in AM50 WM samples were larger, deeper and more
sparsely distributed than those in AZ91 WM samples. This is con-
sistent with localisation of the corrosion process with reducing
Fig. 15. Intergranular fracture surface morphology for a non-welded specimen weld metal Al concentration. It was unclear whether the pits cor-
tested in distilled water. responded to any second phase particles in the substrate.
1958 N. Winzer et al. / Corrosion Science 51 (2009) 1950–1963

Fig. 16. Fracture surface for specimen welded with AM50 insert and tested in distilled water, showing: corrosion damage close to the surface (Detail A), transgranular crack
propagation through the weld metal (Detail B) and intergranular crack propagation through the base metal and HAZ (Detail C).

Fig. 17. Fracture surface corresponding to crack propagation through the base metal for a specimen welded using an AM50 insert and tested in distilled water, showing fine
parallel markings within intergranular facets (Details B and C).

3.5. Surface film analysis indicative of higher corrosion activity in the weld metal relative to
the base metal. Given that, in the case of Mg alloys, increased ano-
Immersion of welded specimens in distilled water invariably re- dic polarisation is accompanied by increased H evolution (the so-
sulted in immediate H bubble evolution at the weld metal. This is called Negative Difference Effect), this is consistent with the role
N. Winzer et al. / Corrosion Science 51 (2009) 1950–1963 1959

Fig. 18. Polarisation curves for weld and base metals, measured with the specimen rotating at 2000 rpm in 0.05 M Na2SO4 solution at pH 9.

Fig. 19. Electrochemical Potential Noise (EPN) and Electrochemical Current Noise
(ECN) results for AM60 WM specimen in NaOH solution at pH 12. Fig. 20. EPN and ECN results showing typical low-frequency transients for AM60
WM specimen in NaOH solution at pH 12.

of the weld metal as the anode in the macro scale. After immersion and tended towards a darker and more uniform appearance with
in distilled water, the surface films covering the weld metals were reducing weld metal Al concentration. The compositions of weld
also visibly darker in appearance than that covering the base metal, metal surface films as measured using XPS are given in Table 5.
1960 N. Winzer et al. / Corrosion Science 51 (2009) 1950–1963

Table 4
Electrochemical noise parameters for base metal, HAZ and weld metal samples.
5
Time frame (s) rI (nA) rV (lV) rI/rV (10 X 1) SI SV WI,0.1Hz (lA2 Hz 1) WV,0.1Hz (lV2 Hz 1)
10 1
AZ31 BM 50–150 0.29 9.30 3.10 0.62 1.06 2.7  10 5.7  10
10 2
250–350 0.27 3.63 7.50 0.69 1.08 2.9  10 4.9  10
10 2
450–550 0.23 3.91 5.93 0.50 0.97 3.1  10 7.2  10
10 0
AZ31–AZ61 HAZ 50–150 0.32 17.2 1.84 1.05 2.13 9.4  10 1.6  10
9
250–350 0.29 22.1 1.30 1.04 2.45 2.0  10 8.7  100
10
450–550 0.22 24.8 0.89 0.84 2.45 2.4  10 1.5  101
9 1
AZ91 WM 50–150 0.45 7.62 5.95 0.85 1.31 1.4  10 4.3  10
9 1
250–350 0.28 7.61 3.72 0.77 1.49 1.5  10 3.9  10
10 1
450–550 0.24 6.56 3.62 0.52 1.47 4.6  10 2.8  10
10
AZ61 WM 50–150 0.39 10.9 3.60 1.10 1.74 7.5  10 8.5  10 1
10
250–350 0.24 7.75 3.08 0.90 1.70 2.7  10 3.4  10 1
10
450–550 0.23 10.5 2.22 0.63 1.76 1.3  10 2.0  100
8
AM60 WM 50–150 1.22 20.8 5.93 1.36 1.91 3.4  10 1.3  101
8
250–350 1.47 37.2 3.96 1.36 2.12 4.2  10 3.3  101
10
450–550 0.39 8.59 4.54 0.63 1.60 5.8  10 4.5  10 1
6
AM50 WM 50–150 8.85 262 33.7 2.69 3.77 1.5  10 1.7  103
7
250–350 3.75 193 19.4 2.89 3.67 8.1  10 2.3  103
8
450–550 0.69 38.8 17.8 1.60 2.66 2.2  10 8.8  101

The results show a reduction in Al concentration (and Al/Mg ratio) the co-existence of other species in the surface film cannot be fully
of the weld metal surface films with reducing Al concentration of ruled out. Table 6 shows the binding energies for peaks in the XP
the substrate. Analysis of the highly resolved XP spectra (measured spectrum correlated with those given in the literature for various
at pass energy of 20 eV) revealed that Mg exists preferentially as Mg and Al compounds. In some cases the spectra contained minor
Mg(OH)2; however, it should be noted that metallic Mg occurs at Mg 2p and Al 2p peaks at binding energies slightly higher than
a very similar binding energy to Mg(OH)2 in the XP spectrum, so those corresponding to the main peaks. The binding energies for
the Mg 2p and Al 2p minor peaks correspond to the MgO/MgCO3
and Al2O3, respectively, indicating that these compounds may also
exist in the surface film.

4. Discussion

Welded and non-welded specimens were susceptible to SCC in


distilled water, with welded specimens being the more susceptible.
An increase in the SCC susceptibility of welded specimens, charac-
terised by a decrease in rSCC, with decreasing weld metal Al (or b
particle) concentration was also apparent, with SCC initiation in
AM20 WM samples occurring at just 0.65 rAIR,p0.2 (Table 3). Such
severe reductions in strength are in line with those reported by
Winzer et al. for Mg–Al alloys in distilled water [2–4]. The mor-
phology of SCC initiation in all specimens was indicative of a mech-
anism involving slip-induced film rupture and localised
dissolution. There were two apparent modes of SCC propagation,
each involving HE, with the predominant mode dependent on
microstructure (i.e. whether the crack propagated through the
weld metal, HAZ or base metal). The occurrence of IGSCC propaga-
tion in the base metal is peculiar for Mg alloys, and is discussed in
Section 4.1.
The increase in the SCC susceptibility of welded specimens with
decreasing weld metal b particle concentration is contrary to the
purported roles of b particles as crack nucleation sites [3,4,7–10]
and in promoting localised corrosion [11–14], both of which imply
a decrease in SCC susceptibility with decreasing b particle concen-
tration. Thus, the increase in SCC susceptibility may be rationalised
in terms of the influences of: (i) stress concentration at the weld
toe due to dissimilarities between the weld and base metal micro-
structures; (ii) residual stresses in the vicinity of the weld inter-
face; (iii) localised corrosion (which is somehow involved in the
mechanism for SCC initiation) in the HAZ and weld metal; or (iv)
H evolution (which occurs under anodic polarisation in Mg alloys).
It seems reasonable to expect that the decrease in weld metal b
Fig. 21. Power Spectral Density (PSD) plots for AM50 WM specimen in NaOH particle concentration corresponds with a decrease in stress
solution at pH 12. concentration at the weld toe, and therefore a decrease in SCC
N. Winzer et al. / Corrosion Science 51 (2009) 1950–1963 1961

Fig. 22. SEM images showing the morphology of pits contained within the anodic surface of the AM50 WM specimen following EN testing.

Table 5 mode IG and TG H-assisted fracture (HAF) has been found to occur
Compositions of weld metal surface films determined by XPS. in pure Mg [33,34]. Bobby Kannan et al. [35] reported IGSCC prop-
O 1s (at.%) C 1s (at.%) Al 2p (at.%) Mg 2p (at.%) Mn 2p (at.%) agation in AZ31 sheet adjacent to laser-beam welds; however, it
AZ91 WM 57.1 20.5 5.7 16.4 0.2
was unclear whether this was due to hydrogen embrittlement or
AZ61 WM 56.5 19.9 5.4 17.8 0.4 micro-galvanic tunneling. In the present case, the IG mode may
AM60 WM 56.9 19.1 5.4 18.2 0.5 be associated with the principle microstructural peculiarity of
AM50 WM 56.9 20.3 4.4 17.9 0.6 the base metal: the orientation of the c-axis normal to the rolling
AM20 WM 58.9 16.6 3.8 19.7 0.9
plane. Agnew et al. [36] used neutron diffraction to show that, even
at low stresses, deformation of hot-rolled AZ31 plate (consisting of
50 lm near-equiaxed grains with the same texture) is manifested
susceptibility. Since this is contrary to the apparent influence of b by composite-like load sharing between differently-oriented
particle concentration (Fig. 8), the influence of stress concentration grains. It seems reasonable to expect that load sharing between
has been neglected. The influences of residual stresses have also differently-oriented grains would be accompanied by elevated
been neglected, as their nature and extent are unknown; however, stress at grain boundaries, which may contribute to the propensity
it is acknowledged that the influences of residual stresses may be for decohesion in the presence of H. Considering the similarities in
significant. The electrochemical potential and EN results given in composition and microstructure of the alloy used by Agnew et al.
Section 3.4 correlate reasonably well with the SCC results, suggest- and that used in the present work, this may explain why IG-HAF
ing that SCC susceptibility might be somehow related to localised occurs in rolled AZ31 but not in other Mg alloys.
corrosion and/or H evolution. These correlations are discussed in The specific mechanism for IG-HAF is unclear. Mechanisms that
Section 4.2. have been proposed for IG-HAF in other materials are generally
derivations of hydrogen enhanced localised plasticity (HELP) [37–
4.1. IGSCC propagation in the base metal 41] and hydrogen enhanced decohesion (HEDE) [42–46]. IG-HELP
is usually associated with crack propagation through the soft pre-
The occurrence of IGSCC propagation in the base metal is in cipitate-free zone adjacent to the grain boundaries, which is not
stark contrast to our present understanding of SCC of Mg alloys. possible in the present case because the grain boundaries were
IGSCC in Mg–Al alloys is usually associated with preferential corro- not furnished by precipitates. HELP is typically characterised by
sion of the a-matrix adjacent to b-precipitates at grain boundaries dimpled intergranular facets, whereas intergranular facets associ-
[15,17,31,32]. The nature of this mechanism is such that it only oc- ated with HEDE are essentially featureless [23]. In the present case,
curs for alloys containing continuous networks of grain boundary it was difficult to identify dimples on the intergranular facets due
precipitates and exposed to aggressive environments. This mecha- to corrosion damage, although the presence of slip traces on the
nism is irresolvable with the present case due to the absence of intergranular facets is indicative of large-scale plasticity in the
grain boundary precipitates in the base metal (Fig. 7) and the pres- vicinity of the crack tip (i.e. consistent with HELP).
ence of slip traces within intergranular facets (Fig. 17), which are
indicative of large-scale crack tip plasticity [23]. It follows that 4.2. Correlation between SCC and corrosion results
the mechanism for IG crack propagation is a form of HE. Mixed-
The comparison of the EN signals and exposed (anodic) surfaces
for weld metal specimens immersed in NaOH solution indicated
Table 6 increasing localised corrosion susceptibility with decreasing Al
binding energies corresponding to XP peaks for weld metal surface films. concentration. This was manifested in the EN signals by: (i) in
the time domain, the increasing amplitude of low-frequency tran-
Mg 2p main Mg 2p minor Al 2p main Al 2p minor
peak (Mg(OH)2 peak (MgO [62]/ peak (Al(OH)3 peak (Al2O3 sients; and (ii) in the frequency domain, the decrease in fn (or in-
[56]/Mg (0) MgCO3 [63]) [64]) [65]) crease in WV,0). In contrast, the EN signal analyses for base metal
[57–61]) (eV) and HAZ samples indicated low localised corrosion susceptibility
AZ91 WM 49.8 – 74.2 – relative to the weld metals. This is consistent with SCC initiation
AZ61 WM 49.6 52.9 eV (30%) 74.2 77.3 eV (30%) within the weld metal, and the relatively low SCC susceptibility
AM60 WM 49.6 – 74.2 – of the non-welded specimens. Given that the mechanism for SCC
AM50 WM 49.4 – 74.0 –
AM20 WM 49.4 52.0 eV (20%) 74.1 76.5 eV (10%)
initiation appeared to involve localised corrosion (tunnelling), it
seems logical that the increase in propensity for localised corrosion
1962 N. Winzer et al. / Corrosion Science 51 (2009) 1950–1963

within the weld metals and the increase in SCC susceptibility of the increase in Al actually mitigates the propensity for b particles
welded samples with decreasing weld metal Al concentration are to behave as local cathodes. This is consistent with the EN results,
related. However, the extent of this relationship would be limited which showed an increasing susceptibility to localised corrosion
due to the influence of stress on the localised corrosion mecha- with decreasing weld metal Al concentration.
nism, which was evidenced by the difference in morphologies for The correlation between Ecorr and SCC susceptibility does not
localised corrosion in EN (Fig. 22) and SCC (Fig. 12B) specimens. hold for AZ61 WM: Ecorr for AZ61 WM is similar to that for AM20
The specific mechanism for localised corrosion in the absence of WM, even though the SCC susceptibility and EN characteristics
stress is unclear. Localised corrosion sites on the exposed surfaces for AZ61 WM are in line with the apparent influences of weld me-
of EN specimens (Fig. 22) did not appear to correspond to underly- tal Al concentration. This behaviour is difficult to explain. It was
ing b particles. It is possible that the localised corrosion mechanism previously proposed [12,54,55] that Zn renders both the a and b
is driven by other microstructural features. Al–Mn particles may phases more cathodic, with Zn having a greater influence on the
also act as local cathodes, albeit not as effectively as b particles b than on the a. In the present case, the comparison between the
[22,54,47]. However, in the present case Mn concentration was rel- polarisation curves for AZ61 and AM60 WM, which have generally
atively consistent for all AMxx weld metals, so this fails to explain similar microstructures except with respect to grain size and the
the apparent trend in localised corrosion susceptibility. Al–Mn par- presence of Zn, suggests that Zn has the opposite effect. It is un-
ticles may be rendered more cathodic by the substitution of Mn by clear what influence, if any, the smaller grain size for AZ61 WM
very small quantities (just a few ppm) of Fe [12,48]. Thus, it may be may have had.
important to note that AM20 WM (which was most susceptible to
localised corrosion) had an Fe concentration higher than the other 5. Conclusions
weld metals and above the tolerance limits for some Mg alloys
[12,48].
There also appeared to be some correlation (with the exception 1. Both welded and non-welded specimens were susceptible to
of AZ61 WM) between the increase in SCC susceptibility of the SCC in distilled water, with welded specimens being the more
welded specimens and the decrease in Ecorr (and corresponding in- susceptible. For welded specimens there was an increase in
crease in galvanic current density when coupled with the base me- SCC susceptibility, characterised by a decrease in rSCC, with
tal) for weld metal samples in Na2SO4 solution with decreasing decreasing weld metal Al (or b particle) concentration. This is
weld metal Al concentration. The decrease in Ecorr and increase in in stark contrast to the purported roles of b particles in promot-
galvanic current density imply an increasing propensity for gal- ing localised corrosion and as crack nucleation sites, both of
vanic coupling between the weld and base metals, with the weld which imply a decrease in SCC susceptibility with decreasing
metal being the local anode. This is indeed consistent with the b particle concentration.
severity of localised corrosion damage and initiation of primary 2. For welded specimens, SCC initiation occurred at the interface
stress corrosion cracks within the weld metal. It may also be between the weld metal and HAZ. The mechanism for SCC ini-
important to note that, for Mg alloys, anodic polarisation is accom- tiation appeared to involve slip-induced film rupture and local-
panied by increased H evolution (the so-called Negative Difference ised dissolution in all specimens. SCC crack propagation
Effect). Thus, the increasing galvanic current between the weld and occurred in the weld metal, HAZ and base metal. The apparent
base metals might be expected to correspond with increasing H mechanisms for crack propagation through the weld and base
activity at the weld metal. Given that H is involved in the SCC prop- metals were TG-HAF and IG-HAF, respectively. The IG fracture
agation mechanism, this is consistent with the increase in SCC sus- mode may be intrinsic to the texture imparted upon the base
ceptibility with decreasing weld metal Al concentration. metal by rolling.
The reduction in Ecorr with decreasing weld metal Al (b particle) 3. The increase in SCC susceptibility with decreasing weld metal
concentration is inconsistent with the results given by previous Al concentration corresponded with: (i) an increasing propen-
workers [49,50] for Mg–Al–Zn alloys, in which increasing b particle sity for metastable pit formation (in NaOH solution); and (ii)
concentration corresponds with Ecorr becoming more negative (at an increasing difference between the corrosion potentials of
least where b particles are discontinuous). Daloz et al. [49] associ- the weld and base metals (in Na2SO4 solution).
ated this effect with the reduction of a-Zn with increasing b parti- 4. The apparent influences of Al concentration on the weld metal
cle volume fraction. This clearly does not apply to the present case, corrosion characteristics may be rationalised in terms of the
in which most specimens contained only small traces of Zn. In con- a-Al concentration, which is coupled with the bulk Al concen-
trast, for homogenised Mg–Al alloys, an increase in Al concentra- tration due to the low volume fraction of b particles. The reduc-
tion is usually associated with Ecorr becoming more positive tion in Ecorr and increase in localised corrosion susceptibility
[11,12]. This is generally attributed to substitution of Mg atoms with decreasing weld metal a-Al concentration are consistent
by Al atoms in the surface film to form Mg1 xAlx(OH)2, which is with: (i) degradation of the Mg1 xAlx(OH)2 surface film integ-
thicker and more protective than Mg(OH)2 [51–54]. In the present rity due to Al depletion; and (ii) an increase in the galvanic cur-
case, there was an increase in the Al concentration of the weld me- rent output of second phase particles, respectively.
tal surface films, measured using XPS, with increasing bulk Al con-
centration. This suggests that the influences of weld metal Al
concentration were related to the influence of Al on the surface Acknowledgements
film integrity rather than to the influence of b particles. This seems
intuitive given that the b particle concentrations in the weld met- The authors acknowledge the contributions of BAM staff to this
als were relatively low. research, in particular Mr. Hans-Joachim Kühn and Mr. Romeo
It should also be noted that the cathodic behaviour of second Saliwan-Neumann. Prof. Andrej Atrens of The University of
phase particles is dependent on the a-Al concentration, such that Queensland, Dr. Stan Lynch of The Australian Defence Science
an increase in a-Al reduces the galvanic current output of b and and Technology Organisation and Dr. James Marrow of The Univer-
Al–Mn phases [54]. In the present case, since the b volume concen- sity of Manchester are thanked for their useful advice. The authors
tration was relatively low even in AZ91 WM, the a-Al concentra- also thank MgF Magnesium Flachprodukte GmbH for their in-kind
tion increased with bulk Al concentration. Thus, it may be that support in the form of AZ31 sheet.
N. Winzer et al. / Corrosion Science 51 (2009) 1950–1963 1963

References [32] A. Moccari, C.R. Shastry, An investigation of stress corrosion cracking in Mg


AZ61 alloy in 3.5% NaCl + 2% K2 CrO4 aqueous solution at room temperature, J.
Mater. Technol. (Zeitschrift für Werkstofftechnik) 10 (1979) 119–123.
[1] N. Winzer, A. Atrens, G. Song, E. Ghali, W. Dietzel, K.U. Kainer, N. Hort, C.
[33] S.P. Lynch, P. Trevena, Stress corrosion cracking and liquid metal
Blawert, A critical review of the stress corrosion cracking (SCC) of magnesium
embrittlement in pure magnesium, Corrosion 44 (1988) 113–124.
alloys, Adv. Eng. Mater. 8 (2005) 659–693.
[34] R.S. Stampella, R.P.S. Procter, V. Ashworth, Environmentally-induced cracking
[2] N. Winzer, A. Atrens, W. Dietzel, G. Song, K.U. Kainer, Comparison of the
of magnesium, Corros. Sci. 24 (1984) 325–341.
linearly increasing stress test and the constant extension rate test in the
[35] M. Bobby Kannan, W. Dietzel, C. Blawert, S. Riekehr, M. Koçak, Stress corrosion
evaluation of transgranular stress corrosion cracking of magnesium, Mat. Sci.
cracking behavior of Nd:YAG laser butt welded AZ31 Mg sheet, Mat. Sci. Eng.
Eng. 472A (2008) 97–106.
444A (2007) 220–226.
[3] N. Winzer, A. Atrens, W. Dietzel, V.S. Raja, G. Song, K.U. Kainer, Characterisation
[36] S.R. Agnew, C.N. Tomé, D.W. Brown, T.M. Holden, S.C. Vogel, Study of slip
of stress corrosion cracking (SCC) of Mg–Al alloys, Mat. Sci. Eng. 488A (2008)
mechanisms in a magnesium alloy by neutron diffraction and modelling,
339–351.
Scripta Mater. 48 (2003) 1003–1008.
[4] N. Winzer, A. Atrens, W. Dietzel, G. Song, K.U. Kainer, Fractography of stress
[37] D.S. Shi, I.M. Robertson, H.K. Birnbaum, Hydrogen embrittlement of a
corrosion cracking of Mg–Al alloys, Metall. Mater. Trans. 39 (2008) 1157–1173.
titanium: in situ TEM studies, Acta Metall. 36 (1988) 111–124.
[5] S. Lathabai, K.J. Barton, D. Harris, P.G. Lloyd, D.M. Viano, A. McClean, Welding
[38] D.H. Lassila, H.K. Birnbaum, The effect of diffusive segregation on the fracture
and weldability of AZ31B by gas tungsten arc and laser beam welding
of hydrogen charged nickel, Acta Metall. 36 (1988) 2821–2825.
processes, in: H.I. Kaplan (Ed.), Magnesium Technology 2003, TMS,
[39] D.H. Lassila, H.K. Birnbaum, Intergranular fracture of nickel: the effect of
Warrendale, 2003, pp. 157–162.
hydrogen–sulfur co-segregation, Acta Metall. 35 (1987) 1815–1822.
[6] A. Munitz, C. Cotler, A. Stern, G. Kohn, Mechanical properties and
[40] D.H. Lassila, H.K. Birnbaum, The effect of diffusive hydrogen segregation on
microstructure of gas tungsten arc welded magnesium AZ91D plates, Mat.
fracture of polycrystalline nickel, Acta Metall. 34 (1986) 1237–1243.
Sci. Eng. 302A (2001) 68–73.
[41] I.M. Robertson, T. Tabata, W. Wei, F. Heubaum, H.K. Birnbaum, Hydrogen
[7] N. Winzer, C.E. Cross, On the role of b particles in stress corrosion cracking of
embrittlement and grain boundary fracture, Scripta Metall. 18 (1984) 841–846.
Mg–Al alloys, Metall. Mater. Trans. 40A (2009) 273–274.
[42] C.J. McMahon Jr., Hydrogen induced intergranular fracture of steels, Eng. Fract.
[8] J. Chen, J. Wang, E. Han, W. Ke, In situ observation of crack initiation and
Mech. 68 (2001) 773–788.
propagation of the charged magnesium alloy under cyclic wet–dry conditions,
[43] C.J. McMahon Jr., Brittle fracture of grain boundaries, Interface Sci. 12 (1994)
Corros. Sci. 50 (2008) 2338–2341.
141–146.
[9] J. Chen, J. Wang, E. Han, W. Ke, Effect of hydrogen on stress corrosion cracking
[44] J. Kameda, C.J. McMahon Jr., Solute segregation and hydrogen-induced
of magnesium alloy in 0.1 M Na2SO4 solution, Mat. Sci. Eng. 488A (2008) 428–
intergranular fracture in an alloy steel, Metall. Trans. A 14 (1983) 903–911.
434.
[45] N. Bandyopadhyay, J. Kameda, C.J. McMahon Jr., Hydrogen-induced cracking in
[10] J. Chen, J. Wang, E. Han, J. Dong, W. Ke, States and transport of hydrogen in the
4340-type steel: effects of composition, yield strength, and H2 pressure,
corrosion process of an AZ91 magnesium alloy in aqueous solution, Corros. Sci.
Metall. Trans. 14A (1983) 881–888.
50 (2008) 1292–1305.
[46] Y. Takeda, C.J. McMahon Jr., Strain controlled vs stress controlled hydrogen
[11] G. Song, A. Atrens, M. Dargusch, Influence of microstructure on the corrosion of
induced fracture in a quenched and tempered steel, Metall. Trans. 12A (1981)
diecast AZ91D, Corros. Sci. 41 (1999) 249–273.
1255–1266.
[12] G. Song, A. Atrens, Corrosion mechanisms of magnesium alloys, Adv. Eng.
[47] O. Lunder, K. Nisancioglu, R.S. Hansen, Corrosion of die cast magnesium–
Mater. 1 (1999) 11–33.
aluminium alloys, SAE Technical Paper Series #930 755 (1993).
[13] G. Song, A. Atrens, Understanding magnesium corrosion, Adv. Eng. Mater. 5
[48] O. Lunder, T.Kr. Aune, K. Nisancioglu, Effect of Mn additions on the corrosion
(2003) 837–858.
behavior of mould-cast magnesium ASTM AZ91, Corrosion 43 (1987) 291–295.
[14] O. Lunder, J.E. Lein, T.Kr. Aune, K. Nisancioglu, The role of Mg17Al12 phase in the
[49] D. Daloz, P. Steinmetz, G. Michot, Corrosion behavior of rapidly solidified
corrosion of Mg alloy AZ91, Corrosion 45 (1989) 741–748.
magnesium–aluminium–zinc alloys, Corrosion 53 (1997) 944–954.
[15] L. Fairman, H.J. Bray, Intergranular stress-corrosion crack propagation in
[50] A. Pardo, M.C. Merino, A.E. Coy, R. Arrabal, F. Viejo, E. Matykina, Corrosion
magnesium aluminium alloys, Br. Corros. J. 6 (1971) 170–174.
behaviour of magnesium/aluminium alloys in 3.5 wt.% NaCl, Corros. Sci. 50
[16] W.K. Miller, Stress-corrosion cracking of magnesium alloys, in: R.H. Jones
(2008) 823–834.
(Ed.), Stress Corrosion Cracking: Materials Performance and Evaluation, ASM
[51] J. Chen, J. Wang, E. Han, J. Dong, W. Ke, AC impedance spectroscopy study of
International, Materials Park, 1992, pp. 251–263.
the corrosion behavior of an AZ91 magnesium alloy in 0.1 M sodium sulfate
[17] W.M. Pardue, F.H. Beck, M.G. Fontana, Propagation of stress-corrosion cracking
solution, Electrochim. Acta 52 (2007) 3299–3309.
in a magnesium-base alloy as determined by several techniques, Trans. Am.
[52] J. Chen, J. Wang, E. Han, J. Dong, W. Ke, Corrosion behavior of AZ91D
Soc. Met. 54 (1961) 539–548.
magnesium alloy in sodium sulfate solution”, Mater. Corros. 57 (2006) 789–
[18] W. Österle, Bundesanstalt für Materialforschung und-prüfing, Private
793.
communication.
[53] L. Fairman, H.J. Bray, Transgranular SCC in Mg–Al alloys, Corros. Sci. 11 (1971)
[19] Z. Keshavarz, M.R. Barnett, EBSD analysis of deformation modes in Mg–3Al–
533–541.
1Zn, Scripta Mater. 55 (2006) 915–918.
[54] S. Mathieu, C. Rapin, J. Steinmetz, P. Steinmetz, A corrosion study of the main
[20] ISO 15472:2001, Surface chemical analysis – X-ray photoelectron spectro-
constituent phases of the AZ91 magnesium alloys, Corros. Sci. 45 (2003) 2741–
meters – calibration of energy scales.
2755.
[21] G.L. Makar, J. Kruger, K. Sieradzki, Stress corrosion cracking of rapidly solidified
[55] D. Daloz, P. Steinmetz, G. Michot, Corrosion behavior of rapidly solidified
magnesium–aluminium alloys, Corros. Sci. 34 (1993) 1311–1342.
magnesium–aluminum alloys, Corrosion 53 (1997) 944–954.
[22] G. Ben-Hamu, D. Eliezer, C.E. Cross, T. Böllinghaus, The relation between
[56] D.E. Haycock, M. Kasrai, C.J. Nicholls, D.S. Urch, The electronic structure of
microstructure and corrosion behaviour of GTA welded AZ31B magnesium
magnesium hydroxide (brucite) using X-ray emission, X-ray photoelectron,
sheet, Mat. Sci. Eng. 452–453A (2007) 210–218.
and auger spectroscopy, J. Chem. Soc. Dalton Trans. (1978) 1791–1796.
[23] S.P. Lynch, Mechanisms of intergranular fracture, Mater. Sci. Forum 46 (1989)
[57] H.C. Halder, J. Alonso, W.E. Swartz, Valence and core electron spectra of Mg in
1–24.
MgO in evaporated thin films, Z. Naturforsch. A 30 (1975) 1485.
[24] J.H. Nordlien, S. Ono, N. Masuko, K. Nisancioglu, A TEM investigation of
[58] L. Ley, F.R. McFeely, S.P. Kowalczyk, J.G. Jenkin, D.A. Shirley, Many-body effects
naturally formed oxide films on pure magnesium, Corros. Sci. 39 (1997) 1397–
in X-ray photoemission from magnesium, Phys. Rev. B 11 (1975) 600–612.
1414.
[59] J.C. Fuggle, XPS, UPS and XAES studies of oxygen adsorption on polycrystalline
[25] Y.F. Cheng, J.L. Luo, M. Wilmott, Spectral analysis of electrochemical noise with
Mg at 100 and 300 K, Surf. Sci. 69 (1977) 581–608.
different transient shapes, Electrochim. Acta 45 (2000) 1763–1771.
[60] H. Pulm, G. Hohlneicher, H.-J. Freund, H.-U. Schuster, J. Drews, U. Eberv, Charge
[26] J.L. Dawson, Electrochemical noise measurement: the definitive in-situ
distribution in some ternary vintl phases as studied by X-ray photoelectron
technique for corrosion applications?, in: J.R. Kearns, J.R. Scully, P.R. Roberge,
spectroscopy, J. Less common Metals 115 (1986) 127–143.
D.L. Reichert, J.L. Dawson (Eds.), Electrochemical Noise Measurement for
[61] P. Steiner, H. Hoechst, S. Hufner, XPS investigation of simple metals: I. Core
Corrosion Applications, ASTM, Philadelphia, 1996, pp. 3–35.
level spectra, Z. Phys. B 30 (1978) 129–143.
[27] A. Legat, V. Doleček, Corrosion monitoring system based on measurement and
[62] X.D. Peng, M.A. Barteau, Spectroscopic characterization of surface species
analysis of electrochemical noise, Corrosion 51 (1995) 295–300.
derived from HCOOH, CH3COOH, CH3OH, C2H5OH, HCOOCH3 and C2H2 on MgO
[28] H.A.A. Al-Mazeedi, R.A. Cottis, A practical evaluation of electrochemical noise
thin film surfaces, Surf. Sci. 224 (1989) 327–347.
parameters as indicators of corrosion type, Electrochim. Acta 49 (2004) 2787–
[63] H.B. Yao, Y. Li, A.T.S. Wee, An XPS investigation of the oxidation/corrosion of
2793.
melt-spun Mg, Appl. Surf. Sci. 158 (2000) 112–119.
[29] R.A. Cottis, M.A.A. Al-Awadhi, H.A.A. Al-Mazeedi, S. Turgoose, Measures for the
[64] A. Hess, E. Kemnitz, A. Lippitz, W.E.S. Unger, D.-H. Menz, ESCA, XRD, and IR
detection of localised corrosion with electrochemical noise, Electrochim. Acta
characterization of aluminum oxide, hydroxyfluoride, and fluoride surfaces in
46 (2001) 3665–3674.
correlation with their catalytic activity in heterogeneous halogen exchange
[30] R.A. Cottis, Interpretation of electrochemical noise data, Corrosion 57 (2001)
reactions, J. Catal. 148 (1994) 270–280.
265–285.
[65] I. Olefjord, H.J. Mathieu, P. Marcus, Intercomparison of surface analysis of thin
[31] D.K. Priest, F.H. Beck, M.G. Fontana, Stress-corrosion mechanisms in a
aluminium oxide films, Surf. Interface Anal. 15 (1990) 681–692.
magnesium-base alloy, Trans. Am. Soc. Met. 47 (1955) 473–487.

Das könnte Ihnen auch gefallen