You are on page 1of 8

This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy

Corrosion Science 53 (2011) 1394–1400

Contents lists available at ScienceDirect

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

Electrochemical aspects of exfoliation corrosion of aluminium alloys: The effects


of heat treatment
T. Marlaud a,b, B. Malki a,⇑, A. Deschamps a, B. Baroux a
a
Laboratoire Science et Ingénierie des Matériaux et Procédés (SIMAP), Grenoble INP, UJF, CNRS, 38402 Saint Martin d, Hères, France
b
Rio Tinto Alcan – Centre de Recherches de Voreppe, BP 27, 38341 Voreppe Cedex, France

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

Article history: Electrochemical approaches are used to investigate the exfoliation corrosion (EFC) of a 7XXX series alu-
Received 10 June 2010 minium alloy that has undergone different tempering treatments. EFC was produced under an artificial
Accepted 8 January 2011 crevice at open circuit potential in neutral chloride solutions, and is found to be associated to current
Available online 18 January 2011
and potential transients. EFC was also produced under galvanostatic control conditions. Observations
made through Scanning Electron Microscopy (SEM) suggest that these transients result from the progres-
Keywords: sion of inter-granular cracks. Last, over-ageing heat treatments that are known to decrease both metal
A. Aluminium
hardness and EFC sensitivity were found to decrease the number of transients.
B. Galvanostatic
C. Exfoliation corrosion
Ó 2011 Elsevier Ltd. All rights reserved.
C. Hydrogen absorption

1. Introduction chloride/oxy-chloride complexes, which have higher molar vol-


umes than that of aluminium [12]. This precipitation occurs along
Aluminium alloys with an elongated grain structure parallel to grain boundaries and may cause the surface grains to lift up [13].
the plate surface are known [1] to be sensitive to exfoliation corro- Other possible mechanisms, such as stresses caused by hydrogen
sion. High strength 2XXX and 7XXX series alloys used in aircraft bubbles or hydrogen embrittlement are barely taken into consider-
are particularly sensitive to this form of inter-granular corrosion ation, despite the fact that hydrogen originating from the cathodic
[2], which is considered to be a major cause of airframe degrada- reaction in an acidic electrolyte can be absorbed by the aluminium
tion [3]. For instance, it has been shown that EFC above a certain matrix, inducing then a degradation of its mechanical properties
critical level could significantly decrease fatigue life [4]. The resis- [14–17]. Therefore, the question arises whether hydrogen gener-
tance of such alloys to EFC is generally increased through over-age- ated by the cathodic reaction may also contribute to EFC through
ing treatments, which unfortunately, significantly decrease the the mechanisms described in the literature [18,19].
mechanical strength [1]. In terms of experimentation, various procedures have been pro-
In the last decade, the effect of heat treatment on the corrosion posed to assess the susceptibility of aluminium alloys to EFC. Alloy
behaviour of aluminium alloys has been extensively studied as evi- and aircraft manufacturers commonly use tests such as those set
denced by the works on filiform corrosion [5–8]. More recently the forth in ASTM G66, ASTM G85, and ASTM G34 (the so-called EXCO,
effect of temper on localised corrosion kinetics of 7XXX alloys has i.e. exfoliation corrosion, test) [20–23]. However, most of these
been studied using the foil penetration technique [9], which pre- tests, although widely accepted, yield only qualitative results.
sents the advantage of measuring corrosion rates under various The EXCO test, for instance, is based on the visual examination of
experimental conditions. Some electrochemical transients were a surface corroded in an acidic medium. Thus, a method that would
observed that the authors attributed to the selective grain attack provide a quantitative assessment of EFC sensitivity would be
combined with inter-granular corrosion, both controlled by the useful to optimise alloys and tempering and to predict lifetimes
alloying elements content [10,11]. Having this in mind, we in- [11,24].
tended in this paper to investigate the influence of tempering on This study is an attempt to better understand the EFC mecha-
the EFC behaviour of some new 7XXX aluminium alloys. nisms at work in 7XXX aluminium alloys with different heat
The mechanisms of EFC generally involve some type of mechan- treatment histories, including the standard industrial tempers T6
ical wedging stresses, created by the precipitation of aluminium (used to maximise metal hardness) and T76 (used to improve
EFC resistance). It examines the electrochemical transients
observed in chloride-based electrolytes at open circuit potential
⇑ Corresponding author. Fax: +33 4 76 82 67 67. (OCP), or under galvanostatic, and in some cases under potentio-
E-mail address: Brahim.Malki@simap.grenoble-inp.fr (B. Malki). static conditions. OCP measurements are advantageous in that they

0010-938X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2011.01.010
Author's personal copy

T. Marlaud et al. / Corrosion Science 53 (2011) 1394–1400 1395

135°C All electrochemical measurements were performed on samples


mechanically polished down to 1 lm. The corrosion attacks were
15°C/min mainly performed on the rolling plane Longitudinal/Long-Trans-
verse (L/LT, see Fig. 2a) at one-fourth the thickness of the initial
As received:
12h 24h 48h 96h ~130h plate. Some transverse corrosion attacks were also performed,
6h at 120°C
where the electrolyte is in contact with the Short-Transverse/
T6 T76 Long-Transverse (ST/LT, see Fig. 2b) plane of the substrate.
Fig. 1. Ageing heat treatments used in the present study: various durations all at The OCP measurements were performed in a 1 M NaCl, 0.25 M
135 °C, following an initial 6-h ageing phase at 120 °C. The T6 and T76 tempers NaNO3 aerated aqueous solution at pH 5.5 [25]. The experimental
correspond, respectively, to 12 and 96 h at 135 °C, and will be the focus of this setup included two working electrodes (WE) and a reference elec-
paper. trode (SCE) (Fig. 3a). The two working electrodes were some disks
measuring approximately 20 mm in diameter, the L/LT plane being
in contact with the electrolyte. On one of the two working elec-
measure current and potential fluctuations under free corroding
trodes an ‘‘artificial’’ crevice was created by pressing a polymethyl
conditions, which are closer to in-service conditions than the stan-
methacrylate (PMMA) cylinder onto the sample (see the assembly
dard ASTM tests. The electrochemical tests were supplemented by
in Fig. 3b). The presence of this crevice-forming system served to
extensive SEM fractographic observations.
greatly accelerate initial exfoliation corrosion under open circuit
conditions. The confined area between the PMMA cylinder and
2. Experimental the surface of the specimen was controlled by applying a torque
of 0.02 N m.
Samples were taken from a 25 mm plate of a non-commercial The two working electrodes were connected through a low
aluminium alloy provided by Rio Tinto Alcan. The chemical compo- impedance ammeter, which made it possible to measure the OCP
sition of this ‘‘Al–Zn–Mg–Cu’’ alloy was: 10.3 wt.% Zn, 2.0 wt.% Mg corrosion current. The difference in potential between the two
and 1.6 wt.% Cu. Small additions of Zr and Sc in this alloy ensure electrodes was less than 250 lV, and their potential with respect
that the grain structure is fully fibrous, i.e. that the grain structure to the reference electrode SCE was measured using a high input
is highly elongated parallel to the sample surface (the rolling impedance voltmeter (1013 O). During the corrosion tests, the data
plane), following a rolling and solution treatment. In accordance sampling rate was equivalent to fs = 18 Hz. An anti-aliasing filter
with standard industrial practices, the alloy first underwent a solu- with a cut-off frequency set to 7.66 Hz was used to eliminate un-
tion treatment and was quenched to room temperature in cold wanted high-frequency interference.
water. It was next subjected to a small amount of plastic deforma- Galvanostatic measurements were performed using a standard
tion and aged 6 h at 120 °C (as received condition). Further ageing three-electrode cell (a single working electrode, a platinum auxil-
treatments were then performed at 135 °C with different holding iary electrode and the reference electrode SCE). The working elec-
times, including the T6 and ‘‘over-aged’’ T76 tempers (Fig. 1). trode corresponded to the crevice-forming device. The electrolyte

Fig. 2. Diagram of the two specimens taken from the sheet of metal. The arrows indicate the surfaces tested during the galvanostatic experiments. The acronyms L/LT and ST/
LT stand for longitudinal/long-transverse and short-transverse/long-transverse.

im
(a) (b)
ZRA
Crevice

sample PMMA support


V1 V2
Fig. 3. (a) Diagram of the experimental setup with two working electrodes (subscripts 1 and 2) connected through a low resistance ammeter (ZRA). im is the measured
current through the ammeter and V1 = V2 = V is the measured potential; (b) a diagram of the screw/wing system of the artificial crevice assembly.
Author's personal copy

1396 T. Marlaud et al. / Corrosion Science 53 (2011) 1394–1400

-708 equal to that obtained at the end of the EXCO test [26–29] and also
corresponds to that of the crack tip solution measured during the
environmentally-assisted cracking of alloy AA7050 [30]. Prelimi-
8 -709 narily experiments were conducted to optimise the applied current
IG, in order to reduce experimental time and achieve noticeable

Potential (mV/SCE)
V(t) exfoliation corrosion morphologies. The IG value of 2.5 mA was
Intensity (µA)

-710 subsequently used for all experimentation.

4
-711 3. Results
incubation time

I(t) 3.1. Open circuit testing


-712
The OCP experiments involved a 64-h period of immersion for
0 the T6-tempered specimen. Analysis of the current and potential
-713 signals (Fig. 4) showed that electrochemical activity occurred after
0 20 40 60
a long period of incubation (20 h). This activity was characterised
Time (h)
by an increase in current, associated with a drop in potential,
Fig. 4. Current and potential transients occurring during the OCP measurements of and followed by a gradual return to steady state values. The first
the T6 specimen in the 1 M NaCl, 0.25 M NaNO3 electrolyte. Exfoliation corrosion idea which comes in mind is to suspect that these transients orig-
occurs after 20 h of incubation. inate from metastable pitting [31–33]. However, micrographic
observations suggest that these transients reveal something more.
Indeed, exfoliation corrosion was clearly evident on the working
electrode by way of the SEM observations performed at the end
of the test (Fig. 5). Moreover, careful observations during the early
stages of the experiment indicated that some blisters appear in the
form of ‘‘protuberances’’ (Fig. 6), in the absence of any visible cor-
rosion product around the site, suggesting a possible effect of
hydrogen.

3.2. Galvanostatic measurements

Galvanostatic measurements were first carried out for 24 h on


both the T6 and T76 samples. In keeping with the EXCO test results
[26], only the T6 samples developed clear exfoliation corrosion
morphology (lifted grains). No visible corrosion layers were ob-
served on the T76 sample, which evidenced only a powder-like
corrosion over the entire surface. Fig. 7 shows the potential, re-
Fig. 5. SEM image of the T6 specimen with the artificial crevice assembly, after 48 h
of immersion in the 1 M NaCl, 0.25 M NaNO3 electrolyte at open circuit potential.
corded over time, for the two samples. The T6 temper shows sev-
eral fluctuations in potential, each of them consisting of a sharp
drop in potential followed by a gradual return to a steady state va-
used was based on that of the OCP measurements and was supple- lue. This variation corresponds, more or less, to the time dependent
mented by 0.033 M AlCl3 in order to buffer the variation of Al3+ cat- function: V 0  DV m :½1  expðt=sÞ where s is a time constant,
ions and to hold the pH stable at 3.2. This value is approximately equalling approximately 100 s. Note also that the asymptotic value

Fig. 6. SEM images of the early stages of exfoliation corrosion on the T6 specimen obtained during the OCP tests. Blisters generally take the form of ‘‘protuberances’’ as shown
in the close-ups.
Author's personal copy

T. Marlaud et al. / Corrosion Science 53 (2011) 1394–1400 1397

(a) -600
Potential (mV/SCE)

-640
T6

-680

T76

-720
0 4 8 12
Time (h)

(b)
-620 3 min
T6
Fig. 9. Potential recorded during the galvanostatic experiments carried out for heat
Potential (mV/SCE)

treatments ranging from peak-aged temper T6 to over-aged temper T76. The details
-640 for the heat treatments are shown in Fig. 1.

-660
4000

T76
-680
3000
Cumulative transients

8000 8100 8200 8300 8400 8500


Time (s)

Fig. 7. (a) Evolution of the measured potential during the galvanostatic experi- 2000
ments for both the T6 and T76 specimens. Note the large fluctuations in potential
for the T6 specimen, as illustrated in close-up (b).

1000
of the potential stays relatively constant throughout the whole
test. For the T76 temper, no such transients were observed. This
transient behaviour was also observed in potentiostatic mode, with 0
applied potentials corresponding to the quasi-steady state values 10 100
achieved during the galvanostatic tests: 620 mV/SCE for T6 and Overaging time (h)
680 mV/SCE for T76. As expected, the recorded current is similar
Fig. 10. Impact of the ageing time on the total number of transients greater than
in both cases (with a mean value close to 2.5 mA), and well-defined 2 mV (threshold) recorded during the 24 h galvanostatic experiment.
current transients (1 min of lifetime) were observed only for the
T6 temper (Fig. 8).
From these initial observations, it is tempting to consider these investigate this possibility, galvanostatic experiments were carried
transients as characteristic of EFC susceptibility. In order to further out at intermediate ageing intervals between the T6 and T76 tem-
pers, specifically after different holding times during the 135 °C
heat treatment, which is known to modify the EFC sensitivity of
3.2 the alloy. The potential transients recorded during these experi-
ments are reported in Fig. 9. As with the previous experiments, it
was observed that both the average potential and the magnitude
2.8 -620 mV/SCE, T6 of the transients decrease as ageing time increases. Additionally,
a logarithmic decrease in the total number of transients, as a func-
Intensity (mA)

tion of time, is also observed (Fig. 10), i.e. as ageing time increases,
the number of transients and EFC sensitivity both decrease.
2.4
1.5 min
3.3. Micrographic observations

2
After the galvanostatic treatment, the specimens were im-
mersed for 24 h in a large volume of pure water before being ob-
-680 mV/SCE, T76 served with SEM. Fig. 11 shows the cross sections of the T6 and
T76 specimens after 90 min of galvanostatic testing. The corrosion
3.38 3.4 3.42 3.44 morphologies are significantly different; they are quite localised
Time (h)
for the T6 temper specimen, and more uniform for the T76 temper
Fig. 8. Transient current recorded during the potentiostatic tests for the T6 and T76 specimen. The corrosion is deeper for the T6 specimen: 100 lm
samples. below the surface of the sample, compared to 50 lm for the T76
Author's personal copy

1398 T. Marlaud et al. / Corrosion Science 53 (2011) 1394–1400

Fig. 11. (a) FEG-SEM images of a cross section of the T6 sample after 3/2 h of galvanostatic testing, evidencing long inter-granular cracks with non-dissolved precipitates
(indicated by arrows) revealed using the chemical contrast imaging mode; (b) FEG-SEM images of a cross section of the T76 sample after 3/2 h of the same test, evidencing
more diffuse inter and inter-sub-granular corrosion.

(a) (b) 3000


a)
T6
2400
Mean depth (µm)

1800

1200

T76
600

0
0 10 20 30 40 50
Time (h)

Fig. 12. (a) Illustration of the corrosion depths measured on the cross sections of both the T6 and T76 samples for statistical analysis of EFC kinetics. (b) Variation of the mean
value as a function of time.

specimen. The long cracks observed in the T6 sample extend in the though there were less events than in the longitudinal tests (L/LT
rolling direction, along the grain boundaries with non-dissolved plane). SEM images of the cross section reveal band-like corrosion
precipitates. In these backscattered electron images, grain bound- zones with some long inter-granular cracks (200–300 lm) that ex-
aries are dark in colour due to the presence of precipitate-free tend far into the unaffected metal (Fig. 13). Moreover, when these
zones with bright spots caused by grain boundary precipitates. samples were aged under air (50% humidity ratio) after the test,
By contrast, the corrosion morphology of the T76 specimen con- cracks still propagated in the rolling direction until they crossed
sists of inter- and inter-sub-granular corrosion paths, which re- the whole sample after 15 days, as shown in Fig. 14a. Fractographic
veals preferential anodic dissolution of the grains (Fig. 11b). No investigations (Fig. 14b, and images at a higher magnification in
visible cracks can be observed. Thus, it appears that the over-aged Fig. 14c and d) show that the damage consists of mostly brittle, in-
T76 specimen is sensitive only to inter- and inter-sub-granular cor- ter-granular cracks with no evident corrosion attack.
rosion. Lastly, a statistical analysis of the average corrosion depth,
as measured on the cross section of both the T6 sample and T76
4. Discussion
sample, was performed and the mean depth was found to be much
less for the T76 specimen than for the T6 specimen (Fig. 12).
Our results show that the appearance of exfoliation corrosion,
either under OCP conditions underneath an artificial crevice or un-
3.4. Effect of surface orientation der galvanostatic or potentiostatic control, systematically corre-
lates with the presence of well-defined voltage or current
In order to evaluate the possible effects of lift-out stress caused transients. The magnitude of these transients, as well as their
by the corrosion products formed during EFC, corrosion tests were amplitude, is shown to be a major parameter that parallels im-
performed under the same conditions along the transverse orienta- proved EFC susceptibility when prolonging heat treatment beyond
tion of the T6 specimen, on the LT-ST plane. In this scenario, grains T6. For the future, this type of measurement opens the door to the
would, presumably, be less prone to lift up since the elongated quantitative evaluation of EFC susceptibility for a wide range of al-
grains are normal to the grain surface (see Fig. 2b for the geometric loys and corrosion media. However, it is first necessary to under-
configuration of the sample). The potential transients recorded stand the mechanisms giving rise to these transients and their
(not reported here) still exhibit well-defined transients, even relation to exfoliation events.
Author's personal copy

T. Marlaud et al. / Corrosion Science 53 (2011) 1394–1400 1399

mely fast once grain boundary strength is exceeded, creating sud-


den contact between the electrolyte and the fresh metallic surface,
promoting in turn the dissolution of the aluminium matrix. Such
an elementary crack advancing step is expected to be at the origin
of the transients, e.g. the intense electrochemical activity observed
during the galvanostatic measurements. The transients can be
regarded as the triggering response of the experimental setup to
a crack occurrence, and the gradual return of the potential to a
steady state value as the result of the formation of aluminium chlo-
ride/oxy-chloride complexes [35,36] on the bare surface thereby
limiting the anodic dissolution kinetics.
The question remains as to whether the stresses causing these
cracks are due to the volumic effect of solid corrosion products.
No such corrosion products were observed in the experiments
reported here, thus suggesting the possibility of a hydrogen effect.
In the acidic medium formed in a corrosion zone due to the
hydrolysis of the aluminium cations, Al dissolution is likely
accompanied by hydrogen emissions, depending on the local pH
and potential at the bottom of the corroding zone [37]. Hydrogen
bubbling was, in fact, clearly visible during and after each test.
The molecular recombination of these hydrogen atoms at or near
grain boundaries is assumed to generate enough pressure to in-
duce crack initiation. This scenario would be consistent with the
fact that very long and extremely thin cracks are observed far
from the corrosion location (in locations where it is highly unli-
kely that uplifting mechanical stresses due to corrosion products
would be have much effect). However, most importantly, it would
be consistent with the observation that similar cracks are
observed when corrosion proceeds along the long direction of
Fig. 13. SEM images of a cross section (on the L/ST plane) of the T6 sample after a the grains normal to the surface of the sample, in which case no
brief period (6 h) of transverse galvanostatic testing. Note the band-like form of the significant external stress is exerted on the grain boundaries by
selective dissolution along grain boundaries. the corrosion products. Lastly, if a hydrogen evolution reaction
is behind the local stress producing the cracks, it is significant that
The average rate of cracks propagation (Fig. 12) corresponds to the presence of a crevice, known to produce local acidification,
around 1–2 lm/min, not unlike what is frequently observed in was found to favour EFC initiation in a nearly neutral electrolyte
stress corrosion cracking [34]. Such cracks may propagate extre- under open circuit conditions.

Fig. 14. (a) Images of a cross section of the T6 sample after 6 h of transverse galvanostatic testing followed by 30 days under air (50% humidity ratio), showing the extension
of thin cracks throughout the entire sample; (b) shows the fractograph after fracturing the sample in (a); (c) and (d) are magnifications of (b).
Author's personal copy

1400 T. Marlaud et al. / Corrosion Science 53 (2011) 1394–1400

From a metallurgical viewpoint, increasing the duration of the [2] M.J. Robinson, N.C. Jackson, Corros. Sci. 41 (1999) 1013–1028.
[3] J.P. Chubb, T.A. Morad, B.S. Hockenhull, J.W. Bristow, Int. J. Fatigue 17 (1995)
ageing treatment beyond the T6 temper modifies the size and com-
49–54.
position of the precipitates and, consequently, the composition of [4] M. Liao, G. Renaud, N.C. Bellinger, Int. J. Fatigue 29 (2007) 677–686.
the surrounding matrix. This modifies in turn the average electrode [5] A. Afseth, J.H. Nordlien, G.M. Scamans, K. Nisancioglu, Corros. Sci. 43 (2001)
potential [38], which may explain the marked decrease of this 2093–2109.
[6] A. Afseth, J.H. Nordlien, G.M. Scamans, K. Nisancioglu, Corros. Sci. 44 (2002)
potential with the ageing holding time. Moreover, the ageing treat- 145–162.
ment also modifies the chemical composition in the neighbourhood [7] A. Afseth, J.H. Nordlien, G.M. Scamans, K. Nisancioglu, Corros. Sci. 44 (2002)
of the grain boundaries, in the so called ‘‘precipitate-free zone’’, 2491–2506.
[8] J.T.B. Gundersen, A. Aytac, J.H. Nordlien, K. Nisancıoglu, Corros. Sci. 46 (2004)
thereby increasing their toughness, resulting in fewer cracks and 697–714.
electrochemical transients. [9] H. Huang, G.S. Frankel, Corrosion 63 (2007) 731–743.
[10] Qingjiang Meng, G.S. Frankel, J. Electrochem. Soc. 151 (2004) 271–283.
[11] X. Zhao, G.S. Frankel, Corros. Sci. 49 (2007) 920–938.
5. Conclusions [12] D.J. Kelly, M.J. Robinson, Corrosion 49 (1993) 787–795.
[13] M.J. Robinson, Corros. Sci. 23 (1983) 887–899.
[14] P.V. Petroyiannis, Al.Th. Kermanidis, P. Papanikos, Sp.G. Pantelakis, Theor.
To summarise, this work shows that:
Appl. Fract. Mech. 41 (2004) 173–183.
[15] N.D. Alexopoulos, P. Papanikos, Mater. Sci. Eng. A 498 (2008) 248–257.
(i) In quasi-neutral electrolyte and under open circuit condi- [16] Amjad Saleh El-Amoush, J. Alloys Compd. 443 (2007) 171–177.
tions, the presence of a crevice favours the development of [17] P.V. Petroyiannis, A.T. Kermanidis, P. Papanikos, S.G. Pantelakis, Theor. Appl.
Fract. Mech. 41 (2004) 173–183.
exfoliation corrosion in the 7XXX series aluminium alloy in [18] H. Kamousti, G.N. Haidemenopoulos, V. Bontozoglou, V. Pantelakis, Corros. Sci.
question. This corrosion is characterised by the presence of 48 (2006) 1209–1224.
current and potential transients with a spectrum depending [19] ASTM-G34 (1974). Standard Test Method for Exfoliation Corrosion
Susceptibility in 2XXX and 7XXX Series Aluminum Alloys.
upon the heat treatment of the alloy. [20] ASTM-G66-99 (2005). Standard Test Method for Visual Assessment of
(ii) Theses transients were also observed under galvanostatic Exfoliation Corrosion Susceptibility of 5XXX Series Aluminum Alloys (ASSET
control conditions. Increasing the ageing time beyond the Test).
[21] ASTM-G85 (2002). Standard Practice for Modified Salt Spray (Fog) Testing.
T6 temper, which is known to drastically decreases the sen- [22] B.W. Lifka, D.O. Sprowls, Corrosion 22 (1966) 7–15.
sitivity to EFC was found to significantly decrease the num- [23] D.O. Sprowls, J.D. Walshal, Simplified exfoliation testing of aluminium alloys,
ber of transients as well as average electrode potential. localized corrosion-cause of metal failure, ASTM STP 516 (1972).
[24] E.A.G. Liddiard, J.A. Whittaker, J. Inst. Met. 89 (1960) 377–384.
(iii) SEM observation performed after the galvanostatic tests
[25] M.C. Reboul, J. Bouvaist, Werstoffe und Korrosion 30 (1979) 700–712.
revealed the presence of inter-granular cracks, indicating [26] T. Marlaud, B. Malki, A. Deschamps, M. Reboul, B. Baroux, ECS Trans. 3 (31)
that the dominant EFC mechanism is discontinuous and sug- (2007) 285–293.
[27] C.J. Baes, R.E. Mesmer, The Hydrolysis of Cations, second ed., R.E. Krieger
gesting that the propagation of exfoliation corrosion is
Publishing Company, 1976.
related to a fracture mechanism. The observed transients [28] A. Turnbull, Chemistry within localized corrosion cavities, Proceedings of the
are believed to result from the crack propagation. The com- Second International Conference on Localized Corrosion 9 (1990) 359–373.
parison between longitudinal and transverse corrosion tests [29] S. Lee, B.W. Lifka, ASTM STP 1134, Am. Soc. Test. Mater. (1992) 1–19.
[30] K.R. Cooper, L.M. Young, R.P. Gangloff, R.G. Kelly, Mat. Sci. F. 331 (2000) 1625–
also suggests that corrosion products have no impact on 1634.
cracks initiation, whereas there is reason to suspect that [31] K. Sasaki, H.S. Isaacs, J. Electrochem. Soc. 151 (2004) 124–133.
hydrogen recombination plays a decisive role. [32] S. Horlé, B. Malki, B. Baroux, J. Electrochem. Soc. 153 (2006) 527–532.
[33] H.S. Isaacs, C. Scheffey, R. Huang, ECS Trans. 11 (22) (2008) 1–12.
[34] W. Dietzel and K.-H. Schwalbe, Application of the rising displacement test to
Scc investigations, slow strain rate testing for the evaluation of
Acknowledgements environmentally induced cracking, in: R.D. Kane (Ed.), Cracking: Research
and Engineering Applications, ASTM STP 1210, Philadelphia, 1993, pp. 134–
148.
C. Henon (Alcan), T. Warner (Alcan) and M. Reboul (SIMAP) are
[35] G.C. Wood, W.H. Sutton, J.A. Richardson, T.N.K. Riley, A.G. Malherbe,
warmly thanked for stimulating discussions. Corrosion 3 (1974) 526–546.
[36] M.C. Reboul, T.J. Warner, H. Mayet, B. Baroux, Corros. Rev. 15 (1997) 471–496.
[37] M. Boinet, J. Bernard, M. Chatenet, F. Dalard, S. Maximovitch, Electrochim. Acta
References
55 (2010) 3454–3463.
[38] E.H. Hollingsworth, H.Y. Hunsicker, Mater. Handb. 9th ed. 1987.
[1] R. Develay, Traitements thermiques des alliages d’aluminium’’ in Techniques
de l’ingénieur, M1290, M1291.

You might also like