Bos 20080311

Prediction of coating durability
Early detection using electrochemical methods
Tom Bos
Cover photo: a pristine silicone coating
Prediction of coating durability

Early detection using electrochemical methods
Proefschrift
ter verkrijging van de graad van doctor
aan de Technische Universiteit Delft,
op gezag van de Rector Magnificus, prof.dr.ir. J.T. Fokkema,
voorzitter van het College voor Promoties,
in het openbaar te verdedigen
op dinsdag 11 maart 2008 om 12:30 uur
door
Wilhelmus Maria BOS

Bachelor of Science in Environmental Technology
geboren te Den Helder
Dit proefschrift is goedgekeurd door de promotor:

Prof. dr. J.H.W. de Wit
Samenstelling promotiecommissie:
Rector Magnificus
voorzitter
Prof. dr. J.H.W. de Wit
Technische Universiteit Delft, promotor
Prof. G.E. Thompson
The University of Manchester
Prof. dr. R. Boom
Technische Universiteit Delft
Prof. Dr.-Ing. G. Grundmeier Universitt Paderborn
Prof. dr. ir. H. Terryn
Vrije Universiteit Brussel
Dr. ir. J.M.C. Mol
Technische Universiteit Delft
Ing. L.G.J. van der Ven
Akzo Nobel Car Refinishes
The research described in this thesis was financially supported by the IOP
Oppervlaktetechnologie research programme of SenterNovem (project number
IOT99001) and TNO Science and Industry.
ISBN 978-90-9022815-0
Copyright 2008 by W.M. Bos
Printed by: Gildeprint drukkerijen B.V., Enschede, The Netherlands
All rights reserved. No part of the material protected by this copyright notice may be reproduced or
utilized in any form or by any means, electronic or mechanical, including photocopying, recording or
by any information storage and retrieval system, without written permission from the author.
Contents
1 Introduction
1.1 Corrosion
1.2 Economic impact of corrosion
1.3 Mitigation of corrosion damage
1.4 Protective coatings
1.5 Research objective and topological breakdown
1.6 Thesis outline
1.7 References
1
1
1
2
3
7
8
10
2 Performance testing of protective coatings

2.1 Introduction
2.2 Outdoor Exposure
2.3 Accelerated tests
2.3.1 Salt spray test
2.3.2 Cyclic testing
2.3.3 Advanced cyclic testing
2.3.4 Overview accelerated tests
2.4 Evaluation methods
2.4.1 Visual evaluation
2.4.2 Electrochemical evaluation
2.5 Acceleration factor
2.6 Correlation
2.6.1 Calculation correlation coefficients
2.6.2 Brief overview of correlation studies
2.7 Conclusions
2.8 References
13
13
13
14
14
16
17
20
22
22
23
23
24
24
26
28
30
3 Electrochemical techniques for coating characterisation

3.1 Introduction
3.2 Electrochemical impedance spectroscopy
3.2.1 History
3.2.2 Measurement principle
3.2.3 Data analysis
3.2.4 Impedance expressions of electrical elements
3.2.5 Electrochemical cell
37
37
37
37
38
41
43
44
Contents
3.2.6 Typical impedance spectra of protective coatings

3.2.7 Physical meaning elements
3.2.8 Constant phase element
3.2.9 Kramers-Kronig transforms
3.3 Electrochemical noise measurements
3.3.1 History
3.3.3 Data analysis
3.3.5 Comparison of noise parameters with impedance spectra
3.4 References
4 Outdoor exposure of coating systems monitored with EIS
4.1 Introduction
4.2 Experimental
4.2.1 Coating material
4.2.2 Measurement equipment
4.2.3 Measurement procedure
4.3 Results and discussion
4.4 Conclusions
4.5 References
45
48
51
56
58
58
59
59
60
61
63
67
67
67
67
67
68
70
83
84
5 Pre-qualification of protective coatings using EIS

5.1 Introduction
5.2 Measurement procedure
5.3 Results
5.3.1 Overall results
5.3.2 Mixing ratio epoxy
5.3.3 Zinc-rich paint
5.3.4 Zinc-rich silicate primer
5.3.5 Corrosion-inhibitive pigments
5.3.6 Chemical resistance moisture-curing silicone rubber
5.4 General conclusions
5.5 References
91
91
92
95
95
97
99
102
105
109
110
113
6 Application of ENM to the study of protective coatings

6.1 Introduction
6.2 Experimental
117
117
117
II
Contents

6.3 Drift
6.4 Asymmetric electrodes
6.5 EIS versus ENM
6.7 References
117
118
123
125
128
129
7 General conclusions
131
Summary
135
Samenvatting
139
Publications related to this work
143
Dankwoord
145
Curriculum Vitae
147
III
1 Introduction
1.1 Corrosion
Although many definitions of corrosion exist, for this thesis, corrosion is defined as
electrochemical degradation, occurring at the metal-solution interface where the metal
is oxidised (anodic reaction) and species from the solution such as oxygen (cathodic
reaction) are reduced [1].
Corrosion occurs because most metals are inherently unstable. Metals are produced
from ores that are rich sources of the requisite elements. Once the ore is mined, the
metals must be extracted, usually by chemical or electrolytic reduction. These
processes are non-spontaneous and it will therefore require a large amount of energy to
obtain the pure metal from the ore. On the other hand, once the metal is refined, it
spontaneously tries to revert to its original stable state [2]. These counteracting
processes are schematically shown in Figure 1.
Refined metal
thermodynamically unstable
man-made
Refining process
non-spontaneous
requires energy
Corrosion
spontaneous
Ore
thermodynamically stable
naturally occurring
Figure 1
Loop of the counteracting refining and corrosion processes.
1.2 Economic impact of corrosion
Corrosion of metals has an enormous economic impact. Cost-of-corrosion studies have

been undertaken by several countries. Although the methodology varies between
studies, all estimates of the total annual cost of corrosion range from 1.5 to 5.2 percent
of each countrys gross national product (GNP, see Table 1) [3].
Chapter 1
In the period from 1999 to 2001, a more systematic study was performed in the U.S. In
this study, the total direct cost of corrosion was determined by analyzing 26 industrial
sectors in which corrosion is known to exist and extrapolating the results for a
nationwide estimate. The total direct cost of corrosion was determined to be $ 276
billion per year, which equates to 3.1 percent of the U.S. gross domestic product
(GDP) [3].
The U.S. study defined the total direct annual corrosion costs as those incurred by
owners and operators of structures, manufacturers of products and suppliers of
services. Indirect costs include factors such as plant downtime (e.g. shutdown of a
nuclear power plant), loss of product (e.g. oil spill), loss off efficiency (e.g. reduced
heat transfer of heat exchangers), contamination (e.g. soluble corrosion products in
pharmaceutical production facilities), etc. [4]. The U.S. study conservatively estimated
the indirect cost to be equal to the direct cost. This suggests that the combined direct
and indirect cost of corrosion probably exceeds $552 billion, representing 6% of the
GDP [5].
Table 1
Overview of cost-of-corrosion studies by country [3].
Country
Total annual cost of corrosion
Percent of GNP
Year
United States
$ 5.5 billion
2.1
1949
India
$ 320 million
1960
Finland
$ 54 million
1965
West Germany
$ 6 billion
3.0
1967
United Kingdom
1.365 billion*
3.5
1970
Japan
$ 9.2 billion
1.8
1974
United States
$ 70 billion
4.2
1975
Australia
$ 2 billion
1.5
1982
Kuwait
$ 1 billion
5.2
1987
$ 276 billion
3.1 (GDP)
2002
United States
*
Not reported in U.S. dollars.
1.3 Mitigation of corrosion damage
Fortunately, the rate at which corrosion processes take place can be reduced, often to a
point where, at least temporarily, corrosion is virtually non-existent.
Introduction
Major corrosion-control methods include barrier protection, cathodic protection and

the use of corrosion inhibitors. In many cases, these methods are used in conjunction.
The concept of barrier protection is to separate the metal substrate from the corrosive
environment. These barriers are frequently applied in the form of organic coatings.
Alternatively, corrosion resistant alloys such as stainless steel spontaneously form a
thin barrier layer that protects the underlying alloy.
Cathodic protection prevents corrosion by making the structure to be protected the
cathode of the corrosion cell. This is achieved by either using a sacrificial anode (of
zinc, aluminium or magnesium) or by impressing a current using an external power
supply and an inert electrode.
Corrosion inhibitors are chemical compounds which, when added in small quantities to
an aggressive environment, are able to reduce the corrosion rate of the exposed metal.
Some inhibitors are adsorbed at the metal surface and thereby produce barrier films,
while other inhibitors reduce the corrosivity of the environment by e.g. pH-buffering
or oxygen scavenging [4,6].
It has been estimated that 25 to 30% of annual corrosion costs in the U.S. could be
saved if optimum corrosion management practices were employed [5].
Many ways have been developed to protect metals from the environment, but most
prominent is the use of organic coatings [1,2,7]. The worldwide production of coatings
in 1996 was about 230 million metric tons and the global coating market is estimated
at about $ 60 billion [in:8].
For economic reasons, the lifetime of these coatings should be as long as possible. The
corrosion protective performance is determined by the quality of the total system,
consisting of the metal substrate, surface pre-treatment, coating type, application
method, curing procedure and the environment [9].
Obviously, the development, selection and qualification of new coatings for a specific
environment, requires a thorough characterisation of the most important coating
properties [10].
1.4 Protective coatings
In this thesis, various properties of protective coatings will be discussed. Therefore,

some general coating characteristics will be addressed first.
Chapter 1
Coatings are mainly a dispersion of pigments and fillers, additives and solvents in a
binder matrix [8-12]. These constituents will be introduced in the following
paragraphs.
Binder
The binder forms the matrix of the coating, the continuous polymeric phase in which
all other components may be incorporated. With a few exceptions (e.g. silicate-based
binders in inorganic zinc-rich primers), binders are a mixture of one or more organic
polymers. Binders form the bulk of the physical barrier a coating usually provides in
the protection of its substrate. Its cross-link density and composition largely determine
such factors as the permeability, chemical resistance and UV-resistance of the coating
[10,12].
Binders can be categorized according to the way the coating transforms from the liquid
phase to a solid state, viz. by physical drying, chemical network formation (curing),
coalescence, or a combination [9].
Physical drying takes place by the evaporation of the solvents from the liquid coating
after application. The solid (non-volatile) parts of the coating remain. The coating can
be re-dissolved in the appropriate solvent. Examples of physically drying coatings are
classic nitrocellulose lacquers and chlorinated rubbers [13].
Coatings that cure by chemical reactions are generally two-component coatings that
polymerize through a chemical reaction initiated by mixing resin and hardener. In
contrast with physical drying, cured coatings cannot be re-dissolved since the reactions
are irreversible. Examples of chemically curing coatings are epoxies that cure with
amines [14] or polyurethanes that are formed by reaction of polyisocyanate with
polyol (polyalcohol) [15].
Film formation by coalescence usually starts with a dispersion of small particles in
water. When the solvent evaporates the polymer, particles blend together and build
an insoluble film. This is true for water-borne acrylic coatings. These one-component
materials cure by solvent evaporation followed by coalescence of the resin particles
[16]. Another example of this type of film formation is the sintering of polymer
particles in a powder coating by heating the coated specimen. The powder coating
particles will melt by the temperature increase and fuse together [9].
Pigments
Pigments contribute to several properties of organic coatings. In fact, a number of
different pigments may be used within the same coating, all with their specific
contribution to the coatings characteristics.
Introduction
Obviously, pigments may provide the coating with almost any colour required, but
pigments are also frequently added for corrosion protection. The mechanical properties
of the coating are also affected by pigments.
Three types of corrosion-protective pigments can be distinguished: barrier, sacrificial
and inhibitive types. Some pigments are involved in more than one protection
mechanism.
Barrier pigments are added to a coating to lower the permeability for water, oxygen
and other corrosive species. Besides this, they protect the binder against UV radiation.
Maximum effectiveness can be reached if the following requirements are met [9,12]:
The pigments are impermeable.
The pigments are chemically inert in the barrier coating.
The pigments have a flake- or plate-like shape that is aligned parallel to the
substrates surface. This way the length of the diffusive pathway is increased.
Wetting of the pigment by the binder must be excellent, and the binder-pigment
adhesion must be resistant to wet conditions.
Examples of barrier pigments are mineral-based materials such as micaceous iron
oxide (MIO) and metallic flakes of, for example, aluminium.
Unlike barrier pigments, sacrificial and inhibitive pigments can become chemically
active in the cured coating.
Metallic zinc, when present in high enough concentration, provides sacrificial
(cathodic) protection. When in electrical contact with the steel surface, zinc acts as the
anode and protects the steel cathode, until metallic zinc becomes depleted [12].
Coatings utilizing inhibitive pigments release soluble species, such as phosphates, into
any water that penetrates the coating. These species migrate to the metal substrate,
where they inhibit corrosion by facilitating the growth of protective surface layers
[9,17,18].
Fillers
Since many of the newer pigments are mostly rather expensive, only a minimal
amount is used. Fillers can further increase the volume of the coating through the
incorporation of low cost materials (chalk, mica, clay, etc.). They also may be used to
improve coating properties such as impact and abrasion resistance and water
permeability. For polyester coatings, fillers are used to minimize the internal stresses
Chapter 1
in the coating. This is particularly important since polyester coatings shrink

considerably during the curing process and the cured film has a high thermal
expansion coefficient [9,10,12].
Additives
With respect to the corrosion protective properties, the most important components of
a coating are the binder and the corrosion-protective pigments. Additives are primarily
related to the manufacturing, application and curing process of the coating.
Additives refer to a large group of components with very specific properties, which
typically are added to paint in small quantities. The functions of the different additives
are very diverse.
Thixotropic agents, for example, are used to control the viscosity of the coating.
Thixotropic fluids are shear-thinning fluids. Coating application by brushing, rolling or
spraying all produce different amounts of shear stress to the coating. For thixotropic
coatings, this temporarily lowers the viscosity, which eases its application. Once
applied, the coating film is subjected to the low shear conditions of gravity.
Consequently, the viscosity of the coating increases back towards its low-shear value.
This process is also desirable as it ensures that the film remains in place during the
drying process.
Another important group of additives are surfactants. Surfactants are used to control
the surface energy of the complete coating or of one of its constituents.
Wetting agents that are used to lower the surface tension of the complete coating
ensure that the coating spreads out and adequately wets the surface of the metallic
substrate, forming a continuous film.
The incorporation of pigments in the polymer matrix is a complicated process due to
the intrinsic incompatibility with the binder. For this reason additives like wetting or
pigment dispersion agents are used.
Other examples of additives are biocides, UV-absorbers, antioxidants, antifoam agents
and corrosion inhibitors. The latter example should not be confused with inhibitive
pigments. The additives are completely soluble in order to provide additional corrosion
protection upon application of the coating [9,10,12].
Solvents
Solvents are used to (temporarily) reduce the viscosity of the coating. Reduction of the
viscosity is necessary to enable homogeneous mixing of the binder and other
components. Furthermore, the reduced viscosity makes it possible to apply the coating
in a thin, smooth and continuous film on a specific surface.
Usually, a mixture of organic solvents is used in order to optimize the viscous
Introduction
properties and to control the drying or curing process. The choice of the solvent
mixture is very important. In the liquid state, prior to application, paint should form a
stable dispersion of binder, pigments and additives in the solvent. All solid
components should be homogeneously distributed in the liquid phase. This requires a
high compatibility between solvent and components. An improper formulation can
diminish the barrier properties of the coating, or can even cause phase separation of
the dispersion [9,10].
Due to tougher environmental and health measures, such as VOC (volatile organic
compounds) regulations, new coating formulations have been developed to reduce the
use of volatile organic solvents. This led to the development of solvent-free epoxy
coatings and waterborne coatings [9,14].
1.5 Research objective and topological breakdown
The objectives of this Ph.D. research are two-fold.

From an industrial perspective, it is economically attractive to develop methods for the
mitigation of corrosion damage of coated structures.
From a scientific point of view, it is interesting to acquire a fundamental understanding
of how protective coatings function and how these coatings deteriorate with time.
Broadly speaking, there are two ways to determine the suitability of a coating for a
specific environment. The first approach involves the use of so-called accelerated
tests. Since the in-service degradation of modern protective coating systems is
extremely slow, accelerated testing aims to reproduce, in a much shorter time than in
the field, natural degradation processes without changing the degradation mechanism.
The second approach utilizes sensitive electrochemical techniques such as
Electrochemical Impedance Spectroscopy (EIS) and Electrochemical Noise
Measurement (ENM) to allow early detection of relevant degradation processes under
non-accelerated conditions.
Both approaches will be addressed in this thesis, as is pointed out in the topological
breakdown shown in Figure 2.
Chapter 1
Mitigation of corrosion damage

of coated metal structures
Early detection of
relevant degradation
processes
Review of
accelerated test
methods
Electrochemical
Impedance
Spectroscopy
Electrochemical Noise
Measurements
General conclusions
Figure 2
Topological breakdown of this thesis.
1.6 Thesis outline
In this thesis, various properties of protective coatings will be discussed in detail.

Therefore, general coating characteristics are introduced in chapter 1.
Chapter 2 reviews the most widely used (accelerated) tests for protective coatings.
The associated evaluation methods are also discussed. Based on this the review,
conclusions will be drawn upon the usability of these tests.
As corrosion of (coated) metals is an electrochemical process, it is sensible to
characterise the performance of protective coatings with electrochemical measurement
techniques. In this thesis, Electrochemical Impedance Spectroscopy (EIS) and
Electrochemical Noise Measurements (ENM) are used to this purpose. Chapter 3
provides a theoretical background of these techniques.
Chapter 4 discusses the results of outdoor exposure of model coating systems,
periodically measured with EIS. The results were analysed and related to a unified
model for coating degradation presented by Nguyen and Hubbard [19].
Though a large amount of EIS data are available on protective coatings, measurement
and analysis procedures vary between research groups. This obstructs direct
comparison of measurement results obtained by different groups. Therefore, in chapter
5, a benchmark measurement protocol is introduced. More than 100 industrial coatings
have been measured in compliance with this protocol. After a general overview of the
results, various examples are discussed in more detail.
Introduction
Currently, EIS is the most commonly used electrochemical method for evaluation of
the protective properties of coatings. Recently, the application of ENM for the same
purpose has become of interest. There are number of advantages of this technique over
EIS, such as the low cost of the required equipment and minimal interference with the
measured system.
Despite these advantages, the application of ENM for corrosion studies is only starting
to mature. In chapter 6, results of ENM will be related to those of EIS measurements.
From this, conclusions will be drawn upon the usability of ENM for the study of
protective coatings.
Finally, in chapter 7 general conclusions will be drawn from the work described in this
thesis.
Acknowledgements
The research described in this thesis was financially supported by the IOP
Oppervlaktetechnologie research programme of SenterNovem (project number
IOT99001) and TNO Science and Industry.
Chapter 1
1.7 References
1
Lenderink, H.J.W. Filiform Corrosion of coated Aluminium Alloy - a study of

mechanisms, Ph.D. Thesis, Delft University of Technology, 1995.
Hamer, W.J., Polypyrrole Electrochemistry - Environmentally friendly corrosion

protection of steel: (im)possibilities, Ph.D. Thesis, Delft University of Technology, 2005.
Koch, G.H., M.P.H. Brongers, N.G. Thompson, Y. P. Virmani, J.H. Payer, Corrosion
Cost and Preventive Strategies in the United States, FHWA-RD-01-156, Federal
Highway Administration, U.S. Department of Transportation, Washington, D.C., 2001.
Jones, D.A., Principles and Prevention of Corrosion, Second Edition, Prentice Hall, New
Jersey, U.S.A., 1995, ISBN 0133599930.
Corrosion - A Natural but Controllable Process, Supplement to Materials Performance,

July 2002, 3.
Moavenzadeh, F. (Ed.), Concise Encyclopedia of Building and Construction Materials, ,

MIT Press, U.S.A., 1990, ISBN 0262132486.
Dobbelaar, J.A.L., The use of impedance measurements in coating research The

corrosion behaviour of chromium and iron-chromium alloys, Ph.D. Thesis, Delft
University of Technology, 1990.
Wicks, Z.W., F.N. Jones, S.P. Pappas, Organic Coatings: Science and Technology Second Edition; Wiley, New York, 1999, 630, ISBN 0471245070.
Westing, E.P.M. van, Determination of coating performance with impedance

measurements, Ph.D. Thesis, Delft University of Technology, 1992.
10 Geenen, F.M., Characterisation of Organic Coatings with Impedance Measurements,

Thesis Delft University of Technology, Delft, The Netherlands, Ph.D. Thesis, Delft
11 Munger, C.G., Corrosion Prevention by Protective Coatings, National Association of
Corrosion Engineers, Houston Texas, 1984.
12 Forsgren, A., Corrosion control trough organic coatngs, 2006, CRC Press ISBN
9780849372780.
13 Hare, C.H., Vinyl and Chlorinated Rubber, Journal of Protective Coatings and Linings,
December 1995, 41-58.
10
Introduction
14 Salem, L.S., Epoxies for Steel, Journal of Protective Coatings and Linings, September
1996, 77-98.
15 ODonoghue, M., R. Garrett, V.J. Datta, Straining at a Gnat and Swallowing a Camel:
Health and Performance Issues with Two-Part Polyurethane Finish Coats, Journal of
Protective Coatings and Linings, December 2006, 28-27.
16 Smith, L.M., Introduction to Generic Coating Types, Journal of Protective Coatings &
Linings, July 1995, 73-82.
17 Hernndez, M., F. Galliano, D. Landolt, Mechanism of cathodic delamination control of
zincaluminum phosphate pigment in waterborne coatings, Corrosion Science, 46 (2004),
2281-2300.
18 Hare, C.H., Inhibitive Primers for Metal: Fundamental Considerations, Journal of
Protective Coatings and Linings, May 1998, 48-62.
19 Nguyen, T., J.B. Hubbard and J.M. Pommersheim, Unified model for the degradation of
organic coatings on steel in a neutral electrolyte, Journal of Coatings Technology, 68
(1996), 45-56.
11
12
2 Performance testing of protective coatings
2.1 Introduction
Protective coatings of many different types and formulations are extensively employed
to protect metal structures against corrosion. Obviously, the development, selection
and qualification of new coatings for a specific environment, requires a thorough
characterisation of the most important coating properties [1]. Therefore, coatings have
been evaluated using a variety of outdoor and laboratory tests for more than 100 years
[2,3]. The ability to accurately predict the service life of organic coatings would be
extremely important to coating manufacturers, suppliers and end-users [4-8].
In this chapter, the most widely used tests will be reviewed, together with associated
evaluation methods. Based on this review, conclusions will be drawn upon the
usability of these tests.
2.2 Outdoor Exposure
The most reliable way of studying the suitability of a coating for a specific substrate
and environment is to actually expose coated substrates to the environment in which
the coating will ultimately be applied [1,7,9-16]. This is generally done by mounting
coated panels to exposure racks as shown in Figure 1.
Though long-term exposure to the actual environment offers a good representation of
the actual service life, natural degradation of modern protective coating systems is
extremely slow. Outdoor exposure tests frequently require 10 to 20 years, before
reliable conclusions can be drawn. Consequently, results are not provided in a
commercially acceptable time period [11,14,15,17,18].
A second drawback to this approach is the variability of the natural environment. No
time-period or location is the same as any other and thus exposed materials are
degrading in constantly changing rates and fashions. This greatly influences the
accuracy of the service life prediction [11,19-21].
13
Chapter 2
Figure 1
Typical setup for outdoor exposure of coated panels.
2.3 Accelerated tests
Many researchers have attempted to speed up the natural coating degradation process
by increasing the physical and chemical stresses (e.g. by altering temperature,
humidity, pH, salt concentrations and the intensity of UV radiation) in so-called
accelerated tests. Ideally, the stresses only cause the system to fail faster than it
normally would, while the mechanism of failure remains the same as in the nonaccelerated conditions [12,22-25].
The ability to relate the performance of a coating exposed in such tests to the field
performance is a universal need. If such a linkage can be made, then it should be
possible to greatly reduce the time-to-market for new products by substituting shortterm laboratory exposure results for results from long-term field exposures
[3,5,6,11,26].
2.3.1 Salt spray test
The most widely used accelerated test for coating evaluation is the salt spray test
[11,27-31]. This has been true despite severe criticisms this type of testing has
received [e.g. 6,11,24,26-37].
14
Performance testing of protective coatings
The salt spray test has its origins in the early 1900s and the procedure was
standardized in 1939 under the designation ASTM B117 [31,38]. Since that time, it
has found its way into all spheres of industry.
Salt spray tests are cabinet tests where a salt solution is pumped into a nozzle where it
meets a jet of humidified compressed air, forming a droplet spray. The standardized
pH-neutral salt spray tests ASTM B117 and ISO 9227* continuously subject test
specimens to a fog of salt particles (5 wt% NaCl) at an elevated temperature of 35 C.
The justification for these extreme conditions has always been that a coating system
that can resist these test conditions should also perform well in aggressive service
environments. The assumption was that the mechanisms of corrosion and degradation
in service would be similar to those in the test cabinet [27].
However, many examples can be found in literature suggesting that little, if any,
correlation exists between the results from salt spray tests and in-service performance.
Most remarkably, negative correlations have been observed quite regularly [13,30,3437,39,40]. This means that certain coating formulations, which performed well in salt
spray tests, actually perform worse in the field and conversely, systems that failed in
the salt spray test show a good durability in field exposure.
The constant stress imposed during the test is frequently mentioned as a cause of the
poor correlation with field performance. More specifically, the test does not allow
relaxation of the stress, such as a drying out period. Because the level of stress
normally changes periodically under natural conditions, constant stress produces
degradation that is not observed in the field [11,24,27,29,31]. For example, in zinc-rich
coatings or at galvanized substrates, zinc is not likely to form a passive film as it does
in the field [41].
The prescribed spray electrolyte and the high temperature are also held accountable for
the poor correlation. For barrier coatings, the osmotic forces are much less than in the
field. In fact, they may be reversed completely from which is seen in reality [41].
Except for the harshest of heavy marine exposure sites, the use of 5 wt% NaCl
produces unrealistic results. The exclusion of chemical species, present under normal
conditions, result in unnatural chemistries of the corrosion products [6,33,42]. The
Current standards:
ASTM B117-03, Standard Practice for Operating Salt Spray (Fog) Apparatus, ASTM
International.
ISO 9227:2006, Corrosion tests in artificial atmospheres -- Salt spray tests, ISO.
ASTM B117 and ISO 9227 are essentially similar. ASTM B 117 is specific to neutral salt spray
(NSS), whereas ISO 9227 also covers acetic acid (AASS) and copper acetic acid (CASS) tests.
15
Chapter 2
high temperature in the test cabinet takes many coatings above their Tgs, rendering
them non-protective and poor in film properties [24,41,43-45].
2.3.2 Cyclic testing
To overcome the deficiencies of continuous salt spray tests, cyclic weathering tests
were developed. An early example is the Prohesion test. The development of this test
started in the 1960s when Harrison and his co-workers argued that salt spray tests
based on sodium chloride spray alone could never simulate corrosion in industrial
atmospheres. Instead, they used a mixture of 0.25 wt% NaCl and 3.25 wt% (NH4)2SO4
dubbed Harrisons mix to replace the 5 wt% NaCl solution used in the salt spray
test. When Harrison used this new mixture in the salt spray test, he observed an
improved correlation with coatings exposed for 14 years in an industrial environment.
Timmins developed Harrisons ideas further. He used a diluted Harrisons mix (0.05
wt% NaCl and 0.35 wt% (NH4)2SO4) for his tests and introduced wet and dry cycling
(one hour of fog at ambient temperature and one hour of drying at 35 C). He assumed
that the addition of wet and dry cycles and lowering the solution temperature to
ambient would correspond more closely to natural weathering. Timmins called this
procedure the Prohesion test, which is an acronym for Protection is Adhesion
[14,27,46,47].
In the 1980s, Lyon further refined this method as did Skerry, to the point when it was
recommended to ASTM for adoption in 1994. It is currently issued as ASTM G85 Annex 5 [46].
Incorporation of cycling steps seems intuitively justified, considering that coatings
exposed to the outdoor environment undergo similar effects on a frequent basis [32].
Indeed, many have claimed the superiority of cyclic tests over conventional salt spray
tests, as these tests produce failures more representative of field results, with better
correlation to actual environments [7,11,29,35,37,48]. In fact, one of the most
important distinctions among exposure tests is whether they are constant stress or
cyclic stress [11]. There are a number of reasons for this:
The glass transition temperature (Tg) of a non-crystalline material is the temperature at which a
material's characteristics change from that of a glass (below Tg: hard and brittle) to that of rubber
(above Tg: elastic and flexible).
Current standard:
ASTM G85-02e1, Standard Practice for Modified Salt Spray (Fog) Testing - Annex A5,
dilute electrolyte cyclic fog dry test, ASTM International.
16
Cyclic variation of temperature allows for some thermally induced expansion

and contraction of the materials causing stresses between the coating and the
substrate [27,49].
Absorption of water into a coating, from humid air or surface deposition causes
a volume expansion resulting in stresses within the material. A wet period
followed by a dry period then causes a volume contraction of the surface layers
setting up more stresses within the coating. Cycling of such stresses can
eventually result in fatigue, setting the stage for further chemical and
mechanical change or degradation [20].
Corrosion and wet-dry cycling causes the successive formation and drying of
corrosion products. This results in a mechanical load at the coating substrate
interface which affects the (wet) adhesion of the coating [33,42].
Corrosion on coated steel substrates (e.g. near scribes) is faster during wet-dry
transitions. On wetting, the corrosion rate rises rapidly as accumulated surface
salts dissolve. The rate then decreases as the surface electrolyte dilutes with
continued wetting. The corrosion rate also rises noticeably during drying
because of both the increasing ionic activity as the surface electrolyte
concentrates and the diffusion layer thickness for oxygen as the condensed
phase becomes thinner. Finally, when the ionic concentration of the electrolyte
layer becomes very high and salts begin to crystallize, the corrosion rate
decreases again. These effects may be even more pronounced on zinc substrates
[41].
This all adds up to impart a more realistic stress onto the coating system in an
accelerated manner. Recognizing this, there has been a trend toward cyclic testing and
several manufacturers have developed their own cyclic test methods [33,39,48,50].
2.3.3 Advanced cyclic testing
Test chambers in which specimens are exposed to ultraviolet radiation (UV) are
widely used to obtain weathering data for a wide range of polymer products, including
coatings. Commercially available UV chambers already began to appear circa 1920
and since then numerous modifications have been made [51].
While cyclic testing was seen as an important advance, it was also suspected that
ultraviolet radiation (UV) plays an important role in natural weathering of coated
metals. This led Skerry and his co-workers to investigate the influence of an added UV
weathering cycle. The results indicated that the corrosion performance characteristics
of organic coatings were markedly affected by the UV-weathering factors in the test
[32].
17
Chapter 2
Furthermore, the combined cyclic salt fog/UV test showed an improved reproduction
of coating performance ranking and failure modes observed in practice
[6,12,24,27,31,32]. In this respect, this cyclic salt fog/UV test is superior to continuous
salt fog, or even cyclic salt fog alone.
The upshot of these and subsequent studies was the publication of ASTM D5894 in
1996 [27,31]. This test standardized procedure, exposes panels to alternating
UV/condensation cycles and wet/dry salt-spray cycles.
The UV/condensation cycle begins with 4 hours of exposure to a UVA-340 fluorescent
lamp at 60 C followed by 4 hours of condensation exposure at 50 C. Condensation
takes place by heating a water reservoir. The produced hot water vapour condensates
on the test panels. The duration of this cycle is one week.
The salt-spray cycle conditions are similar to that of the Prohesion test: one hour of
fog [0.05 wt% NaCl and 0.35 wt% (NH4)2SO4] at ambient temperature and one hour of
drying at 35 C. This cycle is also repeated for one week.
This test protocol is usually repeated three to six times, resulting in a total test time of
six to twelve weeks [27,29,31].
Although ASTM D5894 is highly recommended [27,29,37,40], it is not yet the
pinnacle for accelerated test methods.
For example, too-rapid corrosion of zinc has been cited in literature. The standard
spray solution is inappropriate due to the inherent solubility of zinc sulphate corrosion
products. In addition, the pH of the standard solution is between 5.0 and 5.4. At this
pH, zinc reacts at a significantly higher rate than at neutral pH levels. If the
performance of zinc-based and non-zinc coatings must be compared, an alternate (nonsulphate) electrolyte can be substituted under de guidelines of the standard. More
generally, the electrolyte could be modified for composition or concentration
according to the prevailing atmospheric conditions [27,41].
While ASTM D5894 has been found useful for industrial (onshore) applications, the
test described in NORSOK standard M-501** has been developed particularly for the
offshore industry. The NORSOK M 501 test is a standardised Norwegian weathering
test designed for testing materials intended to bear the tough environmental conditions
Current standard:
ASTM D5894-05, Standard Practice for Cyclic Salt Fog/UV Exposure of Painted Metal,
(Alternating Exposures in a Fog/Dry Cabinet and a UV/Condensation Cabinet), ASTM
International.
**
Current standard:
NORSOK standard M-501 Rev. 5, June 2004.
18
present in the North Sea. Even though it is a national standard, it is today probably the
most recognized standard within the field of offshore coatings [12,28,52].
Common for ASTM D5894 and the NORSOK M-501 test is the cyclic exposure to salt
spray, drying and UV/condensation. They differ in the prescribed spray electrolyte,
number of hours exposed in each chamber and the total number of cycles/weeks for
the whole test.
The former NORSOK M-501 test cycle (revision 4) takes 168 hours. Each cycle
consists of 72 hours of salt spray (with synthetic sea water) at 35 C, followed by 16
hours of drying in air and 80 hours of UV/condensation (4 hours of exposure to UV at
60 C followed by 4 hours of condensation exposure at 50 C). The test runs for 25
cycles, resulting in a total test time of 25 weeks [29,41,48,53].
In the fifth revision of NORSOK M-501, published in June 2004, performance testing
of coating systems was brought in accordance with ISO 20340[54].
This ISO standard combines two well-developed cyclic tests: the Norwegian
NORSOK M-501 and the French Standard, NFT 34-600 [55]. A useful feature to come
out of the French standard was to include the freeze cycle at -20C [56]. Since this
freeze cycle is not obligatory, ISO 20340 makes allowance for cyclic testing under
differing conditions as indicated in Table 1.
Table 1
ISO 20340 Cyclic testing options [57].
Option 1
72 hours Salt spray 5% NaCl at 35 C
24 hours Dry out at -20 C
72 hours Condensation/UV
- 4 hours UV at 60 C
- 4 hours condensation at 50 C
Option 2
72 hours Salt spray 5% NaCl at 35 C
24 hours Dry out at +23 C
72 hours Condensation/UV
- 4 hours UV at 60 C
- 4 hours condensation at 50 C
For both options, one cycle takes 1 week (168 hours). The ISO 20340 test also runs for
25 cycles, resulting in a total test time of 25 weeks [57].
The different dry-out temperatures of -20 C and +23 C can have a significant effect
on the results obtained [56,57].
Mitchell, for example, mentioned the effect of the freeze cycle on barrier coating
systems as compared to zinc-rich primed systems. The freeze cycle basically doubled
the under-film creep of the high solids hydrocarbon modified epoxy, whereas on the
Current standard:
19
Chapter 2
zinc-rich primed systems the freeze cycle had virtually no effect on performance. In
both cases, the incorporation of a freeze cycle gives results much more similar to those
seen in practice [56].
Unfortunately, ISO 20340 has become problematic because of the need to compromise
between various national interests [56,58]. One important difficulty was to obtain
agreement on the scribe dimensions. The French delegation and members of the ISO
committee wanted to sustain the use of a narrow scribe of 0.05 mm, while the
Norwegian delegation -supported by most other members- wanted a 2 mm scribe.
As a result, two different scribes are now prescribed for each panel. The problem is
that these scribes can cause cathodic disbondment on immersion panels, with the large
scribe becoming an anode, showing severe metal loss while protecting the narrow
scribe, the cathode [58].
The latest NORSOK standard allows for the omission of the 0.05 mm scribe specified
in ISO 20340.
2.3.4 Overview accelerated tests
In sections 2.3.1 to 2.3.3 a sequential overview has been given of the most widely used
accelerated tests. Parallel to these developments, many other standardized tests have
emerged and many manufacturers have developed their own tests. Table 2 lists some
available test methods.
Table 2
Overview of some (standardized) accelerated tests.
Test Method
Conditions
Continuous salt spray tests
a
Continuous, pH-neutral salt spray (5 % NaCl at 35 C).
ASTM B117
b
ISO 9227 (NSS)
c
DIN 50021 SS
d
JIS Z 2371 (NSS)
Immersion test
a
ASTM D870
Coated specimens are partially or completely immersed in distilled or demineralized water at ambient or elevated temperatures.
Humidity Tests
a
ASTM D2247
Coated specimens are exposed to atmosphere maintained at approximately
b
ISO 6270
100 % relative humidity with the intention that condensation forms on the test
specimens.
20
ISO 20340:2003, Paints and varnishes -- Performance requirements for protective paint
systems for offshore and related structures, ISO.

Table 2 (continued)
Overview of some (standardized) accelerated tests.
Test Method
Cyclic tests
a
ASTM G85 - Annex 5
Conditions
ASTM G 154
b
ISO 4892-3
Advanced Cyclic Tests
a
ASTM D5894
b
ISO 20340
Automotive
GM 9540P/B (General
e
Motors)
CCT-I, IV (Nissan)
VDA 621-415
HCT (Hoogovens
e
Cyclic Test)
VICT (Volvo Indoor

e
Corrosion Test)
f
SAE J2334
Other tests
Kesternich:
a
ASTM G87
b
ISO 3231
c
DIN 50018
Wet and dry cycling (1 h of fog (0.05 wt% NaCl + 0.35 wt% (NH4)2SO4) at
ambient temperature and 1 h of drying at 35 C).
Coated specimens are exposed alternating UV/condensation cycles.
Alternating UV/condensation cycles and wet/dry salt-spray cycles.

ISO 20340 includes an optional freeze cycle.
Wet/dry and humidity cycling. Electrolytic solution: 0.9% NaCl, 0.1% CaCl2,
0.25% NaHCO3, pH: 6-8.
Total cycle time: 24 h. The typical duration of the test is 80 cycles (1,920 h)
[60].
Wet/dry and humidity cycling. Electrolytic solution: 5 % NaCl.
The typical duration of the test is 200 cycles (1,600 h) for CCT-I and 50
cycles (1,200 h) for CCT-IV [60].
Wet/dry and humidity cycling. High time-of-wetness, poor correlation for zinc
pigments and galvanized steel. Also used for testing heavy infrastructure
paints [41].
Based on actual weathering conditions in The Netherlands. The test
simulates two years of exposure including daily and seasonal variations in
1,680 hours. The temperature an relative humidity are varied from
respectively 25 to 50 C and 50 to 98 %. Test specimens are periodically
dipped in an electrolyte. The composition of the electrolyte differs during
winter and summer simulations [62].
Different variants exist. Stresses used are temperature, humidity and salt
solution (spray or submerged). Tends to produce filiform corrosion at
scribes. The typical duration of the test is 12 weeks [41].
24 h cycle consisting of: 6 h 100% relative humidity at 50 C, 15 min. salt
application (0.5% NaCl + 0.1% CaCl2 + 0.075% NaHCO3) and 17 h 45 min.
of drying at 60 C and 50% relative humidity. The typical duration of the test
is 60 days [41].
8 h exposure to water vapour and sulphur dioxide at elevated temperature
and humidity levels, followed by 16 h of ambient laboratory conditions. Test
originally designed for bare metals exposed to a polluted industrial
environment. The relevance for organic coatings is highly questionable
[41,59].
Accelerated outdoor corrosion test. Test specimens exposed outdoors are
periodically sprayed with a salt solution.
SCAB (Simulated
Corrosion Atmospheric
Breakdown) test:
b
ISO 11474
a ASTM International - American Society for Testing and Materials [http://www.astm.org]
b ISO - International Organization for Standardization [http://www.iso.org]
c DIN - Deutsches Institut fr Normung e. V. [http://www2.din.de]
d JIS - Japanese Standards Association [http://www.jsa.or.jp]
21
Chapter 2
e Corporate standard
f SAE International - Society of Automotive Engineers [http://www.sae.org]
2.4 Evaluation methods
2.4.1 Visual evaluation
Coating degradation is usually evaluated using standardized methods [e.g.

in:6,7,10,29,32,34,61]. The most common evaluation methods are summarized in
Table 3. Unfortunately, the majority of these methods are based on visual observation
[63], rendering them rather subjective.
Table 3
Common standardized evaluation methods of specimens subjected to accelerated

tests.
Standard
ISO 4628-2,
ASTM D714
Aspect
Blistering
ISO 4628-3,
ASTM D610
Rusting
ISO 4628-4,
ASTM D661
ISO 4628-5,
ASTM D772
Cracking
ISO 4628-6,
ISO 4628-7,
ASTM
D42148
ISO 4628-8,
ASTM D1654
Chalking
Flaking
Delamination and
corrosion
Description
These standards describe a method for assessing the degree of
blistering of coatings by comparison with pictorial standards.
The ISO standard has adopted the pictorial standards from
ASTM and includes the correlation between the ISO and ASTM
rating systems.
rusting of coated steel surfaces by comparison with pictorial
standards.
The ISO standard includes the correlation between the ISO and
ASTM rating systems.
cracking of coatings by comparison with pictorial standards.
flaking (scaling) of coatings by comparison with pictorial
standards.
These standards describe methods for assessing the degree of
chalking of coatings by comparison with pictorial standards.
These standards specify methods for assessing delamination

and corrosion around a scribe in a coating on a test panel or
other test specimens. The ISO standard describes one method
which involves the use of pictorial standards.
Both standards include numerical rating of failure.
The results heavily depend on the ability of the operator to translate a visual
observation into a performance rating [12,64]. For high-performance coatings that
22
show almost no visual signs of deterioration after weathering, discrimination between

systems is even impossible [42].
2.4.2 Electrochemical evaluation
There has been considerable effort to develop evaluation methods for organic coatings
that are numerical, reproducible and accurate. The use of electrochemical techniques,
in particular electrochemical impedance spectroscopy (EIS), has been shown to be
very useful. EIS not only provides results in a short time but the obtained data can give
indications on the actual corrosion mechanisms. In addition, corrosion and coating
damage may be determined prior to its visual manifestation [1,7,17,25,36,42,65-71].
This coincides with the present trend towards the development of methods which
enable early prediction of coating performance, even before the occurrence of any
substantial changes in its appearance [7,17,42,67,68,69,71-73].
2.5 Acceleration factor
The amount of acceleration provided by a laboratory test can be determined by

relating the test results to those of an outdoor exposure test. With equation 1, the
acceleration factor of the laboratory test can be calculated [41,74,75]:
A=
xtest t field
x field t test
(1)
where:
A = acceleration factor [-]
xtest = response from accelerated test, e.g. creep [mm]
xfield = response from field exposure, e.g. creep [mm]
tfield = duration of field exposure [h]
ttest = duration of accelerated test [h]
Table 4 shows some acceleration factors found in literature.
23
Chapter 2
Table 4
Acceleration factors of accelerated tests compared to outdoor exposure.

Acceleration factors have been calculated for average scribe creep.
Accelerated test
Outdoor Environment
ASTM B 117 [61]
marine exposure site, Sea Isle City,

New Jersey, USA
New Jersey, USA
offshore field test site, Snorre,
Norwegian sector of the North Sea
offshore field test site, Snorre,
Norwegian sector of the North Sea
New Jersey, USA
Cyclic Salt Fog

(modified version of ASTM G85) [61]
NORDTEST NT BUILD 228
(cyclic salt spray, ASTM G85) [75]
NORSOK M501
(Rev. 1, 1994) [75]
Freeze/UV-condensation/Cyclic Salt Fog
(non standardized) [61]
Acceleration
factor (A)
12.5
10.6
55
14
4.16
Tiemens et al. report different acceleration factors ranging from less than 1 to more
than 8. The amount of acceleration was found to depend on the exposed materials. In
addition, similar materials reacted differently to two different accelerated tests [76].
As mentioned in section 2.2, the outdoor environment is highly variable and exposed
materials are degrading in constantly changing rates. Consequently, acceleration
factors are also found to vary over time [74].
Attempts to increase the acceleration factor by intensifying the laboratory testing
conditions, will result in unnatural changes in degradation mechanism and accounts for
a less realistic comparison with natural weathering [15,20,30,77]. Results of Chong
[61] and Knudsen et al. [75] imply that an inverse relation between the acceleration
factor and the correlation of accelerated tests with natural conditions exists. The
correlation to field exposure was found to decrease with increasing test acceleration.
2.6 Correlation
2.6.1 Calculation correlation coefficients
The significance of an accelerated test is measured against how well it correlates with
field performance [27,48]. Correlation coefficients can be considered as indicators of
the uniformity of acceleration within a batch of samples [41]. Correlations coefficients
are often calculated by using the following equation [61,78]:
r=
( x x )( y y )
(x x) ( y y)
i
24
(2)
where:
r = correlation coefficient [-]
xi = single response value from test A, e.g. creep [mm]
x = average response value from test A, e.g. creep [mm]
yi = single response value from test B, e.g. creep [mm]
y = average response value from test B, e.g. creep [mm]
Equation 2 assumes a linear relationship between the two tests and is calculated for
data from samples run in an accelerated test versus the response of identical samples in
a field exposure [41].
A value of r = 0 indicates no linear relationship, whereas a value of r = 1.0 or r = -1.0
suggests a strong linear relationship between the two tests.
Many variations exist in the calculation of the linear correlation coefficient, including
the use of overall performance indices (sum of the individual ratings, e.g. of blistering,
rusting and delamination) [40] and weight factors that reflect their importance in
practical situations [76].
Alternatively, the Spearman rank correlation coefficient is also used for correlation
analysis [6,40,79]. This use of this coefficient has some advantages. The method does
not assume a linear relationship and is based on the ranks of the performance rather
than on values of physical properties, even when the actual values of the measured
systems are unknown. The Spearman rank correlation coefficient is calculated with
[80]:
rs = 1
6 d i2
n(n 2 1)
(3)
where:
rs = Spearman rank correlation coefficient [-]
di = difference in statistical rank of two corresponding systems [-]
n = number of pairs of values [-]
Table 5 shows a fictitious example of the calculation of rs. Raw data of an accelerated
test and an outdoor exposure test are converted into a performance ranking. Note that
the tied score of 2.2 mm scribe creep is assigned the average of the ranks 2 and 3. The
25
Chapter 2
differences d i of the corresponding ranks are calculated and are squared ( d i2 ). The
sum of d i2 is used to calculated rs with equation 3.
Table 5
Coating
A
B
C
D
E
F
Example of the calculation of the Spearman rank correlation coefficient.

Accelerated test
Outdoor Exposure
di
d i2
Scribe Creep (mm)
3.6
4.9
5.1
5.3
3.7
14.2
Rank
1
3
4
5
2
6
Scribe creep (mm)

2.2
4.3
3.6
2.2
1.3
6.7
Rank
2.5
5
4
2.5
1
6
1.5
2
0
-2.5
-1
0
2
i
2.25
4
0
6.25
1
0
=
13.5
rs =
0.92
2.6.2 Brief overview of correlation studies
In literature, many investigations into the correlation of accelerated test with outdoor
exposure have been described. The number of extensive, long-term studies is much
smaller. A few studies deserve to be mentioned here.
Considerable work has been undertaken by both the SSPC/ASTM (Steel Structures
Painting Council / American Society for Testing and Materials) and CSCT (Cleveland
Society for Coatings Technology).
SSPC/ASTM began a round robin program to evaluate protocols for testing industrial
maintenance coatings [34]. However, some coatings contained ingredients that are
questioned for their environmental impact or are legislatively banned.
In order to broaden the usefulness, CSCT decided to undertake a comparable study,
but would investigate newer compliant coatings [5,6]. In the study, the results of
common accelerated tests were compared to 9 diverse exposure sites throughout the
USA. Table 6 shows the averaged correlation of the tests with those of the outdoor
exposure, using different standardized evaluation methods.
26

Table 6
Overview of Spearman rank correlation coefficients of various tests with 12

months of outdoor exposure. Adapted from Carlozzo et al. [6].
Delamination, corrosion
Rusting
Blistering
a
b
Test
ASTM D1654
ASTM D610 ASTM D714
Salt fog
-0.173
0.045
0.058
Cyclic Salt Fog
-0.050
0.315
0.769
Prohesion
-0.122
0.541
0.688
Prohesion/QUV
0.519
0.481
0.782
Outdoor exposure
0.693
(intercorrelation exposure sites)
a Correlation based on 5 exposure sites. Remaining sites not differentiated enough to report.
b Correlation based on 4 exposure sites. Remaining sites not differentiated enough to report.
The highest correlation for rust creepage (ASTM D1654) was found for the
intercorrelation of the 9 exposure sites (rs = 0.69). Of the accelerated tests,
Prohesion/QUV showed the highest correlation (rs = 0.52) for rust creepage [6].
The performance of coated panels is frequently determined by rust creeping from an
artificially made scribe, as this is presumed the only relatively quick way of
differentiating between the performances of high-quality coating systems.
Unfortunately, this is not the way a coating system usually breaks down in practice.
When a coating system ages and eventually breaks down - it is due to general rusting,
blistering, cracking etc. Acceleration by scribing panels prior to the test, facilitates
assessment of the damage-protective properties of the coating/substrate system and not
the barrier properties of the system [22,52].
Indeed, results of quite some exposure sites of the CSCT study did not show sufficient
rusting and blistering.
Although substantial correlations for surface rusting (ASTM D610) and blistering
(ASTM D714) appear to be present for some sites, some results are actually an
indictment of the unnatural failure modes found in high salt concentration electrolytes
[6].
Knudsen et al. performed tests to identify which accelerated test could be used to
assess the corrosion performance of coatings in marine atmospheres [75]. Scribed test
panels were exposed at an offshore field test site for about 5 years. Analogous test
specimens were subjected to 4 different accelerated tests.
The performance was assessed by measuring the maximum and average scribe creep.
The overall correlation factor (r) for average scribe creep was 0.34 for both the
standard salt spray test and the cyclic salt spray test. The correlation of the Volvo test
(Volvo Corporate Standard 1027, 1375 procedure 2A) was 0.76. The correlation of the
NORSOK test was determined for two test series and was respectively 0.62 and 0.76.
27
Chapter 2
A few coatings developed large scribe creep during the field test while this behaviour
was not predicted by any of the accelerated tests. It appears that the high correlations
for the Norsok and Volvo test are mainly determined by outliers with a relatively
strong scribe creep. When these outliers are not taken into consideration the
correlation coefficients decrease considerably.
In 1999 ECCA (European Coil Coating Association) and TNO (Netherlands
Organisation for Applied Scientific Research) executed an extensive research
programme to assess the reliability of different artificial corrosion tests and their
correlation with 10 years of outdoor exposure. The complete program included 49
different systems, 6 European exposure sites and 9 laboratory tests, including
Prohesion/QUV [76].
Correlations were based on the amount of attack in mm2 per m2 (or in mm2 per m
along edges and scribes) and the use of weight factors that reflect their importance in
practical situations.
Although for specific combinations of substrates, tests and exposure sites correlation
coefficients may exceed 0.90, the variation of coefficients is rather large.
From this exhaustive study it was concluded that none of the accelerated tests assessed
in this study could reliably predict the medium and long-term durability of coil-coated
materials.
Besides the correlation between natural and accelerated weathering, standardized tests
also leave some degrees of freedom regarding the operating conditions of the test. This
can cause a large scatter in test results. Hubrecht et al. detail a similar test with similar
specimens that was executed by 4 different laboratories. The amount of scribe creep
was substantially different on many occasions and even affected the performance
ranking of the systems [66].
2.7 Conclusions
Based upon this review of accelerated tests and evaluation methods, the following
conclusions can be drawn:
The most reliable way of studying the suitability of a coating for a specific
substrate and environment is to actually expose coated substrates to the
environment in which the coating will ultimately be applied. Evidently, this
approach is very time-consuming.
28
The standard salt spray test should not be used for service life predictions or
even performance ranking of coated systems.
Cyclic testing imparts a more realistic stress onto the coating system in an
accelerated manner.
The combined cyclic salt fog/UV tests show a still improved reproduction of
coating performance ranking and failure modes observed in practice.
All (accelerated) tests, including outdoor exposures, are relative tests. They do
not give absolute predictions of how many years a material will last in actual
service. They merely provide an indication of relative performance.
Different outdoor service conditions will lead to different coating durabilities.
Some materials may perform particularly well in one environment but may
produce relatively poor results in other environments. It is therefore not to be
expected that a single test method will be representative for all outdoor
exposures.
Attempts to increase the acceleration factor by intensifying the laboratory
testing conditions will result in unnatural changes in degradation mechanism
and accounts for a less realistic comparison with natural weathering.
Standardized visual evaluation techniques are subjective. The results heavily
depend on the ability of the operator to translate a visual observation into a
performance rating. In case of high-performance coatings that almost show no
visual signs of deterioration after weathering, discrimination between systems is
even impossible.
The performance of coated panels is frequently determined by rust creeping
from an artificially made scribe. Scribing panels prior to the test facilitates
assessment of the damage-protective properties of the coating/substrate system
and not the important barrier properties of the system.
The use of electrochemical evaluation techniques, in particular EIS, has been
shown to be very useful. EIS not only provides more precise results in a short
time but the obtained data can give indications on the actual corrosion
mechanisms. Consequently, the sensitivity of this technique allows the total test
time to be shortened.
29
Chapter 2
2.8 References
1
Geenen, F.M., Characterisation of Organic Coatings with Impedance Measurements,

Ph.D. Thesis, Delft University of Technology (1991).
Brunner, S., P. Richner, U. Mller, Olga Guseva. Accelerated weathering device for
service life prediction for organic coatings Polymer Testing, Polymer Testing, 24 (2005)
25-31.
Martin J.W., T Nguyen, E. Byrd, B. Dickens, N. Embree, Relating laboratory and

outdoor exposures of acrylic melamine coatings - I. Cumulative damage model and
laboratory exposure apparatus, Polymer Degradation and Stability 75 (2002) 193-210.
Ochs, H., J. Vogelsang, Effect of temperature cycles on impedance spectra of barrier

coatings under immersion conditions, Electrochimica Acta 49 (2004) 2973-2980.
Andrews, J., F. Anwari, B.J. Carlozzo, M. DiLorenzo, R.Glover, S. Grossman, C.J.

Knaus, J. McCarthy, B. Mysza, R. Patterson, R. Raymond, B. Skerry, P.M. Slifko, W.
Stipkovich, J.C. Weaver, M. Wolfe., Correlation of Accelerated Exposure Testing an
Exterior Exposure Sites, JCT Journal of Coatings Technology, 66 (1994), 49-67.
Carlozzo B.J., Andrews, J., F. Anwari, M. DiLorenzo, R. Glover, S. Grossman, C.

Harding, J. McCarthy, B. Mysza, R. Raymond, B. Skerry, P.M. Slifko, W. Stipkovich,
J.C. Weaver, G. Wilson, Correlation of Accelerated Exposure Testing and Exterior
Exposure Sites, Journal of Coatings Technology, 68 (1996) 47-61.
Morcillo, M., J. Simanacas, J.M. Bastidas, S. Feliu, C. Blanco, F. Camn, Comparison of

Laboratory Tests and Outdoor Tests of Paint Coatings for Atmospheric Exposure.
Polymeric Materials for Corrosion control, R.A. Dickie, F.L. Floyd (Eds.), (1986) 86100.
Tahmassebi N., S. Moradian, S.M. Mirabedini, Evaluation of the weathering

performance of basecoat/clearcoat automotive paint systems by electrochemical
properties measurements, Progress in Organic Coatings 54 (2005) 384-389.
Munger, C.G., Corrosion Prevention by Protective Coatings, National Association of

Corrosion Engineers, Houston Texas, 1984.
10 Morcillo, M., J. Simancas, J.M. Bastidas, S. Feliu, Comparison between laboratory and
outdoor tests of paint coatings for atmospheric exposure, 378-382.
30

11 Appleman, B.R., Survey of Accelerated Test Methods For Anti-Corrosive Coating
Performance, Journal of Coatings Technology, 67 (1990), 57-67.
12 Fedrizzi, L., A. Bergo, F. Deflorian, L. Valentinelli, Assessment of protective properties
of organic coatings by thermal cycling, Progress in Organic Coatings 48 (2003) 271-280.
13 Fekete ., B. Lengyel, Accelerated testing of waterborne coatings, Progress in Organic
Coatings 54 (2005) 211-215.
14 Zapponi M., T. Prez, C. Ramos, C. Saragovi, Prohesion and outdoors tests on
corrosion products developed over painted galvanized steel sheets with and without
Cr(VI) species, Corrosion Science 47 (2005) 923936.
15 Pospil, J., J. Pila, N.C. Billingham, A. Marek, Z. Hork and S. Neprek, Factors
affecting accelerated testing of polymer photostability, Polymer Degradation and
Stability 91 (2006) 417-422.
16 Hoeflaak, M., Matching marine corrosion - New accelerated test gives good correlation
with outdoor exposure, European Coatings Journal 12 (2004), 20-25.
17 Morcillo, M., S. Feliu, J. Simaneas, J.M. Bastidas, Application of the AC Impedance
Measurements to Predict Paint Systems Performance in Marine Atmospheres, Marine
Corrosion and Fouling, Sixth International Congress, Athens, Greece, 5-8 Sept. 1984.
18 Guseva, O., S. Brunner, P. Richner, Service life prediction for aircraft coatings, Polymer
Degradation and Stability, 82 (2003) 1-13.
19 Martin, J.W., Repeatability and Reproducibility of Field Exposure Results. Service Life
Prediction: Methodology and Metrologies. Proceedings. American Chemical Society
Symposium Series 805. 2002, American Chemical Society, Martin, J. W.; Bauer, D. R.,
Eds, 2-22, 2002.
20 Jacques, L.F.E., Accelerated and outdoor/natural exposure testing of coatings, Progress
in Polymer Science 25 (2000) 1337-1362.
21 Boelen, B., B. Schmitz, J. Defourny, F. Blekkenhorst., A literature survey on the
development of an accelerated laboratory test method for atmospheric corrosion of
precoated steel products, Corrosion Science, 34 (1993), 1923-1931.
22 Bierwagen, G.P., L. He, J. Li, L. Ellingson, D.E. Tallman, Studies of a new accelerated
evaluation method for coating corrosion resistance - thermal cycling testing, Progress in
Organic Coatings, 39 (2000) 67-78.
31
Chapter 2
23 Perrin, F.X., M. Irigoyen, E. Aragon, J.L. Vernet, Evaluation of accelerated weathering
tests for three paint systems: a comparitive study of their aging behaviour. Polymer
Degradation and Stability, 72 (2001) 115-124.
24 Bierwagen, G., D. Tallman, J. Li, L. Hea, C. Jeffcoate, EIS studies of coated metals in
accelerated exposure, Progress in Organic Coatings 46 (2003) 148157.
25 Poelman, M., M.-G. Olivier, N. Gayarre, J.-P. Petitjean, Electrochemical study of
different ageing tests for the evaluation of a cataphoretic epoxy primer on aluminium,
Progress in Organic Coatings, 54 (2005) 55-62.
26 Maxted, J., Short Term Testing and Real Time Exposure, Journal of Corrosion Science &
Engineering, 2 (1999), paper 15.
27 Whitehouse, N, B. Skerry, M.J. Crewdson, M. Bjordal, D.G. Weldon, S.-L. Chong, C.
Campbell, Problem Solving Forum: Which Accelerated Test Is Best?, Journal of
Protective Coatings and Linings, August 2000, 17-28.
28 Valentinelli, L., J. Vogelsang, H. Ochs, L. Fedrizzi, Evaluation of barrier coatings by
cycling testing, Progress in Organic Coatings 45 (2002) 405-413.
29 Aamodt, M., New water-borne products matching solvent-borne coatings in corrosion
protection, Jotun, Technical paper (2004), 1-12.
30 Heffer P., B. Lee, Braving the Elements Analyzing the weathering performance of
chromate-free coating systems, Metal Finishing, 103 (2005), 33-34,60.
31 Grossman, D.M., More Realistic Tests for Atmospheric Corrosion, Journal of Protective
Coatings & Linings, September 1996, 40-45.
32 Simpson, C.H., C.J. Ray, B.S. Skerry, Accelerated Corrosion Testing of Industrial
Maintenance Paints Using A Cyclic Corrosion Weathering Method, .Journal of Protective
Coatings & Linings, May 1991, 28-36.
33 Skerry, B.S., C.H. Simpson, Corrosion and Weathering of Paints for Atmospheric
Corrosion Control, Corrosion 49 (1993), 663-674.
34 Boocock, S.K., Meeting Industry Needs for Improved Tests, Journal of Protective
35 Leeuwen, B. Van, Comparison of Accelerated Test Methods with Exterior Exposure,
Protective Coatings Europe, November 1996.
32

36 Krolikowska, A., Soluble Ion Contamination on Steel Surfaces and its Effects on the
Stability of Coatings, In: European Federation of Corrosion Publications, Organic and
Inorganic Coatings for Corrosion Prevention Research and Experiences, Papers from
Eurocorr 96, L.Fedrizzi, P.L. Bonora (Eds.) (1997) 115-127.
37 Repp, J., Accelerted Corrosion Testing - Truths and Misconceptions,
Performance, 41 (2002), 60-63.
Materials
38 Liu, T., Is the Salt Fog Test an Effective Method to Evaluate Corrosion Resistant
Coatings?, 247-254. In: Corrosion Control by Organic Coatings, H. Leidheiser (ed.),
Houston, Texas, NACE, 1981.
39 Meade, C.L., Accelerated Corrosion Testing, Metal Finishing, 98 (2000), 540-545.
40 Gardner, G., ASTMs New Coating Test Method Addresses Interactive Effects of
Weathering and Corrosion, Journal of Protective Coatings and linings, September 1998,
50-62.
41 Forsgren, A., Corrosion control trough organic coatings, 2006, CRC Press
ISBN:9780849372780.
42 Wit, J.H.W. de, E.P.M. van Westing, D.H. van der Weijde, Evaluation of Coatings - A
total system Approach, Materials Science Forum, 247 (1997) 69-82.
43 Li, J., C.S. Jeffcoate, G.P. Bierwagen, D.J. Mills, D.E. Tallman, Thermal Transition
Effects and Electrochemical Properties in Organic Coatings: Part 1 - Initial Studies on
Corrosion Protective Organic Coatings, Corrosion, 54 (1998), 763-771.
44 Deflorian, F., L. Fedrizzi, and S. Rossi, Electrochemical Impedance Spectroscopy and
Fourier Transform Infrared Spectroscopy of Natural and Accelerated Weathering of
Organic Coatings, Corrosion, 54 (1998), 598-605.
45 Hare, C.H., The Glass Transition, Protective Coatings Europe, April 1996, 10-14, 48-53.
46 Souza, K.M. de, ASTM Prohesion Test Predicts Service Performance of Prepainted
Sheet Steel, Galvatech '04, 6th International Conference on Zinc Alloy Coated Sheet Steel,
Conference Proceedings, April 4-7, 2004 Chicago, USA.
47 Timmins, F.D., Avoiding Paint Failures by Prohesion, Journal of the Oil and Colour
Chemists' Association, 62, 131-135 (1979).
48 Claydon, D.A., Performance Testing of Anti-corrosive Coatings, Technical paper,
AKZO -International Protective Coatings (2002).
33
Chapter 2
49 Miszczyk, A., K. Darowicki, Accelerated ageing of organic coating systems by thermal
treatment. Corrosion Science 43 (2001), 1337-1343.
50 Ochs, H., J. Vogelsang, G. Meyer, Enhanced surface roughness of organic coatings due
to UV-degradation: an unknown source of EIS-artifacts, Progress in Organic Coatings 46
(2003) 182-190.
51 Chin, J.W., E. Byrd, N. Embree, J. Martin, and J. D. Tate, Ultraviolet Chambers Based
on Integrating Spheres for Use In Artificial Weathering, Proceedings of the 7th Annual
Meeting Technical Program of the FSCT, November 5-7, 2001, Atlanta, Georgia.
52 Rasmussen, S.N., Corrosion Protection with Coatings - Will Pre-Qualification Testing
Improve Performance, SFL Congress,
53 Frizzi, M.A., E. Aragon, Correlation Between Natural and Artificial Weathering of
Anticorrosive Paints: Analysis of Some Artificial Weathering Cycles, Journal of
Protective Coatings and Linings, September 2002, 58-62.
54 Lunde, P.G., Setting a Standard, Heres how the Norwegian offshore industry adopted
international standards into its own surface prep and coating standard to make
specifying easier and less expensive. Journal of Protective Coatings and Linings, March
2005, 18-21.
55 Binder G., Examination of Accelerated Laboratory Tests for Corrosion Protection,
Protective Coatings Europe, November 2003, 8-16
56 Mitchell, M.J., Testing Of Offshore Systems Is there a realistic testing scenario or is
there still work to be done, Paper 04004, NACE 2004.
57 Thick, J., Offshore Corrosion Protection of Wind Farms, International Protective
Coatings,Technical Paper, http://www.international-pc.com, 2004, 1-10.
58 Mitchell, M.J., A Look at Work in the U.S. on Specifications for Coatings for Offshore
Structures, Journal of Protective Coatings and Linings, March 2005, 22-25.
59 Altmeyer, F., Choosing an Accelerated Corrosion Test, Metal Finishing, 100,
Supplement 1, (2002), 572-578.
60 Goldie, B., Cyclic Corrosion Testing: A Comparison of Current Methods, Protective
Coatings Europe, July 1996.
61 Chong, S.-L., A Comparison of Accelerated Tests for Steel Bridge Coatings in Marine
Environments, Journal of Protective Coatings and Linings, March 1997, 20-33.
34
62 Weijde, D.H. van der, J.H.W. de Wit, M.P.W. Vreijling, E.P.M. van Westing, G.M.
Ferrari, A. Iversen, F. Zou, D. Thierry, N. Dowling, S. Morel, C. Houpert, N. Philips,
Monitoring the Performance of Organic Coatings in the Automotive Industry,
Collaborate Research Report Delft University of Technology, Netherlands Organisation
for Applied Scientific Research - TNO, Swedish Corrosion Institute, Institute de la
Recherche de la Siderurgie, 1996.
63 Goldie, B., Rating of Paint Defects, Protective Coatings Europe, May 1996, 28-29.
64 Fedrizzi, L., A. Bergo, M. Fanicchia, Evaluation of accelerated aging procedures of
painted galvanised steels by EIS, Electrochimica Acta 51 (2006) 1864-1872.
65 Baldwin, K.R., C.J.E. Smith, Accelerated corrosion tests for aerospace materials:
current limitations and future trends, Aircraft Engineering and Aerospace Technology,
71 (1999), 239-244.
66 Hubrecht, J., A.F. Skenazi, J. Vereecken, D. Coutsouradis, Corrosion Resistance of
Galvan and Coil Coated Galvan in Chloride Containing Atmospheres, ATB Metallurgie
25 (1985), 243-252.
67 Wit, J.H.W. de, D.H. van der Weijde, G. Ferrari, Chapter Organic Coatings, In:
Corrosion Mechanisms in Theory and Practice, P. Marcus (Ed.), Marcel Dekker, Inc.,
New York Basel ISBN 0824706668 (2002).
68 Scully, J.R., Electrochemical Impedance of Organic-Coated Steel: Correlation of
Impedance parameters with Long-Term Coating Deterioration, Journal of The
Electrochemical Society, 136 (1989) 979-990.
69 Kendig, M., J. Scully, Basic Aspects of Electrochemical Impedance Application for the
Life Prediction of Organic Coatings on Metals, Corrosion, January (1990), 22-29.
70 Weijde, D.H. van der, E.P.M. van Westing, G.M. Ferrari, J.H.W. de Wit, The use of
Impedance Spectroscopy in life time analysis of organic barrier coatings during
prolonged immersion, cyclic tests and outdoor exposure. Proceedings of the European
Corrosion Congress EUROCORR97, 22-25 September 1997, Trondheim, Norway,
published by NTNU-SINTEF-NKF, 2, 387-392.
71 Davis, G.D., L.A. Krebs, C.M. Dacres, Coating Evaluation and Validation of
Accelerated Test Conditions Using an In-Situ Corrosion Sensor. Journal of Coatings
Technology, 74 (2002), 69-74.
35
Chapter 2
72 Holyoake, K., J. Yuan, Electrochemical Impedance Spectroscopy Measurements of
Barrier Coatings, Armatec, technical paper [www.armatec.co.nz].
73 ODonoghue, M., R. Garrett, V. Datta, P. Roberts, Electrochemical Impedance
Spectroscopy: Testing Coatings for Rapid Immersion Service, Materials Performance,
September 2003, 36-41.
74 Bauer, D.R. Interpreting weathering acceleration factors for automotive coatings using
exposure models. Polymer Degradation and Stability, 69 (2000) 307-316.
75 Knudsen, O.O., U. Steinsmo, M. Bjordal, S. Nijjer., Correlation between Four
Accelerated Tests and Five Years of Offshore Field Testing, Protective Coatings Europe,
December 2001, 51-56.
76 Tiemens, H.J., M. Hoeflaak, Improved Prediction of Durability of Coated Metal Sheet by
Artificial Testing, ECCA Autumn Congress November 18-20 2001, Brussels, Belgium, 116.
77 Mallon, P.E., Y. Li, R. Zhang, H. Chen, Y. Wu , T.C. Sandreczki, Y.C. Jean, R. Suzuki,
T. Ohdaira, Durability of polymeric coatings: effects of natural and artificial weathering,
Applied Surface Science, 194 (2002) 176-181.
78 Hoeflaak, M., B. de Ruiter, J.H. Maas, Calibrating the climate II: Correlations
between different weathering tests, European Coatings Journal, 3 (2006), 30-35.
79 Schulz, U., P. Trubiroha, U. Schernau, H. Baumgart, The effects of acid rain on the
appearance of automotive paint systems studied outdoors and in a new artificial
weathering test, Progress in Organic Coatings, 40 (2000) 151-165.
80 Siegel, S., Nonparametric statistics: For the Behavioural Sciences, Kosaido Printing Co.
Ltd. Tokyo, Japan, 1956.
36
3 Electrochemical techniques for coating characterisation
3.1 Introduction

techniques [1]. In this thesis, Electrochemical Impedance Spectroscopy (EIS) and
Electrochemical Noise Measurements (ENM) are used to this purpose. This chapter
provides a theoretical background of these techniques.
3.2 Electrochemical impedance spectroscopy
3.2.1 History
The concept of impedance was invented by Heaviside between 1880 and 1887.
Heaviside developed the operational calculus, a method of solving differential
equations by transforming them into ordinary algebraic equations, describing the flow
of current in circuits containing resistors, capacitors and inductors. During the same
period, Heaviside coined the expressions of impedance, admittance and reactance.
Heavisides method for solving differential equations was heavily criticized and
indeed was not proven mathematically until the following century [in:2].
Warburg applied the concept of impedance to electrochemical systems. In 1899,
Warburg published a theoretical paper in which he derived the impedance function for
a diffusion process that still bears his name [in:2;3].
One of the earliest electrical (DC) resistance measurements on coatings for evaluation
of corrosion protection properties was made by J.K. Wirth (1939 to 1942) [in: 4,5].
Not much later, in published work of Bacon, Smith and Rugg (1948), the DCresistances of over 300 coating systems were used to establish three general coating
classifications: good (>108 cm-2), fair and poor (<106 cm-2) [6].
Around 1950 Hartshorn and Brasher used the capacitance to evaluate the uptake of
water by polymer coatings. These were usually measurements at a single or at a few
frequencies around 1000 Hz [in:4,7].
37
Chapter 3
One of the earliest applications of EIS for coating research was published by Menges
and Schneider in 1973 [8]. From this point onwards, an increasing number of authors
recognized the power of EIS in the study of corrosion protection by coatings [9]. From
the 1980s, the number of publications on EIS has grown very fast. This sudden
upsurge has been caused by the fast development of accurate instrumentation,
automated data acquisition and software for quantitative data analysis [1,10,11].
Nowadays, EIS is well established as a powerful tool for the investigation of corrosion
protection by coatings. On the other hand, the technique has its practical limitations
and data analysis requires skilled workers.
The term impedance is derived from the Latin word impedire, which means to hinder
[12]. Analogous to electrical resistance, impedance is a measure of opposition of a
system to a perturbation from its steady state. For impedance measurements the
perturbation is usually applied in the form of an alternating voltage, while the current
response of the system is measured [7]. The applied signal normally consists of a DC
bias potential onto which a sinusoidal perturbation Et is superimposed:
(1)
Et = E 0 sin t
with amplitude E0 and angular frequency :

= 2f
(2)
For a linear, time-invariant (LTI) system this will lead to a sinusoidal current response
in the form:
I t = I 0 sin(t + )
(3)
with amplitude I0 and a phase shift . The relation of Et and It is depicted in Figure 1.
38
Electrochemical techniques for coating characterisation

E
E0
I
I0
Figure 1
Input E and output I signal with phase .
LTI systems have the following properties [7,11]:

They are linear, meaning that if input signal x1(t) leads to output y1(t) and signal
x2(t) leads to y2(t), then the input ax1(t)+ bx2(t) leads to the output ay1(t)+ by2(t).
They are time-invariant, meaning that if the input signal x(t) produces an output
y(t) then any time-shifted input, x(t+), results in a time-shifted output y(t+).
Hence, a system reacts the same to the same input signal at any time;
In practice, electrochemical systems are neither linear nor time-invariant. However,
under well-defined conditions many systems do behave as LTI systems in good
approximation. These conditions are [7]:
The variations of the system with time are relatively slow with respect to the
measurement time. In practice, the system can then be considered timeinvariant.
The input and output signals are only small variations around a certain setpoint.
The relation between the input and output signals can then be assumed linear.
The latter condition is illustrated in Figure 2.
Impedance (Z) is the AC equivalent of DC electrical resistance (R) and can therefore
be defined as:
Z ( ) =
Et
It
(4)
39
Chapter 3
I t = I 0 sin(t + )
I0
E0
E t = E 0 sin t
Figure 2
Approximation of a linear I-E relation by the use of a small amplitude potential

perturbation in a non-linear system. The corresponding Lissajous figure (oval) is
superimposed [Based on 1,12,13].
where Z() is the frequency dependant ratio of perturbation over response.

Since both magnitude and phase of the current response determine the impedance, it is
convenient to introduce complex notation by incorporating the complex number j:
j 2 = 1
(5)
The impedance Z can then be expressed as:
Z ( ) =
E 0 exp( jt )
= Z exp( j ) = Z '+ jZ ' '
I 0 exp( jt + )
(6)
Equation (6) defines the impedance Z at frequency as a vector in the complex plane
of magnitude |Z| with an angle between Z and the real (Z) axis.
These vectors can be plotted in a complex plane. In such plots, the real part of the
impedance (resistance) is represented by Z, whereas the imaginary part (reactance) is
40
represented by Z. It should be noted that for coated systems, most measurement

results will appear in the fourth quadrant. For practical reasons, it is more convenient
to flip the imaginary axis. As a result, Z corresponds with the negative imaginary
axis. This is illustrated in Figure 3.
Z
III
IV
IV
imaginary part (reactance)
II
|Z|
real part (resistance)
Figure 3
Left: The four quadrants of a Cartesian coordinate system. Right: Representation

of the impedance by means of a vector in the fourth quadrant of the complex plane.
As impedance is frequency dependant, Z is normally measured over a wide range of

frequencies. Impedance vectors measured at different frequencies are indicated in this
plane only by their endpoints. This plot of Z() is known as a complex-plane plot, or not quite correct- as a Nyquist plot, an Argand diagram, or a Cole-Cole plot.*
It is also possible to plot the modulus of impedance and the phase angle as a function
of the frequency. This results in a Bode plot.
Section 3.2.6 shows typical impedance spectra of organic coatings using both plot
methods.
3.2.3 Data analysis
In general, there are two ways to interpret impedance data: the fundamental approach
and the phenomenological approach.
*
The terms Nyquist Plot, Cole-Cole plot and Argand diagram are often confused. Impedance is
basically derived from measurements of potential and current at a single input port. Nyquists classical
work dealt with two-port measurements of feedback in amplifiers and involved input and output
voltage determinations. However, the use of the term Nyquist Plot has become so widespread for
representation of EIS data that it will be used in this thesis too. An Argand diagram displays a vector
in the complex plane as shown in Figure 3 (right). Cole-Cole plots are plots of the real and imaginary
part of the permittivity for dielectric systems and should not be used in this context [14,15].
41
Chapter 3
The first approach requires a physical model accounting for the processes occurring in
the measured system in terms of physico-electrochemically valid concepts. Based on
this physical model, a transfer function can be formulated. This transfer function is a
mathematical representation of the relation between the input and output of the studied
system. It includes all parameters of the model and is fitted to the measurement data.
In this way, information on the reaction mechanism is obtained directly from the EIS
data.
A drawback of this method is that the reaction mechanism should be almost fully
understood in order to produce a reliable transfer-function. Moreover, electrochemical
systems are often too complex to arrive at useful transfer functions.
For this reason, impedance data are generally analysed using a phenomenological
approach. Instead of starting with a well-founded physical model, EIS data are
interpreted with the aid of electrical equivalent circuits. These circuits consist of
electrical elements such as resistors and capacitors and are designed to have similar
impedance characteristics as the measured impedance data [2,10,16].
A physically meaningful equivalent circuit, dubbed MPEC (most probable equivalent
circuit), can be constructed by following a number of guidelines:
One usually starts with a qualitative circuit, which has been derived from other
research results.
For simple systems, it is also possible to obtain information by graphical
analysis of the Nyquist and Bode plot. Combinations of resistors, capacitors and
other elements each give a characteristic plot shape in these plots.
Consequently, analysis of the plot shape will not only allow an estimate of
individual component values, but may also indicate how they are combined
together.
The hierarchical structure of the different elements in the circuit should be
brought in accordance with a theoretically meaningful model [12,17,15].
When measurement data are fitted to a possible equivalent circuit, distribution
of the errors over the full frequency range should be random and not follow a
recognisable sine wave. If the latter is the case, the system has an additional
time constant and the circuit should be extended.
The number of elements in the circuit should be as low as possible
[1,2,7,10,16,19]. The addition of elements is tempting because this allows better
fitting results of the measurements. However, the physical meaning of these
42
surplus elements is often unresolved. As a rule of a thumb, the simplest model

that both fits with an acceptable error and has a physical meaning is the most
desirable model. Conversely, each element added should improve the
goodness of fit (i.e. chi square should decrease) [16].
The validity of impedance data can be checked using Kramers-Kronig (KK)
transforms [2,12,18]. The principle and limitations of KK transforms are
explained in section 3.2.9.
After configuring a reasonable circuit one can check the impedance at the
frequencies 0 and Hz. Normal electrochemical systems cannot contain
singularities and should have finite impedance for all frequencies [2,10,11].
This is also one of the four general conditions for a successful KK transform.
The values calculated for the elements of the circuit should show a systematical
change with time, which can be logically attributed to changes of the system
[1,16].
In this thesis, the phenomenological approach is used with respect to these guidelines.
3.2.4 Impedance expressions of electrical elements
As mentioned in the previous section, the phenomenological approach involves the

construction of equivalents circuits consisting of different elements. The impedance
expressions of the elements used in this thesis are given in Table 1.
Table 1 Overview of impedance elements used in this thesis [1,10-12,15,16,20].
Element
Symbol
Impedance expression
Z=R
Resistor
R
Capacitor
CPE
(constant
element)
Warburg
(semi-infinite
diffusion)
Q
phase
W
Warburg
Ws
(finite diffusion length)
Z=
1
jC
Z=
1
( j ) n Y0
Z=
1
( j ) Y0
tanh( jB) n
Z = Rd
B= N
n
D
( jB) with
43
Chapter 3
Table 1 (continued)
Overview of impedance elements used in this thesis.
Where:
R
resistance []
C
capacitance [F or s/]
n
power of the CPE; power of the diffusion impedance [-]
n
Y0 admittance constant [s /]
Rd diffusion resistance []
N Nernst diffusion layer thickness [m]
2 -1
D
average diffusion coefficient of electroactive diffusing species [m s ]
The resistor (R) represents the resistance that charge carriers encounter in a specific
process or material. The capacitor (C) represents accumulation of charged species. The
constant phase element, CPE, is a universal element which can represent a variety of
real elements, such as an inductor (n = -1), a resistor (n = 0) and a capacitor (n = 1) or
non-ideal dielectric behaviour (non-integer values for n between -1 and 1). The
Warburg element, W is used to model linear semi-infinite diffusion which occurs
when the diffusion layer has infinite thickness. This element can also be represented
by a CPE with n = 0.5. The Ws element is used when the diffusion layer has finite
dimensions and the boundary is permeable for diffusing species [1,11,12,15,20].
EIS measurements of protective organic coatings on a metal substrate typically require

a cell as shown in Figure 4. This cell consists of a (electro-) chemically inert cylinder,
which is attached to the coating surface and filled with electrolyte. In the
electrochemical cell, a platinum mesh in parallel to the coating surface serves as the
counter electrode (CE) and the metal substrate is the working electrode (WE). The
counter electrode has a circular hole in the middle where the reference electrode (RE)
is protruding.
During the measurement, the sinusoidal voltage perturbation is applied between CE
and WE. Measurements on coated metals are usually carried out under potentiostatic
control, where the DC potential relative to RE equals the open circuit potential (OCP).
In section 4.2.3 practical limitations of OCP measurements are discussed.
Subscript s refers to a shorted transmission line that corresponds to the impedance response of this
element.
44
Reference electrode
Electrolyte
Counter electrode
Working electrode
Figure 4
Typical electrochemical cell for EIS measurements on coated metals.
3.2.6 Typical impedance spectra of protective coatings
In this section, typical (idealized) impedance spectra measured with the setup depicted
in Figure 4 are discussed.
The impedance plot shape changes considerably over the lifetime of a coated metal
sample. Initially, intact protective coatings display largely capacitive behaviour
[11,21]. In the Nyquist plot, a capacitive arc rises steeply up the imaginary axis while
the Bode magnitude plot (logZvs. log f) shows virtually a straight line with slope -1
(Figure 5, solid lines). At longer immersion times, when the electrolyte penetrates the
coating, the arc becomes a semi-circle (dashed lines). The diameter of this semi-circle
generally decreases with time, corresponding to a decrease in intercept with the
logZ-axis of the Bode plot. This signifies a decreasing coating resistance.
The system now behaves as the equivalent circuit of Figure 5, with the electrolyte
resistance (Re), coating capacitance (Cc) and coating resistance (Rc). With this circuit,
the shape of the Bode magnitude plot is characterized by two breakpoint frequencies:
fRcCc and fReCc where:
f Rx C y =
1
2 RxC y
(7)
Breakpoint frequencies are inversely proportional to the related time constants ( =

RxCy) and lie in capacitive-resistive or resistive-capacitive transition regions.
45
Chapter 3
Evidently, the positions of the breakpoints depend on the magnitudes of the

components. Below fRcCc, the capacitor Cc is blocked and the current runs through the
two resistors in series. The network then behaves as a resistor and Z= Re + Rc.
Since Rc is much larger than Re, it dominates the impedance between f = 0 and fRcCc.
The impedance of a resistor is independent of frequency (i.e. has no reactive
component). As a result, the Bode magnitude plot shows a straight line parallel to the
log-f axis and the Bode phase plot shows a phase angle of zero in this frequency range.
Note that the phase axis has been flipped.
Between fRcCc and fReCc the impedance is dominated by Cc, showing a frequency
dependence of the impedance:Z=1/2fCc. For this frequency range the Bode
magnitude plot shows a straight line with a slope of -1. The Bode phase plot rises
towards -90 corresponding to the capacitive behaviour of Cc.
Above fReCc the impedance of Cc becomes negligible and the impedance is only
determined by the electrolyte resistance: Z= Re. Once again, the Bode magnitude
plot shows a horizontal line and the Bode phase plot shows a phase angle approaching
zero.
It should be noted that the adjective Electrochemical in EIS is not quite correct when
used in combination with the study of early stages of coating degradation. During
these stages no electrochemical, but only physical parameters are measured [10].
However, the use of the term EIS is prevalent and therefore this terminology will be
used here too.
As a result of continuing coating degradation, a second semi-circle will appear in the
Nyquist plot as shown in Figure 6. Usually the high-frequency semi-circle is related to
the coating properties while the low-frequency semi-circle contains information about
the processes related to reactions at the electrode surface. The equivalent circuit
describing this stage is shown in Figure 6 and contains two additional elements: charge
transfer resistance Rct and a double layer capacitor Cdl. With the typical relative
magnitude of the components (Rct > Rc > Re and Cdl > Cc) the four breakpoint
frequencies read: fRctCdl < fRcCdl < fRcCc < fReCc.
Below fRctCdl, both capacitors are blocked and the current runs through the three
resistors in series. The network then behaves as a resistor and Z= Re + Rc + Rct.
Since Rct is the largest of the three, it dominates the impedance in this frequency range.
Between fRctCdl and fRcCdl, the larger capacitor Cdl dominates the impedance. In the
46
frequency range fRcCdl and fRcCc, the impedance of Cdl becomes negligible. For higher
frequencies the situation is similar to that described for Figure 5.
log|Z|
Z
Re
Cc
Rc+Re
Rc
Re
log f
fRcCc
fReCc
log f
Figure 5
Typical Nyquist (left) and Bode plot (right) for a coating in the first stage of
degradation.
log|Z|
Z
Re
Cc
Rct+Rc+Re
Cdl
Rc+Re
Rc
Re
Rct
Re
log f
fRctCdl fRcCdl
fRcCc
fReCc
Rct
Rc
Figure 6
log f
Nyquist (left) and Bode plot (right) for a defective coating under activation
control.
When the rate of diffusion of species through pores in the coating becomes slower than
charge-transfer processes, a low-frequency diffusion tail will appear in the Nyquist
plot inclined at the typical angle of 45. This is shown in Figure 7. In this case, the
equivalent circuit should be extended by placing a Warburg element (W) in series with
Rct. Towards low frequencies, the slope of the Bode magnitude plot will approach -1/2
while the phase plot approaches 45. For even lower frequencies, the diffusion tail
may bend towards the real axis.
47
Chapter 3
log|Z|
Z
Re
Rd+Rct+Rc+Re
Rct+Rc+Re
Cc
Cdl
Rc+Re
Rc
Rct
Re
Ws
Re
Rc
Rct
Rd
45
Z
Figure 7
log f
fRctCdl fRcCdl fRcCc fReCc
log f
Nyquist (left) and Bode plot (right) for a defective coating under diffusion control.
3.2.7 Physical meaning elements
Obviously, it is not the aim of EIS to determine the electrical equivalent of a system.
Ideally, all electrical elements are related to specific physical phenomena, which
together characterise the system. In this section, the physical meaning of the elements
of section 3.2.6 is briefly discussed.
Re
Re is the electrolyte resistance between the coating and the reference electrode and is in principle- determined by Re = de/eAc. Here, de is the distance between the reference
electrode and the coating, e is the conductivity of the electrolyte and Ac is the
immersed area of the coating. However, most electrochemical cells do not have a
uniform current distribution and this relation cannot be used. In addition, Re, also
contains other ohmic contributions of the setup, e.g. the ohmic resistance of the
electrical leads. Therefore, Re is generally used to account for the offset of the highfrequency semicircle rather than for the determination of the true electrolyte resistance.
Cc
The coating capacitance Cc is given by:
Cc =
0 r A
d
(8)
where:
Cc = capacitance of the coating [F]
0 = permittivity of vacuum [ 8.85410-12 Fm-1]
48
r = relative permittivity or dielectric constant of the coating [-]

A = surface area of the coating [m2]
d = thickness of the coating [m]
As the relative dielectric constant of coatings is much lower (roughly 4 to 8) compared
to that of water (80), water uptake by coatings results in a significant increase of Cc
[12]. At least initially, Cc should be a measure of water permeation into the coating.
Indeed, many methods have been presented in literature for the calculation of water
uptake by using the coating capacitance. According to van Westing [1] however, all
these methods only yield an approximation of the real situation. Instead, van Westing
suggests the use of a CPE, accounting for the non-ideal behaviour of coatings. The
physical interpretation of this element will be discussed separately in section 3.2.8.
Rc
Rc has been generally interpreted as the pore resistance due to electrolyte penetration
through microscopic pores, or in areas where more rapid solution uptake occurs due to
inadequate cross-linking of the polymer (also see section 4.3) [11,21]. Thus, the
magnitude of Rc is indicative for the state of degradation of the coating. Rc can also
increase with time, probably as a result of pore or defect blockage by corrosion
product build-up.
Incorporation of conductive pigments (e.g. zinc) also enhances the electric
conductivity of the coating, resulting in a lower coating resistance [1].
Cdl and Rct
At places were the coating has detached or has been removed, an electrochemical
double layer is established at the metal-electrolyte interface. This layer can be
modelled with a double layer capacitor Cdl and charge transfer resistance Rct in
parallel. The double layer capacitance Cdl originates from the difficulty of charge
carriers (electrons and ions) to cross the interface, resulting in a separation of charge.
Charge transfer (i.e. leakage of charge through the double layer), can only occur
through electrochemical reactions. The resistance of this charge transfer is represented
by resistor Rct and is related to the rate of the anodic and cathodic reactions.
The elements Cdl and Rct are often considered as a measure for the area over which the
coating has disbonded. However, Amirudin et al. state that Cdl is more a measure of
the electroactive area than the delaminated area [11]. This implies that Cdl also
depends on the electrochemical state of the surface (i.e. active or passive). For the
49
Chapter 3
same reason, the assumption that Rct is solely proportional to the delaminated area is
too crude.
In many cases, Cdl is also replaced by a CPE to account for non-ideal behaviour of the
double layer.
W
The impedance of the metal-electrolyte interface can be considerably more
complicated when diffusion of reacting species is rate determining. This can be both
diffusion of a reactant towards, or a (corrosion) product away from the electrode
surface.
The mathematical form of the Warburg impedance depends on the physical situation it
should represent. For semi-infinite diffusion, which occurs when the diffusion layer
has infinite thickness, the Warburg element can be described by a CPE with an n value
of 0.5.
However, in many cases the diffusion layer is bounded and the impedance expression
for a finite length Warburg element (see Table 1) should be used. When the metal
electrode surface is polarised, the surface concentration of the species that is being
oxidised or reduced drops to zero. Supplementary species will then diffuse to towards
this region of lower concentration. At high frequencies, the Warburg expression shows
the typical 45 diffusion response in the Nyquist plot. In this case, the frequencydependent distance travelled by the diffusing species (dD=(D/)), is much less than
the DC Nernstian diffusion layer thickness (N). This thickness is the arithmetic
thickness which the diffusion layer would have if the concentration profile were a
straight line coinciding with the tangent to the true concentration profile at the
interface and that straight line were extended up to the point where the bulk
concentration is reached (see the analogy with the conductivity gradient of Figure 9).
Towards lower frequencies, dD approaches N and diffusing species eventually reach
the boundary. This boundary has no concentration gradient and is permeable for the
diffusing species. For even lower frequencies, the influence of the resistive boundary
layer increases and the diffusion tail in the Nyquist plot bends towards the real axis
[11,22].
Another impedance expression for diffusion -not discussed here- assumes no permeability for
diffusing species [15].
50

3.2.8 Constant phase element
The Nyquist plots of (coated) metals almost always display depressed semi-circles (i.e.
their centres lay below the real axis). These depressed semi-circles are better
represented by substituting a CPE in place of the capacitance. Figure 8 shows the
relation between angle [radians] of the rotated coordinate system and the n value of
the CPE. It follows that there is no depression for n = 1 [12].
Z
=(1-n)/2
Figure 8
Illustration of a depressed semicircle (solid line) and the corresponding equivalent

circuit.
Though a CPE transforms correctly according to the Kramers-Kronig transforms, this

element only has a simple electrical analogue for integer values of n. The physical
meaning of all other values of n is still a matter of discussion in literature.
It is generally accepted that CPE behaviour originates from a distribution of time
constants [1,7,11,12,24]. However, different physical interpretations can be found in
literature, as is summarized below.
Roughness of the electrode surface
The ideal capacitive behaviour of a perfectly smooth mercury electrode led to the
suggestion that the CPE could be related to surface roughness. The self-similar fractal
dimension Df was introduced as a roughness parameter and was related to the n value
of the CPE [26,27]:
n=
1
D f 1
(9)
It follows that for a smooth surface (Df = 2) that n = 1, while for a highly contorted
surface (Df = 3) n = 0.5.
51
Chapter 3
Bates et al. have carefully compared measured CPE response for electrode-interface
regions with well-characterized electrode roughness profiles but found no correlation
between n and Df [in:15].
Kerner and Pajkossy distinguish between two types of geometrical irregularities:
surface disorder referring to atomic-scale imperfections and surface roughness for
relatively large geometric imperfections [28]. They point out that these terms are
frequently interrelated: for example, a mechanically coarsened surface carries a large
number of dislocations. Surface disorder involves not only atomic-scale nonuniformities but energetic inhomogeneities along the surface as well. The influence of
surface disorder on CPE behaviour was determined experimentally for noble metals
(Au, Ag, Pt) by EIS measurements before and after annealing. A mild annealing step
does not affect roughness but decreases the level of atomic scale inhomogeneity
(thereby reducing the extent of surface energetic differences). It was clearly shown
that annealing results in higher n-values. Thus, it was concluded that surface roughness
itself is not a sufficient condition for the appearance of CPE behaviour.
Kim et al. calculated Df of different carbon electrodes using atomic force microscopy
(AFM) [26]. The corresponding n values were obtained using Equation 9, which were
compared to those of actual EIS measurements on these electrodes. Surface
inhomogeneity was investigated with X-ray photoelectron spectroscopy (XPS).
The n values calculated from Df barely matched those derived from the EIS
measurements. Remarkably, the n values of EIS measurements were closely related to
the surface inhomogeneity. The surface of carbon is composed of basal planes and
edge planes of which conductive and capacitive properties are different from each
other. It was found that the n value decreased with an increasing amount of edge
orientations.
Surface inhomogeneity
The above results strongly indicate that the influence of surface inhomogeneity is
much larger than the influence of surface roughness.
Dobbelaar [25] used two distribution functions (normal distribution and Cole-Cole
distribution) to show mathematically that distribution of activation energies can cause
CPE behaviour. For both functions, a wider distribution resulted in lower n values. In
addition, Dobbelaar demonstrated experimentally that increasing deviation from ideal
capacitive behaviour coincided with increased porosity of passive films on ironchromium alloys.
52
Concentration gradient
Another approach to describe CPE behaviour is derived from the theory of Young [2932]. This theory was originally developed to explain experimental results for oxide
films on Nb electrodes. Young assumed that the conductivity (x) within the oxide
film decreases perpendicular to the metal surface in an exponential fashion due to nonstoichiometry of the layer:
( x) = (0) exp( x / )
(10)
This relation is shown schematically in Figure 9, where corresponds to a

characteristic length and (d) = k(0)exp(-d/) corresponds to the finite conductivity at
the layer surface.
layer
metal
solution
(0)
()=(0)/e
Figure 9
Model of a layer with exponential decay of conductivity perpendicular to the

surface.
According to Mller and Ghr [in:29], the impedance ZY for such a gradient is given
by the following expression:
1 + j exp( p 1 )
p
ln
ZY =
1 + j
jCY
where:
0 r A
CY =
(11)
: the capacitance of the layer with thickness d and area A,
p = / d : relative penetration depth,
53
Chapter 3
= R CY =
0 r
: time constant, independent of A and d.
(0)
From Equation 11 it follows that the frequency behaviour of the Young impedance
depends on the relative penetration depth. For this impedance, the phase angle is
observed to be frequency-dependent:
= 90 [1 q ] with q
1
ln( ) + p 1
(12)
Comparing the expression for the phase angle of the Young impedance with that of the
CPE ( = 90n), one can conclude that the n value corresponds to the term [1-q].
Considering the Young impedance, deviation from ideal behaviour is mainly caused
by the gradient of conductivity.
However, many practical systems can be fitted properly to a CPE, indicating frequency
dependence is small over a wide frequency range.
Explaining this behaviour with the Young Impedance, one finds that in these systems
the gradient of conductivity is the dominant contribution and therefore d/ >> ln().
In this case, q becomes independent of the frequency:
q p = /d
(13)
Rammelt [30] compared the impedance response of a solvent-borne coating with that
of a water-borne coating. The behaviour of the solvent-borne coating could be
described by a CPE while the phase angle of the waterborne coating clearly showed
frequency dependence according to the Young impedance. He concluded that the
Young impedance was a general impedance for films-covered electrodes and that the
CPE can be considered as a special case of the Young impedance.
Thus, with respect to the theory of Young, non-ideal behaviour of dielectric layers on
metals is characteristic for the layer properties, such as thickness, homogeneity
perpendicular to the surface and penetration depth of ions.
XY and Z distributions of time constants
Apart from of the origin of CPE behaviour, Jorcin et al. [24] distinguish two types of
54
spatial distribution of time constants. XY distribution denotes distributions along the

electrode surface while Z distribution denotes distributions in the dimension
perpendicular to the electrode surface (see Figure 10). A combination of these
distributions is also possible.
Z
Y
electrode surface
X
Figure 10 Spatial dimensions relative to the electrode surface.
For both XY and Z distributions, the global electrode impedance results from an
integration of local impedances along the electrode surface. This overall electrode
impedance can be represented as being the result of the parallel combination of local
impedances.
In the case of a XY distribution, the local impedance does not contain a CPE element.
In the case of a Z distribution, all the local impedances are identical and present CPE
behaviour.
According to Jorcin et al. [24] it should be possible to distinguish between XY and Z
distributions by use of local electrochemical impedance spectroscopy (LEIS)
measurements. In the case of XY distribution, if localized electrochemical impedance
measurements sample a sufficiently small surface area, it should be possible to
measure a single element that does not show CPE behaviour. In the case of Z
distribution, the local element will reveal CPE behaviour with the same exponent as
seen in the overall impedance measurement.
In the experimental study Jorcin et al. [24], conventional and localized electrochemical
measurements were performed on two systems: AZ91 Mg alloy and pure aluminium,
both in Na2SO4 electrolytes.
Jorcin et al. [24] use the terms 2D and 3D to denote the two types of time-constant distribution.
However, 3D distribution is actually used to describe a 1D distribution perpendicular to the electrode
surface. To avoid confusion the terms XY and Z distributions are used instead.
55
Chapter 3
A XY distribution was found for the magnesium alloy. In this case the origin of the
global CPE behaviour was attributed to the distribution of resistance associated with
the geometry of the disk electrode; whereas, the capacitance was independent of
position. Local impedances did not contain a CPE element (i.e. the n value of the local
CPEs 1).
In the case of the aluminium electrode, the global CPE behaviour could be attributed
to a combination of XY and Z distributions. Both the distribution of resistance and
capacitance were associated with the geometry of the disk electrode. The local
impedances also displayed CPE behaviour.
Based on this short overview it can be assumed that protective coatings, will display
both XY and Z distributions of time constants.
Coatings are a dispersion of various constituents in a binder matrix and its local
composition will vary with all spatial dimensions (heterogeneity) resulting in XY and
Z distributions. Moreover, the exponential decay of conductivity perpendicular to the
metal surface (Young impedance) will result in a Z distribution.
3.2.9 Kramers-Kronig transforms
The origin of the KK transforms is based on the fulfilment of four general conditions
[18]:
Causality: The response of the system is due only to the perturbation applied
and does not contain significant components from spurious sources.
Linearity: The perturbation and response of the system is described by a set of
linear differential laws. Practically, this condition requires that impedance
should be independent of the magnitude of the perturbation (see section 3.2.2
on LTI systems).
Stability: The system must be stable in the sense that it returns to its original
state after the perturbation is removed.
The impedance must be finite-valued at 0 and and must be a
continuous and finite-valued function at all intermediate frequencies.
According to the KK transform theory, the imaginary part of the impedance can be
calculated from the real part of the impedance and vice versa. This is illustrated in
Figure 11. The Nyquist plot at the top contains the original measurement data. This
data can be slit up in an real (1A) and imaginary (1B) part and are displayed as a
function of the frequency. Steps 2A and 2B denote the actual KK transforms. Step 2A
involves the transformation of the real part into the imaginary part, while the opposite
56
is taking place for step 2B. Step 3A and 3B denote the combination of the transformed
data into a single Nyquist plot. This plot should match the original data. Modern fitting
software includes KK-transform modules to perform these steps automatically.
Z
Z
1A
2A
Log f
Log f
2A
2B
Log f
3A
Log f
3B
Z
Figure 11 Principle of KK-transform.
A limitation of KK transforms is that a successful transformation does not guarantee

that the measurement data contains no contribution of the instrumental setup.
Impedance of the reference electrode, cables or the internal components of the
equipment might also transform correctly.
Another limitation is related to the fact that the transform equations [18] require the
integration of a complete spectrum from = 0 to = . Obviously, measurement data
57
Chapter 3
are never available over an infinite frequency range and an approximation (e.g. with
the aid of an equivalent circuit) must be used for the missing frequencies.
Measurement data are never perfect. Therefore, no data will transform perfectly. This
introduces yet another disadvantage as it remains up to the user to decide whether the
KK transform is successful.
In summary, with the aid of modern software, KK transforms provides a quick
mathematical tool to check whether impedance data violates the four general
conditions. With respect to the limitations of the technique it should be used with
caution.
3.3 Electrochemical noise measurements
3.3.1 History
In 1961, Hagyard and Williams [34] reported large fluctuations (exceeding 800 mV)
on the corrosion potential of small aluminium electrodes in aqueous chloride solutions,
measured with respect to a reference electrode.
Iverson [35] recorded potential fluctuations between corroding test electrodes of
various metals and alloys and a platinum electrode. He found that the amplitude and
frequency of the voltage fluctuations were characteristic for the corrosion behaviour of
the tested materials. Subsequently, other workers realized that current fluctuations
could also be studied.
In 1987, Eden et al. filed a patent describing a technique for the simultaneous
measurement of potential and current fluctuations [36]. Many workers acknowledged
that combination of potential and current noise was more powerful than the individual
measurements and consequently the electrochemical noise resistance method described
by Eden developed into a widely used electrochemical technique [37].
Eden and Skerry were the first to apply ENM to the study of organic coatings [in:38]
and were soon joined by other laboratories. It was found that changes of the noise
resistance with time were in general agreement with the known performance of several
coating systems. However, Mansfeld and co-workers emphasized the need for analysis
of noise data not only in the time domain, but also in the frequency domain in order to
obtain mechanistic information [39]. Therefore, the spectral noise resistance, obtained
trough Fast Fourier Transforms (FFT) of potential and current fluctuations, was
introduced. This method allows not only evaluation of the frequency dependence of
noise data, but also enables a direct comparison with the modulus of impedance
58
obtained with EIS. This practice has been successfully applied by several workers
[38,39,40].
ENM and related analysis methods are only starting to mature, but many papers
suggest that this technique appears very promising for the study of organic coatings.
Electrochemical noise refers to the spontaneous potential and current fluctuations as a

result of electrochemical reactions. In contrast with EIS, where a well-defined external
perturbation of known frequency is applied, ENM involves the measurement of these
self-generated fluctuations in order to obtain information of the measured system.
Though systems measured with ENM do not behave as LTI systems, useful
information can be obtained by using (statistical) analysis methods.
3.3.3 Data analysis
Analysis in the time domain

Generally, the analysis of noise data is made either in the time domain or in the
frequency domain.
For analysis in the time domain, the most commonly used parameter in ENM studies
of coatings is the noise resistance Rn, which is defined as [36]:
Rn =
(V )
(I )
(14)
where (V) and (I) are the standard deviations of voltage and current time records. If
a moving average over a time shorter than the total recording time is employed, Rn
becomes a function of time and its changes can be used to follow changes in the
electrochemical behaviour of the studied system [40].
For coated metals, it is usually observed that (V) decreases and (I) increases as the
coating degrades. It follows that Rn decreases with the lifetime of the coating [39].
Several attempts have been made to relate Rn to the polarisation resistance Rp. Some
workers (e.g. [41]) have experimentally shown that Rn equates well with Rp, while
others stress that this situation can be only reached conditionally [41,42].
59
Chapter 3
Analysis in the frequency domain

Analysis in the frequency domain requires transformation of the potential and current
time records using Fast Fourier Transform (FFT) or the Maximum Entropy Method
(MEM). The obtained Power Spectral Densities (PSDs) are then used to calculate the
spectral noise resistance (Rsn) [42]:
Rsn ( f ) =
V ( f )
I(f )
(15)
where V and I are the PSDs of potential and current noises. The minimum and
maximum frequencies that can be resolved in the frequency domain are given by [42]:
f min =
1
1
, f max =
N t
2 t
(16)
with N the total number samples and t the time interval at which samples are taken.
From equation 16 it follows that the lowest frequency that can be resolved is
determined by the duration of the noise measurement. At least two samples are
required within each cycle of a given frequency. Therefore the highest frequency, also
known as the Nyquist frequency (N) is only half the sampling frequency.
Analogous to analysis in the time domain, V usually decreases and I increases with
the onset of coating degradation [39].
Drift
Frequently, time records of noise data show a certain drift, which affects the calculated
values of Rn and Rsn(f). Therefore, drift has to be eliminated prior to data analysis. If
the drift is linear, it can be removed simply by subtracting the linear regression line
from the noise data. This is a common method of treating drift, especially prior to
spectral estimation. Non-linear drift can be removed by subtracting more complex drift
functions such as central moving average (CMA) removal, polynomial fitting, and
analogue or digital high-pass filtering. Nevertheless, one should be cautious as this is
accompanied by loss of real low-frequency noise of the system under study [37,43].
The measurement setup for the simultaneous measurement of potential and current
noise can vary somewhat, but the principle remains the same. The current noise is
60
measured between two nominally identical working electrodes using a zero resistance
ammeter (ZRA). The ZRA electronically compensates for the measuring current and
ensures that the two electrodes are at the same potential. The potential noise of the
working electrode pair is measured with respect to a reference electrode. This is shown
in Figure 12.
Reference electrode
Working electrode 1
Working electrode 2
Electrolyte
Figure 12 Typical electrochemical cell for ENM on coated metals.
3.3.5 Comparison of noise parameters with impedance spectra
The noise parameters Rn and Rsn both have the dimension of a resistance. The latter
parameter also is a function of frequency and therefore it has some analogy with the
modulus of the impedance.
Figure 13 shows a comparison between Rn, Rsn and |Z| for a coated metal. According
to Bertocci et al. Rsn(f) is equal to |Z|(f) if the impedances of both working electrodes
are equal and much higher than the solution resistance. In the more likely case of
working electrodes with different impedances, the situation is less obvious. According
to Bertocci et al. both the electrode of the higher impedance and that of the lower
impedance can influence the spectral noise resistance [42]. Experimental work
described in chapter 6 shows that a number of factors determine the correlation
between Rsn and the modulus of impedance of both electrodes.
61
Chapter 3
log|Z|
Rn
Rsn (FFT)
Rsn (MEM)
|Z|
log f
Figure 13 Example of noise resistance (Rn), spectral noise resistance (Rsn) obtained with
FFT and MEM and the modulus of impedance (|Z|). Rn is indicated at the sampling
frequency of the time records.
62
3.4 References

MacDonald, D.D., Reflections on the history of electrochemical impedance

spectroscopy, Electrochimica Acta, 51 (2006) 1376-1388.
Warburg, E., Ueber das Verhalten sogenannter unpolarisirbarer Elektroden gegen

Wechselstrom, Annalen der Physik und Chemie, 67 (1899) 493-499.
Kendig, M., J. Scully. Basic Aspects of Electrochemical Impedance Application for the
Life Prediction of Organic Coatings in Metals, Corrosion 46 (1990) 22-29.
Murray, J.N., H.P. Hack, Testing Organic Architectural Coatings in ASTM Synthetic
Seawater Immersion Conditions Using EIS. Corrosion, 48 (1992) 671-685.
Bacon C.R., J.J. Smith, F.M. Rugg, Electrolytic Resistance in Evaluating Protective
Merit of Coatings on Metals. Industrial and Engineering Chemistry, 40 (1948) 161-167.
De Jong, A., Characterization of Coatings inside Food and Beverage Cans, Ph.D.
Thesis, Delft University of Technology, 2000.
Menges, G., W. Schneider, Schutzverhalten un komplexer Widerstand

hochchemikalienbestndiger polymerer Oberflchen-Schutzschichten, Kunstofftechnik,
12 (1974) 343-347.
Mansfeld, F. Use of electrochemical impedance spectroscopy for the study of corrosion

protection by polymer coatings, Journal of Applied Electrochemistry, 25 (1995) 187-202.
10 Weijde, D.H. van der, Impedance Spectroscopy and organic barrier coatings;
(im)possibilities, Ph.D. Thesis, Delft University of Technology, 1996.
11 Amirudin A., D. Thierry, Application of electrochemical impedance spectroscopy to
study the degradation of polymer-coated metals, Progress in Organic Coatings 26 (1995)
1-28.
12 Geenen, F.M., Characterisation of Organic Coatings with Impedance Measurements,
Ph.D. Thesis, Delft University of Technology, 1991.
13 McKubre, M.C.H., D.D. MacDonald, J.R. MacDonald, Measuring Techniques and
Data Analysis In Impedance Spectroscopy - Emphasizing solid materials and systems,
63
Chapter 3
J.R. Macdonald (ed.), book chapter 3, 133-190. John Wiley & Sons, New York, U.S.A.,
1987, ISBN 0471831220.
14 MacDonald, J.R., W.B. Johnson, Fundamentals of Impedance Spectroscopy In
Impedance Spectroscopy - Emphasizing solid materials and systems, J.R. Macdonald
(ed.), book chapter 1, 1-26. John Wiley & Sons, New York, U.S.A., 1987, ISBN
0471831220.
15 MacDonald, J.R., Impedance Spectroscopy: Old problems and new developments,
Electrochimica Acta, 35 (1990) 1483-1492.
16 Lenderink, H.J.W. Filiform Corrosion of coated Aluminium Alloy - a study of
17 Stoynov, Z., Impedance modelling and data processing: Structural and parametrical
estimation, Electrochimica Acta, 35 (1990) 1493-1499.
18 Bonanos, N., B.C.H. Steele, E.P. Butler, W.B. Johnson, W.L. Worrell, D.D. Macdonald,
M.C.H. McKubre, Applications of Impedance Spectroscopy, in: Impedance
Spectroscopy, J. R. Macdonald (ed.), book chapter 4, 191-339. John Wiley & Sons, New
York, U.S.A., 1987, ISBN 0471831220.
19 Hamer, W.J., Polypyrrole Electrochemistry Environmentally friendly corrosion
20 Zhang, X, Cr(VI) and Cr(III)-Based Conversion Coatings on Zinc, Ph.D. Thesis, Delft
21 Walter, G.W., A review of impedance plot methods used for corrosion performance
analysis of painted metals. Corrosion Science, 26 (1986) 681-703.
22 Raidstrick I.D., Theory In Impedance Spectroscopy - Emphasizing solid materials
and systems, J.R. Macdonald (editor), book chapter 2, 27-83. John Wiley & Sons, New
York, U.S.A., 1987, ISBN 0471831220.
23 Westing, E.P.M. van, G.M. Ferrari, J.H.W. de Wit, Determination of coating
performance using Electrochemical Impedance Spectroscopy. Electrochimica Acta, 39
(1994) 899-910.
24 Jorcin, J.-B., M. E. Orazemb, N. Pbre, B. Tribollet, CPE analysis by local
electrochemical impedance spectroscopy, Electrochimica Acta, 51 (2006) 1473-1479.
64

25 Dobbelaar, J.A.L., The use of impedance measurements in coating research The
corrosion behaviour of chromium and iron-chromium alloys, Ph.D. Thesis, Delft
26 Kim, C.-H., S.-I., Pyun, J.-H Kim. An investigation of the capacitance dispersion on the
fractal carbon electrode with edge and basal orientations, Electrochimica Acta 48 (2003)
3455-3463.
27 Pajkossy, T., Impedance spectroscopy at interfaces of metals and aqueous solutions
Surface roughness, CPE and related issues, Solid State Ionics, 176 (2005) 1997-2003.
28 Kerner, Z., T. Pajkossy, On the origin of capacitance dispersion of rough electrodes.
29 Ghr, H., ber Beitrage einzelner Elektrodenprozesse Zur Impedanz, Berichte der
Bunsengesellschaft Physikalische Chemie, 85 (1981) 274-280.
30 Rammelt, U., G. Reinhard, Impedance studies on iron with a vertical decay of
conductivity, Poster, Third International Symposium on Electrochemical Impedance
Spectroscopy, May 7-12, 1995, Belgium.
31 Oh, H.-J., K.-W. Jang, and C.-S. Chi, Impedance Characteristics of Oxide Layers on
Aluminium, Bulletin of the Korean Chemical Society, 20 (1999) 1340-1344.
32 Schiller, C.A., W. Strunz. The evaluation of experimental dielectric data of barrier
coatings by means of different models, Electrochimica Acta, 46 (2001) 3619-3625.
33 Grundmeier, G., W. Schmidt, M. Stratmann, Corrosion protection by organic coatings:
electrochemical mechanism and novel methods of investigation, Electrochimica Acta, 45
(2000) 2515-2533.
34 Hagyard, T., J.R. Williams, Potential of Aluminium in Aqueous Chloride Solutions,
Part1, Transaction of the Faraday Society, 57 (1961) 2288-2294.
35 Iverson W.P., Transient Voltage Changes Produced in Corroding Metals, Journal of the
36 Eden, D.A., D.G. John, J.L. Dawson, US Patent 5139627 (filed 1987, granted 1992).
37 Cottis, R.A. Interpretation of Electrochemical Noise Data. Corrosion 57 (2001) 265-285.
38 Bierwagen G.P., D.E. Tallman, S. Touzain. Electrochemical Noise Methods Applied to
the Study of Organic Coatings and Pretreatments. Paper 380, Corrosion 98, Reviewed
65
Chapter 3
paper for the 1998 Annual Meeting of the Nat. Assoc. Corrosion Eng. (NACE), NACE
Int., San Diego, CA, March 1998.
39 Mansfeld, F., L. T. Han, C. C. Lee, C. Chen, G. Zhang, H. Xiao. Analysis of
electrochemical Impedance an Noise Data for Polymer Coated Metals, Corrosion
Science, 39 (1997) 255-279.
40 Bertocci, U., C. Gabrielli, F. Huet, M. Keddam, P. Rousseau. Noise Resistance Applied
to Corrosion Measurements- II Experimental Tests, Journal of the Electrochemical
Society 144 (1997) 37-43.
41 Reichert, D.L. Electrochemical Noise Measurement for Determining Corrosion Rates
In Electrochemical Noise Measurements for Corrosion Applications, ASTM STP 1277,
J.R. Kearns, J.R. Scully, P.R. Roberge, D.L. Reichert, J.L. Dawson (Editors), American
Society for Testing and Materials, 1996, 79-89.
42 Bertocci, U., C. Gabrielli, F. Huet, M. Keddam, Noise Resistance Applied to Corrosion
Measurements- I Theoretical Analysis, Journal of the Electrochemical Society, 144
(1997) 31-37.
43 Mansfeld, F., Z. Sun, C.H. Hsu, A. Nagiub. Concerning trend removal in
electrochemical noise measurements, Corrosion Science, 43 (2001) 341-352.
44 Bertocci, U., F. Huet, R.P. Nogueira, P. Rousseau, Drift Removal Procedures in the
Analysis of Electrochemical Noise, Corrosion 58 (2002) 337-347.
66
4 Outdoor exposure of coating systems monitored with EIS
4.1 Introduction
As pointed out in chapter 2, the most reliable way of studying the suitability of a
coating for a specific substrate and environment is to actually expose coated substrates
to the environment in which the coating will ultimately be applied. Electrochemical
Impedance Spectroscopy (EIS) has shown to be very useful in the study of degradation
processes of protective coatings [e.g. 1-5].
In this chapter, the results of outdoor exposure of two model coatings, periodically
measured with EIS, will be presented. Measurement results will be analysed and
related to literature sources to obtain insight in degradation processes under natural
conditions.
4.2 Experimental
4.2.1 Coating Material
Two model systems were selected for outdoor exposure: an epoxy coating on gritblasted steel and coil-coated galvanised steel. The epoxy coating was used in two
varieties: a clear coat formulation (binder) and the commercial formulation.
The clear coat formulation only contains surfactant and defoamer as additives. The
binder and additives are similar for the commercial formulation, but this formulation
also contains pigments and other additives and is used as a marine protective coating.
Both epoxy formulations were air-sprayed. The average dry film thickness of the clear
coat was 167 m and that of the commercial paint was 164 m.
The coilcoat material was used in two thicknesses: a 1-layer polyester of 15 m and a
2- layer polyester of 25 m.
Panels with the model systems were mounted at an exposure rack in a marine
environment (Den Helder, The Netherlands). The rack faced south at an angle of 45.
All EIS measurements were performed with computer-controlled Schlumberger

(Solartron) 1250 or 1255 Frequency Response Analyzers (FRA) in combination with
the Schlumberger 1286 Electrochemical Interface (ECI). The instruments were
67
Chapter 4
controlled via the IEEE-bus of the instruments using Windows-based PCs running
ZPlot from Scribner.
The layout of the electrochemical cell is similar to the one described in section 3.2.5.
The cell consists of a PMMA tube (inner diameter 4 cm) glued with silicon sealant to
self-adhesive PVC foil (3M Scotchmark 7900), which is applied to the coating surface.
The square-shaped PVC foil has a circular holiday of in the middle, which leaves 10
cm2 of the coating exposed. The PVC foil was especially selected for its excellent
adhesion, preventing the formation of crevices at the foil-coating interface. In the
electrochemical cell, a platinum mesh serves as the counter electrode (CE). This ringshaped mesh is positioned parallel to the coating surface to achieve an homogeneous
electric field between the CE and the working electrode (WE) [4,6]. A dedicated
reference electrode (RE; Radiometer XR820, saturated Ag/AgCl) is placed in the
centre of the cell. The built-in pseudo-reference electrode was used to by-pass the
interference of the reference electrode at the highest frequencies. This pseudoreference electrode consists of a platinum wire extending to the tip of the Luggin
capillary [4,6-8]. The sensing tip of the RE is placed between the CE and the WE. The
electrolyte consists of a 3% NaCl solution. Evaporation of the electrolyte was
prevented by using a plastic lid.
The electrochemical cell is placed in a grounded Faradaic cage to reduce electrical
noise.
4.2.3 Measurement procedure
Before each measurement, panels exposed outdoors were transferred to the laboratory
and prepared. Sites with pinholes, defects or deviating layer thickness, were avoided in
the selection of measurement spots. The selected measurement spots were gently
cleaned with moist non-abrasive tissues. EIS measurements were performed after 0,
67, 98, 158, 207, 270, 346, 413, 565 and 714 days of exposure.
Prior to each measurement, the open circuit potential (OCP) was measured. All
measurements were performed under potentiostatic control, where the DC potential
relative to RE equals the OCP.
For coatings that are in an excellent condition, it is difficult to obtain a stable value of
the OCP. Coatings that show purely capacitive behaviour, often show an apparent
OCP that steadily drifts. This is an artefact of the measurement equipment. Even the
best electrometer requires some small input current to perform an OCP measurement.
When this current is applied to a capacitive coating accumulation of charge occurs,
resulting in constantly changing OCP.
68
Outdoor exposure of coating systems monitored with EIS
As the coating deteriorates upon exposure to the electrolyte, the system will show
more resistive behaviour. As a result, no charging occurs and the OCP becomes more
stable. However, also for OCP measurements on resistive coatings, the measured
potential is mainly determined by the resistance of the coating. An ohmic drop over
the resistive coating will mask the electrode potential that is determined by corrosion
reactions. In principle a correct value is measured only if the coatings resistance has
sufficiently decreased [4,6,9].
The applied sinusoidal perturbation was 20 mV for the epoxy coatings and 5 mV for
the coilcoatings. These amplitudes were chosen after examination of the impedance
data obtained with different amplitudes.
An example of the 25 m coilcoat is shown Figure 1. For amplitudes between 1 and
200 mV the response is practically independent of the perturbation and the response of
the system can be considered linear. For perturbations of 500 and 1000 mV the
behaviour clearly starts to deviate. The smallest perturbations of 1 and 2 mV contained
some random noise, while for longer measurement times the larger perturbations are
considered damaging. Therefore, an amplitude of 5 mV was chosen for this coating.
107
1-200 mV
500 mV
1000 mV
|Z|
106
105
104
103
102
101
102
103
104
105
Frequency (Hz)
Figure 1
The response of the 25 m coilcoat to increasing perturbations of 1, 2, 5, 10, 20,

50, 100, 200 (solid lines), 500 and 1000 mV (dashed lines).
All measurements started with a perturbation frequency of 100 kHz. The lower
frequency limit depended on the range over which the response was stable. For failing
coatings, the frequency range was extended down to 10 mHz. Each measurement was
performed at a different part of the coated panels to prevent residual effects of
exposure to the electrolyte.
69
Chapter 4
4.3 Results and discussion
Impedance data were analyzed by constructing equivalent circuits in accordance with

the MPEC (most probable equivalent circuit) guidelines of section 3.2.3. The selected
equivalent circuits A to D are shown in Figure 2. The associated impedance elements
are shown in Table 1. For a detailed description of these elements see sections 3.2.4
and 3.2.7.
A
B
Re
Re
Qc
Qc
Rc
C
Re
Qc
Rc
Re
Qdl
Rct
Figure 2
Qc
Rc
Qdl
Rct
Ws
MPECs used for EIS data fitting.
Table 1
Overview of impedance elements of equivalent circuits A to D.
Symbol
Re
Qc
- Y0
- n
Rc
Qdl
- Y0
- n
Rct
Ws
- Rd
- B
- dN
- D
- n
Description
electrolyte resistance
constant phase element of the coating
- admittance constant
- power of the CPE
coating resistance
constant phase element of the metal double layer
- admittance constant
- power of the CPE
charge transfer resistance
Warburg diffusion impedance (finite diffusion length)
- diffusion resistance
1/2
- dN/D
- Nernst diffusion layer thickness
- average diffusion coefficient of electroactive diffusing species
- power of the diffusion impedance
Unit
-2
[cm ]
n
-1
-2
-1
-2
[s cm ]
[-]
-2
[cm ]
[s cm ]
[-]
-2
[cm ]
-2
[cm ]
1/2
[s ]
[m]
2 -1
[m s ]
[-]
All data were analysed using the CNLLS (complex nonlinear least square) software
ZView from Scribner Associates.
The MPECs that best-fitted to the data as of function of exposure duration are shown
in Table 2. Figure 3 shows the Y0, n and Rc values of the coatings as function of the
exposure duration. Table 4 shows an overview of all the values obtained by fitting the
70
equivalent circuits of Table 2 to the data. The chi square (2) of each fit is also given in
this table. This parameter is a sum of differences between measured and expected
(from the equivalent circuit) outcome, each squared and divided by the expectation.
The measured OCPs are shown in Table 3.
All reported values were determined after 24 h of immersion in the electrolyte and are
normalized to an area of 1 cm2.
71
Chapter 4
Table 2 MPECs for the model coatings as a function of exposure time. The corresponding
panel codes are shown between brackets.
15 m
Coilcoat
[HRCC15]
B
25 m
Coilcoat
[HRCC25]
A
Epoxy
clear coat
[HRECC]
A
Epoxy commercial
formulation
[HRECS]
B
67
98
158
207
270
346
413
565
714
C**
Outdoor Exposure [days]
* No proper measurement data could be obtained for this sample.

** No coating parameters (Qc) obtained for this measurement.
Table 3
Open circuit potential (OCP) vs. saturated Ag/AgCl as a function of the exposure
duration. Shown values are determined after 24 h of immersion in the electrolyte.
15 m
Coilcoat
[HRCC15]
25 m
Coilcoat
[HRCC25]
Epoxy
clear coat
[HRECC]
Epoxy commercial
formulation
[HRECS]
-1.062
-0.323
-0.295
-0.490
67
-0.999
-0.968
-0.447
-0.486
98
-0.913
-0.913
-0.211
-0.500
158
-0.788
-0.472
-0.112
207
-1.028
-0.954
-0.542
-0.498
270
-1.024
-1.011
-0.304
-0.148
346
-1.019
-0.955
-0.112
-0.512
413
-1.024
-1.006
**
**
565
-1.025
-1.011
-0.142
-0.471
714
-1.029
-1.033
-0.573
-0.524
* No proper measurement data could be obtained for this sample.
** No reasonable OCP could be obtained due to charging (see section 4.2.3)
72

HRCC15
1.00E-04
HRCC25
1.00E-05
HRECC
HRECS
Qc-Y0 [sn-1cm-2]
1.00E-06
1.00E-07
1.00E-08
1.00E-09
1.00E-10
1.00E-11
0
100
200
300
400
500
600
700
800
700
800
Outdoor exposure [Days]
1.00
0.80
Qc-n [-]
0.60
0.40
HRCC15
HRCC25
0.20
HRECC
HRECS
0.00
0
100
200
300
400
500
600
1.00E+10
Rc [cm-2]
1.00E+08
1.00E+06
1.00E+04
HRCC15
HRCC25
1.00E+02
HRECC
HRECS
1.00E+00
0
100
200
300
400
500
600
700
800
Figure 3
Qc-Y0 (top), Qc-n (middle) and Rc (bottom) as a function of the exposure duration.
Shown values are determined after 24 h of immersion in the electrolyte.
73
Chapter 4
Table 4
Outdoor
exposure
[d]
Overview of the values obtained from data fitting using the equivalent circuits of
Table 2.
Qc
Qdl
Y0
Rc
Y0
W
n
Rct
Rd
HRCC15
0
2.44E-10
0.93
8.33E+04
5.49E-04
67
6.58E-10
0.99
1.54E+05
3.24E-05
98
1.66E-09
0.94
2.58E+05
7.18E-05
207
4.08E-09
0.88
4.59E+03
3.46E-07
0.77
6.15E+04
1.62E-10
270
1.72E-08
0.78
2.11E+04
1.51E-07
0.63
9.49E+04
1.09E-04
346
2.70E-08
0.88
1.64E+03
8.20E-08
0.86
1.06E+05
8.04E-04
413
9.64E-07
0.53
1.83E+03
4.19E-07
0.78
9.72E+03
9.61E-05
565
1.46E-05
0.42
1.53E+03
6.60E-07
0.82
3.21E+03
6.80E-05
2.12E-05
0.57
1.01E+04
1.02E-04
158
714
HRCC25
0
1.87E-10
0.97
2.42E-03
67
3.35E-10
0.98
1.38E+07
4.06E-03
98
3.37E-10
0.98
2.21E+06
2.54E-03
158
3.54E-10
0.97
6.75E+06
8.30E-04
207
4.00E-10
0.98
4.19E+06
7.10E-04
270
3.76E-10
0.98
1.76E+06
9.74E-04
346
4.79E-10
0.97
1.25E+07
1.49E-03
413
3.92E-10
0.97
3.17E+06
1.24E-03
565
2.08E-10
0.97
6.67E+06
5.96E-04
714
6.50E-10
0.94
8.26E+05
1.86E-04
HRECC
74
3.25E-11
0.99
67
4.47E-11
0.98
2.57E+08
3.80E-03
2.70E-04
98
4.92E-11
0.98
5.61E+07
3.44E-03
158
4.81E-11
0.99
1.02E+06
207
1.55E-10
0.94
4.42E+04
270
3.04E-11
0.99
5.54E+07
7.75E-04
346
2.73E-11
1.00
6.38E+07
2.26E-03
413
1.76E-11
0.98
3.35E+07
3.88E-04
565
2.31E-11
1.00
8.53E+07
714
7.00E-11
0.96
5.70E+06
5.51E-05
2.24E-06
0.49
2.23E+04
1.97E+05
1.72
0.47
1.02E-04
1.81E-03
4.57E-08
0.53
1.71E+06
1.24E+07
32.12
0.45
2.24E-03

Table 4 (continued) Overview of the values obtained from data fitting using the equivalent
circuits of Table 2.
Outdoor
exposure
[d]
Qc
Qdl
Y0
Y0
Rc
Rct
Rd
1.30E-11
0.94
3.91E+08
4.08E-03
67
1.98E-11
0.97
1.32E+07
1.19E-03
HRECS
98
3.78E-11
1.00
4.62E+06
2.40E-03
158
2.67E-11
0.98
3.13E+07
4.66E-04
207
4.33E-11
0.98
270
4.25E-11
1.00
1.60E+07
5.28E-04
346
4.38E-11
0.97
7.51E+06
7.75E-05
413
2.86E-11
0.96
1.18E+06
565
4.99E-10
0.86
4.31E+05
1.15E-06
0.44
9.41E+05
714
5.49E-10
0.86
1.35E+05
2.23E-06
0.38
6.54E+05
6.46E-05
7.55E-05
3.18E-04
1.11E+06
131.6
0.52
2.89E-04
For the epoxy coatings, the OCP oscillated between the -100 and -600 mV indicating
that the true potential of the steel electrode (approximately -650 mV vs. SCE; saturated
calomel electrode [10]) was masked by an ohmic drop over the resistive coatings.
The OCP of the 25 m coilcoat dropped from about -300 mV to -1000 mV, while the
15 m coilcoat remained close to the OCP of bare zinc (approximately -1050 mV vs.
SCE [10]) from the start of exposure.
The time-dependence of the impedance parameters shown in Figure 3 and Table 4
shows quite some scatter. This behaviour is not uncommon for long-term EIS studies
(e.g. see 7,11,12,13,14).
For this study, each measurement was performed at a different part of the panel. Due
to the inevitable sample-to-sample variation, an additional source of scatter was
introduced.
Still, valuable mechanistic information can be derived from the measurement results as
will be discussed in the remainder of this section.
The simplified Randles cell [15] is one of the simplest and most common equivalent
circuits for undamaged coatings [16-28]*. In many cases, the coating capacitance is
replaced by a CPE, resulting in equivalent circuit B. Prior to outdoor exposure (0
days), the 15 m coilcoat and the epoxy commercial formulation could be fitted to this
The experimental work of Randles concerned rapid electrode reactions of mercury in an aqueous
solution [15]. Consequently, the physical interpretation of the elements present in the simplified
Randles cell is different for organic coatings.
75
Chapter 4
circuit. The dielectric behaviour of the coating is expressed by Qc while the Rc element
is related to the resistance that charge carriers (ions) encounter in the coating [4,5].
For the 25 m coilcoat and the epoxy clear coat formulation the inclusion of the Rc
element in the circuit has a negative influence on the goodness of fit. Apparently, the
contribution of the charge carriers is so small for these coatings, that their presence in
the form of a resistor can be ignored at this stage. With respect to the guidelines of
section 3.2.3 for fitting data, these coating varieties were fitted to equivalent circuit A
instead. This circuit is valid for a purely capacitive coating.
It should be mentioned that this circuit violates one of general conditions for a
successful Kramers-Kronig transform (see section 3.2.9). The impedance of the circuit
should be finite-valued at 0 and [29]. However, the impedance of a
capacitor is infinite at = 0. Despite this violation, circuit A is used as no sensible
value for Rc (by using circuit B) can be obtained at this point.
Now, a unified model for coating degradation presented by Nguyen and Hubbard [30]
will be introduced. Together with other sources from literature, this model will be
related to the EIS measurement results.
The model by Nguyen and Hubbard is proposed for the degradation of a defect-free
organic coating on a steel substrate exposed to a pH-neutral electrolytic environment.
For coated materials, water uptake is often mentioned as the first step of the
degradation process. This process is believed to eventually cause the formation of
conductive pathways in the coating, allowing ions to reach the substrate surface.
Nguyen and Hubbard assume that the formation of conductive pathways in an organic
coating during water uptake (Figure 4: step 1. Note: the consecutive steps are indicated
by the circled numbers) is due to an attack by water in the hydrophilic regions in the
film, followed by interconnections of these regions.
In early work of Bacon et al. [31] and Mayne [32], high ionic resistance of coatings
was associated with good corrosion protection. The presumed formation of conductive
pathways is supported by later work of various workers [33-38].
Modes of ion transport through coatings were addressed by Kinsella et al. [33]. They
concluded that coatings exhibit two types of electrical behaviour when immersed in
ionic solutions.
Even visibly non-damaged films show an onset of corrosion after a certain time. However, for
practical systems, the most severe corrosion processes start at local imperfections within the coating
system e.g. at cut edges, defects generated during the lifetime of the coating (scratches, stone
chipping), or due to poor pre-treatment (resulting in blistering, osmosis, coating detachment).
76
H2O, O2
b
H2O, O2, NaCl
hydrophylic
region
coating
Fe2+
anode
H2O, O2
NaCl
anode
e5
Na+, OH-
e-
cathode
Fe2+
e-
Na+, OH-
cathode
cathode
Figure 4
Na+, OH-
OH-
e-
O2 NaCl corrosion products

O2
H2O
O2
corrosion products
Fe2+
conductive pathway
steel
NaCl
anode
cathode
Model for the degradation of an intact organic coating on steel in a neutral NaCl
solution [adapted from 30].
The first type, dubbed I-type, refers to coating areas that have a high resistance as
there are no continuous aqueous pathways through the coating. Another interesting
property, from which the name I-type coating originates, is the inverse relation
between the resistance of the coating and that of the electrolyte [33]. This behaviour
was previously observed by Bacon et al. [31].
Water ingress or egress is controlled largely by the relative thermodynamic activities
of water within the polymer and within the electrolyte.
Kinsella has experimentally shown that the resistance of I-type coatings depends on
the water activity of the external solution and is controlled of the amount of water
taken up [33].
It is likely that charge transport of ions occurs through the polymer matrix by a
random-walk process. The transport process involves the migration of ions through the
matrix into dynamic free volumes caused by thermal motion of polymer segments
[37].
For the second type of behaviour, to which the term D-type has been applied, it is
believed that aqueous pathways through the coating exist and the aqueous phase within
these pathways equilibrates with the bulk electrolyte. Therefore, this type of behaviour
was termed direct. The pathways should not be regarded as pores, as no pores or
The term 'water activity' (aw) describes the amount of water available for hydration of materials. By
definition an aw value of 1 indicates pure distilled water, whereas aw = 0 indicates the absence of 'free'
water molecules Water migrates from areas of high aw to areas of low aw.
77
Chapter 4
capillaries can be detected visually for coatings displaying D-type behaviour

[34,36,38].
Instead, D-type areas should be considered as areas with a different structure, probably
having considerably reduced cross-linking compared to I-type areas
[5,33,34,36,39,40,42,46]. Mayne and Scantlebury performed micro-hardness
measurements and correlated these measurements with regions of I and D type
conduction and found that D-type areas were significantly softer than I-type areas.
This behaviour was indirectly related to inhomogeneities in cross-link density [34].
Not much later, Mayne and Mills showed a good correlation between the local
resistance of the coating and the distribution of underfilm corrosion [35].
Tan et al. measured the inhomogeneity of organic coatings directly using a wire beam
electrode [43].
Thus different sections of the same coating may be either I-type or D-type areas. In
fact, coatings are intrinsically heterogeneous [41] and have been found to contain
degradation-susceptible regions [34,35,38,42,45].
Global ion transport in protective coatings has also been confirmed using radiotracer
techniques. Nmeth et al., followed the diffusion of labelled chloride and sulphate ions
through an alkyd coating with corrosion inhibiting pigments. It was shown that ion
transport processes could be studied with this technique. It will also enable comparison
of different coatings and ion types with respect to diffusion [44].
Following the pathway formation, the next step in the degradation of coated steel is the
transport (Figure 4: step 2) of water, oxygen and ions from the environment to the
metal surface. Although both water and ions can penetrate films, ions tend to penetrate
films only locally by formation of conductive pathways [30].
After 67 days of outdoor exposure, the fit results of the 25 m coilcoat and the epoxy
clear coat formulation were also improved by the addition of the Rc element (resulting
in circuit B). A clear example of this change is shown in Figure 5. The left Nyquist
plot shows capacitive behaviour (straight line) of the epoxy clear coat before exposure
(0 days). These data were analysed using circuit A. However, the right Nyquist plot
(after 158 days of exposure) clearly shows that, towards lower frequencies, the curve
bends to the resistive axis. Therefore, these data were analysed using circuit B.
This supports the idea that the conductive pathways have formed [22,48]. Hence, the
physical model for all coating varieties has changed from Figure 4a to Figure 4b.
Process where covalent bonds are linking one polymer chain to another.
78
-400000
-750000
HRECC 158 days (24 h)

Fit result
HRECC 0 days (24 h)

Fit result
-300000
Z''
Z''
-500000
-200000
-250000
-100000
0
0
250000
500000
Z'
Figure 5
750000
100000
200000
300000
400000
Z'
Experimental () and simulated () Nyquist plots of the epoxy clear coat

formulation before exposure (left) and after 158 days of exposure (right).
After 207 days of exposure, the epoxy commercial formulation was fitted using circuit
A. Once again, the error of the Rc value became unacceptably large when circuit B was
used. The 15 m coilcoat and the epoxy clear coat could not be fitted neither to
equivalent circuit A or B.
For the 15 m coilcoat circuit C provided the best results [13,18,20,21,25,47,49-57].
The Nyquist and Bode plot after 207 days of exposure are shown in Figure 6. Two arcs
can be observed in the Nyquist plot. The first high-frequency arc is characterized by
the coating properties Qc and Rc, similar to circuit B. The larger low-frequency arc is
related to processes taking place at the electrode surface and is characterised by the
added elements Qdl and Rct. Qdl represents the double layer capacitance at the metalelectrolyte interface, while Rct represents the charge transfer resistance of the corrosion
reaction at the bottom of the pores [58].
This stage is represented by step 3 and 4 in Figure 4. Anodes develop on the metal
surface at the base of the pathways while cathodes develop under the coating at the
periphery of the pathways [30].
For the epoxy clear coat equivalent circuit D provided the best results when fitted to
the data after 207 days of exposure [5,8,27,47,48,58,59]. In this circuit, the element Ws
(finite diffusion length) is added to account for diffusion of reacting species [47,58].
The Nyquist plot for this measurement is shown in Figure 7.
79
Chapter 4
-10000
HRCC15 207 days (24 h)
Fit result
HRCC15 207 days (24 h)

Fit result
|Z|
104
-7500
Z''
103
103
104
105
Frequency (Hz)
-5000
-70
-60
theta
-2500
-50
-40
-30
-20
-10
2500
5000
7500
103
10000
Z'
Figure 6
104
105
Frequency (Hz)
Experimental () and simulated () Nyquist and Bode plot of the 15 m coilcoat

after 207 days of exposure. Note: ZView uses theta to indicate the phase angle
instead of phi.
Theoretically, circuit D should show three arcs in the Nyquist diagram (also see
section 3.2.6, Figure 7). The high frequency arc is related to the coating itself; an
intermediate arc is related to the steel-electrolyte interface; and the low frequency arc
is related to diffusion processes.
According to Feliu et al. and Feliu Jr. et al. [47,48] the practical situation is different.
The intermediate arc overlaps with the end portion of the high frequency arc and the
initial portion of the low frequency arc. Occasionally, the intermediate arc is
completely masked, as is also the case in Figure 7 (left). The simulated plots for
different B values -while keeping the other fit parameters unchanged- reveal that the
intermediate arc is indeed obscured by the diffusion tail (Figure 7, right).
The diffusion coefficient of reacting species can be calculated with [60-62]:
n
ZWs = Rd tanh ( jB ) / ( jB ) with B =
N
D
(1)
Where:
Rd = limiting diffusion (mass transfer) resistance [m-2]
N = Nernst diffusion layer thickness [m]
D = average diffusion coefficient of electroactive diffusing species [m2s-1]
80
-100000
-100000
HRECC 207 days (24 h)
Fit result
Fit result
B = 1E1 - 1E5
-75000
-50000
Z''
Z''
-75000
-50000
Ef f ect of
increasing
B value
(QcRc)
-25000
-25000
(QdlRct)
0
0
25000
50000
75000
100000
Z'
Figure 7
25000
50000
75000
100000
Z'
Left: Experimental () and simulated () Nyquist plot of the epoxy clear coat after
207 days of exposure. Right: Simulations with B values of 10, 100, 1000, 10,000,
100,000 s.
An assumption has to be made concerning the effective diffusion layer thickness. It

has been suggested [64] to substitute the average coating thickness, which seems
reasonable. Table 5 lists all the calculated diffusion coefficients together with the
associated parameters.
Table 5
Calculated effective diffusion coefficients.

B [s ]
[cm]
D [cm s ]
207
1.72
0.0122
8.66E-05
714
32.12
0.0122
4.63E-06
131.60
0.0124
1.16E-06
-1
HRECC
HRECS
714
For measurements by Bucharsky et al. the diffusion impedance was related to the
diffusion of oxygen used in the cathodic partial reaction [65]. In addition, Amirudin et
al. state that if an equivalent circuit contains only one diffusion element, it is usually
attributed to the cathodic oxygen diffusion [5].
This seems reasonable in the light of findings by Thomas. In his work, Thomas
concluded that the oxygen consumption rate of uncoated steel is comparable to its
transmission rate through a poor barrier coating and is far higher than the transmission
rate through a good barrier coating [66].
81
Chapter 4
In accordance with the guidelines of section 3.2.3, circuit B could be fitted to the
results of the epoxy clear coat again from 346 to 565 days of exposure. After 565 days
the behaviour of the epoxy commercial formulation was best described by circuit C.
After 714 days, the behaviour of both epoxy formulations was best described by circuit
D.
Depending upon the substrate, blisters may grow due to either cathodic delamination
or anodic undermining.
In case of steel, cathodic delamination is the most likely mechanism [63]. The regions
underneath the coatings where cathodic reactions occur, serve as the initiation sites for
the delamination or blistering. The strong alkalinity at the cathodic sites has been
attributed to the migration of cations (Figure 3, step 5) from the environment to the
cathodic sites, neutralizing the negative charge of the hydroxide ions. The alkalinity of
the solution at the delamination front causes the coating to disbond (Figure 3, step 6).
The capacitance of the electrochemical double layer is generally considered as a
measure for the area over which the coating has disbonded. However, as mentioned in
section 3.2.7, the capacitance is more a measure of the electro-active area than the
delaminated area [5]. This implies that the double layer capacitance also depends on
the electrochemical state of the surface (i.e. active or passive).
While delamination progresses, sodium ions migrate along the coating/metal interface
from the defect to the cathodic sites. This transport process is the rate determining step
[30].
For galvanised steel, the situation depends on whether a zinc layer is still present at the
defect. For an intact zinc layer with a defective coating the coating delaminates due to
cathodic delamination. If the zinc layer has disappeared at the defect, an anodic
delamination mechanism occurs. Zinc is dissolved under the coating, while the steel
surface is cathodically protected [67].
The physical model of a coating that can be fitted with equivalent circuits C and D is
represented in Figure 4d.
The alkaline products from the cathodic reaction reduce the water activity and create
osmotic pressure (Figure 4, step 7) between the environment and the coating/steel
interface at the blisters. This osmotic pressure is probably the principle mechanism for
the growth of cathodic blisters (Figure 4, step 8) [30].
82

4.4 Conclusions
Electrochemical impedance spectroscopy appears to be very useful for the study of

coating degradation mechanisms for a number of reasons.
In order to draw useful conclusions from EIS data, the construction of equivalent
circuits must be carried out with care. By fitting the MPEC to the measurement data,
important parameters such as coating resistance and capacitance, the charge transfer
resistance and the electrochemical double layer capacitance can be determined
quantitatively.
Especially at the early stages of coating degradation, when the coating shows no visual
signs of degradation, EIS clearly has an advantage as various processes related to
coating degradation (e.g. water uptake, formation of conductive pathways) can already
be identified with this technique, yet under non-accelerated conditions.
Despite the fact that the time dependence of the impedance parameters showed quite
some scatter, still valuable mechanistic information has been derived from the
measurement results. Together with other sources from literature, measurement results
were related to the unified model for coating degradation presented by Nguyen and
Hubbard.
83
Chapter 4
4.5 References
1
Feliu, S., J. C. Galvn, M. Morcillo, An Interpretation of Electrical Impedance Diagrams

for Painted Galvanized Steel, Progress in Organic Coatings, 17 (1989), 143-153.
Mansfeld, F., C.H. Tsai, Determination of Coating Deterioration with EIS, I. Basic
Relationships, Corrosion, 47 (1991), 958-963.
Lin, C., T. Nguyen, M.E. McKnight, Relation between AC impedance data and
degradation of coated steel. 1. Effects of surface roughness and contamination on the
corrosion behavior of epoxy-coated steel, Progress in Organic Coatings, 20 (1992), 169185.

Amirudin, A., D. Thierry. Application of electrochemical impedance spectroscopy to

study the degradation of polymer-coated metals, Progress in Organic Coatings, 26
(1995), 1-28.
Geenen, F.M., Characterisation of Organic Coatings with Impedance Measurements,

Ph.D. Thesis, Delft University of Technology, 1991.
Mansfeld, F., Use of electrochemical impedance spectroscopy for the study of corrosion
protection by polymer coatings, Journal of Applied Electrochemistry, 25 (1995), 187202.
Weijde, D.H. van der van der , E.P.M. van Westing and J.H.W. de Wit, Electrochemical
techniques for delamination studies, Corrosion Science, 36 (1994), 643-652.
Loveday, D., P. Peterson, B. Rodgers, Evaluation of Organic Coatings with

Electrochemical Impedance Spectroscopy-Part 2: Application of EIS to Coatings, JCT
CoatingsTech, October 2004, 88-93.
10 Metals Handbook, Volume 13 - Corrosion, ASM International, 9th edition, ISBN

0871700190.
11 Scully, J.R., S.T. Hensley, Lifetime Prediction for Organic Coatings on Steel and
Magnesium Alloy Using Electrochemical Impedance Methods, Corrosion, 50 (1994),
705-716.
12 Gonzlez-Garca, Y., S. Gonzlez, R.M. Souto, Electrochemical and structural
properties of a polyurethane coating on steel substrates for corrosion protection,
Corrosion Science, 49 (2007), 3514-3526.
84

13 Suay, J.J., M.T. Rodrguez, K.A. Razzaq, J.J. Carpio, J.J. Saura, The evaluation of
anticorrosive automotive epoxy coatings by means of electrochemical impedance
spectroscopy, Progress in Organic Coatings, 46 (2003), 121-129.
14 Di Sarli, A.R., E.E. Schwiderke, J.J. Podesta, Evaluation of anticorrosive paint binders
by means of A.C. techniques Influence of the coat thickness, 10th ICMC, Madras, India,
7-11 November 1987, Proceedings, 1091-1100.
15 Randles, J.E.B., Kinetics of rapid electrode reactions, Discussions of the Faraday
Society, 1 (1947), 11-19.
16 Babi, R., M. Metiko-Hukovi and H. Radovi, The study of coal tar epoxy protective
coatings by impedance spectroscopy, Progress in Organic Coatings, 23 (1994), 275-286.
17 Castela, A.S. and A.M. Simes, An impedance model for the estimation of water
absorption in organic coatings. Part I: A linear dielectric mixture equation, Corrosion
Science, 45 (2003), 1631-1646.
18 De Jong, A., Characterization of Coatings inside Food and Beverage Cans, Ph.D.
Thesis, Delft University of Technology, 2000.
19 Deflorian, F., L. Fedrizzi and S. Rossi, Electrochemical Impedance Spectroscopy and
Fourier transform infrared of natural and accelerated weathering of organic coatings,
Corrosion, 54 (1998), 598.
20 Dehri, I., R. L. Howard and S.B. Lyon, Local electrochemical impedance at the cut-edge
of coil-coated galvanized steel after corrosion testing, Corrosion Science 41 (1999), 141154.
21 Grundmeier, G., W. Schmidt and M. Stratmann, Corrosion protection by organic
coatings: electrochemical mechanism and novel methods of investigation, Electrochimica
Acta 45 (2000), 2515-2533.
22 Jttner, K., Electrochemical impedance spectroscopy (EIS) of corrosion processes on
inhomogeneous surfaces, Electrochimica Acta 35 (1990), 1501-1508.
23 Kouloumbi, N., G.M. Tsangaris, A. Skordos and S. Kyvelidis, Evaluation of the
behaviour of particulate polymeric coatings in a corrosive environment. Influence of the
concentration of metal particles, Progress in Organic Coatings 28 (1996), 117-124.
24 MacQueen, R.C., R.R. Miron and R.D. Granata, Method for corrosion inhibitor
mechanism studies in epoxy coated aluminium, Journal of Coatings Technology 68
(1996), 75-82.
85
Chapter 4
25 Marchebois, H., C. Savall, J. Bernard and S. Touzain, Electrochemical behavior of zincrich powder coatings in artificial sea water, Electrochimica Acta, 49 (2004), 2945-2954.
26 Weijde, D.H. van der, E.P.M. van Westing, G.M. Ferrari and J.H.W. de Wit,
Application of Impedance Spectroscopy during the Lifetime of Organic Coatings, in:
Organic coatings for Corrosion Control. Ed. G.P. Bierwagen, Oxford University Press,
Washington DC (1998), 45-57. ISBN 084123549X.
27 Walter, G.W., Application of impedance measurements to study performance of painted
metals in aggressive solutions, Journal of Electroanalytical Chemistry 118 (1981), 259273.
28 Wu, X., W. Zhang and H., Yu, Assessing nonlinearity in impedance models: the model
characteristics of an equivalent circuit with a CPE, Journal of Electroanalytical
Chemistry 398 (1995), 1-4.
29 Bonanos, N., B.C.H. Steele, E.P. Butler, W.B. Johnson, W.L. Worrell, D.D. Macdonald,
M.C.H. McKubre, Applications of Impedance Spectroscopy, in: Impedance
Spectroscopy, J. R. Macdonald (ed.), book chapter 4, 191-339. John Wiley & Sons, New
York, U.S.A., 1987, ISBN 0471831220.
organic coatings on steel in a neutral electrolyte, Journal of Coatings Technology 68
(1996), 45-56.
31 Bacon C.R., J.J. Smith, F.M. Rugg, Electrolytic Resistance in Evaluating Protective
Merit of Coatings on Metals. Industrial and Engineering Chemistry, 40 (1948) 161-167.
32 Mayne, J.E.O., The Mechanism of the Inhibition of the Corrosion of Iron and Steel by
Means of Paint, Official Digest, February 1952, 127-136.
33 Kinsella, E.M., J.E.O. Mayne, Ionic conduction in polymer films, I, Influence of
electrolyte on resistance, British Polymer Journal, 1 (1969), 173-176.
34 Mayne, J.E.O., J.D. Scantlebury, Ionic conduction in polymer films, II, Inhomogeneous
structure of varnish films, 2 (1970), 240-243.
35 Mayne, J.E.O., D.J. Mills, The effect of the substrate on the electrical resistance of
polymer films, Journal of the Oil and Colour Chemists' Association, 58 (1975), 155-159.
36 Mills, D.J., J.E.O., Mayne, The Inhomogeneous Nature of Polymer Films and its Effect
on Resistance Inhibition, In: Corrosion Control by Organic Coatings, H. Leidheiser Ed.
Houston, Texas, NACE 1981, 12-17.
86

37 Leidheiser, H., R.D. Granata, Ion transport through protective polymeric coatings
exposed to an aqueous phase, IBM journal of research and development, 32 (1988), 582590.
38 Wu, C.-L., X.-J. Zhou, Y.-J. Tan, A study on the electrochemical inhomogeneity of
organic coatings, Progress in organic coatings, 25 (1995), 379-389.
39 Kendig, M., J. Scully, Basic Aspects of Electrochemical Impedance Application for the
Life Prediction of Organic Coatings in Metals, Corrosion, 46 (1990), 22-29.
40 Walter, G.W., A review of impedance plot methods used for corrosion performance
analysis of painted metals, Corrosion Science, 26 (1986), 681-703.
41 Corti, H., R. Fernandez-Prini, D. Gomez, Protective organic coatings: Membrane
Properties an Performance, Progress in Organic Coatings, 10 (1982), 5-23.
42 Keene, L.T., M.J. Vasquez, C.R. Clayton, G.P. Halada, Failure of navy coating systems
2: failure pathways of artificially weathered navy coating systems applied to chromate
conversion coated AA2024-T3 substrates, Progress in Organic Coatings, 52 (2005), 187195.
43 Tan, Y.S., S. Yu, The effects of imhomogeneity in organic coatings on electrochemical
measurements using a wire beam electrode - Part II, Progress in Organic Coatings, 19
(1991), 257-263.
44 Nmeth, Z., L. Erdei, A. Kolics, A new approach to studying ion transport in corrosion
protective coatings using an in situ radiotracer method, Corrosion Science, 37 (1995),
1163-1166.
45 Fernandez-Prini, R., H. Corti, Epoxy Coal Tar Films: Membrane Properties and Film
Deterioration, Journal of Coatings Technology, 49 (1977), 62-68.
46 Dicky, R.A., An Introduction to Corrosion Control by Organic Coatings, ACS
Symposium Series, 285, (1985), 773-799.
47 Feliu, S., J. C. Galvn and M. Morcillo, An interpretation of electrical impedance
diagrams for painted galvanized steel, Progress in Organic Coatings 17 (1989), 143-153.
48 Feliu Jr., S. , M. Morcillo, S. Feliu, The reproducibility of impedance parameters
obtained for painted specimens, Progress in Organic Coatings, 25 (1995), 365-377.
87
Chapter 4
49 Bierwagen, G.P., L. He, J. Li, L. Ellingson and D.E. Tallman, Studies of a new
accelerated evaluation method for coating corrosion resistance - thermal cycling testing,
Progress in Organic coatings, 39 (2000), 67-78.
50 DeFlorian F., I. Felhosi, Electrochemical impedance study of environmentally friendly
pigments in organic coatings, Corrosion, 59 (2003), 112-120.
51 Fedrizzi, L., F. Deflorian, G. Boni, P.L. Bonora and E. Pasini, EIS study of
environmentally friendly coil coating performances, Progress in Organic Coatings, 29
(1996), 89-96.
52 Kendig, M., S. Jeanjaquet, R. Brown and F. Thomas, Rapid Electrochemical Assessment
of Paint (REAP), Journal of Coatings Technology, 68 (1996), 39.
53 Kousik, G., S. Pitchumani and N.G. Renganathan, Electrochemical characterization of
polythiophene-coated steel, Progress in Organic Coatings, 43 (2001), 286-291.
54 Lori, L., A. Tamba, F. Deflorian, L. Fedrizzi and P.L. Bonora, Stainless steel as new
substrate for coil coating, Progress in Organic Coatings, 27 (1996), 17-23.
55 Mansfeld, F., The use of electrochemical impedance spectroscopy for the evaluation of
the properties of passive films and protective coatings, Models in Chemistry, 132 (1995),
619-631.
56 McIntyre, J.M., Ha Q. Pham, Electrochemical impedance spectroscopy; a tool for
organic coatings optimizations, Progress in Organic Coatings, 27 (1996), 201-207.
57 Wall, F.D., M. A. Martinez, N.A. Missert, R.G. Copeland, A.C. Kilgo, Characterizing
corrosion behavior under atmospheric conditions using electrochemical techniques,
58 Miszczyk, A., H. Szalinska, Laboratory evaluation of epoxy coatings with an adhesion
promoter by impedance Progress in Organic Coatings, Progress in Organic Coatings 25
(1995), 357-363.
59 Barranco, V., S. Feliu, and Jr. S. Feliu, EIS study of the corrosion behaviour of zincbased coatings on steel in quiescent 3% NaCl solution. Part 2: coatings covered with an
inhibitor-containing lacquer, Corrosion Science, 46 (2004), 2221-2240.
60 Raidstrick I.D. Theory In Impedance Spectroscopy - Emphasizing solid materials and
systems, J.R. Macdonald (editor), book chapter 2, 27-83. John Wiley & Sons, New York,
U.S.A., 1987, ISBN 0471831220.
88

61 MacDonald, J.R., Impedance Spectroscopy: Old problems and new developments,
62 Zhang, X, Cr(VI) and Cr(III)-Based Conversion Coatings on Zinc, Ph.D. Thesis, Delft
63 Lenderink, H.J.W., Filiform Corrosion of coated Aluminium Alloy- a study of
64 Mansfeld, F., L.T. Han, C.C. Lee, C. Chen, G. Zhang, H. XIAO, Analysis of
Electrochemical Impedance and Noise Data for Polymer Coated Metals, Corrosion
Science, 39 (1997), 255-279.
65 Bucharsky, E.C., E.B. Castro, S.G. Real, An Electrochemical Impedance Spectroscopy
Analysis of Protective Behaviour of Final Coatings on Naval Steel, The Journal of
Corrosion Science and Engineering, 2, paper 19 (1999).
66 Thomas, N.L., Coatings for rusty steel, where are we now?, Surface Coatings
International Journal of the Oil and Colour Chemists' Association, 74 (1991) 83-97.
67 Frberth, W., M. Stratmann, Investigation of the delamination of polymer films from
galvanized steel with the Scanning Kelvinprobe, Fresenius Journal of Analytical
Chemistry, 353 (1995), 337-341.
89
90
5 Pre-qualification of protective coatings using EIS
5.1 Introduction
Especially for industrial applications, coatings are required to provide excellent

corrosion protection for a long time. In case of extreme conditions, it is imperative to
select a coating system that is most suitable for site-specific conditions. However,
there is no generally accepted test method for selection of protective coatings [1].
Moreover, in the face of the growing number of new (VOC-compliant) coating
products coming onto the market, coating manufacturers can no longer rely on the
extensive track records of their established products to convince customers of their
suitability [2,3].
Therefore, it has become increasingly important to develop methods for the rapid
evaluation of the corrosion protective properties of these coatings.
As discussed in section 2.4.2., there has been considerable effort to develop evaluation
methods for organic coatings that are numerical, reproducible and accurate. It was
found that -in particular- electrochemical impedance spectroscopy (EIS), has
demonstrated to be very useful in this respect. Since EIS not only provides results in a
short time but also determines coating damage prior to its visual manifestation, it can
be used as a tool to enable rapid determination of relevant coating parameters [4-11].
The protective nature of typical barrier coatings is not solely due to a reduced transport
of oxygen and water to the metal/coating interface, but rather to the prevention of ionic
current flow through the coating. It has been shown that even the best barrier coatings
have a water vapour transmission rate far in excess of that required to maintain
corrosion on uncoated steel.
For oxygen, the consumption rate of bare steel is of the same order as its transmission
rate through a poor barrier coating and is far higher than the transmission rate through
good barrier coatings [12].
Ion solubility within protective coatings is typically very small [13,14] and coatings
inhibit corrosion by virtue of their high resistance.
The pioneering work of Bacon, Smith, and Rugg determined a direct correlation
between coating resistances and their ability of the coating to protect the steel substrate
from corrosion [15]. After measuring the electrical resistances of over 300 coating
systems, three general classifications, based on a sustained resistance value, were
established during this investigation: good, fair and poor. It was found that in
91
Chapter 5
every case good protection was obtained at resistances higher than 108 cm-2 and
poor protection was obtained at resistances lower than 106 cm-2.
Although this work identified the relationship between the ionic resistance of the
coatings and their ability to protect the substrate, the permeability of other corrosive
agents other than ionic species was not considered. Therefore, the results were
presented as providing a measure of the likely protection that could be expected from a
particular coating rather than proposing a mechanism for the protective action of
organic coatings [16].
With EIS, the protective properties of coatings can be quantified by determining a
range of parameters that are related to corrosion protection.
comparison of measurement results obtained by different groups.
Therefore, in this chapter, we will discuss the results of EIS measurements performed
with more than 100 industrial coatings using a benchmark measurement protocol.
After presenting the overall results in section 5.3.1 some characteristic results will be
discussed in more detail in section 5.3.2 to 5.3.6.
5.2 Measurement procedure
Coating material
Most coatings were applied to standard low-carbon, cold-rolled steel Q-panels from
the Q-Lab Corporation. The application method depended upon directions given by the
respective supplier.
The prescribed dry film thickness was 200 m. It is important to note that the relation
between the coating thickness and the coating resistance is not linear [17,18,19,20].
According to Bierwagen et al., the protective performance changes slowly in the range
of thin film thickness, then increases rapidly beyond the first (lower) critical thickness,
and, after a second (higher) critical film thickness, again seems to achieve a
asymptotic value that is almost constant with increasing film thickness [18].
Work in support of the existence of the first critical thickness has been published by
Touhsaent et al. The resistance of polybutadiene coatings was measured as a function
of the coating thickness. It was found that increasing film thicknesses hardly
influenced the resistance up to approximately 8 m. For film thicknesses exceeding
this value, the resistance changed drastically [19].
92
Pre-qualification of protective coatings using EIS
Gonzlez et al., showed evidence of the second critical film thickness. EIS was used to
assess the protective properties of an epoxy primer for three thicknesses. During more
than 1500 hours of immersion in 3% NaCl, the coating resistance of the 100 mcoating showed a continuous decrease with time down to 104 cm-2. The resistances
of the 200- and 500-m coatings remained above 108 cm-2 and are nearly equal [20].
Bierwagen et al. found little influence of the coating thickness for marine epoxy
coatings between 80 and 160 m immersed in 3% NaCl for more than one year [18].
These results suggest that the prescribed film thickness is appropriate (i.e. sufficiently
thick) for most coatings.
After adequate curing, two electrochemical cells (as described in chapter 3 and 4) were
applied for each coating system. Sites with pinholes, defects or deviating layer
thickness, were avoided.
The number of coatings tested sorted by coating type are shown in Table 1. A total of
119 commercial coatings of 17 different coating manufacturers was tested.
The potential perturbation was kept as low as possible and never exceeded 20 mV.
Electrolyte
The composition of the used electrolyte varied with the intended use of the coatings. In
most cases, concentrated artificial rainwater was used, with a composition that is
typical for Dutch coastal sites (Table 1) [21]. The pH of this solution was 6.6 at
ambient temperature. For marine (underwater) applications artificial seawater was
used in accordance with ASTM D1141*. One coating was also tested for its chemical
resistance to 36% HCl, 20% HNO3 and 70% H2SO4 (see section 5.3.6).
Measurement scheme
For each coating system EIS measurements were performed in two-fold after 0, 1, 7
and 21 days of immersion. For a few systems, additional measurements were
performed after several months of immersion.
Current standard:
ASTM D1141-98 (2003) Standard Practice for the Preparation of Substitute Ocean Water,
ASTM International.
93
Chapter 5
Table 1 Number of coatings tested sorted by coating type.
Coating type
100% solid epoxy
Alkyd
Bitumen
Tested coatings [#]
Chlorinated rubber
Coal tar epoxy
Epoxy ester
Epoxy silicone
Epoxy (excl. phenalkamine & novolac)
High solid epoxy - surface tolerant
Moisture-cured urethane
Novolac
Phenalkamine epoxy
Polyester
Polyurethane rubber
1
1
2
1
Silicate modified
2
1
2
19
1
10
11
6
1
1
2
Silicone
Underwater epoxy - 100% solid

Underwater epoxy
Vinyl
Vinyl ester
Water-based epoxy
Zinc silicate + epoxy
Zinc silicate + silicone
Other systems
Total
Table 2
Composition of artificial concentrated (50 X) rainwater, typical for Dutch coastal

sites [21].
Chemical formula
NaCl
Na2SO4
Ca(NO3)24H2O
(NH4)2SO4
MgCl26H2O
NH4Cl
HNO3
KNO3
FeCl24H2O
NH4NO3
NaF
NaHCO3
CuSO45H2O
NiCl26H2O
94
7
1
1
2
8
2
30
119
Concentration [mg/l]
28.03
19.20
17.71
15.14
12.71
7.23
3.74
3.03
0.99
0.85
0.31
0.17
0.05
0.02
Data fitting
Impedance data were analyzed by constructing most probable equivalent circuits
(MPEC) in accordance with the guidelines of section 3.2.3.
5.3 Results
5.3.1 Overall results
The average Y0, n and Rc values for each coating type after 21 days of immersion are
shown in Figure 1 to Figure 3. In these graphs only coatings immersed in artificial
concentrated rainwater were compared. Data that not could be fitted e.g. due to early
failure were also omitted here.
It is obvious that the variation of the Y0, n and Rc values calculated for each coating
type is rather large. The variation found for each coating type is generally larger than
the difference of the average values of two different coating types.
Consequently, no general qualifications should be given merely based upon the
average value of a certain coating type. Therefore, noteworthy measurement results of
individual coatings (systems) will be discussed in the next sections.
Single Y0 value
Average Y0 per coating type
1.00E-07
Y0 (21 d immersion) [sn -1 cm-2 ]
1.00E-08
1.00E-09
1.00E-10
1.00E-11
1.00E-12
1
10 11 12 13 14 15 16 17 18 19 20 21 22 23
Rank coating type [#]
Figure 1
Rank
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Coating Type
Vinyl ester
Polyurethane rubber
Coal tar epoxy
Phenalkamine epoxy
Epoxy ester
Underwater epoxy-100% solid
Polyester (excl. alkyd)
Epoxy silicone
100% solid epoxy
Novolac
High solid epoxy-surface tolerant
Alkyd
Vinyl
Silicone
Bitumen
Water-based epoxy
Chlorinated rubber
Underwater epoxy
Silicate modified
Single and average Y0 values for each coating type after 21 days of immersion in
artificial concentrated rainwater.
95
Chapter 5
Single n value
Average n per coating type
n (21 d immersion) [-]
0.9
0.8
0.7
0.6
0.5
1
10 11 12 13 14 15 16 17 18 19 20 21 22 23
Coating type [#]
Figure 2
Average Rc per coating type
1.00E+10
1.00E+09
Rc (21 d immersion) [cm-2 ]
Coating Type
Vinyl ester
Epoxy ester
Phenalkamine epoxy
Coal tar epoxy
Polyurethane rubber
Silicone
Epoxy silicone
Alkyd
Novolac
100% solid epoxy
Chlorinated rubber
Bitumen
Vinyl
Water-based epoxy
Underwater epoxy
Silicate modified
Single and average n values for each coating type after 21 days of immersion in
Single Rc value
1.00E+08
1.00E+07
1.00E+06
1.00E+05
1.00E+04
1
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Rank coating type [#]
Figure 3
Rank
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Rank
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Coating Type
Coal tar epoxy
Bitumen
Phenalkamine epoxy
Vinyl ester
Silicone
Epoxy ester
Polyurethane rubber
Alkyd
100% solid epoxy
Novolac
Epoxy silicone
Underwater epoxy
Chlorinated rubber
Water-based epoxy
Vinyl
Silicate modified
Single and average Rc values for each coating type after 21 days of immersion in
Reproducibility
The reproducibility of test duplicates was determined by calculating the percentage
error of all coating parameters. The averaged results are shown in Table 3. This table
also contains the values obtained for other electrolytes.
From Table 3 it becomes clear that the scatter of Figure 1 to Figure 3 should not be
associated to the accuracy of EIS measurements, but rather to actual differences in
protective properties of the tested coatings.
In addition, the fitting error of individual components is usually smaller than the
percentage error between replicates. From this observation it can be concluded that
96
error between replicates should be linked to the difficulty in producing identical

coating samples and not to equipment measurement errors (also see [22]). This is
confirmed by the good reproducibility of measurements on dummy cells and
consecutive measurements on one and the same sample (see e.g. section 4.2.3, Figure
1).
Table 3
Average percentage error [%] of test duplicates.
Immersion duration [days]
Y0
n
Rc
0
6.3
0.4
26.4
10.1
0.7
20.0
8.8
0.7
22.3
21
9.4
0.6
22.2
Tait [22] found that the variability of EIS data of a non-corroding coated metal was
approximately one order of magnitude (28 repetitions). It was suggested to use at least
5 replicates for EIS measurements.
However, for our measurements, these extreme variations were only encountered for a
few outliers and were -on average- much smaller. Considering the large amount of
different coatings tested, the use of five replicates was considered impractical and two
replicates were used instead. As mentioned in section 5.2, panels were pre-examined
for pinholes, defects or deviating layer thickness. In case of doubt, the measurement
procedure was repeated with other replicates.
The large percentage errors between replicates for the Rc and Y0 values do not pose a
serious problem as the qualification of these coatings is based on the magnitude of
these values.
5.3.2 Mixing ratio epoxy
Introduction
Epoxy coatings are usually delivered as two-component materials that must be mixed
before use. An epoxy resin should be mixed with a hardener in a fixed stoichiometric
ratio. If there is an excess of either component, a full cross-linked network will not be
achieved since some of one or the other component will remain unreacted in the
polymer network. This will result in inferior coating properties.
97
Chapter 5
On one occasion, a bisphenol A coating cured with TETA (triethylenetetramine)

performed much worse than what should be expected for this type of coating. It turned
out that an incorrect mixing ratio was used. Therefore, the measurement procedure was
repeated using the proper mixing ratio.
Results
The results of both formulations are shown in Figure 4. The mixing ratio has a large
effect on the resistivity of the coating. After 21 days of immersion, the average
resistance of the improper mixing ratio (system A) dropped to 3.0105 cm-2 while
the resistance of the proper mixing ratio (system B) remained 1.8108 cm-2.
Furthermore, the water uptake -signified by the Y0 value of the coating- was notably
affected. The average Y0 value of system A increased to 6.510-9 sn-1cm-2 while the
Y0 of system B only increased to 5.710-11 sn-1cm-2.
1.00E+09
1.00E-08
A1
A1
A2
A2
B1
B2
1.00E+07
1.00E+06
B1
Y0 [sn -1]
Rc [cm -2]
1.00E+08
B2
1.00E-09
1.00E-10
1.00E+05
1.00E+04
1.00E-11
0
10
Time [d]
15
20
1.20
10
Time [d]
20
A1
A1
A2
A2
n [-]
1.00
B1
108
B2
0.80
107
|Z|
0.60
0.40
106
105
104
0.20
103
102
0.00
0
Figure 4
98
B1
B2
10
Time [d]
15
20
103
104
105
Frequency (Hz)
Influence of the resin-hardener mixing ratio upon the protective properties. A

indicates the coating with the improper mixing ratio. B indicates the proper mixing
ratio. The numbers (1,2) indicate replicates of the same system. The Bode
magnitude plot shows the results after 21 days of immersion.
Discussion and conclusions

A proper mixing ratio is very important for the formation of a highly cross-linked
network. Deviation from the stoichiometric ratio will influence both ionic conductivity
and water uptake.
The low coating resistance found for system A is related to the high ion mobility in the
insufficiently cross-linked network. Day et al. found that both excess resin and
hardener will result in higher ion conductivity Deviations from ideal stoichiometry as
small as 1.1:1 produce notable differences in the cured network [23].
Water uptake (increase of Y0) in case of excess epoxy resin is also higher due to an
insufficiently cross-linked network. In case of excess curing agent, the hydrophilic
nature of NH2 groups facilitates water uptake [24].
The EIS measurement results confirm the importance of a proper mixing ratio.
5.3.3 Zinc-rich paint
Introduction
ZRPs (zinc-rich paints) contain high levels of zinc particles (spherical or lamellar
shape, or a combination) dispersed in an inorganic (usually silicates) or organic binder
(usually epoxies) [1,25,26]. These high levels are required to assure electrical contact
between the zinc particles and the steel substrate.
ZRPs are usually highly porous: water can easily reach zinc particles while in
electrical contact with steel substrate [26]. Consequently, zinc can preferentially
dissolve, acting as a sacrificial pigment, and allowing a cathodic protection of the steel
substrate (active protection). This process mainly takes place at the beginning of the
coatings lifetime and gradually slows down [1,25].
Next, (insoluble) zinc corrosion products form and fill up the pores, and the coating
behaves like a barrier-type coating (passive protection, Figure 6b and c).
Undrum [27] states that the dominant mode of corrosion protection can be
distinguished by the measured OCP. The potential of bare steel in seawater is
approximately -650 mV vs. SCE (saturated calomel electrode), and zinc has a potential
of approximately -1050 mV SCE [28]. At an OCP below -800 mV, the coating still
provides cathodic protection. Above this level, the barrier mechanism becomes
dominant.
Monitoring the open circuit potential (OCP) prior to the EIS measurements allows to
follow the electrochemical activity of the coatings. It is commonly [1,26] assumed that
the electrochemical processes that proceed in ZRPs are the oxidation of zinc particles
99
Chapter 5
(Zn Zn2+ + 2e) and the reduction of the dissolved oxygen (O2 + 2H2O + 4e-
4OH).
An organosiloxane ZRP was immersed in artificial concentrated rainwater for 6
months. Its performance was evaluated using both results of OCP and EIS
measurements.
Results and discussion
The increase of the OCP (Figure 5, upper left) corresponds to the decrease of the
electroactive zinc area. This is generally attributed to the isolation of the zinc particles
by the zinc corrosion products in the coating. With respect to the statement of Undrum,
it can be inferred that the protective mechanism shifts from cathodic protection to a
barrier mechanism.
Initially, the coating resistance drops rapidly (down to 2.6106 cm-2) as the
electrolyte penetrates the coating and slowly increases again as corrosion products are
formed that block the voids in the coating. The related average coating resistance
versus time is shown in Figure 5 (upper right). The Y0 value of the constant phase
element of the coating rises in the first week, presumably due to water uptake. After
this period, the Y0 decreases again. It can be assumed that water is gradually being
expelled by zinc corrosion products.
The protective mechanism of ZRP upon immersion is summarized Figure 6.
Compared to the zinc-rich silicate primers (discussed in the next section), the coating
resistance of the ZRP remains relatively high.
Many properties of coatings change abruptly above a certain pigment volume
concentration (PVC). The PVC at which these changes occur is designated as the
critical pigment volume concentration (CPVC) . In ZRPs formulated far below the
CPVC, zinc pigments are isolated from each other by the binder matrix. If the PVC
value approaches the CPVC, an increasing amount of pigment particles will touch
each other. Chains of pigment particles may be formed that stretch throughout the
coatings thickness (this is known as percolation). At PVCs exceeding the CPVC the
zinc particles are closely packed and the coating becomes conductive [1,26,29].
100

0.00
1.00E+08
-0.10
1.00E+07
-0.30
Rc [cm -2]
OCP [V]
-0.20
-0.40
-0.50
-0.60
1.00E+06
1.00E+05
-0.70
-0.80
1.00E+04
-0.90
-1.00
1.00E+03
0
50
100
Time [d]
150
200
1.00E-08
50
100
Time [d]
150
200
1.00
n [-]
Y0 [sn -1]
0.90
1.00E-09
0.80
0.70
1.00E-10
0.60
1.00E-11
0.50
0
50
Figure 5
100
Time [d]
150
200
50
100
Time [d]
150
200
Clockwise from upper left: open circuit potential (vs. sat. Ag/AgCl), coating
resistance Rc, n and Y0 value of the constant phase element of a ZRP as a function
of immersion time.
a
H2O, O2
electrolyte
conductive
pathway
binder matrix
zinc pigment
Zn2+
OH-
steel
cathode
e-
zinc
anode
zinc
corrosion
products
zinc
corrosion
products
Figure 6
Representation of processes taking place upon immersion of steel coated with a

ZRP: a) electrolyte penetration into pores or conductive pathways, zinc pigments when in electrical contact with steel- are preferentially dissolved and provide
cathodic protection to steel. b) Zinc corrosion products form and fill up the pores.
c) Zinc pigments will continue to corrode (possibly non-galvanically) at sites were
these particles come into contact with the electrolyte.
101
Chapter 5
Conclusions
With the combined use of the OCP, EIS measurement results and literature sources it
is possible to derive the protective mechanism of ZRPs.
In this specific case, the studied ZRP is formulated close to, but below the CPVC. The
OCP signifies the presence of zinc particles in contact with the steel substrate, whereas
the relatively high coating resistance indicates that the percolation threshold (i.e. a
well-established electrical conduction) has not been reached.
5.3.4 Zinc-rich silicate primers
Introduction
Compared to epoxies, silicate binders can be pigmented at extremely high levels of
zinc dust. In fact, all zinc-rich silicates are formulated with a PVC that exceeds the
CPVC. This simplifies the realization of metallic zinc-to-zinc contact, resulting in
superior cathodic protection properties.
In addition, on curing, polymerisation of the silicate binder takes place concurrently
with a chemical reaction between the silicate binder, the metallic zinc particles and the
iron substrate. This results in an electrically conductive polysilicate matrix around the
metallic zinc particles. This conductive matrix enhances the galvanic protection
mechanism.
Today, the most important class of zinc-rich silicates are based on solvent-borne alkyl
silicates (typically ethyl silicate) [25,27,30-32].
Zinc-rich silicate primers are often coated with one or more topcoats. These topcoats
are used to extend the life of the inorganic zinc coating by reducing the rate of zinc
depletion through galvanic protection and to protect the primer against physical
damage [1,33,34].
In order to determine the influence of zinc-rich silicate primers upon the performance
of the total system, three systems have been examined with EIS.
All tested systems consisted of a zinc-rich ethyl silicate-based primer and an epoxy
topcoat. Each layer was also tested separately.
The average resistance (Rc) and polarisability (reflected by Y0) for the separate epoxy
layers as well as for the complete system is shown in Figure 7. The average resistance
after 21 days of immersion is shown in Figure 8.
102
For the measurement of single zinc-rich silicate primers no Rc, Y0 and n values could
be obtained. This is not surprising, as these films are formulated above the CPVC, are
highly conductive and thus behave comparable to a corroding zinc substrate. The lowfrequency limit of impedance these primers was estimated to be below 104 cm-2.
After one day of immersion, the average OCP drops to -0.96 V vs. saturated Ag/AgCl
which is close to the OCP of zinc in seawater [28]. The average OCP gradually rises to
-0.84, probably due to the formation of a layer of zinc corrosion products.
After 21 days of immersion the separate epoxy coatings showed an average Rc of
1.8107 cm-2. The complete system (zinc-rich silicate primer + epoxy topcoat)
clearly showed the synergetic effect of the two layers. The average Rc after 21 days of
immersion improved more than on order of magnitude: 2.7108 cm-2 which is much
more than the combined Rc values of the separate layers.
Water uptake of the complete systems, signified by the increase of Y0, was largest for
system B and C. These systems also showed the largest increase of Rc between 7 and
21 days of immersion. From this it can be assumed that elevated water levels in the
coating facilitate the growth of zinc corrosion products. These products block
conductive pathway in the coating and consequently Rc rises.
Though the influence of the zinc-rich silicate primer upon the performance of the total
system is very clear, the cause of this influence needs some clarification.
The galvanic protection offered by the (electrically conductive) zinc-rich silicate
primer is comparable to that offered by galvanised steel. Zinc layers coated with
organic coatings are often referred to as duplex systems [35-38].
The combination of a zinc layer with an organic coating acts synergistically; i.e. the
protection by the duplex system is better than the sum of the protective features of the
zinc layer and the organic coating separately. Initially the organic coating retards the
contact between corrosive species and the zinc surface. Later on, when the corrosion
process occurs, the generated products block the pathways developed throughout the
coating. Consequently, the flux of further corrosive species towards the zinc surface is
slowed down [35,36,38].
103
Chapter 5
A
1.00E+09
1.00E+09
1.00E+08
Rc [cm -2]
Rc [cm -2]
1.00E+08
1.00E+07
1.00E+07
1.00E+06
1.00E+06
1.00E+05
1.00E+05
0
10
Time [d]
15
20
1.00E-09
10
Time [d]
15
20
A
B
C
1.00E-09
Y0 [sn -1cm -2]
Y0 [sn -1cm -2]
1.00E-10
1.00E-11
1.00E-10
1.00E-11
0
10
15
20
Time [d]
10
15
20
Time [d]
Left graphs: average Rc and Y0 values of single epoxy layers (A,B,C), as a function
of immersion time. Right graphs: average Rc and Y0 values of zinc-rich silicate
primers + epoxy topcoats.
Rc (21 days) [cm -2]
Figure 7
1.00E+10
Zinc-rich silicate primer
1.00E+09
Epoxy topcoat
1.00E+08
Zinc-rich silicate primer

+ Epoxy topcoat
1.00E+07
1.00E+06
1.00E+05
1.00E+04
1.00E+03
A1
Figure 8
A2
B1
B2
C1
C2
Influence of a zinc silicate primer upon the resistance of the total system. Three
systems (A,B,C) are shown. The numbers (1,2) indicate replicates of the same
system. The Rc values of the zinc silicate primer could not be resolved and
maximum values are given instead.
In order to obtain a better comprehension of the nature of the synergetic effect, Prez
et al. performed a comparative study between bare galvanised steel and three different
104
duplex systems. Analysis revealed that, after exposure to a cyclic weathering chamber,
the corrosion products of the bare galvanised steel and the duplex systems differed.
Fore bare galvanised steel, hydrozincite, Zn5(CO3)2(OH)6, was the main compound.
For the duplex systems, zincite (ZnO), was the main compound found at the coatingzinc interface. This compound provides better passivation of the zinc surface than
hydrozincite. Apparently, the three coating systems acted as a selective membrane
avoiding CO2 uptake, resulting in superior protective properties [35].
Conclusions
In contrast to the protective properties of the ZRP discussed in section 5.3.3, the
electrochemical properties of the zinc-rich silicate primers on steel can be compared to
that of (hot-dip) galvanised steel. Though the conditions were different from those
used by Prez et al. (and possibly also the protective mechanisms) the synergetic effect
that exists for conventional duplex systems has also been confirmed for zinc-rich
silicate primers with epoxy topcoatings.
5.3.5 Corrosion-inhibitive pigments
Introduction
Protective coatings that do not possess effective barrier properties, such as alkyds and
water-borne epoxies, require an alternative protective mechanism [39-42].
Alkyds are one-component, air-curing coatings (oxidative cross-linking) and,
consequently, are fairly easy to apply [25,39]. Alkyd coatings offer very little barrier
resistance to moisture and oxygen permeation when compared to more modern, fully
synthetic coatings [39].
To control the viscosity of the commercial product, an excess of the alcohol
component usually is used in the reaction mix that forms the resin. These alcohols are
water soluble. The result is that alkyds contain an appreciable amount of water-soluble
material and tend to absorb water and swell.
Alkyds can withstand occasional immersion provided they are permitted to dry.
Repeated wet/dry cycling eventually will cause the coatings to detach [39]. Therefore
alkyd coatings should not be used for immersion service.
Once the film has formed the coating continues to react with oxygen and due to
additional cross-linking, the coating eventually becomes brittle and deteriorates [39].
105
Chapter 5
Growing environmental pressure to reduce the use of organic solvents is focusing

increasing attention on the development of water-borne coatings.
Polymer coatings have an apolar nature and have to be chemically altered to make
them soluble or dispersible in the polar water diluent [25,43-45].
The conversion of solvent-borne resins into equivalent waterborne systems introduces
new problems. By their nature, water-based coatings contain residual hydrophilic
functionality. Consequently, water-borne coatings usually exhibit high water
permeability which tends to enhance their ionic conductivity, a disadvantageous
property for corrosion protection [41,42].
To work against these deficiencies, corrosion inhibitors are frequently added to these
coatings. Zinc phosphate is probably the most widely used corrosion inhibitor [39,46].
Inhibition takes place by phosphate ion donation and protective film formation at
anodic sites.
The performance of three systems containing zinc phosphate (two water-based epoxies
and a two-layer alkyd coating system) and a reference system (solvent-borne epoxy
without inhibitive pigments) was evaluated using EIS. All systems were immersed in
Results
The results of the four coatings are shown in Figure 9.
After 21 days, the average coating resistance Rc of the alkyd coating (B) remained
high: 1.5108 cm-2, almost as high as the solvent-borne epoxy reference system (D):
1.9108 cm-2. The water-based epoxies (A, B) showed lower Rc values of 3.3107
and 4.2106 cm-2.
The Y0 values of Figure 9 (right graphs) have been normalized by fixing the initial
value at 1. The increase of the Y0 is mainly attributed to water uptake [45]. The largest
increase of Y0 was observed for water-based epoxy B which also showed the lowest
coating resistance. All systems containing zinc phosphate showed a larger water
uptake than the reference system.
The alkyd coating and the epoxy reference system were also measured after 379 and
408 days of immersion, respectively. Even after this extended period, the resistance of
the alkyd coating remained remarkably high and is close to that of the epoxy reference
system. However, the normalised Y0 value of the alkyd coating increasingly deviated
from that of the epoxy reference system. The extreme increase of this value obtained
for the alkyd coating signifies that the coating continued to take up large amounts of
water.
106
1.00E+09
12
A
B
B
10
C
D
1.00E+07
1.00E+06
1.00E+05
4
2
0
10
Time [d]
15
20
1.00E+10
10
Time [d]
15
20
18
16
1.00E+09
14
1.00E+08
Y0 normalised [-]
Rc [cm -2]
1.00E+04
1.00E+07
1.00E+06
1.00E+05
12
10
8
6
4
2
0
1.00E+04
1
10
100
Time [d]
Figure 9
Y0 normalised [-]
Rc [cm -2]
1.00E+08
1000
10
100
1000
Time [d]
Coating resistance Rc (left graphs) and normalised Y0 value of the constant phase
element (right right graphs). Systems A and B are water-based epoxies with zinc
phosphate. System C is an alkyd coating with zinc phosphate. D is a solvent-borne
reference epoxy system without inhibitive pigments.
After these additional measurements both panels were inspected visually. The alkyd
coating showed several blisters with a width in the order of 5 mm. One blister formed
between the two coating layers while the other blisters formed underneath both coating
layers. After removal of the detached coating layers the exposed steel substrate
appeared shiny, most probably due to the inhibitive action of zinc phosphate. The
epoxy reference system showed no visual indications of breakdown.
Discussion
Hernandez et al. have shown that performance of waterborne coatings was markedly
affected by the presence of zinc-aluminium-phosphate. The resistance of the coatings
changed more than one order of magnitude due to the addition of the inhibitive
pigment [47].
However, inhibitive pigments cannot function unless they are somewhat soluble in
water [42]. Zinc phosphate is extremely insoluble, and several enhancements to
increase solubility have been commercialized [46]. In the presence of water, a small
fraction of the inhibitive pigments is dissolved which makes them available at the
coating-metal interface. To allow diffusion of the dissolved anticorrosive agent to the
107
Chapter 5
coating-metal interface, the binder should be sufficiently permeable to water. This

requirement clearly contradicts the other requirement of corrosion protective coatings,
i.e. to prevent access of water as a corrosive agent to the metal surface [42].
Consequently, zinc phosphate continuously hydrolyzes and eventually will be lost by
leaching from the paint film [1,25].
The presence of blisters underneath the two layers of the alkyd system suggests that
this type of coating failure would also be reflected in the impedance data. However,
the impedance spectra were dominated by the high impedance of the coating.
Consequently, no reasonable values could be obtained for the electrochemical double
layer (Qdl) and the charge transfer resistance (Rct) (see section 3.2.7 for the physical
meaning of these elements).
Similar behaviour has previously been observed by van der Weijde [48]. An alkyd
coating was applied to a steel substrate that was deliberately contaminated with NaCl.
One sample showed capacitive behaviour for several days, despite the formation of a
small blister after two days of immersion. Immediately after perforation of the blister
with a needle, the Nyquist plot revealed the characteristic shape of a defective coating
system (also see section 3.2.6).
The long-term behaviour of an immersed alkyd coating has been studied by Murray et
al. [49]. Carbon steel panels coated with a thin (approximately 30 m) two-layer alkyd
coating were immersed in artificial seawater (ASTM D1141). Initially, coating
resistances exceeding 108 cm-2 were measured, which decreased gradually with
time. For two panels the capacitance as a function of immersion time was given.
Though the capacitances were approximated by calculating these at a single frequency
(104 Hz), the large increase with time suggests large amounts of water taken up by the
coating. Rustless blisters appeared on all samples, though it should be added that there
were fewer but considerably larger blisters in the cell area compared to other areas of
the panels. Perhaps the high perturbation used (60 mV) influenced the growth of
blisters. The alkyd-coated panel that was immersed for two years, showed a sharp drop
of the low-frequency impedance (10-2 to 10-3 Hz) after almost one year of exposure.
The impedance dropped from 107 to 105 cm-2, indicating coating failure. Based upon
the experimental results Murray et al. concluded that the alkyd coating systems were
unacceptable for immersion applications.
Conclusions
Despite the use of zinc phosphate as an inhibitive pigment, the protective properties of
the tested water-based epoxies are at the lower end of all tested coatings. The
108
protective properties will further deteriorate when zinc phosphate eventually will be
lost due to leaching.
The alkyd coating showed very typical behaviour, previously reported by other
workers. The coating resistance remains high for extended periods and the coating
takes up large amounts of water. Gradually, blisters develop. With EIS, only after the
collapse of these blisters coating failure becomes apparent.
For practical applications it is likely that blisters collapse in an early stage due to
mechanic stresses applied to the coating. Therefore, alkyd coatings should not be used
for immersion applications.
5.3.6 Chemical resistance moisture-curing silicone rubber
Introduction
Novolac epoxies and novolac epoxy-based vinyl esters are known for their chemical
resistance [50]. Novolac epoxies have many reactive groups on the backbone chain,
resulting in high cross-linking density in the coating. This results in a greater chemical
resistance.
To determine the chemical resistance of a moisture-curing silicone rubber coating,
coated panels were exposed to various electrolytes and were examined with EIS.
Measurements were performed with hydrochloric acid (HCl, 36 %), nitric acid (HNO3,
20 %), sulphuric acid (H2SO4, 70%) and concentrated artificial rainwater (reference).
The measurement with sulphuric acid was performed with cell consisting of PVC,
instead of the standard PMMA cell as the latter material is not suitable under these
conditions.
The average resistance as a function of the exposure time is shown in Figure 10.
Throughout the test, the highest average resistance was maintained by the coating
samples exposed to concentrated artificial rainwater: 5.2108 cm-2. The average
resistance of samples exposed to HCl finished somewhat lower: 1.6108 cm-2.
Samples exposed to HNO3 and H2SO4 clearly failed during the test. The average
resistance of samples exposed to HNO3 dropped to 3.3105 cm-2 after 21 days of
immersion. For H2SO4 the average resistance dropped to 1.4104 cm-2 after 7 days
of immersion and no coating resistance could even be determined at the end of the test.
109
Chapter 5
Visual inspection of the coating revealed no blister formation or corrosion products for
immersion in concentrated artificial rainwater, HCl and HNO3. In case of H2SO4, the
coating completely detached.
Art. rain
1.00E+09
HCl [36 %]
Rc [cm -2 ]
1.00E+08
HNO3 [20%]
H2SO4 [70 %]
1.00E+07
1.00E+06
1.00E+05
1.00E+04
1.00E+03
0
10
15
20
Time [d]
Figure 10 Coating resistance Rc as a function of time for different electrolytes.
Conclusions
Based on these results it was concluded that the coating has good protective properties
under neutral conditions and will probably withstand short-term exposure to 36 %
HCl. The coating is considered unsuitable for exposure to 20 % HNO3 and 70%
H2SO4.
The results also confirm the advantage that EIS has over visual inspection. Visual
inspection showed no difference between the coating performances for all electrolytes
except for H2SO4. With EIS it is possible to distinguish between the performances in
all tested electrolytes.
The electrochemical properties of more than 100 industrial coatings have been
determined with EIS, using a benchmark measurement protocol.
From the overall results and the percentage error between replicates it became clear
that large variability of coating parameters for each coating type should not be
associated to the accuracy of EIS measurements, but rather to actual differences in
protective properties of the tested coatings.
In addition, the fitting error of individual components was usually smaller than the
error between replicates. From this it was concluded that error between replicates
110
should be linked to the difficulty in producing identical coating samples and not to
equipment measurement errors. This was confirmed by the good reproducibility of
measurements on dummy cells and consecutive measurements on one and the same
sample.
By discussing some characteristic results more insight has been obtained in the
usability and limitations of EIS.
EIS clearly demonstrated the influence of the mixing ratio of the resin and the
hardener. Deviation from the stoichiometric ratio results in inferior coating properties.
The protective mechanism of a ZRP was analysed with the combined use of the OCP,
EIS measurement results and literature sources.
In order to determine the influence of zinc-rich silicate primers upon the performance
of the total system, three systems were been examined with EIS. All tested systems
consisted of a zinc-rich ethyl silicate-based primer and an epoxy topcoat. Each layer
was also tested separately. In contrast to the protective properties of the ZRP, the
electrochemical properties of the zinc-rich silicate primers on steel were comparable to
that of (hot-dip) galvanised steel. The synergetic effect that exists for conventional
duplex systems has also been confirmed for zinc-rich silicate primers with epoxy
topcoatings.
The performance of three systems containing zinc phosphate (two water-based epoxies
and a two-layer alkyd coating system) was evaluated using EIS. Despite the use of
zinc phosphate as an inhibitive pigment, the protective properties of the tested waterbased epoxies were at the lower end of all tested coatings. The protective properties
will further deteriorate when zinc phosphate eventually will be lost due to leaching.
The alkyd coating showed typical behaviour. The coating resistance remained high for
extended periods and the coating took up large amounts of water. Gradually, blisters
develop. Due to the high resistance of alkyd coating, the double layer capacitance and
charge transfer resistance are masked. Only after the collapse of these blisters coating
failure becomes apparent. In this specific case, EIS clearly fails as a tool to determine
the degradation mechanism.
To determine the chemical resistance of a silicone rubber coating, coated panels were
exposed to various electrolytes and were examined with EIS. Measurements were
performed with hydrochloric acid, nitric acid, sulphuric acid and concentrated artificial
rainwater (reference). Based on these results it was concluded that the coating has
good protective properties under neutral conditions and will probably withstand shortterm exposure to 36 % HCl. The coating is considered unsuitable for exposure to 20 %
HNO3 and 70% H2SO4.
111
Chapter 5
These results also confirm the advantage that EIS has over visual inspection. Visual
inspection showed no difference between the coating performances for all electrolytes
except for H2SO4. With EIS it is possible to distinguish between the performances in
all tested electrolytes.
In summary, EIS provides an objective method to rapidly examine the corrosion
protective mechanisms of coatings as well as other properties such as the mixing ratio
and chemical resistance. However, EIS is not a standalone technique. Only with the
aid other tools (OCP measurement, visual observation, etc.) and sources from literature
meaningful conclusions can be drawn upon the protective mechanisms of the coatings.
112
5.5 References
1
Wicks, Z.W., F.N. Jones, S.P. Pappas, Organic Coatings: Science and Technology Second Edition; Wiley, New York, 1999, 630, ISBN 0-471-24507-0.
Bethencourt, M., F.J. Botana, M.J. Canoa, R.M. Osuna, M. Marcos, Lifetime prediction
of waterborne acrylic paints with the ACDCAC method, Progress in Organic Coatings
49 (2004) 275-281.
Claydon, D.A., International Protective Coatings, technical paper, Akzo Nobel, United
Kingdom (2002).
Morcillo, M., J. Simanacas, J.M. Bastidas, S. Feliu, C. Blanco, F. Camn, Comparison

of Laboratory Tests and Outdoor Tests of Paint Coatings for Atmospheric Exposure, in:
Polymeric Materials for Corrosion Control, R.A. Dickie and F.L. Floyd, Eds., American
Chemical Society, Washington, DC, 1986, 86-100.
Morcillo, M., S. Feliu, J. Simaneas, J.M. Bastidas, Application of the AC Impedance

Measurments to Predict Paint Systems Performance in Marine Atmospheres, Marine
Corrosion and Fouling, Sixth International Congress, Athens, Greece, 5-8 September
1984.
Wit, J.H.W. de, D.H. van der Weijde, G. Ferrari, Chapter Organic Coatings, In:
Corrosion Mechanisms in Theory and Practice, P. Marcus (Ed.), Marcel Dekker, Inc.,
New York Basel ISBN 0824706668 (2002).
Scully, J.R., Electrochemical Impedance of Organic-Coated Steel: Correlation of

Impedance parameters with Long-Term Coating Deterioration, Journal of The
Wit, J.H.W. de, E.P.M. van Westing, D.H. van der Weijde, Evaluation of Coatings A
total system Approach. Materials Science Forum, 247 (1997) 69-82.
Davis, G.D., L.A. Krebs, C.M. Dacres, Coating Evaluation and Validation of Accelerated
Test Conditions Using an In-Situ Corrosion Sensor, Journal of Coatings Technology, 74
(2002), 69-74.
10 Holyoake, K., J. Yuan, Electrochemical Impedance Spectroscopy Measurements of

Barrier Coatings, technical paper, Armatec, New Sealand [www.armatec.co.nz].
113
Chapter 5
11 ODonoghue, M., R. Garrett, V. Datta, P. Roberts, Electrochemical Impedance
Spectroscopy: Testing Coatings for Rapid Immersion Service, Materials Performance,
September 2003, 36-41.
12 Thomas, N.L., Coatings for rusty steel, where are we now?, Surface Coatings
International, Journal of the Oil and Colour Chemists' Association, 74 (1991) 83-97.
13 Mayne, J.E.O., The Mechanism of the Inhibition of the Corrosion of Iron and Steel by
Means of Paint, Official Digest, February 1952, 127-136.
organic coatings on steel in a neutral electrolyte, Journal of Coatings Technology 68
(1996), 45-56.
15 Bacon C.R., J.J. Smith, F.M. Rugg, Electrolytic Resistance in Evaluating Protective
Merit of Coatings on Metals, Industrial and Engineering Chemistry, 40 (1948) 161-167.
16 Greenfield, D. D. Scantlebury, The Protective Action of Organic Coatings on Steel: A
Review, Journal of Corrosion Science & Engineering, 3 (2000), paper 5.
17 Di Sarli, A.R., E.E. Schwiderke, J.J. Podesta, Evaluation of anticorrosive paint binders
by means of A.C. techniques Influence of the coat thickness, 10th ICMC, Madras, India,
7-11 November 1987, Proceedings, 1091-1100.
18 Bierwagen, G., J. Li, N. Davis, D. Tallman, Thickness Dependence of Electrochemical
Properties of Organic Coatings, in: Proceedings of the Fifth Nrnberg Congress, vol. 1,
Nrnberg, Germany, April 12-14, 1999. Vincentz Verlag, Hannover, Germany, 315-335.
19 Touhsaent, R.E., H. Leidheiser, A Capacitance-Resistance Study of Polybutadiene
Coatings on Steel, Corrosion, 28 (1972), 435-440.
20 Gonzlez, S., M.A. Gil, J.O. Hernndez, V. Fox, R.M. Souto, Resistance to corrosion of
galvanized steel covered with an epoxy-polyamide primer coating, Progress in Organic
Coatings 41 (2001) 167-170.
21 Westing, E.P.M. van, D.H. van der Weijde, M.P.W. Vreijling, G.M. Ferrari, J.H.W. de
Wit, Quantification of Coating Aeging Using Inpedance Measurments, In: Organic
Coatings for Corrosion Contol, G.P. Bierwagen (Ed.), American Chemical society,
Washington, DC, USA, 1998, ISBN: 084123549-X, 69-90.
22 Tait, W.S., Coping with Errors in Electrochemical Impedance Spectroscopy Data From
Coated Metals, Journal of Coatings Technology, 66 (1994), 59-61.
114

23 Day, D.R., Effects of Stoichiometric Mixing Ratio on Epoxy Cure-A Dielectric Analysis,
Journal of Reinforced Plastics and Composites, 7 (1988) 475-484.
24 Zhang, J.-T., J.-M. Hu, J.-Q. Zhang, C.-N. Cao, Influence of Curing Agent Content on
the Performance of Epoxy Coatings on Mild Steels, Corrosion, September 2005, 872-879.
25 Forsgren, A., Corrosion control trough organic coatngs, 2006, CRC Press
ISBN:9780849372780.
26 Meroufel, A., S. Touzain, EIS characterisation of new zinc-rich powder coatings,
Progress in Organic Coatings, 59 (2007) 197-205.
27 Undrum, H., Silicate and Epoxy Zinc Primers: A Review, Journal of Protective Coatings
& Linings, July 2006, 52-57.
28 Metals Handbook, Volume 13 - Corrosion, ASM International, 9th edition, ISBN:
0871700190.
29 Hamer, W.J., Polypyrrole Electrochemistry Environmentally friendly corrosion
30 Keijman, J.M., Inorganic and Organic Coatings The difference, PCE 99 Conference,
10-12 March 1999, Brighton, United Kingdom.
31 Parashar, G., D. Srivastava, P. Kumara, Review - Ethyl silicate binders for high
performance coatings, Progress in Organic Coatings, 42 (2001) 1-14.
32 Eccleston, The Effect of Cure Temperature and Humidity on the Properties of SolventBorne Zinc Silicate Coatings, Journal of Protective Coatings & Linings, January 1998,
36-45.
33 Kline, H.H., Inorganic Zinc-Rich, Journal of Protective Coatings & Linings, November
1996, 73-105.
34 Keijman, K., The Use of Novel Siloxane Hybrid Polymers in Protective Coatings,
Protective Coatings Europe, July 1996, 26-32.
35 Prez, C., A. Collazo, M. Izquierdo, P. Merino, X.R. Nvoa, Comparative study between
galvanised steel and three duplex systems submitted to a weathering cyclic test,
36 Amo, B. del, L. Vleva, A.R. Di Sarli , C.I. Elsner, Performance of coated steel systems
exposed to different media Part I. Painted galvanized steel, Progress in Organic Coatings,
50 (2004) 179-192.
115
Chapter 5
37 Deflorian, F., S. Rossi, L. Fedrizzi, P.L. Bonora, EIS study of organic coating on zinc
surface pretreated with environmentally friendly products, Progress in Organic Coatings,
52 (2005), 271-279.
38 Cabanelas, I., A. Collazo, M. Izquierdo, X.R. Nvoa, C. Prez, Influence of galvanised
surface state on the duplex systems behaviour, Corrosion Science 49 (2007) 1816-1832.
39 Byrnes, G., Alkyds, Journal of Protective Coatings & Linings, December 1996, 73-87.
40 Smith, L.M., Introduction to Generic Coating Types, Journal of Protective Coatings &
Linings, July 1995, 73-82.
41 Kendig, M., H.-S Ryang, T. Liao, M. Cunningham, S. Jeanjaquet, Corrosion Induced
adhesion loss - Low volatile organic content (VOC) coating, Report # SC71026.FR,
Rockwell International Science Center (1993), 37.
42 Funke, W., How Organic Coating Systems Protect Against Corrosion, in: Polymeric
Materials for Corrosion Control, R.A. Dickie and F.L. Floyd, Eds., American Chemical
Society, Washington, DC, 1986, 222-228.
43 Decocq, F., M. Slinckx, C. Nootens, A New Technology For Environmentally Friendly,
UV-Resistant, Water-Borne Anticorrosion Paints, Journal of Protective Coatings &
Linings, June 2001, 48-56.
44 Goldie, B., Developments in Formulating Water-borne Coatings, Journal of Protective
45 Westing, E.P.M. van, Determination of coating performance with impedance
46 Hare, C. H., Inhibitive Primers for Metal: Fundamental Considerations, Journal of
Protective Coatings and Linings, May, 1998, 48-62.
47 Hernandez, M., Effect of an inhibitive pigment zinc-aluminum-phosphate (ZAP) on the
corrosion mechanisms of steel in waterborne coatings, Progress in Organic Coatings, 56
(2006) 199-206.
48 Weijde, D.H. van der, Impedance Spectroscopy and organic barrier coatings;
(im)possibilities, Ph.D. Thesis, Delft University of Technology, 1996.
49 Murray, J.N., H.P. Hack, Testing Organic Architectural Coatings in ASTM Synthetic
Seawater Immersion Conditions Using EIS, Corrosion, 48 (1992) 671-685.
50 Salem, L.S., Epoxies for Steel, Journal of Protective Coatings and Linings, September,
1996, 77-98.
116
6 Application of ENM to the study of protective coatings
6.1 Introduction
Currently, EIS (electrochemical impedance spectroscopy) is the most commonly used

electrochemical method for evaluation of the protective properties of coatings.
Recently, the application of ENM (electrochemical noise measurement) for the same
purpose has become of interest. There are number of advantages of this technique over
EIS:
ENM requires relatively low-cost equipment [1,2].
Since no external source of current or voltage is needed, interference with the
system is minimal and long-term monitoring can be continuous [3].
EIS measurement on coatings are normally performed under potentiostatic
control at the OCP (open circuit potential), measured prior to the actual EIS
measurement. If the OCP is time-dependent at a rate that exceeds the lowest
frequency of the EIS measurement, the system is non-stationary and applying
EIS produces errors. For ENM no stabilizing potential is applied [4].
With ENM it is possible to distinguish between various types of corrosion [5].
Despite these advantages, the application of ENM for corrosion studies remains a
controversial subject [6]. In this chapter, results of some experiments will presented
and discussed. Based on these results, conclusions will be drawn upon the usability of
ENM for evaluation of protective coatings.
6.2 Experimental
All noise measurements were performed with computer-controlled Schlumberger

(Solartron) 1286 and 1287 Electrochemical Interfaces (ECI).
Data were collected using Windows-based PCs running CorrWare from Scribner.
The layout of the electrochemical cell is similar to the one described in section 3.3.4.
The cell consists of a PMMA tube (inner diameter 4 cm). Two nominally identical
coated electrodes are glued to each end of the tube with silicon sealant. Similar to the
electrochemical cell for EIS measurements (see section 4.4.2), a self-adhesive PVC
mask is used to ensure that 10 cm2 of the coated electrodes is exposed to the
electrolyte.
117
Chapter 6
A saturated calomel electrode (SCE) is placed in the middle of the tube, between the
two coated electrodes.
With this setup, current fluctuations between two nominally identical electrodes are
measured using a zero resistance ammeter, while the potential fluctuations of both
electrodes are measured with respect to the reference electrode.
6.3 Drift
Introduction
As pointed out by Cottis, there are unresolved fundamental questions about when
variation of potential or current should be regarded as drift and when it should be
regarded as electrochemical noise [7].
In order to find out how drift removal affect the calculated noise parameters a single
dataset has been analysed using two common drift-removal methods.
Experimental
Samples of the investigated bitumen coating (applied to steel Q-panels) have been
immersed in artificial concentrated rainwater (see section 5.2 for the composition) for
more than 1 year. Prior to ENM the coating showed several blisters and rust spots.
Potential and current fluctuations were recorded for 212 seconds at an average rate of
9.7 points per second.
Measurement results
The raw time records of potential and current are shown in Figure 1. A number of
features can be derived from these graphs.
The mean potential of the coupled electrodes (Ecoup) is indicative for the resistivity of
the coating. As the resistivity of coating drops, Ecoup will decrease towards the
corrosion potential of the bare substrate. In this case, Ecoup was -629 mV vs. SCE
which is close to the corrosion potential of bare steel in seawater (approximately -650
mV vs. SCE [8]). This confirms the existence of conductive pathways trough the
coating.
The mean coupling current (Icoup) is measured between the two painted electrodes. A
large Icoup indicates coating damage since virtually no current will flow through intact
coatings. For this coating Icoup was 1.9810-7 Acm-2. This value underscores the poor
quality of the coating, as Icoup of more resistive coatings is found to be almost two
magnitudes smaller (see section 6.5).
118
Application of ENM for the study of protective coatings
The potential and current fluctuations of the time records can also be distinguished. As
mentioned in section 3.3.3, it is usually observed that the fluctuations of the potential
decrease while the current fluctuations increase as the coating deteriorates [2]. These
fluctuations are used for the calculation of the noise resistance (Rn) and the spectral
noise resistance (Rsn) as is shown further on in this section.
8.00E-07
-6.10E-01
-6.15E-01
5.50E-07
I [Acm -2]
E [V]
-6.20E-01
-6.25E-01
-6.30E-01
3.00E-07
5.00E-08
-6.35E-01
-2.00E-07
-6.40E-01
0
50
100
150
200
50
Figure 1
100
150
200
Time [s]
Time [s]
Time records of potential and current (raw data) of a bitumen coating after more
than 1 year of immersion in artificial concentrated rainwater.
Drift removal
If the drift consists of a linear change with time, it can be removed simply by
subtracting the linear regression line from the data, a common method of treating drift
[7,9,10].
A more complex technique for drift removal is the subtraction of the central moving
average (CMA) [9-11]. Here, the average of a group of points symmetrically around a
data value is subtracted from that data value according to:
mn =
1
( x n p + ... + x n 1 + x n + x n +1 + ... + x n + p )
2 p +1
(1)
and:
y n = x n mn
(2)
Here xn denotes the raw signal at time n, mn is the result of the CMA procedure
(consisting in averaging the segment from point n-p to n+p) and yn represents the
CMA-filtered value.
An example of detrending using the CMA is shown in Figure 2. The left graph shows
the raw time record of the current and the CMA of this signal for p = 2. The right
graph shows the detrended data where the CMA is subtracted from the raw data. Note
119
Chapter 6
7.00E-07
I (raw)
7.00E-07
6.00E-07
CMA p=2
6.00E-07
5.00E-07
5.00E-07
4.00E-07
4.00E-07
3.00E-07
3.00E-07
I [Acm -2]
I [Acm -2]
that the detrended data shows reduced fluctuations and has a mean coupling current of
zero.
2.00E-07
1.00E-07
2.00E-07
1.00E-07
0.00E+00
0.00E+00
-1.00E-07
-1.00E-07
-2.00E-07
-2.00E-07
-3.00E-07
-3.00E-07
25
45
65
85
Time [s]
Figure 2
I (raw) -CMA p=2
105
125
25
45
65
85
105
125
Time [s]
Left: current time record (raw data) and the CMA of this signal for p = 2. Right:
result of detrending (i.e. the CMA subtracted from the current time record).
The dataset shown Figure 1 has been detrended using subtraction of the linear
regression line and subtraction of the CMA for p = 250, 50, 25, 10, 5, 2, 1. After
detrending, data were analysed both in the time and the frequency domain.
From equation 1 it can be derived that for large values of p, the range over which the
CMA can be calculated is considerably smaller than the measurement period. In order
to compare the results, the range over which all noise parameters are calculated should
be kept similar and is thus limited by that for p = 250. In addition, the used FFT
algorithm requires that the number of input values must be a power of 2. Within the
available range 210 (1024) was the maximum number of input values that could be
used for analysis in the frequency domain. Consequently, all noise parameters were
based on the measurement results obtained between 26 and 132 seconds.
Analysis in the time domain
The influence of detrending upon analysis in the time domain is shown in Figure 3.
The left graph shows the influence of detrending upon the standard deviations of
potential (V) and current (I). The right graph shows the influence upon the noise
resistance Rn (Rn = V/I).
For this dataset, linear detrending has more influence on V than on I. Indeed, the
time record of the potential (see Figure 1) shows a linear trend. Removing this trend
decreases the calculated V value and, therefore, Rn decreases.
Detrending by subtraction of the CMA clearly influences the calculated V and I
values. Both V and I decrease when smaller values of p are used. How this
influences Rn depends on which parameter is affected the most. For all p values,
120
1.50E-03
7.50E-08
1.00E-03
5.00E-08
5.00E-04
E
V
I
I
0.00E+00
Li
ne Ra
ar w
de da
t
CM tre a
n
A din
p
g
CM =
A 250
CM p =
A 50
CM p =
A 25
p
CM =
1
A 0
CM p =
A 5
CM p =
A 2
p
=
1
0.00E+00
Figure 3
2.50E-08
2.40E+04
2.30E+04
2.20E+04
2.10E+04
Ra
Li
w
ne
da
ar
ta
de
tre
CM ndi
ng
A
p
=
25
CM
0
A
p
=
CM
50
A
p
=
CM
25
A
p
=
10
CM
A
p
=
CM
5
A
p
=
CM
2
A
p
=
1
1.00E-07
Rn [cm-2]
2.00E-03
I [Acm -2]
V [V]
except for p = 1, V was affected significantly more than I. Consequently, those Rn

values were lower than that calculated from the raw data.
Influence of detrending upon calculated noise parameters V, I and Rn.
Analysis in the frequency domain

The influence of detrending upon analysis in the frequency domain is shown in Figure
4. The top graphs show the FFT spectra of potential (VFFT) and current fluctuations
(IFFT). The graphs in the middle show the high-frequency (HF) and low-frequency (LF)
limits of the potential and current spectra (10-point average). The bottom graphs show
the spectral noise resistance (Rsn) and the HF and LF limits of Rsn.
From the top and middle graphs of Figure 4 it becomes clear that detrending mainly
affects the low frequency part of the signal. The HF-limit of the VFFT, for example,
changes relatively little from 6.1610-5 (raw data) to 8.1910-5 V (CMA p = 1).
However, the LF-limit drops considerably from 2.4610-4 to 6.2610-7 V. The LFlimits of both VFFT and IFFT decrease when smaller values of p are used.
The influence of detrending upon Rsn depends on the relative degree to which VFFT and
IFFT are affected. The bottom graphs of Figure 4 show that detrending mainly affects
the low-frequency part of Rsn. The right bottom graph shows that for linear detrending
and detrending by subtraction of the CMA for p = 10 and p = 5 the LF-limit of Rsn is
higher than that of that calculated for the raw data. In these cases, detrending has
relatively more effect on IFFT than on VFFT, resulting in higher LF-limits of Rsn.
121
Chapter 6
1.00E-03
1.00E-07
raw
lin. tr.
1.00E-05
1.00E-06
IFFT [Acm -2Hz -1]
p=250
p=50
p=5
p=1
FFT
VFFT
[VHz-1]
[VHz -1]
1.00E-04
1.00E-07
1.00E-08
lin. tr.
1.00E-09
p=250
raw
1.00E-10
p=50
p=5
p=1
0.01
0.1
1.00E-11
0.001
10
0.01
1.00E-05
1.00E-05
1.00E-07
1.00E-06
1.00E-09
1.00E-07
1.00E-11
1.00E-05
1.00E-05
1.00E-07
1.00E-06
1.00E-09
1.00E-07
1.00E-11
Li
ne Ra
ar w
de da
t
CM tre n a
A din
g
p
CM =
25
A
0
CM p =
A 50
CM p =
A 25
p
CM = 1
A 0
CM p =
A 5
CM p =
A 2
p
=
1
Rsn
[cm--22]]
sn [cm
R
1.00E+06
1.00E+06
raw
p=250
lin. tr.
p=1
Rsn HF
Rsn LF
1.00E+05
1.00E+04
1.00E+03
1.00E+02
1.00E+03
1.00E+02
0.001
0.01
0.1
Frequency [Hz]
10
Ra
Li
w
ne
da
ar
ta
de
tre
CM ndi
ng
A
p
=
25
CM
0
A
p
=
CM
50
A
p
=
CM
25
A
p
=
10
CM
A
p
=
CM
5
A
p
=
CM
2
A
p
=
1
Rsn [cm -2]
I LF
1.00E-04
1.00E+07
1.00E+07
Figure 4
1.00E-03
Li
ne Ra
ar w
de da
t
CM tre n a
A din
p
g
CM =
A 250
CM p =
A 50
CM p =
A 25
p
CM =
1
A 0
CM p =
A 5
CM p =
A 2
p
=
1
Average
[V]
FFT[V]
Average EVFFT
1.00E-04
1.00E+04
10
E
V LF
Average I FFT [Acm-2]
II HF
HF
1.00E+05
1.00E-03
1.00E-03
E
V HF
HF
FFT
Average
[V]
Average E
V FFT[V]
1.00E-03
0.1
Frequency [Hz]
Frequency [Hz]
Average I FFT [Acm -2]
1.00E-08
0.001
Influence of detrending upon spectral noise analysis. Top graphs: FFT spectra of
potential (VFFT) and current fluctuations (IFFT). Middle graphs: high-frequency
(HF) and low-frequency (LF) limits of potential and current spectra (10-point
average). Bottom graphs: spectral noise resistance (Rsn) and the HF- and LFlimits of Rsn. Note: some FFT spectra were omitted for visual clarity.
Conclusions
Detrending of ENM data can have a large effect upon the calculated noise parameters.
If detrending affects potential and current data to a comparable degree, Rn and Rsn may
appear unaffected. This does however not imply that the used detrending method is
appropriate. Therefore one should always look at the effect of detrending upon both
potential and current fluctuations.
122
Though there is no definite answer to what detrending method should be used, CMA
detrending using small p values should be avoided to prevent the loss of lowfrequency noise.
6.4 Asymmetric electrodes
Introduction
In the majority of published work on electrochemical noise it is assumed that both
working electrodes are identical. In practice, for measurements with coated substrates,
this requirement seems hard to fulfil [7]. To find out how asymmetric electrodes affect
the calculated noise resistance (Rn), three experimental conditions have been
evaluated.
Experimental
Two nominally identical coated electrodes were exposed to 3 wt% NaCl. A saturated
calomel electrode (SCE) was used as a reference electrode (RE). The working
electrodes (WE) consisted of a 25-m polyester coilcoat on galvanized steel. The
advantage of coil-coated panels is that coating is applied under controlled conditions,
resulting in a low sample-to-sample variation.
After stabilisation of the open circuit potential (OCP), the noise measurement was
started. After about 1900 seconds, one of the coated electrodes (WE1) was scratched,
exposing the bare zinc layer. After about 2750 seconds the second coated electrode
(WE2) was scratched.
The potential and current as a function of time are shown in Figure 5. The calculated
noise parameters after linear detrending are shown in Table 1.
123
Chapter 6
8.00E-06
-9.00E-01
-9.50E-01
6.00E-06
I [Acm-2]
I [Acm -2]
E [V]
-1.00E+00
-1.05E+00
5.00E-09
4.00E-06
0.00E+00
-5.00E-09
1000
2.00E-06
2000
t [s]
-1.10E+00
0.00E+00
-1.15E+00
-1.20E+00
-2.00E-06
0
1000
2000
3000
4000
1000
2000
t [s]
3000
4000
t [s]
Figure 5
Time records of potential and current (raw data). 0 indicates two intact working
electrodes; 1 indicates one electrode (WE1) scratched, 2 indicates both
electrodes scratched (WE1+WE2).
Table 1
Standard deviation of potential and current fluctuations and the corresponding

noise resistance of a 25-m polyester coilcoat on galvanised steel. Linear trends
were removed prior to calculations.
Both electrodes
undamaged
range: 1000-1500 s
V [V]
-2
-2
Rn [cm ]
WE1 + WE2 scratched
range: 2250-2750 s
range: 3000-3500 s
-3
3.8710
-11
3.7810
1.6010
I [Acm ]
WE1 scratched
3.5010
4.5910
-4
5.0410
-11
3.2110
1.5710
1.0310
-4
-9
5
From Table 1 it becomes clear that after scratching WE1, V decreases while I only
increases slightly. Consequently, the calculated Rn value decreases.
After scratching WE2 V increases slightly but I increases drastically. Accordingly,
Rn decreases drastically.
Conclusions
From the experiment with asymmetric electrodes it can be concluded that the noise
resistance is mainly determined by the electrode with highest resistance. This result is
in agreement with the theoretical work by Bertocci et al. [12]. However, as Bertocci
pointed out, this should not be taken as a general rule for all situations.
124

6.5 EIS versus ENM
Introduction
Given that EIS is the more established technique for the investigation of protective
coatings, it is worthwhile to compare measurement results of EIS with those obtained
by ENM. In this section, results of five industrial coatings are compared and discussed.
Experimental
The five industrial coating that were selected for the measurements are shown in Table
2. All coatings were applied to steel Q-panels and immersed in artificial concentrated
rainwater (see section 5.2 for the composition) for extended periods.
Table 2
Coating
A
B
C
D
E
Coating types and the duration of immersion in artificial concentrated rainwater.

Type
Water-based epoxy
Bitumen
Novolac
Duration immersion [d]

385
222
396
396
396
First, EIS measurements were performed on all coated electrodes. After this, potential
and current fluctuations of the electrode pairs were recorded using the ENM setup.
ENM data were collected for 212 seconds at an average rate of 9.7 points per second.
Only linear detrending has been used to treat the ENM data. FFT spectra were
calculated from 211 (2048) data points.
Ecoup, Icoup (calculated from raw data) and V and I (calculated after subtraction of the
linear trend from the data) are shown in Table 3.
Coatings A to D showed Ecoup values that were close to the corrosion potential of bare
steel in seawater (approximately -650 mV vs. SCE [8]). The Ecoup value of the coating
E was higher due to the high resistance of the coating.
With the exception of coating D, all coatings showed Icoup values between 2 and 3
nAcm-2. Coating D showed a considerably higher Icoup value of 198 nAcm-2.
For coatings A to D V ranged from 1.82 to 3.10 mV. Coating E showed a higher
standard deviation of 9.87 mV.
The standard deviation of the current shows a similar pattern as that of Icoup. With the
exception of coating D, all coatings showed I values between 0.1 and 0.8 nAcm-2.
Coating D showed a considerably higher I value of about 81.7 nAcm-2.
125
Chapter 6
Table 3
Ecoup and Icoup (raw data) of the five industrial coatings.
Ecoup [V]
-2
Coating A
Coating B
Coating C
Coating D
Coating E
-0.625
-0.552
-0.648
-0.629
-0.254
-9
2.8910
-3
3.0310
-10
7.9810
Icoup [Acm ]
2.8410
V (-lin. tr.) [V]
3.1010
-2
I (-lin. tr.) [Acm ]
1.8410
-9
-9
2.3410
-3
2.3110
-10
3.1810
-7
1.9810
2.1110
-3
1.8210
-10
8.1710
-9
-3
9.8710
-8
2.3810
-3
-10
Figure 6 shows a comparison of the impedance spectra, the noise resistance and the
spectral noise resistance of coatings C and D. Rn and Rsn were calculated after linear
detrending.
For coating C, the impedance |Z| of the coated electrodes C1 and C2 (at the Nyquist
frequency fN of the ENM, see section 3.3.3) are roughly one magnitude larger than the
calculated Rn and Rsn values.
Coating D shows much more agreement between the impedance spectra and the noise
parameters.
A comparison of |Z| at fN of coatings A to E with the noise parameters are shown in the
left graph of Figure 7. In all cases, the average |Z| at fN is larger than Rn and Rsn. Two
possible explanations have been found for this discrepancy.
Mansfeld et al. observed that initially Rsn plots of immersed coatings were
significantly lower than the impedance plots. It was stated that instrumentation limits
caused this difference. As the resistance of the coating dropped, Rsn approached the
impedance plots [2].
According to Bertocci et al. the noise resistance Rn is equal to the impedance if the
Bode magnitude plot of |Z| reaches the DC limit in the measurement frequency
bandwidth [12]. This condition was only met for coating D.
In the left graph of Figure 7 |Z| at fN is plotted against the Rn. Apparently, a correlation
exists between the two parameters. Although a numerical correlation between Rn , Rsn
and the DC limit of the impedance for coated electrodes can be only conditionally
reached, it has been confirmed by other workers that there exists a good trend
correlation between both parameters [in:14].
126

Coating C
Coating D
1.00E+06
1.00E+07
|Z| C2
1.00E+06
Rn
Rsn
1.00E+05
1.00E+04
|Z| C1
1.00E+03
1.00E+02
1.00E+01
1.00E+00
1.00E-03
1.00E-01
1.00E+01
1.00E+03
Rn @fs, Rsn , |Z| [cm-2]
Rn @fs, Rsn , |Z| [cm-2]
1.00E+08
1.00E+04
1.00E+03
1.00E+01
1.00E+03
1.00E+05
Frequency [Hz]
Rn (-lin. tr.)
1.00E+09
1.00E+08
1.00E+07
1.00E+06
1.00E+05
1.00E+04
1.00E+03
1.00E+02
1.00E+01
1.00E+00
6.00E+08
|Z|average @ fN
|Z|average @ fN [cm-2]
Rsn @ fN (-lin. tr.)
5.00E+08
4.00E+08
3.00E+08
2.00E+08
1.00E+08
in
g
oa
t
C
oa
t
in
g
B
C
oa
t
in
g
in
g
oa
t
in
g
0.00E+00
oa
t
C
1.00E-01
Comparison of impedance spectra (|Z|), noise resistance (Rn) and spectral noise
resistance (Rsn). The numbers 1 and 2 indicate the impedance spectra for each
coated electrode. Rn is shown at the sampling frequency.
Figure 6
Rn, Rsn@fN, |Z|@fN [cm-2]
|Z| D1
1.00E+01
1.00E+00
1.00E-03
1.00E+05
Rn
Rsn
1.00E+02
Frequency [Hz]
Figure 7
|Z| D2
1.00E+05
0.00E+00 1.00E+07 2.00E+07 3.00E+07 4.00E+07 5.00E+07

-2
(-lin. det.)
[cm-2]
RRnn (-lin.
tr.) [cm
Left: Comparisons of |Z| at fN of coatings A to E with Rn and Rsn. Right: |Z| at fN

versus Rn.
Conclusions
The parameters Icoup and I vary drastically with the degree of coating deterioration.
For defective coating E both values roughly are two orders of magnitude larger than
those of the more resistive coatings.
The parameters Ecoup and V are much less sensitive indicators for coating
deterioration. Ecoup was close to the corrosion potential of the substrate, except for the
resistive coating E. Parameter V varied little between coating A to D. For coating E
V was fairly higher.
In all cases, the |Z| at fN is larger than Rn and Rsn. This difference increases with the
impedance of the electrodes. However, a good correlation was obtained between the
impedance and the noise resistance.
127
Chapter 6
In this chapter, the usability of ENM for the evaluation of protective coatings has been
examined.
Fundamental questions about when variation of potential or current should be regarded
as drift and when it should be regarded as electrochemical noise remain unanswered.
It has been shown that detrending of ENM data can have a large effect upon the
calculated noise parameters. However, when detrending affects potential and current
data to a comparable degree, Rn and Rsn may appear unaffected. This does not imply
that the used detrending method is appropriate. Therefore one should always look at
the effect of detrending upon both potential and current fluctuations.
detrending using small p values should be avoided to prevent the loss of lowfrequency noise.
Concerning the experiment with asymmetric electrodes, it can be concluded that in this
particular case the noise resistance is mainly determined by the electrode with highest
resistance. Nevertheless, this outcome should not be taken as a general rule for all
situations.
For five industrial coatings, examined with both EIS and ENM, it was found that
modulus of the impedance at the Nyquist frequency was always larger than the noise
parameters Rn and Rsn. This difference was found to increase with the impedance of
the electrodes. Despite this difference, a good correlation was obtained between the
In general, the parameters Icoup and I vary drastically with the degree of coating
deterioration. The parameters Ecoup and V are much less sensitive indicators for
coating deterioration.
On the basis of the correlation that has been found between Rn, Rsn and the DC limit of
the impedance, it is possible to use ENM to monitor changes of protective properties
of coatings.
128
6.7 References
1
Le Thu, Q., Bierwagen G.P., S. Touzain, EIS and ENM measurements for three different
organic coatings on aluminium, Progress in Organic Coatings, 42 (2001) 179-187.
Mansfeld, F., L.T. Han, C.C. Lee, C. Chen, G. Zhang, H. Xiao, Analysis of
Electrochemical Impedance and Noise Data for Polymer Coated Metals, Corrosion
Science, 39 (1997) 255-279.
Mills, D.J., S. Mabbutt, Investigation of defects in organic anti-corrosive coatings using

electrochemical noise measurement, Progress in Organic Coatings, 39 (2000) 41-48.
Bierwagen, G.P., X. Wang, D.E. Tallman, In situ study of coatings using embedded
electrodes for ENM measurements, Progress in Organic Coatings, 46 (2003) 163-175.
Dawson, J.L., Electrochemical Noise Measurement: The Definitive In Situ Technique

for Corrosion Applications? In: Electrochemical Noise Measurements for Corrosion
Applications, ASTM STP 1277, J.R. Kearns, J.R. Scully, P.R. Roberge, D.L. Reichert,
J.L. Dawson (Editors), American Society for Testing and Materials, 1996, 3-35.
Kearns, J.R., J.R. Scully, P.R. Roberge, D.L. Reichert, J.L. Dawson (Editors),
Overview In: Electrochemical Noise Measurements for Corrosion Applications, ASTM
STP 1277, American Society for Testing and Materials, 1996, ix-xvii.
Cottis, R.A., Interpretation of Electrochemical Noise Data, Corrosion, 57 (2001) 265284.
Metals Handbook, Volume 13 - Corrosion, ASM International, 9th edition, ISBN

0871700190.
Mansfeld, F., Z. Sun, C.H. Hsu, A. Nagiub. Concerning trend removal in

electrochemical noise measurements, Corrosion Science, 43 (2001) 341-352.
10 Lowe, A.M., Estimation of Electrochemical Noise Impedance and Corrosion Rates from
Electrochemical Noise Measurements, Ph.D. Thesis, Curtin University of Technology,
2002.
11 Bertocci, U., F. Huet, R.P. Nogueira, P. Rousseau, Drift Removal Procedures in the
Analysis of Electrochemical Noise, Corrosion 58 (2002) 337-347.
129
Chapter 6
12 Bertocci, U., C. Gabrielli, F. Huet, M. Keddam, Noise Resistance Applied to Corrosion
Measurements- I Theoretical Analysis, Journal of the Electrochemical Society, 144
(1997) 31-37.
13 Huet, F., A. Bautista, U. Bertocci, Listening to Corrosion, The Electrochemical Society,
Interface - Winter 2001, 40-43.
14 Greisiger, H., T. Schauer, On the interpretation of the electrochemical noise data for
coatings, Progress in Organic Coatings, 39 (2000) 31-36
130
7 General conclusions
In this thesis, two ways to determine the suitability of a protective coating for a
specific environment have been addressed.
The first way involves the use of so-called accelerated tests. Since the in-service
degradation of modern protective coating systems is extremely slow, accelerated
testing aims to reproduce, in a much shorter time than in the field, natural degradation
processes without changing the degradation mechanism.
In Chapter 2 common (accelerated) tests and evaluation methods for protective
coatings have been reviewed.
Based upon this review, the following conclusions were drawn:
The most reliable way of studying the suitability of a coating for a specific
substrate and environment is to actually expose coated substrates to the
environment in which the coating will ultimately be applied. Evidently, this
approach is very time-consuming.
The standard salt spray test should not be used for service life predictions or
even performance ranking of coated systems.
Cyclic testing imparts a more realistic stress onto the coating system in an
accelerated manner.
The combined cyclic salt fog/UV tests show a still improved reproduction of
coating performance ranking and failure modes observed in practice.
All (accelerated) tests, including outdoor exposures, are relative tests. They do
not give absolute predictions of how many years a material will last in actual
service. They merely provide an indication of relative performance.
Different outdoor service conditions will lead to different coating durabilitys.
Some materials may perform particularly well in one environment but may
produce relatively poor results in other environments. It is therefore not to be
expected that a single test method will be representative for all outdoor
exposures.
Attempts to increase the acceleration factor by intensifying the laboratory
testing conditions will result in unnatural changes in degradation mechanism
and accounts for a less realistic comparison with natural weathering.
Standardized visual evaluation techniques are subjective. The results heavily
depend on the ability of the operator to translate a visual observation into a
performance rating. In case of high-performance coatings that almost show no
131
Chapter 7
visual signs of deterioration after weathering, discrimination between systems is

even impossible.
from an artificially made scribe. Scribing panels prior to the test facilitates
assessment of the damage-protective properties of the coating/substrate system
and not the important barrier properties of the system.
The use of electrochemical evaluation techniques, in particular EIS
(electrochemical impedance spectroscopy), has been shown to be very useful.
EIS not only provides more precise results in a short time but the obtained data
can give indications on the actual corrosion mechanisms. Consequently, the
sensitivity of this technique allows the total test time to be shortened.
Another way to determine the suitability of a protective coating utilizes sensitive
electrochemical techniques such as EIS to allow early detection of relevant
degradation processes under non-accelerated conditions.
In chapter 4, the results of outdoor exposure of two model coatings, periodically
measured with EIS, have been presented. Measurement results have been analysed and
related to literature sources to obtain insight in degradation processes under natural
conditions.
It was concluded that EIS appears to be very useful for the study of coating
degradation mechanisms for a number of reasons.
In order to draw useful conclusions from EIS data, the construction of equivalent
circuits must be carried out with care. By fitting the MPEC (most probable equivalent
circuit) to the measurement data, important parameters such as coating resistance and
capacitance, the charge transfer resistance and the electrochemical double layer
capacitance can be determined quantitatively.
signs of degradation, EIS clearly has an advantage as various processes related to
Despite the fact that the time-dependence of the impedance parameters showed quite
measurement results. Together with other sources from literature, measurement results
were related to the unified model for coating degradation presented by Nguyen and
Hubbard.
Especially for industrial applications, coatings are required to provide excellent
corrosion protection for a long time. In case of extreme conditions, it is imperative to
132
General conclusion
select a coating system that is most suitable for site-specific conditions. However,
there is no generally accepted test method for selection of protective coatings.
Therefore, it has become increasingly important to develop methods for the rapid
evaluation of the corrosion protective properties of these coatings.
There has been considerable effort to develop evaluation methods for organic coatings
that are numerical, reproducible and accurate. It was found that -in particular- EIS, has
demonstrated to be very useful in this respect.
5 the electrochemical properties of more than 100 industrial coatings have been
determined using a benchmark EIS measurement protocol.
From the overall results and the variation between replicates it became clear that large
variability of coating parameters for each coating type should not be associated to the
accuracy of EIS measurements, but rather to actual differences in protective properties
of the tested coatings.
The measurement results confirmed that the benchmark EIS measurement protocol is
an objective method to rapidly examine the corrosion protective mechanisms of
coatings as well as other properties such as the mixing ratio and chemical resistance.
However, EIS is not a standalone technique. Only with the aid other techniques
(measurement of the open circuit potential, visual observation, etc.) and sources from
literature meaningful conclusions can be drawn upon the protective mechanisms of the
coatings.
the protective properties of coatings. Recently, the application of ENM
(electrochemical noise measurement) for the same purpose has become of interest.
However, the application of ENM for corrosion studies remains a controversial
subject. In chapter 6, results of some ENM experiments have been presented and
discussed.
Based on these results, conclusions have been drawn upon the usability of ENM for
evaluation of protective coatings.
Fundamental questions about when variation of potential or current should be regarded
as drift and when it should be regarded as electrochemical noise remain unanswered.
It has been shown that detrending of ENM data can have a large effect upon the
calculated noise parameters. However, when detrending affects potential and current
data to a comparable degree, the noise resistance (Rn) and the spectral noise resistance
(Rsn) may appear unaffected. This does not imply that the used detrending method is
133
Chapter 7
appropriate. Therefore one should always look at the effect of detrending upon both
potential and current fluctuations.
detrending (subtraction of the central moving average) using small p values (p is
related to the number of points over which the CMA is calculated) should be avoided
to prevent the loss of low-frequency noise.
Concerning the experiment with asymmetric electrodes, it can be concluded that in this
particular case the noise resistance is mainly determined by the electrode with highest
resistance. Nevertheless, this outcome should not be taken as a general rule for all
situations.
For five industrial coatings, examined with both EIS and ENM, it was found that
modulus of the impedance at the Nyquist frequency was always larger than the noise
parameters Rn and Rsn. This difference was found to increase with the impedance of
the electrodes. Despite this difference, a good correlation was obtained between the
In general, the mean coupling current (Icoup) and the standard deviation of the current
(I) vary drastically with the degree of coating deterioration. The mean potential of the
coupled electrodes (Ecoup) and the standard deviation of the potential (V) are much
less sensitive indicators for coating deterioration.
On the basis of the correlation that has been found between Rn, Rsn and the DC-limit of
the impedance, it is possible to use ENM to monitor changes of protective properties
of coatings.
134
Summary
Summary
Corrosion of metals has an enormous economic impact. Fortunately, the rate at which
corrosion processes take place can be reduced, often to a point where, at least
temporarily, corrosion is virtually non-existent.
One of these, corrosion-control methods is barrier protection. The concept of barrier
protection is to separate the metal substrate from the corrosive environment. These
barriers are frequently applied in the form of coatings.
In this thesis, various properties of protective coatings will be discussed. Therefore,
some general coating characteristics are introduced in chapter 1.
The objectives of the research described in this Ph.D. thesis are two-fold.
From an industrial perspective, it is economically attractive to develop methods for the
mitigation of corrosion damage of coated structures.
From a scientific point of view, it is interesting to acquire a fundamental understanding
of how protective coatings function and how these coatings deteriorate with time.
Two ways to determine the suitability of a coating for a specific environment can be
distinguished.
The first approach involves the use of so-called accelerated tests. Accelerated testing
aims to reproduce, in a much shorter time than in the field, natural degradation
processes without changing the degradation mechanism.
The second approach utilizes sensitive electrochemical techniques such as
Electrochemical Impedance Spectroscopy (EIS) and Electrochemical Noise
Measurement (ENM) to allow early detection of relevant degradation processes under
non-accelerated conditions.
Both approaches will be addressed in this thesis.
Chapter 2 reviews the most widely used (accelerated) tests for protective coatings.
The associated evaluation methods are also discussed.
The most reliable way of studying the suitability of a coating for a specific substrate
and environment is to actually expose coated substrates to the environment in which
the coating will ultimately be applied. Evidently, this approach is too time-consuming.
Therefore, many researchers have attempted to speed up the natural coating
degradation process by increasing the physical and chemical stresses in accelerated
tests.
135
Summary
The most widely used accelerated test for coating evaluation is the salt spray test. This
has been true despite severe criticisms this type of testing has received. The constant
stress imposed during the test is frequently mentioned as a cause of the poor
correlation with field performance. The prescribed spray electrolyte (5 wt% NaCl) and
the high temperature (35 C) are also held accountable for the poor correlation.
To overcome the deficiencies of continuous salt spray tests, cyclic weathering tests
were developed. Incorporation of cycling steps seems intuitively justified, considering
that coatings exposed to the outdoor environment undergo similar effects on a frequent
basis. Indeed, these tests produce failures more representative of field results. The
development of more advanced cyclic tests, resulted in a further improvement of the
correlation with field performance.
Coating degradation is usually evaluated using standardized methods. The majority of
these methods are based on visual observation, rendering them rather subjective. The
use of electrochemical techniques, in particular EIS, has been shown to be very useful.
EIS not only provides results in a short time but the obtained data can give indications
on the actual corrosion mechanisms. In addition, corrosion and coating damage may be
determined prior to its visual manifestation.
In literature, many investigations into the correlation of accelerated tests with outdoor
exposure have been described. A review of extensive and long-term studies revealed
that:
from an artificially made scribe. Acceleration by scribing panels prior to the
test, facilitates assessment of the damage-protective properties of the
coating/substrate system and not the barrier properties of the system.
High correlations found for some tests were mainly determined by outliers with
a relatively strong scribe creep.
In one research programme, specific combinations of substrates, tests and
exposure sites showed correlation coefficients that exceeded 0.90, but the
variation of coefficients was rather large. From this exhaustive study it was
concluded that none of the accelerated tests assessed in this study could reliably
predict the medium and long-term durability of coil-coated materials.
techniques. In this thesis, EIS and ENM are used to this purpose. Chapter 3 provides a
theoretical background of these techniques.
136
Summary
Chapter 4 discusses the results of outdoor exposure of model coating systems,

periodically measured with EIS. Panels with the model systems were mounted at an
exposure rack in a marine environment (Den Helder, The Netherlands) for 2 years.
Before each measurement, panels were transferred to the laboratory and prepared for
the measurement. Impedance data were analyzed by constructing equivalent circuits in
accordance with MPEC (most probable equivalent circuit) guidelines.
signs of degradation, EIS clearly shows an advantage as various processes related to
Despite the fact that the time-dependence of the impedance parameters showed quite
measurement results.
Measurement results are related to the unified model for coating degradation presented
by Nguyen and Hubbard.
5, a benchmark measurement protocol is introduced. More than 100 industrial coatings
have been measured in compliance with this protocol. For each coating system EIS
measurements were performed after 0, 1, 7 and 21 days of immersion.
From the overall results and the variation between replicates it became clear that large
variability of coating parameters for each coating type should not be associated to the
accuracy of EIS measurements, but rather to actual differences in protective properties
of the tested coatings.
The measurement results confirmed that the benchmark EIS measurement protocol is
an objective method to rapidly examine the corrosion protective mechanisms of
coatings as well as other properties such as the mixing ratio and chemical resistance.
However, EIS is not a standalone technique. Only with the aid other techniques
(measurement of the open circuit potential, visual observation, etc.) and sources from
literature meaningful conclusions can be drawn upon the protective mechanisms of the
coatings.
the protective properties of coatings. Recently, the application of ENM for the same
purpose has become of interest. In chapter 6, the usability of ENM for the
characterisation of protective coatings is studied.
137
Summary
Frequently, time records of noise data show a certain drift, which affects the calculated
noise parameters. Therefore, drift has to be eliminated prior to data analysis. Two
methods for detrending have been examined. However, fundamental questions about
when variation of potential or current should be regarded as drift and when it should
be regarded as electrochemical noise remain unanswered.
In the majority of published work on electrochemical noise it is assumed that both
working electrodes are identical. For measurements with coated substrates, this rarely
the case. From an experiment where the influence of asymmetric electrodes was
examined it can be concluded that -in this particular case- the noise resistance is
mainly determined by the electrode with the highest resistance. Nevertheless, this
outcome should not be taken as a general rule for all situations.
Five industrial coatings were examined with both EIS and ENM. It was found that the
modulus of the impedance at low frequencies was always larger than the noise
parameters. This difference was found to increase with the impedance of the
electrodes. Despite this difference, a good correlation was obtained between the
impedance and the noise resistance. On the basis of this correlation, it is possible to
use ENM to monitor changes of protective properties of coatings.
The final chapter of this thesis summarises the conclusions drawn in the preceding
chapters.
138
Samenvatting
Samenvatting
Corrosie van metalen heeft enorme economische consequenties. Gelukkig kan de
snelheid van corrosieprocessen worden gereduceerd, vaak tot een niveau waar, in ieder
geval tijdelijk, corrosie nagenoeg niet optreedt.
En van deze corrosiebeschermingsmethoden berust op barrirewerking. Het concept
van barrirewerking is om het metalen substraat fysiek te scheiden van het corrosieve
milieu. Deze barrires worden veelvuldig toegepast in de vorm van coatings.
In dit proefschrift worden verscheidene eigenschappen van beschermende coatings
besproken. Daarom worden in hoofdstuk 1 eerst algemene coatingeigenschappen
gentroduceerd.
De doelstellingen van het onderzoek beschreven in dit proefschrift zijn tweeledig.
Vanuit een industrieel perspectief is het economisch aantrekkelijk om methoden te
ontwikkelen om corrosieschade van gecoate constructies tegen te gaan.
Vanuit wetenschappelijk oogpunt is het interessant om fundamenteel inzicht te krijgen
in hoe beschermende coatings functioneren en hoe deze coatings degraderen als
functie van de tijd.
Er kunnen twee manieren worden onderscheiden om te bepalen of een coating geschikt
is voor toepassing in een bepaald milieu.
De eerste benadering betreft het gebruik van zogeheten versnelde testen. In versnelde
testen wordt geprobeerd de natuurlijke degradatie versneld te laten optreden zonder dat
daarbij het degradatiemechanisme veranderd.
Bij de tweede benadering wordt gebruik gemaakt van gevoelige elektrochemische
technieken als elektrochemische impedantie spectroscopie (EIS - Electrochemical
Impedance Spectroscopy) en elektrochemische ruismeting (ENM - Electrochemical
Noise Measurement) om in een vroegtijdig stadium relevante degradatieprocessen vast
te stellen onder niet-versnelde omstandigheden.
Beide benaderingen worden behandeld in dit proefschrift.
In hoofdstuk 2 wordt een overzicht gegeven van de meest gebruikte (versnelde)
testen voor beschermende coatings. De gerelateerde evaluatiemethoden worden ook
besproken.
De meest betrouwbare manier om te bepalen of een coating geschikt is voor toepassing
in een bepaald milieu is door gecoate substraten te exposeren in een soortgelijk milieu.
Het is duidelijk dat deze benadering te tijdrovend is. Daarom proberen vele
139
Samenvatting
onderzoekers het natuurlijke degradatieproces te versnellen door fysische en

chemische stressfactoren te verhogen in versnelde testen.
De meest wijdverbreide versnelde test voor coatingevaluatie is de zoutsproeitest. Dit is
het geval ondanks de zware kritiek die deze manier van testen heeft ontvangen. De
constante stress die wordt opgelegd tijdens de test wordt vaak genoemd als oorzaak
voor de slechte correlatie met de praktijk. De voorgeschreven elektrolyt (5 % NaCl)
waarmee wordt gesproeid en de hoge temperatuur (35 C) worden ook gezien als
factoren die een negatieve invloed hebben op de correlatie.
Om de nadelen van de constante zoutsproeitest weg te nemen, zijn cyclische
verweringstesten ontwikkelt. Het opnemen van cyclische stappen is gevoelsmatig juist,
aangezien coatings die buiten worden toegepast vaak aan gelijksoortige wisselende
condities worden blootgesteld. Het is inderdaad gebleken dat deze testen leiden tot
faalmechanismen die dichter bij de praktijk liggen. De ontwikkeling van meer
geavanceerde cyclische testen, resulteerde in een verdere verbetering van de correlatie
met de coatingprestaties in praktijk
Coatingdegradatie wordt normaal gevalueerd met standaardmethoden. De
meerderheid van deze methoden is gebaseerd op visuele inspectie, waardoor deze
methoden subjectief zijn. Het toepassen van elektrochemische technieken, EIS in het
bijzonder, kan erg waardevol zijn. EIS levert niet alleen snel resultaten, maar daarnaast
kan de verkregen data indicaties geven over het corrosiemechanisme. Bovendien kan
het optreden van corrosie en schade aan de coating worden vastgesteld voordat dit
visueel waarneembaar wordt.
In de literatuur zijn er vele onderzoeken beschreven over de correlatie tussen versnelde
testen en buitenexpositie. Uit een overzicht van uitgebreide en lange-termijn studies is
gebleken dat:
De prestatie van gecoate panelen veelvuldig wordt bepaald door de aantasting
(a.g.v. corrosie) vanuit een aangebrachte kras in de coating. Versnelling door
het aanbrengen van een kras voorafgaand aan de test, faciliteert het vaststellen
van de beschermende eigenschappen ter plaatse van defecten. De barrireeigenschappen worden niet op deze wijze vastgesteld.
De hoge correlaties die soms werden gevonden voor sommige testen werden
voornamelijk bepaald door uitbijters met een relatief hoge onthechting vanuit
de kras.
In n onderzoeksprogramma werden voor specifieke combinaties van
substraten, testen en expositielocaties correlatiecofficinten van meer dan 0.90
gevonden. De variatie van cofficinten was echter erg groot. Op basis van deze
omvangrijke studie werd geconcludeerd dat geen van de versnelde testen een
140
Samenvatting
betrouwbare voorspelling van de duurzaamheid van coilcoatings op de middelen lange-termijn kon geven.
Aangezien corrosie van (gecoate) metalen een elektrochemisch proces is, is het logisch
om de prestatie van beschermende coatings te karakteriseren met behulp van
elektrochemische meettechnieken. In dit proefschrift worden EIS en ENM met dit doel
toegepast. Hoofdstuk 3 geeft de theoretische achtergrond van deze technieken.
Hoofdstuk 4 bespreekt de resultaten van de buitenexpositie van model
coatingsystemen, periodiek gemeten met EIS. Panelen met de modelsystemen werden
gedurende 2 jaar bevestigd aan een expositierek in een mariene omgeving (Den
Helder) . Voor elke meting, werden de panelen overgebracht naar het laboratorium en
geprepareerd voor de meting. Impedantiedata werden geanalyseerd door
equivalentcircuits te construeren in overeenstemming met de MPEC (most probable
equivalent circuit) richtlijnen.
Vooral in vroegtijdige stadia van coatingdegradatie, wanneer de coating nog geen
visuele tekenen van degradatie vertoont, biedt EIS duidelijk voordelen. Verscheidene
processen gerelateerd aan coatingdegradatie (e.g. wateropname, vorming van
geleidende paden) kunnen dan al met deze techniek worden gedentificeerd, onder
niet-versnelde omstandigheden.
Ondanks het feit dat de tijdsafhankelijkheid van de impedantieparameters wat
spreiding laat zien, kan toch waardevolle mechanistische informatie worden afgeleid
uit de meetresultaten.
De meetresultaten zijn gerelateerd aan het genificeerde model voor coatingdegradatie
van Nguyen en Hubbard.
Hoewel er een grote hoeveelheid impedantiedata van beschermende coatings
beschikbaar is, variren meet- en analyseprocedures tussen onderzoeksgroepen. Dit
verhindert een directe vergelijking van meetresultaten tussen deze groepen. Daarom
wordt in hoofdstuk 5 een standaard meetprotocol gentroduceerd. Meer dan 100
industrile coatings zijn volgens dit protocol gemeten. Voor elk coatingsysteem zijn
EIS metingen uitgevoerd na 0, 1, 7 en 21 dagen immersie.
Uit de algehele resultaten en de variatie tussen replicas werd duidelijk dat de variatie
van coatingparameters voor elk coatingtype niet moet worden gerelateerd aan de
meetnauwkeurigheid van EIS, maar aan de feitelijke verschillen in beschermende
eigenschappen van de geteste coatings.
De resultaten bevestigen dat het standaard EIS meetprotocol een objectieve methode is
om snel de corrosiebeschermende mechanismen van coatings vast te stellen. Ook
141
Samenvatting
wordt inzicht gekregen in andere eigenschappen zoals de invloed van de

mengverhouding (epoxy/harder) en de chemische resistentie. Echter, EIS is niet een op
zichzelf staande techniek. Alleen met behulp van andere technieken (het meten van de
open circuit potentiaal, visuele observaties, etc.) en aanvullende literatuur kunnen
zinvolle conclusies worden getrokken over de beschermingsmechanismen van
coatings.
Momenteel is EIS de meest gebruikte elektrochemische techniek voor evaluatie van
beschermende eigenschappen van coatings. Recentelijk is de toepassing van ENM met
hetzelfde doel van belang geworden. In hoofdstuk 6 wordt de bruikbaarheid van ENM
voor de karakterisatie van beschermende coatings onderzocht.
Ruissignalen als functie van de tijd laten vaak een bepaalde trend zien. Aangezien deze
trend de berekende ruissignalen benvloed, moet deze vr de data-analyse worden
verwijderd. Twee methoden om de trend te verwijderen zijn onderzocht. Desondanks
blijven fundamentele vragen over wanneer variatie van potentiaal of stroom moet
worden gezien als onderdeel van de trend en wanneer als elektrochemische ruis
onbeantwoord.
In het grootste deel van de publicaties over elektrochemische ruis wordt aangenomen
dat beide werkelektrodes identiek zijn. Voor metingen aan gecoate substraten is dit
zelden het geval. Uit een experiment waar de invloed van asymmetrische elektrodes is
onderzocht, kan worden geconcludeerd dat in dit specifieke geval de ruisweerstand
voornamelijk wordt bepaald door de elektrode met de hoogste impedantie. Echter,
deze uitkomst moet niet worden gezien als een regel voor alle situaties.
Vijf industrile coatings zijn onderzocht met zowel EIS als ENM. Het werd duidelijk
dat de laagfrequente impedantie altijd groter was dan de ruisparameters. Dit verschil
bleek toe te nemen met de impedantie van de elektrodes. Ondanks dit verschil werd
een correlatie gevonden tussen de impedantie en de ruisweerstand. Op basis van deze
correlatie is het mogelijk om ENM in te zetten om veranderingen van beschermende
coatings te monitoren.
In het laatste hoofdstuk van dit proefschrift worden de conclusies uit de voorafgaande
hoofdstukken samengevat.
142

W.M. Bos, G.M. Ferrari and J.H.W. de Wit, Accelerated corrosion tests and
qualification methods for organic coatings A new approach, Proceedings
EUROCORR 2004, Nice, France, September 2004.
W.M. Bos, G.M. Ferrari, J.H.W. de Wit, Accelerated corrosion tests and qualification
methods for organic coatings - Determination of degradation mechanisms,
Proceedings EUROCORR 2005, Lisbon, Portugal, 2005
W.M. Bos, G.M. Ferrari, J.H.W. de Wit, Determination of degradation mechanisms of
Organic Coatings, Proceedings 16th International Corrosion Congress, Beijing, China,
2005.
W.M. Bos , A. Homborg, G.M. Ferrari, J.H.W. de Wit, Complementary use of EIS and
ENM for Qualification of Organic Coatings, Proceedings EUROCORR 2006,
Maastricht, The Netherlands, September 2006.
J.H.W. de Wit, W.M. Bos, J.M.C. Mol, G.M. Ferrari, Chapter 22: Coatings for marine
and shipping applications, in: High performance organic coatings: Selection,
application and evaluation, A.S. Khanna (Ed.) Autumn 2008, ISBN 1 84569 265 9
Chapter 4 and 5 of this thesis have been submitted for publication in scientific
journals.
143
144
Dankwoord
Twee mensen hebben mij overtuigd promotieonderzoek te gaan doen.
Allereerst wil ik mijn promotor, Hans de Wit, hiervoor bedanken alsmede voor het
vertrouwen en de vrijheid die ik heb genoten tijdens het schrijven van dit proefschrift.
Ondanks je overvolle agenda wist je altijd tijd voor me vrij te maken.
Het begon echter allemaal bij TNO Industrie en Techniek in Den Helder toen Gabriele
Ferrari aan mij dacht toen er een promotieplaats beschikbaar kwam. Ringrazio
Gabriele per avere creduto nelle mie capacit, aver creato la possibilit di fare la
ricerca e la tesi e sopratutto di avermi convinto a farla.
De tweede Hans die ik wil bedanken is Hans Sonke. Zonder je praktische
coatingkennis en je netwerk in de coatingwereld was met name hoofdstuk 5 niet wat
het nu is. Op het uitvoerende vlak wil ik Sibo Buter en Patrick Ruig van TNO
bedanken. Jullie hebben bergen werk verzet! Ik ben blij te horen dat jullie het werk
voortzetten.
Axel Homborg van de Koninklijke Marine heeft tijdens zijn afstudeeropdracht een
belangrijke bijdrage geleverd op het gebied van ruismetingen.
Wouter Hamer, bedankt voor je scherpe commentaar op hoofdstuk 3. We zijn nu weer
collegas!
Glen Donnelly, thank you very much for reviewing part of my thesis. As a native
English speaker you have corrected many typical Dutchman-tries-to-write-English
mistakes.
Ineke Glaser-Muller, bedankt voor het verwerken van mijn vele literatuuraanvragen.
Ook een artikel uit 1899 wist je moeiteloos(?) op te duiken.
Verder wil ik de begeleidingcommissie van het IOP-project bedanken voor hun
betrokkenheid. De commissie bestond uit: Hans de Wit, Jan van Turnhout, Gabriele
Ferrari, Leo van der Ven, Cees van Vliet, Jan Breen en Patrick van Veenendaal.
Massimiliano Giacomelli Penon was mijn mede- promovendus in het IOP-project.
Max, nu is het mijn beurt om je te bedanken voor de samenwerking!
Mijn ouders, Sandra, Rob, Melle, Janna, Pomme en Zoey wil ik bedanken voor hun
steun gedurende het promotieonderzoek. Ik heb nu weer wat meer tijd voor jullie.
145
Dankwoord
Tot slot wil ik Arina, Richard en Sunny bedanken voor hun gastvrijheid tijdens de
periodieke uitstapjes naar Curaao. Ayo!
Tom Bos
Amsterdam, februari 2008
146
Curriculum Vitae
Wilhelmus Maria Bos
Date of birth: June 7, 1974
Place of birth: Den Helder
Education
2006 TNO Innovators Course (BonSpirit, Coach & Commitments)

1994-1998 Bachelor of Science in Environmental Technology (B. Sc.)
Institute for Petroleum and Gas Technology "Noorderhaaks", Hogeschool
van Amsterdam.
Summary of work experience
2007 - present
Shell Global Solutions International B.V.
Upstream Materials Engineer
2003 - 2007
PhD research
Delft University of Technology
Department of Materials Science and Engineering
1998 - 2007
Project leader
TNO Science and Industry
147
148

Bos 20080311

Uploaded by

Document Informationclick to expand document informationsteel

Document Informationclick to expand document information

Copyright:

Available Formats

Bos 20080311

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Bos 20080311

Uploaded by

Copyright:

Available Formats

Prediction of coating durability

Early detection using electrochemical methods

Cover photo: a pristine silicone coating

Prediction of coating durability

Wilhelmus Maria BOS

Dit proefschrift is goedgekeurd door de promotor:

2 Performance testing of protective coatings

3 Electrochemical techniques for coating characterisation

3.2.6 Typical impedance spectra of protective coatings

5 Pre-qualification of protective coatings using EIS

6 Application of ENM to the study of protective coatings

6.2.1 Measurement equipment

Publications related to this work

Loop of the counteracting refining and corrosion processes.

1.2 Economic impact of corrosion

Corrosion of metals has an enormous economic impact. Cost-of-corrosion studies have

Overview of cost-of-corrosion studies by country [3].

Total annual cost of corrosion

Not reported in U.S. dollars.

1.3 Mitigation of corrosion damage

Major corrosion-control methods include barrier protection, cathodic protection and

1.4 Protective coatings

In this thesis, various properties of protective coatings will be discussed. Therefore,

in the coating. This is particularly important since polyester coatings shrink

1.5 Research objective and topological breakdown

The objectives of this Ph.D. research are two-fold.

Mitigation of corrosion damage

Topological breakdown of this thesis.

1.6 Thesis outline

In this thesis, various properties of protective coatings will be discussed in detail.

Lenderink, H.J.W. Filiform Corrosion of coated Aluminium Alloy - a study of

Hamer, W.J., Polypyrrole Electrochemistry - Environmentally friendly corrosion

Corrosion - A Natural but Controllable Process, Supplement to Materials Performance,

Moavenzadeh, F. (Ed.), Concise Encyclopedia of Building and Construction Materials, ,

Dobbelaar, J.A.L., The use of impedance measurements in coating research The

Westing, E.P.M. van, Determination of coating performance with impedance

10 Geenen, F.M., Characterisation of Organic Coatings with Impedance Measurements,

2 Performance testing of protective coatings

2.2 Outdoor Exposure

Typical setup for outdoor exposure of coated panels.

2.3 Accelerated tests

2.3.1 Salt spray test

Performance testing of protective coatings

2.3.2 Cyclic testing

Performance testing of protective coatings

Cyclic variation of temperature allows for some thermally induced expansion

2.3.3 Advanced cyclic testing

Performance testing of protective coatings

ISO 20340 Cyclic testing options [57].

2.3.4 Overview accelerated tests

Overview of some (standardized) accelerated tests.

Performance testing of protective coatings

Overview of some (standardized) accelerated tests.

VICT (Volvo Indoor

Alternating UV/condensation cycles and wet/dry salt-spray cycles.

2.4 Evaluation methods

2.4.1 Visual evaluation