Você está na página 1de 6

September 1972

ANTI-CORROSION

THE THEORY OF STRESS


CORROSION CRACKING IN ALLOYS
Dr J. C. Scully, Department of Metallurgy;
The Houldsworth School of Applied Science
From the book of the same title issued by the Nato Scientific Affairs Division. For a full review
of the book, please turn to page 26.
Introduction
Stress corrosion cracking is a phenomenon that is of interest
to a wide range of metal users. When it occurs under service
conditions, often without any prior indication of impeding
failure, its effect may be catastrophic.
Large sums of money may be lost as a result of a stress
corrosion failure, arising perhaps from the enforced shutdown
of an operating plant, e.g. a nuclear reactor generating station,
or from the loss of a vehicle at sea or in the air. The use of
expensive non-susceptible alloys for situations where stress
corrosion may occur and where the possibility of failure cannot be tolerated is another source of economic loss. An easing
of the problem has not come about, and it has so far proved
nearly impossible to build structures according to a design
specification which will ensure that failure does not occur
as has been done with other forms of failure. Instead, failures
occur and much trouble has been caused during a long period
of time.
Stress corrosion has been as a recognised source of failure for
over 25 years. Despite the expenditure of large sums of money
on research during that period of time, however, it is still a
relatively little understood phenomenon about which there is
considerable ambiguity and argument. Furthermore, the problem cannot be said to be a diminishing one, since the number
of alloys known to be susceptible to stress corrosion cracking
and the number of environments that cause cracking have
both risen during that period. In each decade the problem has
been extended.
There are a number of reasons for this. With the development
of new alloys there nearly always appears to be some in a
given series or one in a particular heat treatment which is
found to be susceptible to stress corrosion cracking. In addition, greater demands are being made upon alloys that have
been available for some time and have been considered to be
immune. More severe service conditions may reveal susceptibilities or newly employed environments that were not previously even suspected may cause failure. With titanium alloys
susceptibility in sea water was discovered not in service but
through the employment of a new form of test. In other
situations alloy requirements are being extended not by exacerbating the operating conditions but by demanding predictable long lives of alloy components in environments in which
failure within a period of 20 years would result in a serious
economic disaster, e.g. nuclear reactor generating stations.
Over the last 25 years or so a large, widely dispersed and
mainly unco-ordinated effort has been made in the Western
World to cope with stress corrosion cracking. This has been
attempted on two broad, overlapping fronts. Firstly, there
have been attempts to cope with the problem pragmatically by
establishing testing programmes designed to reproduce the

specific operating conditions and to reduce susceptibility by


examining variables upon times to failure in a range of stress
rupture tests, often designed by the institution in which the
work is performed. One consequence of this has been the creation of bodies of data that are derived from tests often sufficiently incompatible as to prevent meaningful comparison. Secondly, there have been many attempts to investigate the problem
mechanistically in an attempt to resolve the separate reactions
or events that act in concert to produce crack initiation and
failure. Many institutions have employed both approaches at
the same time; such practice allows for possible interaction
between the practising engineer and the scientist in the laboratory. Communication between the user and the investigator is
of paramount importance to both, and it should be encouraged whenever possible.
Despite the large amount of work and thought that has
been put into stress corrosion mechanisms, a full understanding of them has still to be obtained. It is extremely difficult,
for example, to predict whether a new alloy will be susceptible
and in what environment, or to define what must be avoided
in designing a new alloy which might be stressed while in an
aqueous environment. This lack of predictability arises at
least partly from the complexity of the stress corrosion
phenomenon. It is difficult to avoid a certain diffuseness in
approach to the problem, since it has not been established unequivocally what the principle variables are. It is difficult
therefore to conduct experiments in which one variable alone
can be examined.
Within NATO the engineering aspects of stress corrosion
have been examined periodically by AGARD. What has not
been attempted previously is to provide a situation in which
workers within the NATO countries who are concerned with
the fundamental theories of stress corrosion cracking could meet
in order to discuss at length the limitations of current understanding, and thereby to focus attention upon those aspects
that urgently need to be examined in order to determine
definitely the various mechanisms of stress corrosion cracking.
The NATO Science Committee sponsored the Research
Evaluation Conference with that primary objective. There is
good reason for this. Much fundamental work on stress corrosion cracking is carried out in laboratories within the N A T O
countries. The major proportion of the important advances in
the fundamental understanding of stress corrosion has been
made in the past in those laboratories. The prospects for the
achievement of a definitive advance in understanding in the
future are, therefore, good. The achievement of the primary
objective should be seen in the next few years in the accellerated
pace with which fundamental concepts are clarified and the
mechanisms of cracking in the various systems become understood.

ANTI-CORROSION

The problem
Stress corrosion cracking in metallic lattices is concerned
with the nucleation and propagation of cracks that are induced by an environment. I n order to understand the mechanisms of cracking in any particular alloy system, it is necessary
to determine clearly:
(i) the environmental reaction or reactions that result in
(a) crack nucleation, and (b) crack propagation;
(ii) the role of the lattice structure in contributing to these
same reactions, e.g. the properties of grain boundaries,
dislocations, precipitates and the interfaces and regions
adjacent to them; and
(iii) the relationship between the load and the type of stress
that it creates, and how that stress affects both crack
initiation and propagation.
The crack initiation and propagation processes both consist
of a number of simultaneously-occurring electrochemical reactions. The initiation process commonly includes the breakdown of a protective film. For this to occur, there may be
a solid state reaction occurring in regions where (perhaps)
the conductivity of the film is raised prior to dissolution. The
influence of the metallic lattice structure in aiding such breakdown processes has sometimes been explained as arising from
solute segregation, sometimes demonstrated, sometimes hypothesised. This may cause differences in the film over a surface and
render it less resistant to the environment at certain sites. The
relationship between the applied load and the type of stress
that it produces can be considered on one scale that requires
the application of fracture mechanics to the specific configuration and on another scale that requires the examination of
dislocation patterns and their interaction with microscopical
structural features. Combined with this there is the need for
a careful design of the experimental arrangement, which must
include also an appreciation of why the different experimental
configurations and procedures employed in stress corrosion
studies do not always give the same result.

Interaction of separate disciplines


The division of the stress corrosion problem into the three
factors listed indicates clearly the complexity of the problem.
In the most general sense the three factors come under the
heading of (i) corrosion, (ii) physical metallurgy, and (iii)
fracture mechanics. Each of these is a separate and demanding
discipline and most workers in stress corrosion cracking have
been trained in one of them, some possibly in two and only
a few in all three. Inevitably, and unavoidably, there is a
tendency for individual workers to think along the lines of
their training. Electrochemists, physical metallurgists and
fracture specialists all tend to emphasise the significance to
stress corrosion mechanism of features that dominate their
separate disciplines. The potential at the tip of a crack, the
configuration of disclocations and the type of stress developed
in a specimen are all important, yet the understanding and
explanation of a crack propagation mechanism lies almost certainly not in any one measurement of whatever kind but in
the dynamic interaction of these factors and others. T h e
electrochemist must be aware of the difference between the
properties of an ideal static surface and the surface at the tip
of a propagating crack. There is the obvious problem here
that it is difficult to characterise that surface. The physical
metallurgist must consider whether dislocation patterns or
precipitate properties would affect the surface reactions that
occur at the tip of a crack. The fracture specialist must relate
continuum mechanics to surface reactions.
Such interactions will require discernment of a high calibre,
and will probably produce explanations of great subtleties.
Certainly in the last few years the necessary mutual awareness has been manifest, and this is a welcome sign. Hopefully

September 1972

that aspect of stress corrosion studies will increase as a result


of the Conference. A better understanding of stress corrosion
mechanisms would appear to be very dependent upon it, since
the interaction of disciplines requires also the interaction of
those trained in those diciplines.
The intellectual problem of determining the interaction of
three different disciplines cannot be over-emphasised. It is
difficult to determine the balance or the relative importance
of each discipline for any particular system of stress corrosion,
since this would appear to require, a priori, a good understanding of the cracking process. Furthermore, each stress
corrosion system appears to be different, and the relative importance of each discipline probably varies from system to
system. This is not altogether surprising, however. There is
no reason to expect a fixed relationship, and all the evidence
suggests that it varies considerably.
There appears to be a wide range, extending from stress
corrosion systems, where the stress play a predominant role,
to others where it serves little purpose other than to pull the
crack sides apart. In some of the higher strength alloys, for
example, stress corrosion fractures may be very similar to
mechanical failures. While the corrosion step is necessary, the
stress corrosion mechanism can be regarded as a particular
kind of recognised mechanical failure. In the lower strength
alloys, however, the type of fracture observed usually bears
no similarity to any type of fracture observed in these alloys
under non-stress corrosion conditions. Fractographic interpretation or even description is not easy, and explanation
only of a very simple kind can be attempted.
The role of the corrosion processes in these two systems is
rather different. In one case the corrosion process appears to
aid or promote a type of mechanical failure that can occur
under other circumstances in the absence of corrosion. In
the other, the corrosion process promotes a type of fracture
that is typical of stress corrosion fracture in these alloys
but is not otherwise observed. It can be understood only by
analysing the configuration at the tip of a stress corrosion
crack. The complex interaction of the microstructure with the
environment determines the type of fracture that is observed,
and a wide variety is obtained. In some systems the role of
the stress is minimal. It produces failure, but appears to contribute very little to the kinetics of the chemical reactions
occurring between the alloy and the environment. Titanium
in methanol would be a good example of this type of system.
The corrosion process, once initiated, penetrates the total
thickness of specimens even in the absence of an applied
stress or of a residual stress, so that specimens eventually fall
apart under the action of their own weight. The requirement
might appear to be a completely soluble corrosion product, yet
this does not appear to be always necessary. Oxide films may
form right through unstressed brass specimens in ammonia,
presumably in a situation where they are extremely porous.
In both cases the main function of the stress is to pull the
specimens apart. The type of fracture produced is usually
very simple, since it is associated with intergranular separation.
The relative importance of the applied stress is very low.
Perhaps the answer to the problem of attempting to achieve
an interaction of disciplines is to ensure that workers trained
in different disciplines are at least made aware of some of the
problems that arise outside their particular speciality. I n this
way they may gain insight not only into the problems of
others but see more clearly the limitations of their own
discipline when applied to stress corrosion. An electrochemical approach cannot by itself provide a complete mechanistic understanding any more than a purely metallurgical
approach.

The role of time


An additional factor of all cracking processes that must be

September 1972

ANTI-CORROSION

included in determining the mechanisms of stress corrosion


cracking is the role of time. A quantitative determination of
a stress corrosion mechanism must include not only what is
happening either in metallurgical or electrochemical terms but,
perhaps more importantly it must indicate the rates at which
such occurrencies are operating. It has been suggested1 in very
general terms that the important feature in a stress corrosion
crack propagation process is the time in which a sequence
of events occurs, e.g. the rate of slip step emergence or the
rate of repassivation.
Apart from those systems where there is a large mechanical
component to the propagation process, stress corrosion crack
propagation usually occurs rather slowly, c. 3/sec. These
rates may only be average values, and the precise physical
significance of them cannot be described if the propagation
process is intermittent. Such a process probably occurs
universally, perhaps varying in the actual physical scale of
the intermittent event, since it seems clear from fractography
that propagation processes do not occur at a constant rate in
polycrystalline specimens. At the crack tip new surface areas
of metal are created as the crack propagates, and the reaction
of them with the relatively small volume of liquid available
at the crack tip represents an extremely important stage in any
attempt to understand the cracking mechanisms.
A large number of events may be speculated about. Depend
ing upon how rapidly the fresh metal is created and upon the
ionic composition of the liquid in the region of the crack tip,
the crack may either move forward, halt, branch or exhibit
a change in propagation rate. The surface may repassivate, per
haps partially, thus forming a somewhat protective film. The
film may be broken or penetrated by the underlying deform
ing metal. Hydrogen discharge may result in a volume of
metal being embrittled. Hydrogen may help the destruction
of protective films by causing blistering or in general having
an adverse effect upon their adherence. It may also hinder
their formation. Electrochemical reactions on freshly exposed
metal may occur at rates that are orders of magnitude
greater than those occurring under equilibrium conditions.
The high reactivity of freshly exposed metal may promote
non-equilibrium reactions or unusual metastable species.
Around the crack tip plastic relaxation will occur and will
recur as the crack propagates. This will cause continual
changes in the crack tip shape. The character of such de
formation will depend upon the structural properties of the
specific alloy at the temperature of the test and upon the type
of test employed. It may be affected significantly by minor
alloying additions of either substitutional or interstitial ele
ments. It will be dependent upon the value of the stress acting
across the region at the tip of the crack. It may be increased
or reduced by the corrosion process. If, for example, atoms
are removed from the region of the crack tip, plastic deform
ation in that region might be expected to occur more readily.
If, however, thick corrosion film growth occurs in this region,
then it may hinder the plastic relaxation occuring at and
underneath the crack tip surface.

Significance of repassivation kinetics


The importance of film formation and film fracture during
stress corrosion crack propagation has long been recognised,
as is evident in stress corrosion literature. A number of pro
posed mechanisms include as an essential requirement the
formation of a film which reduces the reactivity of the under
lying metal surface. Others place emphasis upon the fracture
of a film as an integral part of a propagation mechanism.
One general model was proposed in 1966 which distinguished
the crucial step causing propagation as a critical rate of film
formation 1 ' 2 . The model was proposed to cover situations in
which the film might be either a metal compound or a noble

metal film. To emphasise the necessary protection afforded by


such films, the word 'repassivation' was employed.
The situation envisaged is drawn in Fig. 1. A passive sur
face exists in equilibrium with an environment. Under the
action of a tensile stress a slip step is produced. Since the
stable surface condition is a passive film this might be ex
pected to form on the newly-created metal surface. The ex
tent to which this occurs will depend upon the 'aggressive
ness' of the environment and upon the passivation characteris
tics of the metal, two general terms that cover many factors,
e.g. temperature, pH, electrode potential, alloy composition, en
vironment composition, etc. If the rate of repassivation is too
high, the high reactivity of a crack tip will be replaced by the
lower reactivity of a pit; if the rate is too low, attack over a
large part of the surface will occur and the highly localised
attack that is necessary for crack propagation will not occur.
The repassivation kinetics will be all important and will
clearly be affected by events occurring in the metal and by
events occurring within that small volume of the environ
ment at the crack tip.
This matter has been considered in detail2, although there is
little experimental knowledge of the situation. At the tip of
a stress corrosion crack it was postulated that the freshlycreated metal surface can be expected to adsorb from solution

species that will passivate that surface and species that will
maintain that surface in the active condition 1-3 . This will be
an important process, and the precise manner in which it
occurs should affect the propagation process. The ratio of
freshly exposed metal surface area to available solution volume
will also be important. It has been suggested1 that the rapid
increase in fresh area may itself cause localised depletion of
the solution of passivating species, thereby rendering the en
vironment increasingly aggressive and extending the repassiv
ation time. While the physical evidence for such events is poor,
such considerations do demand a simultaneous consideration
of electrochemical and metallurgical events. T o that extent
the model is of interest. Furthermore it is possible to consider
the reactions on various parts of the emergent step by attempt
ing to depict static polarisation curves for them. Again the
physical reality of such ideas is poor, but conceptually they
provide valuable insight into events at the crack tip.
The general argument was based upon the effect of the
rapid production of a wide slip step. In austenitic stainless
steels, -brasses and -titanium alloys, dislocation patterns in
thin foils of susceptible alloys made from specimens that have
been lightly strained exhibit co-planar arrays, which suggest
that coarse slip and therefore wide slip steps may occur on
free surfaces. Whether such steps occur at all stages of crack
propagation is unlikely but the creation of fresh metal in the

ANTI-CORROSION

vicinity of the crack tip is not in dispute. Thus the formal


representation of a step can be taken as indicative of fresh
metal area. The important process is that the step that forms
is so wide that repassivation of at least a partial nature cannot
occur within the time available on the whole surface.
In Fig. 2A, curve 1 represents the polarisation curve obtained from a susceptible alloy that exhibits passivity in a
neutral solution for that part of the surface that does not pit.
It is applicable to the sides of the crack. Curves 2 and 3
represent the behaviour of the surface under conditions of
increasing chloride ion concentration, falling pH and diminishing dissolved oxygen content. These are the conditions that
develop within a pit. The distinction between Curve 1 and 3
probably has no real meaning, since there will be a gradual
transition from Curve 1 representing passivity to Curve 3
representing conditions near the crack tip. The surface in this
region is covered with a film affording little protection, a situation described as 'partial passivation'. The potential at this
surface will be low, and hydrogen discharge will occur. Curve
4 represents the dissolution of the active metal on an emergent
slip step as drawn in Figs. 1 and 2B. The difference in
current density on Curves 3 and 4 at points in the region of
A and B must be large enough to propagate a crack. If it is
small, then the aspect ratio (length/width) of the crack will

September 1972

be low, and as this difference is reduced the crack will tend


to develop into a broad crack, then a fissure and then a deep
pit. Alternatively, the ratio of current densities obtained by
the intersections of Curves 4 and 1 with the cathodic polarisation curve may be the determinant of whether a general
pitting situation represented by Curve 3 becomes a crack.
On this elementary model the continuous failure to passivate at least partially (Curve 3) the base of an emergent step
which maintains the chloride solution at the tip on a concentrated de-aerated non-protective condition accounts for crack
propagation. The model proposed placed emphasis upon the
role of the electrode potential in determining repassivation
kinetics. It was suggested2 that the critical delay might be expected to occur over narrow potential ranges associated with
transitions, e.g. the active to passive transition. In addition
the environmental composition and the alloy composition
were considered with respect to the electrode potential in
determining the critical delay in repassivation.
Cracking was considered to be a very special form of pitting
arising from the effect of metal surface changes upon the
local chemistry. It was demonstrated 4 that in austenitic stainless steels, for example, the principle effect of pitting is a
chemical one and not a physical one, and that in 1N HC1 at
room temperature pitting is a pre-requisite for cracking to
occur.
Two important ratios appear to be important by this type
of analysis:
(i) the area of freshly created metal surface: the volume
of solution in the crack tip region
and
(ii) concentration of aggressive species: concentration of
passivating species in the crack tip region.
The general concept that more metal surface area is created
than the solution can passivate can be divided into finer points.
The description of plastic deformation occurring at the crack
tip by an emergent slip step is purely formal. Such a process
may occur in ductile alloys at low stresses, but the deformation
occurring about a crack tip is likely to take a number of
forms; which occurs will depend upon the alloy composition,
heat treatment, temperature, etc. On the environmental side,
whether a surface passivates will depend not only on the
solution composition but also upon the surface composition.
It is conceivable, therefore, that the rate of formation of new
metal surface could be the same for two alloys, only one of
which was susceptible to stress corrosion cracking. One alloy
would crack and one would suffer only from pitting.
The importance of the ratio of species was considered in
some detail 3 in 1968. The aggressive species can be divided
into two classes, one of which causes cracking, the other one
of which causes corrosion. In general, for cracking to occur
there must be some ratio, R, of cracking to passivating species
above which cracking occurs. If, in addition, a non-cracking
aggressive species is present, then the corrosion front will be
widened. Cracking will be accompanied by corrosion on the
crack sides, with a consequent change in the crack morphology. The presence of non-cracking aggressive species is
helpful in creating and maintaining the necessary acidity at
the tip of a propagating crack; its effect is 'additive' in the
sense that the concentration of cracking species that is required is less than in its absence. From what has already been
discussed the variation of these ratios with time is also of
importance.
Incorporated in these two simple ratios are many complexities. On the metallurgical side, the deformation and fracture
mode and how these may be affected by surface films or
anodic dissolution, and the creation of active sites, have long
been recognised as important. On the electrochemical side,
repassivation kinetics, transport processes through the electrolyte and the electrochemistry of strained and straining metal

September 1972

ANTI-CORROSION

are important. Included in such analyses' must be an explana


tion of the relationships observed between crack velocity and
stress intensity factor. The Region I relationship, for example,
in which the stress intensity factor is proportional to the
logarithm of the crack velocity, appears to occur only in those
situations in which some corrosion occurs. It is not observed
with titanium alloys in sea water. An explanation for such
a relationship may lie in a consideration of repassivation
events.
The model of repassivation has been applied inferentially to
the stress corrosion of titanium alloys3. Stress corrosion
cracking occurs in Ti-5Al-2.5Sn alloys when specimens are
strained dynamically while exposed to aqueous NaCl and to
methyl alcohol containing 1 per cent HC1. At high strain
rates, obtained by employing a high crosshead speed on the
testing device, plastic deformation rapidly leads to ductile
fracture, because there is insufficient time for crack initiation
and propagation to occur. As the crosshead speed employed
is lowered cracking occurs, and the elongation to failure is
considerably reduced, as shown in Fig. 3. When the crosshead speed is further lowered the observed behaviour of the
specimen depends upon the environment employed. The alloy
is usually passive in aqueous solutions, and at very low crosshead speeds the rate of deformation of the surface is so low
that film repair can occur in a shorter time than crack initia
tion. The result is an absence of cracking and a return to the
air fracture observed at high crosshead speeds. The methanol
solution, however, produces a different effect at low crosshead
speeds, since it is aggressive to the alloy. Film repair cannot
occur. As the crosshead speed is reduced there is a longer
time for crack initiation and propagation. The result is a
gradually diminishing elongation to fracture.
Such an explanation needs to be examined very carefully.
At low strain-rates the frequency of occurrence of slip steps
will be lower, but the speed of emergence of any individual
step may be very similiar to that occurring at high strainrates. On a localised scale, therefore, the effect of strain-rate
may be negligible, in which case crack initiation might be
expected. Until such events are examined more closely, there
is no completely satisfactory answer. In titanium alloys crack
ing is initiated by cleavage, for which it is likely that a
critical stress is necessary for it to occur. To explain results
such as that shown in Fig. 3 requires a consideration of re
passivation in association with whatever is causing cleavage,
e.g. hydrogen. Thus to consider the repassivation process in
isolation may not always be worthwhile. Experimentation and
interpretation must be based upon that process together with
the consequent fracture process. For titanium alloys a qualita

tive attempt was made to do this by considering simultaneous


ly the increasing stress, the repassivation behaviour and the
hydrogen process as a function of strain-rate 0 . The interpreta
tion was based upon film formation creating an important
barrier to rapid hydrogen entry. With so many processes
occurring much more data are required before quantitative
models can be formulated, but it does seem to the author
that this is the correct approach.
The basic model proposed 1 has also been applied to the
formation of tunnels 7 . The general conclusion was drawn that
tunnels are nucleated at the active/passive transition and occur
in environments that do not cause cracking as well as in those
that do. The difference was thought to arise from the depth
of tunnel that was formed. In environments that caused crack
ing repassivation was more difficult, and the observed depth
of attack was greater than in non-cracking environments. It
has also been applied to the stress corrosion of alloys in the
active potential range 6 , in which tunnelling is observed as the
principle transgranular fracture mode. The film in such
situations is considered to be an adsorbed inhibiting layer of
halide ions 8 . Recently cracking has been observed9 in mild
steel exposed to aqueous CO/CO 2 mixtures, which might be
ascribed to the breakdown of an adsorbed layer of inhibitive
CO molecules. It is possible therefore that the protective film
that is thought to form in the proposed repassivation model
may not only be some form of passive film or noble metal
film but may be an inhibitive layer formed from some species
in the environment.
More recently, the model proposed in 1966 of a critical
delay in repassivation has been applied to stainless steel by
Staehle. The results of straining electrode experiments in that
system seem to confirm that the model proposed is correct,
viz. that a critical repassivation delay is observed only over
the potential range within which cracking occurs, and that
susceptible alloys exhibit somewhat more pronounced effects.

The crack path


It has been a matter of considerable discussion which path
a crack follows, intergranular or transgranular, and what de
termines it. The early simple idea that there is a pre-existing
path that is preferentially corroded applies to titanium metal
in methanol and perhaps also to iron in nitrate solutions. It
is by no means clear why the grain boundaries in those metals
should be selectively attacked. General concepts of grain
boundary impurities are readily involved and may well apply,
but it is difficult to obtain definite proof. It is also not clear
why only certain environments attack these grain boundaries
in such a selective way. There are clear metallurgical and en
vironmental variables which, to a limited extent, can be in
vestigated separately. With most systems of stress corrosion
the interaction is more complex. Cracking does not follow a
pre-existing path because it does not exist. In the unusual
case of stressed -titanium alloys in methanol, the pre-existing
path of the grain boundaries does not provide the only crack
path, since this undergoes a transition at a certain stress level
and subsequently follows a transgranular cleavage mode at
much higher velocities than could be attained by a grain
boundary dissolution mode.
I n general, to explain the path it is necessary to invoke the
effect of the stress (a) in promoting reactive features in the
structure, or (b) in providing emergent slip steps which
destroy surface films that might be tending to form at the
crack tip, as has already been discussed. The reactive features
may be plastically-induced precipitates or dislocation pile-ups
that have postulated solute segregation about them, or merely
the freshly-exposed surfaces of wide emergent slip steps. The
formation of films that might be fractured depends upon the
environmental conditions, particularly the pH, which is prob
ably characteristic for a given stress corrosion system.

10

ANTI-CORROSION

In most systems of stress corrosion cracking does not follow


one mode exclusively. Instead one path predominates. This
has not always been recognised, and the observation will
render inadequate mechanistic hypotheses or theories that
exclude either the intergranular or transgranular mode. The
transition may depend upon the potential, the solution com
position, or the level of stress or possibly combinations of
these. It is not a subject that has been widely investigated.
The stress corrosion nomenclature tends to be confused in
describing what is a general phenomenon. Stress corrosion
and hydrogen embrittlement are commonly distinguished
between, even in situations where the critical corrosion re
action in the stress corrosion process is not clear. It is
difficult to see what justification there can be for this. If,
for example, a delay in repassivation is responsible for
allowing hydrogen to promote transgranular cleavage in
-titanium alloys then the role of the stress in affecting the
corrosion process in this critical way can just as well be called
stress corrosion as hydrogen embrittlement. The argument
can be made, however, that hydrogen can cause failure under
conditions where no corrosion occurs as, for example, in the
cathodic charging of high strength steels. The path of such
fracture may be different, since it may not be associated with
sites of preferential anodic dissolution which would provide
entry points for hydrogen under conditions of open circuit
or anodic polarisation. It is not always clear, however, what
advantages are gained by attempting to separate failure
phenomena occurring under open circuit conditions, and per
haps such distinctions in nomenclature should be dispensed
with.
A failure that occurs by cracking only when a stressed
specimen is exposed to a certain environment is a stress corro
sion failure, and it little matters whether the main corrosion
reaction is anodic or cathodic. That hydrogen embrittlement
occurs in a variety of environments where as stress corrosion
occcurs in rather specific environments also seems rather a
hollow distinction. Increasingly, the range of environments
that cause stress corrosion cracking is extending and has there
by become less specific in nature. The term stress corrosion
embrittlement also appears to serve little purpose. The process
involves embrittlement of an engineering kind (a reduced elong
ation to failure) but on an atomistic scale there may be
marked amounts of plastic deformation. Cracking seems such
a plain unequivocal term that it should not be replaced by
embrittlement.
So many methods of testing are employed in stress corrosion
studies that inevitably some confusion exists over which is the
best method. Not all give the same result and, while this can
usually be explained, the differences promote doubt and un
certainty. They hamper communications between workers very
considerably. Hopefully this confusion over stress corrosion
testing will soon be dispelled. The attractiveness and advant
age of an agreed stress corrosion test is very great, and both
in Europe and in the U.S.A. attempts are being made at
least to rationalise the type or types of tests that are employed.
With the literature full of many different types of tests em
ploying different loading devices, surface finishes, pre-exposure
times, notched or plain specimens and many other variables,
any rationalisation cannot help but be an improvement.

Methodology and mechanisms


The general complexity of the stress corrosion cracking pro
cess accounts for the difficulty that there has been in analys
ing in an unambiguous fashion the main components of the
mechanisms in any single system. In the literature associated
with stress corrosion a number of mechanisms have been pro
posed which have subsequently been modified and/or com
bined in a number of ways. Many experimental techniques
of considerable ingenuity and originality have been employed
in order to examine cracking processes.

September 1972

Stress corrosion is notable for controversies and arguments.


It is also notable for hypotheses. Elementary scientific method
ology is required in order to settle many of the arguments.
Hypotheses must be tested. These tests must take the form
of finding out whether or not the deductive consequences of
any particular hypothesis are statements that correspond to
reality. The predictions arising from the hypothesis about
what is not yet known are of crucial importance. With the
wide range of analytical tools that is available both to the
chemist and to the metallurgist, the time must shortly come
when some commonly assumed situations can be firmly estab
lished or rejected. The physical reality of the schematic draw
ing of Figure 2 from which Figures 3 and 1 are derived would
be a typical example. Practically nothing is known about the
shape of the crack tip or the mechanical properties of the
material immediately ahead of the crack tip.
Other mechanistic ideas are not so far advanced. One example
will be examined, but it is not unique. The mechanism of
stress sorption has been hypothesised by several schools of in
vestigators and for a wide range of susceptible alloys. The
essence of this mechanism is that the adsorption at the tip
of the crack of the specific species that causes stress corrosion
cracking in susceptible alloys lowers the bonding energy of
atoms in that region sufficiently to cause rupture under the
action of the externally applied stress.
Phenomena of a similar kind appear to occur in nonmetallic solids and in liquid metal embrittlement. In metals
the situation is less clear. The term pseudo-Rebinder has been
proposed 10 to account for adsorption phenomena on metal
surfaces which result in apparent decreases in flow stresses,
but which may be attributable to either adsorption-induced
decreases in the hardness of the oxide film on the metal surface
or to dissolution of a work-hardened surface layer. Either
such effect might result in a lower flow stress. A genuine
adsorption-induced reduction in strength has not been
demonstrated clearly. This particular mechanism of stress
corrosion is still therefore at the hypothesis stage. This does
not in any way invalidate it. Neither does the difficulty of
obtaining experimental evidence. Nevertheless it remains in
a curiously undeveloped state when consideration is given to
the number of papers that have been published which con
clude by explaining stress corrosion mechanisms by the
occurrence of such a process. Furthermore, unlike most
hypotheses concerning stress corrosion mechanisms, no pre
dictions have been made about stress-sorption phenomena as
applied to stress corrosion which can be put to test. The
mechanism remains a hypothesis after 15 years of invocation.

References
1.
2.
3.
4.

J. C. Scully, Brit. Corros. J., 1, 355, (1966).


J. C. Scully, Corros. Sci., 7, 197, (1967).
J. C. Scully, Corros. Sci., 8, 771, (1968).
I. S. McCollough and J. C. Scully, Corros. Sci, 9, 651,
(1969).
5. J. C. Scully and D. T . Powell, Corros. Sci., 10, 371,
(1970).
6. D. T . Powell and J. C. Scully, Corrosion, 24, 151, (1958).
7. I. S. McCollough and J. C. Scully, Corros. Sci., 9, 707,
(1969).
8. J. D. Harston and J. C. Scully, Corrosion, 25, 493, (1969).
9. A. Brown, J. T . Harrison and R. Wilkins, Corros. Sci.,
10, 547, (1970).
10. R. M. Latanision and A. R. C. Westwood, 12th Tech
nical Report to ONR, RIAS, Baltimore, Md. (1959).
General
Fundamental Aspects of Stress Corrosion Cracking (ed.
R. W. Staehle. A. J. Forty and D. van Rooyen), N.A.C.E.,
Houston, Texas (1969), contains an extremely compre
hensive set of papers covering most aspects of stress
corrosion cracking.

Você também pode gostar