C H AP T E R 2

Fracture Mechanics
Background

INTRODUCTION

At the most fundamental level, brittle fracture occurs when stresses

reach a magnitude needed to break the bonds between the atoms in the
material. However, if there were no means of concentrating stress,
loads necessary to cause failure would be extremely large. It is the
presence of small defects that concentrate applied stresses to a magni-
tude sufcient to cause them to grow. The materials we discuss in this
book are brittle because plastic ow mechanisms in them are insuf-
cient to relieve the stress concentration at the defect.
The science of fracture mechanics allows us to calculate the forces
needed to cause defects to grow based on knowledge of specimen
geometries and the applied loads. In this chapter, we provide some
historical perspective and a summary of the basic principles of fracture
mechanics. The reader is encouraged to consult Anderson (1995) and
Munz and Fett (1999) for more details.

EARLY BRITTLE FRACTURE RESEARCH

Although fracture studies of ceramics can be traced as far as back as

1867, our current understanding of brittle fracture can be traced to

The Fracture of Brittle Materials: Testing and Analysis, First Edition. Stephen W. Freiman
and John J. Mecholsky, Jr.
2012 The American Ceramic Society. Published 2012 by John Wiley & Sons, Inc.

6
 Early Brittle Fracture Research 7

2b

2a

Figure 2.1 Void in a plate in which stresses are concentrated at point A.

Inglis (1913), who initiated the concept of stress concentration at a void

in a material as shown in Figure 2.1.
Assuming an elliptically shaped cavity, the concentrated stress, c,
due to the presence of the void, is given by the following expression.
c = (1 + 2 a / ), (2.1)
where 2a is the long axis of the ellipse and is its radius of curvature.
As approaches zero, that is, the elliptical cavity begins to resemble
a crack whose tip radius is of atomic dimensions. The atomistic nature
of materials and the nonlinearly elastic or plastic deformation that
occurs in the vicinity of a crack tip avoids the problem of 0, but
this approach does not yet allow us to quantify the stress state at an
actual crack tip, and says nothing about when the void is likely to grow.
The second important advance in our understanding of brittle
failure was made by Grifth (1920) who postulated that brittle failure
in glass is a result of the growth of small cracks when the material is
subjected to a large enough tensile stress. He put forth the hypothesis
that these cracks are present in all glasses in a distribution of sizes,
8 CHAPTER 2 Fracture Mechanics Background

leading to the concept that the smaller the volume (or area) under stress,
the less likelihood of nding a large aw, and therefore the higher the
strength. Grifth demonstrated this concept experimentally by measur-
ing the strength of glass bers of varying diameter, and showing that
strength increased with decreasing ber diameter. The aw that eventu-
ally grows to failure is determined by its severity as well as its location
with respect to the highest tensile stress, thereby giving rise to the
statistical nature of brittle failure.
Grifth also hypothesized that a materials resistance to the growth
of a crack is determined by the energy required to create the two frac-
ture surfaces produced by its extension. This approach assumes that
fracture occurs in an equilibrium manner, that is, in the absence of any
kinetic effects, and that no energy is lost due to plastic ow or heat. It
also neglects the possible effects of the test environment in which the
aw is growing, for example, water. Grifths expression for glass
fracture based upon this approach is given by:

2 E f
1/ 2

f =
a
, (2.2)

where f is the fracture strength, E is Youngs modulus, f is the energy

required to form the crack surfaces (i.e., the fracture energy), and a is
the critical aw size. While Grifths calculations were not entirely
accurate because of his assumption that the fracture energy of the glass
could be extrapolated from measurements of surface energy carried out
at elevated temperatures, the form of Equation 2.2 accurately depicts
the relationship between strength, aw size, and fracture energy.
However, what was still required was a way of translating the external
loads on a part into knowledge that could be used to predict its resis-
tance to crack growth. This awaited the development of fracture
mechanics.

DEVELOPMENT OF FRACTURE MECHANICS

Credit for the development of the science of fracture mechanics is

rightfully give to George Irwin and his colleagues at the U.S. Naval
Research Laboratory (Irwin, 1958). Irwin rst introduced the concept
of a strain energy release rate, G, which has nothing to do with time
 Development of Fracture Mechanics 9

dependence, but is the change in strain energy with crack extension.

Equation 2.2 then becomes:
1/ 2
EG (2.3)
f =
a
G = 2 f . (2.4)
There are three ways that stress can be applied to a crack (Fig. 2.2).
The predominant situation with respect to the stressing of brittle
materials leads to crack growth under mode I loading, that is, pure
tension across the crack face.
A schematic of the stress state at the crack tip produced by an
externally applied mode I load is shown in Figure 2.3.

Mode I Mode II Mode III

(Opening) (In-Plane Shear) (Out-of-Plane Shear)

Figure 2.2 Modes of fracture (Anderson, 1995).

Ki
y

r1/2
a

Figure 2.3 Schematic of the stress eld at a crack tip.

10 CHAPTER 2 Fracture Mechanics Background

The stress intensity factor is dened through the following

expression:
KI
y = ( f ()). (2.5)
2 r
y is the applied, far-eld, stress, is the angle in the plane with respect
to the crack face, and r is the distance from the crack tip. Note that
Equation 2.5 is only valid outside of the singularity-dominated zone,
within which the material is nonlinearly elastic or in which some per-
manent deformation has taken place. The beauty of linear elastic frac-
ture mechanics is that it can be used to explain fracture in spite of the
existence of this singularity.
The relationship between KI and applied stress can take many forms
depending on the type of loading and the location of the crack relative
to the load. The simple example of a surface aw, the predominant
source of failure for brittle materials (Fig. 2.4), gives rise to the
relationship:

K I = Ya1 / 2, (2.6)

Fracture
Surface

2c

Figure 2.4 Surface aw.  is used in Equation 2.8.

 Critical Fracture Toughness 11

where Y is a numerical factor that depends on the loading method and

the crack geometry, and a is the depth of the crack into the interior.
The surface aw is typically elliptic rather than semicircular in
shape, and is usually elongated along the tensile surface. The expres-
sion for stress intensity factor for such a aw is given by:
a
1/ 2
K I = M 2 , (2.7)

where is an elliptical integral that accounts for the fact that the
stress intensity factor varies around the perimeter of the aw, and is
given by:
/2 1/ 2
b2 a 2 2

= 1
0
b
2 sin d.

(2.8)

The stress intensity is a maximum at the tip of the minor axis of the
aw and is a minimum at the tip of the major axis. M is a surface cor-
rection factor whose value is approximately 1.12 as determined by
nite element analyses (Raju and Newman, 1979).

CRITICAL FRACTURE TOUGHNESS

Crack extension takes place when the applied load rises to a level such
that the stress intensity factor produces crack-tip stresses large enough
to rupture atomic bonds at the crack tip. Fracture toughness is dened
as the value of the stress intensity factor, KI, at the point where KI = KIc,
the critical stress intensity factor.

K IC ( EGC ) ( 2 E f ) .
1/ 2 1/ 2
(2.9)

Equation 2.9 provides the link between fracture mechanics and Grifths
fracture energy.
How do we measure criticality? Grifth dened f in thermody-
namic terms, but for most materials, the point of reversible crack exten-
sion and contraction is not observed. Experimentally, we can increase
the stress on a crack to the point at which the crack grows to failure.
However, most brittle materials are sensitive to the effects of the atmo-
sphere in which they are tested. Subcritical crack growth, otherwise
known as stress corrosion, or environmentally enhanced crack growth,
12 CHAPTER 2 Fracture Mechanics Background

is very difcult to avoid (See Chapter 3). The key is the relative humid-
ity of water in the environment, that is, its partial pressure. No liquid,
no matter how small is the solubility of water in it, can be considered
to be completely inert. Even in relatively neutral external environments,
for example, dry gases and vacuum, some materials will still undergo
crack growth prior to catastrophic fracture due to other crack tip mecha-
nisms. KIC, for brittle materials has at times been dened as the point
at which rapid crack extension occurs, usually seen as a drop in the
load on a testing machine. But, then how rapid is rapid?
There are two options with respect to dening fracture toughness;
(1) One can assume that KIC is a fundamental material property, and
that each fracture mechanics technique represents an attempt to esti-
mate its value by a different method, or, (2) one could assume that
fracture toughness can only be dened in terms of the measurement
itself, and that there is nothing particularly fundamental about each
determination. For the purposes of this document, we will assume the
former, namely that there is a number that we can call KIC for each
material, but whose determination will depend on many factors, includ-
ing test procedure, environment, etc.
Nonetheless, we are clearly dealing with an operational denition
(as is also true for metals). For a given material, KIC will be a function
of environment, testing rate, and test geometry. For most of the tests
described in this document, KIC is calculated from the maximum load
reached during the test. The crack velocity at which this point occurs
will be a function of the compliance of both the specimen and the
testing machine, but is usually in the range of 102 to 100 m/s. Fortu-
nately, crack growth curves are typically very steep, so any uncertainty
in crack velocity translates into only a small uncertainty in fracture
toughness.
Another issue is the stress state at the crack tip, that is, plane strain
or plane stress. Whether there exists a state of plane stress or plane
strain at the crack tip depends on the constraints. Where there is sig-
nicant constraint, for example, thick specimens, and little or no plastic
deformation, a state of plane strain will exist. In all specimens and
materials, the outer surface of the specimen will be in a state of plane
stress, but the depth of the plane stress zone could be quite small. For
metals, the condition of plane strain can be dened in terms of the size
of the crack tip singularity, which is related to the ratio of KI to the
yield stress in the material. Given that yield stresses for brittle materials
 Mixed Mode Loading 13

are so large (basically immeasurable), zone sizes will be vanishingly

small. It can be assumed that for almost any brittle material and for any
specimen size, the crack will be in a state of plane strain.
The issue of plane strain or plane stress arises in particular when
one transforms measurements between a fracture energy and a stress
intensity approach. It should be noted that both approaches are equiva-
lent; neither is more fundamental. The relationship between fracture
toughness and strain energy release rate is given by:
In plane stress: K I = GE (2.10)

In plane strain: K I = GE (1 2 ), (2.11)

where is Poissons ratio. However, historically, the (1 2) term has
been ignored in the literature in discussing brittle fracture. For most
materials neglecting this term will introduce an 4% uncertainty into
the absolute value of either G or KI depending on the primary
measurement.
Given the above discussion, we suggest the following as a good
working denition of fracture toughness.
For a brittle material in which no subcritical crack growth occurs,
KIC is the stress intensity factor at which crack growth begins.

MIXED MODE LOADING

However, mixed mode fracture can also occur if the face of the aw
is oriented at an angle to the tensile stress. A schematic of such a situ-
ation is shown in Figure 2.5. Both KII and KIII components of the stress
eld can act on the aw, with KII being greatest at the tensile surface,
and KIII being largest at the maximum aw depth.
Two issues arise in mixed mode loading. The most important is the
choice of a model that accurately combines the modes; the other issue
is the possible existence of slow crack growth. Effects of mode III

A A

Figure 2.5 Schematic of mixed mode loading.

14 CHAPTER 2 Fracture Mechanics Background

loading generally are not a signicant factor in brittle fracture, and

usually can be ignored.
Petrovic and Mendiratta (1976, 1977) discussed their results on the
basis of three expressions for combined mode I and mode I1 stress-
intensity factors, the maximum normal stress criteria (Kordisch et al.,
1976), the coplanar, maximum strain-energy release rate criteria
(Erdogan and Sih, 1963), and the noncoplanar, maximum strain-energy
release rate criteria (Paris and Sih, 1965) (Eqs. 2.122.14,
respectively).

K c = cos( / 2)[ K I cos2 ( / 2) 3 / 2 K II sin ] (2.12)

K c = ( K I2 + K II2 )1/ 2 (2.13)

K c = [2 / (3 + cos2 )][(1 / ) /(1 + / )]/ 2 [(1 + 3 cos2 )K I2

(2.14)
+ 8 sin cos K I K II + (9 5 cos2 )K II2 ]1/ 2
where is the angle between the initial crack plane and the plane in
which the crack propagates. Sih (1974) analyzed mixed mode fracture
and concluded that a crack will propagate in the direction of minimum
strain-energy density, Scr, which can be expressed as
Scr = (1/ )(all K I2 + 2 a12 K I K II + a 22 K II2 )
(2.15)
= [( 1)K Ic2 ]/ (8G ),
so that
2
K IC = [8G/ ( 1)](all K I2 + 2 a12 K I K II + a22 K II2 ), (2.16)
where G is the shear modulus, = 34, for plane strain conditions,
and a11, a12, and a22 are given by:

a11 = (1/16G )[(1 + cos )( cos )]

a12 = (sin /16G )[ 2 cos ( l)] (2.17)
a22 = (1/16G )[( + 1)(1 cos ) + (l + cos )(3 cos 1)].

In the study by Petrovic and Mendiratta (1976, 1977), the aws were
nearly semicircular. In many cases, it is often necessary to use an
expression to calculate KI and KII that takes into account the degree of
ellipticity of the aw. For KI, this expression is shown in Equation 2.18,
 Mixed Mode Loading 15

and is a relatively simple procedure, well known in the fracture mechan-

ics literature (Randall, 1967).

K I = (Y / )c1/ 2 sin 2 , (2.18)

where Y = (1.2)1/2 for a surface aw, is the angle between the stress
direction and the initial crack plane, is an elliptical integral of the
second kind involving the major (b) to minor (a) radii ratio of the aw,
and c is the smaller of a or b. For KII, an expression used by Petrovic
and Mendiratta (1976, 1977), can be used:

K II = [ 4 /1/ 2 (2 )]c1/ 2 sin cos . (2.19)

Equations 2.7 and 2.8 yield values of KI and KII at the boundary of the
aw and the specimen surface. Kc can also be calculated by measuring
fracture-mirror size (Mecholsky et al., 1974, 1977):

K c = [( K bj / )(c /rj )1/ 2 ], (2.20)

where Kbi is the branching stress intensity constant (i.e., fracture

mirror constant) for the material (= Yri1/ 2), corresponding to either
the mirror-mist (j = 1) or mist-hackle (j = 2) mirror radius, r, and c is
(ab)1/2, and is an elliptical integral of the second kind and is a function
of the a/b ratio.
Marshall (1984) studied the behavior of indented cracks with and
without residual stress. He found contrasting responses of as-indented
and annealed indentations to mixed mode loading. As indented aws,
which are inuenced by residual contact stresses, extend stably during
failure testing and align normal to the maximum tension prior to failure.
The critical conguration does not involve mixed mode loading, but
the strength is inuenced by the resultant kinking of the crack. Annealed
indentations, which are free of residual stresses, extend unstably at a
critical loading with an abrupt change of fracture plane from that of the
initial aw. These results would be expected to apply generally to brittle
materials as long as crack growth is not disrupted by microstructural
variations on a scale comparable with the crack size. He found that all
of the fracture criteria that are consistent with observed crack paths
provide underestimates of the strengths of stress-free indentation aws
in mixed mode loading.
16 CHAPTER 2 Fracture Mechanics Background

Marshall (1984) observed that the most soundly based fracture

criterion (maximum strain energy release rate) underestimates the
strength of the annealed aws in mixed mode loading, as does the
maximum normal stress criterion. The strain energy density and copla-
nar strain energy release rate criteria also underestimate the strengths
of the annealed indentation aws, but both criteria give predictions that
are consistent with the data for the annealed machining damage. Mar-
shall attributed the strength deviation in the annealed specimens to
possible shear tractions due to surface asperities. Glaesemann et al.
(1987) suggested that mixed mode cracks will reorient themselves to
be perpendicular to the maximum principal tensile stress due to slow
crack growth processes for both indented and stress free cracks. In
materials that exhibit slow crack growth, this would explain the reason
for the strength to be less than that predicted by mixed mode theory.
Furthermore, Marshall (1984) determined that an empirical correlation
between the strengths of as-formed aws in mixed mode loading and
predictions based on the coplanar strain energy release rate criterion:

/ 0 = (cos )1/ 2. (2.21)

Although there is no physical basis for this correlation (as-formed aws

are perpendicular to the applied stress at the point of failure), it is useful
for making predictions of strength due to mixed mode loading. Obvi-
ously, more work is needed in the area of mixed mode loading to
translate from testing to the understanding of mechanisms.
Another more recent and promising criterion for predicting the
behavior of brittle elastic bodies under the inuence of mixed mode
loading is known as the generalized maximum tangential stress (GMTS)
criterion (Ayatollahi and Aliha, 2009; Smith et al., 2001). The tangen-
tial stress, , in a linear elastic cracked body, can be written as an
innite series expansion (Williams, 1957):

= [1/ (2 r )1/ 2 ][cos / 2]{K I cos / 2 K II sin } + T sin 2 + O(r 1/ 2 ),

(2.22)

where r and are the crack-tip coordinates. T is usually called the T-

stress, and is a constant, nonsingular term independent of the r distance
from the crack tip. The O(r1/2) terms are higher-order terms and are
usually negligible near the crack tip. According to the GMTS criterion,
 Questions for Chapter 2 17

crack growth initiates radially from the crack tip along the direction of
maximum tangential stress, 0. Also, the crack extension takes place
when the tangential stress along 0 and at a critical distance rc from
the crack tip attains a critical value c. Both rc and c are assumed
to be material constants.
The criterion for crack propagation under mixed mode loading
according to the GMTS theory is:

K IC = cos / 2[ K If cos2 ( 0 / 2) 3 / 2 K IIf sin 0 ] + (2 rc )1/ 2T f sin 2 ( 0 / 2),

(2.23)

where KIf, KIIf, and Tf are the critical values of the stress intensity, and
T stress for mixed mode fracture. The GMTS criterion has only been
compared with a few materials. It should be used to test the results of
previous researchers to see if this criterion can explain previous results.
The question of how the effects of the various modes should be
combined mathematically to calculate fracture toughness is still a ques-
tion of interest. Test procedures that can be used to determine crack
growth characteristics under mixed mode loading are described in
Chapter 4.

SUMMARY

The strength of brittle materials is limited by the size and geometry of

surface and volume defects. The most common defects are small surface
aws introduced during machining or handling. Linear elastic fracture
mechanics provides a way to analyze and understand brittle fracture by
linking applied stress to aw severity through a stress intensity factor.
We dene critical fracture toughness as the stress intensity factor at
which crack growth begins. Cracks can propagate under mixed modes
of loading as well.

QUESTIONS FOR CHAPTER 2

1. Compare and contrast Grifths crack growth model and that developed by
Irwin.
18 CHAPTER 2 Fracture Mechanics Background

2. If a material fails with a strength of 14 MPa and a fracture energy of

8 J/m2 in tension, what is the size of the fracture-initiating defect according
to Grifth theory? According to Irwins theory, using the stress intensity
approach? Assume E = 450 GPa and Y = 1.26.
3. Show that the singularity dominated zone, r, is directly related to the crack
size for an innitely sharp crack in a brittle material.