Corrosion Science 48 (2006) 1209–1224

www.elsevier.com/locate/corsci

Corrosion-induced hydrogen embrittlement

in aluminum alloy 2024
H. Kamoutsi a, G.N. Haidemenopoulos a,*
,
V. Bontozoglou a, S. Pantelakis b
a
Department of Mechanical and Industrial Engineering, University of Thessaly, 38334 Volos, Greece
b
Department of Mechanical and Aeronautical Engineering, University of Patras, 26110 Rio, Greece

Received 18 August 2004; accepted 13 May 2005

Available online 19 August 2005

Abstract

The present paper focuses on the observed corrosion-induced embrittlement of alloy 2024
and tries to answer the key question on whether the observed embrittlement is attributed to
hydrogen uptake and trapping in the material. Hydrogen is produced during the corrosion
process and is being trapped in distinct energy states, which correspond to different micro-
structural sites. The formation of a hydrogen-affected zone beneath the corrosion layer is sup-
ported by fractographic analysis. Removal of the corrosion layer leads to complete restoration
of yield strength but only partial restoration of ductility. Additional heat treatment to release
the trapped hydrogen leads only to complete restoration of ductility.
 2005 Elsevier Ltd. All rights reserved.

Keywords: A. Aluminium; A. 2024-T351; B. AFM; B. SEM; C. Exfoliation corrosion; C. Hydrogen

embrittlement; C. Hydrogen trapping; C. Fractography

*
Corresponding author. Tel.: +30 24210 74061/74062; fax: +30 24210-74061.
E-mail address: hgreg@mie.uth.gr (G.N. Haidemenopoulos).

0010-938X/$ - see front matter  2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2005.05.015
1210 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224

1. Introduction

The structural integrity of aging aircraft structures can be affected by corrosion.

As the time of an aircraft in service increases, there is a growing probability that cor-
rosion will interact with other forms of damage, such as single fatigue cracks or mul-
tiple-site damage. The aging aircraft may have accumulated corrosion damage over
the service life and its residual strength depends on possible degradation stemming
from corrosion-induced embrittling mechanisms. One characteristic example where
failure was attributed to multi-site damage (MSD) has been the Aloha Airlines acci-
dent in 1988. Damage was attributed to growth and linkage of multiple fatigue
cracks, emanating from rivet holes [1]. Recent investigations on fire fighting planes
of the Hellenic Aerospace Industry has also shown considerable corrosion damage
around rivet holes [2].
There are two key questions regarding this issue: (1) Is there a corrosion-induced
degradation of ductility, which in turn degrades damage tolerance and the residual
strength of aerostructures? and (2) What is the underlying corrosion-induced embrit-
tling mechanism? The answer to the first question has been given by a series of
experiments [3–5], involving mechanical testing of pre-corroded alloy 2024. It was
shown that (i) degradation of ductility and of fatigue life increases with corrosion
exposure time and (ii) removal of the corrosion layer restores strength but not
ductility. These results were attributed to the operation of a bulk corrosion-induced
embrittlement mechanism, and it was suggested that hydrogen might be a possible
underlying cause.
Other researchers have also considered hydrogen as an embrittlement mecha-
nism in Al-alloys. Studies by Scamans et al. [6] of Al embrittlement in humid
air, argued for a major role of hydrogen. In particular, the intergranular crack path
and the reversibility of the phenomenon (recovery of ductility after degassing) sup-
ported a hydrogen rather than an anodic dissolution mechanism. Also, Scamans
and Tuck [7] measured hydrogen permeability and stress corrosion resistance of
the Al–Mg–Zn alloy, as functions of quench rate and aging treatment, and found
similar trends. Speidel [8] reviewed results up to 1992, mainly for Al–Mg–Zn alloys.
More recently, Young and Scully [9] considered the kinetics of crack growth of
aluminum alloy 7050 in a humid air environment and confirmed that hydrogen
embrittlement was the controlling mechanism. Jones [10] summarized evidence
for hydrogen uptake and its contribution to crack growth for the low-strength alloy
5083.
Hydrogen is produced by surface corrosion reactions and part of it is absorbed in
atomic form into the material [11]. In particular, the production of atomic hydrogen
by a single-electron transfer process, according to the reaction
H2 O + e ! OH + H ð1Þ
makes water an aggressive environment for aluminium alloys [9]. Absorbed hydro-
gen diffuses towards the interior of the material and may be retained at various pref-
erential locations. More specifically, it has been shown [12,13] that lattice defects
(vacancies, dislocations, grain boundaries) and precipitates provide a variety of trap-
 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224 1211

ping sites. Hydrogen traps are mechanistically classified as reversible and irreversible
[14], depending on the steepness of the energy barrier needed to be overcome by
hydrogen to escape from them.
Thermal desorption has been successfully used to study hydrogen diffusion and
trapping in pure aluminium [15,16], Al–Cu, Al–Mg2Si [13] and Al–Li alloys [17].
It has also been combined with accelerated corrosion tests in order to characterize
corrosion and hydrogen absorption in alloy 2024 [18,19]. In the last two works,
the existence of multiple trapping states was verified and the quantity and evolution
pattern of hydrogen was discussed. The goal of the present study is to link hydrogen
uptake and trapping to material embrittlement.

2. Experimental procedures

The material used for the present study was alloy 2024-T351, supplied in thick-
nesses 1.6–3.0 mm. The chemical composition of the alloy (wt.%) is: Al–4.35Cu–
1.5Mg–0.64Mn–0.5Si–0.5Fe. Exfoliation corrosion testing (EXCO) was performed
according to ASTM specification G34-90 [20]. It included exposure at 25 ± 0.5 C,
in a solution containing 234 g NaCl, 50 g KNO3 and 6.3 ml concentrated HNO3
(70 wt.%) diluted to 1 L of distilled water. Exposure times in the EXCO solution ran-
ged from 15 min to 96 h. Specimen cleaning after removal from the corrosive solu-
tion involved soaking in concentrated HNO3 for 5 min, rinsing in water, then in
acetone and thoroughly drying in a purge of warm air. The cleaning process aimed
at the complete removal of corrosion products and its duration did not exceed
10 min.
The early stages of corrosion were studied by AFM, while the later stages by
SEM and metallographic sectioning. An in-house thermal desorption/gas chroma-
tography system was employed in order to determine hydrogen being trapped in
the alloy during corrosion. The corroded specimens were introduced in a thermo-
stated furnace and held under inert nitrogen purge. A constant heating rate of
5 C/min was applied, and the amount of hydrogen evolved was measured as a
function of specimen temperature. The following measures were taken to guarantee
that detected hydrogen originates from outgassing from the alloy microstructure: In
order to exclude possible effect of containment, we conducted blind experiments
with all the containment in place but no specimens, and detected no measurable
amount of hydrogen. In order to exclude possible effect of corrosion products, we
used the above described meticulous cleaning procedure. Optical microscopy before
and after cleaning confirmed removal of all visible corrosion products. Further,
hydrogen measurements of (uncorroded or slightly corroded) specimens before
and after cleaning produced identical results. This is taken to confirm lack of inter-
ference of both the cleaning method and the amount of corrosion products with
hydrogen measurement. Finally, to exclude possible effect of carrier gas, we have
used chromatographic purity N2, and have regularly confirmed the absence of O2
from the containment atmosphere. Calibration of the measurements was performed
1212 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224

by the use of standard mixtures (1000 ppm H2 in N2) certified by specialty gas
providers.
Hydrogen profiling was also performed, by taking hydrogen measurements from
multiple corroded specimens that were subjected to progressively deeper material re-
moval. The removal was performed by grinding on SiC abrasive wheels of 1000 grid.
Five identical specimens from the 2.4 mm thick plate were exposed to the EXCO cor-
rosion test for 24 h. The first specimen was measured for hydrogen immediately after
cleaning, while the others were ground on all six sides by 50 lm, 100 lm, 200 lm and
350 lm respectively.
Microhardness testing was performed in corroded and reference specimens, which
were heated to various temperatures simulating the thermal desorption cycle of
hydrogen measurements. These data were set in perspective with the characteristics
of hydrogen trapping states. Microhardness measurements in the corroded speci-
mens were taken after mechanical removal of the corroded layer. In order to mini-
mize possible hardening effect of the removal process, a slow speed, low-pressure
polishing technique was employed. Tensile testing of corroded specimens was per-
formed according to ASTM E8m-94a [21] specification. For the tests a 200 KN
Zwick universal testing machine and a servohydraulic MTS 250 KN machine were
used. The strain rate was 104 s1. The tensile tests were conducted in the University
of Partas and have been previously reported [3,22]. The results were adopted here.
Fractography of tensile specimens was performed using scanning electron micro-
scopy in order to identify the mode of fracture and provide evidence for the relation
of hydrogen uptake to the deterioration of ductility.

3. Results and discussion

3.1. Microstructural description of corrosion

Corrosion in this alloy starts in the form of pitting. Pits begin to appear as early as
after 15 min of exposure and are mainly located at intersections of cracks in the pro-
tective surface oxide layer (Fig. 1a). With exposure time pits become deeper and start
to be connected by a network of intergranular corrosion paths (Fig. 1b). This process
of pit-to-pit interactions leads to pit clustering and coalescence (Fig. 1c). From that
point on, corrosion does not penetrate much deeper but instead spreads beneath the
surface and causes exfoliation of surface layers (Fig. 1d).
The above type of damage evolution—and more specifically the intergranular net-
work following pit growth—is responsible for the transport of corrosive solution
deep into the material, where it reacts producing hydrogen. Thus, hydrogen is gen-
erated at the front of the corrosion layer (for example at the bottom of pits) and
spreads to the adjacent unaffected material establishing a hydrogen diffusion zone
below the corrosion zone. The depth of attack, i.e. the deepest location of the corro-
sion front as a function of time, is determined by metallographic sections. The value
after 24 h of exposure to EXCO is 350 lm, as shown in Fig. 1e.
 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224 1213

3.2. Hydrogen evolution with heating

Results of thermal desorption of hydrogen from corroded specimens are plotted

in Fig. 2a. The pattern of evolution is a strong function of the specimensÕ tempera-
ture, indicating the existence of multiple trapping sites. Different curves correspond
to specimens with varying exposure time to the exfoliation solution. It is evident that
the quantity of hydrogen increases drastically with exposure. Fig. 2b is a magnifica-
tion of the y-axis and shows that a similar dependence on exposure time also holds
for the low-temperature part of the spectrum.
Four major peaks are identified, in agreement with the experiments of Charitidou
et al. [19]. The onset of the peaks labelled as T2, T3 and T4 occurs respectively at
200, 410 and 495 (±10) C. The existence of critical temperatures, below which no

Fig. 1. (a) Alloy 2024-T3 (LT plane), 15 min in the corrosive solution EXCO. 3D depiction of corrosion,
initiation of pitting (arrow A) (50 · 50 lm scan). (b) Alloy 2024-T3 (LT plane), for 4 h in the corrosive
solution exco. Initiation of intergranular corrosion (arrow A). (c) SEM topographies of Alloy 2024 (LT
plane) exhibiting pit-to-pit interactions and pit clustering after exposure in EXCO for (a) 8 and (b) 12 h.
(d) Alloy 2024, 24 h EXCO exposure, showing exfoliation (LT direction). (e) Determination of depth of
attack (LT direction).
1214 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224

Fig. 1 (continued )

hydrogen evolution takes place, classifies these traps as irreversible [14]. Trapping
state T1 is found to release hydrogen continuously at temperatures close to ambient
and is considered a reversible trap.
The total quantity of hydrogen liberated from each trap is estimated by integrat-
ing the area under the respective peak. Results of total hydrogen quantity as a func-
tion of exposure time in the corrosive solution are shown in Fig. 3. The three
irreversible traps saturate, indicating depletion of available trapping sites, in contrast
to the reversible trap, T1. Trapping state T4 saturates first and also has the highest
desorption temperature. These observations support the conclusion that T4 has the
highest binding energy and thus is energetically favored.
The tentative microstructural origin of the above traps has been previously dis-
cussed [19]. Trapping state T1 has been associated with interstitial lattice sites, in
agreement with detailed measurements of Young and Scully [16]. Trap T2 has been
associated with the semi-coherent interfaces of the strengthening phases or the inco-
herent interfaces of dispersoids and the matrix lattice. Trap T3 has been attributed to
the formation of Mg hydride, as proposed by Scamans and Tuck [7]. Finally,
 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224 1215

7.5
T4
7.0 2 Hours
6.5 4 Hours 48h
6.0 8 Hours
5.5 12 Hours
5.0 24 Hours
4.5 96 Hours 96h
4.0 48 Hours
µgH2 /min

3.5 T3 8h
3.0
2.5
2.0 12h
1.5 24h
1.0 T2
T1 4h
0.5
0.0 2h
-0.5

0 100 200 300 400 500 600

a Temperature (oC)

2 Hours T4
1.0
T3
4 Hours
8 Hours
12 Hours 8h
24 Hours
96 Hours
12h
T2
48 Hours
µgH2 /min

0.5 96h 24h

4h

T1
48h
2h

0.0

0 100 200 300 400 500 600

b Temperature (oC)

Fig. 2. (a) Desorption of H2 in specimens of aluminium alloy 2024-T3 for continuous heating up to
600 C. Thickness of material 1.8 mm. (b) Magnification of (a) in the region up to 1.0 lgH2/min.

trapping state T4 starts to release hydrogen at a temperature coinciding with the

dissolution of the strengthening Al2CuMg (S) phase. The ability of all the above
microstructural elements to serve as hydrogen traps has been demonstrated by Sai-
toh et al. [13] using tritium autoradiography.
1216 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224

10
8
8
6
6
µgH2

µgH2
4
4

2 2

0 0
T1 T2

0 20 40 60 80 100 120 0 20 40 60 80 100 120

Corrosion time, h Corrosion time, h

80 400
70 350
60 300
50 250
40
µgH2

200
µgH2

30
150
20
100
10
50
0
T3 T4
-10 0
0 20 40 60 80 100 120
0 20 40 60 80 100 120
Corrosion time, h Corrosion time, h
Fig. 3. Amount of hydrogen desorbed from the four trapping states (T1–T4) as a function of corrosion
exposure time.

A closer look at Fig. 2a indicates the possible existence of additional trapping sites,
manifested by two shoulders between traps T2 and T3. The amount of hydrogen asso-
ciated with these states is evidently too small compared with the amount related to the
main traps T3–T4, and the former are masked by the latter. The origin of these minor
states may be related to dislocations and vacancies respectively of the matrix, as has
been demonstrated by thermal desorption measurements in high purity aluminum [16].
Another key question addressed by the present experiments is how deep into the
material does hydrogen penetrate. Sequential sectioning of specimens exposed to
EXCO for 24 h provides the hydrogen profile. The quantity of hydrogen remaining
in the specimens as a function of the depth of material removed is depicted in Fig. 4.
The amount of hydrogen in the uncorroded material is also shown, and indicates
that the hydrogen affected zone extends beyond 350 lm, which is the corrosion depth
of attack shown in Fig. 1e. Based on the above data, an estimate of the depth of
hydrogen penetration is 400–500 lm.
There are some fundamental difficulties in the application of diffusion analysis to
interpret/predict the above results. First, hydrogen is generated at the front of the
 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224 1217

60

µgH2 40

20

uncorroded

0
-50 0 50 100 150 200 250 300 350 400 450
Depth (µm)

Fig. 4. Total amount of H2 (lg) for different removal depths. 24 h exposure to EXCO solution.

corrosion zone and part of it diffuses towards the unaffected material, establishing a
hydrogen-affected zone. However, as the experiment progresses, the boundary be-
tween the two zones is continuously displaced towards the interior: hydrogen-
affected material becomes corroded (possibly releasing the trapped hydrogen), while
new material is affected. Second, because of the highly non-uniform shape of the cor-
rosion zone (see Fig. 1e), uncorroded but hydrogen-affected material is removed, to-
gether with corrosion products, during sectioning. Thus, the amounts of hydrogen
shown in Fig. 4 correspond mostly to the deeper regions of the affected zone. Third,
there is a four-order-of-magnitude scatter (1011–1015 m2/s) in reported values of
the diffusivity of hydrogen in aluminum [23,24].
Keeping the above complications in mind, we may use the experimentally deter-
mined penetration depth of 400–500 lm to estimate by naive transient diffusion ana-
lysis an effective diffusivity of

d2
Deff ¼  1013 m2 =s ð2Þ
16t
where d = hydrogen penetration depth and t = diffusion time ( exposure time
24 h).
This value should be interpreted as a trap-affected diffusivity, because the speci-
mens in the present work are initially free of hydrogen and become charged during
the corrosion process. Recent experiments by Young and Scully [16] have helped ex-
plain the large scatter of H2 diffusivity values, and in particular have shown that the
activation energy of dislocation traps in pure polycrystalline aluminium leads to a
room-temperature diffusivity of order of 1013. As discussed above, the main trap-
ping states of the presently considered aluminum alloy 2024 act similarly to disloca-
tions, in that they produce elastic stress fields at the interfaces of the strengthening
phases and dispersoids with the matrix lattice. In view of the above, the value of
1218 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224

diffusivity estimated by the experimentally determined hydrogen penetration depth

appears reasonable.

3.3. Microhardness measurements

The hydrogen spectra presented in the previous paragraphs provide some clues
about trapping states, based on the temperatures where hydrogen evolution takes
place. However, during the thermal cycle of hydrogen desorption the material under-
goes microstructural changes that affect its hardness. A comparison between the hard-
ness of corroded and uncorroded material that experience the same heating history is
shown in Fig. 5, and differences are set in perspective with the hydrogen spectra.
As expected, the hardness of the uncorroded material decreases with temperature
up to 400 C due to overaging (a slight increase at 250 C is attributed to a secondary
ageing peak), while it rises again above 400 C due to dissolution and reprecipitation.
The corroded material exhibits a similar behavior in general. However there are two
distinct differences in the hardness values. The corroded material appears softer than
the uncorroded up to 300 C and harder above 350 C. The first temperature region
incorporates states T1 and T2. Hydrogen in the T1 state is arguably trapped in inter-
stitial sites while T2 hydrogen is trapped at particle/matrix interfaces. Recent re-
search indicated that hydrogen might aid dislocation motion by altering the
binding energy between dislocations and solute atoms [25,26]. On the other hand,
hydrogen at matrix/particle interfaces might alter the interaction energy between dis-
locations and particles for coherency hardening in a way similar to that proposed for
lattice dislocation. The result is that hydrogen in T1 and T2 states probably aids
dislocation motion in overcoming either the lattice resistance or semicoherent parti-
cle resistance and the corroded material appears softer.

Uncorroded
Corroded
140 6
Hydrogen spectum
130 5

4
120
3
Microhardness, HV

110
µgH 2/min

2
100
1
90
0
80 -1

70 -2

60 -3
0 100 200 300 400 500 600
Temperature (oC)

Fig. 5. Microhardness profile versus temperature and hydrogen spectrum for alloy 2024.
 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224 1219

The second temperature region (above 300 C) incorporates states T3 and T4. It is
presently argued that T3 hydrogen is trapped as Mg hydride, while T4 hydrogen is
trapped in the main hardening phase (h or S). As is apparent from Fig. 5, hydrogen
affects the dissolution and reprecipitation process or even the overaging (coarsening)
process in the alloy and the corroded alloy appears harder. However more work is
needed to clarify this behavior.

3.4. Mechanical testing and fractography

In order to establish the effect of corrosion and hydrogen trapping on strength

and ductility, tensile testing of corroded samples [22] and corresponding fracto-
graphic analysis were undertaken. All corroded specimens were exposed for 24 h
and the following conditions were considered: Corroded specimens (C), corroded
specimens from which the corrosion layer was subsequently removed (CR), corroded
specimens heat treated at 495 C (CH), and corroded specimens with both corrosion
removal and heat treatment (CRH). The reference material consisted of uncorroded
specimens subjected to the respective material removal and heat treatment (condi-
tions U, UR, UH and URH respectively).
Elongation to fracture and energy density are the measurements used to assess
tensile ductility of the specimen. The energy density, U, is defined as:
Z ef
U¼ r  de ð3Þ
0

and represents the energy per unit volume absorbed for the deformation and fracture
of material. Energy density is presently evaluated as the area under the engineering
stress–strain curve as an approximation, since no significant necking is observed.
Tensile test results are given in Table 1. The tensile results for each test series rep-
resent the average value of four tensile tests. The deviations typically did not exceed
1% for ultimate tensile strength (UTS), 2.5% for yield strength, 6% for elongation
and 5% for energy density. The data are presented as percentages of the respective
reference values in Fig. 6. Corroded material (condition C) exhibits a yield strength
92% and an elongation 39% that of the uncorroded material. Heating of the speci-
men at 495 C (condition CH) does not affect yield strength, however it restores

Table 1
Results of tensile tests
Condition Yield strength (MPa) UTS (MPa) Elongation (%) Energy density (MJ/m3)
U 376 489 15 72
UH 327 496 17 81
UR 358 472 16 74
URH 296 447 14 59
C 349 425 6 25
CH 303 433 10 40
CR 363 468 11 49
CRH 287 438 14 57
1220 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224

Yield strength
UTS
100 Elongation
Energy Density

80
%Value of property

60

40

20

0
Ref C CH CR CRH
Condition

Fig. 6. The results of Table 1 presented as percentages of the respective reference values.

Fig. 7. Fracture surfaces of condition C (corroded 24 h) specimen. Individual photos correspond to

different modes of fracture from the specimen surface (on start of corrosion) through the centre of the
specimen.
 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224 1221

ductility to 56% of the uncorroded heat treated (condition UH). Removal of the cor-
rosion layer (condition CR) restores the yield strength to 100%, while ductility is
only partially restored to 66% that of the reference value. Removal of the corrosion
layer followed by heating at 495 C fully restores both the yield strength and elonga-
tion to the reference value (condition CRH).
The tensile test results are interpreted as follows: Corrosion affects a surface layer
of the material by introducing defects such as pitting, exfoliation and intergranular

Fig. 8. Fracture surfaces of condition CR (corroded 24 h + removal of 350 lm) specimen. Magnification
of the embrittled region and, transition to ductile fracture.
1222 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224

cracks. These defects degrade the strength of the material and lead to a deterioration
of yield strength, due to a decrease of the load carrying capacity of the corroded
layer as well as due to stress concentration caused by corrosion notches. Removal
of material up to the maximum depth of attack removes all the above defects and
thus restores strength. Ductility is also degraded by corrosion, but is only partially
restored by removal of the corroded layer. This constitutes evidence of a bulk mech-
anism of embrittlement. Subsequent heating to a temperature where all hydrogen
traps are activated and release hydrogen fully restores ductility, implicating hydro-
gen embrittlement as the responsible mechanism.
Additional evidence in favor of a hydrogen mechanism comes from the results of
fractographic analysis of tension specimens (condition C) presented in Fig. 7. The
most interesting finding is that a quasicleavage (embrittled) zone always appears be-
tween intergranular corrosion and the ductile uncorroded matrix. This observation is
consistent with the existence of a hydrogen affected zone, which degrades the ductil-
ity of the material even when the corrosion layer has been removed.
The proposed mechanism of hydrogen induced embrittlement of alloy 2024 is
sketched in Fig. 8. Corrosion creates damage, which penetrates in a non-uniform
way from the surface to the interior of the material, resulting (for 24 h exposure in
EXCO) in a depth of attack of 350 lm. Below the corrosion layer, a hydrogen
zone is established due to the diffusion of hydrogen released by the corrosion reac-
tions (Fig. 8a). Removal of a uniform layer equal to the depth of attack, removes
all the corrosion damage and all hydrogen trapped in the 350 lm surface layer in
the alloy. The result is an elongation increase to 66% of the reference value. How-
ever, part of the hydrogen zone still remains (Fig. 8b). Heating of the alloy to
495 C liberates all hydrogen (Fig. 8c) and restores ductility back to the original
value.

4. Conclusions

The experiments performed in this work lead to the following conclusions regard-
ing corrosion-induced hydrogen embrittlement in aircraft Al-alloys:

1. Corrosion damage starts with pitting and proceeds to pit-to-pit interactions, inter-
granular attack and exfoliation.
2. Hydrogen is produced during the corrosion process and is being trapped in dis-
tinct states in the interior of the material.
3. The temperature of hydrogen evolution and the variation of hardness with heat-
ing provide indirect evidence of the microstructural nature of hydrogen traps.
4. The hydrogen front advances with the corrosion front and establishes a hydrogen-
affected zone beneath the corrosion layer.
5. Removal of the corrosion layer leads to complete restoration of yield strength but
only partial restoration of ductility. Removal of the corrosion layer and heating to
activate all hydrogen traps leads not only to complete restoration of strength but
also to complete restoration of ductility.
 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224 1223

6. Detailed fractographic analysis shows the existence of a quasicleavage transition

zone between the intergranular corrosion zone and the ductile corrosion-un-
affected material. This quasicleavage zone has been embrittled by hydrogen diffu-
sion and trapping.

These results constitute evidence of hydrogen embrittlement in aluminum alloy 2024.

Acknowledgment

Part of this work has been financially supported by the Greek Secretariat of
Research and Technology (GSRT) and by AirBus.

References

[1] Aloha Airlines, Flight 243, NTSB Report Number AAR-89-03, Washington, DC, adopted on June
14, 1989.
[2] Z. Riga, Hellenic Aerospace Industry, private communication, 2000.
[3] S.G. Pantelakis, N.I. Vassilas, P.G. Daglaras, Effect of corrosive environment on the mechanical
behavior of the advanced Al–Li alloys 2091 and 8090 and the conventional aerospace alloy 2024,
METAL 47 (1993) 135–141.
[4] S.G. Pantelakis, P.G. Daglaras, C.A. Apostolopoulos, Tensile and energy density properties of 2024,
6013, 8090 and 2091 aircraft aluminum alloy after corrosion exposure, J. Theor. Appl. Mech. 33
(2000) 117–134.
[5] A.T. Kermanidis, Ph.D. Thesis, Department of Mechanical Engineering and Aeronautics, University
of Patras, Greece, 2003.
[6] G.M. Scamans, R. Alani, P.R. Swann, Corros. Sci. 16 (1976) 443.
[7] G.M. Scamans, C.D.S. Tuck, Embrittlement of aluminium alloys exposed to water vapour, in: Z.A.
Foroulis (Ed.), Proceedings of the Symposium on Environment—Sensitive Fracture of Engineering
Materials, The Metallurgical Society of the AIME, 1979, pp. 464–483.
[8] M.O. Speidel, Hydrogen embrittlement and stress corrosion cracking of aluminum alloys, in: R.
Gibala, R.F. Heheman (Eds.), Hydrogen Embrittlement and Stress Corrosion Cracking, ASM,
Materials Park, OH, 1992, pp. 271–296.
[9] G.A. Young, J.R. Scully, The effect of test temperature, temper and alloyed copper on the hydrogen-
controlled crack growth rate of an Al–Zn–Mg–(Cu) alloy, Metall. Mater. Trans. A 33 (2002) 101–
115.
[10] R.H Jones, The influence of hydrogen on the stress-corrosion cracking of low-strength Al–Mg alloys,
JOM—J. Miner. Met. Mater. Soc. 55 (2) (2003) 42–46.
[11] D.E Azofeifa, N. Clark, A. Amador, A. Saenz, Determination of hydrogen absorption in Pd coated
Al thin films, Thin Solid Films 300 (1997) 295–298.
[12] G. Itoh, K. Koyama, M. Kanno, Evidence for the transport of impurity hydrogen with gliding
dislocation in aluminium, Scr. Mater. 35 (6) (1996) 695–698.
[13] H. Saitoh, Y. Iijima, K. Hirano, Behaviour of hydrogen in pure aluminium Al–4 mass% Cu and Al–1
mass % Mg2Si alloys studied by tritium electron microautoradiography, J. Mater. Sci. 29 (1994) 5739–
5744.
[14] G.M. Pressouyre, A classification of hydrogen traps in steel, Metall. Trans. A 10 (1979) 1571–
1573.
[15] R.A. Outlaw, D.T. Peterson, F.A. Schmidt, Hydrogen partitioning in pure cast aluminium as
determined by dynamic evolution rate measurements, Metall. Trans. A 12 (1981) 1809–1816.
[16] G.A. Young, J.R. Scully, The diffusion and trapping of hydrogen in high purity aluminium, Acta
Mater. 18 (1998) 6337–6349.
1224 H. Kamoutsi et al. / Corrosion Science 48 (2006) 1209–1224

[17] P.N. Anyalebechi, Hydrogen diffusion in Al–Li alloys, Metall. Trans. B 21 (1990) 649–655.
[18] G.N. Haidemenopoulos, N. Hassiotis, G. Papapolymerou, V. Bontozoglou, Hydrogen absorption
into aluminium alloy 2024-T3 during exfoliation and alternate immersion testing, Corrosion 54 (1)
(1998) 73–78.
[19] E. Charitidou, G. Papapolymerou, G.N. Haidemenopoulos, N. Hasiotis, V. Bontozoglou, Charac-
terization of trapped hydrogen in exfoliation corroded aluminium alloy 2024, Scr. Mater. 41 (12)
(1999) 1327–1332.
[20] G34-90, A Standard test method for exfoliation corrosion susceptibility in 2xxx and 7xxx series
aluminium alloys (EXCO Test), in: Annual Book of ASTM Standards, 1994, p. 129.
[21] ASTM E8M-94a, in: Annual Book of ASTM Standards, West Conshohoken, PA, 1994, p. 129.
[22] S.G., Pantelakis, P.V. Petroyiannis, H. Kamoutsi, G.N. Haidemenopoulos, V. Bontozoglou, Evidence
on the corrosion-induced hydrogen embrittlement of the 2024 aluminum alloy, in: International
Conference on Influence of Traditional Mathematics and Mechanics on Modern Science and
Technology, Messini, Greece, 2004.
[23] T. Ishikawa, R.B. McLellan, The diffusivity of hydrogen in aluminium, Acta Metall. 34 (6) (1986)
1091–1095.
[24] H.-J. Schlüter, H. Züchner, R. Braun, H. Buhl, Diffusion of hydrogen in aluminium, Zeit. Phys.
Chem. 181 (1993) 103–109.
[25] P. Sofronis, Y. Liang, N. Aravas, Hydrogen induced shear localization of the plastic flow in metals
and alloys, Eur. J. Mech.—A/Solids 20 (6) (2001) 857–872.
[26] Y. Liang, P. Sofronis, R.H. Dodds Jr., Interaction of hydrogen with crack-tip plasticity: effects of
constraint on void growth, Mater. Sci. Eng. A 366 (2) (2004) 397–411.