FAILURE MECHANISMS AND METALLOGRAPHY: A REVIEW

I. Le May*

INTRODUCTION

In spite of the best efforts of design engineers and materials specialists, and in
spite of well-thought-out procedures for quality assurance and control, engineering com-
ponents still fail in service from time to time. In the majority of cases the failure does
not lead to serious consequences and the component is merely replaced as having served
its useful life and worn out, and in such cases optimization of the procedures of design
and quality assurance have probably not been of unduly great concern, economic con-
siderations dictating that avoidance of failure is not of overriding importance. However,
in other cases failure may lead to serious consequences which can cause serious financial
loss, environmental contamination or loss of life: in such situations and to minimize
the risk of failure, the design and quality assurance procedures must have been very care-
fully studied and controlled, and in the event of such a failure occurring it is frequently
necessary to establish the root causes in terms of design, choice and quality of material,
fabrication procedure, and so on. Thus, the metallographer may be required to conduct
an investigation to determine the mechanisms of failure, based on the appearance of the
fracture surfaces and the microstructure of the components which have failed. It is the
purpose of this paper to review the various mechanisms of fracture and to discuss the
identifying morphological and microstructural features appropriate to each.

BASIC FRACTURE MECHANISMS

There are two basic and distinct mechanisms of fracture, namely cleavage and duc-
tile fracture. The former occurs under tensile stress and involves separation along cry-
stallographic planes with little or no plastic flow taking place, and leads to a brittle ap-
pearance of the fracture surface, with a small amount of energy being dissipated: the
latter involves plastic deformation by sl ip and the energy dissipation involved is much
greater, but depends on the extent of the plastic flow. It will be large when extensive
shear, necking or void formation occurs, but relatively small when plastic flow is local-
ized to the tip of a propagating crack, in which case the fracture may be termed brittle
in an engineering sense, but is ductile on a strictly mechanistic basis.

However, it is insufficiently informative to classify fractures strictly in terms of

the mechanisms of cleavage or ductile fracture, and we must look at the various factors
which allow and cause cracks to propagate gradually or in a stepwise manner until they
reach a critical length at which rapid failure takes place under load, either by cleavage,
ductile fracture or some combination of the two. Indeed, it is relatively uncommon to
find fractures of a simple overload nature outside of the laboratory, and much more

*University of Saskatchewan, Saskatoon, CANADA S7N OWO

J. L. McCall et al. (eds.), Metallography in Failure Analysis

© Plenum Press, New York 1978
2 I. LE MAY

common to find that failure in service has taken place over a period of time under the
normal working conditions of loading. Thus, we shall identify fatigue crack growth, inter-
granular cracking and environmentally-assisted cracking as additional mechanisms leading
to failure.

Fatigue crack growth occurs under repeated or cyclic tensile stress and is by far the
commonest cause of failure in engineering components and structures. Intergranular
cracking may arise in several ways: it may take place during creep deformation at high
temperature either as a result of grain boundary sliding or from the growth of voids at
grain boundaries as a result of vacancy deposition; or it may arise from segregation of
specific elements at grain boundaries, so leading to an embrittled structure which can
fracture under impact or monotonic loading. Environmentally-assisted cracking relates
to fracture under the combined action of stress and an environmental effect such as cor-
rosion, the presence of an embrittling fluid such as hydrogen, or the adsorption of a
chemical species on to the fracture surface. Thus, environmentally-assisted cracking in-
cludes the phenomena of stress corrosion cracking, corrosion fatigue, hydrogen embrittle-
ment and liquid metal embrittlement. In addition, we may extend it to include failure
caused by radiation damage in conjunction with the service loading.

Two additional mechanisms of failure may be identified, corrosion and wear. These
are both extremely important from a practical standpoint, although they are of a some-
what different nature to the mechanisms of fracture identified above.

INVESTIGATIONAL PROCEDURES

The determination of the failure mechanisms which were operative in causing a

component to fracture or cease to function involves consideration of the design, specifi-
cations and service conditions for the component, the evidence of witnesses to the fail-
ure, and detailed examination of the fracture surfaces on both macro- and micro-scales,
together with possible destructive tests to determine the microstructure and mechanical
properties of the part. It is rare that evidence provided by the fracture surfaces alone
will furnish sufficient information to determine unequivocally the cause(s) of failure,
and such evidence may be quite misleading, for example when artifacts are present due
to rubbing or subsequent damage inflicted on the fracture surfaces. The overall proce-
dures involved in a failure analysis are discussed fully in appropriate reference texts [1,21.

It should be emphasized that it is absolutely essential that all non-destructive exam-

inations and metallography are completed and documented before any destructive test-
ing is begun.

Macro examination, both visual and under a low power microscope (preferably a
stereo microscope), can be of great value in determining the overall nature of the frac-
ture (e.g., whether it is essentially brittle or ductile on an engineering scale), the relation-
ship of the fracture plane or planes to the operative directions of stressing, the point of
initiation of the failure, and damage which may indicate an impact or other load. Thus,
Fig. 1 illustrates a fatigue failure with the origin of the fracture pinpointed by means
of the beach marks or arrest lines, while Fig. 2 shows clearly the path of a brittle frac-
ture arising from impact loading. In many such cases it is unnecessary to conduct de-
tailed examination of the fracture surfaces on a micro-scale, and one may move directly
to metallographic study of the microstructure at the crack initiation point.

Detailed microscopic examinations, the results of which we shall relate to the

specific mechanisms of fracture, include metallography using both replica techniques
and transmission electron microscopy (TEM) and direct observation under the scanning
electron microscope (SEM), as well as optical and electron microscopic examinations
FAILURE MECHANISMS AND METALLOGRAPHY 3

(al

(bl

Fig. 1. Matching surfaces, (a) and (b), of fatigue failure in a crank-shaft. Origin of frac-
ture is arrowed.

Fig. 2. Brittle fracture in connecting rod which broke suddenly following fatigue failure
of bolts at big end bearing. Arrow indicates crack propagation direction.
4 I. LE MAY

of metallurgical specimens cut from the failed component. In addition, it may be neces-
sary in some cases to determine the distribution of specific elements adjacent to the frac-
ture surfaces, or to identify particular precipitates which lie on these, using more special-
ized analytical techniques such as Auger spectroscopy or electron microprobe analysis,
these and related techniques being discussed in a previous symposium volume [31.

Analysis using the SEM has considerable advantages regarding simplicity and the
avoidance of the somewhat tedious procedures involved in replication, but in practice
failed components are generally much too large to be examined directly in the SEM, and
so the component must be cut up for such analysis: consequently, examinations using
replicas and the TEM may be more appropriate if the fracture surface and the component
are not to be destroyed [41.

The techniques involved in preparing replicas for TEM are well documented and,
rather than discuss them here, the interested reader is referred to the appropriate sources
for detailed information and instructions (see e.g., [5,6,71 ).

SPECIFIC FAILURE MECHANISMS AND METALLOGRAPHIC EVIDENCE

In examining the various failure mechanisms and metallographic evidence associated

with each, we shall rely heavily on TEM studies using replicas, as these provide much
finer detail than do optical microscopy or most SEM studies. Where appropriate, these
will be supplemented by additional metallographic evidence based on sections through
the microstructure.

Cleavage

Cleavage may occur in BCC and HCP metals and is associated with rapid loading
and low temperature. It is facilitated by the presence of high triaxial stresses which
allow a high tensile stress to exist for a given value of uniaxial yield stress. Cleavage can
be a major problem in low strength structural steels, and has in the past led to the catas-
trophic failure of many large structures including the classic case of the tanker Schenec-
tady in calm, cool weather while at her fitting-out dock moorings [8] (see Fig. 3).

Since the late 1940's extensive study has been made of the conditions leading to
brittle fracture in steel, and our knowledge of the transition from transgranular cleavage
to ductile fracture as temperature is raised and of the effect on this transition tempera-
ture of metallurgical and microstructural variables has become reasonably complete.
Nonetheless, such failures still do occur today, and are of particular concern when deal-
ing with structures for low temperdture environments such as the Arctic.

Because of its crystallographic nature, cleavage leads to a fracture surface composed

of flat, shiny facets. The crack is essentially flat and lies approximately normal to the
operative tensile stress axis, but changes its orientation slightly when it crosses a grain
boundary because of the differing orientations of adjacent grains and because the crack
follows specific planes (e.g., { 100} in a-iron). If the boundary is a high angle or tilt
one, the crack may restart in the adjacent grain as a single one; however, in the case of
a twist boundary, many small parallel cracks may form, with cleavage steps between
them (Fig. 4). These steps run together to form larger ones and lead to the characteris-
tic river patterns seen in cleaved polycrystalline metals (Fig. 5). Cleavage steps may also
form at the intersection of a cleavage crack and a screw dislocation [91, as shown in
Fig. 6.

Another characteristic feature observed in cleaved iron is that of tongues [101,

which are formed by local fracture along twin-matrix interfaces, the mechanism being
FAILURE MECHANISMS AND METALLOGRAPHY 5

Fig. 3. The T -2 tanker Schenectady after sudden, brittle failure while in her fitting·out
dock. (Courtesy of the U.S. Ship Structure Committee).

Twist boundary

Fig. 4. Schematic showing cleavage steps produced when cleavage crack crosses a twist
boundary.

illustrated in Fig. 7, while the appearance of the resulting tongues on a microfractograph

is shown in Fig. 8. Tongue formation occurs as a result of the high velocity at which a
cleavage crack propagates (it has a limiting velocity between 0.4 and 0.5 of the speed of
sound) which produces a local strain rate which is too high for slip processes to provide
all the accommodation required, and hence many twins are normally formed just ahead
of the moving crack tip.
6 I. LE MAY

Fig. 5. River pattern on cleaved mild steel bolt.

Direction of

Fig. 6. Schematic showing a cleavage step initiated at a screw dislocation intersecting

the cleavage plane.

Although cleavage is crystallographic in nature, in metals local plastic flow will

begin when the local stress reaches the yield stress for shear, and a cleavage crack can
initiate after some critical displacement has occurred at the crack tip. Above the due·
tile-brittle transition temperature (DBTT), yield will take place rather than cleavage.
FAILURE MECHANISMS AND METALLOGRAPHY 7

Fig. 7. The formation of a tongue at a twin-matrix interface in a-iron.

Fig. 8. Microfractograph showing tongues (arrowed) in cleaved mild steel.

During the propagation of a cleavage crack below DBTT, plastic flow always takes place
at the crack tip where a high stress concentration exists. Thus, the energy which is re-
quired to create the new surfaces is greater than that strictly required for separation of
atomic planes.

A number of models have been formulated to explain the nucleation of cleavage

cracks on the basis of dislocation movement, the simplest one envisaging dislocations at
8 I. LE MAY

the head of a pile-up coalescing to form a wedge-shaped crack nucleus [11] as shown in
Fig. 9. The resulting stress concentration may cause initiation of a crack on a plane
roughly normal to the sl ip plane, when the stress on this reaches a value equal to the
theoretical strength. An alternative model was proposed by Cottrell [12] and is shown
in Fig. 10. In this, intersecting dislocations on 110 slip planes coalesce to form an im-
mobile dislocation on a 100 cleavage plane, the crack nucleus growing as more dislo-
cations pass down the sl ip planes.

T-TO~

.1. .1.. J..

Fig. 9. Cleavage crack nucleated from coalescence of dislocations at head of a pile-up.

Cleavage cracks may also be initiated at twins [13] or in second phase particles
such as carbides in steel. Smith [14] has discussed the case of cleavage being initiated
in a grain boundary carbide by plastic deformation in the neighboring ferrite grain as
shown in Fig. 11. This may, in turn, initiate cleavage in the next grain if the local grain
boundary stress is sufficiently high and crack formation is energetically favorable. Smith's
theory is of particular value as it allows us to examine the important effect of carbide
thickness on the brittle fracture of steels in a semi-quantitative manner.

Ductile Fracture

Ductile fracture involves plastic flow, and the general term covers awide range of
failure modes. Under monotonic loading, and in the absence of cleavage, the ductile
fracture may take place with a large degree of variation in the extent of plastic deforma-
tion, and when this is small the fracture may still be considered brittle in an engineering
sense, with the crack propagating in a plane normal to the applied tensile stress. In such
a case, however, the localized plastic deformation at the crack tip precludes the forma-
tion of cleavage facets and steps.

Various types of ductile fracture are illustrated schematically in Fig. 12, ranging
from tensile fracture in pure, ductile metals which can result in necking down to a
point, to creep fracture at high temperature by intergranular cavitation. In the present
discussion, we shall concentrate on the rupture of engineering materials containing
second phase particles or inclusions, and defer discussion of intergranular creep fracture
to a later section. This restriction is hardly a severe one, as engineering materials always
FAILURE MECHANISMS AND METALLOGRAPHY 9

(001 )

(a)

(b)

Fig. 10. Cottrell's model for cleavage crack initiation in BCC metals from intersecting
dislocations forming immobile dislocation on the (001) cleavage plane. Additional dis·
locations moving down slip planes cause nucleus to grow as in (b).

contain large quantities of second phase particles, ranging from fine precipitates which
are only detectable using thin foil TEM as in precipitation hardened alloys, to large par-
ticles visible under the optical microscope and consisting of various alloying elements
which have been added to improve some properties, such as machinability.

The mechanism of ductile fracture is generally related to the initiation of micro-

voids at second phase particles, and their subsequent growth and coalescence. As a con-
sequence of this process, the fracture surfaces contain dimples which show evidence of
the initiation points at the particles (see Fig. 13). Depending on the mode of fracture,
viz., tension, shearing or tearing, the dimples may have equiaxed or elongated shapes,
and their formation is illustrated in Fig. 14, while the two different dimple shapes are
apparent in Figs. 13 and 15 respectively. Obviously, in order to distinguish between
shear dimples and tear dimples, it is necessary to prepare matching replicas from adjacent
surface areas. When the particles are large and brittle, microvoid initiation may take
place by the fracture of the particles rather than decohesion at the particle-matrix inter-
face. Thus, cracked particles may be observed during metallographic examination of
longitudinal sections through the material, and cleaved particles may be seen on the
fracture surfaces adjacent to the dimpled surface of the matrix.
10 I. LE MAY
a a

t t

a a
Fig. 11. Cleavage crack initiated in grain boundary carbide by plastic flow in neighboring
ferrite grai n.

A simple model for microvoid initiation by decohesion at the particle-matrix inter-

face was proposed by Gurland and Plateau [15], being based on the concept that the
strain energy relieved by void formation must be sufficient to produce the required energy
for the newly created surfaces. This lead to an expression for the applied uniaxial tensile
stress, 0, to produce decohesion as
1 E'Y %
o = - ~) (1)
q a
where q is the stress concentration factor at the inclusion, 'Y the specific surface energy
of the crack, E a weighted average of the elastic moduli of inclusion and matrix, and a
is the particle size (diameter). This model has been extended by Lui and leMay [16]
to take into account the plastic work being dissipated around the particle, and the follow-
ing relation is obtained
FAILURE MECHANISMS AND METALLOGRAPHY 11

...
, .. .., ..
(a) (b) (e)

(d)

Fig. 12. Schematic illustration of different types of ductile fracture in tension : (a) Neck-
ing to a point in very pure metal ; (b) Failure by shearing off; (c) Cup and cone fracture
caused by void nucleation and coalescence with final shear; (d) Intergranular creep frac-
ture showing different void nucleation points.

(2)

in which or is the average yield stress of the matrix for a given particle shape and volume
fraction of precipitates, and is assumed to be independent of particle size and strain. V
is the volume of a particle, while 6V IV indicates the volume deformed around the parti-
cle in relation to the particle, and is regarded as being approximately constant for given
particle shape.

Equation (2) indicates that the stress to initiate fracture is a function of (1 /a)%,
and it has also been shown that the UTS varies in a like manner for some high strength
quenched and tempered steels and for spheriodized carbon steels, for a given carbide
shape [16]. Hence, it has been suggested that the controll ing step in ductile fracture in
such materials may be microvoid initiation, this occurring at a point close to the maxi-
mum on the tensile load-extension curve.
12 I. LE MAY

Fig. 13. Equiaxed dimples on fracture surface of quenched and tempered 4140 steel.
Note the carbides at which local fracture initiated (arrowed).

Where extensive work hardening takes place, voids grow under the influence of a
tensile stress as well as of a lateral stress, as occurs during necking, and it has been demon-
strated that the fracture ductility can be calculated on the basis of the voids meeting each
other, and is seen to be a function of the volume fraction of voids [171. In high strength
materials, where work hardening capacity is generally low, the microvoids may link up
with a propagating crack by shear fracture along slip bands [181, as illustrated in Fig. 16.
This can give rise to a distinctly jagged surface as may be seen on the weld metal shown
in Fig. 17 which failed in a tensile stress-rupture test.

Fatigue

The great majority of engineering components fail by fatigue, i.e., under repeated
or cyclic loading, but one cannot assume that fatigue is the cause of failure without the
accumulation of appropriate evidence.

In examining the nature of fatigue fractured surfaces it is important to distinguish

between the three stages of fatigue failure, viz., crack initiation, crack propagation, and
final rupture which occurs when the crack reaches a critical length at which point it
propagates rapidly to cause sudden failure during a single tensile load application. Fatigue
is not in itself a mechanism of fracture, and we must look at the separate mechanisms of
crack initiation and crack propagation.

Fatigue cracks generally initiate at a surface discontinuity or stress raiser, their

formation being aided by the presence of a notch or inclusion; however, they may in
some cases initiate at a subsurface defect such as a casting defect, inclusion, or quench
or hydrogen-assisted crack. They are formed through the occurrence of local cyclic plas-
tic deformation, which generates high local dislocation densities on slip planes, and can
lead to the formation of small surface intrusions and extrusions which may be seen either
by optical or replica electron microscopy. Figure 18 illustrates these schematically. The
nominal stresses at which fatigue cracks can form may be well below the elastic limit of
the material, but on a micro-scale the local stresses may be much higher as a result of the
stress concentrations around defects and inclusions, leading to highly localized plastic
deformation which is not apparent on a normal engineering scale.
FAILURE MECHANISMS AND METALLOGRAPHY 13

t
G
t la)

-laa ~ ~\ cv=J IecceI

- ~ I ::>J~j
(b)

t ~ Iecce I
B t::::J leccel
t Ie)

Fig. 14. Formation of dimples by (a) tension; (b) shear; (c) tearing. Variation in frac-
ture morphology is shown at the right.

Fig. 15. Elongated dimples on fracture surface of quenched and tempered 4140 steel.
14 I. LE MAY

t
a a

t
•
•
Vo id from particle - matrix

a a

Fig. 16. Microvoids linking by shear along slip bands in a high strength alloy with limited
work harden ing capacity.

Fig. 17. Fracture surface of 316L weld metal specimen tested at high temperature in
tension, showing jagged appearance from linking of microvoids by shear.
FAILURE MECHANISMS AND METALLOGRAPHY 15

t
Surface ....
,",
X,

~
.. ~
.. "
,
", ,"
...
~, "
", "' ,

t t
Fig. 18. Formation of intrusions and extrusion at surface during fatigue loading.

A fatigue crack propagates as a result of local plastic deformation at the crack

tip, this occurring during each tensile loading period. Hence, we may note that fatigue
cracks will not propagate under compressive loading conditions. The crack grows initially
along a slip band where the initial nucleus was formed in the highly dislocated structure,
and after some time it changes from this Stage I crack propagation mode to Stage II mode,
which is roughly normal to the tensile axis, as indicated in Fig. 19. The length of the
Stage I crack is generally small, and its surfaces are often essentially featureless; how-
ever, during Stage II crack propagation, which extends over a much greater distance, dis·
tinctive features termed fatigue striations (see Fig. 20) are frequently formed on it. The
morphology of these striations varies widely with material and environment, and to
understand the reasons for this, we shall look more closely at the mechanisms of fatigue
crack propagation.
Laird and Smith [19] proposed their "plastic blunting model" for the formation
of striations during stress reversal, and this is illustrated in Fig. 21. A number of other
workers have proposed mechanisms for striation formation based both on a continuum
mechanics approach and on consideration of slip processes at the crack tip. In particu-
lar, Neumann [20], Pelloux [21], and Broek and Bowles [22] have made detailed
examination of the slip processes, and the main features of the coarse slip model which
has been proposed as a result of their studies are shown in Fig. 22. It is seen that the
crack propagates by slip occurring on intersecting slip planes meeting at the current
crack tip, producing crack blunting. Upon load reversal, slip takes place in the reverse
direction, but the crack does not reweld because of surface oxidation, and crack tip
resharpening occurs. Hence, a series of fine slip steps appear on the fracture surface
and a larger step or striation is produced during each stress cycle.

In real engineering materials slip will not always be possible on slip planes which
exactly pass through the crack tip as they may be blocked by means of dislocations
producing a back stress, second phase particles, and so on. Lui [23] has considered
this situation for FCC metals, and the mechanisms involved and resulting fracture sur-
16 I. LE MAY

Stage II

Fig. 19. The three stages of fatigue crack propagation.

Fig. 20. Fatigue striations on quenched and tempered 4140 steel.

face morphology are similar to those proposed independently by Neumann [24]. Lui's
model for crack growth is shown in Fig. 23, and it is capable of explaining the wide
variety of striation shapes observed in practice, depending on the number of active for-
ward and reverse slip planes. The model can also be considered to be applicable to Bee
metals although the geometry of slip will be different. The formation of slip band
cracks, which occur particularly in more ductile materials, and which are shown in Fig.
24, can be explained in terms of restraint imposed on reverse slip by large carbides or
other second phase particles in conjunction with the large extent of forward slip taking
place at high stress levels (see Fig. 25).
FAILURE MECHANISMS AND METALLOGRAPHY 17

(a)

(e)

(c)

Fig. 21. The plastic blunting model of Laird and Smith [19] for the formation of fatigue
striations. One complete stress cycle is illustrated.

Fatigue crack propagation in vacuum is characterized by a much decreased (by 5

to 10 times) crack growth rate and by an absence of striations as shown in Fig. 26 [21].
This may be explained by the fact that reverse slip can occur on the same slip planes as
were involved in the crack advance, and rewelding can also take place, there being no
surface oxide film. If rewelding were complete, no crack propagation would take place
from cycle to cycle, and it has been suggested that rewelding is complete only during
the first closure, becoming progressively poorer during subsequent compressive parts of
the cycle [25]. Thus, a crack would first open as in air, take a series of cycles (5-10)
before it ceased to be rewelded, and then move forward by another step. This process
would give rise to periodic, but irregular, slip steps as have been observed in practice in
vacuum, these having a spacing approximating that of striations formed in air [25]. Ob-
jections to the mechanisms may be raised on the basis that the crack should move for-
ward during each cycle by the width of the area not rewelding, but this ignores the
strain hardening occurring in the plastic zone ahead of the crack, and it seems more
probable that the crack will progress very little after each partial rewelding on closure,
until it jumps forward on breaking through the cyclicly hardened zone at its tip.

Striations are not always discernible on Stage II fatigue fracture surfaces formed
in air, particularly in high strength ferrous alloys. The reasons may be that they are too
fine (at very low crack propagation rates) or too poorly defined in terms of height
18 I. LE MAY

..
Increasing tension: crack opening

Initial
crack II
Slip band

Compression: crack closing

Compression: next cycle

First cycle

Second cycle

Fig. 22. The coarse slip model for fatigue crack growth.
FAILURE MECHANISMS AND METALLOGRAPHY 19

( 110) (111 )
(111 )

(100) SLIP GEOMETRY

FOR FCC METALS

Slip bands
JJ
Fig. 23. The model of Lui [23] for fatigue crack growth on multiple slip planes.

Fig. 24. Slip band cracks on quenched and tempered 4140 steel.
20 I. LE MAY

Slip band cracks

Fig. 25. Schematic of striation profile with slip band cracks formed by restraint of re-
verse sl ip from the model of Lui [231.

AIR
"..

Fig. 26. Fatigue fracture surf-ace of 2024-T3 aluminum tested successively in air, 216
cycles in vacuum of 5 X 10-6 torr, in air. From Ref. [211 by courtesy of R.M.N.
Pelloux and the American Society for Metals. Copyright (1969) by American Society
for Metals.

because of the very limited plastic flow occurring in high strength steels, or else the
loading mode may not be conducive to their formation (e.g., tension-tension plane
stress conditions (261). They have, however, been observed even on fatigue crack sur-
faces which propagated along the prior austenite grain boundaries in as-quenched high
strength steel at the interface with grain boundary carbides [27,281, as shown in Fig.
27.
FAILURE MECHANISMS AND METALLOGRAPHY 21

Fig. 27. Fatigue fracture along prior austenite boundaries in as-quenched 4140 steel,
with striations apparent in places on the boundaries.

The presence of striations on the fracture surface is not unique to fatigue crack
propagation, as will be emphasized later when discussing stress corrosion cracking. How-
ever, one feature which is sometimes observed on fracture surfaces and is unique to fa-
tigue is tire track formation. Figure 28 shows some typical examples, and these features
are formed when small second phase particles are either detached and embedded in a
film of metal dust which moves forward as the crack propagates, or else by projecting
particles from one surface which indent the other during crack closure, as the surfaces
become offset during stress cycling after the crack tip has moved ahead by some dis-
tance [29). Such features are very valuable in determining fatigue fracture to have
been operative, as they are frequently observed when the crack propagation rate was
high and few (or no) striations may be visible as in low-cycle fatigue.

Fig. 28. Tire tracks on 4140 steel quenched and tempered at 700°C.
22 I. LE MAY

Stage III of the crack propagation process (see Fig. 19) is a stage characterized by
rapid crack growth, microvoid coalescence, and dimple formation, and is a precursor to
final rupture. Before this occurs, stretching may take place at the crack tip upon loading
[31], leading to the formation of a stretched zone as shown in Fig. 29. This is generally
similar in appearance to the stretched zone which may be seen on the surface of a frac-
ture mechanics specimen between the prior fatigue crack and the fast fracture region [32].

Fig. 29. Stretched zone just prior to final shear fracture in fatigued high strength steel.

Intergranular Fracture

As temperature is raised and rate of straining decreased, polycrystalline metals

which normally fracture in a transcrystalline ductile manner undergo fracture of an
intergranular nature. Thus, the majority of creep service failures are intergranular, and
a typical fracture is shown in Fig. 30.

The development of intergranular creep fracture depends on the nucleation,

growth and subsequent linking of voids on the grain boundaries, and Gifkins [33] pro-
vided a useful distinction between the two types of cavities seen on graJn boundaries:
the first is usually associated with cracking at triple-points and is designated as a "w"
(for wedge) type cavity; the second type is the isolated rounded cavity, termed an "r"
cavity. These are illustrated in Figs. 31 and 32 respectively.

Wedge cracks form at triple points due to grain boundary sliding, and may be
promoted by decohesion at interfaces between grain boundary precipitates and the ma-
trix as shown in Fig. 33. Fracture produces a rough surface, and grain boundary pre-
cipitates may be identified on this in appropriate circumstances as shown in Fig. 34.

The nucleation and growth of r-type voids at high temperature is strongly depend-
ent upon the stress state. The effective stress may lead to their nucleation, while sub-
sequent growth is dependent on the maximum tensile stress. The latter promotes the
stress-induced flow of vacancies to voids [34], and these grow preferentially on boun-
daries having high tensile stress acting on them and, indeed, their formation can be
suppressed by superposition of hydrostatic compressive stress [35]. At lower levels of
FAILURE MECHANISMS AND METALLOGRAPHY 23

/ ,
- J '_

Fig. 30. Intergranular creep fracture in a nickel-base superalloy gas turbine blade.

Fig. 31. Wedge-type cavities in 316 stainless steel which failed by creep.

stress or somewhat lower temperature, fracture may take place by the shear of material
between voids, giving rise to a mixed intergranular-transgranular fracture surface [361.

Intergranular fracture may also take place under monotonic loading at ambient
temperatures in embrittled materials. Embrittlement may arise from segregation of spe-
cific elements (from Groups IV to VI in the Periodic Table, in the case of steels) to grain
boundaries, or from precipitation of brittle phases such as intermetallics or carbides on
grain boundaries. In such cases the fracture has a faceted appearance with the crack
following the grain boundaries, and detailed analysis of the surface and immediately
underlying material may be necessary to prove conclusively the nature of the embrittling
24 I. LE MAY

.--:
.....,..
"
:,t'
,
.. ~
.J
80.,.m
.~

Fig. 32. 316L weld metal which failed by creep at high temperature, showing r-type voids .

J.'.
/,J'f . .J' -1 / if
o
...J , ' , . . . • ," ~ ~.I, '-----I' • v / . ,

.J ~ -~

rr.
,- ,j) .. 0

:. ~_r,/'
' _.1 , \ ()
·4 ' .t,./~ I J ~

.'J
l) 0 t • .... ' • I
'It' ___- - ~'). ( ~
l , I" , If t ~
~ ~ J 1
\:..1 t.~'~.
11.,. ... . , '" {..
J " • :' . ' .• ~., ,,\' } •
,
, •
..J -< ~
',
o <'f •
~
.-.0 . , ' •~ ",;.. '.

" <'"
.0):
II •
\. ... , ,..... .,. . • ¥ ,

.' ,.' cal ' ~_" F e.!:')

I , ...- , )(.I .~ ,~,
v ) J
f l .
~ ,.,G,'
.1(.,:,
"""'.
4

_. 0" • I\. _ ~'r""'. " "J

'-..r - J '
'. ',,':.. ~ .
,p. ... •

ooc::,,~J. ' t
~
.~ '1: C
~, -~\ ~
~ij f O<l / , " ~~
....> ",'
~ ..~ " ,.
0 ' .... •• ( )4
• ~f) .. -
,0
' ,., .
I'
:,.~.
,....;l'I) '-~ (. " '15um' ,..c.
~ - Cf) '
,T,--
.....;,,-. ~
f)
'~r . ·I.!'"
v . ..
.-l_of;
,r-. /.

Fig, 33. Formation of w-type cavities by decohesion at precipitate·matrix interfaces in

316 stainless steel.

effect, Some examples of grain boundary precipitates are included in an earlier review
paper prepared by the author [37] ,

Environmentally-Assisted Fracture

This, as indicated earlier, is a large and complex subject, and we may identify the
following cases where the influence of environment coupled with stress may lead to fail·
ure: stress corrosion cracking (SCC); corrosion fatigue; hydrogen embrittlement; liquid
metal embrittlement; and radiation damage, In the space available it is not possible to
FAILURE MECHANISMS AND METALLOGRAPHY 25

Fig. 34. Microfractograph of 316 stainless steel which failed in creep, showing carbides
and a-phase particles on fracture surface.

look at these in any great detail, so reference will be made, where appropriate, to recent
reviews where more detailed discussion may be obtained.

The mechanism of stress corrosion cracking is illustrated in Fig. 35. Essentially,

corrosion is localized at the tip of a propagating crack which may be either intergranular
or transgranular, depending on the material and the corrosive environment. The surface
of the crack becomes passivated or covered by a protective film, and this is ruptured
when slip takes place at the crack tip, leaving a fresh surface. This in turn is attacked
and subsequently either passivated or otherwise protected, the crack tip having moved
forward. The process continues on a stepwise basis. The fracture morphology produced
can vary widely, and ma\:, include striations as well as linear features running parallel to
the direction of crack propagation, an example being shown in Fig. 36, which is taken
from the fracture surface of an AISI 316L orthopaedic implant [381. Stress corrosion
fractures are characterized by little ductility, and in many cases an oxide coating may
be seen on the crack surfaces. A typical photomicrograph of a crack tip is shown in
Fig. 37.

Corrosion fatigue and SCC are closely allied, and the latter may be thought of as
a special case in which the stress level is maintained constant. The fatigue limit of ma-
terials which have a poor resistance to corrosion may be very low indeed in an aqueous
environment, perhaps less than one tenth of the ultimate tensile strength. Again, a wide
range of fracture morphologies may result, and it should be noted that corrosion fatigue
is not confined to aqueous or liquid environmental conditions. It may be the operative
mechanism of failure in humid air, which can produce greatly reduced fatigue lives as
compared with those obtained in dry air [391. A recent review by Schijve [401 serves
to emphasize the importance of the problem and to pinpoint some of the gaps in our
knowledge. Similarly, two recent Conferences have attempted to review some of the
areas of activity at the present time [41,421.

Hydrogen embrittlement can be a severe problem in ferrous materials in particu-

lar, because of its very high mobility in the BCC iron lattice at ambient temperatures.
When present in sufficient quantity, hyd rogen lowers the cohesive strength of the lattice
[431, and it also tends to accumulate at defects, producing a high internal pressure
26 I. LE MAY

Fig. 35. Schematic illustration of stress corrosion crack propagating by rupture of the
protective film on plastic deformation at the crack tip, subsequent corrosion, repassiva·
tion, and build up of stress to repeat the cycle.

Fig. 36. Stress corrosion cracking striations on the fracture surface of 316L stainless
steel, with linear features parallel to crack growth direction (arrowed), indicative of
"tunnelling".
FAILURE MECHANISMS AND METALLOGRAPHY 27

: \
r-:- --
• I

Fig. 37. Stress corrosion crack in 316 stainless steel.

which may cause rupture, particularly in high strength alloys. Hydrogen pickup may oc,-
cur in several ways: it may take place from water in the scrap used in steelmaking, the
gas becoming trapped during solidification; it may arise from pickling or plating opera-
tions; it may be picked up from the service environment; it may be a product when
welding with electrodes having moist coating; or it may be produced at the tip of a
propagating stress corrosion or corrosion fatigue crack. In the last mentioned case, it
can be generated either from reaction of the corrosive medium with the freshly exposed
metal surface or from dissociation of a gas (such as water vapor) on this surface: the
subsequent absorption and embrittlement at the tip of the crack is considered to be a
significant and frequently occurring mechanism in both SCC and corrosion fatigue.

Entrapped hydrogen can give rise to a number of features which may be detected
metallographically. For example, on a macroscopic scale blisters may be produced in
ferrite, probably being nucleated at sub-surface inclusions [441. Flakes or small internal
fissures may sometimes be detected by ultrasonic examination .or, destructively, by opti-
cal metallography as shown in Fig. 38. These give rise to shiny facets on the fracture
surface. When the fracture surface of hydrogen embrittled material is examined using
electron microfractography, small microcracks may sometimes be seen.

Hydrogen embrittlement does occur in materials other than steel, such as titanium,
zirconium and the refractory metals. Titanium and zirconium form brittle hydrides
while the refractory metals are thought to become embrittled from supersaturation of
hydrogen in solution [451.

Liquid metal embrittlement has become of increasing interest in the light of the
developments around the world on liquid metal fast breeder reactor (LMFBR) technol-
ogy, as normally ductile metals may fail in a catastrophic brittle manner by cleavage or,
more commonly for polycrystalline materials, by rapid intergranular crack propagation.
The mechanism postulated is that a liquid metal atom reduces the bond strength of the
atoms of the solid material by chemisorption at the tip of a crack or at the end of a
dislocation pile-up near a surface obstacle. For more detailed discussion of the mechan-
isms, see the reviews by Westwood et al. [46] and by Kamdar [47] .
28 I. LE MAY

Fig. 38. Hydrogen cracking in HAZ of armour-plate welded with improperly dried elec-
trodes.

Radiation damage may lead to void formation, locking of dislocations through

interaction with radiation-induced defects, or excess precipitation of a second phase.
Generally, the effect is to cause the material to lose ductility and to become more prone
to brittle fracture. A comprehensive review of the relations between fracture me,chanisms
and irrad iation damage has been given by Bement et al. [48], to wh ich the interested
reader is referred.

Corrosion and Wear

These two failure mechanisms are apparently straightforward and can be recognized
with relatively little need for significant metallographic examination. However, it is often
very difficult to provide solutions to corrosion problems except in straightforward cases
where the wrong materials for the environment or dissimilar metals in contact have been
chosen. Equally, while we can recognize wear, it is often difficult to provide a suitable,
alternative choice of material to stand up to a particular abrasive service condition, and
our knowledge of the detailed mechanisms of wear as they relate to microstructure
leaves a considerable amount to be desired.

One particular situation which combines corrosion and wear is the insidious prob-
lem of fretting corrosion. In this regard, Waterhouse has prepared an excellent mono-
graph of this subject [49], which discusses the mechanisms of, and metallographic evi-
dence for, fretting corrosion, together with methods of combatting it.

CONCLUDING REMARKS

Hopefully, this paper has touched on the mechanisms of failure and the metallo-
graphic evidence for each, which are of relevance to practicing metallographers involved
in, or interested in becoming involved in the analysis of service failures. Once again,
however, it must be emphasized that all the evidence must be weighed and that metallo-
graphic examination is an ancillary tool which, when properly used, can provide definite
confirmation or otherwise of mechanisms which may be postulated on the basis of the
environmental conditions, overall appearance of the fracture and so on.
FAILURE MECHANISMS AND METALLOGRAPHY 29

ACKNOWLEDGEMENTS

The author is grateful to his former graduate students and associates, many of
whose names appear in the References, for stimulating discussions on the subject of fail-
ure mechanisms. In particular, it is appropriate to acknowledge the assistance provided
by Dr. W.E. White, now of the University of Calgary in this respect and also in the prep-
aration of this paper.

Acknowledgement is also made to the National Research Council of Canada, for

support of the author's studies in the areas of deformation and fracture.

REFERENCES

1. "Failure Analysis and Prevention", Metals Handbook, 8th Ed., Vol. 10, ASM, Metals
Park, Ohio, (1975).
2. Colangelo, V.J., and Heiser, F.A., Analysis of Metallurgical Failures, John Wiley, New
York, (1974).
3. McCall, J.L., and Mueller, W.M. (eds.), Microstructural Analysis: Tools and Tech -
niques, Plenum Press, New York - London, (1973).
4. Lui, M.-W., Le May, I., and Coote, R.I., "Nondestructive Metallurgical Examination
of a Damaged 100 MW Generator Rotor", Praktische Metallographie, 12, pp. 319-
326, (1975).
5. "Fractography and Atlas of Fractographs", Metals Handbook, 8th Ed., Vol. 9 ASM,
Metals Park, Ohio (1974).
6. Warke, W.R., Nielsen, N.A., Hertzberg, R.W., Hunter, M.S., and Hill, M., "Techniques
for Electron Microscopic Fractography", in Electron Fractography, STP 436, ASTM,
Philadelphia, pp. 212-230 (1968).
7. Brammer, I.S., and Dewey, M.A.P., Specimen Preparation for Electron Metallography,
Blackwell Scientific Publications, Oxford (1966).
8. Biggs, W.D., The Brittle Fracture of Steel, Macdonald and Evans, London, (1960).
9. Friedel, J., Dislocations, Pergamon Press, Oxford, p. 321 (1964).
10. Berry, J.M., "Cleavage Step Formation in Brittle Fracture Propagation", Trans. ASM,
51, pp. 556-581 (1959). See also discussion of this paper by Crussard, C., Plateau,
J., and Lean, J.B., on pp. 581-584.
11. Zener, C., "The Micro-Mechanism of Fracture" in Fracturing of Metals, ASM, Clevel-
land, pp. 3-31 (1948).
12. Cottrell, A.H., "Theory of Brittle Fracture in Steel and Similar Metals," Trans AIME,
212, pp. 192-203 (1958).
13. Hull, D., "Nucleation of Cracks by the Intersection of Twins in a-Iron", Phil. Mag.,
Eighth Series, 3, pp. 1468-1469 (1958).
14. Smith, E., "The Nucleation and Growth of Cleavage Microcracks in Mild Steel", in
Physical Basis of Yield and Fracture, Inst. of Physics and the Physical Society,
London, pp. 36-46 (1966).
15. Gurland, J., and Plateau, J., "The Mechanism of Ductile Rupture of Metals Contain-
ing Inclusions", Trans. ASM, 56, pp. 442-454 (1963).
16. Lui, M.-W., and Le May, I., "The Relation Between Ultimate Tensile Strength and
Carbide Size in Steel", J. of Eng. MatIs. and Technology, Trans. ASME, 98, pp.
173-179 (1976).
17. McClintock, F.A., "Local Criteria for Ductile Fracture", Int. J. Fracture Mech., 4,
pp. 101-129 (1968).
18. Knott, J.F., "Micro-Mechanisms of Fracture and the Fracture Toughness of Engin-
eering Alloys", in Fracture 1977, D.M.R. Taplin (ed.), Vol. 1, Univ. of Waterloo
Press, Waterloo, Ont., pp. 61-92 (1977).
19. Laird, C., and Smith, G.C., "Crack Propagation in High Stress Fatigue", Phil. Mag.,
Eighth Series, 7, pp. 847-857 (1962).
30 I. LE MAY

20. Neumann, P., "Bildung und Ausbreitung von Rissen bie Wechselverformung", Z.
Metallkunde, 58, pp. 780-789 (1967).
21. Pelloux, R.M.N., "Mechanisms of Formation of Ductile Fatigue Striations", Trans.
ASM, 62, pp. 281-285 (1969).
22. Broek, D., and Bowles, C.O., "The Study of Fracture Surface Profiles in the Electron
Microscope", Int. J. Fracture Mech., 6, pp. 321-322 (1970).
23. Lui, M.-W., "Relations Between Microstructure, Mechanical Properties and Fracture
Morphology of AISI 4140 Steel," Ph.D. Thesis, University of Saskatchewan, Saska-
toon, (1973).
24. Neumann, P., "The Geometry of Slip Processes at a Propagating Fatigue Crack - II",
Acta Met., 22, pp. 1167-1178 (1974).
25. Broek, D., "Some Contributions of Electron Fractography to the Theory of Frac-
ture", Review No. 185, International Metallurgical Reviews, 19, pp. 135-182 (1974).
26. Plumbridge, W.J., and Ryder, D.A., "The Influence of Specimen Geometry on the
Mode of Fatigue Crack Growth in Aluminum", Acta Met., 17, pp. 1449-1452
(1969).
27. Le May, I., and Lui, M.-W., "Fractographic Observations of Fatigue Fracture in
High-Strength Steels", Metallography, 8, pp. 249-252 (1975).
28. Lui, M.-W., and Le May, I., "Effects of Grain Boundary Carbides on the Fatigue
Fracture of an AISI 4140 Steel", in Grain Boundaries in Engineering Materials,
J.L. Walter, J.H. Westbrook and D.A. Woodford (eds.), Claitor's, Baton Rouge,
LA, pp. 397-407 (1975).
29. Beachem, C.D., "Microscopic Fatigue Fracture Surface Features in 2024-T3 Alu-
minum and the Influence of Crack Propagation Angle upon their Formation",
Trans. ASM, 60, pp. 324-343 (1967).
30. Lui, M.-W., and Le May, I., "A Fractographic Study of Fatigue in AISI4140
Steel", IMS Proceedings 1971, pp. 227-241 (1972).
31. Lui, M.-W., Shaw, W.J.D., Whelan, J.M., and Le May, I., "A Comparison of Transi-
tion Zones on the Fracture Surfaces of High Strength Steels", in Microstructural
Science, Vol. 3A, P.M. French, R.J. Gray and J.L. McCall (eds.). Elsevier, New
York, pp. 3-20 (1975).
32. Spitzig, W.A., Pellissier, G.E., Beachem, C.D., Brothers, A.J., Hill, M., and Warke,
W.R., "A Fractographic Analysis of the Relationships Between Fracture Tough-
ness and Surface Topography in Ultrahigh-Strength Steels", in Electron Fractog-
raphy, STP 436, ASTM, Philadelphia, pp. 17-30 (1968).
33. Gifkins, R.C., "Mechanisms of Intergranular Fracture at Elevated Temperatures",
in Fracture, B.L. Averbach, D.K. Felbeck, G.T. Hahn, and D.A. Thomas (eds.).
MIT Press, Cambridge, Mass., pp. 673-623 (1959).
34. Hull, D., and Rimmer, D.E., "The Growth of Grain-Boundary Voids Under Stress",
Phil. Mag, Eighth Series, 4, pp. 673-687 (1959).
35. Needham, N.G., Wheatley, J.E., and Greenwood, G.W., "The Creep Fracture of
Copper and Magnesium", Acta Met., 23, pp. 23-27 (1975).
36. White, W.E., "A Study of Creep in Stainless Steel Weldments", Ph.D. Thesis, Uni-
versity of Saskatchewan, Saskatoon, (1977).
37. Le May, I., "Microstructural Analysis and Fractography", in Microstructural Analy-
sis: Tools and Techniques, J. L. McCall and W.M. Mueller (eds.), Plenum Press,
New York - London, pp. 153-168 (1973).
38. White, W.E., Postlethwaite, J., and Le May, I., "On the Fracture of Orthopaedic
Implants", in Microstructural Science, Vol. 4, E.W. Filer, J.M. Hoegfeldt and J.L.
McCall (eds.), Elsevier, New York, pp. 145-158 (1976).
39. Shaw, W.J.D., and Le May, I., "Fatigue Crack Propagation in High Strength Steel
Under Controlled Humidity Conditions", in Fatigue Testing and Design, Vol. 2,
R.G. Bathgate (ed.), Soc. of Environmental Engineers, London, pp. 31.1-31.24
(1976).
FAILURE MECHANISMS AND METALLOGRAPHY 31

40. Schijve, J., "Fundamental and Practical Aspects of Crack Growth Under Corrosion
Fatigue Conditions", Proc. Inst. Mech. Engrs., London, 191, pp. 107-114 (1977).
41. Proceedings of the Conference on the Influence of Environment on Fatigue, Inst.
Mech. Engrs., London, (1977) (in press).
42. ASTM Symposium on Corrosion Fatigue, Denver, (November 1976) (to be published).
43. McMahon, C.J., Briant, C. L., and Banerji, S.K., "The Effects of Hydrogen and Impuri-
ties on Brittle Fracture in Steel", in Fracture 1977, Vol. 1, D.M.R. Taplin (ed.).
University of Waterloo Press, Waterloo, Ont., pp. 363-385 (1977).
44. Evans, U.R., The Corrosion and Oxidation of Metals, Ed. Arnold, London, (1960).
45. Lawley, A., Liebmann, W., and Maddin, R., "Effect of Hydrogen on the Yielding
Behavior of Molybdenum", Acta Met., 9, pp. 841-850 (1961).
46. Westwood, A. R.C., Preece, C.M., and Kamdar, J.H., "Adsorption-Induced Brittle
Fracture in Liquid-Metal Environments", in Fracture: An Advanced Treatise,
Vol. III, H. Liebowitz (ed.). Academic Press, New York, pp. 589-644 (1971).
47. Kamdar, M.H., "Embrittlement by Liquid Metals", Prog. Mat. Sci., 15, pp. 289-374
(1973).
48. Bement, A.L., Hoagland, R.G., and Smidt, F.A., "Fracture Mechanisms and Radia-
tion Effects", in Fracture: An Advanced Treatise, Vol. III, H. Liebowitz (ed.).
Academic Press, New York, pp. 535-587 (1971).
49. Waterhouse, R.B., Fretting Corrosion, Pergamon Press, Oxford (1972).