You are on page 1of 10

Corrosion product analysis on crevice corroded Alloy-22 specimens

P. Jakupi
a
, F. Wang
b
, J.J. Nol
a
, D.W. Shoesmith
a,
a
Department of Chemistry, The University of Western Ontario, London, Ontario, Canada N6A-3K7
b
Department of Physics, University of Science and Technology Beijing, Beijing 100083, PR China
a r t i c l e i n f o
Article history:
Received 12 May 2010
Accepted 12 January 2011
Available online 25 January 2011
Keywords:
A. Alloy-22
A. Molybdenum
A. Polymeric molybdates
C. Crevice corrosion
A. Nickel-based alloys
B. Raman spectroscopy
a b s t r a c t
Surface analytical techniques were applied to characterize corrosion products formed during the crevice
corrosion of the NiCrMo(W) Alloy-22 in 5 mol/L NaCl at 120 C. Micro-Raman spectroscopy demon-
strated the formation of polymeric molybdates within the crevice corroded region where intergranular
corrosion dominated. The location and chemical speciation of the Mo and W species formed was inves-
tigated by Raman mapping. Crevice corrosion was found to propagate preferentially across the alloy sur-
face rather than to penetrate deeply at localized sites, a feature which appears to be linked to the
formation and build-up of polymeric molybdates.
2011 Elsevier Ltd. All rights reserved.
1. Introduction
NiCrMo alloys are distinguished for providing corrosion resis-
tance in extremely corrosive environments, and are consequently
used in a wide range of corrosive industrial applications. The
anti-corrosion and electrochemical properties of these alloys arise
from the passive, thin (25 nm) protective oxide layer that covers
the alloy, with the alloying components giving the oxide its unique
properties.
Ni-based alloys, in general, are more corrosion resistant than
stainless steels due to their tolerance for extensive alloying. Cr
and Mo are known to impart resistance against oxidizing and
reducing acids, respectively [1,2]. In terms of passive corrosion
and breakdown behavior, it has been suggested that Cr plays the
key role in maintaining the passivity of the oxide [2]. Conse-
quently, the breakdown potential is primarily inuenced by the
Cr content, while repassivation occurring after a breakdown event
is inuenced by the Mo content [3]. From a localized corrosion
viewpoint, enhancing the passive nature of the oxide lm should
render the alloy immune to initiation. However, once initiation
and subsequent propagation have occurred, a knowledge of how
the surface composition evolves and affects repassivation is key
to determining localized corrosion mechanisms.
Although alloying with Mo is well known to increase the corro-
sion resistance of iron and Ni-based alloys, the mechanisms are
still not fully understood. It has been suggested that Mo locates
at defect sites that would otherwise preferentially dissolve [4,5],
leading to a modication of the rate of anodic dissolution due to
the greater relative MoMo bond strength found in alloys [6,7].
Lloyd et al. [2] observed, with increasing applied potential, segre-
gation of the alloying elements, NiCr to the inner alloy-oxide
interface and MoW to the outer oxideelectrolyte interface for a
series of commercial NiCrMo alloys. The greater the Mo content
of the alloy, the greater the extent of segregation resulting in lower
passive current densities. It has also been suggested that Mo is sta-
bilized at the oxide/electrolyte interface as MoO
2
4
in stainless
steels [8,9] and gives the outer regions of the passive lm a cation
selective character that discriminates against the incorporation of
the aggressive Cl

anion into the passive lm. Bastidas et al. [10]


observed the formation of Mo-chloro complexes on AISI 316 stain-
less steels that may potentially decrease the free chloride concen-
tration close to the alloy surface and further inhibit Cl

ingress in
the protective passive lm. Inhibition of localized corrosion on
stainless steels [1114] and Ni-alloys [15,16] due to Mo has been
studied extensively. In both alloys, the inhibiting effect of alloyed
Mo has been shown to be primarily due to the species MoO
2
4
. Even
when added as a salt to the electrolyte this species has been shown
to increase corrosion resistance [1517].
This study focuses on the crevice corrosion of the NiCrMoW
Alloy-22 and especially the inuence of Mo on propagation and
repassivation. Crevice corrosionproducts have beenstudiedprimar-
ily by Raman spectroscopy, although other surface analytical tech-
niques were also used. To determine whether W behaved similarly
to Mo, a few experiments were performed on the W-free Alloy-
2000. The nominal compositions of these alloys are given in Table 1.
0010-938X/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2011.01.028

Corresponding author. Tel.: +1 519 661 2111x86366, fax: +1 519 661 3022.
E-mail address: dwshoesm@uwo.ca (D.W. Shoesmith).
Corrosion Science 53 (2011) 16701679
Contents lists available at ScienceDirect
Corrosion Science
j our nal homepage: www. el sevi er . com/ l ocat e/ cor sci
2. Experimental procedure
Crevice corrosion experiments were conducted using an elec-
trochemical cell built within a Teon (PTFE)-lined Hastelloy

pres-
sure vessel [18]. A homemade Ag/AgCl (saturated KCl) reference
electrode [19] was used to measure the potential of the working
electrode. All potentials reported in this study are quoted against
this electrode unless stated otherwise. The working electrode
was cut from an Alloy-22 plate which was 0.317 cm (1/8
00
) thick.
The alloy composition is given in Table 1. The Alloy-22 specimen
was machined and bent into a V-shape to ensure that only one cre-
vice was formed in contact with the electrolyte solution, . The cre-
vice assembly was held together with threaded Alloy-22 bolts and
nuts. The crevice former was a small Teon wafer sandwiched be-
tween the at metal surface of the working electrode and a poly-
sulfone coupon. This assembly dened a creviced area of 4 cm
2
.
The working electrode had a threaded tapped hole in one end to
accept a nickel alloy welding rod used to make electrical contact
to external circuitry. Udel (Polysulfone) bushings were placed be-
tween the working electrode and the metal bolts and nuts to insu-
late the bolts and any additional creviced areas at these locations
from the working electrode. The crevice-forming face of the work-
ing electrode was grinded with a series of wet silicon carbide pa-
pers (320, 600, 800, 1000, and 1200 grits). All parts of the crevice
assembly were degreased by sonication in methanol and de-ion-
ized water, and then rinsed with de-ionized water, and air dried.
Prior to assembly, the V-shaped crevice electrode and Teon cre-
vice former were submerged in the electrolyte solution to be used,
to ensure wetting of the crevice interior. The crevice tightness was
adjusted using a Teon feeler strip cut from the same sheet as
the Teon crevice former.
A 5 mol/L NaCl solution prepared using de-ionized (DI) Milli-
pore water (18.2 MXcm) and reagent grade sodium chloride from
Caledon Chemicals (99.0% assay) was used in all experiments. The
solution was naturally aerated by agitation in air. To prevent boil-
ing of the electrolyte, the cell was pressurized to 414 kPa with ultra
high purity argon gas. The temperature was controlled at 120 C in
all experiments. During heat-up, the corrosion potential of the
creviced specimen was measured using a Solatron model 1284
potentiostat. Once the desired temperature was reached (6 h),
crevice corrosion was initiated and propagated under constant cur-
rent (galvanostatic) control. As discussed previously [20], this was
necessary since all attempts to initiate and propagate crevice cor-
rosion using an Alloy-22 creviced electrode galvanically coupled to
a counter electrode of the same alloy under the same conditions
were unsuccessful.
On completion of each experiment, the creviced specimen was
rinsed with de-ionized water and methanol and then dried and
stored in a desiccator prior to surface analysis. The specimens were
photographed and examined with an optical microscope (Olym-
pus, IX70). A Hitachi S-4500 eld emission Scanning Electron
Microscope (SEM) equipped with an EDAX energy dispersive X-
ray (EDX) system was used to examine the surface topography of
the creviced specimens.
X-ray photoelectron spectroscopic (XPS) analyses were per-
formed using a Kratos Axis Ultra XPS. Spectra were analyzed using
the commercial CasaXPS software, v. 2.2.107 [21]. XPS spectra
were corrected for charging by taking the C 1s spectrum for adven-
titious carbon to be at a binding energy of 284.8 eV. High resolu-
tion spectra were tted with a mixed Gaussian/Laurentian
function using the parameters published by Biesinger et al. [22].
Confocal Raman spectroscopy was performed under ambient
conditions with an Alpha SNOM, (WITec). A linearly polarized
YAG laser (Verdi 5, Coherent Inc.) with a wavelength of 532 nm
was used for Raman excitation. A 50 objective lens was used to
focus the laser beam onto chosen regions of the specimen. Raman
spectra were recorded at each image pixel by an air-cooled back-
illuminated CCD camera. A typical image consisted of 50 50 pix-
els that made up a total of 2500 spectra. Raman images of the cor-
roded specimens were constructed by integrating the intensities of
the characteristic Raman bands within the scanned area in the XY
orientation. A laser power of 10 mW was used to enhance Raman
scattering. Creviced specimens were checked after analysis to en-
sure no damage had occurred due to surface heating by the laser.
3. Results
3.1. Surface analytical results within propagated region
Fig. 2 shows the creviced region of an Alloy-22 specimen on
which crevice corrosion was galvanostatically initiated and
propagated at 200 lA. Corrosion damage is conspicuous near the
edge (i.e. mouth) of the creviced region and appears to have prop-
agated laterally across the surface much more rapidly than it pen-
etrated into the alloy. An optical magnication (Fig. 3) reveals that
the crevice corroded region is decorated with pits and preferential
Table 1
Elemental composition of Alloy-22 and Alloy-2000 (wt%).
Element Ni Cr Mo W Fe Co Mn Si C S Cu
Alloy-22 Bal. 22 13 3 3 2.5 0.5 0.08 0.01 0.02
Alloy-
2000
Bal. 23 16 3 2 0.5 0.08 0.01 0.01 1.6
working
electrode
electrolyte
level
Alloy-22 nuts
and bolts
Teflon crevice
former
Udel bushing
Udel block
Fig. 1. Schematic of the creviced working electrode (not to scale).
teflon crevice former
propagated region
creviced region
Fig. 2. Alloy-22 specimen on which crevice corrosion was initiated and propagated
galvanostatically at an applied current of 200 lA.
P. Jakupi et al. / Corrosion Science 53 (2011) 16701679 1671
corrosion along the grain boundaries. Intergranular corrosion (IGC)
was shown to dominate the damage morphology for crevice corro-
sion via galvanostatic control on Alloy-22. Fig. 4 shows the mea-
sured potential of the creviced specimen at the applied constant
current of 200 lA: For the rst 50 s the measured potential in-
creases rapidly before decreasing to a steady state value for the
duration of the experiment (>300 s). As discussed in more detail
previously [20,23], the steady-state potential indicates the occur-
rence of crevice propagation and reects both the development
of aggressive chemistry and a resulting IR drop within the active
region of the crevice. The steady-state potential observed, as well
as the lack of potential excursions and transients associated with
it, indicates that once initiation and stabilization of an active loca-
tion has occurred, no new corrosion sites initiate. This suggests
that propagation involves the growth of a single (or very small
number) of initiated sites. Since the area damaged by propagation
increases with time under constant current conditions, the current
density must decrease, at least at some locations within the cre-
vice. This implies that the spread of corrosion damage is more
readily achieved than the accumulation of damage at deep local
sites. If the critical conditions for propagation are to be maintained,
then the current density and hence critical chemistry, must be
maintained at the spreading edges of the damaged area, strongly
suggesting that the current density decreases at the already dam-
aged locations left behind. It seems likely that these trailing loca-
tions undergo repassivation.
Fig. 3 shows that corrosion products are present within the
creviced area and tend to accumulate on the grain surfaces. During
dismantling of the creviced specimen once the experiment was
completed, the corrosion products appeared to have stayed intact
on the alloy surface even after rinsing and drying. Similar behavior
is observed over the applied current range, 20200 lA.
To characterize the corrosion products within the creviced re-
gions, a variety of surface analytical techniques were applied. Figs.
5 and 6 show SEM images and corresponding EDX maps of IGC
areas and grain surfaces, respectively, for a specimen corroded at
an applied current of 200 lA. The elemental analyses show similar
trends in both areas. EDX signal intensities for Mo, W, and O were
all greater on the corroded grain boundaries and for the corrosion
products on the grain surfaces compared to the bare grain surfaces.
By contrast, the signal intensities for Ni and Cr were depleted in
these regions relative to the bare grain surface. To further examine
the chemical nature of the corrosion products within the corroded
regions Raman spectroscopy was employed.
Fig. 7 shows optical images and the corresponding Raman maps
recorded within the corroded region, specically at locations
where IGC had occurred and corrosion products had accumulated.
Contrast in the Raman signal intensities recorded at grain bound-
aries and on grain surfaces was clearly observed. Also, a distinctive
signal contrast was observed between bare grain surfaces and
those covered by corrosion products. The signal intensities in
Fig. 7 are the integration of the total Raman signals observed with-
in the spectrum.
Low-resolution Raman spectra (Fig. 8) of the mapped regions in
Fig. 7 show similar signals for areas subjected to IGC and those
where corrosion products accumulated on the grains, indicating
the accumulation of corrosion products has also occurred within
the corroded grain boundaries. Peaks were observed and centered
at 355, 835, and 965 cm
1
. The peaks observed at 350 and
835 cm
1
were consistently broader than the sharper peak at
965 cm
1
. Depth proles and 3D Confocal Laser Scanning Micros-
copy (CLSM) images [24] also show the accumulation of corrosion
products within the corroded grain boundaries, which have been
previously shown to be locations of preferential attack on creviced
specimens [25].
In a multi-component alloy such as Alloy-22 (Table 1) care
should be taken in assigning Raman peaks to specic chemical spe-
cies. According to the Critical Crevice Solution (CCS) model, crevice
propagation is characterized by accelerated oxidation and dissolu-
tion, and the subsequent hydrolysis of metal cations leading to a
low pH and an increase in chloride content within the creviced re-
gion [26,27]. Precise pH measurements within the creviced region
are not practical due to the constricted geometry however, acidic
values (<3) are undoubtedly expected during propagation
[28,29]. From the elemental surface analyses (Figs. 5 and 6) the rel-
ative EDX signal intensities for Mo, W, and O were enhanced at cor-
roded sites and in corrosion products within the corroded region,
suggesting that the Raman signals observed from the mapped
images are due to a mixture of these species formed within an
acidic environment.
3.2. Assigning Raman peak signals to molybdates and tungstates
Molybdates and tungstates are known to aggregate to form
polynuclear species in low pH solutions [30,31]. At near-neutral
pH, in oxic solutions the stable molybdate species is MoO
2
4
. Within
the range 4.5 < pH < 6.5 the paramolybdate species Mo
7
O
6
24
is
known to co-exist with MoO
2
4
via the proton-consuming
equilibrium:
7MoO
2
4
8H

Mo
7
O
6
24
4H
2
O 1
Larger aggregates such as the octamolybdate anion Mo
8
O
4
26
are
known to be formed upon further acidication into the pH range
2.9 to 1.5,
intergranular
corrosion
corrosion
product
Fig. 3. Optical micrograph of the Alloy-22 specimen in Fig. 1. This magnied image
was taken within the crevice corroded region to show the preferred attack on the
grain boundaries and the resulting.
0.6
0.4
0.3
E

(
V
)
0.2
V
A
g
/
C
l
)
0.0
E
E

(
V
400 500 600 700
-0.3
0.0
0 100 200 300 400 500 600 700
Time (s)
0 2
0 10000 20000 30000 40000
-0.2
0 10000 20000 30000 40000
Time (s)
Fig. 4. Potential measured on a creviced Alloy-22 specimen corroded at an applied
constant current of 200 lA.
1672 P. Jakupi et al. / Corrosion Science 53 (2011) 16701679
8MoO
2
4
12H

Mo
8
O
4
26
6H
2
O 2
and, for pH < 1.5 (i.e. the conditions anticipated within a propagat-
ing creviced region), further aggregation may lead to the polymo-
lybdate [3133] anion, Mo
36
O
8
112
. At pH 0.9, molybdate is known
to reach its isoelectric point and form molybdic acid (MoO
3
H
2
O),
and on further acidication, positively charged species such as
MoO

2
, MoO
4+
and HMo
2
O
4
6
have all been reported [34,35].
To study the inhibitive effect of nitrates on Alloy-22 corrosion
under acidic conditions, Gray et al. [36] utilized in situ Raman spec-
troscopy to characterize transpassive dissolution products. Signi-
cant Raman peaks were observed at 321 and 349 cm
1
, which were
attributed to the MoO bond stretching mode and the Mo(2)O
bond bending mode, respectively. The Raman spectra reported clo-
sely matched the spectra for Mo(V) chloride solution standards,
but the authors reported the detected species as MoO
4
. No evi-
dence for Mo-polymerization was observed.
In this study a broad peak is observed spanning the wavenumber
region 175570 cm
1
, but centered at 355 cm
1
. According to lit-
erature values, this broad feature can be assigned to the r MoO
bending mode characteristic of MoO
2
and MoO
3
oxides [3639].
Two distinct peaks are also observed at 835 and 965 cm
1
, with
the latter usually being the most intense of the three observed peaks
in the spectra. Although sharper than the broad peak at 355 cm
1
,
these peaks are still muchbroader thanthe sharppeaks observedfor
pure crystalline materials [37,40]. Published studies [39,41,42] sug-
gest that the peak observed at 835 cm
1
is due to the asymmetric
stretch of MoOMo bonds, characteristic of MoO
2
4
. However, this
peak was observed to be relatively weak compared to the peak ob-
served at 897 cm
1
, which is primarily used to characterize
MoO
2
4
. Dieterle and Mestl [37] observed signicant peaks at 835
and 843 cm
1
and attributedthemto the bridging MoO
2
vibrations
of the intermediate molybdenumoxide, Mo
4
O
11
. Cross and Schrader
[43] also observed a sharp peak at 834 cm
1
, which they attributed
to the oxide intermediate Mo
4
O
11
formed upon heating MoO
3
thin
lms. In a Raman study of aqueous molybdates [44] as a function
of pH, no peaks in the wavenumber region 830850 cm
1
were ob-
servedfor a near-neutral pHof 6.6, whileacidicationof theaqueous
solutionto pH2.2 and pH0.95 led to the appearance of Ramanpeaks
at 844 and 835 cm
1
, respectively. It was suggested these peaks
could be due to the formation of the highly aggregated molybdate
species, Mo
36
O
8
112
. Acidication of aqueous molybdates to pH 2.2
also led to the appearance of a broad peak at 363 cm
1
attributed
to the octamolybdate anion Mo
8
O
4
26
.
The sharpest and most intense peak observed in the Ramanspec-
tra (Fig. 8) at 965 cm
1
has been assigned to the symmetrical
Fig. 5. SEM image and corresponding EDX maps of a crevice corroded region on Alloy-22 specimen.
Fig. 6. SEM image and corresponding EDX maps of the corrosion products accumulated on the grain surfaces of a crevice corroded Alloy-22 specimen.
P. Jakupi et al. / Corrosion Science 53 (2011) 16701679 1673
stretching of the MoO bond [39]. Commonly for molybdates,
the Raman bands observed in the ranges 9501006 and
918945 cm
1
are characteristic of octahedrally and tetrahedrally
coordinated compounds respectively, and are key in accurately
identifying molybdate species, since the Raman peaks are usually
the most intense within these frequency ranges [39,4547]. Shifting
of these peaks is linked to the molybdate chain length; for example,
symmetrical t (MoO) stretching increases from 897 to 943 to 965
cm
1
, characteristic of the molybdate species MoO
2
4
, Mo
7
O
6
24
, and
Mo
8
O
4
26
, respectively [45]. Aqueous molybdate studies [40,41,44]
haveshownthat the Ramanbandlocatedat 897 cm
1
, whichis char-
acteristic of the species MoO
2
4
, decreases inintensityconcomitantly
with the appearance of Raman bands centered within the range
9501000 cm
1
when the pH is decreased. In this study (Figs. 8
and 9), the sharp asymmetric peak has been consistently observed
withinthe region 915970 cm
1
, whichis assigned to the polymeric
Mo
8
O
4
26
(965 cm
1
) species [39,40]. The shoulder on the low fre-
quency side of the peak suggests that stretching modes for other
polymeric molybdate species [40], such as Mo
7
O
6
24
(940 cm
1
) and
Mo
3
O
2
10
(950 cm
1
) may also contribute to this peak. Broadening
of the peak centered at 355 cm
1
, may also be attributed to a combi-
nation of Raman active species, specically the oxides MoO
2
and
MoO
3
and the Mo
8
O
4
26
species formed at low pH.
The broadness of the peaks (Figs. 8 and 9) suggests a lack of
crystallinity and order of the Raman-active species. Also, Raman
bands arising from tungstate species [42,44] with similar
stretching frequencies to those of molybdates may also contribute
to the broadness of the peaks in Figs. 8 and 9.
For comparison the commercial alloy, Alloy-2000, which con-
tains an increased Mo content, but no W, was also analyzed to
Fig. 7. Optical images (a and c) and the corresponding Raman maps (b and d) of crevice corroded regions on an Alloy-22 specimen. Spectra were recorded in regions (outlined
in dash-line) showing IGC (a and b) and accumulated corrosion products on grain surfaces (c and d).
(a) (b)
400 400
300 300
200 400 600 800 1000 1200
200
200 400 600 800 1000 1200
200
200 400 600 800 1000 1200
(cm
-1
)
(cm
-1
)
200 400 600 800 1000 1200
I
n
t
e
n
s
i
t
y

(
A
r
b
i
t
r
a
r
y

U
n
i
t
s
)
I
n
t
e
n
s
i
t
y

(
A
r
b
i
t
r
a
r
y

U
n
i
t
s
)
Fig. 8. Low-resolution Raman spectra of areas exhibiting (a) IGC and (b) corrosion product accumulation recorded on crevice corroded Alloy-22 at an applied current of
200 lA.
1674 P. Jakupi et al. / Corrosion Science 53 (2011) 16701679
determine whether W played a similar role to Mo. Fig. 9 shows
high-resolution spectra recorded on corroded grain boundaries
for both Alloy-22 and Alloy-2000. Only minor differences, such as
minor peak shifts and variations in broadness, especially for the
peak centered at 838 cm
1
, were observed. A full spectral decon-
volution would be required to accurately locate and determine
whether there was any signicant effect of W on the spectra. As
for molybdates, the tungstate, WO
2
4
, is known to be the predomi-
nant species at near-neutral pH and further acidication leads to
polymerization (Reaction (3)) and then dimerization [31] (Eq. (4)).
6WO
2
4
7H

HW
6
O
5
21
3H
2
O 3
2HW
6
O
5
21
4H
2
OW
12
O
36
OH
10
10
4
However, the kinetics of reactions (3) and (4) are known to be
slower than the formation of the polymerized molybdates
[31,46], via reactions (1) and (2). Raman peaks similar to those in
Fig. 9 have been observed for aqueous polymeric tungstate solu-
tions (as a function of pH [46]) at 838 and 960 cm
1
and attributed
to WO
2
4
and the polymeric species in Eqs. (3) and (4). However, no
signicant differences in the spectra recorded on Alloys-22 and
2000 are observed, Fig. 9, and it is likely that the polymeric molyb-
dates dominate the corrosion product content within the corrosion
damaged region since the Mo content of the alloys is considerably
higher than the W content.
Fig. 10 shows Raman images of the corroded grain boundary
shown in Fig. 7b. The image intensities were integrated and sepa-
rated into the wavelength ranges characteristic for the three Ra-
man bands in an attempt to determine the distribution of
Raman-active species within the damaged area. Based on intensity
(yellow being the most intense), it appears that the Raman bands
residing in the 940985 cm
1
(Fig. 10c), attributed primarily to
the species Mo
8
O
4
26
(discussion above), dominate the damaged
areas followed by the band in the region 325364 cm
1
, also prob-
ably attributable to Mo
8
O
4
26
. Fig. 10 does not necessarily imply
these species are the most concentrated, but rather, shows how
these species are qualitatively distributed across the damaged
region. In general, the Raman active species observed in the
820856 cm
1
range was always the least distributed, i.e. most
localized. Quantifying these species would be difcult due the
complexities of in situ measurements within the constricted geom-
etry of a crevice.
3.3. Corrosion products outside the corrosion-damaged region
Fig. 11 shows micrographs of the solid corrosion products
formed just outside the damaged region on an Alloy-22 specimen
crevice corroded at 70 lA. These were optically different in size
and color from the corrosion products observed within the dam-
aged area. In this study the corroded region is dened as the area
40000
Alloy-2000 (without W)
Alloy-22 (with W)
30000
20000
10000 10000
250 500 750 1000 1250 1500
0
250 500 750 1000 1250 1500
cm
-1
cm
R
e
l
a
t
i
v
e

I
n
t
e
n
s
i
t
y

(
a
r
b
i
t
r
a
r
y

u
n
i
t
s
)
Fig. 9. High-resolution Raman spectra comparing the spectra recorded on corrosion
products on crevice corroded Alloy-22 (with W) and Alloy-2000 (without W)
crevice corroded at an applied current of 200 lA to an accumulated charge of 3 C.
Fig. 10. Raman images constructed from Raman signals recorded within the wavenumber ranges: (a) 325364 cm
1
, (b) 820856 cm
1
, and (c) 940985 cm
1
on the
corroded grain boundary regions shown in Fig. 6b.
crevice mouth
propagated region
(a) (b)
outside corroded
region/stained region
Fig. 11. (a) Alloy 22 specimen on which crevice corrosion was initiated and propagated galvanostatically at a constant applied current of 70 lA: (b) the area just outside the
crevice corroded region.
P. Jakupi et al. / Corrosion Science 53 (2011) 16701679 1675
in which corrosion damage is clearly observed. Upon disassembly
and rinsing of the creviced specimen, the corrosion products
observed in Fig. 11b stayed intact on the alloy surface and were
consistently observed outside the crevice corroded location on
specimens galvanostatically corroded at a variety of applied cur-
rents. The corrosion products were a greenish brown in color,
and cracked. The organization of the insoluble corrosion products
suggest that it was once a uniform layer and that cracking occurred
during drying. Fig. 12a shows a Raman intensity (yellow being
most intense) map of the corrosion products. The intensity is not
uniform across the insoluble layer and the resulting spectrum,
Fig. 12b, taken from a location of high intensity, showed similar
peaks to those observed within the damaged region, Fig. 8,
although they were not as sharp. This, and the observation that
the Raman intensity decreases with distance from the damaged re-
gion, suggests that Mo corrosion products were transported out of
the damaged region onto the non-corroded region. SEM/EDX was
used to determine the elemental contrast in alloying components
in this region. This analysis, Fig. 13, shows there is an enhancement
of chromium and a depletion of nickel on the region of the surface
covered with corrosion product akes (Fig. 13b) and vice versa for
a region where the bare alloy surface is exposed (Fig. 13c). Subse-
quent XPS analysis (Fig. 14) showed that the chemical content of
the akes is predominantly chromium, in particular the species,
Cr(OH)
3
(95%). A small amount of Cr
2
O
3
and Cr was also observed.
Both chromium(III) compounds are Raman active [48,49] but have
(a) (b)
400
300
200
1000 2000 3000 4000
(cm
-1
)
I
n
t
e
n
s
i
t
y

(
A
r
b
i
t
r
a
r
y

U
n
i
t
s
)
Fig. 12. (a) Raman intensity map of the creviced area immediately outside the corrosion damaged region; (b) the corresponding spectrum for an intense signal (yellow) area.
(For interpretation of the references in color in this gure legend, the reader is referred to the web version of this article.)
Fig. 13. SEM micrograph showing (a) the area just outside the crevice corroded region and the corresponding EDX spot analyses for the region covered with akes (b) and the
uncovered region (c).
1676 P. Jakupi et al. / Corrosion Science 53 (2011) 16701679
a low Raman scattering cross-section and are known to exhibit
signicant spectral broadness [50]. Mo(VI) was the predominant
molybdenum species observed in the XPS spectrum, which is con-
sistent with the presence of MoO
2
4
, and Mo
8
O
4
26
species identied
by Raman spectroscopy.
4. Discussion corrosion product impact on crevice corrosion
propagation
Crevice corrosion studies on Alloy-22 conducted by Shan and
Payer [51] showed a similar distribution of alloying elements as
observed in this study. Despite observing congruent dissolution
in the active region, corrosion products in the corroded region
were rich in O, Mo, and W and depleted in Ni and Cr. This was
attributed to a difference in solubilities of corrosion products at
low pH. Also, dark green deposits similar to those observed in
Fig. 11 were found on the uncorroded metal surface around the
crevice contacts.
Corrosion products such as H
2
gas and insoluble deposits
formed within the active region have been shown to promote ini-
tiation and subsequent propagation [52,53] by decreasing the cre-
vice gap and increasing the crevice electrolyte resistance R,
allowing the IR drop criterion to be satised [54].
Lillard et al. [16] compared the corrosion resistance of commer-
cial Ni-alloys in terms of their Mo content: Alloy-276 (1517 wt%
Mo), Alloy-625 (810 wt% Mo), and Alloy G-3 (68 weight% Mo).
They demonstrated that the corrosion resistance decreased in the
order Alloy-276 > Alloy-625 > Alloy G-3, which corresponds to
the decreasing Mo content. Alloying with Mo was shown to de-
crease the passive current density and lower the anodic dissolution
rate in the alloys active polarization region, consequently, decreas-
ing crevice corrosion propagation rates. It was shown for a series of
NiCrMo alloys (in 0.1 M H
2
SO
4
+ 1 M NaCl) that alloying with Mo
suppresses dissolution [2]. Lillard et al. [16] attributed the inhibit-
ing affects of Mo to the dissolution product MoO
2
4
, but were still
unclear as to the exact mechanism. Addition of sodium molybdate
(Na
2
MoO
4
) to the creviced alloys bulk electrolyte even caused a
delay in crevice initiation and a signicant decrease in propagation
rates [11,1517]. To ensure the inhibiting effect of MoO
2
4
was not
simply due to competitive migration with Cl

into the creviced re-


gion, Lillard et al. [16], for comparison, added Na
2
SO
4
in a separate
crevice experiment and showed that SO
2
4
did not display the same
inhibiting effects as MoO
2
4
. Many hypotheses have been proposed
[16] to explain the inhibiting nature of MoO
2
4
including that it is
thermodynamically unstable with respect to Mo polymeric species
at low pH, reactions (1) and (2). It has also been proposed that salt
lms, such as MoO
2
Cl
2
, contribute to the inhibiting behavior [15].
The bi-polar lm model proposes that the passive lm is com-
posed of an inner anion-selective and an outer cation-selective
layer. Alloying with Mo [55], as well as the addition of MoO
2
4
into
the electrolyte [17] was shown to enhance the inner barrier layer
and to improve the corrosion resistance of the Fe19Cr9Ni alloy.
This effect was attributed to the cation selective property of MoO
2
4
and its ability to deprotonate Cr(OH)
3
thereby facilitating the
formation of the stable Cr
2
O
3
inner barrier layer. Also, the outer
molybdate layer was shown to hamper the ingress of Cl

and
OH

effectively eliminating rehydration of this barrier layer.


Although this role has not been veried explicitly for the polymeric
molybdates identied in this study, it is plausible that the in-
creased negative charge on these species would enhance cation
selectivity and inhibit ingress of anions into the outer lm. Addi-
tionally, the thermodynamic instability of MoO
2
4
at low pH with
respect to Mo polymeric species comes at the expense of H
+
con-
sumption, reactions (1) and (2), which would reinforce the ten-
dency for an active site within the creviced region to repassivate.
This study suggests the inhibiting nature of alloyed Mo on active
crevice propagation is due to the formation of insoluble polymeric
molybdates, Mo
7
O
6
24
and Mo
8
O
4
26
. It is likely these species existed
as a polymeric gel during crevice corrosion and then dried into so-
lid form upon dismantling and drying of the crevice, Fig. 3.
Experimental evidence suggests that Mo has a greater impact
on the transpassive state then the passive state of NiCrMo oxi-
des. On Alloy-22, the protective barrier layer has been character-
ized primarily as Cr
2
O
3
[2], and with increasing applied potential,
from 200 to 500 mV (versus 0.1 M KCl, Ag/AgCl) in acidic solution
the oxide showed an increase in segregation resulting in an inner
CrNi layer and an outer MoW layer consistent with a stronger
inuence of Mo and W at more positive potentials. In studies on Al-
loy-22 under open-circuit conditions in acidic chloride environ-
ments [56,57], the impedance response indicates that the
composition of the oxide changes with pH and exposure time
and, in a complimentary study, the authors characterized the ma-
jor transpassive dissolution product as Mo [36]. These studies sug-
gest that once the integrity of the oxide is threatened, in acidic
solutions and at positive potentials the inuence of Mo becomes
more pronounced.
Although the presence of Mo containing corrosion products has
been demonstrated, there is still uncertainty as to the exact Mo
dissolution product. Since MoO
2
4
is thermodynamically unstable
at pH < 3 [58] it is unlikely that this species is the major dissolution
product during crevice propagation. Wanklyn [11] demonstrated
that crevice corrosion of stainless steels (Type 430 and 316) and
the Ni-alloy, Inconel 625, were inhibited only when Mo
6+
, rather
than Mo
3+
and Mo
4+
were added to the crevice electrolyte. It has
been suggested [16] that during passive dissolution within the
creviced region, MoO
2
4
would be formed and eventually exert an
inhibiting effect once a decrease in pH and crevice propagation oc-
curs. Additionally, it was shown that Mo
6+
could be formed by ano-
dic dissolution of MoO
2
grown oxides at pH 2.4 [11],
MoO
2
2H
2
O !MoO
2
4
2H

2e

5
However, the above reaction suggests that Mo is dissolved from
a barrier layer which is unlikely for NiCrMo alloys which have
been shown to have predominantly Cr barrier layers. This would
then suggest that the Raman signal at 355 cm
1
observed in this
study would only be attributable to the oxide MoO
3
, and the poly-
meric species, Mo
8
O
4
26
.
Fig. 15 summarizes the distribution of alloying elements ob-
served in the present study. The SEM micrograph depicts a region
7000
Cr 2p
3/2
6000
C p
3/2
Cr(OH)3
5000
4000
3000
575 580
Binding Energy (eV)
i
n
t
e
n
s
i
t
y

(
A
r
b
i
t
r
a
r
y

U
n
i
t
s
)
Fig. 14. The Cr 2p
3/2
XPS band measured on the corrosion products shown in
Fig. 12a. The dashed line shows the t for Cr(OH)
3
.
P. Jakupi et al. / Corrosion Science 53 (2011) 16701679 1677
just outside the corrosion damaged region of the creviced speci-
men shown in Fig. 11, and shows three distinct regions: (1) a dam-
aged/corroded region decorated with intergranular corrosion; (2) a
region covered with insoluble chromium hydroxide akes; and (3)
a region, still underneath the crevice former, that resembles the
freshly polished passive alloy. Based on the pH and its inuence
on the solubilities of the major alloying components, Ni, Mo, and
Cr, we can say that Fig. 15 illustrates the consequences of a pH gra-
dient stretching from the crevice corrosion-damaged region 1, with
low pH and concentrated in insoluble polymeric molybdates,
through region 2 covered by chromium hydroxides insoluble at
higher pH, to region 3, which remains unaffected by crevice corro-
sion. Given the pH dependence of the solubility of chromium(III)
[58], its precipitation as an insoluble corrosion product would be
expected after only a small increase in pH. Also, as suggested by
the CCS model [26,27], extremely low pH values are required to
maintain the localized active site. Therefore the location of the
insoluble chromium hydroxides delineates the extent of the cor-
roded region, as shown in Fig. 15.
5. Conclusions
Insoluble corrosion products were shown to accumulate within
the corroded regions of creviced Alloy-22 and Alloy-2000 speci-
mens, especially within corroded grain boundaries. The damaged
regions were more concentrated in Mo and O (and to a lesser de-
gree, W, in the case of Alloy-22) relative to the non-damaged re-
gions, while Ni and Cr were depleted. The Mo-containing
compounds were identied via Raman spectroscopy as the oxide
MoO
3
, as well as the polymeric species Mo
7
O
6
24
and Mo
8
O
4
26
. Pub-
lished literature suggests that the Raman peak centered at
835 cm
1
could be attributed to the oxide Mo
4
O
11
and/or the poly-
meric species Mo
36
O
8
112
. However, this study alone could not re-
solve whether these species coexisted with the identied
polymeric species. For Alloy-22, analogous tungstate compounds
may also have been formed within the corroded region, but similar
Raman signals for Alloy-22 and Alloy-2000 (no W content) could
not easily be de-convoluted to verify this. Polymeric species were
formed within the corroded region as a result of the relative ther-
modynamic instability of MoO
2
4
(and possibly WO
2
4
for Alloy-22)
at low pH suggesting that alloying Mo plays a signicant role in
controlling crevice propagation on NiCrMo alloys.
Acknowledgements
The authors thank the Science and Technology Program of the
Ofce of the Chief Scientist (OCS), Ofce of Civilian Radioactive
Waste Management (OCRWM), and the United States Department
of Energy (DOE) for support. The work was performed under the
Corrosion and Materials Performance Cooperative, DOE Coopera-
tive Agreement No. DE-FC28-04RW12252. The views and opinions
of the authors expressed herein do not necessarily state or reect
those of the United States Department of Energy.
The authors would also like to acknowledge Zhifeng Ding for
use of the Raman apparatus and Surface Science Western for the
use of their facilities.
References
[1] D.C. Agarwal, Adv. Mater. Proc. 158 (2000) 2731.
[2] A.C. Lloyd, J.J. Nol, S. McIntyre, D.W. Shoesmith, Electrochim. Acta 49 (2004)
30153027.
[3] J.R. Hayes, J.J. Gray, A.W. Szmodis, C.A. Orme, Corrosion 62 (2006) 491500.
[4] R.C. Newman, Corros. Sci. 25 (1985) 331339.
[5] R.C. Newman, Corros. Sci. 25 (1985) 341350.
[6] P. Marcus, V. Maurice, in: M. Schutze (Ed.), Corrosion and Environmental
Degradation, Wiley-VCH, New York, 2000 (Chapter 3).
[7] P. Marcus, Corros. Sci. 36 (1994) 21552158.
[8] I. Olefjord, L. Wegrelius, Corros. Sci. 31 (1990) 8998.
[9] A. Schneider, D. Kuron, S. Hoffman, R. Kirchheim, Corros. Sci. 31 (1990) 191
196.
[10] J.M. Bastidas, C.L. Torres, E. Cano, J.L. Polo, Corros. Sci. 44 (2002) 625633.
[11] J.N. Wanklyn, Corros. Sci. 21 (1981) 211225.
[12] K. Sugimoto, Y. Sawada, Corros. Sci. 16 (1976) 323337.
[13] J. Postlethwaite, R.J. Scoular, M.H. Dobbin, Corrosion 44 (1988) 199203.
[14] W. Yang, R.C. Ni, H.Z. Hua, A. Pourbaix, Corros. Sci. 24 (1984) 691707.
[15] M.A. Cavanaugh, J.A. Kargol, J. Nickerson, N.F. Fiore, Corrosion 39 (1983) 144
150.
[16] R.S. Lillard, M.P. Jurinski, J.R. Scully, Corrosion 50 (1994) 251265.
[17] Y.C. Lu, C.R. Clayton, A.R. Brooks, Corros. Sci. 29 (1989) 863880.
[18] X. He, J.J. Nol, D.W. Shoesmith, Corros. Sci. 47 (2005) 11771195.
[19] J.J. Nol, Ph.D. Thesis, Department of Chemistry, University of Manitoba,
Winnipeg Manitoba, 1999.
[20] P. Jakupi, D. Zagidulin, J.J. Nol, D.W. Shoesmith, ECS Trans., Cancun, Mexico,
29-3, Oct./Nov., ECS, 3 (2007) 259271.
[21] www.casaxps.com (last accessed April 2010).
[22] M.C. Biesinger, C. Brown, J.R. Mycroft, R.D. Davidson, N.S. McIntyre, Surf.
Interface Anal. 36 (2004) 15501563.
[23] P. Jakupi, J.J. Nol, D.W. Shoesmith, Corros. Sci., in press.
[24] P. Jakupi, J.J. Nol, D.W. Shoesmith, Corros. Sci., in press.
[25] P. Jakupi, J.J. Nol, D.W. Shoesmith, Electrochem. Solid State Lett. 13 (2010)
C1C3.
[26] J.W. Oldeld, W.H. Sutton, Brit. Corros. J. 13 (1978) 1322.
[27] J.W. Oldeld, W.H. Sutton, Brit. Corros. J. 13 (1978) 104111.
[28] N. Sridhar, D.S. Dunn, Corrosion 50 (1994) 857872.
[29] F. Bocher, F. Presuel-Moreno, J.R. Scully, J. Electrochem. Soc. 155 (2008) C256
C268.
[30] C. Heitner-Wirguin, R. Cohen, J. Inorg. Nucl. Chem. 26 (1964) 161166.
[31] C.F. Baes, R.E. Mesmer, The Hydrolysis of Cations, rst ed., Wiley-Interscience,
1976.
[32] K.F. Jahr, J. Fuchs, Angew. Chem. Int. Edn. 5 (1966) 689699.
[33] K.H. Tytko, B. Schoenfeld, B. Buss, O. Glemser, Angew. Chem. 85 (1973) 305
307.
[34] J. Aveston, E.W. Anacker, J.S. Johnson, Inorg. Chem. 3 (1964) 735746.
[35] Y. Sasaki, Acta Chem. Scand. 15 (1961) 175189.
[36] J.J. Gray, J.R. Hayes, G.E. Gdowski, C.A. Orme, J. Electrochem. Soc. 153 (2006)
B156B161.
[37] M. Dieterle, G. Mestl, Phys. Chem. Chem. Phys. 4 (2002) 822826.
[38] M.A. Py, K.A. Maschke, Physica B 105 (1981) 370374.
[39] P.A. Speavack, N.S. McIntyre, J. Phys. Chem. 97 (1993) 1102011030.
[40] S. Himeno, H. Niiya, T. Ueda, Bull. Chem. Soc. Jpn. 70 (1997) 631637.
[41] O.F. Oyerinde, C.L. Weeks, A.D. Anbar, T.G. Spiro, Inorg. Chim. Acta 361 (2008)
10001007.
[42] N. Weinstock, H. Schulze, J. Chem. Phys. 59 (1973) 50635067.
[43] J.S. Cross, G.L. Schrader, Thin Solid Films 259 (1995) 513.
[44] K.Y.S. Ng, E. Gulari, Polyhedron 3 (1984) 10011011.
[45] C.C. Williams, J.G. Ekerdt, J.-M. Jehng, F.D. Hardcastle, I.E. Wachs, J. Phys. Chem.
95 (1991) 87918797.
[46] K.Y.S. Ng, E. Gulari, J. Catal. 92 (1985) 340354.
[47] C.P. Cheng, G.L. Schrader, J. Catal. 60 (1979) 276294.
[48] C.A. Melendres, M. Pankuch, Y.S. Li, R.L. Knight, Electrochim. Acta 37 (1992)
27472754.
[49] J. Zhao, G. Frankel, R.L. McCreery, J. Electrochem. Soc. 145 (1998) 22582264.
[50] B.M. Weckhuysen, I.M. Wachs, J. Phys. Chem. 100 (1996) 1443714442.
Fig. 15. SEM micrograph of a crevice corroded Alloy-22 specimen showing: (1) the
corrosion-damaged region with corroded grain boundaries containing polymeric
molybdates; (2) insoluble chromium hydroxides consistently found just outside
region 1; and (3) the non-corroded creviced region still showing polishing marks.
1678 P. Jakupi et al. / Corrosion Science 53 (2011) 16701679
[51] X. Shan, J.H. Payer, J. Electrochem. Soc. 156 (2009) C313C321.
[52] J.N. Al-Khamis, H.W. Pickering, J. Electrochem. Soc. 148 (2001) B314B321.
[53] A.M. Al-Zahrani, H.W. Pickering, Electrochim. Acta 50 (2005) 34203435.
[54] H.W. Pickering, J. Electrochem. Soc. 150 (2003) K1K13.
[55] C.R. Clayton, Y.C. Lu, J. Electrochem. Soc. 133 (1986) 24592473.
[56] J.J. Gray, C.A. Orme, Electrochim. Acta 52 (2007) 23702375.
[57] J.J. Gray, J.R. Hayes, G.E. Gdowski, B.E. Viani, C.A. Orme, J. Electrochem. Soc. 153
(2006) B61B67.
[58] M. Pourbaix, Atlas of Electrochemical Equilibria, Section 10.2, Pergamon,
Oxford, 1966.
P. Jakupi et al. / Corrosion Science 53 (2011) 16701679 1679

You might also like