Sunteți pe pagina 1din 12

Ekctrochimica

Acta, Vol. 39, No. 14, pp. 2225-2236, 1994


Copyripht Q 1994 Ellcvicr scktn
Ltd.
Printedin &at Britain. AU righti mervcd
OolS-4686i94 s7.00 + 0.00

0013486@4)E0171-U

RAMAN SPECTRAL AND ELECTROCHEMICAL


STUDIES
OF SURFACE FILM FORMATION ON IRON AND ITS
ALLOYS WITH CARBON IN Na&OJNaHCO,
SOLUTION
WITH REFERENCE TO STRESS CORROSION CRACKING
M. ODZIEMKOWSKI,*
J. FLIS~ and D. E. IRISH*
*Guelph-Waterloo Centre for Graduate Work in Chemistry, Waterloo Campus,
Department of Chemistry, University of Waterloo, Waterloo, Ontario, Canada N2L 3Gl
TInstitute of Physical Chemistry of the Polish Academy of Sciences, 01-224 Warsaw, Poland
(Received 24 January 1994)

Abstract-In

situ and ex situ Raman microspectroscopy and electrochemical measurements on rotating


disc electrodes have been used to identify surface films formed on iron and its alloys with carbon in 0.5 M
Na,CO, + 0.5 M NaHCO, at 75C and potentials where susceptibility and resistance to SCC occur. A
good quantitative correlation between data of in situ Raman spectroscopy, galvanostatic surface film
reduction, and SCC tests was found. At active dissolution potentials, where transgranular SCC was
observed, the iron-carbon alloys surfaces were blocked by FeCO, or FeCO, . H,O salt films. At potentials of the active-passive transitions, where intergranular SCC was noticed, in situ Raman spectroscopy
revealed the formation of Fe,O, for both decarbonized iron and iron-carbon alloys. The magnetite film,
formed on iron-0.7% C at potentials of the maximum SCC susceptibility, in contrast to films formed on
decarbonized iron, was found to be strongly heterogeneous and often incorporated FeCO,. At more
positive potentials, ie -0.6V (see) at which the resistance to SCC was observed, galvanostatic reduction
and in situ Raman spectroscopy revealed formation of FeOOH. At this potential Raman signals from
Fe,O, were not detectable. It is suggested that intergranular SCC is favoured by the electrochemical
conditions for which ferric oxides or oxyhydroxides are in equilibrium with FeaO, and/or Fe(H) species.
Key words: stress corrosion cracking, iron-carbon alloys, Raman spectroscopy.

INTRODUCTION
The mode of stress corrosion cracking (SCC) of low-

carbon steels at elevated temperatures in 0.5 M


Na,CO, + l.OM NaHCO,
solution varies with
potential: it is transgranular in the active potential
region and intergranular in the active-passive transition region[l, 23. Sutcliffe et aI.[l] have suggested,
on the basis of ex situ investigations, that salt layers
(ie iron carbonate) favour transgranular SCC, while
magnetite films favour intergranular SCC. According
to Wendler-Kalsch[2] the SCC suceptibility sharply
decreases, and eventually disappears at the potential
threshold between the formation of magnetite Fe,O,
and iron oxyhydroxide, y-FeOOH. The very recent
semi in-situ surface enhanced Raman spectral studies
of passive films formed on silver-coated iron electrodes in 0.05 M Na,CO, + 0.75 M NaHCO, confirmed the presence of an iron(I1) carbonate in the
passive film, and for particular conditions, Fe,O,[3].
The formation of magnetite together with a radioactive species in the primary recirculation systems of
water-cooled nuclear power plants is detrimental; it
increases the exposure of personnel to radiation
during maintenance and repair. Much effort has
been spent to reduce the build-up and/or effective
EA 39-14-I

removal of activity, ie the decontaminationdissolution of magnetite from a pipeline steel[4].


Carbon is an essential metallurgical component of
steels which significantly affects SCC and film formation in various electrochemical systems[5, 63; it is
believed that it can also affect the film formation and
repassivation processes in the CO:-/SCOT
buffer
solution.
Under severe conditions of SCC, the loss of reflectivity and roughening of the electrode eliminate the
applicability of many in situ optical techniques for
the study of surface film formation. The intrinsic low
sensitivity of conventional normal Raman spectroscopy was improved by the development of the
Raman microprobe and multichannel
detection
which have allowed in situ identification of corrosion
products on iron in 1 M NaOH[7]. Some experiments have been reported in which surface enhanced
Raman scattering has been achieved by deposition of
a discontinuous silver layer over films on pure iron,
allowing identification of film components[3, 8, 91.
With the exception of Fe-Cr alloys[lO, 111, little
work utilizing in situ Raman spectroscopy has been
done with iron alloys, particularly at elevated temperatures.
In this work we have used in situ micro-Raman

2225

M. ODZIEMKOWSKI et al.

2226

spectroscopy and electrochemical measurements on


the rotating disc electrode to gain a better understanding of the influence of carbon on the passivation of iron in 0.5 M NaCO, + 0.5 M NaHCO, at
75C, at potentials where susceptibility and resistance to SCC occur.
EXPERIMENTAL
Experiments were carried out on decarbonized Fe
and two Fe-C alloys: Fe + 0.25% C, and Fe + 0.7%
C. The quantities of impurities were as follows: S
(0.003), P (O.OOS),Si (0.013), Mn (0.051), Cr (0.023), Ni
(0.027), Cu (0.027) and Al (O.O03)wt.%. Cylindrical
samples were annealed in purified argon at 950C for
0.5 h and then quenched in oil. The quenching was
performed in an attempt to retain carbon in the iron
matrix in a fairly uniform solid solution, ie a martensitic structure. Quenched specimens with high
carbon concentrations can simulate material of grain
boundaries in mild steel, enriched in carbon as a
result of segregation processes. Carbon-free samples
were obtained by annealing the cylindrical specimens
of Fe-0.008% C in flowing wet hydrogen at 850C
for 60 h. For electrochemical measurements peripheral surfaces of each cylindrical specimen (10mm
in diameter) were insulated with a thin layer of
Teflon and the specimen was then pressed into the
Teflon holder. After electrochemical measurements,
exactly these same specimens, but 4.75mm in diameter embedded in an epoxy resin, were used for
Raman spectral studies. The front surface was polished with 1200 grade Sic emery paper, washed in
copious amounts of Milli-Q or double distilled
water, and ultrasonically cleaned three times. All
measurements were carried out at 75C in deaerated
0.5 M Na,CO, + 0.5 M NaHCO, (pH 10.05) prepared from double distilled or Milli-Q water and AR
chemicals.
Raman spectra were obtained with a Dilor
OMARS-89 spectrometer equipped with a 512
channel diode-array optical multichannel analyser
and an Olympus microscope Model BHT. The spectrometer was operated by using the 18OOgmm-
gratings with two gratings operating in subtractive
mode and a third grating dispersing the output onto
the 512 diode array detector. For data acquisition
the spectrometer was interfaced to an IBM PC-AT
computer. The spectra were transferred to a PC
WAILAB486 microcomputer for spectral treatment
and analysis. When required, spectra were baseline
corrected and bandfitted, imposing a Lorentzian
and/or Gaussian profile on the bandshape. All
spectra presented here are given with f 1 cm- 1 accuracy and without smoothing.
Excitation was achieved using the 514.53nm line
from a Coherent Innova 70 argon ion laser. The
objective lens used in the spectral studies has an
objective magnification
50 (OLYMPUS
IC50/
MSPLAN 50, co/Of= 180, NA =0.55). The output
laser power was 150 or 60mW; this translates to ca.
26mW and ca. 1OmW at the sample surface, respectively. The powers were obtained by removing the
objective and placing a power meter at the sample
location. To minimize local heating for in situ

measurements a slightly defocused 1OmW laser


beam was used.
A three-compartment (ie three electrode) spectroelectrochemical cell used previously by Stolberg[ 121
was redesigned to allow work at elevated temperatures under the Raman microscope. Potential
was measured against the saturated calomel electrode (see). All measurements were started after
cathodic cleaning at a potential of - 1.05 V for 300s
for electrochemical and 1800s for spectroelectrochemical measurements. Specimens for Raman spectral studies were placed horizontally at a distance of
1Omm from an optically flat window for the
cathodic cleaning. After 0.25 h from the onset of the
anodic potential, without losing potential control (ie
continuous polarization), the specimens were moved
to a distance of ca. 0.5 mm from the optical window.
The
anodic
potentiodynamic
polarization
measurements were carried out on stationary electrodes (Orps) and on rotating electrodes with the
rotation velocity of 49 rps and a potential sweep rate
of 1 mVs_. Cathodic reduction of surface films was
performed after potentiostatic anodic polarization of
iron-carbon alloys at -0.8, -0.75, -0.7, - 0.65 and
-0.6V (see) for various times; a potentiostat was
disconnected (disconnection lasted 10e6 s) and the
potential changes were recorded during cathodic
reduction at a current density of 20pA cm- .

RESULTS
Potentiodynamic

anodic polarization measurements

Typical potentiodynamic
anodic polarization
curves for iron-carbon
alloys in 0.5M Na,CO,
+ 0.5 M NaHCO, buffer are shown in Fig. 1 for a
stationary electrode and Fig. 2 for a rotating disc
electrode, respectively. For the stationary electrodes
Fig. 1 shows that in the active region the anodic
current was almost independent of the increasing
carbon content, whereas in the passive region the
current rose when the carbon content increased. The
anodic currents in the active-passive transition region
were considerably higher for the rotating electrodes
than for the stationary ones. This demonstrates that
some soluble corrosion products or loose deposits,
which can be removed by rotation, inhibit anodic
dissolution and promote passivation. Davies and
Burstein[13] found that soluble complex anions,
Fe(CO&-, form in CO:-/HCO;
solutions and that
these decrease the dissolution rate of iron. Probably,
the effect of the rotation (Fig. 2) is partly due to a
decrease in concentration
of Fe(CO&at the
surface. From SCC tests under constant low extension rates and from scanning electron microscopy
studies[14] we know that, in the solutions used,
intergranular SCC is most distinct at a potential of
-0.7 V (see). At this potential, the anodic current for
stationary Fe-0.7% C was larger than for decarbonized Fe and Fe-0.25% C, indicating stimulation of
anodic dissolution of iron by carbon.
The possible reactions of iron in the solution used
can be deduced from the equilibrium potentials in
the Fe-H,0 system. Vertical lines in Fig. 1 mark the
potentials for the formation of Fe(OH), (line l),

2227

Surface film formation of iron and its alloys

lo2i

I~~~~~25ri~
- ------_______
60-

Fe + 0.25%C

50

1. _._._.-.-.-.-

10-l
1

-0.6

-0.6

-0.4

-0.2

._._.

--A--

Fe + 0.7 % C

_._._._._._.zr~=
I

0.2

0.4

0.6

06

E I VISCEI

Fig. 1. Potentiodynamic anodic polarization curves for FeC alloys. Stationary electrodes, rotation speed OHz. Vertical lines indicate equilibrium potentials at pH 8.5
(Ref.[lS]) for the couples: (1) Fe/Fe(OH),; (2) Fe/Fe,O,;
(3) Fe(OH)JFe,O,; (4) Fe(OH)Ja-FeOOH; (5) FesOJ
a-FeOOH; (6) Fe(OH),/y-FeOOH.

0;
Fe,O, (lines 2 and 3), and a- or y-FeOOH (lines 4-6)
calculated for pH8.5 (this is the corrected pH for
75C for a solution of pH 10.05 at 25C)[15, 163.
Solid corrosion
products
on steel in hot
CO:-/HCO;
can involve FeCO, and Fe,O, in the
active region, and additionally a-Fe,O, or other
Fe(W) species in the passive region[17].
The active dissolution potential region; potential
-0.8 v (SIX)
The rotating disc electrode is particularly useful in
kinetic studies of heterogeneous reactions because
the thickness of the diffusion layer is independent of
the disc radius[18]. In order to determine the
kinetics of dissolution of Fe-C alloys, the anodic
current densities were measured as a function of the
disc rotation speed (Fig. 3). These currents were
obtained after a fast potential step from -1.05 to
-0.8V (see) and represent steady state values. The
shape of the i vs. (o/2n) /* dependence suggests (a)
diffusion control of a corrosion process where only a
part of the entire disc surface is diffusion-active, or
(b) mixed kinetics of the corrosion process, ie when
the rate of the chemical transformation at the inter-

10-y, , , , , , , , , , , , , , , , , 1
-0.6

-0.6

-0.4

-0.2

0.2

0.4

0.6

0.6

E.VISCEl

Fig. 2. Potentiodynamic anodic polarization curves for


FCC alloys. Rotating electrodes, rotation speed 49 Hz.

11

31

L1

61

71

1 d2

kd2nl*

Fig. 3. Corrosion rates of FeC alloy discs vs. square root


of the disc speed, rps, in deaerated 0.5 M Na,CO, + 0.5 M
NaHCO,, 75C.

face is comparable with the transport rate[18]. The


results of Fig. 3 can be further understood using an
approximate equation derived by Landsberg and
Thiele[19] for the case of diffusion control with a
partially blocked surface and at a moderate disc
rotation speed, ie r 6 6 (see Fig. 4):
1
_=-hm

1.61~~
nFD213co

1
&

: LA
nFDco

Here the radius r of the active surface fragments and


the mean distance between their centres 2r are
smaller than the diffusion layer thickness, 6; v is the
kinematic viscosity of the solution, D is the diffusion
coefficient of the reagent species, co is the concentration of the diffusing species in bulk solution, A, is a
constant, and the other symbols have their usual
meaning. The linearity of i- vs. (~/27r)-/~ (see Fig.
4) might suggest the diffusion control of the anodic
dissolution with partial blocking of the disc surface.
The different behaviours of Fe-C alloys compared to
pure Fe (Figs 3 and 4) might have their origin in a
diminished radius of the active part of the disc and
an increase of the distance between active parts in
samples with a large amount of carbon and/or in the
different surface films formed on Fe-C alloys compared to purified Fe. The results of in situ Raman
spectroscopy for identification of the substances
blocking the iron surface are now presented.
Raman spectroscopy
In Fig. 5, Raman spectra in
region
of deaerated
0.5M
NaHCO, solution at 75C (Fig.
in situ Raman spectra obtained
decarbonized Fe (Fig. 5b) and

the 400-1200cm-1
Na,CO, + 0.5M
5a) are compared to
from surfaces of the
Fe + 0.25% C (Fig.

2228
0.075

E=

0.8 V ISCEI

,$

/.."
,<.'
,,<..'
Fe + 0.25% C
;<:
Fe + 0.7% C
,,$"
,,,."
,+:."
,.p ,"

Fe decrrbonizrd

..m.
-.-

0.05
i
R

ODZIEMKOWSKI
et al.

M.

5---

L
0.025

0
0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.6

/ s2

1w/2n1-2

Fig. 4. The results of Fig 3 plotted in the coordinates:


according to equation ( 1).
l/i,, = f( l/J&

1065

E - - 0.6 V (SCEI

1062

bicarbonate ions, respectively[20, 213. The weak


band at 842 cm-, observed at 841 cm- by Oliver
and Davis[21] in 3.5 M KHCOs , may arise from the
COs out-of-plane deformation of bicarbonate ion.
Note that the very weak[20, 211 lower frequency
bands at 681cm- for saturated Na,CO,[20], and
672 and 632cm- for 3.5M KHCO,[Zl]
were not
observed in 0.5 M Na,COs + 0.5 M NaHCOs with
the low laser powers employed in these experiments.
The Raman spectrum of the surface of decarbonired
Fe after polarization at -0.8 V (see) is characterized
by the appearance of two very weak bands at 541
and 565cm-. A broad band at 545cm-* was
recently detected on silver-coated iron electrodes in
0.05 M Na,CO, + 0.75 M NaHCO, and was attributed to the presence of Fe(OH), species in the
surface film[3]. The appearance of a prepassive
Fe(OH), film at the active potential region for iron
electrodes was previously suggested by many
authors[3, 13, 221. Thus on the basis of these works
and our previous ellipsometric studies[23] we can
possibly attribute these bands to the presence of
Fe(OH), on the surface of decarbonixed iron. This
assignment is rather tentative since these bands are
very weak and the expected stronger (O-H) band
above 36OOcm- was not detected. The Raman
spectrum of a surface of Fe + 0.25% C (Fig. 5c) is
remarkably different from that of decarbonixed iron.
Namely, the weak bands at 541 and 565cm- are
missing; instead a strong band at 1082cm- was
observed for surfaces from both Fe + 0.25% C and
Fe + 0.7% C alloys (see Fig. 6). This new band at
1082 cm- , which appears upon active dissolution, is
assigned to the symmetric stretch of carbonate
anion; it originates from the presence of FeCO, or

70

dscrrbonized

600

Fe

600

Fe + 0.7% C, E - - 0.6 V (SCE)


1065

642

1000

Raman Shift / cm-


Fig 5. Raman spectra in the region 400-12OOcm- of: (a)
deaerated 0.5 M Na,CO, + 0.5M NaHCO,, 75C; (b)
decarbonized Fe in situ;(c) Fe + 0.25% C in situ.
5c), after 1 h* polarization at -0.8 V (see). As visible
in Fig. 5, bands at 1065, 1010 and 842~11~ orig-

inate from the electrolyte. The strongest bands at


1065 and lOlOcm- have been assigned to the symmetric stretching vibration of the carbonate and
l In this publication
with reference to Raman spectral
studies, time t represents the time interval from the onset of
the anodic polarization to the moment when the collection
of Raman spectra was started. A minimum time of 20min
was required to collect Raman spectra with an acceptable
signal-to-noise ratio.

650

950

1050

1150

Raman Shift / cm-


Fig. 6. Development of the iron carbonate signal at
1082cm- with the time of anodic polarization for
Fe + 0.7% C electrode.

Surface film formation of iron and its alloys

FeCOJ * Hz0 in the surface films formed on ironcarbon alloys. The Raman spectrum of a natural
crystal of FeCO, (siderite) is well known; it is characterized by two strong bands-an
internal mode at
1089 and an external lattice vibrational mode at
289cm-[24].
Only the strongest fundamental, ie
the symmetric stretch, was observed in our studies.
The mapping of the surface of decarbonized iron
and iron-carbon alloys with the cc. 4/.~rn laser beam
was also performed. These studies revealed a strong
heterogeneity of the surface film formed on Fe-C
alloys, in contrast to purified iron. For purified iron
the Raman spectrum shown in Fig. 5b was representative of the entire surface, ie for ten arbitrarily
chosen areas of the electrode surface the spectra were
quantitatively
very similar. For iron-carbon,
the
spectra taken from different areas of the electrode
quantitatively differed from each other. For example,
after 1 h of polarization at -0.8 V (see), ca. 50% of
the area of the Fe + 0.7% C electrode was free of
detectable 1082cm- Raman signal from FeCO, .
With the increase of the polarization time the surface
coverage increased; after cc. 2 h almost the entire
surface of the Fe + 0.7% alloy was covered with an
iron carbonate film.
Raman spectra of magnetite

Raman
spectra
of magnetite
have
been
reportedC25, 261. The discrepancy between the data
of various authors is evident by example of the position of the strongest A,, mode of magnetite; it has
been reported at posittons ranging from 663 to
706cm-[27$
In this study we used three types of
magnetite, namely (a) thick-porous magnetite crystallized from solution, (b) a thin layer magnetite
grown by a solid state 8lm mechanism, and (c)
natural magnetite from the sodalite mine in Bancroft, Ontario, Canada. The Raman microprobe
spectra of these reference materials were obtaind ex
situ in the laboratory atmosphere. A Raman spectrum of the thick-porous magnetite was not detectable. Material decomposition under even a very low
laser power (less than 1OmW) was evident from
microscope observations. Raman spectra of a thin
layer of magnetite, grown by a solid state mechanism, reveal the presence of two bands at 669 and
54Ocm- (Fig. 7a) and weak higher frequency
signals possibly from unknown impurities. The
theoretically-predicted
bands in the low frequency
region were not observable (Fig. 7a inset). Only the
Raman spectrum of the natural magnetite exhibited
all the predicted phonon energies at 294, 319, 415,
540 and 669cm- (Fig. 7b, c) with symmetries TZpr
E , T,,, T,, and A,,, respectively[26]. The observed
Pi onon frequencies, while different from many reported data, are in good agreement with theoretical
considerations and data reported by Hart et aI.[26].
For strong light absorbing materials such as magnetite, as for Raman spectra of carbon fibres[28],
localized sample heating is expected to distort the
lattice modes. Indeed, as visible in Fig. 8, the
increase
of laser power
from
1OmW (ca.
200mWmm-2)
to ca. 26mW (ca. 520mWmm-)
shifts the Tzs mode at 540 and the A,, mode at
669 cm- to higher frequencies by ca. 5 cm-. For
natural magnetite (Fig. 8a, b), the full width at half

2229

L
.g

Ramrn

Shift
669

I ea
-

I
400

500

600

700

600

Raman Shift / cm-


Fig. 7. Raman spectra of magnetite reference samples in
the 400-800an- region: (a) a thin layer of magne.tite
grown by a solid state tilm mechanism; (b) natural magnetite from the sodalite mine in Bancroft, Ontario, Canada.
Inset shows Raman spectra of magnetite reference samples
in the 200-SOOcm- spectral region.

maximum (FWHM) also increases from 34 to


42cm- for the A, mode. This unusually high
FWHM value for a phonon is in excellent agreement
with data reported by Hart et aI.[26]. Furthermore
when 100 x magnification was used (ie higher power
densities, cu. 800mW mm-), for both types of magnetite the symmetry of the A, band was lost (Fig.
8~). The computer-generated band shape revealed
the presence of a shoulder at ca. 692 cm- i. The band
position and microscopic observation suggests that
this spectral feature is not due to the magnetite
phonon but it arises from a new phase, y-FezOs. In
the presence of oxygen y-Fe,O,, not haematite aFe,OB, is expected to be the principal oxidation
product of magnetite[lO]. These results, together
with previous observations[lO] lead us to conclude
that for a correct Fe,O, band identification not only
well-defined reference material is necessary, but laser
power (possibly low), its stability, and the time of
illumination (possibly short) and the presence or
absence of oxygen are critical factors.
Cathodic reduction of surfacefilms

The potential
changes during
galvanostatic
reduction of surface films formed on decarbonized
Fe and its alloys with C, after 300 s and 2 h anodic
polarization, are presented in Figs 9 and 10, respectively. Cathodic charging curves showed three potential arrests A, B and C which, on the basis of the
equilibrium potentials (Fig. 1). can be attributed to
the reduction of Fe(II1) species, FesO, or Fe(OH), .

M.

2230

ODZIEMKOWSKI

Arrest A occurs after polarization at -0.65 and


-0.6OV, or -0.7V (see) for a rotating Fe-0.7% C
electrode; it lies between the equilibrium potentials
for the couples Fe(OH),/y-FeOOH
and Fe,O,/aFeOOH. Evidently, this potential arrest is associated
with the reduction of Fe(II1) species. Arrests B and C
are close to the equilibrium potentials of the couples
FesO,/Fe and Fe(OH),/Fe, respectively, hence they
can be associated with the reduction of Fe,O,
and/or Fe(OH), to iron. These arrests appear after
anodic polarization in the active and active-passive
transition regions.
From the cathodic charging curves the following
can be concluded :
(i) Fe(III) species (probably FeOOH) form at
about -0.7V (see) and at nobler potentials; the
amount of these species increases with the polarization time. The charging curves do not exhibit an
arrest B, so it can be supposed that FeOOH is
reduced to Fe(OH), , but not to Fe,O, . The rotation
of the disc electrodes results in an increase of the
amount of these species; this might be due to
removal of Fe&O,):or loose deposits which
hinder the passivation. The presence of carbon in the
material hinders the formation of these species
(shorter arrest A, Fig. 10).
(ii) The charge of the arrest A provides some information on the thickness of the film composed of
Fe(II1) species. Assuming the roughness coefficient
equal to unity, and density of Fe,O, and FeOOH
equal to 5.1 and about 4.Ogcm-, respectively, the
would correspond
to the
charge of 1 Cm-

et al.

0.5M

Na2C03

0.5M No%03

75c,3005
---Fe+O.ZS%C

-Fe

-0.6;

_0.7_
z
2 -0.6
T
aLi
-0.9

12

2L

Orps

\ 5.
\ .*...
\, -....
%_
:L

Bsn*n
0
12

36
-0.7

. . . . . . Fe*O.7%C

2C

36

...... . . . . . . . . . . . . . . .
----___

Fig. 9. Galvanostatic cathodic reduction curves of surface


films formed on Fe-C alloy electrodes, 300s after the onset
of anodic polarization. Arrows indicate arrests A, B and C,
and equilibrium potentials for the couples given in Fig. 1.

669
671

A+/

shouldel
at 692

-0.6

600

Raman

700

Shift

600

I cm-

Fig. 8. Influence of laser power density on the 7s and the


modes of magnetite; (a) ZOOmWmm-*; (b)
AI
52bmWmm-*; and (c) 100x magnification objective,
800mW mm-*. These spectra represent an average original
(ie no mathematical data treatment) signal from 150 accumulations with 6 s accumulation time.

16

30

42
oc,

0
6
cm-2

16

30

42

Fig. 10. Galvanostatic cathodic reduction curves of surface


film/films formed on F&J alloy electrodes, 2 h after the
onset of anodic polarization. Designations are given in Fig. 9.

2231

Surface film formation of iron and its alloys

reduction of 0.16nm of the Fe,O, layer or 0.23nm


of the FeOOH layer.
(iii) Potential arrests B and C [Figs 9 and 10,
potentials -0.75 and -0.7OV (see)] were shorter for
rotating electrodes than for stationary ones (for stationary electrodes the reduction appeared to be
uncompleted for the charge passed). This might indicate that the amount of Fe,O, and/or Fe(OH), was
smaller on the rotating electrodes, or that the rotation facilitated the reduction of these species.
(iv) The charge of the arrest B suggests that at
-0.7OV (see) the amount of Fe,O, increases with
the increasing carbon content, whereas at -0.75V
(see) the amount of Fe,O, and/or Fe(OH), is larger
for decarbonized Fe than for Fe-C alloys.
The method of galvanostatic reduction is widely
used to study the composition of passive films.
However, as mentioned above, the appearance of
two potential arrests might well suggest a two step
reduction process, namely the reduction of an Fe(III)
type oxide or oxyhydroxide to Fe,O, and further
reduction to Fe(OH), or Fe. Thus, in situ Raman
spectroscopy, providing information at a molecular
level, is a particularly useful additional tool for direct
in situ studies of a surface film on iron.
Raman spectral studies ofjlm formation at primary
passive and active-passive transition potentials

Raman spectra were interpreted using band frequencies for Fe,O, reference compounds presented
in Fig. 7; for other oxides and oxyhydroxides of iron,
Table I from Ref. [7] was used. Figure lla and b
show Raman spectra of a decarbonized Fe surface
after 0.4 and 2 h anodic polarization at -0.75 V
(see), respectively. Except for the bands from the
electrolyte at 1065, 1010 and 842cm-, new bands at
672 and ca. 550cm- are clearly visible after a 0.4
and 2 h polarization time. In contrast to decarbonized Fe after 0.4h polarization, spectra for Fe-0.7%
C are characterized by two bands at higher frequencies (Fig. llc), a very weak band at 728 and
a stronger one at 1082 cm- . As discussed earlier for
-0.8V (see), these bands originate from FeCO, or
FeCO, . H,O in the surface film formed on the Fe-C
alloy. The increase of the polarization time to 2 h for
Fe-0.7% C (Fig. lld) resulted in the increase of the
iron carbonate signal at 1082cn- and the appearance of a new band at 668 cm- and a very weak
feature at ca. 545cm-. The comparison of the
spectra of natural Fe,O, with that of decarbonized
Fe (Fig. lla, b) and F&0.7% C (Fig. lld) indicates
that magnetite Fe,O, formed on these materials.
Thus, the bands located at 668 and 672 cm- and
weak ones at 545 and 550cm- are assigned, respectively, to A,, and T& symmetries of magnetite. The
correct in situ identification of magnetite is further
confirmed by an excellent correlation
between
FWHM values for the 668 (FWHM = 42cm-) and
672cm- (FWHM = 35cm-) bands and that of
natural
magnetite. As indicated
earlier, these
FWHM values are in excellent agreement with data
reported for magnetite by Hart et al.[26]. For Fe,O,
the strongest band at ca. 670cm- appears at a frequency similar to that of the strongest band of 6FeOOH[7,29]; thus the observation of a weak band
at ca. 545 cm-[29] as well as the calculation of the

>
.=

670

400

600

600

1000

1200

Raman Shift / cm-

Fig. 11. In situ Raman spectra during anodic polarization


at -0.75 V (see): (a) decarbonized Fe after 0.4h polarization; (b) d&a~b&zed
Fe after 2 h polarizaiion; (c)
Fe + 0.7% C after 0.4h polarization; and (d) Fe + 0.7% C
after 2 h polarization.
FWHM for the A,, mode are of particular importance in order to discriminate between Fe,O, and
&FeOOH.
As indicated earlier, from SCC tests under constant low extension rates and from scanning electron
microscopy studies[ 141 we know that in this particular solution, intergranular SCC is most distinct at a
potential of -0.7 V (see); thus passivation at this
potential was further studied. Figure 12a and b represents in situ Raman spectra taken from the surface
of decarbonized Fe and Fe-0.7% C, respectively,
after 2 h anodic polarization at - 0.7 V. Both Raman
spectra (Fig. 12a and b) are characterized by the
appearance of magnetite peaks at ca. 545 and
668cm-. Mapping of the surface of purified iron
and iron-carbon alloys with the 4pm laser beam
was performed. These studies again revealed strong
heterogeneity of the surface film formed on Fe-C
alloys, in contrast to purified iron. For purified iron,
the Raman spectrum shown in Fig. 12a is representative of the entire surface, ie for ten arbitrarily
chosen areas of the electrode surface, the spectra
were quantitatively very similar, indicating a uniform
layer of Fe,O,, ie no patches of Fe,O, existed. For
iron-carbon, the spectra taken from different areas
of the electrode quantitatively differed from each
other, as presented in Fig 12b and c. Figure 12c
shows the in situ Raman spectrum taken after a 1 h
anodic polarization but from a different electrode
area than the spectrum of Fig. 12b (for 2 h
polarization). For three independent Raman microprobe mapping experiments, results were quantitatively similar to those presented in Fig. 12b and c,
leading to the conclusion that the time of polarization was not a factor and these observations are a

2232

M. ODZIEMKOWSKIet

70

35
E -

0.7

V (SCE)

1065

666

al.

Fe

+ 0.7%

E = -

0.7

V ISCE)

660
1062

1066

a
.=
:
E

L
.=
P
E

0
4

Raman Shift I cm-

Fig. 13. Comparison of the Raman spectrum for


Fe + 0.7% C, obtained in situ after 1 h anodic polarization
at -0.7V (s& Fig. 12~) to the ex situ Raman spectrum of
the specimen treated as above and additionally rinsed with

0
400

water.
600

600

1000

1200

Raman Shift 1 cm-

Fig. 12. In situ Raman spectra during anodic polarization


at -0.7 V (see) of: (a) decarbonized Fe after 2 h polarization; (b) Fe + 0.7% C after 2 h polarization; and (c)
Fe + 0.7% C after 1h polarization. The spectrum (Fig. 12~)
was obtained in the same experiment but from a different
area of the electrode surface than that shown in Fig. 12b.

consequence of strong heterogeneity of the surface


film formed on Fe-C alloys. This heterogeneity of
the surface film results in weaker or stronger
(dependent on the electrode area) Fe,O, signals at
548 and 668cm- accompanied by (or lacking) an
additional band at 1082cn-
from FeCO, (Fig.
12~). These differences in signal intensity suggest the
existence of magnetite patches on the Fe-0.7% C
alloy, and the areas contain both Fe,O, and
FeCO,. Such a heterogeneity of the surface film
formed on Fe alloys was suggested by Melendres et
~I.[301 without experimental evidence. We agree
with those authors[30] that with Raman spectroscopy it is almost impossible to distinguish an inner
and outer thin passive layer. However, these experiments indicate that heterogeneity of surface films can
easily be studied in situ with the Raman microprobe.
After 1 h anodic polarization an Fe-0.7% C electrode was removed from the electrolyte and rinsed
with copious amounts of Mini-Q water. In Fig. 13,
the comparison of the in situ spectrum (from Fig.
12~) with the ex situ spectrum is presented. Exposure
of the Fe-0.7% C electrode to the laboratory atmosphere resulted in the broadening and shift of magnetite bands to higher frequencies. These effects are
accompanied by appearance in the ex situ Raman
spectrum of additional, well-distinguished shoulders
at 665 and 708 cm- 1 and, not seen on this figure, a
very weak band at ca. 4OOcm-. The spectral
changes indicate that the air exposure of Fe,O,,
formed at the active-passive transition potential,
results in its partial transformation to y-Fe,O, (the
665 and particularly
708cm-
lines are the
evidence). The brown colour of the surface suggested

that 6-FeOOH also formed. Note, these ex situ and


in situ results are in agreement with earlier works of
Goetz et al.[31] and Bardwell et al.[32] indicating
that ex situ surface analysis gives valid results only
when a passive film is formed at sufficiently anodic
potentials, while films formed at more negative
potentials (ie in the active-passive transition) are not
stable with respect to ex situ analysis. Furthermore,
the fact that rinsing the electrode in water did not
remove the FeCO, signal might indicate that FeCO,
observed in situ can be incorporated into the structure of magnetite.
Galvanostatic cathodic reduction of surface films
formed after a 2 h polarization at -0.7V resulted in
a diminished Fe,O, signal (at ca. 670cm-I), but, as
visible in Fig. 14, did not lead to its complete disappearance. Potentiostatic cathodic polarization for
about 0.5 h at - 1.05 V (see) led to the almost total
disappearance of the magnetite band (see the righthand side of the figure in Fig. 14), indicating almost
the complete reduction of Fe,O,. Similar experiments, giving this same result, were performed on
samples anodically polarized at -0.65 V. All the
above explain why on the E-Q curves for stationary
electrodes (see Figs 9 and 10) we did not observe the
potential arrest characteristic of reduction of Fe,O,.
Raman spectra obtained during cathodic reduction
are more noisy than those obtained during anodic
polarization due to hydrogen evolution as well as
shorter spectral accumulation times (1 s for cathodic
reduction vs. 6 s and more for anodic polarization).
Following two hours of anodic polarization at
more positive potentials (-600 and - 650mV), no
FeCO, signal was observed but other ferric compounds formed. This is visible in Fig. 15 which presents spectra for decarbonized Fe after a 2 h anodic
polarization at -0.65 and -0.6V (see). The spectrum at -0.65 V (see) exhibited the usual Fe,O,
bands at 548 and 67Ocn- together with a broad
band at 388-4OOcm-. At more positive potential, ie
-0.6V, bands from magnetite were not detectable
even after long polarization times. Since all oxyhy-

2233

Surface film formation of iron and its alloys

2234

M. ODZIEMKOWSKI
et al.
applied potentials. In this figure, the potential
regions of transgranular and intergranular SCC are
also indicated. Figure 16 should only be considered
as a semi-quantitative illustration of the situation.
More quantitative
calculations
were impossible
because none of the electrolyte bands could be used
as an internal intensity standard for surface signal
normalization.
All three electrolyte signals were
observed to be time dependent, probably due to
involvement of both the anions (CO:-, HCO;) in
surface film formation
processes as suggested
earlier[13]. Furthermore, the introduction of a third
anion as an internal intensity standard to allow for
direct comparison of Raman spectral results with
both electrochemical and SCC measurements had to
be avoided.

70

i -

-0.65

E = -0.6

DISCUSSION

800

600

Raman

1000

Shift

1200

cm-

Fig. 15. In situ Raman spectra for decarbonized Fe in the

200-650 (inset) and 400-1200 cm- regions after 2 h anodic


polarization at -0.65 and -0.6 V.
droxides of iron are characterized by appearance of a
band at cu. 390cm-[7], a positive identification of
the surface film formed at - 0.6 V&e) is difficult. On
the basis of cathodic reduction data (Figs 9 and 10)
and the presence of a band at cu. 386 cm- we can
suppose the formation of an unidentified FeOOH
phase at potentials of -0.65 and -0.6 V.
Summary of Raman spectroscopic data

Figure 16 summarizes our Raman results. It illustrates the effect of a 2 h polarization at the applied
potentials on the Raman spectra obtained from
decarbonized Fe (open symbols) and Fe-0.7% C
(solid symbols) in 75C 0.5 M Na,CO, + 0.5 M
NaHCO, electrolyte, by plotting the net integrated
intensity of observed Raman bands against the
900
w
::
2
al

800

700

OSM

Na,CO,

+ 03.4

NaHCOl

Tranrgranular
Intergranular

SCC

see

, 75

i l 670 cm-
670 cm-_
4.
550 cm-
550 cm- -

j 0
j n 1082
A 541

-0.75

Potential

-0.70

V vs.

-0.65

cm-
cm-

-0.6

(SCEI

Fig. 16. Plot of the integrated Raman band intensity for


decarbonized Fe (open symbols) and Fe + 0.7% C (solid
symbols) after 2 h polarization as a function of the potential.

From Fig. 16 it is evident that Raman spectral


data are in good qualitative agreement with both
electrochemical data and SCC tests. Namely, at
active dissolution potentials, corrosion of Fe-C alloy
electrodes is blocked by an iron carbonate film (a
signal at 1082cm- is evidence) resulting in an
observed lower anodic dissolution current (see Figs 1
and 3). At these active dissolution potentials, as
reported by Sutcliffe et al.[l]. transgranular SCC
was observed. At more positive potentials Fe,O,
magnetite forms (bands at ca. 550 and 67Ocm-) and
intergranular SCC takes place. The maximum of the
Fe,O, signal at -0.7V, for both decarbonized Fe
and Fe-0.7% C (in spite of the observed Elm
heterogeneity), is in good correlation with the Fe,O,
potential arrests on cathodic reduction curves
(especially for rotating disc electrodes, Fig. 9). At the
potential -0.7 V (see) the most distinct intergranular
SCC was observed. At potentials more positive than
-0.65V, as indicated both by cathodic reduction
(Fig. 10) and Raman spectroscopy (Fig. 15), magnetite did not form on stationary electrodes.
Despite the significant formation of Fe,O, at the
potential of the distinct intergranular
SCC, this
product does not appear to be critical for the cracking, since it also formed at -0.75 V (see) at which
SCC was slight. It is supposed that the specific electrochemical
conditions
for intergranular
SCC
involve the onset of the formation of the passivating
Fe(III) species, such as ferric oxide Fe,O, or more
probably FeOOH. Under such conditions, grains of
steel can be protected by the passivating film,
whereas grain boundaries can be active due to segregated elements, such as carbon, which hinder the formation of the Fe(II1) passivating film[S, 61. The
ferric oxide and/or oxyhydroxide should be in equlibrium with Fe,O, or Fe(H) species, since at more
noble potentials the passivating film can cover the
entire surface, including grain boundaries.
In this work the formation of Fe(II1) species was
manifested by the arrest A (Figs 9 and 10) which
occurred at the potential of the maximum SCC. The
arrest A indicated that the potential of SCC corresponded to the reversible reaction of Fe(I1) = Fe(W).
At this potential Fe(III) species should coexist with
Fe,O, and Fe(H).

2235

Surface film formation of iron and its alloys

At the potentials of -0.7V (for all materials) and


-0.65V (see) (for Fe-C alloys) the Fe(III) species
were not detected by Raman spectroscopy, obviously
due to a low thickness of this film; judging from the
arrest A for the stationary electrodes, at the potential
of - 0.65 V (see) the FeOOH film was about 0.6 nm
thick on Fe-0.7% C and about 2.1 nm thick on pure
iron (Fig. 10, FeOOH signal detected at ca.
4OOcm-, see Fig. 15 inset).
The smaller amount of Fe(II1) species on Fe-C
alloys than on Fe (Fig. 10) demonstrates that carbon
hinders the passivation of iron. It is suggested that
the intergranular
SCC of plain steel in the
carbonate/bicarbonate
solution results from poorer
passivation of the grain boundary material, due to
its enrichment in carbon and/or other solutes, which
hinder the formation of the Fe(II1) passivating film.

SUMMARY

conditions when Fe(II1) species are in equilibrium


with Fe,O, and/or Fe(II) species.
(7) Raman spectroscopy is clearly a valuable technique for studying the corrosion processes at elevated temperatures. The Raman microprobe has
the unique ability for in situ mapping of surface films
with a horizontal spatial resolution of 4pm; thus it
is a particularly useful tool for studies of localized
corrosion, such as SCC or pitting corrosion.
Acknowledgements-The authors are grateful to Dr D. W.
Shoesmith from Atomic Energy of Canada Ltd., Pinawa,
Manitoba and Professor J. Lipkowski from University of
Guelph, Ontario, for supplying magnetite reference
samples. Funding for this research was provided by the
Natural Sciences and Engineering Research Council of
Canada. Part of the work has been performed within the
Research Project No. 3 12989101 sponsored by the State
Committee for Scientific Research (KBN), Poland.

AND CONCLUSIONS
REFERENCES

In situ and ex situ Raman spectroscopy and electrochemical measurements were performed on pure
iron and its alloys with carbon in OSM Na,CO,
+ 0.5 M NaHCO, at 75C at potentials where the
susceptibility and resistance of plain steel to intergranular XC occur. The main results and conclusions are as follows:
(1) The presence
of carbon
in iron caused a
decrease of anodic current for iron in the active
region, but an increase of the current in the passive

region.
(2) Raman spectroscopy revealed the presence of
corrosion products as follows:
Active region:
_ possibly Fe(OH), on pure iron;
- FeCO, or FeCO, . H,O on Fe-C alloys.
Active-passive
- Fe,O,

transition

region:

on pure iron;
- Fe,O, and FeCO, (or FeCO,. H,O) on Fe-C
alloys.

Passive regions:
- FeOOH.

(3) The presence of carbon promoted the formation of FeCO, (or FeCO,. H,O) in the active and
active-passive regions, and Fe,O, at a potential of
-0.7V (see) at which the susceptibility of intergranular SCC of low carbon steel was highest.
(4) Similar to Raman spectroscopy, galvanostatic
cathodic reduction of corrosion products indicated
that the formation of Fe,O, is more intense on Fe-C
alloys than on Fe at the potential of -0.7V (see),
but less intense than on Fe at the potential of
- 0.75 V (see).
(5) Potential vs. cathodic charging curves of specimens after polarization in the passive and activepassive regions showed a potential arrest at about
-0.7 V (see), suggesting the reduction of FeOOH to
Fe(OHJ . This arrest was shorter for the Fe-C alloys
than for pure iron, demonstrating that the formation
of passivating Fe(II1) species on Fe-C alloys is more
difficult than on pure iron.
(6) Intergranular
SCC was most intense at the
potential of the reduction of Fe(II1) species; this suggests that the SCC is favoured by the electrochemical

1. J. M. Sutcliffe, R. R. Fessler, W. K. Boyd and R. N.


Parkins, Corrosion 28,318 (1972).
2. E. Wendler-Kalsch, Werkstofl und Korrosion 7, 535
(1980).
3. J. C. Rubim and J. Diinnwald, J. electroanal. Chem.
258, 327 (1989).
4. D. S. Mancey, D. W. Shoesmith, J. Lipkowski, A. C.
McBride and J. Noel, J. electrochem. Sot. 140, 637
119931.
5. J. Flis. Corros. Sci. 15. 553 (1975).
6. J. Flis; Corros. Sci. 25; 317 (1985).
7. A. Huaot-Le Golf. J. Flis. N. Boucherit, S. Joiret and J.
Wilinsii, J. elect&hem. sot. 137, 2684 (1990).
8. J. Gui and T. M. Devine, Corros. Sci. 32, 1105 (1991).
9. J. Gui and T. M. Devine, J. electrochem. Sot. 138, 1376
(1991).
10. I. C. G. Thanos, Electrochim. Acta 31,811 (1986).
11. C. A. Melendres, M. Pankuch, Y. S. Li and R. L.
Knight, Eleclrochim. Acta 37,2747 (1986).
12. L. Stolbere Ph.D. Thesis. Universitv of Gueloh (1990).
13. D. H. DaLies and G. i. Burstein; Corrosion &, 416
1----r

(1980).

14. M. Odziemkowski and J. Flis, unpublished results.


15. T. Misawa, Corros. Sci. 13,659 (1973).
16. V. Ashworth and P. J. Boden, Corros. Sci. 10, 709
(1970).
17. R. N. Parkins, C. S. ODell and R. R. Fessler, Corros.
Sci. 24, 343 (1984).
18. Z. Zembura, in Corrosion of Metals and HydrogenRelated Phenomena (Edited by J. Flis), Chap. 5. Elsevier
Science, New York (1991).
19. R. Landsberg and d. Thiele, Electrochim. Acta 11, 1243
(1966).
20. J. B. Bates, M. H. Brooker, A. S. Quist and G. E. Boyd,
J. phys. Gem. 76, 1565 (1972).
21. B. G. Oliver and A. R. Davis, Can. J. Chem. 51, 698
(1973).
22. J. OM. Bockris, M. A. Genshaw, V. Brusic and H.
Wroblowa, Electrochim. Acta 16, 1859 (1971).
23. M. Odziemkowski and J. Flis, Proceedings of the 4th
Comecon

Conference

on Methods

of Corrosion

Protec-

tion, Varna, Bulgaria, 1985, Vol. 4, pp. 16-19.


24. R. G. Herman, C. E. Boadan, A. J. Sommer and D. R.
Simpson, Appl: Spectroscy41,437 (1987).
25. P. R. Graves. C. Johnston and J. J. Campaniello, Mat.
Res. Bull. 23,1651 (1988).
26. T. R. Hart, H. Temkin and S. B. Adams, in Light Scattering in Solids (Edited by M. Balkanski, R. Leite and S.
Portal.
~ I
--~-I,ro. 254.
~~ Flammarion
~~~~~~
~~~ Sciences.I Paris 11976).

2236

M. ODZ~MKOWSIU
er al.

27. C. Johnston, Vibrational Spectroscopy 1,87 (1990).


28. D. E. Irish, 2. Dcng and M. Odzicmkowski, J. Power
Sources, submitted. (7th Int. Meeting on Lithium Batt&s, Boston, May 15-20.1994, p. 47.)
29. A. Hugot-Lc Gaff and C. Pallota, J. electrochem. Sot.
132,2805 (1985).
30. C. A. Melcndrcs, M. Pankuch, Y. S. Li and R. L.
Knight, Electrochim. Acta 37,2747 (1992).

31. R. Go&, D. F. Mitchell, B. MacDougall and M. J.


Graham, J. electrochem. Sot. 134,535 (1985).
32. J. A. Bardwell, G. I. Sproule and M. J. Graham, in
Oxide Films on Metals and Alloys (Edited by B. R.
MacDoupll, R. S. Alwit and T. A. Ramanarayanan), p.
195. The Electrochemistry Society, Pcnnington (1992).

S-ar putea să vă placă și