UNIT 1

MICROMECHANISMS OF FRACTURE

Objectives:

To discuss and classify fracture types from a Material Engineering point of view. To study
characteristic features associated with different types of fracture. To explore factors that
can influence fracture behaviour such as strain rate, temperature and environment. To
develop fracture mechanism maps.

Introduction:

This unit considers the types of failure mode that a cylindrical test specimen of an
engineering material can experience when subjected to tensile loads under a range of rate,
temperature and environmental conditions. It demonstrates that characteristic fracture
features are associated with the different modes. Experimental measurements of this type
of specimen are then used to construct fracture maps which illustrate the test regimes
under which each mode is most likely to operate.

Ideal Strength:

The maximum stress required to fail a perfect crystal can be determined by calculating the
tensile force necessary to pull atoms apart. With reasonable assumptions it can be shown
that the upper limiting strength is approximated by the expression

E
C 


with E the elastic modulus. Such high strengths can be approached in fibres and whiskers:

Material Measured Stress Modulus Theoretical stress

F (GPa) E (GPa) E/ (GPa)

Silica Fibres 24.1 97.1 30.9

Iron Whiskers 13.1 295.2 93.9
Silicon Whisker 6.47 165.7 52.7
Alumina Whisker 15.2 496.2 157.9
Steel Wire 2.75 200.1 63.7

4
In general, however, the measured fracture stress is found to be significantly lower than the
stress required to separate two atomic planes. The magnitude of this discrepancy, though, is
dependent on test piece section size. For extremely small diameters, fibre strengths can
approach the theoretical value.

Materials have low fracture strengths compared with theoretical values because in many
cases they deform plastically at lower stresses. The plastic deformation results in the
accumulation of irreversible damage which eventually precipitates failure. In addition, real
materials contain defects that can promote early failure. A purpose of the present unit is to
discuss these alternative ways in which materials can fail.

Classification of Fracture Mechanisms:

Fracture behaviour is temperature dependent. However, it is possible to place the

observations into a low temperature regime where plastic flow is not strongly dependent on
temperature or time and a high temperature regime in which materials creep. The dividing
line is arbitrarily taken at 0.3TM with TM the absolute melting temperature of the material.

At low temperature, a crystalline solid pulled in tension may fail by cleavage or brittle
intergranular fracture. Alternatively, it may fail in a ductile manner with extensive
deformation and the formation of a neck.

5
At high temperatures, creep fractures can occur in both a transgranular and intergranular
manner. In both regimes, the failures can vary from brittle to fully ductile. A schematic
diagram illustrating the range of failures is shown below –

It is possible to see the full range of fracture in one material.

Other failure modes can occur. A particularly important one is fatigue. Fatigue occurs when
the applied stress (or strain) varies in a cyclic fashion. It involves the progressive growth of
cracks until failure is precipitated. Another important failure mode is stress corrosion. In
this case, cracks can grow progressively in metals and alloys when they are exposed to a
constant stress and particular environments (e.g. salt water).

Characteristic features of each fracture type are discussed next.

6
Ductile Fracture:

The most common type of ductile fracture is observed when a test piece is overloaded in
tension. Usually it is associated with the formation of a neck and has a characteristic cup
and cone appearance. It is invariably associated with the occurrence of plastic deformation.

In very pure materials that are free of inclusions and second phase particles it is possible for
necking to continue until an arrow point is reached representing 100% reduction in area. In
some single crystals, necking proceeds until failure occurs by shear along a single slip plane.

Engineering materials always contain second phase particles. These can vary in size from
large inclusions to fine precipitate particles specifically developed by heat treatment to give
the alloy the required mechanical properties. Under deformation, particularly the intense
strain conditions in the neck, these particles give rise to microvoid formation. The voids
form either by fracture of the inclusion itself or through decohesion of the particle/matrix
interface.

Slip then allows the material between the microvoids to neck down giving rise to the
characteristic dimpled fracture surface.

7
Microvoid formation occurs under other states of stress as well. The morphology of the
fracture surface dimples varies with the stress system. Under tension the dimples are
equiaxed. Pure shear generates microvoids elongated in the shearing directions so that the
voids point in opposite directions on the two halves of the fracture surface. Tearing due to a
non uniform stress field produces elongated dimples that point back to the crack origin on
both fracture surfaces.

Another type of ductile fracture, which is planar in character, has been found in high carbon
steels at room temperature and some pure metals at very low temperatures. Care must be
exercised in interpreting these because at first sight they appear to be brittle. The difference
can be illustrated by reference to the eutectoid structure in steel which consists of
alternating regions of brittle cementite (Fe3C) and ductile ferrite.

8
Brittle Fracture (T  0.3TM):

Brittle fracture is a catastrophic event in which a material fails after little or no prior plastic
deformation. It is often the cause of dramatic service failures. This contrasts with ductile
failures which are rare under service conditions and usually only occur where there is a
gross design fault or an unexpected overload.

Brittle fractures can be inter- or transgranular. Intergranular failures are readily recognised
from the fracture surface since the cracks follow the grain boundaries which then stand out
in three dimensional relief. In the absence of an aggressive environment the decohesion
between adjacent grains is due to the presence of impurity elements at the boundary.
Examples of this include segregation of sulphur to grain boundaries in low carbon iron, the
formation of aluminium nitride at boundaries in Al-killed steels and the formation of
chromium carbide (Cr23 C6) in stainless steels under particular heat treatments. In many
cases the problem can be minimised or averted by changes in alloy composition or heat
treatment. In some cases, however, it is due to precipitation and segregation of particular
elements during extended service operation. The presence of aggressive environments,
such as hydrogen gas, or of corrosive species can produce intergranular failures in situations
where they might not otherwise be expected. Some of these will be discussed later.

In general, the most common mode of brittle fracture involves transgranular crack growth.
The cracks propagate along specific crystallographic planes. These are called cleavage planes
and the failure process is usually called cleavage. The crack chooses these planes because
they present low energy paths to its movement. The mechanism is observed in body
centred cubic materials (e.g. iron), hexagonal close packed materials (e.g. titanium, zinc),
ionic crystals (sodium chloride) and covalently bonded substances (e.g. silicon nitride). The
face centred cubic metals (e.g. copper, aluminium) are only susceptible to cleavage events
under severe environmental conditions.

Since the cleavage crack follows a specific crystallographic plane within a grain it tends to be
relatively flat and to have a high reflectivity. Usually, neighbouring grains have a slightly
different orientation so that the crack changes direction at the grain boundary in order to
continue propagating on the preferred cleavage plane.

Cleavage crack
traversing several
grains

9
Close examination of the cleavage ‘facets’ reveals a number of characteristic features. In
particular, a river pattern with associated tributaries is often observed.

Direction of crack

propagation

The river pattern is caused by steps on the fracture surface where the crack has jumped
across parallel cleavage planes. River patterns frequently form where the propagating crack
traverses a grain boundary. The direction of crack propagation can be determined from the
river pattern as indicated.

It is possible to identify three discrete types of cleavage failure:

The first, designated ‘Cleavage 1’, occurs without general plasticity. It is attributed to the
existence of small, inherent flaws (defects) in the material. If the flaws are sufficiently large,
a cleavage crack can propagate at an applied stress that is less than general yield. The
strength of the material is determined by the largest flaw it contains.

‘Cleavage 2’ is activated if pre-existing flaws are very small or absent. It is now necessary to
initiate the starter defect. This requires the applied stress to equal or exceed the stress for
slip or twinning since these deformation processes can nucleate cracks. Cleavage 2,
therefore, is preceded by micro-plasticity. Even so the fracture is brittle since ductility is less
than 1%.

In contrast, ‘Cleavage 3’ is preceded by substantial plastic strain (1 - 10%). Failure is now

encouraged by work hardening which acts to limit deformation and by the larger nucleated
cracks formed by the extensive slip deformation. This mode is more prevalent at higher
temperatures.

10
Ductile - Brittle Transition:

Toughness is the ability of a material to absorb energy before and during fracture. If the
energy absorption is high, the material is said to be tough or to have a high fracture
toughness. Such materials tend to behave in a ductile manner. If the absorption is low, the
material is brittle. A measure of the toughness of a material can be obtained from the
energy absorbed in an Izod or Charpy impact test. The interesting observation from such
measurements is the fact that many materials, and body centred cubic ones in particular,
show a marked transition in energy absorption with temperature. In essence, the materials
are brittle at low temperature but ductile at high temperature.

In addition to the energy absorption, the fracture surface shows a transformation with the
amount of facetting falling dramatically as the transition temperature is traversed. The
ductile/brittle transition temperature is sensitive to other factors such as purity,
composition, heat treatment, grain size and strain rate. The effect of grain size and strain
rate at a fixed temperature is shown below.

11
 UNIT 2
CRACKS IN STRUCTURES
- ENERGY BALANCE

Objectives:

To assess why a crack or defect in a structure can become unstable and cause catastrophic
failure. To develop a failure criterion based on energy balance. To explore the effect of
plasticity on fracture. To define the strain energy release rate criterion - the toughness, G.
To examine theoretical and experimental methods of determining G.

Introduction:

Many engineering structures contain pre-existing flaws or defects that have escaped
detection by the non destructive testing techniques used. Alternatively, they may be
stressed highly enough for cracks to develop during normal service operation. Even if these
cracks are not initially dangerous they can grow by a number of processes, such as fatigue,
to a size at which they can cause catastrophic failure of the structure. History is littered with
disasters resulting from such failures. Even today with our greater awareness, catastrophes
can still occur.

It is important for the engineer to accept that his structure/materials does contain defects,
to quantify the effect that the defects might have and to apply safety standards that
minimise the risk of service failure. This might well mean that the fast fracture design
criterion takes precedence over other ‘failure’ criteria such as those for yielding or plastic
collapse with a resultant reduction in design stresses and/or design lives. The procedures
that allow such decisions to be taken rationally come under the general heading of fracture
mechanics.

Fracture mechanics can provide answers to the following questions:

(i) the effect of crack size on residual strength.

(ii) the maximum crack size that the structure can tolerate.

(iii) the length of time it takes for a crack to grow from some initial size to failure.

(iv) the initial size of flaw that can be allowed to enter service (i.e. NDT inspection
threshold).

(v) frequency of inspection during service.

12
Fracture mechanics encompasses the whole range of engineering from the microscopically
small to the major industrial plant:

Griffith Criterion:

In 1921, A.A. Griffith during work on the fracture behaviour of glass was concerned with
explaining its highly brittle behaviour. He reasoned that if a crack was introduced into a
stressed plate of elastic material, for equilibrium to be maintained, a balance must be struck
between the decrease in potential energy within the material/structure and the increase in
surface energy due to the new crack faces. The released potential energy comes from stored
elastic strain energy within the material and the work done by the resultant movement of
the applied external loads.

If the material contains an existing crack, then it could grow if the system supplies the
required additional surface energy. Where the system supplies more energy than is required
by the new surfaces, the crack will propagate almost instantly in a catastrophic manner.

The surface energy arises from the non equilibrium configuration of atoms at any surface in
a solid. Griffith studied the behaviour of a through crack in an infinitely large plate.

13
The specific surface energy was defined to be  per unit area. For the crack in the geometry
above, the total cracked surface area is (2a x 2t). The total surface energy, therefore, is (4 at
s).

Griffith then used the stress analysis of Inglis to calculate the decrease in potential energy
due to the introduction of a crack into an infinitely large plate. He found this to be ( a2 t)
2/E The change in potential energy due to the introduction of a crack is, therefore:-
  2 a 2t
U – U0 =   4at s
E

with U = Potential energy with crack

U0 = Potential energy without crack

 = applied stress

a = one half crack length

t = thickness

E = elastic modulus

 = specific surface energy

The equilibrium condition is now obtained by differentiation and equating the result to zero.

U 2  2 at
 4t s  0
a E

(U0/a = 0 of course). On rearrangement

  2t
2 s 
E

The second derivative

 2u   2t

a 2 E

The negative value means that the equilibrium condition is unstable so that the crack will
always grow.

The above energy balance can be shown schematically:

14
Griffith rearranged the energy balance for the equilibrium condition to give

2 E s

a
for plane stress conditions. For plane strain the equation is modified to

2 E s

a 1   

with Poissons’s ratio  = 0.25 to 0.33 in many materials.

The Griffith equations should be regarded as necessary but not sufficient conditions for
fracture. The geometry of the crack tip is not taken into consideration and it too has an
effect on fracture behaviour.

Orowan Modifications:

The Griffith approach is applicable to materials that are fully brittle. He was able to
demonstrate the effectiveness of the theory by means of tests on precracked glass samples.
However, in practice many materials are not ideally brittle exhibiting some plastic
deformation during fracture. This plastic deformation is associated with the tip of the crack
and is responsible for increased toughness in these materials. In terms of the Griffith
equations it is found that the fracture energy is several orders of magnitude greater in
metals and polymers than s suggests.

15
In recognition of this fact, Orowan reasoned that the true surface energy term in Griffith’s
equations should be replaced by an effective surface energy term. The effective surface
energy is the sum of the true surface energy and the work of plastic deformation (p), i.e.

e = s +p

with e = Effective surface energy

s = True (specific)surface energy
p = Work of plastic deformation

Experimental measurements show that p, >> s

However, their combined effects change the Griffith equation to:

2 E s   p 

a

This is the Orowan modification of the Griffith equation. It is p in metals and some polymers
that enhances toughness. The equation has limited value from an engineering point of view
because it is difficult to obtain reliable values of s and p. However, it does illustrate the
important factors in fracture and shows how improvements in toughness might be
achieved.

Irwin approach:

Irwin approached the influence of plasticity from a different viewpoint. Instead of

considering the energy absorbed by the new crack surface and by plastic deformation he
used the energy released by the system, i.e. the energy source. Specifically he adopted the
elastic energy released per unit crack length increment, u/a. This was given the notation
G and he was able to show that

EG

a
Comparing the Irwin and modified Griffith approach it is apparent that

G = 2(s + p)

At the point of instability when the crack propagates catastrophically to failure, the elastic
energy release rate achieves a critical value Gc. This critical value is a material property and
can be measured in the laboratory by means of sharply notched specimens. The Irwin
approach, therefore, is of much greater value from an engineering point of view.

16
Energy release rate:

The Griffith and related energy criteria for fracture consider that crack growth will occur if
the energy required to form an additional crack length, da, can just be supplied by the
system. The system in this case encompasses the release of strain energy and the work done
by the boundary (loading) forces. The relative importance of each depends on the manner in
which the external loads are applied. If the loading ends are fixed, the external load cannot
do work so the energy required for growth must come from the release of elastic energy.
However if the ends of the plate can move during crack extension, work is done by the
external load. Now the elastic energy of the plate increases in addition to the increase in
crack length.

First of all, take the general case of an elastically loaded body containing a crack ‘a’.

Load Displacement

P = Mu

(M = Body stiffness)

Compliance C = 1/M

u=CP

C depends on a

If the displacement of the applied load, P, is u, then the amount of stored strain energy is

1
U3  Pu
2

with U3 the stored strain energy and M the body stiffness for crack length a.

17
If the crack extends by da, the additional surface energy required is dependent on the
release of strain energy and the work done by the external body forces. Let the work done
by the external forces be represented by U1 = Pu. Consider now the energy changes
associated with the increase in fracture area, A, as the crack grows such that dA = Bda with
B the thickness of the body. On this basis we may define G as

dU dU 1 dU 3
G  
dA dA dA

dU 3 1 du 1 dP
where  P  u
dA 2 dA 2 dA

dU 1 du
and P
dA dA

1  du dP 
so that G  P u 
2  dA dA 

NB

a = Crack size
A = Crack area
u = Displacement
U = Energy (distinguish between U1 and U3)

The load-displacement relationship can be redefined as

u
C (1)
P

with C the Compliance of the body. Thus U = cP and :

du dC dP
P C (2)
dA dA dA

Substituting (1) and (2) in the equation for G and re-arranging we obtain

1 2 dC
G P
2 dA

which on re-expressing in terms of crack size, a, using the equation for area/crack size
above, gives

P 2 dC
G
2 B da

18
This relationship applies to bodies of uniform thickness. The value of G is obtained from P, B
and dc/da all of which are measurable.

The above derivation of G takes into consideration the work done by the end loads. It
relates to fixed load conditions in which the ends of the plate are free to move during crack
growth thus allowing the external load to do work. In consequence, the elastic energy
content of the plate increases giving the observed positive value of G. If the plate is
evaluated with fixed ends (rigidly gripped), the external load cannot do work. In this case
the energy required for crack growth comes entirely from the release of elastic energy so
that

P 2 dc
G
2 B da

The two cases are shown schematically below

Consider the fixed end situation. The graph shows load-displacement lines for a body with a
crack of size a and for the same body when the crack has grown to a + da. The crack
extension takes place at load F, but since the ends are fixed, the displacement remains
constant so that the load drops (P). The resultant release of elastic energy which is equated
to G is shown by the hashed area.

For the fixed load situation, the major difference is the additional work done by the load.
This is represented by the rectangle on the right hand side of area P,u. As a result, the
strain energy content of the plate increases as shown by triangles OBD and OEF. The
difference between the two is the triangle OBE which, in area, is, ½Pu (i.e. ½ BEFD). The
extra strain energy, therefore, is supplied by the load doing work. However, even after this
strain energy has been accounted for, there remains energy equal to ½Pu available for
crack growth. An analysis shows that the magnitude of this energy is the same as for the
fixed end situation but the sign is reversed because the source of supply is through the
external load doing work.

19
In summary, therefore, for the fixed grips case the energy required for crack growth is
provided by the elastic strain energy. Under constant load, it is supplied by the load. For
both conditions, the elastic energy release rate is the same (only the sign is reversed). It can
in fact be shown that the magnitude of G is independent of the type of load application (e.g.
fixed grip, constant load, combinations of load and displacement, and machine stiffness).

At instability, when the crack grows rapidly to failure, the critical strain energy release rate
is
P 2 max dC
Gc 
2 B da

So once the compliance - crack length relationship has been determined for given
specimens, G can be found and Gc obtained by establishing the load at fracture provided the
amount of plastic deformation is kept to a minimum.

Experimental determination of’ G:

A wide range of specimen types are used in the determination of G. One particularly
important one is the Double Cantilever Beam (DCB). This geometry is considered below in
order to illustrate the general principles of the experimental method.

The technique involves the construction of a load displacement diagram for a range of crack
sizes. The cracks can be introduced mechanically or by means of fatigue pre-cracking. The
slope of each line is then used to construct a graph showing the dependence of compliance
on crack size. From the slope of the resultant line, the derivative of compliance with respect
to crack length, dC/da can be established thereby allowing G to be determined.

The procedure is shown schematically below.

20
Toughness, Gc, Values

Materials Gc, KJm-2

Pure ductile metals 100 – 1000
High Strength Steels 15 – 118
Mild Steel 100
Titanium alloys 26 – 114
Aluminium alloys (high strength) 8 – 30
Polyethylene 6–7
Epoxy 0.1 – 0.3
Silicon carbide (S.C.) 0.05

Example (from Knott & Elliott):

In the compliance calibration of an edge cracked fracture toughness test piece of an

aluminium alloy, it was observed that a load of 100 KN produced a displacement between
the loading pins of 0.3000 mm when the crack length was 24.5 mm and 0.3025 mm when
the crack length was 25.5 mm. The fracture load of an identical test piece containing a crack
of 25.0 mm is 158 KN. Calculate the critical value of the energy release rate at fracture if all
test pieces were 25 mm thick.

21