Sunteți pe pagina 1din 19

Electrochimica Acta 45 (2000) 25152533

Corrosion protection by organic coatings: electrochemical


mechanism and novel methods of investigation
G. Grundmeier
a,
* , W. Schmidt
b,1
, M. Stratmann
b
a
Thyssen Krupp Stahl AG, Forschung, Zentrales Qualitats- und Prufwesen, Eberhardstr. 12, 44120 Dortmund, Germany
b
Uni6ersitat Erlangen-Nurnberg, Lehrstuhl fur Korrosion und Oberachentechnik, Martensstrasse 7, 91058 Erlangen, Germany
Papers received in Newcastle, 20 December 1999
Abstract
The application of electrochemical techniques for corrosion studies of organic coatings on reactive metals is
considered from the analytical and mechanistic standpoint. Techniques such as electrochemical impedance spec-
troscopy (EIS), scanning vibrating electrode and scanning Kelvinprobe (SKP) are powerful tools to better understand
the fundamental processes of corrosion at defects and underneath coatings. In the rst part of this paper these three
techniques are discussed in more detail as they present a very complementary approach to understand the ensemble
of coating degradation, processes in defects and corrosion underneath coatings, respectively. The second part of this
paper focuses on the two important mechanisms of cathodic delamination and liform corrosion (FFC). Since both
forms of corrosion are characterised by certain electrochemical reactions underneath coatings and are localised in
nature, the discussion focuses on the application of the SKP to give new insights in these corrosion phenomena.
2000 Elsevier Science Ltd. All rights reserved.
Keywords: Corrosion protection; Organic coatings; Electrochemical impedance spectroscopy; Kelvinprobe
www.elsevier.nl/locate/electacta
1. Principal aspects of corrosion protection by organic
coatings
Protecting reactive metals by covering their surface
with organic coatings is a smart way to take advantage
of mechanical properties of metals such as steel or
aluminium while preventing them from corrosion and
introducing one or multiple requested surface proper-
ties in one step. These properties might be colour, wear
resistance, formability, noise reduction and electronic
insulation.
Organic coatings consist of
a binder or vehicle,
pigments, and
additives such as dryers, hardening agents, stabilising
agents, surface activating compounds, and dispersion
agents.
The vehicle, usually polymers of relatively low molec-
ular weight, determines the basic physical and chemical
properties of the coating. However, additional pigments
can signicantly inuence the properties of the coating.
Their role is to provide colour, act as a barrier for
corrosive species and to serve as actively corrosion
inhibiting species. Nowadays steel is usually covered by
metallic coatings, which are in most cases zinc or zinc
alloys. An inorganic conversion layer is deposited on
top of the metallic coating to generate a corrosion
resistant interface and to provide a link to the organic
primer. For automotive applications usually a cathodic
electrocoat is used as a corrosion resistant organic
primer. The topcoat gives the system its appearance
and acts as a barrier between the corrosive medium and
the inner layers. Both the primer and the topcoat
contain corrosion active pigments.
* Corresponding author.
E-mail address: grundmeier@tks.thyssen.com (G. Grund-
meier)
1
Present address: Max-Planck-Institut fu r Eisenforschung
GmbH, Max-Planck-Strasse 1, 40237 Du sseldorf, Germany.
0013-4686/00/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved.
PII: S0013- 4686( 00) 00348- 0
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2516
It is obvious from the combination of these multiple
layers that the mechanism of corrosion for technical
systems especially under conditions of natural weather-
ing is a very complex one. However, it is the aim to
understand the fundamental processes of coatings
degradation as a tool to optimise new coating systems
with improved corrosion behaviour. Concerning the
details of organic coating technology the reader is
referred to the numerous reviews and monographs [1
5]. An overview of the different corrosion phenomena
on polymer-coated metals can be found in Leidheisers
review [6]. Some of the methods of investigating poly-
mer coatings on metals are outlined in Refs. [79].
While more and more functionality is introduced into
coatings, still the aspect of corrosion protection espe-
cially for steel and aluminium are of great interest in
research and development. This is due to the fact that
hazardous compounds such as Cr(VI) which nowa-
days guarantee excellent corrosion protection prop-
erties have to be replaced with alternative
environmentally friendly compounds;
the introduction of new light metals such as magne-
sium with specic corrosion behaviour require spe-
cially adopted coatings;
use of water based or 100% solvent free coatings will
replace solvent based coatings;
application of new curing technologies such as UV
or electron beam (EB) curing might lead to new
specic reactions at the metal/polymer interface; and
the trend to sell pre-coated steel sheet to the automo-
tive industry to omit secondary corrosion protection
procedures and to reduce the costs caused by expen-
sive paint shops raises new demands for thin organic
coatings.
The corrosion protection properties of an organic
coating most often result less from the barrier proper-
ties but more from the maintenance of adhesion to the
substrate under chemical and electrochemical condi-
tions imposed by the environment. It could be shown
that for typical organic coatings used for corrosion
protection the diffusion rate of H
2
O and O
2
far exceeds
the diffusion limited value for oxygen reduction [10
12]. However, ion solubility within the coating is typi-
cally very small due to the low dielectric constant of
common coatings [13].The following roles of organic
coatings provide corrosion protection:
Barrier for ions leading to an extended diffusive
double layer.
Adhesion of the coating.
Blocking of ionic paths between local anodes and
cathodes along the metal/polymer interface.
Vehicle of corrosion active pigments and inhibitors
which are released in the case of coating damage.
Fig. 1 schematically shows the inuence of an or-
ganic coating on the electrode potential and the size of
the double layer in front of the metal surface.
Fast metal dissolution due to the large electric eld
(10
7
V/cm) leads to a negative electrode potential in the
defect area in contrast to the polymer-coated area
where metal dissolution is strongly inhibited and a
diffuse double layer is observed.
Corrosion of a polymer-coated metal requires the
penetration of water molecules, ions and oxygen to the
interface. Even visibly non-damaged lms show an
Fig. 1. Schematic of the shape of the double layer for electrolyte covered metal (top) and polymer-coated metal (bottom).
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2517
onset of corrosion after a certain time. Thus the diffu-
sion of these species probably follows small pores or
pathways within the polymer that facilitate this trans-
port. However, the most severe corrosion processes
start at local defects within the coating. These might be
introduced during production e.g. at cut edges, or are
generated during the lifetime of the coated material, for
example caused by stone chipping or scratching. Such
defects lead to the contact between the corrosive elec-
trolyte and the bare metal. Subsequently, coated sys-
tems corrode normally along the polymer/metal
interface. The corrosion mechanisms are controlled by
the metal substrate or its metallic coating. For the
reactive metals steel, zinc-coated steel and aluminium
the perdominant mechanisms are the cathodic delami-
nation and liform corrosion (FFC), respectively. The
existence of these two different corrosion mechanisms
and the fact that iron can show both mechanisms
depending on the composition of the environment,
verify that the mechanism of corrosion depends on the
metallic substrate and the composition of its passive
layer and corrosion products as well as on transport
phenomena. Moreover, the fact that corrosion starts at
the metal surface and that progress of corrosion hap-
pens at the interface between the metal and the polymer
makes electrochemical methods the most suitable tools
to understand the corrosion of organic coatings on
reactive metals.
In the following rst part of this contribution those
electrochemical methods which are most interesting and
popular for polymer-coated metals are described in
short. The second part presents the current knowledge
of corrosion mechanisms for polymer-coated steel and
aluminium.
2. Electrochemical methods of investigation
High ohmic resistance of organic coatings impedes
the use of dc-type electrochemical measurements, since
during a potentiostatic polarisation of a polymer-
coated metal the potential drop across the polymeric
layer is orders of magnitude larger than the potential
drop across the metal/polymer interface. This has led to
the development of a number of advanced electrochem-
ical techniques [14]. Electrochemical impedance spec-
troscopy (EIS) and the scanning Kelvinprobe (SKP)
have found widespread application for electrochemical
studies of the degradation of polymer-coated metals.
New electrochemical scanning techniques such as the
scanning vibrating electrode technique (SVET) and lo-
calised impedance spectroscopy (LEIS) recently have
found application for assigning integral corrosion be-
haviour to local defects in the coating and for measur-
ing corrosion processes within the defect as a function
of the adjacent coating. While the galvanostatic pulse
method, relaxation voltammetry, EIS and electrochemi-
cal noise method are valuable techniques for studying
dielectric properties, onset of defect formation and
processes of coating degradation, recent scanning refer-
ence electrodes techniques such as SVET and SKP
enable localisation of defects (SVET and SKP) with
high spatial resolution of a few tens of micrometers.
The SKP moreover allows study of localised corrosion
processes underneath insulating coatings [14].
In this contribution the electrochemical techniques
are ordered according to the following principal pro-
cesses relevant for the corrosion of polymer-coated
metals:
swelling and ion incorporation into the coating and
formation of intrinsic defects such as pin holes;
onset of localised corrosion in defects; and
spreading of corrosion underneath the coating which
can be caused by on anodic or cathodic reactions.
2.1. Measuring the deterioration of organic coatings by
electrochemical impedance spectroscopy
Practically speaking EIS provides a measure of the
resistance of the organic coating to aqueous and ionic
transport. The technique is based on the measurement
of the current response on small sinusoidal perturba-
tions of the electrode potential as a function of the
frequency of the perturbation [15,16]. In the following
we will focus on the possibility of detecting the onset of
corrosion and the progress of corrosion underneath the
polymer coating.
Different models have been proposed to analyse EIS
measurements obtained for model systems and techni-
cal coatings. These models have been applied for indus-
trial screening of organic coatings on bare and
phosphated steel in 0.5 M NaCl solution [17]. Fig. 2
shows a typical impedance spectrum of a quasi-ideal
coating which does not exhibit any indication of corro-
sion attack even after exposure time of up to half a
year.
The Bode plot shows a pure capacitative behaviour
over a wide frequency range and the polarisation resis-
tance at low frequencies is in the order of 10
11
ohm
cm
2
. This is the simple case of a homogeneous 3-D lm
which is well illustrated in Fig. 3.
The impedance of the coated electrode is described
by a parallel combination of the capacitance, C
L
; and
the resistance, R
L
, of the layer
Z
L
( j) =
R
L
1+jR
L
C
L
(1)
The coating capacitance is given as
C
c
=mm
0
A
d
(2)
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2518
Fig. 2. Impedance of a quasi-perfect coating on steel. Experi-
mental data () and optimum t ( ) using the EC in Fig. 3 [18].
partially damaged coating which thereby caused cor-
rosive undermining of the coating.
The rst case has been studied recently by van West-
ing et al. [18]. Measurements were carried out for
pigmented and unpigmented high resistance epoxy
coatings on cold rolled steel. The authors tted the
impedance of the coating using a constant phase ele-
ment (CPE). Fig. 4 shows the plots of the CPE parame-
ters Y
0
and n during a long term exposure in 3% NaCl
solution. The parameter Y
0
reects the total polarisabil-
ity and n is a measure of the interaction between the
polarisable groups [19].
The rst steep change in both parameters is due to
the swelling and ion incorporation of the polymer
coating. The CPE-constant Y
0
increases to a stationary
value due to saturation of the coating but shows a bend
around 400 h. The corresponding features were ob-
served for the CPE-power n. An observation of the
metal surface with a stereo microscope after 400 h
showed the existence of small corrosion spots under-
neath the coating. The same type of behaviour, preser-
vation of high impedance with changing
CPE-parameters, was found also for pigmented
coatings.
Impedance spectra for defect containing polymer-
coated metals exposed to corrosive electrolytes can be
tted according to the electronic circuit (EC) shown in
Fig. 5. R
po
has been called the pore resistance in [20],
which it is considered to be due to the formation of
ionically conducting paths in the polymer. R
de
is the
polarisation resistance at the metal surface in contact
with the ionically conducting paths and C
dl
is the
corresponding capacitance. The resulting impedance
plot is schematically shown in Fig. 6.
Scully and Hensley calculated the theoretical effect of
various percentages of defect area on the impedance
[21]. The impedance parameters were obtained from an
iterative procedure, whereby synthetic spectra were
compared with experimental spectra and updated to a
good correspondence. In this procedure, pore resistance
(here: R
d
), charge transfer resistance (here: R
t
) and
double layer capacitance were held constant. The defect
area was varied. The resulting synthetic EIS data as
shown in Fig. 7 were strikingly similar to that observed
during the degradation of actual coatings. The high (fh)
and low () frequency breakpoints increase linearly
with the defect area with f
l
saturating above 0.001%
defect area.
In the case of spreading corrosion underneath the
coating, a transmission line model as schematically
shown in Fig. 8 represents the system as described by
Kendig et al. [22].
Resistance values R
s(i)
characterise the ohmic resis-
tance between the defect and the respective active part
underneath the coating. Only if the ohmic resistance
R
s(i)
is sufciently low the effective capacitance C
dl
where m is relative dielectric constant, m
0
is dielectric
constant in vacuum, A is coating area, and d is coating
thickness.
Thus the capacitance measurement by EIS can
provide information on the water uptake, since this
incorporation of polar molecules leads to an increase in
the dielectric constant of the coating.
Three cases might now be distinguished:
corrosion underneath the coating without any defect
in the coating itself,
partially damaged coating with cracks reaching the
metal surface, and
Fig. 3. Impedance model of a defect free organic coating on a
metal surface in contact with an electrolyte.
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2519
Fig. 4. CPE-parameters Y
0
(left axis) and n (right axis) vs. time during a long time exposure in 3% NaCl solution (Reprinted with
permission from [18]. 1994 Elsevier Science).
which is equal to the sum of the capacitance within the
pore and those underneath the coating, does indeed
scale with the disbonded area as assumed by many
authors.
Theoretically, it is obvious that the determination of
charge-transfer resistance and double layer capacitance
is possible only if the time constants of the interfacial
reaction and the polymer lm are clearly separated such
that the impedance of the coating is smaller or equal to
the impedance of the interfacial reaction. This might be
the case for very thin polymers and highly inhibited
interfaces. If the impedance of the interfacial reaction is
signicantly smaller than the impedance of the polymer
lm, it is unlikely that the impedance of the interface
can be deduced from the overall impedance of the
polymer-coated metal. To circumvent the problem of
the high coating impedance between the reference elec-
trode and the metal polymer interface, Feser and Strat-
mann developed a set-up where the reference electrode
is directly placed at the interface [11].
This experimental set up is schematically shown in
Fig. 9 where one reference electrode is located at the
metal/polymer interface and a second one is positioned
in front of the coating showing a signicant difference
due to the large impedance of the coating. Both spectra
exhibit just one time constant, however, solely spectrum
B depends strongly on a change of the oxygen activity
in the electrolyte as shown in Fig. 10. The resistance at
low frequencies increases with decreasing oxygen activ-
ity [11]. Thus spectrum A is determined mainly by the
properties of the polymer, whereas spectrum B shows
the impedance of the interface, which depends on the
activity of oxygen. While this set-up is appropriate for
fundamental studies, it is too sophisticated to be ap-
plied to the evaluation of technical systems. Moreover,
there is still no spatial resolution in this measurement.
However, there is a need to measure the corrosion
properties of polymer-coated metals locally due to the
fact that mostly only a very small part of the coating is
damaged, which dominates the electrochemical be-
haviour of the whole exposed surface. In this case
scanning reference electrodes should help to assign
defects and to separate anodic and cathodic areas of
galvanic couples.
Fig. 5. Impedance model of a defect containing organic coat-
ing on a metal surface in contact with an electrolyte.
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2520
Fig. 6. Schematic impedance spectrum of a defect containing
polymer on a metal.
Fig. 8. Equivalent circuit schematics for an organic coating in
case of a coating with a disbond starting from a scratch [22].
2.2. Detection of local defects by means of scanning
electrode techniques
Defects in organic coatings may originate from the
production process (e.g. cut edges, forming induced
defects) or from mechanical impact (e.g. stone chip-
ping). However, coatings which may possess ionic con-
ductive pathways or ionic residuals are located at the
interface so that corrosion starts at sites of the coating
which are not visibly damaged. Since EIS is a priori an
integral method it can detect the existence of such
defects but it can not assign them to certain points on
the examined surface. The idea of a local measurement
is to examine areas separately which differ in their
activity on one sample.
Fig. 7. Bode magnitude plot showing the theoretical effect of various percentages of defect area on the simulated impedance
behaviour of polymer-coated steel with a total area of 10 cm
2
. Hypothetical defect areas are indicated using both area% and the
ASTM D610 scale [21].
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2521
Fig. 9. Experimental set-up used for the impedance analysis
with a reference electrode at the metal/polymer interface [11].
the inuence of inhibitors and pigments on the activity
of these defects.
Close to the metal surface, current lines follow radial
curves into the solution. Assuming that the specic
resistance z of the electrolyte is constant, the current
lines lead to hemispheres of constant potential. The
current lines intersect these hemispheres perpendicu-
larly. The values of the potential are a function of the
current, the specic resistance of the electrolyte z, and
the distance of the hemisphere to the current source or
sink.
By measuring the potential differences DV between
point A and point B, which are separated by a distance
2d, local currents can be calculated according to
i
local
=
1
z
DV
2d
(3)
Scanning reference electrodes enable the measure-
ment of potentials as a function of the location. The
utilisation of glass capillaries in combination with refer-
ence electrodes such as a calomel electrode enable
measurement of corrosion potentials while pseudo-ref-
erence electrodes such as Pt-wires are used to measure
the potential difference between two points in solution.
Included in this latter category are closely spaced refer-
ence electrodes that give a direct measure of the current
density from the potential gradient and distance be-
tween them as shown in Fig. 11(a). Such electrodes are
used for the scanning reference electrode technique
(SRET).
The vibrating probe of the SVET gives a direct
measure of the electric eld, or from Ohms law, the
component of current density at the point in the direc-
tion in which the electrode vibrates (Fig. 11(b)). The
vibration may be parallel or perpendicular to the sur-
face of the investigated sample.
The advantages with respect to the SRET are the
higher local resolution and sensitivity for small currents
based on the applicable lock-in technology. As a stan-
dard microprobe, a thin Pt-wire, which is isolated ex-
cept at its tip, is used to map the surface and its
potential is measured versus a reference electrode,
which is immersed into the same electrolyte but far
away from the sample surface. Usually, the Pt-tip is
platinised to reduce the interfacial impedance of the
probe.
The local galvanic current and the measured poten-
tial difference perpendicular to the sample surface are
again correlated according to Eq. (3).
Although Eq. (3) gives a direct correlation of mea-
sured potential difference and vertical current, usually a
calibration of the reference electrode is done which
results in a coefcient which takes into account the
experimental conditions such as the impedance of the
microelectrode, real amplitude of the vibration and
resistance of the electrolyte. For the calibration process,
Measuring local electrode potentials by means of
scanning reference electrodes has a long history in
electrochemistry. Recently, the development of the
SVET led to the utilisation of localised current density
maps to detect local defects in organic coating after
forming or to measure the activity of cut edges in
corrosive environments [2325]. The local measurement
of currents can not overcome the inherent difculty in
measuring corrosion underneath high resistance coat-
ings but helps to understand the origin of defects and
Fig. 10. Impedance spectra (Bode plots) on the coated iron
sample after a change from N
2
to O
2
-purged electrolyte (0.1 M
NaCl). (a) Total impedance, (b) partial impedance; 1, N
2
, 0
min; 2, O
2
, 15 min; 3, O
2
, 65 min; 4, O
2
, 110 min; 5: O
2
, 320
min [11].
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2522
Fig. 11. Schematic of the two electrode probe conguration of the SRET (a) and the one electrode conguration of the SVET (b)
[14].
usually a point current source of known current and
size is measured by the SVET. In recent years the
application of the SRET and SVET covers also the eld
of polymer-coated metals.
Recently Zou et al. investigated the degradation of
coil-coated galvanised steel at the cut edge [25]. The
authors focused on the inuence of the chromate con-
tent in the coating on the active zinc dissolution at the
cut edge. Chromate clearly led to a rapid diminishing of
anodic activity of the exposed zinc. The chromate free
primer did not lead to the equivalent inhibition of the
zinc dissolution (see Fig. 12). Not surprisingly, the
inhibition of the Zn dissolution measured with the
SVET was in accordance with a high corrosion resis-
tance of the chromate containing primer.
Recently, local impedence spectroscopy (LEIS) based
on the SRET has been applied to study the local ac
solution current density above polymer-coated metals
[26,27]. The magnitude of the local impedence Z(v)
local is derived from equation Eq. (4).
Z()
local
=
V(
applied
)
I()
(4)
where V()
applied
is the magnitude of the applied
voltage pesturbolon between the reference and working
electrode.
Twenty-six carbon steel samples were polished,
cleaned and then contaminated by dropping a small
amount of NaCl solution on the surface of the speci-
men. After drying the sample was coated with an epoxy
resin. Impedance measurements were done in a dilute
NaCl solution. The authors revealed that even above a
visible blister underneath the coating an impedance
spectrum almost equal to that of the intact area is
measured by LEIS as long as the coating itself is intact.
The reason is the high impedance of the coating in
series with the low impedance of the interface in the
contaminated region. A smaller change observed di-
rectly above the blister was assigned to a local change
in the capacitance of the coating.
A new and promising way of LEIS measurement
based on the SVE technique was developed by Bayet et
al. [28].
2.3. Measuring corrosion at the metal /polymer
interface by means of a scanning Kel6inprobe
The SKP allows the difculty of conventional refer-
ence electrode techniques which require a conducting
path between the reference electrode and the working
electrode to be overcome. In principle the Kelvinprobe
measures the work-function of a sample using the vi-
brating condenser method [29,30]. A schematic of the
SKP is shown in Fig. 13. Under certain circumstances
the work-function is determined by the electrode poten-
tial and therefore the Kelvinprobe is able to measure
local electrode or corrosion potentials. The major ad-
vantage of the Kelvinprobe in comparison to conven-
tional electrochemical devices such as electrochemical
reference electrodes is the fact that the Kelvinprobe
measures electrode potentials without touching the sur-
face under investigation across a dielectric medium of
high resistance.
In principle the Kelvinprobe consists of a metallic
reference electrode, which is separated from the sample
by a dielectric medium and connected to the sample by
a metallic wire.
As after connection of the two metals, the electro-
chemical potential of the electrons within both phases
will be identical a charging of one sample with respect
to the other (Volta-potential difference) will be ob-
served. Therefore, for a given and constant work func-
tion of the reference metal the work function of the
sample can be determined by a measurement of the
G
.
G
r
u
n
d
m
e
i
e
r
e
t
a
l
.
/
E
l
e
c
t
r
o
c
h
i
m
i
c
a
A
c
t
a
4
5
(
2
0
0
0
)
2
5
1
5

2
5
3
3
2
5
2
3
Fig. 12. The current density distribution over a cross section of hot-dip galvanised steel in 10 mM NaCl. The primer P4 (chromate containing) was applied on one side of the
zinc while the primer P6 (chromate free) was applied on the other side. (a) 10 min in 10 mM NaCl, (b) 100 min in 10 mM NaCl, (c) contours of (a) [25].
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2524
Fig. 13. Schematic of the SKP.
2.3.1. System metal polymerhumid air
For non-highly-oriented polymers with a rather small
dipole potential is, the following result is obtained:
E
Corr
=
!W
ref
F

Pol
m
1/2
ref
"
+Dc
Pol
ref
(8)
where W
ref
is the work function of the reference metal,

Pol
the surface dipole potential of the polymer and E
1/2
ref
is the half-cell potential of the reference metal.
Similar to an electrolyte covered metal surface, an
electrode potential of the inner interface would be
measured. However, the physical meaning of this elec-
trode potential is not as obvious, as it cannot be
interpreted by conventional electrochemical kinetics.
The electrode potential may in the absence of any
faradaic current be determined by dipole orientation of
segments of the polymer chain. If, however, faradaic
currents like the reduction of oxygen are possible at the
inner interface, then the interface will be polarised until
the rate of the oxygen reduction is negligible. This is
true e.g. for polymer-coated gold surfaces.
2.3.2. System metal oxidepolymerhumid air
The Volta-potential difference is given by
Dc
Pol
ref
=Db
Ox
Me
+Db
Ox
+Db
Pol
Ox

1
F
v
e
Me

W
ref
F
+
Pol
(9)
and if the Volta-potential drop across the oxide layer is
substituted by the corresponding change in chemical
composition, then for an iron substrate, Dc
ref
pol
is:
Dc
Pol
ref
=
Dv
Fe
3+
/Fe
3+
0
F
+Db
Pol
Ox

W
ref
F
+
Pol
+
RT
F
ln
Fe
3+
Fe
2+
n
(10)
Thus the Volta-potential difference represents the
oxidation level within the oxide scale at the metal/poly-
mer interface [39].
2.3.3. System metal /metal
oxideelectrolytepolymerhumid air
This system is typical for a delaminated interface
where an electrolyte layer is existent between the sub-
strate and the polymer. This potential drop is called the
membrane or Donnan-potential and is directly associ-
ated with the preferential incorporation of ions into the
polymeric matrix [38,40,41]. As a result:
E
Corr
=
W
ref
F

Pol
m
1/2
ref
+Db
D
+Dc
Pol
ref
(11)
Therefore the Volta-potential difference Dc
ref
Pol
allows
the corrosion potential at the inner metal/electrolyte
interface buried below the polymeric coating to be
measured only if the Donnan-potential is known or
small. Usually the Donnan-potential is of specic im-
Volta-potential difference. Dc
ref
sample
is usually measured
by the vibrating condenser method of the Kelvinprobe
[3032]. In this technique, sample and reference elec-
trode form a condenser and the reference electrode is
forced to vibrate in a mean distance to the sample d( by
an amplitude Dd. Then the change of the capacitance C
is given by:
C=mm
0
A
d(+Dd sin(t)
(5)
where m is dielectric constant of the medium, m
0
is
electric eld constant, is frequency of vibration for
the reference metal, and A is surface area of the refer-
ence plate; and therefore a current results in the exter-
nal circuit which is given by:
i =Dc
sample
ref
dC
dt
(6)
If an external voltage U is switched into the external
circuit (see Fig. 13), then:
i =(DcU)
dC
dt
(7)
with i =0 for Dc=U.
Therefore, in a conventional approach, the voltage U
is changed until the current i vanishes, and for this
condition (nulling technique) Dc
ref
sample
is measured.
The voltage between sample and probe necessary to
reduce the AC-signal to zero is regarded as the Volta-
potential difference Dc
ref
sample
. During measurement the
chamber of the SKP is kept at very high relative
humidity (typically \95%) in order to keep the electro-
chemical reactions running.
A relation between the Volta-potential difference
Dc
ref
sample
and the corrosion potential E
corr
of the cor-
roding interface exists, which must be derived for differ-
ent interfaces of interest. For polymer-coated metals the
Kelvinprobe allows to measure the potential distribu-
tion at the inner buried interface [3338].
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2525
Fig. 14. Top and middle sketch: Delamination mechanisms on
polymer covered steel and zinc; bottom: FFC on aluminium
alloys.
reference material. The calibration procedure has
therefore to be performed frequently and typically only
with a simple metal/metal cation system like Cu/
CuSO
4
.
3. Corrosion mechanisms of polymer-coated steel,
galvanised steel and aluminium
As far as the substrate is concerned, the instability of
a metal polymer interface is governed by three crucial
properties: the electron-transfer properties at the inter-
face, the redox properties of the oxide between the
metal and the polymer and the chemical stability of the
interface with respect to those species, which are
formed during the electron transfer reaction (ETR).
The rate of ETRs is strongly inuenced by the
surface composition of the metal. As most materials
are covered by oxides their electronic properties will
determine the rate of ETR. Therefore, metals which
are covered by electron conducting or semiconduct-
ing oxides such as iron or zinc will show a different
reactivity at the substrate/polymer interface in com-
parison to those materials which form highly insulat-
ing oxides such as aluminium.
Certain oxides are characterised by a xed ratio of
anions and cations (e.g. Al
2
O
3
) whereas others have
a strongly potential dependant composition such as
iron-oxides due to the ipping of valence states
(Fe
2+
, Fe
3+
) in the cation sublattice. Any change of
the electrode potential is reected in a change of the
valence states and this will change the semiconduct-
ing properties, as e.g. Fe
2+
-states can be regarded as
donors of the n-type semiconductor [45,46]. Further-
more, during reduction of the oxide the base mate-
rial will be oxidised and this redox reaction may
reduce the adhesion of the coating.
During the electron transfer very reactive intermedi-
ates and reaction products are formed which will
chemically react with the material itself. It is shown
below that major reaction products are OH

ions
which are generated during the reduction of molecu-
lar oxygen. Certain metals such as iron are very
stable under those conditions whereas others like
aluminium or zinc are covered by oxides which are
highly soluble in alkaline electrolytes [47].
Accordingly, it must be expected that steel, zinc-
coated steel, and aluminium will behave rather differ-
ently due to the different electronic, redox and chemical
properties of the interface. In this paper the delamina-
tion reactions on three materials will be compared,
which symbolise limiting cases (Fig. 14):
Iron, with highly electron conducting oxides, which
are stable in alkaline media.
Zinc, with semiconducting oxides, which are not
stable in alkaline media.
portance for polymers with a high density of xed
charges (ion exchange membranes), as polymers with
xed cationic functional groups will exchange anions
exclusively and vice versa [41,42]. Lacquers used for
corrosion protection may have some xed ionic groups;
however their concentration is orders of magnitudes
lower than the one of typical ion exchange membranes.
In most equations of this section the Volta-potential
difference Dc is related to the electrode or corrosion
potential in a linear manner, however a calibration
constant is needed in order to calculate the electrode
potential from a Volta-potential measurement. For
aqueous electrolytes the calibration constant is easily
obtained by measuring the Volta-potential of a metal
electrode which is exposed to an electrolyte containing
the metal cation in a dened concentration [43,44].
Then the electrode potential of the metal/metal cation
system is known from the Nernst equation and the
calibration constant is obtained e.g. by plotting the
known electrode potential versus the Volta-potential
for different metal/metal cation systems. However, it
should be kept in mind, that the calibration constant
contains the work function of the reference metal and
depends strongly on the actual surface condition of the
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2526
Aluminium, with insulating oxides, which are not
stable in alkaline media.
3.1. The delamination of organic coatings on iron
In the presence of oxygen the electrode potential of
the metal/polymer interface changes in a well dened
manner with increasing distance from the defect: close
to the defect the potential is negative, whereas far away
from the defect rather anodic potentials are observed
(Fig. 15). For most coatings the steep increase of the
electrode potential also marks the delamination fron-
tier, the region of anodic potentials representing the
intact interface. The physical origin of the sudden
change of the electrode potential however is given by
the migration of ions from the defect into the interface
which results in the polarisation of the highly polaris-
able interface to the potential of the non-polarisable
defect [34]. This is clearly seen in potential proles
measured in the presence of oxygen, which are similar
to the ones shown in Fig. 15. However, in the presence
of oxygen no steady potential increase is observed
within the already delaminated zone (Fig. 16).
As a result the potential step in Fig. 16 only marks
the incorporation of ions into the interface and delami-
nation is a subsequent reaction. Three regions are
clearly seen in Fig. 15 which should be discussed
separately:
The intact interface is characterised by an anodic
potential plateau. This plateau results from the high
electronic conductivity of the oxide covered iron surface
which allows ETR but no ion transfer reactions. There-
fore, oxygen will be reduced at this interface and this
reaction is balanced by the anodic oxidation of the
oxide. As the electrode potential of the oxide is given
by the activity of Fe
2+
and Fe
3+
states, any oxidation
results in an anodic potential shift which is accompa-
nied by a steady decrease of the donor density and
therefore by a decreasing rate of the ETR. Above a
certain anodic potential the rate of the oxygen reduc-
tion is extremely small and no further anodic potential
shift is observed. This is the nal potential as measured
by the Kelvinprobe. The transient of the anodic poten-
tial shift therefore marks the capability of the surface
towards electron transfer. It has been shown, that a
proper surface treatment decreases the rate of this
anodic potential shift dramatically to a point where
almost no anodic shift is observed due to a completely
blocked interface (Fig. 17).
The sudden potential step marks the most interesting
position, as here reactions will occur which are respon-
sible for the loss of adhesion. As discussed before, the
potential step is caused by the incorporation of ions
into the interface and the galvanic coupling of the
interface to the defect. The cathodic potential step also
marks the reduction of the previously oxidised oxide
Fig. 15. Typical potential distribution of a polymer covered
iron substrate in humid air for different delamination times
(times as indicated, electrolyte in the defect: 0.5 M KCl)
(Reprinted with permission from [36]. 1999 Elsevier Sci-
ence).
and the increase in its donor density. Obviously, this
must result in an increase in the rate of the electron
transfer. Surface analysis reveals no anodic activity in
this area as besides a passive lm no thick oxide layers
are found [35,37]. The anodic counter reaction of the
oxygen reduction therefore must be the dissolution of
the base material within the defect. Indeed a galvanic
current has been measured between both sites [36] and
oxygen is reduced within the zone marked by the
Fig. 16. Potential proles for different delamination times as
indicated. The defect was covered by 0.5 M NaCl. The atmo-
sphere in the SKP was water saturated argon [36].
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2527
Fig. 17. Transient of the potential relaxation in air after
cathodic polarisation at 1 V SHE in borate buffer surface
treatment of the steel substrate as indicated.
potential shift instead of an anodic one. Experiments
have indeed proven that the anodic potential shift is
linked to the galvanic current between defect and the
frontier of incorporation of ions and is measured only
if oxygen will be reduced within the latter zone [3436].
The electrochemical situation is summarised in Fig.
19. The galvanic element is dominated by the electronic
properties of the oxides being present at the interface as
any shift in the electrode potential reects a corre-
sponding shift in the electronic properties and therefore
in the rate of the oxygen reduction at the interface.
During oxygen reduction a strongly alkaline electrolyte
is formed which stabilises the oxide on the metal.
Therefore anodic metal dissolution is never observed
within the zones described above. As the galvanic ele-
ment does obviously not destroy the metallic substrate
the delamination of the organic coating is only caused
by bond breaking within the adjacent organic layer. It
has indeed been proven that intermediate radicals
formed during oxygen reduction are responsible for an
oxidative destruction of the interface and therefore on
iron the instability of the substrate/polymer interface is
directly linked to the rate of oxygen reduction.
3.2. The delamination of organic coatings on zinc
The situation described for iron and steel is also
typical for a zinc/polymer interface (Fig. 20) [48]. Zinc
is also covered by electron conducting oxides and there-
fore allows oxygen to be reduced. Again at an intact
interface the anodic partial reaction is missing and
therefore the oxygen reduction leads to an oxidation of
the oxide scale until the electronic properties are such
that no further electron transfer is possible. This is true
for potentials which are approximately 500 mV more
negative than those of oxide covered iron.
If however the atmosphere is changed to an oxygen
free one, then a rapid decrease of the electrode poten-
tial is observed also for the intact interface (Fig. 21),
whereas for iron the potential is stable for rather long
times. This different behaviour is not well understood
up to now but may be caused by the rather different
donor densities of both oxides. In order to reduce the
iron oxide at the metal/polymer interface a large frac-
tion of Fe
3+
states have to be transformed into Fe
2+
.
In-situ Mo bauer studies have shown Fe
2+
concentra-
tions up to 30% for electrode potentials of 0.3V
H
[31]
. This limits any cathodic potential shift at a stable
iron/polymer interface signicantly. As oxides of com-
parably mixed valence-states are unknown for zinc, a
cathodic potential shift may be possible for a rather
negligible charge.
As soon as ions are incorporated into the interface a
galvanic element is set up, the potential of the interface
is polarised cathodically and oxygen will be reduced.
However, zinc oxides are not stable within the induced
potential increase. In order to compensate the charge
cations will migrate to the zone of oxygen reduction.
This is conrmed by spatially resolved ESCA-measure-
ments (Fig. 18).
Between the defect and the steep potential increase a
steady potential increase to more anodic values is ob-
served. This steady potential increase is observed only
in the presence of oxygen and therefore is linked to the
oxygen reduction below the organic coating. The rea-
son for the potential increase may be twofold: either the
anodic potential shift marks a different chemistry at the
interface or it results from an ohmic potential drop
caused by the galvanic current. The change of the
chemical composition at the interface can be ruled out,
as during the oxygen reduction an alkaline pH is
formed and the open circuit potential of a passive iron
surface in an alkaline medium would show a cathodic
Fig. 18. Substrate: iron. ESCA-concentration prole of the
metal surface after the mechanical removal of the polymer
layer [70].
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2528
alkaline environment. AES sputter proles measured
within the delaminated zone prove the signicant
growth of the oxide scale which indicates anodic reac-
tions. This behaviour is also reected in potential pro-
les measured with the Kelvinprobe after a sudden
removal of oxygen from the atmosphere (Fig. 22). For
zinc potential proles are measured which show a
strong cathodic potential shift within the zone, which
had been assigned to the locus of the oxygen reduction
before. The potential shift even inverts the potential
difference between the defect and the frontier of ion-in-
corporation: now the potential in the latter position is
400 mV more negative than the potential within the
defect. This observation is not caused by an inversion
of the galvanic element as in the absence of oxygen no
galvanic current ows between the defect and the inter-
face. The electrode potentials are only dened by the
thermodynamic equilibrium potential. Within the defect
the equilibrium is given by the Zn/Zn
2+
couple
whereas at the metal/polymer interface the couple Zn/
Zn(OH)
4
2
will prevail and this explains the observed
potential inversion. In the absence of oxygen the Kelv-
Fig. 19. Principle corrosion model explaining the formation of a galvanic element. Upper part: cross section through a metal
polymer interface with a defect in the polymer coating. Central part: overview of the polarisation curves at the defect (left side), the
intact interface (right side) and the situation after galvanic coupling of the parts (lower graph).
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2529
Fig. 20. Typical 2-dimensional potential proles on a polymer
covered zinc substrate as measured with the SKP in air (95%
r.h.) for different corrosion times (as indicated) with 0.5 M
NaCl at the defect (Reprinted with permission from [48].
2000 Elsevier Science).
inprobe is therefore a tool to measure a local pH at the
buried interface and the observed time dependence of
the potential proles are dominated by the diffusion of
OH

along the interface.


Interestingly, zinc/polymer interfaces may therefore
also be destroyed by an anodic metal dissolution trig-
gered by the cathodic oxygen reduction, but the combi-
nation of both reactions will buffer the pH and
therefore limit any destruction of the interface due to
extremely high OH

concentrations. For iron the op-


posite is true.
In comparison to zinc, galvanised steel is of consider-
ably higher technological interest. As long as the defect
only penetrates the coating and zinc is still present at
the defect, the situation is identical to the one of pure
zinc. If, however, the scratch also penetrates the gal-
vanised zone and iron is exposed to the electrolyte at
the defect, the galvanic element will change and now
the defect will be the anode and zinc at the interface
will be the cathode and will protect the defect. Now the
galvanic element between defect and interface is indeed
inverted but this is possible only if the zinc/polymer
interface has been delaminated before. Careful studies
have shown that indeed also for the zinc/iron couple in
a rst step the zinc/polymer interface is destroyed by
cathodic delamination and only then zinc starts to
protect iron while acting as a sacricial anode [48].
3.3. Filiform corrosion (FFC) of polymer-coated
aluminium
FFC of metals is characterised by a thread-like un-
dermining of a coating [4952]. Two different regions
of the progressing laments can be observed: the liquid
lled active head and a tail of corrosion products.
Various metals such as Al, Fe and Mg show this kind
of corrosion underneath a (polymer) coating. An ex-
tended literature review of FFC investigations on alu-
minium is given by Bautista [53], earlier reviews
combined with new experiments have been made by
Hahin [54], Hoch [55] and Ruggeri and Beck [56]. Some
recent results on FFC on aluminium alloys can be
found in Refs. [5764].
The systems polymer-coated Fe and Zn which were
discussed before are characterised by a substrate, which
is covered by electron conducting oxides and therefore
allow electrochemical reactions to occur at any place on
the surface. Delamination is caused by the incorpora-
tion of ions into the interface and the subsequently
formed reaction products of the oxygen reduction,
which will cut bonds at the interface and therefore
extend the delaminated zone. As the incorporation of
ions into an intact interface may be difcult due to the
low dielectric constant a reaction zone will be estab-
lished between the frontier of ion incorporation and the
chemical destruction of the interface. Due to the limited
Fig. 21. 2-dimensional potential map of a polymer-coated zinc
surface; defect covered with 0.5 M NaCl; top: 1640 min
exposure in air; bottom: 21 h after exchange of atmosphere
from air to N
2
. (Reprinted with permission from [48]. 2000
Elsevier Science).
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2530
Fig. 22. Potential proles as measured with the SKP on partly delaminated polymer-coated zinc sample after a change of the
atmosphere from air to pure nitrogen; corrosion time before the change: 2710 min; corrosion times since change as indicated.
(Reprinted with permission from [48]. 2000 Elsevier Science).
spatial resolution of the Kelvinprobe (up to a few tens
of a micrometer), little information exists about the
spatial extension of this reaction zone. Considering the
barrier properties of most coatings the reaction zone
may be in the order of some 10 nm and therefore
cathodic delamination is possible only for the surface of
homogeneous electronic properties.
For Al based alloys this is denitely not the case, as
Al is covered by an insulating oxide and all ETR are
limited to highly localised reactive spots, which can be
identied as intermetallic inclusions of various compo-
sitions, depending on the alloying elements and the
impurities. Therefore, Al-based alloys electrochemically
may be considered as partially blocked electrodes and
the reactivity is determined by the surface area fraction
of the inclusions and the conductivity of their specic
oxide layers. If the spacing between the inclusions
(some mm) is larger then the dimension of the reaction
zone then the reaction sequence as discussed before is
interrupted and cathodic delamination should not be
observed anymore. This is indeed true and under com-
parable experimental conditions blistering and anodic
undermining is observed instead of cathodic
delamination.
If, however, the relative humidity in the storage
chamber is reduced to values of approximately 85%
another type of destruction of the metal/polymer inter-
face is observed: FFC (Fig. 23).
This highly localised attack is not only observed for
Al-alloys but also for steel under comparable experi-
mental conditions. Obviously, cathodic delamination
requires a high water activity besides an electron con-
ducting surface, probably in order to allow ion migra-
tion along the interface. On iron, cathodic delamination
fails with decreasing humidity due to the lack of ion
migration into the interface whereas on Al cathodic
delamination is not observed even at high humidities
due to the lack of electronic conductivity.
There is agreement in the literature on FFC, that two
different areas of a corrosion lament can be distin-
guished: the small, so-called active head (concentrated
electrolyte solution) at the front and the tail of solid
corrosion products.
The corrosion laments start to grow from a scratch
(containing an initiating electrolyte like NaCl solution)
in the polymer coating in perpendicular direction to the
scratch. At their very front, the active head, they carry
an acidic solution of the metal cations and the initiating
anions [56]. The laments do not cross each other
[5052,65], their width increases with increasing RH
[5052] and with thicker coatings [50,51]. The latter
two effects have only been investigated for iron and
steel.
The driving force for the directed growth process is
believed to be an oxygen concentration cell between the
front and the back of the active head [50,51,56]. There-
fore, FFC is closely related to crevice and pitting
corrosion. Ruggeri and Beck [56] calculated transport
rates of oxygen through the polymer coating and
through the tail of corrosion products (FFC on iron)
and thereafter proposed that diffusion through the tail
is dominating. Furthermore, in FFC underneath metal
platings [50], the transport of oxygen through the coat-
ing is impossible.
However, some authors speculated about FFC also
being a type of cathodic delamination, i.e. the cathode
being the propagating reaction site [6668].
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2531
Fig. 23. Photograph of FFC attack on epoxy-coated aluminium alloy AA2024-T3. Filaments start to grow from the articial defect
(scratch) and from the samples edge.
Slabaugh [52] was the rst to attempt measurement
of the corrosion potentials of different parts of the FFC
track to detect anodic and cathodic areas. He did this
by cutting the polymer along the FFC track open,
lifting the organic coating off the surface and immedi-
ately measured the electrode potentials by inserting
micro reference electrodes. This method however means
a major disturbance to the fragile FFC system, and the
obtained results may therefore be doubted.
By using the SKP, one can overcome these difculties
by measuring the corrosion potential in-situ without
destroying or even touching the surface of the polymer
coating above the lament. Therefore, experimental
evidence can be provided for the long proposed, but
not yet proven theory of the electrochemical mechanism
of FFC by measuring the potential distribution.
The different mechanism of FFC (anodic undermin-
ing instead of cathodic delamination) is reected in
rather different electrode potentials around the
laments head. Whereas for cathodic delamination the
delamination front is positively polarised with respect
to the already delaminated zone the head of the liform
Fig. 24. Photograph (a) and corrosion potential distribution (b) of a FFC sample (AA2024-T3). The potential scale is 300 mV from
black (low potential) to white (hig potential) (Reprinted with permission from [69]. 2000 Elsevier Science).
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2532
shows a negative potential with respect to the tail (Fig.
24) [69]. Therefore, the head can be identied as the
local anode and the local cathode is situated behind the
anode within the tail. Nisancioglu [58] was the rst who
considered the role of the intermetallic particles as local
cathodes during FFC of Al alloys.
The expansion of the disbonded zone is therefore not
caused by a chemical destruction of the organic phase
by the intermediates of the reaction products of the
oxygen reduction but by a non-electrochemical attack.
It is believed that the reason for delamination is the
pressure of the highly concentrated electrolyte within
the liform head which is separated from the tail by a
dense membrane. This pressure causes a mechanical
disbonding of the interface as it was proposed by van
der Berg et al. [65] for iron FFC, and therefore allows
the electrochemical reaction front to jump from one
reactive site to the next one without the need of any
electrochemical activity in between.
Scheck [64] reported a correlation between the extent
of FFC and the glass transition temperature (T
g
) of the
polymer coating. He observed a sharp rise in the extent
of FFC attack, if the storage temperature exceeded T
g
,
which provides further evidence for a purely mechanical
deadhesion mechanism.
References
[1] Z.W. Wicks, F.N. Jones, S. Peter Pappas, Organic Coat-
ings: Science and Technology, Wiley, New York, 1999.
[2] G. Fettis, Automotive Paints and Coatings, VCH, Wein-
heim, 1995.
[3] R. Lambourne (Ed.), Paint and Surface Coatings, Ellis
Horwood, New York, 1987.
[4] J.F.H. van Eijnsbergen, Duplexsystems, Elsevier, Amster-
dam, 1994.
[5] A.D. Wilson, J.W. Nicholson, H.Y. Prosser (Eds.), Sur-
face Coatings Vols. 1 and 2, Elsevier Applied Science,
Amsterdam, 1987.
[6] H. Leidheiser Jr., Corrosion 38 (1982) 374.
[7] H. Leidheiser, Jr., (Ed.), Corrosion Control by Organic
Coatings, Science Press, Princeton, NJ, 1979.
[8] H. Leidheiser, Jr., (Ed.), Corrosion Control by Organic
Coatings, NACE, Houston, TX, 1981.
[9] M.W. Kendig, H. Leidheiser, Jr., (Eds.), Corrosion Pro-
tection by Organic Coatings, The Electrochemical Soci-
ety, Pennington, NJ, 1987.
[10] R.A. Dickie, ACS Symp. Ser. 285 (1985) 773.
[11] R. Feser, M. Stratmann, Steel Res. 61 (10) (1990) 482.
[12] N.L. Thomas, J. Prot. Coat. Linings 6 (12) (1989) 63.
[13] R. Feser, M. Stratmann, Werkst. Korros. 42 (1991) 187.
[14] G. Grundmeier, K. Ju ttner, M. Stratmann, Novel electro-
chemical techniques in corrosion research, in: M. Schu tze
(Ed.), Corrosion and Environmental Degradation, Wiley-
VCH, submitted for publication.
[15] J.R. Scully, D.C. Silverman, M.W. Kendig (Eds.), Elec-
trochemical Impedance Analysis and Interpretation,
ASTM, Philadelphia, 1993.
[16] J.R. Mac Donald, Impedence Spectroscopy, Wiley, New
York, 1987.
[17] J. Titz, G.H. Wagner, H. Spahn, M. Ebert, K. Ju ttner,
W.J. Lorenz, Corrosion 46 (1990) 221.
[18] E.P.M. van Westing, G.M. Ferrari, J.H.W. de Wit, Cor-
ros. Sci. 36 (6) (1994) 979.
[19] E.P.M. van Westing, G.M. Ferrari, J.H.W. de Wit, Cor-
ros. Sci. 34 (1993) 1511.
[20] F. Mansfeld, H. Shih, H. Greene, C.H. Tsai, in: J.R.
Scully, D.C. Silverman, M.W. Kendig (Eds.), Electro-
chemical Impedance Analysis and Interpretation,
ASTM, Philadelphia, 1993, p. 37.
[21] J.R. Scully, S.T. Hensley, Corrosion 50 (1994) 705.
[22] M.W Kendig, S. Jeanjaquet, J. Lumdsden, in: J.R. Scully,
D.C. Silverman, M.W. Kendig (Eds.), Electrochernical
Impedance Analysis and Interpretation, ASTM,
Philadelphia, 1993, p. 407.
[23] H.S. Isaacs, Corros. Sci. 28 (6) (1988) 547.
[24] H.S. Isaacs, J. Electrochem. Soc. 138 (3) (1991) 723.
[25] F. Zou, C. Barreau, R. Hellouin, D. Quantin, D. Thierry,
Mater. Sci. Forum 289292 (1998) 83.
[26] F. Zou, D. Thierry, Electrochim. Acta 42 (2022) (1997)
3293.
[27] F. Zou, D. Thierry, H.S. Isaacs, J. Electrochem. Soc. 144
(6) (1997) 1957.
[28] E. Bayet, F. Huet, M. Keddam, K. Ogle, H. Takenouti, J.
Electrochem. Soc. 144 (4) (1997) L87.
[29] Lord Kelvin, Philosoph. Mag. 46 (1898) 82.
[30] F. Kohlrausch, Praktische Physik, vol. 12, Teubner-Ver-
lag, Stuttgart, 1968, p. 320.
[31] M. Stratmann, Die Korrosion von Metalloberachen un-
ter du nnen Elektrolytlmen, VDI-Verlag, Du sseldorf,
1994.
[32] H.D. Liess, R. Maeckel, J. Ren, Surf. Interface Anal. 25
(1997) 855.
[33] M. Stratmann, M. Wolpers, H. Streckel, R. Feser, Ber.
Bunsenges. Phys. Chem. 95 (1991) 1365.
[34] A. Leng, H. Streckel, M. Stratmann, Corros. Sci. 41
(1999) 547.
[35] A. Leng, H. Streckel, K. Hofmann, M. Stratmann, Cor-
ros. Sci. 41 (1999) 579.
[36] A. Leng, H. Streckel, M. Stratmann, Corros. Sci. 4 (1999)
599.
[37] M. Stratmann, R. Feser, A. Leng, Electrochim. Acta 39
(1994) 1207.
[38] M. Cappadonia, K. Doblkofer, M. Jauch, Ber. Bunsen-
ges. Phys. Chem. 92 (1988) 903.
[39] G. Grundmeier, M. Stratmann, J. Appl. Surf. Sci. 141
(1999) 43.
[40] K. Doblhofer, R.D. Armstrong, Electrochim. Acta 33
(1988) 453.
[41] K. Doblhofer, Bull. Electrochem. 8 (1992) 96.
[42] W. Kutner, Electrochim. Acta 37 (1992) 1109.
[43] M. Stratmann, H. Streckel, Corros. Sci. 30 (1990) 681.
[44] S. Yee, M. Stratmann, R.A. Oriani, J. Electrochem. Soc.
138 (1991) 55.
[45] U. Stimming, J.W. Schultze, Ber. Bunsenges. Phys. Chem.
80 (1976) 1297.
[46] U. Stimming, J.W. Schultze, Electrochim. Acta 24 (1979)
859.
G. Grundmeier et al. / Electrochimica Acta 45 (2000) 25152533 2533
[47] M. Pourbaix, Atlas of Electrochemical Equilibria in
Aqueous Solutions, NACE, Houston, 1974.
[48] W. Fu rbeth, M. Stratmann, Corros. Sci. 2000, in press
[49] C.F. Sharman, Nature 153 (1944) 621.
[50] M. Van Loo, D.D. Laiderman, R.R. Bruhn, Corrosion 9
(1953) 277.
[51] H. Kaesche, Werkst. Korros. 11 (1959) 668.
[52] W.H. Slabaugh, W. Dejager, S.E. Hoover, L.L. Hutchin-
son, J. Paint Techn. 44 (1972) 76.
[53] A. Bautista, Prog. Org. Coat. (1996) 49
[54] C. Hahin, in: L.J. Korb, D.L. Olson, J.R. Davis (Eds.),
Metals Handbook, vol. 13, 9th edn., ASM International,
Metals Park, OH, 1987, p. 104.
[55] G.M. Hoch, in: R.W. Staehle, B.F. Brown, J. Kruger, A.
Agarwal (Eds.), Localized Corrosion, NACE, Houston,
TX, 1974, p. 134.
[56] R.T. Ruggeri, T.R. Beck, Corrosion 39 (1983) 452.
[57] H. Hagen, K.-H. Rihm, Farbe Lack 96 (1990) 509.
[58] K. Nisancioglu, H. Leth-Olsen, O. Lunder, Proc. Corr.
Prev. Org. Coat. (1994) 51.
[59] J.E. Pietschmann, H. Pfeifer, Aluminium 69 (1993) 1019.
[60] J.E. Pietschmann, H. Pfeifer, Aluminium 69 (1993) 1081.
[61] J.E. Pietschmann, H. Pfeifer, Aluminium 70 (1994) 82.
[62] A. Rudolf, W.-D. Kaiser, Aluminium 72 (1996) 726.
[63] A. Rudolf, W.-D. Kaiser, Aluminium 72 (1996) 832.
[64] K. Scheck, Thesis, Stuttgart, 1991.
[65] W. van der Berg, J.A.W. van Laar, J. Suurmond, Proc. 3rd
Int. Conf. Org. Coat. Sci. Techn. 1 (1979) 18.
[66] W. Funke, Prog. Org. Coat. 9 (1981) 29.
[67] M.J. Schoeld, J.D. Scantlebury, G.C. Wood, J.B. John-
son, 8th Int. Congr. Metall. Corros. 2 (1981) 1047.
[68] W. Funke, Ind. Eng. Chem. Prod. Res. Dev. 24 (1985) 343.
[69] W. Schmidt, M. Stratmann, Corros. Sci. 40 (1998) 1441.
[70] M. Dannenfeldt, Dissertation, Erglaugen 2000.
.

S-ar putea să vă placă și