Sie sind auf Seite 1von 8

Corrosion Science 52 (2010) 37743781

Contents lists available at ScienceDirect

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

Effect of heat treatment on anodic activation of aluminium by trace element indium


Brit Graver a,1, Antonius T.J. van Helvoort b, Kemal Nisancioglu a,*
a
Department of Materials Science and Engineering, Norwegian University of Science and Technology, N-7491 Trondheim, Norway
b
Department of Physics, Norwegian University of Science and Technology, N-7491 Trondheim, Norway

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

Article history: The presence of trace elements in Group IIIAVA is known to activate aluminium anodically in chloride
Received 30 April 2010 environment. The purpose of this paper is to investigate the surface segregation of trace element In by heat
Accepted 25 July 2010 treatment and resulting surface activation. Model binary AlIn alloys, containing 20 and 1000 ppm by
Available online 1 August 2010
weight of In, were characterized after heat treatment at various temperatures by use of glow discharge
optical emission spectroscopy, electron microscopy and electrochemical polarization. Heat treatment
Keywords: for 1 h at 300 C gave signicant segregation of discrete In particles (thermal segregation), which activated
A. Aluminium
the surface. Indium in solid solution with aluminium, obtained by 1 h heat treatment at 600 C, also acti-
B. Polarization
B. TEM
vated by surface segregation of In on alloy containing 1000 ppm In, resulting from the selective dissolution
C. Passivity of the aluminium component during anodic oxidation (anodic segregation). The effect of anodic segrega-
tion was reduced by decreasing indium concentration in solid solution; it had negligible effect at the
20 ppm level. The segregated particles were thought to form a liquid phase alloy with aluminium during
anodic polarization, which in turn, together with the chloride in the solution destabilized the oxide.
2010 Elsevier Ltd. All rights reserved.

1. Introduction after annealing, thereby removing the surface modied by heat


treatment prior to electrochemical characterization. It was still
Anodic activation of aluminium alloys by the presence of trace concluded that anodic activation could result from the segregation
elements in Group IIIAVA has been a recent subject of interest be- of pure In particles by annealing, suggesting In segregation in the
cause of its signicance in corrosion and surface engineering for bulk, possibly along the grain boundaries. However, segregated in-
corrosion protection of recycled aluminium alloys [1]. Anodic acti- dium in the form of secondary phase particles did not activate as
vation in chloride solution manifests itself in the form of a signi- much as indium in solid solution [11,12]. In one study [13], in
cant negative shift in the pitting potential relative to pure which the thermally modied surface of the AlIn samples was
aluminium and a signicant increase in the anodic current output not disturbed after annealing, In was reported to activate by heat
at potentials where aluminium is normally expected to be passive. treatment at temperatures signicantly lower (300 C) than that
Previous work in this laboratory showed that Pb in small con- required for Pb (600 C) [2].
centration (ppm level) in the bulk of the material can segregate The purpose of the present paper is to investigate if In, which
to the surface by heat treatment and activate the surface anodically may be present as a trace element in aluminium alloys, can be-
in chloride environment [1,2]. Several commercial alloys, such as come enriched and segregated in sufcient amounts at the alumin-
AA8006 [1,3] and 3005 [4], can be activated by heat treatment at ium surface as a result of heat treatment to be able to activate the
600 C because of the presence of Pb as a trace element. These al- surface anodically. Furthermore, it is intended to compare the
loys were reported to become activated by heat treatment also at mechanisms of activation by low and high temperature annealing
signicantly lower temperatures, possibly due to the added pres- and specic effects of In as trace element in relation to more signif-
ence of In, Sn or both. icant impurity compositions of In. This work is restricted to inves-
It is well known that aluminium can be activated anodically in tigation of binary model alloys of super pure aluminium alloyed
chloride solution by adding In ions to the solution [5,6] and by with 20 and 1000 ppm by weight In. The results are compared to
alloying the metal with In [7,8]. It was suggested that activation the corresponding results for binary AlPb alloys [2].
occurred by the formation of an AlIn amalgam which detached In preliminary work [14,15], the electrochemical behaviour of In
the oxide lm from the surface [9,10]. In the case of Al alloyed with compositions between 20 and 1000 ppm were found to be interme-
In [7,8,11,12], the material was often metallographically polished diate to the behaviour of these two extreme compositions in NaCl
solution. The two extreme compositions 20 and 1000 ppm were
* Corresponding author. Tel.: +47 73 59 40 35; fax: +47 73 59 40 83.
thus selected for a more comprehensive surface-analytical and elec-
E-mail address: kemal.nisancioglu@material.ntnu.no (K. Nisancioglu). trochemical investigation in the present work for a better under-
1
Present address: Det Norske Veritas, NO-1322 Hvik, Norway. standing of segregation and activation mechanisms. In a related

0010-938X/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2010.07.029
B. Graver et al. / Corrosion Science 52 (2010) 37743781 3775

study, we have investigated the activation of aluminium by trace cially for In in Al, were lacking. However, the settings of the instru-
element Sn [16]. ment were identical for all measurements, such that the proles in
different plots are comparable. Since the Al and O proles were
similar for both AlIn20 and AlIn1000 at the same heat treatment
2. Experimental
condition, the proles of these elements for only one of these alloys
are shown in each gure.
Binary AlIn specimens containing nominally 20 and 1000 ppm by
The data indicate enrichment of indium near the surface of
weight indium, referred to as AlIn20 and AlIn1000 in the remainder
300 C-annealed specimens of both alloys AlIn20 and AlIn1000,
of the paper, were prepared from high-purity aluminium (99.99%)
as shown in Fig. 1a. Since the In proles followed the oxygen pro-
and indium (99.99%) and cast into 2.5 cm thick moulds. After cool-
les with depth, and by inspection of the O relative to the Al pro-
ing, the cast specimens were scalped and cold rolled down to
les, In enrichment appeared to exist in the oxide and at the oxide
2.2 mm thickness. Indium compositions of the two alloys were sub-
metal interface. The In depth proles on samples heat treated at
sequently determined by inductively coupled plasma optical emis-
600 C were uniform relative to those treated at 300 C (Fig. 1b).
sion spectrometry (ICP-OES) as 15.7 and 1010 ppm, respectively.
However, an appreciable enrichment of In was still observed at
The materials were cut into 2  2 cm pieces, and metallograph-
the surface relative to the bulk, although this was about an order
ically polished through 1 lm diamond paste as the nal polishing
of magnitude smaller than the corresponding magnitudes obtained
step. The specimens were thereafter heat treated for 1 h at various
at 300 C. This indicated that a signicant amount of In in the bulk
temperatures in an air circulation furnace and then quenched in
of the specimen was segregated to the surface by annealing at
distilled water. Most samples were characterized electrochemi-
300 C, while heat treatment at 600 C gave more uniform proles
cally in this condition. Some samples were also slowly cooled in
caused by homogenization of In in the samples. The oxide thick-
air after heat treatment. Selected samples were etched in 10 wt.%
nesses, again judged by the trends in the Al and O proles, were
NaOH for 10 s at 60 C and then de-smutted for 1 min in concen-
estimated to be about 10 and 30 nm for the samples annealed at
trated nitric acid at 25 C before electrochemical characterization.
300 and 600 C, respectively. Other differences between alloy
Surface-analytical work was performed by use of a Hitachi S-
AlIn1000 annealed at 300 and 600 C were longer decay length of
4300SE eld emission gun scanning electron microscope (FEG-
the proles to the bulk levels and higher noise level for the higher
SEM), equipped with Inca (Oxford Instruments) energy dispersive
annealing temperature. These differences resulted from the pres-
spectrometer (EDS). The electron beam acceleration voltage was
ence of a thicker oxide and higher roughness of the surface of
15 kV. Selected samples of alloy AlIn20 were analysed also by a
the samples annealed at 600 C. Surface roughness resulted from
JEOL 2010F eld emission gun transmission electron microscope
microscale slip caused by internal stresses introduced by water
(FEG-TEM) at an acceleration voltage of 200 kV. The microscope
quenching. Roughness was introduced also by nonuniform sputter-
was equipped with a scanning unit (STEM), for which the probe
ing of different grains during depth proling [2].
size used was 1 nm, and EDS. Cross-sectional TEM specimens were
prepared by glueing two identical surfaces together with epoxy,
3.2. FEG-SEM
followed by mechanical thinning in the direction perpendicular
to the surfaces and Ar ion milling on a Gatan Precision Ion Polish-
It was difcult to detect indium on alloy AlIn20 annealed at
ing System (PIPS) at 4.5 kV. The TEM specimens were studied on a
300 C by use of EDS in SEM. Indium was detectable when located
cold stage held at about 160 C to reduce beam damage. Elemen-
primarily in pits, which were formed along and near the grain
tal depth proling was performed by use of a Horiba JobinYvon
boundaries during annealing. Such pits are shown in Fig. 2. No in-
glow discharge optical emission spectrometer (GD-OES) operated
dium was detected on alloy AlIn20 annealed at 600 C. However,
in the radio-frequency mode.
the pits were still present, but they were more randomly distributed
The specimens were polarised potentiodynamically and poten-
throughout the grains rather than on or near the grain boundaries.
tiostatically in 5 wt.% NaCl solution maintained at 25 1 C and ex-
Segregated In was more readily detectable on the AlIn1000
posed to ambient air. The cell geometry, including the positioning of
specimen in all tempers. Segregations in the form of particles were
the electrodes, and the solution volume were identical in all the
located both along the grain boundaries and in the grain bodies, as
runs. The solution was stirred mildly by use of a mechanical stirrer
shown for the sample annealed at 300 C in Fig. 3a. Segregations in
at a constant rate. The potentiodynamic sweep rate was 0.1 mV/s
the grain bodies were, however, closer to the grain boundaries than
in the positive potential direction, starting from 50 mV more nega-
the bulk of the grains. Most of the grain-boundary segregations ap-
tive to the open circuit potential. The potential used for the potentio-
peared to be in the form of continuous lm segments near the tri-
static measurements was 0.8 VSCE. Dissolved oxygen was not
ple points. Submicron size In particles were observable on the
purged from the solution because the rate of oxygen reduction
grain boundaries and the bulk of the grains also on the 600 C-an-
was too small to affect the high anodic current densities investigated
nealed sample, as shown in Fig. 3b, although the equilibrium solu-
on highly activated aluminium surfaces. The choice and justication
bility of In in Al at this temperature is higher than 1000 ppm [17].
of the experimental conditions are discussed in detail elsewhere [2].
The observed particles were probably those which did not go into
Corrosion morphology resulting from electrochemical polariza-
solid solution due to the short homogenization time of 1 h.
tion was investigated by SEM observation. Samples were prepared
Since the pits were not always associated with In particles on
by potentiodynamic polarization, as described above, up to prede-
samples annealed and then quenched in water, it was not possible
termined potentials, at which the specimens were removed from
to determine if the pits were related to segregated In in all cases.
the test solution, rinsed, dried and taken to the microscope.
Considering the possibility that segregated In was mechanically re-
moved from the surface during water quenching, a new set of spec-
3. Results imens were prepared by annealing for 1 h at 300 C, followed by
cooling in ambient air, instead of water quenching. The surface
3.1. GD-OES morphology was similar to the quenched specimens, exhibiting
pits along and near the grain boundaries. However, there were
Qualitative GD-OES depth proles obtained for the elements In, more pits on the air-cooled samples than water-cooled samples.
Al and O for the as heat-treated specimens are shown in Fig. 1. The A larger number of pits were observed on the low temperature an-
concentration scale is in arbitrary units as calibration data, espe- nealed samples than the high temperature annealed samples. In
3776 B. Graver et al. / Corrosion Science 52 (2010) 37743781

Fig. 1. GD-OES depth proles for AlIn model alloys heat treated for 1 h at (a) 300 C and (b) 600 C, followed by quenching in water.

of heat treatment. Fig. 5a shows examples of closed and open blis-


ters (pits). EDS point, line and area scans in and around such blis-
ters, as indicated by the analysis path in Fig. 5a, failed to indicate
the presence of segregated indium. EDS analysis along the path is
shown in Fig. 5b. Similarly, no segregated In was observed in the
blisters on AlIn1000. The TEM data indicated that the cover on
the blisters was aluminium oxide without any trace of a metallic
component. The blisters were the precursors for the pits. The
underlying pits became visible by destruction of the oxide cover.
This may have happened during annealing (when the blisters be-
came larger than a critical size), quenching and rinsing of the sur-
face after heat treatment.

3.4. Electrochemistry

The polarization curves for the AlIn alloys heat treated for 1 h
at different temperatures are shown in Fig. 6. As-polished AlIn20
alloy was only slightly activated in comparison with the behav-
Fig. 2. FEG-SEM secondary electron images of alloy AlIn20, heat treated for 1 h at iour of pure aluminium, as shown in Fig. 6a. Heat treatment re-
300 C and quenched in water. The small brighter spots in the pits (inset) are In sulted in a signicant activation of alloy AlIn20 with increasing
particles.
temperature, and at 300 C a maximum level of activation was at-
tained. By increasing the annealing temperature further to 600 C,
some cases, the pits were covered by the surface oxide, giving the
activation was gradually reduced down to the level of the as-pol-
appearance of blisters, as shown in Fig. 4. Formation of such blis-
ished condition. This is in contrast with the behaviour of the
ters by annealing aluminiumindium alloys has been reported ear-
AlPb20 alloy, which exhibited signicant activation only by
lier without giving an explanation for their origin [18].
annealing for 1 h at the highest temperature (600 C) selected
[2].
3.3. FEG-TEM AlIn1000 alloy was activated for all tempers investigated
(Fig. 6b). Note that the potential scale in Fig. 6b is expanded rela-
Cross-sectional specimens of alloy AlIn20, annealed at 300 C tive to Fig. 6a, and the differences between the tempers are not
and air cooled, were analysed by FEG-TEM with the purpose of as large as for the AlIn20 sample. Also for this case, annealing at
throwing more light on the nature of blisters formed as a result 600 C gave reduced activation relative to the lower annealing

Fig. 3. (a) FEG-SEM secondary electron image of alloy AlIn1000 heat treated for 1 h at 300 C and (b) a backscatter electron image of the same alloy heat treated for 1 h at
600 C. The samples were quenched in water after heat treatment. The white arrows show indium particles.
B. Graver et al. / Corrosion Science 52 (2010) 37743781 3777

In order to investigate the role of segregated In at the surface


and In in homogeneous solid solution with bulk aluminium, sur-
face thickness of the order of a few lm was removed either by
metallographic polishing or alkaline etching and de-smutting of
the heat-treated samples. After polishing, the anodic behaviour of
the 300 C-annealed AlIn20 sample became similar to that of the
as-rolled and polished variant (as-polished in Fig. 6a), as shown
in Fig. 7a, i.e., the surface exhibited passive behaviour relative to
the as-annealed sample, indicating that In, which was segregated
to the surface during annealing, was effectively removed by polish-
ing. However, the etched sample exhibited only a limited reduction
of activation relative to the as-annealed sample, indicating that
caustic etching was not as effective as polishing in removing the
segregated surface layer.
Fig. 4. FEG-SEM secondary electron image of a grain boundary on alloy AlIn20, heat For the sample annealed at 600 C, polishing seemed to reduce
treated for 1 h at 300 C, decorated with blisters and pits. This sample was air
cooled.
the passivity slightly in relation to the as heat-treated sample, as
shown in Fig. 7b. Caustic etching activated the surface in relation
to the as heat-treated sample as a result of In segregation at the
surface by selective etching of the more active aluminium compo-
nent. Subsequent de-smutting did not remove the segregated In.
Similar to alloy AlIn20, activation of the AlIn1000 sample, ob-
tained by heat treatment at 300 C, was reduced by subsequent
mechanical polishing or caustic etching, as shown in Fig. 8a. Fur-
thermore, the polished and etched conditions were less active than
the as-rolled and polished specimen (no heat treatment), shown in
Fig. 6b. This indicated that annealing of AlIn1000 at 300 C also re-
sulted in a surface layer, which was rich in indium, probably in the
form of segregated particles, as indicated by the surface-analytical
results reported above. This layer could be removed at least par-
tially by polishing or etching. However, removal of the surface
layer was less effective in passivating the AlIn1000 alloy as it
was for the AlIn20 alloy.
The anodic behaviour of the AlIn1000 samples, which were an-
nealed at 600 C, was not as signicantly altered by mechanical
polishing or caustic etching as those annealed at 300 C, as shown
in Fig. 8b. The activated behaviour of all three surface conditions
for this case indicated that the activating element In was more or
less homogeneously distributed in the sample and accessible at a
more or less similar concentration (or activating state) at the sur-
face in agreement with the GD-OES data (Fig. 1b).
Potentiostatic measurements were performed at 0.8 VSCE,
Fig. 5. (a) Bright eld FEG-TEM image of a blister (I) and a pit (II) on a cross- which is about 50 mV below the pitting potential of pure alumin-
sectional AlIn20 sample, heat treated for 1 h at 300 C and air cooled. (b) EDS line ium. Samples annealed at 300 C exhibited a high anodic output at
scan across blister (I) along the vertical dashed line shown in (a). The curve for C Ka1 the outset for both AlIn20 and AlIn1000 samples, as shown in
originates from the glue in (a), which was used in specimen preparation. Fig. 9a and b, respectively. The current density for the AlIn1000 al-
loy was higher than for the AlIn20 alloy in agreement with the
temperatures. Independent of heat treatment, however, all potentiodynamic data (Fig. 6). The current density decreased with
AlIn1000 samples tested exhibited higher activation than the most time of polarization for both cases, indicating the removal of super-
active AlIn20 specimen. cial active sites by corrosion. The active sites removed must be

Fig. 6. Effect of annealing temperature on potentiodynamic polarization of AlIn samples containing (a) 20 and (b) 1000 ppm In. The numbers in the graphs represent heat
treatment temperature in C. The annealing time used at each temperature was 1 h, followed by quenching in water.
3778 B. Graver et al. / Corrosion Science 52 (2010) 37743781

Fig. 7. Potentiodynamic polarization data for as-rolled and polished alloy AlIn20 in the heat treated (HT), heat treated and polished (HT + polished), and heat treated and
etched (HT + etched) condition. Heat treatment was performed for 1 h at (a) 300 C and (b) 600 C, followed by quenching in water.

Fig. 8. Potentiodynamic polarization data for as-rolled and polished alloy AlIn1000 in the heat treated (HT), heat treated and polished (HT + polished), and heat treated and
etched (HT + etched) condition. Heat treatment was performed for 1 h at (a) 300 C and (b) 600 C, followed by quenching in water.

segregated In at the surface, in accordance with the surface-analyt- indicated by GD-OES analysis (Fig. 1b). Polishing the AlIn1000
ical results, in particular GD-OES depth proles for the same type sample heat treated at 300 C gave a signicant reduction in the
of samples, as discussed above (Fig. 1). The steady-state current potentiostatic current density, suggesting that most of the In in
density attained (ca. 0.2 mA/cm2) indicated a relatively passive the sample was probably segregated to the surface, also in accor-
surface for the AlIn20 specimen whereas the AlIn1000 specimen dance with the GD-OES data (Fig. 1a).
indicated still an active surface at about 15 mA/cm2 after nearly In summary, potentiostatic results indicated time-dependent
3 h of polarization. All these observations were in accordance with changes occurring on the surface by removal of an active layer,
the potentiodynamic data (Fig. 6), which indicated a maximum causing passivation, or enrichment of In by selective corrosion of
activation for samples annealed at 300 C and reduction of the acti- aluminium, causing increased activation.
vation as a result of mechanical removal of the active layer by pol-
ishing (Figs. 7 and 8). 3.5. Corrosion morphology
For samples annealed at 600 C, the AlIn20 sample exhibited a
nearly passive behaviour in relation to the 300 C-annealed equiv- Corrosion morphology was investigated on specimens polarised
alent for the entire period of potentiostatic polarization, after potentiodynamically to predetermined end potentials in the range
exhibiting a small initial current density of 1 mA/cm2, which de- 1.25 and 0.8 VSCE. In this range, alloy AlIn20, heat treated at
cayed within the rst 100 s. As-rolled and polished specimens 600 C, was passive based on the results presented above
had a smaller passive current at the outset with appreciable in- (Fig. 6a). Alloy AlIn20, annealed at 300 C, showed supercial etch-
crease with time (Fig. 9a), which probably was caused by enrich- ing that appeared to initiate both at the grain boundaries and in the
ment of indium on the surface due to selective dissolution of the grains and propagate along preferred crystallographic planes, as
aluminium matrix. The AlIn1000 specimens annealed at 600 C, shown in Fig. 10.
however, exhibited active surface throughout the polarization per- Alloy AlIn1000, heat treated at 300 C, corroded at and near the
iod. A certain increase of the current density was observed with grain boundaries in the form of etching and pitting after sweeping
time before a steady-state was attained. The non-heat treated, the potential just above the pitting potential, as shown for the end
as-rolled and polished specimen exhibited a similar behaviour. potential of 1.2 VSCE in Fig. 11a and at a higher magnication in
The AlIn20 sample, which was polished after heat treatment, Fig. 11b. Pitting around the grain boundary precipitates of In could
exhibited passive behaviour similar to the as-polished variant be clearly observed. The particles along the grain boundaries were
(not heat treated), as shown in Fig. 9a. The AlIn1000 sample, which discrete and rounded after corrosion, as opposed to lm-formed
was polished after heat treating at 600 C, indicated a slight de- segregations obtained after heat treatment and before corrosion
crease in activation in relation to the unpolished equivalent (Fig. 3a), indicating that the lm pulled together into discrete par-
(Fig. 9b), probably as a result of removal of the In enriched layer ticles as a result of corrosion. This may have occurred also as a
B. Graver et al. / Corrosion Science 52 (2010) 37743781 3779

Fig. 9. Current density transients on alloys (a) AlIn20 and (b) AlIn1000 at an applied constant potential of 0.8 VSCE. All samples were initially polished, and some were
polished again after the heat treatment, as indicated in the graphs. Samples were quenched in water after each heat treatment step.

4. Discussion

Most effective enrichment and activation of aluminium due to


presence of In in the metal occurred as a result of heat treatment
at 300 C. This result is in agreement with earlier results by Fukuo-
ka [13]. The present results indicate further that the activation oc-
curs as a result of surface segregation of In by annealing at 300 C
(thermal segregation) for both 20 and 1000 ppm In concentrations.
The segregation appeared to occur to such an extent that the bulk
In concentration became signicantly reduced, as observed directly
by the GD-OES data and indirectly by the polarization data.
With increasing annealing temperature, activation was reduced
for low In alloy (AlIn20), attributable to dissolution of In in solid
solution with aluminium. Because of the low concentration of al-
Fig. 10. FEG-SEM backscattered electron image of alloy AlIn20, heat treated at loyed In, the concentration of In at the surface became too small
300 C, after potentiodynamic polarization 0.88 VSCE.
in the homogenised temper to give appreciable activation com-
pared to the situation where there was surface segregation of In.
For the 1000 ppm alloy, heat treatment temperature had a
result of softening or melting by Joule heating during oxidation of smaller signicance because of sufciently high concentration of
surrounding activated aluminium. Small pits were formed also in In relatively homogeneously distributed throughout the metal.
the grains close to the grain boundaries, as indicated in Fig. 11a Nevertheless, annealing at 300 C was more effective than anneal-
and shown at higher magnication in Fig. 11b. A ner debris of cor- ing at 600 C also in activating the surface of this high-In alloy. In
rosion products can also be observed near the grain boundaries, all cases, the lower temperature caused segregation of In to the
probably resulting from corrosion around the ner In particles dis- surface and thereby increased activation in relation to the homog-
tributed densely near the grain boundaries, as can be observed on enised condition obtained at the higher temperature. For the
the uncorroded surface (Fig. 3a). At higher end potentials, larger homogenised condition, however, the entire sample thickness be-
pits developed, as shown in Fig. 12a, which exhibited anodically came effectively activated. Activation was maintained in situ by
segregated In nanoparticles at the pit bottoms, as shown in selective anodic dissolution of aluminium during corrosion in chlo-
Fig. 12b. Corrosion on alloy AlSn1000, heat treated at 600 C, re- ride solution, thereby enriching the surface with indium (anodic
sulted in larger and more discrete pits in the grains with anodically segregation of indium). In the 300 C condition, the bulk of the
segregated In particles in the pits [19] (not shown). 1000 ppm alloy was more passive than the surface layer with seg-

Fig. 11. FEG-SEM backscattered electron images of alloy AlIn1000, polished and heat treated at 300 C and corroded by potentiodynamic polarization up to 1.2 VSCE. (a)
Corrosion morphology at and near grain boundaries and (b) a magnied grain boundary with indium particles and pits nearby in the grains.
3780 B. Graver et al. / Corrosion Science 52 (2010) 37743781

Fig. 12. (a) FEG-SEM secondary electron image of a pit on alloy AlIn1000, which was heat treated at 300 C and corroded by potentiodynamic polarization up to 1.05 VSCE.
(b) Backscattered electron image of the pit bottom at a higher magnication showing segregated In particles as bright spots.

regated In, as the bulk appeared to become depleted of In as a re- uum at temperatures as low as 100 C [18], which is lower than the
sult of thermal segregation of In to the surface. melting point for indium. A possible cause of pit formation is evap-
In contrast to the AlIn alloys, thermal segregation of Pb occurred oration of segregated In particles at the surface, especially in vac-
only after annealing at the higher temperature 600 C [2]. Thermal uum. However, the vapour pressure of In at the temperatures
segregation of Pb was not signicant at the lower annealing tem- investigated is not high enough to justify this explanation [22]. An-
peratures. The annealing temperature required for thermal segre- other possible explanation is the appearance of blisters as a result
gation of Pb was, therefore, much higher than that required for of diffusion of segregated In back into the aluminium substrate by
thermal segregation of indium. In contrast to Pb, indium was heat treatment. In particular, our results seem to agree with the
homogenised in the bulk of AlIn alloy by heat treatment at work of Kesten et al. [23] on the response of In implanted alumin-
600 C. Indium, in solid solution with aluminium in sufcient ium alloys to heat treatment in vacuum and air. The indium dif-
amount, could also be effective in activating the surface as dis- fused from the implantation sites to the surface at around 550 K
cussed above. While 20 ppm In in solid solution was not sufcient, and back into the bulk at temperatures above 650 K. No indium
1000 ppm In was more than adequate in causing activation. In this evaporation was detected. The location of the pits at and near
case, activation occurred by selective dissolution of the more active the grain boundaries can be a result of grain growth in combina-
aluminium, thereby enriching the surface with indium and causing tion with indium segregating along the grain boundaries during
segregation of the metal (anodic segregation), as is well known annealing. Blister formation by release of hydrogen trapped in
from earlier work on In in aluminium alloys [710]. The difference the material during preparation and processing of the samples is
in the effect of In and Pb in this respect can be explained by the fact yet another possibility. Pit formation on aluminium alloys contain-
that In is much more effective than Pb as an activator in general. ing low melting point trace elements as a result of annealing is a
While the presence of a given concentration of Pb in solid solution subject of continuing research.
with Al did not cause a discernable activation of the surface, the Corrosion morphology observed on alloy AlIn20, which was
same amount of In in solid solution could give appreciable activa- supercial etching for the 300 C temper, is in accordance with
tion. Only a signicant segregation of Pb at the surface by heat the enrichment of the surface by In as a result of heat treatment
treatment (thermal segregation) caused an appreciable activation at this temperature. The nature of enrichment could not be deter-
of the Pb containing material. Pb segregation by selective anodic mined, although the crystallographic nature of the attack may give
dissolution of aluminium does not activate the surface. clues for future work in this direction. The tendency for grain-
Solid solution solubility of In in Al is very small (nearly zero) at boundary and near grain-boundary attack on alloy AlIn1000 after
300 C [17], at which maximum activation of a surface layer oc- heat treatment at 300 C demonstrate the tendency for grain-
curred. Indium particles were observed to segregate easily along boundary segregation of In at this temperature. The fact that the
the grain boundaries and diffuse to the surface at this temperature. grain surfaces near the grain boundaries are also susceptible to
Aluminium can also be activated by the addition of In to the chlo- etching indicates that some of the In segregating to the grain
ride solution [5,6], contrary to the experimental ndings for Pb [2]. boundaries become incorporated again into growing grains. In-
These observations and failure to detect other forms of In segrega- dium in solid solution appears to cause activation mainly in the
tion near the surface by TEM indicated that segregated In particles, form of pitting, as is the case for alloy AlIn1000 heat treated at
rather than In segregating in other forms, such as the formation of 600 C. However also in this case, pit propagation occurs preferen-
a thin lm at the thermally grown oxidemetal interface in the tially along the triple point junctions, indicating the prevalence of
case of Pb [20], was responsible for activation. These results sup- grain boundary sensitization.
port the theory [10] that low melting point In forms a liquid alloy The change in the morphology of the In particles on alloy
with aluminium, assisted by the heat generated by the oxidation AlIn1000, which are segregated thermally by heat treatment at
process and melting point depression caused by the size of the par- 300 C, from sharp edged lm-like morphology at the grain bound-
ticles [21]. In combination with the presence of chloride in the aries into rounded particles during corrosion, was attributed to
solution, the passivity of the oxide is signicantly reduced, culmi- possible melting and pulling together of the lm by surface ten-
nating in an effective depression of the pitting potential and in- sion. The In particles observed in the bottom of the pits, which
creased corrosion rates. must have formed by anodic segregation, were probably also liquid
The pits and blisters, which formed as a result of heat treat- droplets during corrosion. The observation indicates that such
ment, appeared to be associated with segregated indium and grain droplets probably form and become removed by undermining con-
growth. Blisters were reported to form after annealing in high vac- tinuously during anodic polarization on the corroding areas and
B. Graver et al. / Corrosion Science 52 (2010) 37743781 3781

contribute to the activation process in accordance with the amal- Acknowledgements


gamation theory.
Hydro Aluminium assisted with GD-OES measurements and
5. Conclusions preparation of model alloys. This work was supported by The Nor-
wegian Research Council under Contract No. 158585/431.
Activation of aluminium in chloride solution is a result of segre-
gation of indium to the alloy surface. This can occur thermally by References
heat treatment (thermal segregation) or in situ by selective anodic
[1] Y.W. Keuong, J.H. Nordlien, S. Ono, K. Nisancioglu, J. Electrochem. Soc. 150
oxidation of the aluminium component of the alloy during corro-
(2003) B547.
sion (anodic segregation). Annealing at 300 C, which is a temper- [2] . Svik, Y. Yu, J.H. Nordlien, K. Nisancioglu, J. Electrochem. Soc. 152 (2005)
ature higher than the melting point of indium, caused signicant B334.
[3] Y. Yu, . Svik, J.H. Nordlien, K. Nisancioglu, J. Electrochem. Soc. 152 (2005)
thermal segregation of indium to the surface of AlIn alloys contain-
B327.
ing 20 and 1000 ppm by weight In. This caused signicant anodic [4] A. Afseth, J.H. Nordlien, G.M. Scamans, K. Nisancioglu, Corros. Sci. 44 (2002)
activation especially of the 20 ppm AlIn aluminium alloy in chlo- 145.
ride solution. The bulk of both alloys became depleted of In by sur- [5] W.M. Carroll, C.B. Breslin, Corros. Sci. 33 (1992) 1161.
[6] H.A. El Shayeb, F.M. Abd El Wahab, S. Zein El Abedin, J. Appl. Electrochem. 29
face segregation, and mechanical removal of the active layer (1999) 473.
caused increased passivity. [7] B. Breslin, W.M. Carroll, Corros. Sci. 34 (1993) 1099.
By annealing at 600 C, all In in the 20 ppm alloy and most In in [8] A. Venugopal, V.S. Raja, Corros. Sci. 39 (1997) 1285.
[9] S.B. Saidman, S.G. Garcia, J.B. Bessone, J. Appl. Electrochem. 25 (1995) 252.
the 1000 ppm alloy were dissolved in solid solution with the alu- [10] J.B. Bessone, D.O. Flamini, S.B. Saidman, Corros. Sci. 47 (2005) 95.
minium matrix. Activation of the 1000 ppm In alloy was nearly [11] J.C. Lin, H.C. Shih, J. Electrochem. Soc. 134 (1987) 817.
unaffected by the high temperature treatment showing that activa- [12] R. Despic, D.M. Drazic, M.M. Purenovic, N. Cikovic, J. Appl. Electrochem. 6
(1976) 527.
tion was nearly equally possible, independent of whether indium [13] K. Fukuoka, Sumitomo Light Metal Technical Reports, vol. 42, 2001, pp. 131.
was in solid solution or segregated at the surface. Activation of [14] B. Graver, . Svik, Y. Yu, K. Nisancioglu, in: P. Marcus, V. Maurice (Eds.),
the 20 ppm In alloy was, however, nearly eliminated by high tem- Passivation of Metals and Semiconductors, Elsevier, Amsterdam, 2006, p. 627.
[15] B. Graver, A. van Helvoort, J.C. Walmsley, K. Nisancioglu, Mater. Sci. Forum
perature heat treatment because the surface concentration of In
519521 (2006) 673.
was signicantly reduced by homogenization. Even enrichment [16] B. Graver, A.M. Pedersen, K. Nisancioglu, ECS Trans. 16 (52) (2009) 55.
of In by selective anodic dissolution of aluminium was not suf- [17] L.F. Mondolfo, Aluminium Alloys: Structure and Properties, Butterworths,
London, 1976. pp. 304.
cient for surface activation in this case.
[18] S.K. Kailasam, S.P. Murarka, M.E. Glicksman, J. Electrochem. Soc. 147 (2000)
The blisters, which are observed on or near grain boundaries 4318.
after heat treatment could not be related to any In segregation [19] K.F. Graver, Ph.D. Thesis, Norwegian University of Science Technology,
by the characterization techniques used here. A possible cause of Trondheim, Norway, 2009.
[20] J.C. Walmsley, . Svik, B. Graver, R. Mathiesen, Y. Yu, K. Nisancioglu, J.
their formation is tentatively suggested to be related to release Electrochem. Soc. 154 (2007) 28.
of trapped hydrogen in annealing. [21] H.W. Sheng, K. Lu, E. Ma, Acta Mater. 46 (1998) 5195.
The present results support the theory that anodic activation of [22] R.C. Weast (Ed.), CRC Handbook of Chemistry and Physics, 66th ed., CRC Press,
Boca Raton, FL, USA, 1985, p. D-215.
aluminium is probably a result of segregated indium forming an [23] J. Kesten, P. Wodniecki, M. Uhrmacher, K.P. Lieb, Hyp. Int. 39 (1988) 129.
amalgam with the aluminium during the oxidation process.