Você está na página 1de 18

Appl. Phys.

A 38, 1-18 (1985)

Applied
Surfaces
Physics A ""
9 Springer-Verlag 1985

Corrosion of Stainless Steels in Chloride Solution"


An XPS Investigation of Passive Films
P. Brfiesch and K. Mfiller
Brown-Boveri Research Center, CH-5405 Baden, Switzerland
A. Atrens
University of Queensland, St. Lucia, Queensland, 4067 Australia
H. Neff
North Carolina State University, Raleigh, NC 27650, USA
Received 13 February 1985/Accepted 12 June 1985

Abstract. Five commercial steels ranging from the martensitic stainless steel containing
12% chromium to the superferrite containing 29% chromium, 4% molybdenum, and 2%
nickel have been studied by XPS. In addition, a pure iron-chromium alloy containing 7%
chromium has been investigated. Armco iron and pure chromium (99.99%) were included
as references. The formation of the passive films (or corrosion)occurred in deoxygenated
0.1 M NaC1 solution (pH = 5.6), from which the samples were transferred directly to the XPS
chamber under controlled atmosphere (Ar). Concentration profiles (at.-%) of the alloy
constituents in their oxidized and metallic states have been determined separately from the
measured XPS depth profiles. For c > 12% chromium the passive films have the following
structure: there is a depletion of Cr in the inner region, followed by an enrichment
(concentration maximum) in the central region of the films. The height of this maximum
increases, and its position shifts towards the surface with increasing chromium content in
the alloy. The outermost monolayers are rich in water and hydroxyl groups. Various
significant properties of the films change drastically at the critical chromium concentration
of about 12%. This behaviour is rather independent of the other components (Mo, Ni, Cu)
present in the alloys and is discussed in terms of a phase transition in the films which is
controlled by the chromium concentration.

PACS: 68.45.v, 81.60.Bn

Detailed investigations of blade failures in lowpressure steam turbine stages have indicated that the
breakdown of passivity is a critical step in the damage
process [1]. As a first step in understanding passivity
breakdown, it is necessary to gain insight into the
structure and chemical composition of the passive
films which form on relevant technical alloys under
conditions chosen to simulate those found in practice.
The relevant environment within the low-pressure
steam turbine stages is the Wilson zone where first
condensation occurs. The exact composition of the

corrosive medium is somewhat difficult to measure,


but it is certain that we are dealing with a deaerated
chloride solution (temperature ranging between 50 ~
and 100 ~ The blade alloys in most common use at
present are essentially martensitic stainless steels containing 12% chromium, although precipitation-hardened martefisitic stainless steels and duplex stainless
steels have also been used.
This paper reports an XPS investigation of commercial
alloys. They range from 12%-chromium martensitic
stainless steel to the superferrite containing 29% Cr,

P. Briiesch et al.

4% Mo, and 2% Ni. In addition, a pure ironchromium alloy containing 7% chromium has been
investigated. As end members, pure Armco iron and
pure chromium have also been studied. In order to
simulate the conditions found in practice, the surface
films which were studied by XPS have been formed
under open-circuit conditions and not under potentiostatic control and the chloride electrolyte was deaerated. From XPS measurements alternating with Ar-ion
sputtering, depth profiles have been obtained for the
main alloying elements in both their metallic and
oxidized states, as well as for oxygen. From a semiquantitative analysis absolute concentrations in at.-%
have been derived. The distortions of the depth profiles
originating from the finite sampling depth of XPS have
been corrected. By this procedure we obtained "differential concentrations" which approximately reflect
the true concentrations as a function of depth. To our
knowledge only "integral concentrations" of the elements in the passive films have been obtained in the
literature. Such "integral concentrations" comprise
contributions from all levels below the surface; the
presence of steep concentration gradients in the passive films implies that depth profiles derived on the basis
of "integral concentrations" are considerably
distorted.
Additional information about the chemical nature of
the passive films has been obtained by angledependent XPS measurements. In contrast to ion
sputtering, this method is nondestructive and serves to
check the reliability of the results obtained by ion
sputtering. The combination of the two methods
yields, not only a detailed characterization of the
surface films but also a correlation of film properties as
a function of chromium content in the different
alloys.
The paper is organized as follows: Sect. 1 describes the
experimental procedure, including materials, electrochemical measurements, XPS measurements and data
reduction. In Sects. 2 and 3 we present the results of the

electrochemical and XPS measurements, respectively.


The last section is devoted to a discussion of the
experimental results in the light of present knowledge
from related studies of passive films. Special attention
is given to the drastic changes of various film properties observed near the critical chromium concentration of about 13at.-% (about 12wt.-%), and
these surface properties are correlated with the electrochemical proPerties of the alloys.

1. Experimental
1.1. Materials
Table 1 contains designations and structures of the
materials studied.
The compositions of the samples given in Table 2 have
been determined by optical emission spectroscopy,
x-ray fluorescence and wet-chemical analysis. The
alloys were studied in a heat-treated condition consistent with normal use. The surfaces were polished in the
manner usually practiced in metallography. This involves polishing them with ever finer polishing
material, terminating with alumina of less than 0.3 gm
size and a mirror-quality surface finish.

1.2. Electrochemical Setup


The electrochemical setup was used,
(i) to form films for XPS studies under open-circuit
conditions,
(ii) to measure the corresponding open-circuit potentials as functions of time,
(iii) to measure current-potential curves.
The electrochemical setup has been described in detail
in a recent publication [2]. Here we shall briefly give
the essentials (Fig. 1). The electrochemical preparation
chamber is a horizontal glass cylinder directly flanged
to the fast insertion lock of the KRATOS ES 300 XPS
spectrometer. Two ports allow gas and electrolyte

Table 1. Designations and structures of materials studied


No.

Commercial
designation

Our
designation

Structure

1
1A
2

Armco iron
X20CrMoV12
1

Fe
7Cr
12CrlMo

X7CrNiCuNb
15 6 2

15Cr6Ni

4
5
6
7

X3CrNiMol7 13 3
X4CrMnNiMoN2646
29-4-2
99.99Cr

17Crl3Ni
26Cr6Mn
29Cr4Mo
Cr

ferrite
ferrite
martensite
precipitationPH
hardened
austenite
duplex
ferrite
ferrite

C o r r o s i o n o f Stainless Steels in Chloride Solution


Table 2. W t . - % (upper lines) and at.-% (lower lines) o f elements contained in the materials studied
No.

Fe

99.85

1A

93.00
92.52

7.00
7.48

85.52
84.36

11.9
12.60

0.92
0.53

0.66
0.62

0.48
0.48

0.28
0.30

75.28
74.86

14.8
15.81

1.65
0.95

5.2
4.92

1.53
1.34

0.8
0.81

65.98
65.67

17.8
19.03

2.3
1.33

11,5
10,89

62.04
61.42

25.9
27.53

2.22
1.27

42
3.95

64.24
63.58

29.0
30.83

3.6
2.07

2,44
2.30

99.99
99.99

.
.

INSERTION

Cr

Mo

Ni
.

.
.

Cu

Si

0.24
1.09

0.23
0.14

0.06
0.27

0.45
0.89

1.7
1.72

0.09
0.10

0.05
0.03

0.029
0.13

0.55
1.09

5.6
5.63

0.039
0.18

0,07
0,07

0.12
0.13

0.05
0.03

0.005
0.02

0.48
0.95

.
.

.
.

Mn

.
.

.
.

.
.

.
.

.
.

LOCK

.
.

.
.

.
.

.
.

.
.

.
.

ELECTROCHEMICAL
PREPARATION
-

Auxiliary

Nb

.
.

CHAMBER

Por t

Sampleholder

to U H V - P r e p a r a t i o n
C h a m b e r and
Photoelectron
Spectrometer

g Electrode
Electrode Electrolyte
Contact
Platinum
Counter - Electrode
Reference - Electrode
(Sat. C a l o m e l - Electrode)
to Potentiostat

Glass Tube ~ _ ~

IIIIIIIIIIIIIIIIII
I

Fig. 1. Schematic d i a g r a m o f the electrochemical p r e p a r a t i o n c h a m b e r in c o m b i n a t i o n with the fast insertion lock

4
solution to be admitted. The working electrode is
mounted on a sample holder, which is a rod that can be
moved horizontally from the atmosphere through the
preparation chamber into the vacuum system (UHV
preparation, clean-sputtering) and back to the chamber. A glass tube mounted with its opening beneath the
electrode holds a certain volume of the electrolyte
(0.1M NaC1, pH about 5.6, room temperature, made
with BAKER organic-free water). This volume can be
lifted so that the meniscus of the liquid contacts the
electrode when desired. The electrolyte is deaerated in
advance by argon bubbling. A slight argon overpressure in the chamber keeps oxygen out during the
electrochemical measurements and film-forming exposure. Said glass tube also holds the reference electrode
with Luggin capillary. The counterelectrode is a wire
(Pt) wound around the Luggin capillary's thick portion. In view of the very small volume of electrolyte
used, the "deaerated" condition may not meet the
most rigorous criteria but is expected to be consistent
with conditions in the Wilson zone.
The three electrodes were connected to a PAR 123
potentiostat to record dynamic current-potential
curves. Open-circuit potentials (OCP) were monitored
via the same system and recorded as functions of time,
up to periods of 5 to 10ks.
All electrochemical measurements were performed
after surface preparation of the mechanically polished
electrode samples by argon-ion sputtering in U H V
(10rain, 4keV, 12 gA/cm 2, incidence angle 45~ The
dynamic current-potential curves were recorded with a
scan rate of 100mV/s. This unusually high speed was
selected in order not to mask the effect of the pretreatment by growth of secondary passivation films. In the
same way we could avoid the accumulation of perturbing concentrations of metal ions in the electrolyte
originating from the active-dissolution region. Repeat
curves were always recorded after repeated cleansputtering in UHV. By continuous cycling steady-state
current-potential curves could be obtained. All sweeps
were started from the cathodic region. The potentiostat was set close to the hydrogen evolution potential
before the circuit was closed by the surging electrolyte.
The sweep was triggered one or two seconds after
the meniscus touched the electrode.
The results obtained in these potentiodynamic current
measurements were checked in separate experiments.
A much larger volume of the deaerated electrolyte was
used. The electrodes were rotating disks pretreated
mechanically and electrochemically. The curves were
recorded at 100mV/s, as before, and at the much
slower scan rates of 20 and 2 mV/s. Without discussing
these results in further detail at this point, we wish to
point out that, (i) at 100 mV/s the shape of all curves
(Fe, Cr, and six alloys) is essentially the same as

P. Brfiesch etal.
reported here for the sputtered samples, (ii) at slower
scan rates, essential new features do not arise in the
forward scans (there are cases of considerable hysteresis in the reverse scans following transpassive dissolution which are not relevant to the present work), and
(iii) most conspicuously, the typical features of passivation curves, viz., a region of high currents (active
dissolution) followed, beyond a Flade potential, by a
region of drastically lower currents, are not evident in
the curves recorded in chloride solutions; sections
where the characteristic is negative are either totally
absent or quite insignificant. Hence all discussion
based on the results of Fig. 9 will be unaffected by the
unconventional procedure adopted, by necessity, in
the electrochemical cell of Fig. 1.
Prior to the XPS measurements the eight samples were
exposed to the electrolyte for a period of 15h. The
conditions were exactly the same as described above,
when measuring open-circuit potentials; the electrodes
were truly at open circuit to simulate practical conditions. After the exposure time the electrolyte was
separated from the electrode and the latter was
carefully rinsed with deaerated BAKER water while
keeping the electrochemical cell under argon atmosphere. The electrode with the surface film was then
transferred directly into the XPS spectrometer by
using the fast insertion lock. Residual water remaining
on the electrode, and the argon introduced from the
electrochemical cell, were pumped off after transfer in
the UHV system.
1.3. X P S Measurements and Data Reduction
XPS measurements were carried out in a KRATOS ES
300 Electron Spectrometer in the FAT mode of
operation using MgK~I,2 excitation (1253.6 eV). The
base pressure was a few times 10-9tort. Binding
energies of electrons were determined using a gold
standard; the overall resolution was about 1.3eV.
Samples having a surface area of 5 16mm 2 and a
thickness of 2ram were used. Depth profiles were
obtained by sputtering with argon ions (4keV,
12 jxA/cm 2, angle of incidence 45~ The sputter rate
depends on the composition and will, in general,
change with depth. As will be shown below, the passive
films on the alloys to a large extent consist of Cr203.
We therefore determined the sputter rate of Cr20 3 by
using a film of known thickness prepared by reactive
sputtering in a Balzers rf/dc machine. The thickness of
the film has been measured by an Alpha-Step Tencon
instrument. The sputter rate was about 30A/min; a
similar sputter rate has been obtained for Fe203.
The absolute concentrations ei(z) of atoms i at depth z
from the surface have been evaluated by means of
argon-ion sputtering alternating with XPS measure-

Corrosion of Stainless Steels in Chloride Solution

merits. By definition

c,(z) = n~(~)
Zjnj(z)'

Cr20 ~

Cr

(1)

where ni(z) is the number density (number of atoms per


unit volume) of atoms of type i at a depth z and the sum
extends over all types of atoms in the sample. It is
shown in the Appendix that under certain simplifying
approximations ni(z) is given by

20s

60s

h i ( z ) : RiXz) - 2,, dg~s(z)

aT'

(2)
120s

where

Ris(z ) = n~176

(3)

Here I~ is the intensity of photoelectrons originating


from atoms of type i in state s of the pure element i, and
I~(z) is the corresponding intensity from the sample
after sputtering off a layer of thickness z. 2,~ is a mean
escape depth (spatial average) and n o is the number
density of atoms i in the pure element i; n~176
where z ~ is the number of atoms i per unit volume vo.
Choosing vo = 100 A we have n~ = 8.549, nor = 8.372,
n~
and n~
For oxygen we put
n~
which closely corresponds to the number
density of oxygen ions in Cr203 or Fe20 3. The
intensities I~ for the metals have been determined from
the observed intensities of the alloys after sufficiently
long argon-ion sputtering, until saturation values of
the intensities were attained, and by normalizing these
intensities to 100% of the corresponding metal. In this
connection it should be mentioned that a pure Fe-Cr
alloy does not undergo preferential sputtering when
sputtered with Ar ions [3].
Assuming a mean sputter rate f such that z = gt where t
is the sputter timel (2) can also be written in the form

n,(t) = Ri,(t ) - ei,Ri~(t),

(4)

where Ri~(t)=n~176
and fi~=Xijf. The escape
depths for Ni2p, Fe2p, Cr2p, Ols, and Mo3d are
estimated to be 13, 15, 16, 17, and 19/k, respectively
[4, 5]. For the practical calculations we have assumed
a mean escape depth 2-=15A for all 2'~. In our
experiments ~ - 0 . 5 A / s giving i , ~ 3 0 s . The second
term in (2) and (4) corrects for the deformation of the
depth profile associated with the finite sampling depth
[6] (2-effect). In XPS 95% of the total photoelectron
peak intensity originates from a layer of depth
6=32-sinO, where O is the take-off angle of the
electrons; in our case O = 90 ~ and fi = 45 A. Neglecting
the second term one obtains "integral concentrations",
as mentioned in the Introduction. At t = 0 the slope
/~(0) in (4) usually is not well defined. For this reason
and because of unavoidable contaminations present at

l"~ 300s
~'q~
670

675

680

600 s
685

Eki n (eV)

Fig. 2. Evolution of the Cr2p3/2 XPS line of sample 2 as a


function of sputtering time

the surface, the depth profiles shown in Figs. 10 to 14


are uncertain for t < 1 0 s or for z<5,a,.
Concentration profiles of Fe, Cr, Mo, Ni, and oxygen
have been determined by observing the intensities of
the Fe2p, Cr2p, Mo3d, Ni2p, and Ols lines. Figure 2
for instance shows the evolution of the Cr2p3 /2 line of
sample 2 as a function of sputtering time. Only Cr is
observed at t = 0 and t = 2 0 s ; the structure then
gradually changes, and after 600s the spectrum is
dominated by metallic chromium. Another example is
given in Fig. 3 which shows the spectra of the Mo3dand Ni2p lines of sample 4. These spectra will be
discussed in Sect. 3. The evaluation of the intensities
involves background subtraction, [7] deconvolution
of the composite structures into their metallic and
oxidic components, and measurements of the areas of
the individual components. Figure4 for instance
shows the deconvoluted spectra of the Fe2p3/2 ,
Cr2p3/2, and O ls peaks of sample 6. In view of possible
sputter-induced effects (see below), the spectra recorded after sputtering have been decomposed in a
more qualitative manner. The Fe2pa/2 structure, for
example, was decomposed only into the metallic
component Fe" and a single oxidic component
Fe ox= Fe 2 + + Fe a +.
In order to obtain some qualitative information about
depth profiles without argon-ion sputtering we have
measured the XPS intensities of the Fe2p, Cr2p, and
O ls lines as functions of the electron emission angle O.
Figure 5 for instance shows the angular variation of
the Fe2pa/2 line of sample 6. The significance of these
3+

"

P. Briiesch et al.
a

Mo 6+

Mo

Fe 2 p 3 / 2

f . . . . uo~ ud'f . . . . . . .
0s
540

545

550

20s

1015

t020

1025
Eki n (eV)

1030

675

680

Ni~p

Os

720

725
F: kin ( EV )

Fig. 4. Deconvolution of the Fe2p3/2, Cr2p3/2, and O ls XPS lines

of sample 6 before sputtering


20s

Fe s* Fe a+

60s

Fe

180s

595

400

405

Ekin(eV)

90*

Fig. 3a and b. Evolution of(a) the M03d, and (b) the Ni2p XPS

lines as a function of sputtering time

60*
45*

spectra will be discussed in Sects. 3 and 4. At this point


we use the observed angle-dependent intensities I~,(O)
in order to check the reliability of depth profiles ci(z)
obtained by argon-ion sputtering. The procedure
involves calculating the I~(0) from the concentrations
c~(z) as input data, and comparing the I~,(O)c,~ with
Is(O)obs- Since in (1) the total number density
N = Ejnj(z) is nearly independent ofz (as can be verified
numerically) we have

I~(0) ~ gK~s(O ) o~q(z) e-~/xi~ "sinOdz"

(5)

50"
t5"

54O

545

550

Ekin(eV)

Fig. 5. Angular variation of the Fe2p3/2 XPS line of sample 6


before sputtering

Forming the ratio

Ri(O)~

= I~

I~

(6)

+ I.~(0) '

where I~
is the intensity observed for the total
oxide (Cr 3+, Fe 2+ + F e 3+) and I.~(0) is the intensity
observed for the metal (Cr, Fe) we obtain
c~

e -~z/,z, sinO d z

Ri(O)~,z~ = o

c~(z)e-~:"si"~
0

(7)

where q(z) = c~
+ cm(z). It should be mentioned that
Ri(O)obs is independent of sputter-induced effects
because the intensities in (6) have been measured
before sputtering; on the other hand, Ri(O)c,lc is
affected by sputtering via the concentrations c~(z).
Figure 6 shows a comparison between the Ri(O)obs and
Ri(O)~alc- The agreement is satisfactory. It should be
mentioned, however, that less satisfactory results have
been obtained especially for the iron profiles in the case
of very thin films. The deviations resulting in these
cases are related to effects such as preferential reduc-

Corrosion of Stainless Steels in Chloride Solution

i'ox
R (O)/

7
I

i~

-1oo

4.O

~ |

-zoo ~

0.9
0.8
O.i

R (o)

t.0

!
-

-400

'

'

5ooi-.,/
~-

"%

-~ooF
I,

0.9-

2000

0.7
0

I
15

I
30

I
45

I
60

75

I
9000

Fig. 6. Comparison of experimental (x, o) and calculated (--)


R(O) values for the Fe2p3n and Cr2pa/2 XPS lines &samples 1,

2, and 6 (see text)


tion of iron by argon-ion sputtering, nonuniform film
thickness and/or sputter-induced ion mixing [8]. All
these effects will be especially impOrtant in very thin
films. That some reduction of iron oxide by argon-ion
sputtering must be expected has been established by
Frankenthal [3], who also showed that for Cr203 the
reduction is significantly less than for Fe203. The
influence of sputtering on the composition of the films
is further discussed in Sect. 4.1.

2. Electrochemistry
The surface films studied in the present work by XPS
were produced at open circuit. For this reason opencircuit potentials (OCP) of the samples were measured
under identical conditions. This was done in order to
simulate the conditions existing in the Wilson zone and
also in order to obtain the data required for comparison of our present results with those obtained by
other workers at films produced under potentiostatic
control.
The OCP-time curves show spontaneous passivation
for the chromium-containing samples. The values of
potential and the trends in the potential-time curves
provide a measure for nobility (stability against corrosive action) of a given alloy.
Potentiodynamic curves were recorded in order to
have a frame of reference (active/passive region) for the
open-circuit potentials. The absolute values of the
currents measured can be taken as a measure of
reactivity (electrochemical dissolution). It should be
pointed out that the double-layer charging currents in
our systems are of the order of 5 gA.

4000
TIME

0.8-

29 Cr

4Mo

'

<~
I

Cr

~A

' 112 Cr IMo


I ~ I ~5 c, ONi
l " I~7 c~ 15Ni
/ 5 126 C r 6Mn

6000

8000

/ s

Fig. 7. Open-circuit potentials as functions of time for samples 1


to 7 in 0.1 M NaC1 solution (deaerated with Ar, see: saturated
calomel electrode)

The curves in Fig. 7 illustrate the long-time behaviour


of the OCP. Alloys 2 to 6 show the expected tendency
of increasingly noble character. In the logarithmic
plots (Fig. 8) one can see that alloys 3 to 6 occupy a
relatively narrow band of potentials over the first
1000 s, the values being significantly more noble than
those of alloys 1A, 2 and of pure iron. Alloy 2
spontaneously passivates after a delay of about 400 s.
Only pure iron and alloy 1A end up in the active
region.
The extrapolated limiting values attained at long times
by the alloys with 12 to 20% Cr are consistent with the
values reported by Uhlig [9]. The dependence of these
potentials on chromium concentration will be discussed below (Fig. 16a).
The reproducibility of these data was such that trends
were clearly preserved within a single series of runs
covering all alloys when the series was repeated at a
later date (with the same procedure), while absolute
differences in potential at intermediate times were as
high as (though not typically) 100mV in individual
cases. Such lack of precise definition probably is typical
for almost oxygen-free systems where the potential is
relatively readily upset by minor effects. Such irregularities are leveled out at long exposure times.
Figure 9 shows anodic potentiodynamic currentpotential curves, (a) on fresh surfaces upon first contact
with the electrolyte, immediately after argon-ion sputtering in UHV, and (b) in the tenth anodic sweep under
cyclic potential control. The potential-sweep limits
were selected to coincide with the start of hydrogen
evolution on the cathodic side, and with the start of
transpassive dissolution on the anodic side (at currents
of about 1 mA).
The reasons for selecting a high potential-sweep rate
(100mV/s) have been discussed. The steady surface
state eventually attained under open circuit certainly is

P. Brfiesch et al.
t

0
-100

-2oo

Fe

;A T c ~

__
~
4
5

12
t5
17
26
29

Cr tMo
Cr 6 Ni
Cr,3Ni
Cr 6 M n
Cr 4Mo

++
J
+(~++~/""
+ . . - ~

iI
i

++~"/" J ~ " 7 ~

.........

: -300

>
E

... -400
o -500

-600
-700

t0 t

t02
TIME

~/PA/r

t03

/ s

~t500

/.." o

[/

//'-; 500 /..-

..Tf ~

-500

500

t000

t500

U / mY vs SCE

-500

i/pA/cm2:1500
-t000

500

t04

Fig. 8. Open-circuit potentials (vs sce) as functions of the


logarithm of time for samples 1 to 7 in 0.1 M NaC1
solution (deaerated with At)

/(ig)

=~

7 and 9b still reveal parallels in electrochemical


behaviour of the alloys under the two sets of conditions. During cycling again, alloys are more noble
when the chromium content is higher. Alloy 2 displays
the largest change in the voltammogram between the
first and tenth sweep. The OCP condition of the alloys
corresponds to the potential region in the sweep just to
the left of the coordinate origin in the plots of
Fig. 9.
The charge passed during first sweeps (Fig. 9a)
amounts to values of 1-2mC/cm 2, and thus corresponds to something like three monolayers of di- or
trivalent compound formed on top of the surface
sputtered clean in UHV.
The open-circuit potentialsattained after 4 ks according to Fig. 7 and the currents recorded, according to
Fig. 9, at U = 0 V (sce) have been plotted in Fig. 16a
against the chromium content of the alloys. The trends
seen in these plots will be discussed below in conjunction with the XPS results of Figs. 16b and c.

, i
3. X P S Results

7""
-tO

.~,~..~

.............
:56o
o

""J

500

1ooo

t5oo

U / mV vs SCE

-500

Fig. 9a and b. Potentiodynamic current-potential (vs sce) curves


for samples 1 to 7 in 0.1 M NaC1 solution (deaerated with At),
recorded at a scan rate of 100 mV/s. (a)Anodic scan using fresh
surface (sputtered in UHV, transferred directly to the cell (Fig. 1),
electrolyte contact established at -1000 to -1100mV. (b)The
same after 10 anodic-cathodic cycles

not the same as the state of the surface to which Fig. 9b


refers, since in the latter case, alternating oxidation and
reduction of the surface are induced externally by
current flow (with a current of 100 IxA, some 0.3 atomic
layers can be dissolved anodically per second). Figures

Figure 10 shows the depth profiles of chromium for the


different alloys, as measured after exposure to the
electrolyte under OCP conditions during 15 h. In the
alloy substrates (at sputter times t > 600 s) the concentrations are relatively close to the bulk values listed
in Table 2. For alloy 1A (7% Cr) the profile is
qualitatively different compared with the other alloys:
only a very small accumulation of Cr, of the order of
1%, is observed near z ~ 20 A. The situation is quite
different for the other alloys containing 12% or more
chromium: In the interface region between the alloy
and the passive film there is a depletion of chromium
which is followed by a pronounced maximum within
the passive film. For alloy 2 (12.6at.-% Cr) the
maximum of Oct(Z) is at a depth of z ~ 30 A, while for
the other alloys containing more than 12.6 at;-% Cr

Corrosion of Stainless Steels in Chloride Solution

9
/k----

Ccr(t)
( at %)

50

.....
.o

._+';.'-:.

'_,;./2
. . . .

. . . . . . . . .

'10

I
0

40
I

20

I
80

I
t20

t80
I

40

60

90

Sputter time

I
(s) 5 0 0

Approx, depth

(~) t50

//~
600
/,,

I ,,~

300

Fig. 10. Depth profiles of the total chromium concentration


(Cr ~ + C r ' ) of samples 1A to 6. @ 7.0 Cr; @ 12.6 Cr, 0.6 Mo,
0.6Ni; @ 15.8Cr, 0.9Mo, 4.9Ni; @ 19Cr, 1.3Mo, 10.9Ni;
@27.5Cr, 1.3Mo, 4Ni, 5.6Mn; @30.8Cr, 2Mo, 2.3Ni
(Table 2)

the maximum is located between 10 and 15 A below the


surface. With increasing chromium content in the bulk,
the maximum becomes considerably higher but the
increments become smaller at the higher chromium
concentrations; this indicates that a saturation value
of the peak concentration near about 35-40 at.-% Cr
will be reached (normalized with respect to all components, metal plus oxygen). Note that the chromium
concentration for pure Cr203is 40 at.-%. The decrease
of the chromium concentration in the outermost layers
is due to the presence of hydroxides and water in this
region of the passive films, as well as to some dissolution of chromium into the electrolyte. That appreciable amounts of hydroxides and water are contained in
the films can be deduced from the shape of the Ols
XPS line (Fig. 4c). Angle-dependent measurements
actually reveal that hydroxides and water are accumulated in the outermost layers of the films (Fig. 15).
In addition to these species, the surface will hold some
unavoidable oxygen-containing contamination also
contributing to the decrease in chromium concentration very close to the surface.
On the basis of deconvolution of the Fe2p and Cr2p
XPS lines it is possible to derive individual concentration profiles for the metals (Fe", Cr") as well as
for the oxides (Fe ~ Cr~ These profiles are shown in
Figs. 11 to 13. In these figures Cr ~ stands for Cr 3+
while Fe ~ represents the sum of Fe 2+ and Fe 3+. In
pure Armco iron (sample 1, Fig. 11a) the broad concentration profile of Fe ~ reflects a relatively thick
oxide film of the order of 50 A. In this case the oxide
film contains about 90% Fe 3+ and 10% Fe z+ (Fig.
16c). For pure FezO3 the concentration of iron would
be 40at.-%, the 10% Fe 2+ would raise this figure to
40.8 at.-%. These values agree well with the mean iron

concentration in the film of sample 1 (Fig. 11a). From


Figs. 11 to 13 it is seen that with increasing chromium
concentration in the alloy the iron and chromium
peaks become narrower and also shift towards the
surface. The coexistence of iron and chromium in the
passive films indicates that an iron-chromium oxide
exists at least in the inner part of the passive films, while
the outer part is enriched in hydroxides and water as
mentioned above. It should be emphasized, however,
that no sharp interface between the outer "hydroxidewater region" and the inner "oxide region" exists; this
can be seen from the angle-dependent XPS measurements (Fig. 15).
Figures 11 to 13 seem to indicate that metallic iron and
chromium coexist with the iron-chromium oxides.
Such an overlap between oxide and metal profiles
would be expected if the oxide contained metallic
clusters. Nothing is known about the stability of such
metallic clusters in very thin oxide films. On the other
Ci

/k---~

(t) i a

L-

(at % )

..../'

70

,i

60
50
40 ~

3oy / / /

"\\\~

/ Fe~

2O 10
I

40

80

4~.0

20

40

60

0
Ci

I -..we"

t 8 0 Sputter time (s) 5 0 0


I

90

~ql---

]<600L-i/

Approx. depth(~) 150

I ~_

300
//--~

(t)

(at % ) - -

._..._.~/.'---~

-gO6O
5O

/"(

4O

Fgx

3O
2O

'

Or..

-"

l0

. . . .

crT-

t----/r
0
j

40
I

80
I

t~-0
I

~80 Sputter time (s) 5 0 0


I
l

//

600
I],,_

O
20
40
60
90 Approx. depth (~) t50 500
Fig. 1la and b. Depth profiles of (a) sample 1 (pure Armco iron)
and (b) sample 2 (12.6 Cr, 0.5 Mo, 0.6 Ni). Fe m, Cr'~: metallic iron
and chromium, Fe ~ = Fe z + + Fe 3 +, Cr ~ = Cr 3 +

10

P. Brfiesch et al.
lk-'--

cl (t)

ci (t) ]
(el % }F
/

(at % ) ' a

I/---"

,,,,-.---

70
60

60

50

50

,'I~

,~ F=m

40

40
30

20

%.

t0

0
I

cox

zog^i
Fl

~...~j.--.

Cr

l , ~ " ~ ' r - ~ "Is-" ~


4O
80
'120
I
I
l

2O

4.0

60

. . . . .

"r
t 8 0 Sputter

300

c; It}

80

i/-'-

I.

t20

%%0

t80 Spulter time

20

40

60

90

(e) 3(~0
I
Approx. depth(n) t 5 0

I:

300

let % )

70
60

70

//
60

,,

40

(at % )

I~. w-.f.

--'I"-=- / / ~
time (s) ~ 0 0
600
1
//
I

90 Approx. depth(.~) '~50

c, ~
I\

IoI-/~ ".~-"
VJ'~Y-,_3"-.,--..,_.,

//---

//----

cl (t)

30

.Fe~
J

t~....~.-==~-~'~

~'F."

50

Fern

40
40

3~ / . / y c , ' "
.

3 0

... .... ~."-- .......

--

. . . .

,i

t...

c~

20

rm

..-----~--

I J L-:~_
0
40

80

J_
420

20

40

60

t "'-.+-..,

I
I
t80 Sputter time (s) 300
I

/;v; :->_..
I ' " I " *---~-.1~

90 Approx. depth(~) 450

....

L5,,4 ",J

600
l/

~_

300

Fig. 12. Depth profiles of(a) sample 3 (15.8Cr, 0.9Mo, 0,4 Ni)
and (b) sample 4 (19 Cr, 1.3 Mo, 10.9Ni). Fe'~, Cr": metallic iron
and chromium, Fe~
2+ +Fe 3+, Cr~
3+

t 8 0 Sputter time (=1 3 0 0


i
l

2o

40

60

90 Approx. deplh(l) 450

80

420

600
ii

i=

]00

c; (t)
(It

'&

90

.f"

80

70

60

50

hand, it should be kept in mind that, although we have


derived differential depth profiles, our depth resolution
still is limited due to intrinsic effects such as inhomogeneous film thickness and sputter-induced mixing of
the atoms and ions. The limitations of depth resolution
due to these intrinsic effects will be particularly severe
for the very thin films where the metal-oxide overlap is
found to be most pronounced. In reality the metallic
phase is expected to be better separated from the oxide
phase than the figures indicate; a complete separation
is unlikely.
The depth profile of oxygen is shown in Fig. 14a. For
pure iron (sample 1) the oxygen profile decreases very
slowly with increasing distance from the surface.
Already for the alloy containing 12.6 at.-% Cr (sample 2) the oxygen concentration Co(Z) decreases more
rapidly, and this tendency continues to hold with
increasing chromium content. This observation indicates that the mean film thickness decreases considerably with increasing chromium concentration in the
bulk. It is also worth noting that the slope of Co(Z ) for

40

/F--

40
30

r,~ cSx

20
10
1

0
I

/'"~

40
l
20

"- ~

80
I
40

.. 4.---,~, . . . . . .

t20
I
60

4---

480 $putter time (e)


I
90 Approx. depth (~i

Fig. 13a~z. Depth profiles of(a) sample 5 (27.5Cr, 1.3 Mo, 4Ni,
5.6Mn), (b)sample 6 (30.8Cr, 2Mo, 2.3Ni), and (c) pure
chromium. Fem, Cr": metallic iron and chromium.
Feox=Fe z+ +Fe 3+, Crox=Cr 3+

samples 1 and 2 changes considerably near z ~ 10A.


This indicates excess oxygen very close to the surface
due to the accumulation of hydroxides and water as
well as other oxygen-containing contaminations. Finally, it should be noted that even after long sputtering
times an appreciable oxygen level of the order of
5-8 at.-% is observed in the alloys and also in pure
chromium. This might be surprising at first sight;
however, it is a well-known fact that the complete

Corrosion of Stainless Steels in Chloride Solution

11

Co It) /

tt---

,.,;3 a

40

1
20

BO

I
40

t20

t 8 0 Spulier

I
60

I
90

time (t) 3 0 0
I
Approx. depth(~) ~50

;;__

600

II

300
#---

c.o"(t )

(at %1

t.4

-/,--

t.2

t.0

./

0.8

.,,,m

f t . ~ |

0.6
0.4

,' ! - -

0.2
/ I

I I I
4080

I
t20

20

40

60

c/

0
c~p(t)
(at %

I
I
1 8 0 Spulter time Is) 300
I
I
9 0 Approx. depth (~) t 5 0

/,L_

600
I;

I_

300
#---

44

tZ
t0
8
6

/...----~>- . . . . . . . . . . . . . . . . . . .

l/_ _._'tit__

2
'I

0
I

I
40
I
20

I
80
I
40

I
420
I
60

I
I
t 8 0 Sputter time (e) 3 0 0
I
I
9 0 Approx. depthll) t 5 0

H--600
l/ I .~
300

Fig. 14a-c. Depth profiles of (a) oxygen, (b) metallic molybdenum, and (c) metallic nickel of selected samples @ pure

Armco iron; @ 12.6Cr, 0.5Me, 0.6Ni; @ 15.8Cr, 0.9Me,


4.9Ni; |
1.3Mo, 10.9Ni; @27.5Cr, 1.3Me, 4Ni,
5.6 Mn; @__30.8Cr, 2 Me, 2.3 Ni; @ pure chromium
removal of oxygen is difficult or even impossible using
argon-ion sputtering of metals and alloys with high
oxygen affinity [10].
Figure 3a shows the Mo3d spectra of sample 4 at
different sputtering times. Before sputtering, at t = 0 s,
the structure of the Mo3d doublet is complex and
different from the structures observed after sputtering
at 20, 60, and 180s. Before sputtering the "Mo3d3/2

structure" is broader and slightly shifted towards


higher binding energies compared to the spectra after
sputtering; in addition a shoulder is observed at the
high-binding-energy side of this structure. These features imply that the metallic Mo3d3/2 line overlaps
with lines of oxidic 3ds/z states indicating the existence
of small amounts of molybdenum oxides and/or
oxyhydroxides [5]. The fact that after a sputtering time
of only 20 s the spectrum already shows the features of
metallic molybdenum confirms that molybdenum
oxides are rapidly reduced by argon-ion bombardment
[11-13].
In contrast to molybdenum, essentially no nickel in the
oxidic state has been observed in the passive films.
Even in sample 4 containing as much as 13% Ni, only
metallic nickel has been observed before and after
sputtering (Fig. 3b): Figures 14b and c show the depth
profiles of metallic Me and Ni. There is a strong
depletion in the region of the passive films. Concerning
the coexistence of metallic Me and Ni with mixed ironchromium oxides the same remarks apply as made
above for metallic Cr and Fe. For samples 4 and 6 we
observe a pronounced accumulation of metallic nickel
in the region of the interface between the passive film
and the alloy.
While no nickel oxides and only relatively small
amounts of molybdenum oxides and/or oxyhydroxides have been observed in the passive films, the
situation is different for the manganese contained in
sample 5 (Table 2). In this alloy manganese is observed
in both the metallic and oxidic state (MnO, MnO2, and
possibly oxyhydroxides), even after 180 s of argon-ion
sputtering.
The surface sensitivity of XPS which is a consequence
of the limited mean free p a t h of the electrons in the
solid, can be increased by lowering the electron
emission angle O. Figure 15 shows the angular dependence of the ratio I1(0)/I2(0), where 11(0) is the
intensity due to O H - and H 2 0 and I2(O) is the
intensity of 0 2 - bound to metal ions in the oxide.
These intensities have been obtained by deconvolution
of the Ols lines as shown in Fig. 4c, and the angular
dependence of these intensities is given by (5). Figure 15
reflects the fact that O H - and H 2 0 are enriched in the
outer part of the passive films, the enrichment being
most pronounced in the high-alloyed sample 6 containing 31 at.-% chromium.
In order to investigate chloride penetration into the
passive films we have studied two extreme cases,
namely pure iron (sample 1) and the alloy containing
31 at.-% chromium (sample 6). In pure iron we do find
small amounts of chlorine incorporated in the oxide
film, even after argon-ion sputtering of 180 and 300 s;
since at these sputter times no sodium has been
detected we can exclude the possibility that the

12

P. Brtiesch et al.

I1

12

0ts

2.0
1.8
t.6

t.4
t.2
t.0
0.8
0.6
0.4
0.2

t5

30

45

60

75

90 O

Fig. 15. Angular dependence o f l l ( O ) of samples 1, 3, and 6. I1:


intensity of O H - and H20,12: intensity of 02 - bound to metal

ions in the oxide


chloride originates from residual traces of NaC1 present at the surface of the sample. In sample 6, however,
no detectable incorporation of chlorides into the
passive filmhas been observed.

4. Discussion

4.1. The Composition of the Passive Films


In the following we summarize the most important
results concerning the composition of the passive films
and compare them with the results from the
literature.
Angle-dependent XPS measurements have shown that
hydroxides and water are accumulated in the outermost layers of the passive films and that the central
region of the films consists mainly of iron-chromium
oxides. The transition from the outer "hydroxidewater" region to the inner "oxide region" is continuous. These results agree with those obtained by
Olefjord [4, 5]; Okamoto [14] and Bockris et al. [15,
16] have also established that hydroxides and water
are essential components of the surface films.
The passive films are enriched in chromium. In the
region of the critical chromium concentration of about
13%, the chromium concentration in the films increases nonlinearly with the chromium content in the
bulk. For alloys containing more than 20% chromium
the central region of the passive films consists mainly of
Cr~O3. Our depth profiles show pronounced maxima
in the chromium concentration which are located close
to the surface. These maxima are more pronounced
than those found in the literature [17-20] and are

qualitatively similar to the profiles observed for airoxidized iron-chromium alloys [21]. Only very few
indications are found in the literature for a minimum in
the chromium concentration at the interface between
the passive film and the underlying alloy [18]. A
possible explanation for this observation is discussed
at the end of Sect. 4.3.
Since in Figs. 11-14 we show absolute concentrations
of the elements, one might be tempted to derive the
stoichiometry of the passive films as a function of depth
below the surface. This is, in principle, possible, but it
should be kept in mind that the observed stoichiometry may deviate considerably from the true stoichiometry due to sputter-induced effects, in particular to
selective sputtering of oxygen. We have estimated the
stoichiometry in the central region of the surface films
(i.e., at the maximum of the Fe ~ or Cr ~ profiles in
Figs. 11-13) and have found a composition M20~
(M2= Fe~176
where x~1.8-2.3 for % < 2 0 %
and x ~ 3 for % > 2 0 % . These results agree qualitatively with Frankenthal's sputter experiments of Fe20 3
as discussed at the end of Sect. 1.3. To find the true
composition may require the use of significantly milder
sputter conditions. On the other hand, the ease of
sputter-induced reduction of the films with less than
20% Cr is a further indication for their insufficient
chemical stability.
Nickel and copper have been observed only in the
metallic state; these elements are strongly depleted in
the passive films. On the other hand, we observe an
accumulation of metallic nickel in the interface region
between the passive film and the alloy; this accumulation is most pronounced in the nickel-rich alloy 4
containing 13% nickel. An accumulation of nickel in
the interface region has also been observed by
Olefjord [22].
As expected, metallic molybdenum is strongly depleted
in the passive films. Molybdenum in oxidic states is
present before sputtering, similar to the results of
Olefjord and Brox [5]. The distribution of these species
in the passive films cannot be deduced since argon-ion
sputtering leads to instantaneous reduction [11-13].
An enrichment of molybdenum was found in the
passive film by Goetz and Landolt [23] for an alloy
containing 11% Mo; this concentration is considerably higher than the molybdenum concentrations in
our samples, so that a direct comparison with these
results is not possible. In contrast to Olefjord [22] but
in agreement with Lumsden and Staehle [24], we do
not find an enrichment of metallic molybdenum below
the passive film. This might be due to the fact that it is
difficult to detect such accumulations in narrow
regions by a method involving sputtering; in addition,
in comparing different results it should be borne in
mind that the preferential dissolution of alloying

Corrosion of Stainless Steels in Chloride Solution

components depends to some extent on the electrolyte


used for the electrochemical experiments.
In sample 5 (duplex steel) containing 5.6% Mn, the
passive film contains manganese in the oxidized state.
Since the Mn2p3/2 binding energies of MnO, Mn203,
and Mn30 4 are only slightly different [25], it is not
possible to establish the actual oxidation state; the
spectra are compatible with Mn / and Mn 3 but not
with Mn 4+.
Small amounts of chloride originating from the
electrolyte could be observed in the surface film on
pure iron. Under our electrochemical conditions
(OCP) this film is not a passive film and allows chloride
to penetrate to a considerable extent. This is not the
case for the chromium-rich sample 6 containing 31%
Cr, and probably neither for the other alloys containing more than 12% Cr; the absence of chlorides from
the passive films has also been reported by Elfstr6m
[19] and Szklarska-Smialowska [26].
4.2. Passive Films
and Critical Chromium Concentration
It is a well-known fact that below a critical chromium
concentration Coritof about 12% the corrosion rate of
iron-chromium steels increases drastically with decreasing chromium content [27]. Alloys with c < corlt
are nonnoble, that is, they normally show strong
corrosion, while alloys with c > co,it are noble, that is,
they form a passive film which strongly suppresses
corrosion. This critical limit is observed, not only in
pure iron-chromium alloys but also in the commercial
steels investigated in this study. Figure 16a shows the
open-circuit potentials (OCP) at 4 ks from Fig. 7 as
well as the currents at U = 0 V (sce) from Fig. 9b. One
can see important changes of both the OCP and the
corrosion currents at c ~ corit. The results for OCP are
consistent with those of Uhlig [9] obtained under
similar conditions (deaerated NaC1 solution).
The questions now arise to what extent the properties
of the passive films change at corot and how these
properties can be correlated with the electrochemical
behaviour.
Figure 16b illustrates the chromium accumulation,
dccr, as defined in Fig. t0 as well as the mean thickness
d of the surface films as functions of the chromium
concentration in the alloys. The film thicknesses d have
been estimated from the depth profiles (Figs. 11 to 13
and 14a). Both quantities change drastically at Cc~it.
According to Fig. 16b corrosion-resistant alloys
(c > co,it) are characterized by thin passive films which
are strongly enriched in chromium. On the other hand,
nonnoble steels (c < C,~t) are characterized by relatively
thick oxide or hydroxide films, the chromium enrichment of which is only small or nil. As mentioned in

13
(, a

+100

t.5 "-"...............

I
I

ul
0

-I00

>"
E

cn

-300

.%.

~0.5
-5oo

Z/
0
0

- - t
8

i
24

t6

i -70 0
52

Ccr ( at% )lnalloy


kd b A )

ACc,

(at %1- ~14

60.

12

5(> 9

40

I0

30

,o I j ,
0

~)

to

o,o

30

20

Ccr ( at% )inagoy

c,C':~176C

--.I---:

t.O

Fe 3+

Fe2++ Fe 3+

0.9

0.8
0.4

0.7

0,2
0,2

--,,4"
~

~'"
0

Ccrit.

I ~
10

20

30

C c r ( a t % )lnalloy

Fig. 16. (a) Open-circuit potentials from Fig. 7 (points at 4ks;


right-hand scale) and potentiodynamic currents (ipa) from Figs.
9a (e) and 9b (0) (points at 0 V, left-hand scale) as functions of the
chromium content in the alloys. The vertical arrows indicate the
tendency towards 10ks; the lower part of the OCP curve is
interpolated according to Uhlig [9]. (b) Mean thickness d of the
surface layers and chromium enrichment Accr (Fig. 10) as
functions of the chromium content in the alloys. | according to
Elfstr6m [19]. (c)Normalizcd concentration of Cr 3+
[Cr~
~ + Fe~ (-- from depth profiles; . . . . . from integral
concentrations) and normalized concentration of Fe 3+
[Fe 3 +/(Fe z + + Fe 3 +)] as functions of the chromium concentration in the alloys

14
Sect. 4.1, the latter films do not constitute passive films,
since they were formed under OCP conditions not
leading to a point in the passive region at long
exposure times. This statement applies in particular to
the films formed on pure iron and alloy 1A whose film
thicknesses are not well reproducible and probably
also depend on the electrolyte [28]. The film thicknesses of samples 1 and 1A plotted in Fig. 16b are
estimated mean thicknesses with an uncertainty of at
least __20 A. We do not believe, however, that these
films were formed only after the end of exposure of the
samples to the electrolyte, because all manipulations
were carried out under controlled conditions (transfer
of the sample in argon atmosphere). We cannot rule
out that precipitation of corrosion products from the
solution has contributed to the films found on samples
1 and 1A and is responsible for some of the irreproducibility in film thickness. On the other hand, it is known
that the passive films (c>%it) formed under OCP
conditions after a long exposure time (about 15 h) are
quite stable [29], and for this reason their film
thicknesses are expected to be well reproducible and
unaffected by precipitation effects. At this point it
should be mentioned that a qualitatively similar
behaviour of film thickness as a function of chromium
concentration has also been found for pure ironchromium steels by means of ellipsometric
methods [30].
Figure 16c shows the normalized cation concentration
ofCr 3+, that is the ratio R = Cr~
~ + Fe~ as well
as the normalized concentration of Fe 3+, that is
P = Fe 3+/(Fe 2+ + Fe 3+), in the passive films, as functions of chromium concentration c in the alloys. The
values of P have been determined by deconvolution of
the Fe2p3/2 spectra before argon i o n sputtering
(Fig. 4a). Alloy 2 with c = 12.6 at.-% Cr is just at %~t
9 and R-~ 40%. For the true passive films with c > corotwe
obtain R > 50% in agreement with Fischmeister et al.
[31] and with Asami et al. [28]. A qualitatively similar
behaviour for P(c) has been found by Asami using
1N H2SO 4 as the electrolyte [32]; his absolute values
are, however, significantly different from our values,
possibly because of the different electrolyte or the
different method used for data evaluation. The fact that
alloy 2 is just at the limit between active and passive
behaviour is also reflected in Fig. 7 (compare the
corresponding remarks made in Sect. 2).
The general behaviour of the normalized concentration of Fe a +, that is of P(c), can be explained if one
assumes that Fe 3+ ions lost from the film by dissolution in the electrolyte are partially replaced by Cr 3
supplied from the alloy by diffusion. Note the relatively large deviation of the data point for sample 5
containing 27.5 at.-% Cr and 5.6 at.-% Mn (Table 2).
This deviation would be expected under the assump-

P. Brfieschet al.
tion that the substitution of Fe by Mn occurs mainly
in the divalent state. According to Pourbaix [33] the
system is just at the boundary between Mn 2+ and
Mn 3+.
It is very remarkable that the functions ACc~(C), d(c),
R(c), and P(c) shown in Figs. 16b and c are largely
independent of alloy structure and the presence of the
alloying elements Mo, Ni, and Cu (Table 2). This is
understandable if one remembers that Ni and Cu are
very strongly depleted in the passive films and that Mo
occurs only in very small quantities. The functional
dependences shown in Figs. 16b and c are therefore
almost completely determined by the main elements
iron and chromium. Deviations are expected to occur
only if the alloying elements are present in the oxidic
state in appreciable quantities, as is the case for
manganese in sample 5 (Fig. 16c).

4.3. The Hypothesis of a Phase Transition


in the Passive Film
The properties of the surface films depicted in Figs. 16b
and c all show a drastic change near the critical
chromium concentration c=c~rit~12.5at.-%. This
strongly suggests that the transition from corrosive
behaviour at c < eeritto the passive behaviour at c > co,it
is due to a phase transition in the surface film which is
driven by the chromium concentration c in the alloy, or
more precisely, by the corresponding chromium concentration x(c) in the films.
Transmission-electron-diffraction studies b y McBee
and Kruger [-34] indicate that the anodically formed
surface films are composed of iron-chromium oxides
of the spinel type with the composition
3+
3+
Fe 2 + Fe2-x(c~Crx(c)O,.
If this is the case the phase
transition is expected to occur at a critical concentration x = xcrit. The results obtained by XPS and
SAM show, however, that the composition of the
surface films is more complicated, because the films not
only contain oxides but also hydroxides and water, the
latter species being accumulated in the outermost
monolayers.
Consider now an alloy with chromium concentration c
whose clean surface (such as obtained after argon-ion
sputtering) is exposed to the electrolyte. In the first
monolayers certain reaction products, such as hydroxides will be formed. Further oxidation of the alloy
below the surface film can continue only if the oxygencarrying species (O) can migrate from the electrolyte
through the surface film already formed. In the stationary state the rate of oxidation at the alloy-film interface
is equal to the dissolution rate at the film-electrolyte
interface so that the film thickness and composition
remain constant. The oxygen current is therefore
equivalent to a corresponding current of iron and

Corrosion of Stainless Steels in Chloride Solution


chromium into the electrolyte. Let Do be an effective
diffusion coefficient for oxygen (which includes possible effects of electrical fields across the film). Do will
in general depend on x (and hence on c): Do = Do(x).
From the depth profiles of oxygen (Fig. 14a) it follows
that Do(x) is large for x < x~rlt (slowly falling profiles),
but that Do(x) is small for x >Xcr~t (rapidly falling
profiles). Thus, the phase transition mentioned above
manifests itself also in a strong change of Do(x) near
x = xcrir The nature of this phase transition is not
known; it might be a structural phase transition
(crystalline to amorphous, for example [34]) or a phase
transition in a mixed conductor in which the electronic
and ionic conductivity are strongly coupled and
strongly depend on chromium concentration x in the
film.
In the case of a slowly falling depth profile of oxygen
(X<Xorit) we also expect a flat depth profile for
chromium, but hardly an enrichment of chromium in
the surface film. At x = xorit the oxygen profile already
falls more rapidly (sample 2 in Fig. 14a), and correspondingly we observe a relatively broad maximum in
the chromium profile (sample 2 of Fig. 10). For values
of x considerably larger than xcrit the oxygen depth
profiles fall sharply, and we therefore expect that
pronounced maxima in the chromium concentrations
will occur in narrow regions even closer to the surface
than for x = xc~t; this is indeed borne out by experiments (Fig. 10). We can think in terms of a chemically
driven segregation in which, due to its high oxygen
affinity, chromium is driven from the outermost layers
of the alloy towards the oxygen barrier. This loss of
chromium atoms in the alloy surface layers leads to the
formation of vacancies. In general, only part of these
vacancies will be reoccupied by chromium, others will
be occupied by iron or other available atoms, or
remain empty. Hence a net loss of chromium is
expected which explains the chromium depletion in the
interface region between the passive film and the alloy
(Fig. 10). It must be emphasized, however, that in
addition to this chemically driven segregation there is
another important mechanism which contributes to
the enrichment of chromium in the passive films,
namely the selective dissolution of iron into the
electrolyte as studied by Kolotyrkin et al. [35, 36].

4.4. Implications of Passive Film Structure


on Corrosion Properties
This work has examined the structure and composition of the passive film for a range of alloys in one
particular environment. The alloys were carefully
chosen. They represent the range of resistance to
corrosion currently available in commercial stainless
steels. Corrosion resistance in this particular instance

15
is taken to mean resistance to pitting and crevice
corrosion in aqueous chloride solutions.
It is known that resistance to pitting corrosion is a
complex nonlinear function of both the chromium and
molybdenum contents. For example Horvath and
Uhlig [39], using the pitting potential as a measure of
pitting resistance, showed that the pitting resistance
increased very sharply, for Fe-Cr alloys, at a
chromium content of about 30%. They also showed
that increasing nickel contents in the range 0-60%
increased the pitting potential only very slowly, but
that small molybdenum additions had a very large
beneficial effect. Subsequently, the synergistic effect of
both chromium and molybdenum has been shown
repeatedly [40-44]. For some compositions of
F e - C r - M o - N i alloys (and alloy 5 would be expected
to be with this range), N additions can further increase
the pitting potential [45, 46]. No significant increases
of the pitting potential have been observed by additions of V, Si, Co, A1, Ti, Ag, Ce, La, Nb, Zr, and Cu
[40]. However, it has been often argued that the pitting
potential is in fact a poor predictor of actual corrosion
behaviour and that for this purpose the protection potential is to be preferred [44, 4749]. Unfortunately, there is less data available to record how the protection potential depends in composition. Although
a correlation between the pitting and protection potentials is expected, it has been shown [44] that alloys
with very noble pitting potentials can have quite active
protection potentials. Nevertheless, these data [44]
also indicate the beneficial effect of both chromium
and molybdenum as well as the synergistic effect when
both elements are present together in the alloy.
These electrochemical studies can be summarized as
predicting an increased resistance to pitting corrosion
as the chromium and molybdenum contents are increased. These can be correlated with observations of
corrosion behaviour during exposure tests which show
that alloys with higher chromium and molybdenum
contents are generally able to withstand higher temperatures [50, 51] and higher chloride concentrations
[43, 51-55] before becoming susceptible to pitting or
crevice corrosion. The actual rates of corrosion during
passive corrosion are usually very low and therefore
are very rarely measured. For these reasons the
expression % Cr + 3.3 x % Mo has evolved as a useful
guide to the corrosion resistance of stainless steels.
The stainless steels used in this study have been ranked
by this means, with the expectation that a higher
number reflects a higher corrosion resistance, manifest
by the ability to resist pitting in more concentrated
chloride solutions at higher temperatures.
The solution to which the steels were exposed was
chosen with the expectation that all the stainless steels
would exhibit passive corrosion behaviour with the

16

P. Br/iesch et al.

stainless steel number 2 being on the margins of


showing pitting corrosion [48]. Indeed, the electrochemical results have shown that these expectations
were fulfilled, with all the stainless steels as well as the
pure Cr showing passive corrosion behaviour.
Under the experimental conditions considered in this
study alloy 2 can be considered a transition between
the nonpassive iron and the iron-chromium alloy
containing 7% Cr on the one hand and the fully passive
behaviour exhibited by alloys 3 to 6 on the other hand.
If these alloys 3 to 6 are then said to exhibit stable
passivity, it is clear that the stable passive film is largely
independent of alloy composition. It can be characterized by a four-layer structure consisting in turn of a
hydrated layer containing much oxygen and OH - ions
in contact with the solution, a layer composed of a
mixed iron-chromium oxide, a layer approximating
Cr203 which is in immediate contact with a metallic
layer depleted in chromium and enriched in nickel. The
outer 3 layers are only 1.5 to 2.0 nm in depth and thus
represent only several monolayers: the actual number
ofmonolayers depends on whether oxygen or Cr203 is
taken as the standard of measurement. Within this
general structure there are two additional noteworthy
features: no chloride is incorporated into the passive
film; and metallic chromium and iron coexist within
the oxide layers. This last observation can be interpreted as either metallic clusters within the passive film or
as inherent (probably localized) inhomogeneities in
film thickness.
These similarities in the structure and composition of
the stable passive film for alloys 3 to 6 would be
expected to lead to similar properties. Thus, for
example, the corrosion rate should be governed by
either the diffusion rate through the film or alternatively by leakages through localized spots where the
film is particularly thin. These expectations are consistend with the experimental observations that the corrosion rate, as measured by ipa, is virtually the same for
alloys 3 to 6.

On the other hand, when the corrosion resistance is


considered in terms of resistance to pitting corrosion in
more concentrated chloride solutions at higher temperatures, then an additional consideration comes into
play. This is the consideration of the film stability
under the more aggressive conditions. The differences
in this sort of corrosion resistance must reside in
differing abilities of the different stainless steels to
maintain the stability of their passive film in environments of different aggressivity. As this sort of corrosion
resistance increases with both chromium and molybdenum contents, it appears that the availability of
chromium in the alloy is one of the critical factors in
passivity, and that the role of molybdenum must be in
facilitating chromium availability either by easing
chromium transport to sites where it may be oxidized
to CrzO3 or by decreasing the activity of environmental species which would tend to solvate the
chromium.
Since the film structure and composition as measured
in this study is very similar to that determined in other
studies [-4, 5, 14-24, 26, 56] it appears that a stable
passive film will have similar structure and composition under all conditions of alloy composition and
environmental exposure. This film structure will then
produce comparable corrosion rates. This is documented in Table 3 where the passive current during
polarization studies has been taken as a measure of
the corrosion rate. The values in this table do indeed
show that a wide range of stainless steels under a wide
range of environmental conditions, have passive current densities within a relatively narrow range of values.
However, it should also be noted that there is a
systematic increase in passive current density with a
decrease in pH, a decrease in molybdenum content or
an increase in temperature. These relatively minor
differences in passive current density must then be
accounted for by the rather subtle differences in the
passive film structure and composition. What is
needed is an in-depth study relating passive corrosion

Table 3. Passive current densities for a wide range of stainless steels under a wide range of environmentalconditions
Material

Environment

Passive current
[ga/cm 2]

Scan rate
[mV/s]

Ref.

Fe, 19Cr, 20Ni, 3-7Mo


304
316
Fe, 26Cr, 1Mo
Fe, 16Cr, 14Ni, 2Mo
Fe, 25Cr, 0-3.5Mo
Fe, 17Cr, 0-3Mo
Fe, 19.8Cr
Fe, 28Cr, 4Ni, 2Mo

3.5% NaC1, pH3, 10-90 ~


3% NaC1, pH 1-8, 23 ~
3% NaC1, pH 1-8, 23 ~
3.5% NaC1, pH 7, 40 ~
3.5% NaCI, pH 7, 40 ~
1N HzSO,;, 29.8 ~
0.1N HC1, 29.6 ~
0.5M H2SO4, 25 ~
4N NaC1, pH 2, 80 ~C, [O] < 10 ppb

0.8-3
0.2-2
0.4-4
0.5
0.3
2-8
2-9
10
2

0.3
0.2
0.2
0.2
0.2
0.2
0.2
1
1

57
58
58
59
59
60
61
5
62

Corrosion of Stainless Steels in Chloride Solution


resistance to the structure a n d c o m p o s i t i o n of the
passive film a n d t r a n s p o r t t h r o u g h the film or at
localized t h i n spots.

Appendix
Depth profiles have been obtained using the following procedure.
For photoelectrons of atoms i in state s the intensity is given by

17
surements. We also thank Drs. R. K6tz, S. Stucki, P. Pfluger, and
H. J. Wiesmann for interesting discussions and Dr. H. R. Zeller
for critically reading the manuscript. In addition we thank Prof. I.
Olet]ord, Prof. H. Fischmeister, and Dr. U. Roll for valuable
discussions during Conference meetings and Prof. H. J. Grabke
for providing us with the Fe7Cr-alloy.

References

cc

Iis(z) = K~ S n~(z') e-<~'-z)/z's(~')dz''


z

Here K~s= I0%f~s is a constant which depends on the x-ray flux


I0, the photoelectron cross-section ais (assumed to be independent of depth z') and an instrumental factor f~s. This factor is
proportional to the instrumental transmission T(E~s), which in
general depends on the kinetic energy E~s. For the KRATOS
ES 300 instrument it is said that in the FAT mode T ( E J varies
with E~ ~ (37). It has been shown, however, that it is a better
approximation to take T(E~)/T(Ej~,) as unity for all values of the
kinetic energy (38). Hence we put f ~ = f independent of the
kinetic energy, ni(z' ) is the number density of atoms i at depth z'
and Adz' ) = 2i~(z')sin O, where2~(z') is the escape depth and O is
the take-offangle of the electrons. The integral extends from the
actual sputter depth z to infinity. The origin of both z and z' is
chosen to be at the original surface before sputtering. In our
experiments O = 90~ hence Ai~(z') = 2~(z'). In general, 2a depends
on z' if the composition changes; since this function is not known
we replace 2i~(z')by a spatially averaged escape depth 2i~ which
we set equal to the mean escape depth 2o for the pure element i.
We then obtain
co

Ii~(z) = Ki~ ~ ni(z')e-(~'-~)/~'~dz',


z

and
oo

O
zl~,i~

0 -Ii~Kiln ~0 ~ e o

0
0
d z -_ _ K~n~2~s

for the intensity of the homogeneous element i. Defining the


quantity

o Idz)
Ri~(z)=ni
we obtain
1

~o

Rdz) = T6e ~/~'~~ ni(z')e -~ m~dz'.


"~is

Forming dR~(z)/dz = R'~(z) and using


d ~

dz

! f(z')dz'= --f(z)

gives

hi(Z) = Ri~(z)-- 7.1~R'i~(z),


where ~ ~ 2~

Acknowledgements. The authors are indebted to Mrs. M. Sch/ifer


for the chemical analysis and to Mr. R. Baumann for the surface
polishing of the samples. Special thanks are given to Mr. W.
Foditsch, Mr. R. Weder, and Mr. P. Untern/ihrer for excellent
technical assistance during the electrochemical and XPS mea-

1. A. Atrens, H. Meyer, G. Faber, K. Schneider: In Corrosion in


Power Generating Equipment, ed. by M.O. Speidel and
A. Atrens (Plemtm, New York 1984) p. 299
2. H. Neff, W. Foditsch, R. K6tz: J. Electron Spectrosc. Rel.
Phen. 33, 171 (1984)
3. R.P. Frankenthal, D.E. Thompson: In Passivity of Metals,
ed. by R.P. Frankenthal and J. Kruger (ECS, Princeton
1978) p. 262
4. I. OlefJord, B.O. Elfstr6m: Corrosion 38, 46 (1982)
5. I. Olefjord, B. Brox: In Passivity of Metals and Semiconductors, ed. by M: Froment (Elsevier, Amsterdam 1983)
p. 561
6. S. Hofmann: In Oberfldchenanalytik in der Metallkunde, ed.
by H. L Grabke (Deutsche Gesellschaft ffir Metallkunde,
Oberursel 1983) p. 229
7. D.A. Shirley: Phys. Rev. B5, 4709 (1972)
8. S. Hofmann, J.M. Sanz: Fresenius Z. Anal. Chem. 314, 215
(1983)
9. H.H. Uhlig, N.E. Carr, P.H. Schneider: Trans. Electrochem.
Soc. 79, 111 (1941)
10. A. Spitzer: Private communication (1984)
11. R. Holm, S. Storp: Appl. Phys. 12, 101 (1977)
12. S. Storp: Microchim. Acta (to be published)
13. K.S. Kim, W.E. Baitinger, J.W. Amy, N. Winograd: J.
Electron Spectrosc. Rel. Phen. 5, 351 (1974)
14. G. Okamoto: Corros. Sci. 13, 471 (1973)
15. O.J. Murphy, J. O'M. Bockris, T.E. Pou, D.L. Cooke, G.
Sparrow: J. Electrochem. Soc. 129, 2149 (1982)
16. R.W. Revie, B.G. Baker, J.O'M. Bockris: J. Electrochem. Soc.
122, 1460 (1975)
17. W.R. Cieslak, DJ. Duquette: In Passivity of Metals and
Semiconductors, ed. by M. Froment (Elsevier, Amsterdam
1983) p. 405
18. M. da Cunha Belo, B. Randot, F. Pons, J. Le H6ricy, J.P.
Langeron: J. Electrochem. Soc. 124, 1317 (1977)
19. B.O. Elfstr6m: Mater. Sci. Eng. 42, 173 (1980)
20. A.E. Yaniv, J.B. Lumsden, R.W. Staehle: J. Electrochem. Soc.
124, 490 (1977)
21. R.P. Frankenthal, D.L. Malm: J. Electrochem. Soc. 123, 186
(1976)
22. I. Oletjord: Mater. Sci. Eng. 42, 161 (1980)
23. R. Goetz, D. Landolt: Electrochim. Acta 29, 667 (1984)
24. J.B. Lumsden, R.W. Staehle: ASTM Spec. Tech. Publ. 596, 39
(1976)
25. Perkin-Elmer Corporation: Handbook of X-ray Photoelectron Spectroscopy (Eden Prairie, Minnesota 1977)p.74
26. Z. Szklarska-Smialowska, H. Vielhaus, M. Janik-Czachor:
Corros. Sci. 16, 649 (1976)
27. H.H. Uhlig, R.B. Meats: In Corrosion Handbook, ed. by
H. H. Uhlig (Wiley, New York 1948) p. 20
28. K. Asami, K. Hashimoto, S. Shimodaira: Corros. Sci. 18, 151
(1978)
29. H. Fischmeister, U. Roll: Private communication (1984)
30. K. Sugimoto, S. Matsuda: Mater. Sci. Eng. 42, 181 (1980)

18
31. H. Fischmeister, S. Hofmann, U. Roll: Fresenius Z. Anal.
Chem. (to be published)
32. K. Asami, K. Hashimoto, S. Shimodaira: Corros. Sci. 16, 386
(1976)
33. M. Pourbaix: Atlas of Electrochemical Equilibria (NACE,
Houston 1974) p. 290
34. C.L. McBee, J. Kruger: Electrochim. Acta 17, 1337 (1972)
35. Ya.M. Kolotyrkin, V.M. Knyazheva: In Passivity of Metals,
ed. by R.P. Frankenthal and J. Kruger (ECS, Princeton
1978) p. 678
36. Ya.M. Kolotyrkin: Electrochim. Acta 25, 89 (1980)
37. A. Barrie: In Handbook of X-ray and UV Photoelectron
Spectroscopy, ed. by D. Briggs (Heyden, London 1977)
Chap. 2
38. Y.M. Cross, J.E. Castle: J. Electron Spectrosc. Rel. Phen. 22,
53 (1981)
39. J. Horvath, H.H. Uhlig: J. Electrochem. Soc. 115, 791 (1968)
40. N. Pessel, F.C. Hull, C. Liu: Westinghouse Research and
Development Report 478 (1969)
41. A.P. Bond: J. Electrochem. Soc. 20, 603 (1973)
42. E.A. Lizlovs, A.P. Bond: J. Electrochem. Soc. 122, 719 (1975)
43. M.A. Streicher: Corrosion 30, 77 (1974)
44. J. Hochmann, A. Desestret, P. Jolly, R. Mayoud: NACE-5,
p. 956
45. W.H. Richardson, P. Guha: Br. Corros. J. 14, 167 (1979)

P. Briiesch et al.
46. J.E. Truman, M.J. Coleman, K.R. Pirt: Br. Corros. J. 12, 236
(1977)
47. A. Atrens: Proc. 8th Intern. Congr. on Metallic Corrosion
(Dechema, Frankfurt 1981) p. 206
48. A. Atrens: Corrosion 39, 483 (1983)
49. A. Broli, H. Holtan, M. Midjo: Br. Corros. J. 8, 173 (1973)
50. R. Brigham, E.W. Tozer: Corrosion 30, 161 (1974)
51. A. Garner: Corrosion 37, 178 (1981)
52. B.C. Syrett, D.D. Macdonald, H. Shih: Corrosion 36, 130
(1980)
53. H.P. Hack: Mater. Perf. (June 1983) p. 24
54. L.S. Redmerski, J.J. Eckenrod, C.W. Kovach: Mater. Perf.
(June 1983) p. 31
55. T.J. Lennox, M.H. Peterson, C.W. Billow: Mater. Perf. (June
1983) p. 49
56. J.R. Cahoon, R. Bandy: Corrosion 38, 299 (1982)
57. P~J. Brigham: Corrosion 28, 177 (1972)
58. V. Scotto, G. Ventura, E. Traverso: Corrosion Sci. 19, 237
(1979)
59. D. Sinigaglia, G. Rondelli, G. Taccani, B. Vicentini, G.
Dallaspezia, L. Galelli, B. Bazzoni: Werkstoffe und Korrosion 31, 851 (1980)
60. E.A. Lizlovs, A.P. Bond: J. Electrochem. Soc. 118, 21 (1971)
61. E.A. Lizlovs, A.P. Bond: J. Electrochem. Soc. 116, 574 (1969)
62. A. Atrens: Z. WerkstofRech. 15, 309 (1984)

Você também pode gostar