Corrosion is an irreversible phenomenon which results from the basic
thermodynamic characteristics of the materials and the nature of their
environment. In this lecture we will look at the water corrosion phenomena
applicable to metals and alloys, but excluding both high temperature oxidation
phenomena and corrosion of non-metallic materials.
Corrosion phenomena may concern the most of the devices of our day to day life,
when they are immersed insufficiently severe environments, or made from an
insufficiently resistant material (which may happen for cost consideration, even when
the appropriate solution is known, but expansive). Corrosion risks may also arise from
inappropriate secondary processing conditions, such as welding or drawing. The
conjunction
The corrosion of metals takes two different forms:
1) Dissolution of the metal and therefore loss of matter. This consists of passage of
the metal cations into an aqueous solution (anodic dissolution), the electrons
p^roduced by this dissolution being consumed by a cathodic reaction. As such, the
dissolution of metals
differs greatly from ordinary dissolution such as when sugar is added to water.
Aqueous corrosion is above all an electrochemical phenomenon.
2) The formation of rust, which is the common name given to certain iron oxides. In
this case, the product of corrosion is not soluble but solid and often considered
unsightly.
The corrosion of metals may take different forms:
1) Oxidation in oxygen containing atmospheres at high temperatures (higher than the water
dew point in this atmosphere). This produces a gain of matter (the oxygen trapped in the
oxide) and will be investigated later. Morever, It is well known that most metal ores are
found in their oxide form. However, the overall reaction involving formation of the oxide
splits into several simple reactions including the dissociation of its cations and electrons.
2) An oxide can also form by interaction with water (passivation reaction). In fact, this
consists of both an oxidation reaction and an acid-base reaction which may include several
steps.
3) In certain conditions, metal cations are transferred to solution in hydrated form (anodic
dissolution), resulting in a loss of matter. In aqueous corrosion, the electrons produced by
the anodic dissolution reaction having to be consumed by a cathodic reaction. As such, the
dissolution of metals differs greatly from ordinary dissolution such as when sugar is added
to waterfor instance. Aqueous corrosion is above all an electrochemical phenomenon.
4) Lastly, the cations anodically dissolved may cause hydrolysis, leading to the formation of
a hydroxide or an oxide which can precipitate in the form of rust, which is the common
name given to certain iron oxides and often considered unsightly.
This 3 forms have in common to initiate from a chemical oxidation process , namely de-
electronation , involving an oxidising element (for example oxygen, or ferric iron) which
is able to trap the electrons from a reducing species (for instance oxhydrile ion, or ferrous
ion). The tendency of a metal to be oxidized is related to its capacity to combine with the
oxidizing agent.
The predominant role of water in corrosion phenomena must be underlined. Theoxygen in the iron oxides is drawn essentially from the molecules of water and not atall or only very slightly from the oxygen in the air, as is most often thought (the oxygenonly encourages the cathodic reaction as will be explained later). In our temperateclimates, atmospheric corrosion may be considered a water corrosion phenomenondue to the presence of a film of water on the surface of the metal which generallyconstitutes a cold surface encouraging the formation of condensation. Moreover, theexample of very old iron parts found in hot and dry climates which have not rustedbecause the ambient moisture content is too low to cause condensation of water onthe surface of a metal illustrates this point.
1) The dipolar water molecule adsorbs to metal surfaces due to either physical
or chemical interactions.
2) Adsorbed water can combine with the metal by loosing a proton and an
electron, giving birth to a MOH type adsorbed complex. This second step
(formation of the absorbed complex) is common between water corrosion and
passivation mechanisms described below.
3) In turn, the complex MOH may (or not) oxidise by loosing an electron. The
oxidised form MOH+ is unstable and dissolves by giving up the cation to the
water environment.
This a typical water corrosion process (Bockris mechanism for divalent iron
dissolution).
4) If, instead of oxidizing and dissolving, the MOH complex is again
deprotonated, we observe the formation of a stable species closely linked to
the metal and inhibitor of any subsequent aqueous corrosion (the water is no
longer absorbed on the metal but on the MO oxide).
This the typical water passivation process
In ferrous alloys, the passive film growth proceeds from cation hopping and electromigration
of the resulting vacancies toward the metal
In chloride containing environments, the adsorbed water can be replaced by a
chloride ion. A similar mechanism to the one described for water corrosion is then
possible but the MCl complex can only dissolve (by restituting the chloride ions in
solution after dissolution of the cation) and cannot form a protective passivating
compound. Finally, the chloride ion functions like a cation pump and the mechanism
can be repeated. This mechanism is known as chloride dissolution.
Water is not only an acid/base amphoter The hydrogen and oxygen evolution
reactions are in fact two aspects of the oxydoreduction of water
The deprotonated form of water (hydroxyl ion) behaves like a reducing agent, able to
give up electrons with release of oxygen. Its protonated form (hydronium ion)
behaves like an oxidizing agent, capable of trapping electrons with release Hydrogen
These reactions are controlled by the solution pH and, in the presence of an electron
reservoir (for example metal), by the difference of potential between this reservoir and
the solution.
The charge transfers between metal and solution lead to a difference of
potential between the metal and the core of the solution.
The figure has been drawn in the case of a net positive excess of
charge on the metal surface and of a net negative excess of charge in
the electrolyte.
One has assumed that the solution excess of charge was concentrated
at the immediate vicinity of the metal (outer Helmoltz plan), which is true
for sufficiently conductive solutions. In this case , the interface behaves
as a capacitor , the two plates of this capacitor beeing the metal surface
(Inner Helmoltz plan) and the Outer Helmotz plan
Both IHP and IHP form the so called Electrochemical double layer
Remark: In the more general case, the charge transfered to the
lectrolyte is not concentrated at the interface and form the so called
« diffuse layer »
The anodic dissolution results in a transfer of cations (positive charges) from the
metal to the solution
From another hand, the metal forms a reservoir of electrons which it can exchange
with other oxidizing species present in solution. This transfer is called "cathodic
reaction".
In an acid environment, the usual cathodic reaction is the one of so-called hydrogen
evolution which reduces proton to hydrogen.
In a neutral oxygenated environment, the oxygen is consumed (evolution of oxygen)
thereby increasing the local pH.
In ferric salt environments, the ferric ions can be transformed into ferrous salts.
And so on …
In each case, the cathodic reaction operates below the redox standard potential
(above it is the opposite reaction which can take place, for instance release of oxygen
by decomposition of the water).
The typical reactions of metal-water interactions involve both electrons
(oxydoreduction) , protons (acidobasicity), and solute cations (solution chemistry)
1) Anodic disssolution: M M++ + e-
2) Precipitation of rust M+++H20 MO + 2H+
3) Passivation: M+H20 MO + 2H+ + 2e-
cst.pHq
kT.V
cst.(c) pH
cq
kTcstV
32 )3(
log2
1 )2(
)log(2
3.2. (1)
kT
qVc
kT
qVH
exp.
exp
With c being the concentration in
solute cation and V the metal
solution potential difference
The Mass action law writes:
Which can be represented in a potential pH diagram (the Pourbaix
diagram) with c being a parameter. Let us note that MO is protective
when formed from passivation reaction (3) but much less or not at all
when formed from rust precipitation reaction (2).
The electochemical activities of solute species are:
The actual Pourbaix Diagrams are more complex, due to
the co-existence of several electrochemical reactions
The Al diagramme shows 2 soluble species
(3) log (Al+++) = 5,7 - 3 pH
(5) V = -1,663 + 0.02 log (Al+++)
(6) V = -1,2662 - 0.08 pH + 0,02 log (AlO2-)
For Iron, several forms of solute iron (Fe++, Fe+++ , HFeO2--, …) have to be considred ,
defining several corrosion regions in the diagram…. (see figure).
The intensity of the cathodic and anodic reactions is related to the difference of
potential between the metal and the electrolyte. This is set using a potentiostat and
the current density is measured; we obtain then a polarisation curve (see above).
The overall measured current iG is in fact the sum of the anodic current iA (positive)
and the cathodic current iK (negative). This overall current is nil at the free corrosion
potential Vcor (no potential applied). The cathodic and anodic currents are then equal
in absolute values to the corrosion current icor
The corrosion potential and current correspond to the intersection of the individual
polarisation curves for the anodic and cathodic reactions. The slope of the curve i(V)
at the corrosion potential has a dimension which is the inverse of a resistance (by unit
surface) called polarisation resistance. There are different techniques for measuring
the polarisation resistance and deducing the corrosion current.
Lorsque les deux réactions cathodique et anodique sont controlées par une loi de
Tafel, le courant global résultant est une différence d ’exponentielles. On peut
reformuler la loi composée ainsi obtenue en rapportant les potentiels au potentiel de
corrosion (courant global nul), c ’est à dire en utilisant la surtension définie plus haut.
Dès que l ’on s ’éloigne du potentiel de Corrosion, l ’une des deux réactions anodique
ou cathodique devient dominante et l ’une des exponentielles peut etre négligée
devant l ’autre. La surtension dépend alors logarithmiquement du courant global (Loi
de Tafel). La Constuction d’ Evans permet alors de déterminer graphiquement le
potentiel et le courant de corrosion
Near the corrosion potential, the current varies linearly with the electrode potential. A
polarisation resistance is defined. The Stern-Geary law gives then the corrosion current as a
fonction of this resistance and Tafel coefficcients
10
2 materials electrochemically different exhibit different polarization curves when
placed in tyhe same electrolyte.
When these 2 materials are in contact each of the other, it results in an
electrochemical coupling and the so called « galvanic corrosion ».
Folowing the galvanic corrosion rules, the less noble material specialises in anodic
reaction, then corrodes, while the nobler specialises in cathodic reaction, then is
protected
This is only a schematic view of galvanic corrosion problems and things may be more
complex in real situations :We can give the example of corrosion due to contact
between two different metals. In certain cases the most noble metal is protected but
accelerates the corrosion of the less noble metal (galvanic corrosion). In other cases
nothing happens, although the less noble metal remains passive in the considered
environment, for instance aluminium joinery fixed by stainless-steel screws does not
cause corrosion after disappearance of the protective polymer washer.
We must also consider that bi-metallic corrosion is not limited to only the galvanic
couple: an active metal in a given environment may lead to corrosion of a metal which
should normally be passive in this environment, simply because the products of
corrosion of the former acidify the solution!!!
Lorsque deux métaux différents comme l ’Aluminium et l ’Acier sont mis en contact électrique
dans un même milieu conducteur (ionique), ils forment une pile qui débite du courant en
consommant le métal le plus anodique (voir schéma). Comme l ’Aluminium est anodique par
rapport à la plus part des métaux usuels, il est souvent la victime de ces assemblages.
La corrosion galvanique nécessite trois conditions :
-deux métaux différents
-un contact électrique(électronique)
-un électrolyte conducteur (contact électrolytique)
La suppression de l ’une de ces trois conditions supprime le phénomène.
Bien que certains éléments comme le cuivre déplace le potentiel de l ’Aluminium vers les
valeurs plus nobles, le choix de l ’alliage d ’Aluminium ne permet pas d ’éviter ce problème. Si
l ’on s ’en tient a la série galvanique ci-dessus, on pourrait croire qu’il suffit de remplacer un
assemblage aluminium (1050)/acier inoxydable par un assemblage aluminium (2024)/acier
ordinaire, pour régler le problème. Malheureusement l ’expérience montre que si on diminue
effectivement le courant de corrosion galvanique avec ce nouvel assemblage, on n ’augmente
pas pour autant la pérennité de l ’assemblage. La solution utilisée remplace deux éléments
résistants à la corrosion par deux éléments corrodables, ce qui est gagné sur la corrosion
galvanique est perdu par l ’auto-corrosion des éléments du couple. L ’expérience montre que
dans l ’air, l ’assemblage de tôles Aluminium par des vis en acier inoxydable est préférable (cf..
les assemblages de mâts en Aluminium avec vis inox. sur les voiliers). Par contre s ’il y a un
électrolyte; cas des parties immergées par exemple, il faut trouver d ’autres solutions.
Remarks:
1) In most cases the most noble metal is protected but accelerates the
corrosion of the less noble metal (galvanic corrosion). In other cases
nothing happens, since the less noble metal remains passive in the
considered environment, for instance aluminium joinery fixed by
stainless-steel screws does not cause corrosion after disappearance of
the protective polymer washer.
2) We must also consider that bi-metallic corrosion is not limited to only
the galvanic couple: an active metal in a given environment may lead to
corrosion of a metal which should normally be passive in this
environment, simply because the products of corrosion of the former
acidify the solution!!!
In very aggressive media (hydrochloric acid in the example), the anodic
current (iron dissolution) increases continuously when the electrode potential
increases
In less aggressive media (sulphuric acid for instance) the polarisation curve of
iron exhibit several different domains.
First (active behaviour) the anodic current increases up to a maximum (The
critical passivation current).
Second, it decreases down to a very low intensity plateau (passive region)
before increasing again for high potential (transpassive region). The onset of
passivity in a potential range is due to the presence at the metal surface of a
very thin but protective oxide layer (the passive film).
This film is unstable at too high potentials (transpassive domain).
The figure above shows the typical behaviour of Fe-Cr alloys at different Cr contents
in sulphuric acid (1M).
The fall in passive current with the Chromium content increase is particularly
noticeable up to 12% Cr which is generally considered as the practical limit for a steel
to be stainless.
The composition and the thickness (around a few nanometres) of a passive film can be
measured by different surface analysis techniques. The film is rich in the most oxidizable
elements of the alloy (in this case chromium). It is also hydrated and often heterogeneous
(the parts furthest towards the outside being more hydrated than the inner part, closer to an
oxide). In addition, the thickness and composition depends largely on the metallurgical
history of the surface. They also evolve in time, generally ensuring an increasingly protective
nature with regard to the different forms of corrosion.
The factors involved in the potential destabilisation of a passive film are the acidity (which
consumes oxygen and hydroxil ions) and the chloride ions (or all other similar ions) which
combine with the cations without forming a protective species. The chloridized acid
environments are particularly dangerous. Neutral non-chloridized or quasi-neutral
environments (by this we mean where the pH is greater than a critical pH called the
depassivation pH) present no risk, the non-chloridized acid environments can lead to uniform
corrosion of the surface by disappearance of the passive film if the critical current is higher
than the capacity of the oxidising agent to consume the electrons produced. As we shall see
later, the neutral chloridized environments (pH above the depassivation pH determined in the
presence of chlorides) can lead to a local form of corrosion which only affects a small part of
the surface, the remainder remaining passive. The relevant notion is the critical chloride
content, but it is difficult to implement since many other parameters (such as the potential
reached) may intervene and other quality criteria will generally be preferred.
On this typical polarisation curve, we can clearly distinguish 3 potential ranges:
1) The peak of anodic dissolution (also called activity range) with an anodic currentwhich may reach several mA/cm²)
2) The passive range where the current is les than 1 µA/cm² and often of the order of1 nA/cm² )
3) The transpassive range where the dissolution current increases again.
The passive domain corresponds to a quasi absence of dissolution, i.e. in practicenon-oxidisability of the material (1 µA/cm² is approximately 10 µm/year). Thisslowing of the anodic dissolution is due to the presence of a thin film of oxide (a fewnanometres) called passivating film or passive layer or PASSIVE FILM whichconsiderably slows down the kinetics of the ionic transfer (by a factor often greaterthan 1000).
The electrochemical behaviour may differ in the presence of an oxidising agent other
than the proton. In the example shown in the figure, the oxidising power is high
enough for the intersection of the anodic and cathodic curves to be found in the
passive range of potential (Point P). This situation occurs when the critical
passivation current is below the current which can be supplied by the cathodic
reaction. If not (for instance when oxidation is due to the single protons as described
earlier), the system operating point is situated in the active domain and there is no
protection by the passive film.
The effect of the pH can be measured by plotting the polarisation curves and by measuring
the critical passivation currents in relation to pH. Above a certain pH, called the
depassivation pH, the activity peak is no longer observed and the metal is therefore passive
whatever the potential (provided it stays below the pitting potential as we shall see later, and
of course provided we avoid the high potentials corresponding to the transpassive domain, or
the decomposition of water, or the change in the degree of oxidation of certain cations).
In a chloride containig neutral environment, increasing the potential can lead to a sudden
increase in the current long before the transpassivation potential range is reached.
Micrographic examination after testing shows that local failure of passivity has arisen and that
pitting has developed.
The above polarisation curves have been obtained with a 430Ti steel in a NaCl aqueous
solution with 6.6 pH.
The pitting potential measured as described above diminishes linearly with the logarithm of
the chromium content. It is also subject to wide scatter and this leads to the use of statistical
methods. This scatter seems inherent in the pitting phenomenon, which appears to be
probabilistic, at least insofar as initiation phase is concerned.
Si le milieu contient un oxydant permettant une réaction cathodique, le potentiel d ’abandon devrait
s ’établir à lintersection de la courbe anodique déterminée à l ’aide du potentiostat et de la courbe
cathodique (supposée connue) correspondant à la réduction de cette oxydant. Suivant le pouvoir oxydant
de cette réaction (courbes cathodiques K1 ou K2), le potentiel d ’abandon s ’établira en dessus ou en
dessous du potentiel de piqûration et il y aura ou non corrosion localisée. Le potentiel de piqûres est
donc bien un critère de qualité pour la résistance à la corrosion par piqûres et les alliages ayant un
potentiel de piqûre élevé sont ceux qui résistent le mieux à la piqûration.
Chloride acid environments cumulate trends to passivation failure. Diminishing the pH
increases the critical current and increasing the content of chlorides brings down the pitting
potential. This assembly considerably reduces the passivity range
Such environments must really be considered constituting the most hazardous.
Although pitting corrosion constitutes the typical type of local corrosion, and certainly the
moste investigated in the past, it is often not the major concern of the engineer when
maintaining the integrity of a structure.
This major concern is often associated with other factors which have an accelerating role in
terms of corrosion priming. These factors are multiple and we give a few examples below:
- geometrical factors: crevice corrosion, corrosion under deposits
- mechanical factors: stress corrosion, fatigue corrosion, friction corrosion, etc.,
- metallurgical factors: intergranular corrosion (today considered controlled in most
instances), etc…
- complex situations where several phenomena interact and which often require the
intervention of an expert if only to determine the various causes of corrosion.
Observation of pits at 3 different scales.
The last observation (right hand down corner) is performed using Scanning electron
microscopy (X1000). It shows a typical view of a semi developped pit at the end of the
mesoscopic stage:. A thin metallic film is still present and covers a part of the pit. Secondary
pits are visible all around the main hole. The pit may either go on or repassivate if the thin
"top" film breaks down
Close correspondence is observed between the potentiokinetically measured pitting
potential and the number of pits after exposure to a chloride solution in identical and
clearly defined conditions.
Such a result confirms the role of pitting potential as a corrosion quality criterion in a
chloride neutral environment.
The straight (red) line represents the rest potential of a sample immersed in aerated water.
Alloys with poitting potential larger than this rest potential are acceptable. Others are not
Crevice corrosion takes place in a confined area (joint, anfractuosity, under an inert deposit,etc.) called crevices. The main characteristic of a crevice is the absence of easy convectiveexchanges with the outside.
The corrosive solution penetrates in the crevice, possibly by capillarity. Due to the absence ofconvectiion and of sufficient diffusionin the crevice, the chemical composition can be verydifferent in the crevice than outside
In addition, it will be noted that the oxygen content (supposed here to be the majorityoxidising agent in the environment) falls very quickly in the crevice leading to a differentialaeration solution between the inside (anodic zone) and the outside (cathodic zone). Contraryto what occurs with ordinary steels, this coupling is insufficient to initiate crevice corrosionwhich occurs later by breakdown of the passive film (acid corrosion, when the pH falls belowthe depassivation pH, or pitting corrosion when the chloride content has become highenough).
The left side figure shows an experimental device generating crevice corrosion. The resultingcorrosion is shown on the right side figure
Grade : AISI 316 (1.4401)
Using conditions : thermal exchange between molasses and chlorined juice
Causes : there is an enclosed zone at each contact point between the two plates, and under
the joint crevice corrosions appear.
Solutions : avoid crevice configurations …. or use a grade more resistant to crevice
corrosion (for example 1.4462 (Uranus 45N)).
The cations produced by slow dissolution through the passive film quickly enrich the crevicewith positive electrical charges.
An electrical field is then created from the crevice inside to the surrounding zone. The resultof this is the electromigration of the cations from the crevice to the outside solution and of themajority negative species contained in the solution, i.e. the chloride ions, from the solution tothe inside of the crevice. Finally, in the stationnary state, the crevice contains an amount ofpositive cations and of negative anions (chlorides)
In addition, the cations are hydrolysed and the solution pH inside the crevice, regularly falls.
The conditions are thus united for depassivation, either by acid corrosion in a hydrochloricmedium when the pH falls below the depassivation pH
Due to the increase in the chloride content pitting corrosion may also occur, when the piitingpotential falls below the abandon potential. This pitting corrosion causes a catastrophicincrease in the dissolution of cations in the crevice and there quickly occurs aciddepassivation of all the film below the crevice.
Based on the geometry of the crevice, the kinetics involved in chloride enrichment and
acidification can be calculated ( see figure ). An excellent correlation has been found in
experimental conditions between the duration of actual incubation and the one deduced from
the acidification kinetics. Invariably, the crevice pH falls as the cations content increases.
This leads to corrosion starting after a certain period of time (in practice around a few days or
months), a period which increases as the depassivation pH falls.
The depassivation pH in a chloridized environment is therefore a quality criteria for the
crevice corrosion resistance of stainless steels. To be complete, we should add a second
criteria to take account of the possibilities of pitting beneath the crevice, i.e. the potential for
pitting in acid chloridized environments, a notion which until now has been applied less.
Since in an ideal crevice pH should irreversibly decrease and Cl- concentration
irreversibly increase, the best way of preventing crevice corrosion consists in avoiding
crevice corrosion conditions!! or if not at least dismantling the installations for cleaning at
periodicities which are less than the depassivation time.
However, in actual practice the situation is not as catastrophic as it might appear.
Indeed, the severity of crevice corrosion (which governs the acidification kinetics) is
associated with the possibility or otherwise of exchanges with the exterior. Fortunately,
these exchanges always exist to a small extent (otherwise the water would not wet even
the inside of the crevice and no corrosion would take place). Increasing the acidity and
the chloride content are therefore limited by the diffusion factor. The result is that a
material with a fairly low depassivation pH (and fairly high potential for pitting in a
chloridized acid environment) will remain insensitive to crevice corrosion (for a given
crevice geometry).
The most usual occurrence of intergranular corrosion is due to the precipitation
of Cr carbides at grain boundaries (location of preferential precipitation in the
metals) as a result of heat treatment such as occurs for instance on cooling a
metal product or in the neighbourhood of a weld. If this formation of Cr carbides
is not followed by homogenization treatment, there will be a depletion of Cr for
the formation of the carbides in the surrounding area (for instance in the event of
cooling too fast to allow diffusion from the surrounding metal). This chromium
depletion may lead to a very low content locally in the steel which is below 12%.
Its resistance to all forms of corrosion then falls.
Intergranular corrosion occurs as a result of a change in the composition of the steel at the
grain boundaries, a phenomenon which makes the steel more sensitive to several types of
corrosion (pitting, stress, etc.).
The most usual occurrence of intergranular corrosion is due to the precipitation of Cr
carbides at grain boundaries (location of preferential precipitation in the metals) as a
result of heat treatment such as occurs for instance on cooling a metal product or in the
neighbourhood of a weld. If this formation of Cr carbides is not followed by
homogenization treatment, there will be a depletion of Cr for the formation of the carbides
in the surrounding area (for instance in the event of cooling too fast to allow diffusion
from the surrounding metal).
This chromium depletion may lead to a very low content locally in the steel which is
below 12%. Its resistance to all forms of corrosion then falls.
There are several metallurgical solutions for overcoming the IGC (see table above). They
cover either the composition of the steel or the heat treatments used after sensitisation.
Stress on corrosion is a complex phenomenon. Its concrete manifestation is the appearance of
damage which neither the mechanics nor the science of corrosion taken on their own could
have generated. In simple terms, it may be said that stress corrosion can appear in hot
chloride containing environments, and that in situations where there is an element of risk, the
use of ferritic stainless steels (body centred cubic), or austeno-ferritic steels, is much more
preferable than the use of austenitic steels (face centred cubic).
Several mechanisms are envisaged for taking account of the different observations. By
simplifying to the extreme, it may be considered that stress causes mechanical failure of the
passive film which then creates a zone of active dissolution which evolves like a pit.
Stabilising this type of corrosion will then depend on this supposed competition between local
failure of the film (emergence of the slip plane) and its re-passivation or otherwise in a
chloridized environment.
Hence, we can envisage an interaction between the failure generated by residual dissolution
through the passive film and the dislocations present in the metal. Lastly, once local corrosion
has begun, the behaviour at the bottom of the crack is doubtless driven by the evolution of the
hydrogen produced by the cathodic reaction, or its interactions with the dislocations present in
the metal.
In this paper we have summarily described a limited number of ways in which local corrosionbegins. In actual fact, there are many reasons for corrosion starting and they sometimescombine. For instance atmospheric corrosion, which from a distance resembles pittingcorrosion, is in fact a very complex phenomenon. Similarly, corrosion beneath deposits (areason for proper cleaning) is different depending on whether the deposit is chemically inert(in which case similar to crevice corrosion) or active.
In all cases, careful analysis of the service conditions (composition of the corrosiveenvironment, temperature, etc.) and the sequences involved in operation of the consideredsystem are essential, including the phases which are often neglected such as cleaning, non-useof the installation, etc.) or long-term trends (climatic cycle in atmospheric corrosion, road orother urban use for a vehicle). Such analysis is essential when carrying out an assessment butalso for any recommendations (an exhaustive examination of the predicted conditions of useis then necessary) and often permits preventive measures to be adopted.
The critical analysis should also cover the tests used in the laboratory to qualify the solutionsadopted and their representativity. This work is not always performed correctly. Indeed it isalways necessary to question the relevance of the tests, even when accelerated. The absenceof such a procedure will result in illusory and economically unreasonable solutions.
Lastly, the constant and necessary reduction of industrial costs requires the choice of stainlessproducts exactly matching a given use with minimum "over qualities". In order to avoid aconcomitant reduction in safety, a better understanding of the corrosion mechanisms willalways be needed. This justifies both the need for constant research into the science ofcorrosion and the existence of a body of corrosion engineers responsible for providing the linkbetween this research and industrial practices.