Frac To Graphy
Frac To Graphy
Frac To Graphy
Fractography of
Ceramics and
Glasses
George D. Quinn
http://dx.doi.org/10.6028/NIST.SP.960-16e2
N I S T
Special
Publication
960-16e2
NIST Recommended Practice Guide
Fractography of
Ceramics and Glasses
George D. Quinn
Materials Measurement Laboratory
May 2016
The goal of this Guide is to make fractographic analysis of brittle materials less
an art and more an engineering practice for scientists and engineers. This guide
emphasizes practical approaches for problem solving and failure analyses.
This Guide is a free public service of NIST for the benefit of American
industry, and the worldwide ceramic and glass communities. The demand for
this book has been very strong since it was first published in 2007. The first
print run of 3,000 copies was distributed around the world and the feedback has
been very positive. Users say that they appreciate the content, the careful
layout and organization, and the wealth of colorful illustrations and examples.
This book is often used as a supplement to courses on fractography of ceramics
and glasses. I am grateful to the National Institute of Standards and
Technology for printing the book on high-quality paper and in color. Although
the book is also available electronically, most users prefer the printed form so
that it is handy on their desks or laboratories
“From the beginning of time to its end, no two cracked surfaces will fit
each other, unless originally they were part of the same piece. Cracks are
as distinctive as fingerprints. … Now the important thing about cracks is
that they must be propagated. They do not originate all over the final
fracture surface, but at one tiny spot, and from thence are propagated out
iii
Fractography of Ceramics and Glasses
to the rest of the area. This fact and the fact that the telltale marks or
“fingerprints” of the process are left on the surface, provide the
groundwork for the subject of Fracture Diagnosis, which in the last
decade has become an important minor branch of physical science.”
This guide emphasizes practical approaches for problem solving and failure
analyses, but it also distills the state of the art in our “important minor branch
of physical science.” I get gratification from the free dissemination of all the
know-how developed over decades by many conscientious researchers,
materials scientists, and engineers. I salute them and have done my best to
credit their contributions by liberal use of citations and the extensive
bibliographies in this Guide.
This edition is about 15% larger than the first edition. It corrects some of the
errors, including the description of vicinal illumination. There are 300 new
illustrations; many of which are from my colleagues around the world. All
told, there are approximately 1000 illustrations and tables in this second
edition. In Chapter 1, Introduction, and Chapter 2, Resources, I added a little
about the History of Fractography and the new documentary standards. The
fracture mirror size procedures in Appendix D were adopted in an ASTM
standard 1678 in 2009, and the text in this Guide has been adjusted
accordingly. Chapter 3 on Equipment has updated microscopy techniques, a
rapidly evolving field. A major correction was made to the description of
“vicinal illumination,” which was described erroneously in the first edition.
Chapter 4 on Patterns has new material on sketches and drawings, edge
chipping, thermal stresses and thermal shock, compression fracture, and
laboratory-scale tests on medicinal vials and dental crowns. Chapter 5 on
Fracture Surfaces has new information on mechanical fatigue, slow crack
growth mechanisms, scarps, glue chips, and a new marking: “step hackle.” The
term “microstructural hackle” has been refined. Chapter 6, Origins, has 28
additional pages with dozens of new figures, and new information on bubbles,
processing and firing flaws, cusps and geometric sharp points. There is a new
definition of “Griffith flaw.” Chapter 7, Quantitative Analysis, has a number of
additions including a figure showing flow lines of stress around bubbles or
flaws; new text about “far-field stress;” new text about the accuracy of the
Newman Raju stress intensity factors and how to extend them to a/c >1; and
Bansal’s approximation. Chapter 9, Composites, has new examples of highly-
filled resin matrix composite examples from the dental materials field. Strictly
speaking, they are not ceramic or glasses, but they behave brittlely and have
many of the same fractographic features. There are several new cases studies
in Chapter 10, including an AlON coarse-grained IR dome; ceramic dental
crowns, and a single crystal sapphire spacecraft refractor. Chapter 11,
iv
Preface
Acknowledgements
The author derived much inspiration from the late Professor Van Derck
Fréchette of Alfred University. Van made many important contributions to the
science of fractography of brittle materials, not the least of which was his 1990
seminal book: Failure Analysis of Brittle Materials. Van solved many
practical, vexing, and spectacular cases and trained hundreds of fractographers
over his long and distinguished career.
The author has had a continuing and fruitful collaboration with Professor James
Varner at Alfred University. Dr. Varner and the author continue the Van
Fréchette legacy by training a new generation of fractographers.
The author thanks Dr. Jeffrey Swab, of the U.S. Army Research Laboratory;
Aberdeen, Maryland and Mr. Michael J. Slavin (formerly of ARL) for their
help in crafting a framework for fractographic analysis that was used to create
the U.S. Army Military Handbook (MIL HDBK) 790 in 1990. This MIL
HBDK was the precursor to American Society for Testing and Materials
standard (ASTM) practice C 1332 adopted in 1996. This was the first formal
standard for fractographic analysis of ceramics and brittle materials in the
world. Many of the concepts in this Guide stem from our collaboration on
those standards. Dr. Swab and I learned much about how fractographers work
from a Versailles Advanced Materials and Standards round robin on ceramic
fractography that we organized in 1993-1994.
The author has learned much from discussions with and readings of the prolific
writings of Mr. Roy Rice, Prof. Jack Mecholsky, and Dr. Stephen Freiman.
They have significantly advanced the state of the art of materials science
through their fractographic work. I note with sadness the passing of Mr. Roy
Rice in 2011. He was one of the most respected fractographers in the world
and from whom I learned much. His books on mechanical properties and
microstructure of ceramics and composites are very valuable. Roy gave me
many of the single crystal fractographs in Chapter 8 of this Guide.
v
Fractography of Ceramics and Glasses
Dr. Roger Morrell of the National Physical Laboratory has been an inspiration,
particularly with his astute practical applications of fractography to failure
analysis and his own Guide to Fractography.
I also acknowledge the insightful work of Dr. Janet B. Quinn formerly of the
American Dental Association, who made her own contributions to the science
of brittle materials fractography and sometimes presented challenging
fractographic problems to her spouse. About one year after the first edition
appeared in 2007, just in time for her own course on Dental Materials
Fractography, she suddenly became ill and died. At the time, she was applying
fractography to ceramic, glass-ceramic, and composite dental restorations such
as crowns and bridges. Janet and her dear friend Dr. Susanne Scherrer of the
University of Geneva have taught a new generation of dental materials
fractographers around the world how to use fractography effectively.
vi
Table of Contents
vii
TABLE OF CONTENTS
Preface
1. Introduction 1-1
2. Resources 2-1
Documentation 3-2
Microscopes 3-14
viii
Table of Contents
ix
Fractography of Ceramics and Glasses
5.9.2 Stable extension from local residual stress or R-curve effects 5-73
6. Origins of Fracture
6.1 Origins, Flaws, and Defects 6-1
x
Table of Contents
bending 7-30
xi
Fractography of Ceramics and Glasses
analysis 7-69
xii
Table of Contents
Appendix A Bibliography
Index
xiii
Introduction
1. Introduction
Why did it break? What was the origin of failure? Did it break as expected or
from an unexpected cause? Was there a problem with the material or was the
part simply overloaded or misused? Why did this part break and others not?
What was the stress at fracture? Was the laboratory strength test successful or
was there a misalignment? These are common practical questions and the
fractographer can often give straightforward definitive answers.
The curse of brittle materials is that they are prone to catastrophic fracture.
Brittle fracture is fracture that takes place with little or no plastic deformation.
Nature has partially compensated for this shortcoming by furnishing clear
fracture patterns and fracture surface markings that provide a wealth of
interpretable information. Indeed, in many respects, fractographic analysis of
ceramics and glasses is easier and can produce more quantitative information
than the fractographic analysis of metals or polymers.
Figure 1.1 shows broken glass and ceramic rods. Using the techniques
described in this Guide, fractographers are able to determine that the rods were
broken in bending, from surface flaws, and even determine the fracture stress.
(a) (b)
Figure 1.1 Fractured glass (a) and ceramic (b) rods.
This Guide takes a broader view of fractography than merely the examination
of the fracture surfaces. The author wrote the definition that is in ASTM
standard 1322 1:
1-1
Fractography of Ceramics and Glasses
Fréchette 2 also took broad view and described the science of fractography as
the study of fragments and their interpretation in terms of material properties
and conditions leading to failure.
It is surprising how many instances there are in the literature where undue
attention was spent on small fracture surface regions that were probably not
even in the area of fracture initiation. It behooves one to look at the overall
breakage pattern first, before one jumps to a microscopic examination of the
fracture surfaces.
Most people recognize that fractography is a valuable tool for failure analysis,
but fewer appreciate its value in routine mechanical testing or support of
materials processing. Although thousands of ceramic or glass items and test
specimens are broken daily, only a tiny fraction is examined fractographically.
A wealth of information is lost about the causes of fracture and the nature of
the material. The flaw type may be just as important as the fracture stress in a
strength test. Rice rued this state of affairs in 1977: 3
"The most significant experimental procedure that can aid the understanding of
mechanical properties is a study of fracture surfaces, especially to identify
fracture origins…. It is indeed amazing the number of mechanical properties
studies conducted that were extensively concerned directly or indirectly with
the size and character of flaws and microstructure from which failure
originated in which no attempt was made to experimentally observe and verify
the predicted or implied flaw character."
1-2
Introduction
What are the skills that a fractographer needs? Fractographers ideally should
have knowledge of materials science, microscopy, mechanics (stresses and
strains), and an aptitude for problem solving and troubleshooting. Puzzle-
solving skills can be helpful as well as an ability to be able to sketch or draw as
discussed in section 4.7. They should be comfortable with microscopy, since
many of the features to be studied are smaller than the unaided eye can discern.
Some knowledge of materials science is necessary since strength-limiting flaws
and crack propagation behavior are controlled by processing and the
microstructure.
a
This is reminiscent of Arthur C. Clarke’s Third law, which is: “Any sufficiently
advanced technology is indistinguishable from magic.” Profiles of the Future: An
Inquiry into the Limits of the Possible, A. C. Clarke, Bantam, 1973.
1-3
Fractography of Ceramics and Glasses
σ2 σ2 σ2
σ1 σ1 σ1 σ1
σ2 σ2 σ1 τ
σ2
σ1 σ1 σ1
τ
Uniaxial Biaxial Equibiaxial Torsion
Tension Tension Tension σ1 = σ2 σ2 = -σ1
Most fracture problems the reader will encounter will be uniaxial or biaxial
stress cases, but there are very important cases (projectile-impacted ceramic or
glass armor plates, or localized Hertzian or sharp contact problems) where the
stress state is triaxial; that is, there are stresses in all three directions, σ1, σ2,
and σ3. Some of these are tension stresses that are denoted with a positive sign
(e.g., +100 MPa), and some may be compressive stresses, denoted by a minus
sign (e.g., -300 MPa). Mechanical engineers are very comfortable with
multiaxial stress problems, and can convert the stresses in any arbitrary
orientation at a given point to the specific orientation that has the principal
stresses. (The X-Y-Z axes at any point can be rotated and are reoriented so that
the small differential volume element has three normal stresses σ1, σ2, and σ3
1-4
Introduction
and no shear stresses acting on its faces.) Mechanical engineers commonly use
a Mohr’s circle analysis or finite element analysis to find this orientation and
the maximum principal stresses. Maximum principal stresses usually control
fracture in ceramics and glasses. Only in specialized cases do shear stresses
cause fracture. This is an important difference in how ceramics and glasses
break as compared to metals and polymers.
It should also be borne in mind that most parts and laboratory specimens have
stress gradients and may even have internal residual stresses. It is rare that a
component has a uniform stress throughout. Simple loading configurations are
typically covered in the first chapters of an elementary strength of materials
engineering textbook, which may be consulted to gain a basic familiarity with
stresses and strains. It is beyond the scope of this Guide to delve into
multiaxial stress fracture criteria, but the reader should be aware that what
causes fracture in a tensile-loaded ceramic part may be completely different
than for a compressively-loaded ceramic part, or for a part with mixed tension
and compression stresses. Engineers often use stress diagrams with fracture
envelopes to illustrate fracture versus no-fracture boundaries in multiaxial
stress problems. Mechanical engineers not versed in ceramic or glass behavior
do not appreciate the nuances of multiaxial failure criteria for ceramics and
glasses. For example, many finite element modelers who are new to the
ceramic field erroneously use the Von Mises (or maximum distortion energy)
criterion in their models. While this criterion works quite well for metals, it is
completely wrong for ceramics and glasses. Many mechanical engineers are
surprised at how different the Poisson’s ratios for glasses and ceramics are as
compared to those for metals!
There are two special stress conditions that warrant explanation. Mechanical
engineers may say that parts are loaded in plane stress or plane strain
conditions. Plane stress refers to a case where there are only two principal
stresses, σ1 and σ2. The third stress σ3 = 0. In other words, all the stresses lie in
a flat plane. This is commonly the case with thin parts, such as a thin-walled
cylinders or pressurized bottles. The principal stresses are parallel to the
surface. In contrast, with thick parts, there can be a σ3 internal stress. If the
part is very thick, lateral thinning strains may be inhibited. The strain through
the thickness, ε3 = 0. There are only two principal strains ε1 and ε2 and they are
parallel to the surface. This is called the plane strain condition and it is
especially important for fracture toughness test specimens. There are stringent
requirements on the thickness of metallic fracture toughness specimens to
ensure that they are in a plane strain condition. The matter is not important for
ceramics and glasses, since the plastic yield zones in ceramics are tiny.
1-5
Fractography of Ceramics and Glasses
Part of the skill of a fractographer is knowing where to look and how to look.
A specimen that breaks into only two pieces may not be too difficult to
interpret. Medium to high strength specimens with multiple fractures can
initially be confusing, but with a little experience and a few simple precautions,
one can find the primary fracture. Component or service failures may be
difficult since the loading conditions may be unknown. Key pieces may be
missing or the evidence incomplete or contradictory. Multiple crack systems
from different causes and events may intersect and confound an interpretation.
The author suspects one reason that fractography is sometimes held with
suspicion or is deemed subjective is that some incautious fractographers have
jumped to conclusions and reached questionable interpretations. It is unwise to
make conclusions based on limited or incomplete evidence. Novices often
force fit a fracture scenario to their limited experience base. Some fracture
patterns may be difficult to detect or be very subtle and may not be recognized.
Good fractographers recognize their strengths and the limitations of their
experience base and do not overreach or extrapolate. They should be alert to
unusual or new fractographic markings or to failure modes with which they are
not familiar and should be ready to search the literature, consult colleagues, or
try to create comparable markings in the laboratory. No fractographer is born
with a built-in data base of fractographic patterns in the brain, so step-by-step
accumulation of experience is necessary. One may consult textbooks, reference
articles, and this Guide to help acquire knowledge, but there is no substitute for
hands-on direct eyeball and microscopy experience.
Fractography is a tool for the broader topic of failure analysis. Figure 1.3
suggests a simple but apt analogy. The fractographer is called upon to solve a
fracture mystery. The detective uses his or her powers of observation to study
the scene of the fracture and meticulously collects and preserves the available
evidence. The clues are contemplated and weighed against the available
background information furnished by witnesses. Comparable cases are
1-6
Introduction
1-7
Fractography of Ceramics and Glasses
This is a technical book. It is beyond the scope of this book to delve into issues
of litigation, fault, or blame. Readers who are interested in these matters are
referred to Chapter 11 “The Expert Witness” in Van Fréchette’s book,2 an
article by B. Adams in the second Alfred conference on fractography, 7 an
article by C. O. Smith in Volume 11 of the ASM handbook, 8 or a review by
Bradt in the third Conference on Fractography of Advanced Ceramics in
Slovakia. 9
1-8
Introduction
1-9
Fractography of Ceramics and Glasses
Chapter 1 References
1-10
Introduction
1-11
Resources
2. Resources
Each chapter of this Guide also contains a list of references for topics covered
in the chapter.
2-1
Fractography of Ceramics and Glasses
2-2
Tools and Equipment
The most important tools of the fractographers craft are the binocular
stereoptical microscope and the scanning electron microscope. The stereoptical
microscope affords to the eyes a magnified, naturally-appearing, three-
dimensional view of the fracture surface that retains all color and reflectivity
information. Fracture origins and flaws in ceramics and glasses can often be
detected with the stereoptical microscope, but higher magnifications are often
needed to see the flaw more clearly. The scanning electron microscope serves
this need and has good depth of field, high magnification capability, and
chemical analysis features. Other tools such as hand magnifiers, compound
optical microscopes, or even atomic force microscopes play supportive roles.
The “marriage” of the microscope with digital cameras and computers has
revolutionized the field. The new technologies have rendered film photography
obsolete and are opening up marvelous new capabilities not heretofore
available. The shortcomings of conventional compound microscopes (e.g.,
limited depth of field) or of scanning electron microscopes (e.g., flat-appearing
images) can be overcome. Virtual three-dimensional images can be
constructed with automatically-rastering optical microscopes or by analysis of
multiple scanning electron microscope images taken with slightly different
specimen tilts. The pseudo three-dimensional images can be displayed, tilted,
and rotated to afford different perspectives. Various quantitative numerical
analyses of surface topography (e.g., surface roughness or even fractal
dimension analysis) can be performed at almost no extra effort since the surface
topography has been recorded digitally.
Simple low-cost software programs are now available that can simply and
conveniently overcome the depth of field limitations of virtually any
microscope. A series of photos at the same magnification are taken while
slightly readjusting the focus in sequential steps. The software interprets the
regions that are in focus and stitches these together to create a single image
with an infinite depth of field.
In this chapter, traditional and new tools of the trade and are reviewed with an
emphasis on their applicability to fractography. Digital cameras are discussed
in four sections of this chapter: first in section 3.2 as simple tools for
photographing an overall part, then in sections 3.4 and 3.5 as recording devices
attached to conventional microscopes, and finally in section 3.7 as components
of new digital microscopes which do not even have eyepieces for the viewer to
look through.
3-1
Fractography of Ceramics and Glasses
The examination of fractured pieces begins with a visual examination using the
unaided eye. Some fractographer prefer to use lint-free gloves when handling
fragments, but gloves usually are not necessary. Ceramics and glasses are hard,
durable materials and may be cleaned fairly easily.
Small jeweler’s loupes as shown in Figures 3.2 and 3.3 also are handy,
particularly in the field. Their working distances range from 50 mm for the
low power (5X) loupe, to about 25 mm for a 10X loupe. Some practitioners
may be adept at holding a watchmaker’s loupe in one eye without using a hand,
thereby freeing up both hands. Most fractographers will use one hand to hold
3-2
Tools and Equipment
the loupe and the other to hold the examined fragment. Machinist loupes and
pocket optical comparators with magnifications up to 20X (two are shown on
the right of Figure 3.2) have built-in measuring scales with graduations as fine
as 0.1 mm (0.005 in). Rapid measurements of mirror sizes or even fracture
origin sizes may be made with these inexpensive tools. Remember, the more
powerful the magnification, the smaller the field of view and the shallower the
depth of field, so a good general duty loupe or comparator with a 7X or 10X
magnification may be perfectly suitable and more versatile than one with a 20X
magnification. At these low magnifications, illumination is usually not a
problem and ambient light is adequate.
Figure 3.2 Hand loupes and optical comparators. From the left, 5X and 7X
watchmaker loupes; a 20X loupe, a 7X comparator with an internal 20 mm
scale marked in 0.1 mm increments, and a 8X to 16X zoom comparator with an
16 mm scale marked in 0.1 mm increments.
Figure 3.3 Loupes (5X to 20X) may either be held by the eye or by hand. The
part is a single crystal silicon rod.
3-3
Fractography of Ceramics and Glasses
Simple holders or clamps may be useful as shown in Figures 3.5 and 3.6,
particularly if the fractographer is examining repetitive examples of a particular
type specimen or component. It can be exasperating to hold a part in one hand
at just the right angle, while holding a magnifier with the second, but then need
an extra hand to apply some alcohol with a cotton swab to clean the part.
Similarly, the fractographer may wish to jot down some important information
as part of a written record.
Figure 3.4 A jeweler’s or hobbyist’s head mounted magnifier (2X to 3.5X) can
be helpful.
Figure 3.5 Specimen holders. Clockwise from the lower left are: a bend bar
holder made with an alligator clip on a bendable (10 gage electrical) wire with
electrical tape wrapped around the teeth to cushion the specimen, two
aluminum jigs with slots and grooves for bend bars or tension strength
specimens, an aluminum block with hole, and three aluminum holders
comparable in size to a glass microscope slide with groves to hold bend bars
both on their flats and on edge for chamfer examination.
3-4
Tools and Equipment
Conventional clays should not be used for mounting specimens unless there is
no alternative. Clay is a pernicious material that can easily contaminate the
fracture surfaces and blend into the fracture features. Natural clays have colors
and compositions similar to ceramic products. They are next to impossible to
clean off. Their electrostatic surface charge and extremely fine plate-like
structure cause them to cling to the fracture surface. Clays often have an oil to
keep them plastic that makes things even worse, since the oils can smear over
the fracture surfaces even on glasses. Clay often gets on one’s hands and then
tends to get on everything. More information on clay contamination is in
section 6.10.
(a) (b)
Figure 3.6 Component or specimen holders. (a) Clockwise from the lower
left: a homemade plastic platform mounted on a swivel base; a pivoting
wooden holder, a cork ring; a clamp on a stand, an alligator clip on a stand
(with electrical tape on the teeth to cushion the specimen), an inverted
miniature wood flower pot with a hole, and two wooden candleholders with
dolls heads and polymer mounting clay. The doll’s head balls can be rotated
to obtain optimum illumination angles onto the specimen’s surface. (b) shows
a polymeric clay that is safe to use and is easily removable.
a
One example of a polymeric clay is Sculpey III Polymer, “Oven bake clay” made by
Polyform products, Elk Grove, Ill. It is easily dissolved by paint thinner or acetone.
3-5
Fractography of Ceramics and Glasses
Simple cleaning tools and markers such as shown in Figure 3.7 should be
conveniently located at the examination work station. Acetone, ethanol, or
methanol with tissues and cotton swab applicators are handy for cleaning.
Compressed air from air lines should be used with caution since there may be
traces of oil mixed with the air. Water soluble ink pens (with fine tips) are
handy for marking the specimen. Ceramic fracture surfaces should never be
cleaned with metallic tools or probes. It may be tempting to try to scrape or
poke off a contaminant, but metallic tools are abraded by the harder ceramic
causing metallic traces to be smeared on the fracture surface. Fractured dental
restorations often have plaque on them. The plaque can even be on a fracture
surface if it has been exposed to the oral environment for any period of time.
An example of what plaque looks like is shown in section 6.10 of this guide,
where contaminants are discussed. Plaque can be removed by using an
ultrasonic bath with a 16% solution of NaOCl (“bleach”) for 8 minutes, then
rinsing well with water (courtesy S. Scherrer).
There are two instances where specimens should be examined before cleaning.
Firstly, staining or discoloration of a flaw may provide important information
about the flaw itself. Secondly, fine debris of the material itself on the fracture
surface may be a clue that cyclic fatigue has occurred. These are discussed at
the very beginning of Chapter 5, Fracture Surface Examination.
3-6
Tools and Equipment
(a) (b)
3-7
Fractography of Ceramics and Glasses
Modern digital single lens reflex (SLR) cameras and even simple consumer
digital cameras with a macro option can be used at very close distances (~100
mm or less) from the object. Digital cameras, either professional high-
resolution models or simple consumer versions such as shown in Figure 3.9,
have revolutionized fractographic photography. They have auto focus and
exposure controls and can capture good images in ambient lighting without a
flash. Liquid crystal display (LCD) monitors on the rear of the cameras allow
fractographer to see what the lens will capture when the image is snapped.
The digital camera instantly displays the captured image. Remedial action can
be taken immediately and a new image captured. This is very efficient and led
to the rapid obsolescence of film-based photography. Another advantage is that
images taken with digital cameras seem to be less sensitive to the illumination
source color. Color film often produced odd-colored images if film balanced
for outdoors was used indoors with tungsten filament bulbs or fluorescent
lamps. Another great advantage of the digital camera is that it needs less light
to make a good quality image Ambient illumination usually suffices. Flash
lamps used with old film cameras often over- or underexposed an object and
created harsh shadows.
Figure 3.9 Digital camera with macro capability, auto exposure, and auto
focus.
Experience has shown that a digital camera chip with 2 million to 5 million
pixels is adequate for most applications. Some publications require no less
than 3 megapixel resolution and images must be presented at 120 dots per cm
(300 dots per inch). The LCD monitor should be bright and large (at least 40
mm). The camera should have auto focus and auto exposure options and, most
importantly, have a macrophotography option for close ups at distances as short
as 100 mm. File storage format should include a Joint Photographic Experts
Group (JPEG) or Tagged Image File Format (TIFF) option. Sometimes Bitmap
(BMP) format files are used with Microsoft Windows operating system
computers. Some professionals prefer the raw image format (RAW), which is
3-8
Tools and Equipment
JPEG images have an option that they may be saved with variable amounts of
compression, but users should record images with little or no compression. A
copy of the image can be made at a later time with some compression to reduce
the file size so that electronic documents or presentations with many images do
not become unwieldy. JPEG files are as much as 1/5th to 1/10th the size of an
uncompressed TIFF file, with very little or no loss of detail. JPEG is available
only for true color or grayscale images. JPEG reduces the number of bits in the
image by eliminating repetitive image data or image data that is hard to see.
The software may allow various compression levels and the higher the level of
compression, the lower the quality of the image. JPEG may compress more
than other methods and the results may not be fully reversible. At higher
compression ratios (> 20:1 or a 50% setting), JPEG can degrade the quality of
the image. A JPEG file is automatically uncompressed when it is opened.
The author typically saves and presents his images in JPEG format with no
compression. Virtually all the photographic images in this Guide were saved in
the JPEG format. Line art and graphs were saved in LZW compressed TIF
files.
3-9
Fractography of Ceramics and Glasses
now cost less than one hundred dollars. More elaborate single lens reflex
(SLR) digital cameras have more options, more versatile lenses, and higher
resolutions. Some may have extremely high resolutions, but it should be borne
in mind that the stored image file sizes may be very large. SLR cameras should
be ordered with close up lenses as an option.
(a) (b)
(c) (d)
Figure 3.10 Cell phones can be used for fractographic analysis (a) shows an
obsidian arrowhead as viewed at 2X magnification in a cell phone; (b) a street
photo near Santo Stefano Basilica in Bologna, Italy in 2014; (c) a close-up of
the tempered glass fracture pattern, and (d) the close-up at 5X of the internal
inclusion origin (arrow). The author was on his hands and knees for (d), which
was shot at 5X about 60 mm above and aiming through the glass at an angle.
3-10
Tools and Equipment
3-11
Fractography of Ceramics and Glasses
Figure 3.11 Stereo binocular microscope with a fiber optical light source for
directional illumination. This is an indispensable tool for the ceramic and
glass fractographer. “Low-angle, grazing illumination” is illustrated in the
picture. In other cases, a ring light mounted on the objective lens of the
microscope may be suitable. This particular microscope has interchangeable
0.63X, 1.0X and 2.0X objective lenses, a body zoom magnification range of
.65X to 4X, and 25X eyepieces for an overall magnification range of 10X to
200X . The moderately tall (300 mm) mounting post enables specimens up to
150 mm tall to be placed under the microscope.
The stereomicroscope is designed such that each eyepiece views the object
from a slightly different angle, thereby creating two different images. The
brain interprets these as a three-dimensional view. The stereomicroscope is in
reality two separate microscopes. Older rudimentary Greenough design
stereoscopes had two completely different optical paths, one for each eye.
They had very limited capabilities. Most contemporary stereomicroscopes use
a common objective lens and then two internal optical light paths in the body
up to the eyepieces. Parfocality, whereby the object remains in focus while the
magnification is changed, is a very advantageous feature of the microscope.
Follow the manufacturer’s guidelines to set the eyepieces for Parfocality.
3-12
Tools and Equipment
depth of field make it possible to clearly see surface features despite the
inherent roughness or curvatures of fracture surfaces. Some stereomicro-
scopes also have diaphragm apertures that can be stopped down to enhance the
depth of field. Many stereomicroscopes are modular and easily modified or
expanded.
3-13
Fractography of Ceramics and Glasses
for size measurement purposes, either by the use of a graduated reticule in the
eyepiece, or by a camera mounted on the microscope.
Final magnification is provided by the wide field eyepieces that are usually
10X, 16X, or 25X. Stereomicroscopes may be used either with or without
eyeglasses. The distance between the eyepiece tubes usually can be adjusted
for the interpupilary distance. Eyepieces may have rubber boots that attach to
the eyepieces for screening out stray light coming into the eyes. These can be
removed or slid out of the way for eyeglass wearers. Typical overall
magnifications can range from as low as 5X to as high as 200X. Some modern
stereomicroscopes can achieve as much as 300X to 400X magnification with
16X or 20X body magnifications and 1.6X objectives.
A very valuable but expensive optional feature is a tilting eyepiece head that
may be adjusted up or down so that the viewer does not have to extend or bend
his or her neck to see through the eyepieces, or to have to move a chair or stool
up or down. A tilt head can cost between $ 3,000 and $ 4,000.
3-14
Tools and Equipment
appropriate internal lens to match the image size to the camera chip size.
Some microscope bodies have a lever switch that directs the light to either the
eyepiece or to the camera. This is extremely inconvenient. A much better
scheme is a beam splitter that directs 50 % of the light to the eyepiece and 50 %
to the camera so that the image can be viewed at the same time it is projected
onto a monitor.
Instant film cameras are now obsolete. They were supplanted first by video
cameras, but now almost exclusively by digital cameras connected to a
computer. Video cameras which had fast image capture capabilities had to be
used with frame grabbers to snap an image. They had very low 640 x 480 or
768 x 500 picture element resolutions so that many images per second could be
taken. They have been rendered obsolete as digital cameras became more
capable, less expensive, faster even with low light, and easier to use.
3-15
Fractography of Ceramics and Glasses
The camera input can be fed to free software provided with the digital camera
as part of a complete package, or to a separate image processing software
package. In the past there inevitably were compatibility problems with
software and cameras and all the usual annoying nuances of computer
programs. Contemporary digital cameras and software packages are much
more user friendly. Nevertheless, it probably is best to have digital cameras
and their software installed by a camera or microscope company representative
who has experience trouble shooting installation and compatibility problems.
The simpler the software the better! The author has used digital software
packages that are excessively complicated, in large part since they were written
with lots of options and clever additions for experts. They are slow and
cumbersome, and have so many menus and submenus that they are a nuisance
to use. (Users should beware of salespersons who are eager to sell expensive
software packages that the user does not really need.) A lot of time is lost in
training students and new users.
3-16
Tools and Equipment
Figure 3.12 A silicon nitride 6 mm diameter rod broken in flexure (801 MPa).
(a) and (b) show stereoptical microscope digital images. Even though the
microscope has great depth of field, the cantilever curl and holder are out of
focus in (a). (b) shows the holder in focus. (c) shows the digitally
reconstructed composite image from 15 separate images.
3-17
Fractography of Ceramics and Glasses
Figure 3.13 A 25 mm glass fragment. (a) and (b) show typical images made
by a stereoptical microscope at low magnification. Although the microscope
has excellent depth of field, portions of the fragment are out of focus in each
view. (c) shows the digitally reconstructed image made from 16 separate
images focused from the top to bottom of the fragment. The digital
reconstruction took less than 1 minute.
3-18
Tools and Equipment
Figure 3.15 Digital microscopes are a new tool for the fractographer. This
simple digital microscope (arrow) is mounted in a stand on the right. It is
focused on the fracture surface of a broken glass rod. The monitor and
computer (not shown) are separate components
3-19
Fractography of Ceramics and Glasses
that is becoming increasingly common. They can range from inexpensive low-
power, hand-held models to elaborate, high-power, expensive models mounted
on a rigid platform with full Z-axis control and digital image stitching
capabilities. Figure 3.15 shows a mid-priced model (~$ 1,000 plus $130 for the
stand) with variable magnifications from 20X to 250X and corresponding
working distances of 48 mm to 2 mm. It may be difficult to obtain sharp,
focused images with the simpler models that are hand held, so a stand is
advisable. The digital microscope is not parfocal. As the magnification is
changed, the image must be completely refocused and the working distance is
changed significantly. This is a nuisance. Another limitation is illumination,
which is usually provided by built-in light-emitting diodes surrounding the
lens. This limits their ability to highlight or even discern critical fracture
surface features. For example, shadowing, or low-angle grazing illumination,
which is essential for ceramic examination and fracture mirror examination,
cannot be done. A separate light source is recommended.
At the other end of the spectrum are high-end digital microscopes mounted on
very rigid fames and bases and which may have built-in vibration dampers.
They may have fully-motorized positioning stages and automated Z-axis
focusing and even have motorized tilting heads. Camera and stage controls are
via joysticks, electronic knobs, or even touch-screen monitor manipulation.
These systems can automatically conduct Z-axis focus stacking analysis and,
combined with the stage control, can generate quantitative three dimensional
profiles of the fracture surface. The virtual three-dimensional image can be
displayed on the monitor and rotated and viewed from different perspectives.
Despite this extraordinary versatility, the illumination systems with such
microscopes (usually a ring of LEDs) are very limited. Many of the techniques
described in this Guide require fully-independent illumination control. It is
recommended that independent illumination sources such as shown in the next
section be acquired for use with such high-end digital microscopes. The
working distances between the lens and the piece should be large enough that
independent light can be directed onto the specimen at variable angles.
Figure 3.16 shows some options for specimen illumination. Bright light
sources are advisable to facilitate the examination of dark ceramics and the use
of high magnifications. The most common illumination source for fracto-
graphic analysis is a ring light that clamps to the objective lens. The author
rarely uses it. It provides uniform lighting, but may not be effective in
3-20
Tools and Equipment
Figure 3.17 Dual gooseneck fiber optic light guides allow flexibility in
directional illumination.
3-21
Fractography of Ceramics and Glasses
(a) (b)
Figure 3.18 Low-angle grazing illumination (a) is effective in highlighting
fracture surface topography as shown in (b), which is the fracture surface of a
ceramic rod broken in flexure.
This is one major revision in this 2nd edition of this Guide as compared to the
first edition. Low-angle grazing illumination was erroneously termed “vicinal
illumination” in the first edition. Vicinal illumination is different.
3-22
Tools and Equipment
(a) (b)
Figure 3.20 Bright field reflected light (a) and vicinal illumination (b) of a
multilayer ceramic capacitor. Both images are at the same magnification.
Only one crack is apparent in (a). Vicinal illumination revealed the second
crack on the right in (b). These are from a compound optical microscope. The
aperture was stopped down for the image in (b). (Courtesy of S. Hull, NASA,
Goddard Research Center.)
3-23
Dark field illuminators such as shown in Figure 3.21 may also be handy. The
dark field illuminator is similar to a normal ring light, but the light is directed
sideways from the ring towards an object in the middle. It is in essence low-
angle grazing illumination from all directions. This sideways illumination
accentuates subtle surface features and scratches.
Figure 3.21 Dark field illumination of the fracture surface of a fractured steel
railroad coupler pin. The arrow points to a fatigue crack.
3-24
Tools and Equipment
(a) (b)
Figure 3.22 Transmitted light can be from an illuminator plate as shown in (a)
or in (b), an improvised stage made of Plexiglas, and white paper or a mirror
for transmitted illumination. A fiber optic light source can be directed onto
the tiltable inclined plane so that light is reflected up through a specimen
mounted on the top. (Thanks to Dr. D. Shechtman for suggesting this to the
author.)
Figure 3.23 Crossed polarizing filters or plastic sheets. Polarizers can either
be dedicated accessories for the stereoscope, or can be improvised from
ordinary single lens camera lens attachments or sheets of plastic polarizing
film.
3-25
3.9 Coatings
Coatings may be applied to specimens that have poor reflectivity or which are
translucent and scatter light from beneath the surface, thereby washing out the
surface detail. A thin layer of gold or platinum may be sputtered onto the
specimen surface in a small vacuum chamber. Such equipment is commonly
used to prepare a specimen for scanning electron microscope examination, but
the coating may enhance optical examinations as well. Figure 3.24 shows an
example.
Gold and platinum coatings may be removed without harm to most ceramic
specimens by soaking the pieces for about 1 minute in “aqua regia.” This is a
combination of 1 part by volume of nitric acid and 3 to 4 parts hydrochloric
acid. This is a hazardous combination and appropriate precautions must be
taken (e.g., gloves and eye protection). Only mix the amount to be used in a
cleaning session. Never store it, since it releases gasses that build up pressure
in a sealed container. The specimens should be rinsed in water afterwards.
When disposing of used aqua regia, water should not be added to the acid mix.
Instead, the acid mix should be carefully poured into a large container of water
or a sink with ample flowing water.
Aqua regia can cause harm to a specimen if the specimen contains anything
metallic, such as a metallic-inclusion fracture origin. The author once had a
highly-filled resin matrix dental composite bend bar that had some trapped
contamination covering the origin during the initial SEM examination. The
original gold coating had to be removed, the specimen recleaned and recoated
for a second SEM examination. The fracture origin was a calcium rich
inclusion. The aqua regia not only removed the gold coating, but it also reacted
with the fracture origin forming calcium chloride. The EDX analysis during
the second SEM session showed chlorine at the origin, which was puzzling
until the author remembered the aqua regia cleaning step and realized that the
HCl had reacted with the flaw.
3-26
Tools and Equipment
(a)
(b)
3-27
Fractography of Ceramics and Glasses
Figure 3.25 A simple but effective coating technique for white ceramics is to
stain the surface with a green felt tip pen. The staining can be watched in real
time through the microscope. The green stain can be easily removed with
ethanol alcohol on a cotton swab applicator and the process repeated.
3-28
Tools and Equipment
A simple expedient that works very well for many white or gray ceramics is a
common felt tip pen as shown in Figures 3.25 and 3.26. Green felt tip pens
work the best, partly because the eye is most sensitive to green wavelengths,
but probably also due to the character of the green dye. b The dyes can easily
be removed with ethanol within a few seconds or minutes and the process
repeated in a matter of seconds. Most of the dye will come off, but some may
penetrate into crack branches or undercuts in the fracture surface and may be
more persistent. If the green dye is allowed to sit for longer periods of time, it
can be removed by soaking the part for 1 to 2 minutes in a 30% solution of
H2O2 (hydrogen peroxide). Note this is a much greater concentration than is
use for topical cleaning of wounds. Gloves and eye protection should be used.
Additional dye and staining techniques are discussed in section 3.24 later in
this chapter.
(a) (b)
Figure 3.26 An example of green dye staining on a dental glass ceramic bend
bar. (a) shows the entire fracture surface illustrating how the stain drains into
valleys and accentuates hackle lines and ridges and how easily the fracture
mirror region stands out (arrow). The bottom view shows the fracture origin:
a surface crack (arrows) made by Knoop indentation for a fracture toughness
measurement. The pictures were medium resolution video prints and the thin
horizontal line is an artifact from a defect in the print head.
b
Papermate Flair, 1.1 mm, Sanford Corporation, Division of Newell, Rubbermaid.
3-29
Fractography of Ceramics and Glasses
(a)
(b)
Figure 3.27 Two stage micrometers (a). A close-up of the bottom one is in (b).
The numbers show 1 mm spacings. The thin lines on the top right are 100 µm
apart. The thin lines on the top left (arrow) are 10 µm apart. This model was
accurate to an amazing 0.01 %.
3-30
Tools and Equipment
An additional advantage for the X-Y stage measurement is that the size
readings are independent of the magnification used to view the specimen. In
contrast, images captured by the camera must be calibrated for the exact
magnification.
Figure 3.28 A X-Y traversing stage with digital micrometer heads is very
convenient for positioning a piece and making accurate size readings.
3-31
Fractography of Ceramics and Glasses
One drawback to the traversing stage is that it takes up vertical space and hence
the microscope body must be moved up higher and possibly may limit the
working range under the objective if the microscope is mounted on a short post.
As noted above, stereomicroscopes for fractographic examination should be
mounted on tall posts for maximum versatility.
Compound optical microscopes such as shown in Figures 3.29 and 3.30 may be
used for fractographic analysis, but are less versatile than stereo-microscopes.
Compound microscopes are best suited for small specimens with relatively flat
surfaces. They magnify much more (up to 1600X) than a stereomicroscope,
but three-dimensional topographical effects are lost. Virtually all work is now
done with air lenses. Oil immersion lenses are no longer needed.
Objective lenses from 5X to 100X may be used in the lens turret, but in practice
the limited depth of field makes it difficult to see a fracture surface in focus at
larger magnifications. Only portions of a fracture surface may be in focus
even at lower magnifications. Working distances are also very small, and care
must be taken not to contact the objective lens with a specimen. Long focal
length objectives are ideal. These have greater working distances, but at the
cost of some loss of light collecting power and resolution. The author uses a
compound microscope with long focal length lenses. The 100X objective has a
1.3 mm clearance and 0.8 numerical aperture. Some 100X lenses are available
with working distances as long at 13 mm, but these have lower numerical
apertures (0.5 to 0.6) and thus less resolving capability. The clearance between
a stage and the objective lenses may also preclude examination of large
fractured pieces. Low-angle grazing illumination is difficult to achieve with
compound microscopes, especially at greater magnifications.
3-32
Tools and Equipment
3-33
Fractography of Ceramics and Glasses
3-34
Tools and Equipment
Compound microscopes are useful for measuring very small features, such as
tiny fracture mirrors in optical fibers or other strong test specimens.
Compound microscopes can also be used in transmitted illumination mode.
These microscopes are also optimal for measuring fracture mirror sizes in
glasses.
In some cases, fracture surface features may actually be easier to observe on the
replica than on the original surface. In many transparent or translucent
ceramics such as porcelains or coarse grain ceramics, light scattering from
below the surface can obscure the fracture surface details. For example, subtle
wake hackle lines behind pores may be easier to see on replicas. Curved
surfaces can be made more amenable to high magnification examination, since
a replica can be flattened. Since some replicas are a negative of the fracture
surface, depressed features such as pores or cracks are converted to raised
features on the replica that sometimes are more easily imaged in the scanning
electron microscope.
There are a variety of materials that may be used to make a replica as shown in
Figures 3.31 to 3.35. Two excellent summary papers with practical tips are by
Varner. 1,2 The replica must be made without the introduction of artifacts. The
replica material must also not damage the original material nor chemically react
with it!
3-35
Fractography of Ceramics and Glasses
Figure 3.31 Thin cellulose acetate replica tape, acetone, applicator, and bend
bar specimens. These are used to make reversed (negative) replicas.
There are two types of replicas: “positive replicas” and “negative (or reversed)
replicas.” Some replica materials are laid over a fracture surface then removed
after they dry sufficiently to retain the critical surface features. The material
preserves a mirror image of the fracture markings: what was left is now right,
and so forth. This is a “reversed replica” or a “negative.” Other replicas may
be made in multiple steps to create a “positive replica.” So for example, a
silicone impression or mold may be made around an object, and then pulled off
the fracture surface. An epoxy resin is then cast into the mold and allowed to
dry. The mold is removed, and now the epoxy represents a positive replica
(i.e., a clone) of the original part. The dental materials community has a good
line of products suitable for making mold impressions of teeth and these
materials are suitable for general fractographic applications as well, as
discussed by Scherrer et al. 3 Fréchette also describes replication techniques in
his book. 4
3-36
Tools and Equipment
(a)
(b)
Porcelain
(c)
Origin
Glaze
3-37
Fractography of Ceramics and Glasses
Since “reversed” replicas like the cellulose acetate tapes are flipped versions of
the fracture surface, it is prudent to label them clearly as to their location on the
component, lest the fractographer confuse top-bottom, or left-right orientations.
Special care should be taken to avoid artifacts in the replicas such as trapped air
pockets, tears, or handling artifacts. Replicas should be kept clean and
carefully stored in containers.
Positive replicas may be made with filled silicone rubbers, unfilled (transparent
silicone rubber, and polyvinylsiloxane (PVS).1,2,3 The latter material is
commonly used in dentistry to make impressions in patient’s mouths as shown
in Figures 3.33 and 3.34. Extraction of a damaged dental crown from the
mouth typically requires its destruction. Therefore it is prudent to make a
3-38
Tools and Equipment
(a)
(b)
Figure 3.33 Dental replica equipment. (a) shows a typical PVS applicator
gun which has interchangeable tips and compounds. (b) shows Dr. S. Scherrer
demonstrating the technique in a dental materials fractography class.
3-39
Fractography of Ceramics and Glasses
Once the mold has set, an epoxy resin may be poured into it. One interesting
advantage of this procedure is that the mold can be reused to make multiple
epoxy replicas. They are useful as teaching aids since every student examines
the same object with the same features. Gold or carbon coatings can be
sputtered onto positive replicas to enhance viewing in reflected light or to
prepare the replica for SEM examination.
3-40
Tools and Equipment
3-41
Fractography of Ceramics and Glasses
(a)
1 2 3
cellulose
acetate
replica epoxy replica
(b) (c)
(d)
Figure 3.35 Replicas of a broken whiteware plate. (a) shows both a cellulose
acetate tape replica on the left, and a dental PVS mold and epoxy positive
replica on the right. The PVS mold was cut into 3 pieces (labeled 1,2,3) and
the middle one gold coated. (b) shows the cast epoxy resin positive from the
plate rim area. (c) shows a close-up of the epoxy replica. Telltale small wake
hackle lines (described in chapter 5) are seen in the glaze (arrows). These
little “weathervanes” are more easily seen in the glaze than in the whiteware
body and they help direct one’s attention back to an origin area shown in (d) .
Coarse hackle lines (chapter 5) lead back to the origin site, a contact damage
crack on the rim marked by the arrow. Coating the clear epoxy with a thin
gold coat would have made the markings even easier to interpret.
3-42
Tools and Equipment
Modern SEMs are not difficult to use and materials scientists and engineers can
perform their own routine examinations without the need of an expert operator.
The advent of commercial SEMs and transmission electron microscopes in the
1960s led to a quantum jump in the quality of fractography of ceramics and
glasses. Researchers and engineers could finally see and characterize the
strength-limiting flaws that previously were too small to be clearly discerned
with optical microscopy. It is fair to say that the SEM opened up a whole new
3-43
Fractography of Ceramics and Glasses
The SEM uses electrons as a signal source and, since the probe size can be
made smaller than the wavelength of light, smaller features can be resolved and
much greater magnifications achieved. A vacuum is necessary to remove gas
that would interfere with the electron beam. The SEM has superb depth of
field over a broad magnification range. For a working distance of 20 mm, the
depth of field may be 0.4 mm at 100X or 40 µm at 1000X and 4 µm at
10,000X. 6 Sometimes the SEM is preferred for this reason at even low
magnifications (100X to 500X) that are well within the range of optical
microscopes. On the other hand, SEM images often look flat and tend to de-
emphasize large surface topography changes. The SEM image can also be
deceptive and it may be confusing to interpret whether a feature projects above
the surface or is a depression below the surface. Stereo SEM viewing,
described below, can help solve this dilemma. A great advantage of the SEM is
that compositional analysis can also be done using accessories that detect X-
rays emitted from the fracture surface.
Specimens must fit into the vacuum chamber and are usually cut and mounted
on a small aluminum stub that will be placed into the SEM as shown in Figure
3.37. The SEM should have a large opening to accommodate broken
fragments and test specimens without having to cut or trim them to small sizes.
The stage has external controls that permit the specimen to be positioned in X,
Y, and Z directions, rotated, and tilted. The size of the chamber and the mount
station, plus the clearance to the electron optics all put limits on the size of
specimens than can be examined. A good SEM for fractography has a large
chamber, a large entry port, and can pump down with a minimum of difficulty.
c
The coincidental advent of modern fracture mechanics in the late 1950s, described
in chapter 7, also led to major breakthroughs in the study of ceramics and glasses.
3-44
Tools and Equipment
(a)
(b)
Figure 3.37 (a) shows an open SEM vacuum chamber showing the specimen
station on the left and the positioning controls on the right, outside of the
chamber. (b) shows a close up of the two fracture specimens on their stub on
the examination mount. Both fragments had to be cut from longer pieces to fit
in the chamber. This model has a large opening and the stage can
accommodate specimens as tall as 25 mm.
3-45
Fractography of Ceramics and Glasses
Figure 3.38 Two matching fracture halves of a broken round tension specimen
on a stub in a protective plastic box. Silver paint was used to attach the
specimens to the stub. Ideally, the halves should have been cut to similar
lengths to save time refocusing in the SEM.
Figure 3.39 SEM specimen preparation tools. Clockwise from the top left are
two SEM length calibration standards, two silver paint bottles, a double-sided
conductive adhesive tape roll, two sheets of double-sided conductive adhesive
disks, and an assortment of aluminum stubs.
3-46
Tools and Equipment
Figure 3.39 shows some common accessories and tools for SEM examination.
The specimens must be cleaned carefully of greases, dust, or debris. It can be
exasperating to find a tiny piece of dust right on top of the origin site when the
specimen is in the chamber. Soaking the specimen in a small beaker of acetone
in an ultrasonic bath for a minute or so is often effective. After removal, a
gentle blast of compressed air can eliminate any acetone residue. Sometimes it
may be sufficient to swipe the fracture surface with a cotton swab soaked in
acetone. Small trays or plastic boxes are convenient for holding the specimens
and protecting them from contamination prior to examination. Special
containers are available for mounting stubs. The best procedure is to keep the
specimens as clean as possible from the moment the fragments are recovered.
There must be a good electrical contact between the specimen and the stub.
The specimen can be affixed to the stub by silver paint or by conductive carbon
double-sided sticky tape. If multiple specimens are placed on a stub, it is a
good idea to make a hand drawn sketch or map to show which specimen is
where on the stub, lest the viewer become confused when viewing the array in
the SEM. Furthermore all the specimens should be oriented in a similar way.
For example, if a series of bend bar fracture surfaces are being examined, they
all should have their tensile surfaces facing the same way. Otherwise, much
time can be lost getting oriented or searching the wrong areas during the SEM
session. A set of optical photos of the fracture surface can be handy when the
specimens are examined in the SEM.
3-47
Fractography of Ceramics and Glasses
some edges of the fracture surface may enhance the image by promoting
contrast. Coating along the specimen sides may be incomplete, so sometimes it
is wise to add a thin silver paint line along the specimen sides from the fracture
surface to the stub. Of course, the silver paint should be kept well away from
the fracture origin or any features of interest on the fracture surface. The
conductive coating step may be omitted for some modern SEMs that operate at
low accelerating voltages and low vacuums as described in a later section on
environmental SEMs. Additional details and tips on coatings for SEM
examination may be found in the paper by Healy and Mecholsky. 7
3-48
Tools and Equipment
Figure 3.41 The electron beam interacts with the specimen surface and volume
creating emissions that may be analyzed. Signals are generated in all
directions, but they come from different depths of the material. The beam is
rastered over the surface and an image is sequentially formed.
The initial working distance can be adjusted to enhance the field of view and
the depth of field. Long working distances allow a lower minimum
magnification which is sometimes helpful in initial orientation or in
photographing the whole part or as much as possible. A longer working
distance also allows the specimen to be tilted more. On the other hand, longer
working distances reduce resolution and may reduce X-ray signal acquisition.
Image clarity is also affected by specimen tilt. As noted above, having the
surface perpendicular to the incident beam minimizes foreshortening, but
secondary electron and X-ray collection can be maximized by tilting the
specimen towards the detectors.
Initial inspections on older SEMs with a cathode ray tube (CRT) display often
are typically with 500 line scans, but when a image is recorded on film, a
higher resolution CRT with 2000 or more line scans is used. The raster of the
beam on the specimen is synchronized with that of the electron beam that
rasters across the display CRT. Modern SEMs with digital monitors also have
differences in resolution between “live image” displays and captured images.
3-49
Fractography of Ceramics and Glasses
The secondary electron mode of operation provides the best spatial resolution
and produces a surface topographical image. Secondary electrons are low
energy electrons (10 eV to 50 eV) primarily produced from the material close
to the surface. Since they are low energy, a positively charged detector off to
one side can capture the emitted electrons for the image. These electrons may
even follow a curved trajectory from the surface to the detector. Edge charging
can be a problem with very rough surfaces or those with sharp corners and
cracks, in which case gamma adjustments to the image may help. Gamma
adjustments can suppress bright areas and enhance dark areas. For some edge
charging cases, backscattered imaging may be preferred.
Figures 3.42 through 3.45 show matching secondary electron and backscattered
electron images of fracture origins. Figure 3.42 shows how the backscatter
mode can even help highlight a controlled semi-elliptical flaw made by a
Knoop indenter.
3-50
Tools and Equipment
200 µm
200 µm
(a) (b)
Figure 3.42 Comparative secondary electron (a) and backscattered electron
(b) images of the fracture surface of a Knoop semi-elliptical precrack in a
silicon nitride fracture toughness test specimen. The edge charging in (a)
emphasizes certain features, but (b) shows that the flattening effect of the
backscattering mode tends to better reveal the overall crack (arrows).
(a)
(b) (c)
3-51
Fractography of Ceramics and Glasses
(a)
(b)
(c)
3-52
Tools and Equipment
(a) (b)
(c)
Figure 3.45 An iron inclusion in a silicon nitride rod (684 MPa). Matching
secondary electron (a) and backscattered electron images (b). (c) is the EDX
pattern of the inclusion revealing it is iron.
3-53
Fractography of Ceramics and Glasses
An X-ray scan may require several minutes as the electron beam illuminates the
surface and the detector accumulates a sufficient signal of a range of emitted X-
rays. A typical scan appears as shown in Figure 3.45c where signal intensity is
plotted versus X-ray energy or wavelength. The specimen response is
compared to known characteristic wavelengths of the various elements to
identify which elements are present. A common procedure is to scan the
critical feature, then rescan a different area away from the feature and compare
the patterns. The accelerating voltages should be high enough to activate
emissions from the elements. This should be 1.5 times the threshold energy of
the element. So for example, to detect copper which has Kα X-rays of 8.0 keV,
it is best to use an accelerating voltage of 12 keV or greater. Otherwise if the
voltage is too low, then the X-rays will not be emitted or detected, even if
copper is present. 10 Similarly, for zirconium, which has Kα X-rays of 15.7
keV, it is best to use an accelerating voltage of 25 keV or greater.
3-54
Tools and Equipment
Figure 3.46 Electron backscatter diffraction map of the flat exterior surface of
a tungsten carbide/11 mass % cobalt fracture toughness test specimen. The
black crack has propagated through the microstructure. Individual grains with
different crystallographic orientations are shown in color. Crack propagation
was under static loading and was mixed inter- and transgranular, with the
� 0] slip planes. (Image courtesy of R. Morrell
latter through the grains on [101
and K. Mingard, NPL)
in Figure 3.46 12 which images the exterior surface of a fracture toughness test
specimen. A focused electron beam is scanned across a crystalline material and
at each point an electron diffraction pattern is created and analyzed. At the
present time EBSD is not amenable to analysis of rough fracture surfaces.
3-55
Fractography of Ceramics and Glasses
The SEM takes pictures that make the surface look flat and do not show the
topography clearly. A much better sense of the full topography can be revealed
by the simple process of taking matching stereo photographs. Two photos of
the same area are taken, but with the specimen tilted 5o to 15o in the SEM
chamber between shots. A minor adjustment to the brightness and contrast
may be necessary for the second photo, but is not essential. Some people are
able to hold the two images in front of their eyes and focus the left eye on the
left photo and the right eye on the right photo. The distance between the photos
is adjusted and the eyes are “relaxed.” The brain merges the images and
interprets them as a single three-dimensional view. Most people find it easier
to put the images under a stereoscope, which is a stereo viewing optical device
as shown in Figure 3.48. The fractographer looks into two eyepieces and the
photos are spaced apart and adjusted slightly back and forth until the three-
dimensional effect is achieved. It may take a little practice to get the two
images to register, but the usual trick is to focus each eye on an identical
feature or spot in the two photos. Initially the two eyes will discern the spot as
two separate images, but by carefully sliding one or the other photos around,
the spots can be made to merge. At that instant the three-dimensional effect
should be detected since each eye will be viewing the same surface, but from
different perspectives. This may take a little practice, but the viewer helps
make it easy for most people. One curious effect is that the perception of what
3-56
Tools and Equipment
features are up or down may change depending upon which photo is on the left
and which is on the right. It may be necessary to swap the photos to get the
correct rendition. The orientation of the photos may have to be turned 90o as
well in order to match the tilt angle used in the SEM.
Figure 3.48 Stereoscope. The two photos are placed under the viewer and the
viewer looks into the two eyepieces. The photos are moved around slightly
until they appear as one image in three dimensions.
Some modern SEMs can digitally capture the pair or of tilted images, or even
three images at different tilts. Image analysis software can then interpret the
multiple images and constructs a pseudo three-dimensional rendition that is
displayed on a monitor. The computer image can be rotated or displayed from
different vantage points and height profiles through various sections can be
displayed. Quantitative analysis may be done by a procedure known as
stereophotogrammetry. A few applications to ceramics so far were hampered
by contrast limitations 13 or lack of computing power, 14 but the latter problem
probably is less of a concern now.
3-57
Fractography of Ceramics and Glasses
3-58
Tools and Equipment
3-59
Fractography of Ceramics and Glasses
The AFM is a powerful microscope that uses a tiny tip (radius about 2 nm to 20
nm) at the end of a cantilever to probe the surface. The tip is brought into
close proximity to the sample surface and either gently contacts the surface or
taps it while the tip is oscillated close to the cantilever’s resonant frequency.
The probe is held at a fixed location in the X-Y plane and the surface to be
examined is rastered in the X-Y plane. The Z-deflection of the cantilever is
magnified by a laser beam that reflects off the cantilever into a position
sensitive photo detector or interferometer. A computer records the X-Y and Z
displacements and a three-dimensional image of the surface can be constructed.
The AFM is capable of extremely fine resolution (10 pm) and does not require
any special specimen preparation or testing environment, but the specimen
must be very nearly flat and has to be mounted flat on a holder. The AFM is
most frequently used in microelectronic and biological studies. It has been
applied to the study of fatigue crack markings in metals and fracture mirror
roughness measurements in glasses. There are limitations to the AFM. The
scanned area is usually quite small, depending upon the resolution it can be as
small as 1 µm square to as large as 150 µm square. The vertical surface
features can only be a few micrometers in size. Acquisition times may be long.
Color and reflectivity information are lost. Therefore the AFM is almost never
used to search for fracture origins or other telltale fracture markings, but
instead is used for very close up, high-resolution examination of surface
roughness detail or coating evaluations. Hull used it to study the nature of mist
and hackle lines in brittle epoxies.6
3-60
Tools and Equipment
(a)
(b)
Figure 3.50 Matching (a) optical microscope and (b) lateral scanning optical
profilometer images of a fracture mirror in a ground glass rod. (b) shows an
area 1.0 mm tall x 1.3 mm wide. The colors correspond to the scale range
shown on the right (+7.9 µm to -16.3 µm). Wallner lines and mirror mist are
readily detected and displayed. The origin is a surface crack from a scratch.
(b is courtesy of R. Gates)
3-61
Fractography of Ceramics and Glasses
3-62
Tools and Equipment
Origin
(a) (b)
Trace 1, Mirror
Peaks from Wallner lines
(c)
1 µm
50 µm
Trace 2, Mist
Peaks from Wallner lines
(d)
5 µm
40 µm
Figure 3.51 Matching stereoptical (a) and confocal microscope (b) images of
a glass rod fracture mirror. Height profiles for tracks 1 and 2 are shown in (c)
and (d). The major peaks are from Wallner lines. The minor peaks show the
inherent roughness in the mirror and mist regions. (b) and traces are courtesy
of J. López-Cepero Borrego and A. Ramirez de Arellano Lopez, Univ. Seville.)
3-63
Fractography of Ceramics and Glasses
Figure 3.52 Fracture in a glass plate. Fracture originated at the left and
accelerated to the right as attested to by the increased spacing between the
timing ripples. (Courtesy H. Richter)
3-64
Tools and Equipment
TH
Fracture surface
(a) (b)
3-65
Fractography of Ceramics and Glasses
Dye penetration is a staple of nondestructive testing, but it also has value for
fractographic analysis. It is commonly used to search for hidden cracks in a
ceramic part prior to usage or after exposure to some loading or environmental
conditions. Simple staining with common dyes such as the felt tip pens shown
in section 3.9 can aid post-fracture interpretation in porous or coarse-grained
materials that are otherwise very difficult to interpret. The dye can be applied
to the fractured parts. For example, Figure 3.54 shows a portion of a fractured
porcelain toilet tank. Simple staining revealed tiny incomplete crack branches
that gave a clear indication of the direction of crack propagation. Figure 4-30
shows thermal shock cracks highlighted by red dye.
Figure 3.54 Fractured porcelain part with dye penetration by a common felt-
tip pen. The dye reveals crack branching patterns (small arrows) and thus the
direction of crack propagation (large arrow).
3-66
Tools and Equipment
Simple colored staining dyes or felt tip pens may be used with white ceramics.
Level 4 ultrahigh sensitivity fluorescent dyes are used with dark ceramics.
Ultrahigh penetrating fluorescent dyes are needed to get into the tiny cracks.
Special precleaning and soaking steps sometimes are necessary to facilitate dye
penetration. Sometimes it may even be necessary to heat or vacuum treat a part
to eliminate adsorbed gasses on the crack faces prior to dye penetration. After
penetration, ultraviolet or “black lamp” illumination (Figure 3.55) is used to
excite the dye such that it fluoresces and can be seen by the eye. Parts may be
examined under a stereoscope, but the images are dim since the amount of light
emitted is very slight. It is difficult or impossible to photograph features under
these conditions. Compound optical microscopes, on the other hand, may be
used with intense xenon light sources and fluorescent filters so that images may
be captured as shown in Figure 3.56. The photo shows an image made on an
older generation instant developing film that was exposed for several minutes.
Digital cameras would have made a better recorded image. Figure 3.57 shows
a C contact damage crack in a large notched silicon nitride ball prior to and
after a controlled fracture. 23 Fischer et al. 24 have used fluorescent dyes on
flaws in dental ceramics.
3-67
Fractography of Ceramics and Glasses
a b
Figure 3.57 C contact damage crack in a silicon nitride ball. (a) shows the
outer surface and the crack highlighted by fluorescent dye penetrant. (b) shows
one half of the notched ball after it was broken. (c) shows the fracture surface.
(Courtesy T. Lube and R. Danzer, Mountain University Leoben).
3-68
Tools and Equipment
Machining cracks are small and tight, and difficult to dye penetrate. Part of the
problem with them is that there usually is a thin zone (of the order of only a
few micrometers) on the surface that is in residual compression. It clamps the
cracks tight at the surface. Mizuno et al. 25 devised a clever procedure to use a
metallic solution with Pd(NO3)2 impregnation in a vacuum. The solution
normally would not penetrate the as-ground surfaces, but this was overcome by
two methods. One was to polish off the surface compression layer. The
alternative method was to apply a small amount of bending to open up the
cracks for penetration on the tensile surface. The palladium-penetrated cracks
were evident on the fracture surfaces when viewed by SEM, presumably in the
backscattered mode. Kanematsu used this methodology but with an electron
micro probe to study machining cracks by using cold-isostatic pressing to
infiltrate the dye. 26,27,28, He further used plasma etching to highlight the
surface and subsurface damage around grinding damage and indentations.27,29
Figure 3.58 from reference 28 shows matching SEM images and palladium
maps for three different specimens ground with a 100 grit, 75% concentration,
resin-bonded diamond wheel.
(a)
(b)
(c)
3-69
Fractography of Ceramics and Glasses
The Nagaoka University of Technology in Japan has done fine work with novel
dye penetration methods to investigate three-dimensional structures and flaws
in powdered compacts and fired ceramic structures. For example, Professors
Uematsu and Tanaka and students have used potassium iodide and iodine in
diluted water missed with a saturated boric acid solutions to study the
uniformity and lack of uniformity in pressed compacts. 30,31
3.25 Polariscope
(a) (b)
Figure 3.59 (a) shows an old but still very useful polariscope. (b) shows a twin
band (arrows) in a complex single crystal sapphire part. The photo was taken
by a simple digital consumer camera held up to the eyepiece.
3-70
Tools and Equipment
Polariscopes are also valuable with single crystals. Twins may be revealed as
shown in Figure 3.59b.
A common scanner such as shown in Figure 3.60a is handy for converting print
images to digital JPEG or TIFF files.
Figure 3.60b shows a simple ultrasonic cleaner that is helpful for cleaning
small specimens prior to SEM examination. The specimens may be placed into
small beakers with a solvent such as acetone and then placed into the sonic
cleaner bath which is partially filled with water. Methyl alcohol is effective in
removing finger prints. A few minutes vibration is usually sufficient to remove
common contaminants and cutting debris.
(a) (b)
3-71
Fractography of Ceramics and Glasses
Simple cutoff or wafering machines (Figure 3.61) are handy for cutting large
specimens to sizes that can be mounted in optical or scanning electron
microscopes. Cutting should be done wet to flush debris away and to minimize
heating of the specimen.
The merger of the microscope with digital cameras and computers will
continue and virtual imaging with computer-generated topographical images
will become commonplace. Despite all the advantages of the new tools and
advanced technologies, they will never replace direct optical imaging. The
stereoptical microscope will always be a primary tool since it is easy to use and
retains color and reflectivity information and can capture internal reflections.
Most of the new digital technologies emphasize topographic detail to the
exclusion of all else. Very often a key feature will be fleetingly observed
through the microscope eyepieces during the rotation of a specimen or the
angle of lighting.
New computer software and digital camera technologies will enhance the
capabilities of the conventional compound optical microscope so that it
3-72
Tools and Equipment
Perhaps one of the greatest current limitations of the art of image storage and
display is that it is all two-dimensional. Fracture surfaces are by their very
nature three-dimensional, so something is lost when an image of a fracture
surface is captured and stored in a computer’s memory. This is partially solved
by new image analysis computer programs used in conjunction with multiple
SEM images (taken with different tilts) or optical microscopes with Z-axis
rastering such that a computer generated model or depiction of the surface can
be displayed. Such virtual images can be color enhanced and can be tilted,
rotated, or viewed from different angles. Yet, they are all displayed on two-
dimensional computer monitors.
One can imagine the future fractographer donning special eyepieces or a visor
that enable him or her to traverse over or “fly over” a virtual fracture surface.
These will certainly have considerable novelty value, but one wonders whether
they will lead to better diagnoses than those made by an experienced observer
looking through a simple stereoptical microscope. It is very common for the
fractographer, while wiggling and tilting a fracture surface under a stereoptical
microscope, to detect a fleeting reflection or stray subtle fracture line, which
upon further examination unlocks the secret of a particular problem. Will
future computer software and image analysis programs ever match this skill?
Perhaps the great breakthrough will occur when technology is invented that can
capture genuine three-dimensional images that retain full color, reflectivity, and
internal reflection information and display them with full fidelity in three
3-73
Fractography of Ceramics and Glasses
dimensions. The author can personally attest to the hours of desperate struggle
to acquire the best two-dimensional image to capture a critical detail, only to
feel despair when showing the image later and realizing that it “just doesn’t
show what the eyes saw.”
Chapter 3 References
3-74
Tools and Equipment
3-75
General Examination
Fréchette defined the fracture origin as: “that flaw (discontinuity) from which
cracking begins.” 1 A slightly broader definition considers that the fracture
origin is both a location and a flaw:
Even if the primary objective is to find the fracture origin, the overall breakage
pattern should be examined first. Most of this chapter is concerned with
breakage patterns in general. Examination of the fracture surfaces is covered in
Chapter 5. An experienced fractographer may simultaneously examine the
general breakage pattern and the fracture surfaces. Chapter 6 covers the types
of fracture origins that one might find.
Most but not all origins in glass are from surface- or edge-located origins.
Ceramics fracture from volume, surface, or edge sources. The fracture origin
is the site that had the worst combination of tensile stress and flaw severity.
The latter is determined by the flaw size, shape, and thermoelastic or chemical
interactions with the matrix. A large flaw may not necessarily trigger fracture,
especially if it is located in a portion of a part that has low stress or it is very
blunt such as a spherical pore. The severity of a flaw and whether local stresses
are sufficient to trigger fracture are covered by principles of fracture mechanics
presented in Chapter 7. This chapter begins with some generic breakage
patterns and shows some common examples, both for laboratory tested
specimens and component fractures. The end of the chapter has some general
observations about fractures caused by mechanical overload, impact or contact,
thermal, corrosion or oxidation, and some observations on how residual stresses
may affect fracture patterns.
4-1
Fractography of Ceramics and Glasses
The general examination often begins with specimen reassembly. Ideally all
fragments are available. Even if the pieces with the origin are lost, valuable
information can be gleaned from the fragments. With some experience and
with a familiar breakage pattern, irrelevant shards or ordinary secondary
fracture fragments may be ignored. At this point the fractographer should take
care not to introduce contamination to the fragments and especially the fracture
surfaces.
A common concern is whether it is safe to put the pieces together. This step is
discouraged when conducting analysis of polymer or metal fractures. In
contrast, ceramic and glasses are hard and durable materials. The pieces
usually fit perfectly together. It is acceptable to fit pieces together gently,
taking care to avoid abrasion or chipping. Single crystals and glasses are more
sensitive to chipping than polycrystalline ceramics and must be handled more
carefully.
Simple adhesive tapes may be used to temporarily hold pieces together. Apply
the tape sparingly and only on an external surfaces and not between fracture
surfaces. Small strips of masking tape may be used judiciously, but it should
be borne in mind there is a risk of contamination. Whatever tape is used, it
should be a type that is easily removable. Keep in mind that a reassembled
body held together by tape strips may be unwieldy.
a
Duco cement or Scotch “All purpose adhesive, Super strength”.
4-2
General Examination
(“super”) glues may be used since they can set with very thin bonds such that
cumulative misfits are held to a minimum. Such glues can be removed by
soaking in acetone or nail polish remover. In any case, glues should be used
sparingly and every precaution taken to prevent them from coming into contact
with fracture origin areas.
Crack branching and radiating patterns may be interpreted before the fracture
surfaces are examined. Much of the fractographer’s craft is in the observation
and interpretation of these patterns. Crack branch patterns can lead back to an
origin and also tell much about the stress state and stress magnitude. A simple
example is a glass biaxial strength test specimen shown in Figure 4.2. The
radiating pattern is intuitive and leads the observer back the origin which is in
the disk middle. Sometimes multiple fractures can be simultaneously triggered.
Examples are severe thermal shock fractures, or mid-body fractures on each
side of an impacted bottle. Secondary origins are often present in components
that initially break from one origin, but then the fragments themselves fracture.
4-3
Fractography of Ceramics and Glasses
(a) (b)
(c)
Figure 4.1 Transillumination. (a) and (b) are an extracted human tooth from
an approximately 80 year old patient. Transillumination from the side shows
internal enamel cracks. (courtesy S. Scherrer) (c) shows transmitted (through)
illumination of a chemical vapor deposited SiC micro tensile specimen. Very
coarse grains acted as light pipes through the body. Relatively pure SiCs are
green. (d) – (f) show an aluminum-rich inclusion origin, in a highly-filled
resin-matrix dental composite. (d) is a trans-illumination view with light from
below. Although the fracture surface features are blurred by the material’s
translucency, the origin flaw stands out at the chamfered corner. The SEM
images (e) and (f) show the inclusion.
4-4
General Examination
(a)
(b) (c)
Once fracture has commenced, the propagating crack can fork or branch into
multiple propagating cracks. These may be at non-normal angles to the far-
field stress direction. Nevertheless, the initial fracture starts out as a single
radiating crack that separates the part into two pieces. Often the search for the
fracture origin entails studying the branching patterns and tracing them back to
the region of first fracture where the two pieces with the origin join across a flat
surface as shown in the center of Figure 4.2b.
Branching may not occur at all if the failure stresses are low, or if they drop off
in the direction of crack propagation, or if the opening up of the crack makes a
specimen more compliant such that the stresses are relieved.
4-5
Fractography of Ceramics and Glasses
θ θ
origin
(a) (b)
Another very useful pattern is that the angle of forking varies with the stress
state as shown in Figure 4.4. The trend shown by the solid line was reported by
Preston in 1935 for the fracture of glass articles. 2 Uniaxially-stressed parts,
such as direct tension strength or bend bar specimens, fork at about 45o.
Equibiaxially-stressed parts, such as a uniformly-loaded thick window that has
an origin near the center, fork at as much as 180o. Pressurized bottles, which
have circumferential hoop stresses that are double the axial stresses, fork at
about 90o. Torsion fractures, wherein the ratio of the principal stresses is
negative (one stress is tensile, the other is compressive as shown in Fig. 1.2),
have very small forking angles. Preston said that the angles were very
consistent and varied by only a few degrees, but other investigators have
observed much greater variability.
It now appears that systematic variations can occur due to component shape,
the exact loading configuration, the stress level, and possibly the material
properties. For example, the 180o branching angle for equibiaxial tension
(“drum skin tension”) reported by Preston occurs for pressure loaded, thick
windows that fracture at high loads. Shinkai 3 shows such a fracture (his Figure
22a) in a 1.7 m x 0.9 m x 6 mm building window. On the other hand, most
laboratory scale disk specimens tested with ring-on-ring equibiaxial
4-6
General Examination
Figure 4.4 Branching angle versus stress state. The solid line is the trend for
glass shown by Preston.2 The shaded bands show Rice’s data4 ranges for
several ceramics and Bullock and Kaae’s range for a glassy carbon.5 The
author’s data for silicon nitride ring-on-ring disks are also shown.6
loading (Figure 4.2) have smaller angles from 120o to 155o and only approach
180o if the specimen is very strong and has much stored elastic energy at
fracture. Laboratory scale specimen data by Rice 4 on a range of ceramic
materials, Bullock and Kaae 5 on glassy carbon, and the author on silicon
nitride, 6 tend to confirm Preston’s general trend, but with smaller angles. Most
of Bullock and Kaae’s uniaxial angles were 30o. The exact biaxial stress ratio
for Rice’s data for his thermally-shocked and ring-on-ring loaded disks is
unclear and therefore is shown as a band in Figure 4.4. Some of the fractures
were outside the inner loading ring where the hoop and radial stresses were not
the same and thus σ2/σ1 < 1. Some of his data was for ball-on-ring loaded
specimens which were also not equibiaxial unless fracture began at the exact
center. It is also unclear how he defined his angles. They may have been the
angle between any two crack segments or the included angle between the two
crack segments that were the furthest apart. Based on Figure 12 of his paper, it
may have been the former. Nevertheless, Rice made a number of useful
4-7
Fractography of Ceramics and Glasses
Rice suggested that branching patterns tend to split the broken component into
regions of approximately similar area.4 This implies that the cracks partition
the component into pieces with comparable areas and strain energies, a
reasonable supposition. Bullock and Kaae also noted that the branching angles
diminished with progressive branching.5
So although there are discrepancies in the reported data, and the optimum
manner of measuring the angles is unresolved, branching angles provide at least
a qualitative indication of the stress state. There may be material property or
component shape effects. This may be a good topic for a graduate thesis.
4-8
General Examination
(a) (b)
σ Rb = Ab (4.1)
4-9
Fractography of Ceramics and Glasses
A general qualitative assessment about the stresses in a part can be made from
the number of fragments. Quantitative analysis is discussed in Chapter 7.
Low stress, low energy fractures create minimal branching and hence few
fragments. High energy fractures cause extensive fragmentation. The extent
of fragmentation depends upon the stress state throughout the body and the
total energy available for fracture. A small pebble or BB gun shot creates only
localized damage in a window, but a hurled brick will cause window bending in
addition to localized impact damage. Sometimes a general observation about
fragmentation patterns is sufficient for a diagnosis. For example, if a fracture
occurs at an unexpectedly low stress the fractographer might suspect and search
for a gross material, handling, or impact flaw that has weakened a part. Many
thermal breaks are low energy fractures. High local or transient stresses may
precipitate initial crack propagation, but the stress levels may be low in most
other portions of a part. Low energy fractures often have few or no telltale
markings on the fracture surfaces, as discussed in the next chapter.
On the other end of the spectrum, high-energy fracture may be so virulent and
create so many fragments that reconstruction may be impractical. Critical
pieces may be lost. That is not to say fractographic analysis is hopeless.
Sometimes telltale pieces that contain the origin can be culled out of a mass of
4-10
General Examination
fragments and the critical information gleaned while the majority of pieces are
ignored. Analysis of the fracture surfaces can reveal whether fracture was
caused by tension or bending, as will be discussed in the next chapter.
Uneven stress states may be manifested in many ways. For example, Fréchette1
showed an example of a tempered, high-powered lamp that cracked
spontaneously as a result of heating in-service. In the fracture origin area, the
fragments were large, attesting to a low to moderate stress level. Elsewhere in
the lamp, extensive fragmentation and dicing typical of tempered glass was
observed. Fréchette concluded that the intense heating in service caused the
middle of the lamp to lose its temper and weaken such that it was more
susceptible to fracture from thermal stresses.
4-11
Fractography of Ceramics and Glasses
4-12
General Examination
This category includes machined specimens with grip holes or shoulders, glass
optical fibers pulled in tension, or long tubes or rods that are pulled apart.
Direct tension is conceptually one of the simplest loading configurations, but in
practice it can be tricky to achieve. Slight loading misalignments or test
specimen irregularities can create superimposed bending stresses. Breakage
patterns in properly aligned specimens are usually quite simple as shown in
Figure 4.7. Fractures initiate and run perpendicularly to the loading axis until
and if they branch. Moderate-to-high strength specimens may branch with the
telltale 30o to 45o angle as discussed in section 4.3. Fractures at a
nonperpendicular angle to the loading direction (or with compression curls as
discussed in the next section) should be cause for concern and suggest
misalignments in the load train. It is especially important to ascertain whether
fracture origins are located on the specimen surface, in the interior, or at an
edge as will be discussed in the next chapter. Breakages in a group of test
specimens should be randomly distributed within the gage section. An
occasional break near the end of the gage section is probably not too serious,
but if all specimens break from the end, then the tester should review the test
specimen design. The stress concentration at the end of the gage section, the
fixture and grip alignments, and the machining of the blend junction at the end
of the gage section should be checked.
High strength fibers have the added complication that secondary fractures are
common. After initial fracture, the intense stored elastic energy in the fiber
releases, sending stress waves through each broken portion. Reverberations
and fiber whipping cause additional breakages. Fragmentation may be so
thorough that it may be impractical to find a primary fracture. Testing in oil or
coating the fibers with grease may minimize the whipping and secondary
breakage, at the risk of contaminating the primary fracture and altering the
environmental sensitivity of fiber strengths.
4-13
Fractography of Ceramics and Glasses
Flexural strength testing with rod or bar specimens is often done in lieu of
direct tension testing. Flexure is much easier to do. Specimens and testing
equipment cost considerably less. A drawback is that much less volume and
area are exposed to the full tensile stresses. Hence, flexural strengths are
typically greater than direct tension test strengths. Reference 10 is a review of
flexural strength testing of ceramics. For our purpose here, it is convenient to
briefly review the stress state as shown in Figure 4.8. The bending loading
creates a stress distribution such that the maximum tensile stress is on the
bottom as shown in (b) and (c). The stresses diminish into the interior until
they are zero in the middle, at the “neutral axis.” The tensile stresses are
balanced by compression stresses on the opposite side. The maximum stress
exists only directly opposite the middle loading point in three-point loading and
diminishes linearly with distance from the middle loading point to the outer
support points. In four-point flexure, the maximum stress occurs over a larger
region between the two inner loading points.
(c) (c)
Figure 4.8 Three- and four-point flexure test configurations are shown in (a).
The stress distribution on a cross section is illustrated in (b) and (c).
4-14
General Examination
A very telltale feature of bend fractures is the compression curl, also known as
the cantilever curl, shown in Figures 4.9 and 4.10. Once the crack has
propagated from the tensile half of the specimen into the compression side, it
slows down and changes direction. In stronger specimens the crack can branch,
creating a double curl. The existence of a compression curl is an important
sign that the specimen either was loaded primarily in bending or had a strong
(a)
force
(b) (c)
Bovine dentin
Round glass rod rectangular bend bar
(d) (e)
4-15
Fractography of Ceramics and Glasses
Kolsky, 12 using very high speed photography in glass, showed that the
compression curl in beams in bending is formed late in a fracture sequence,
often due to elastic wave reverberations interacting with the slowly moving
crack in its final stages of breakthrough. This may seem counter-intuitive since
the fracture surface shows that fracture very quickly reaches terminal velocity
(if a fracture mirror and hackle form as discussed in the next chapter), but
cracks may in fact slow down very rapidly once they get to the compressively-
loaded half of the beam. Furthermore, Kolsky showed that the local stress state
around the crack is dramatically altered during the fracture event. It is very
much a dynamic situation. Most of this is of secondary importance, however.
The key point here is quite simple: a compression curl indicates a part was
loaded in bending.
Figure 4.10 shows a variety of fracture patterns in bend bars. Low strength
fractures are often the easiest to interpret since the specimen breaks into only
two or three pieces. Often a small “T” shaped piece will be opposite the origin,
but this fragment can be ignored since it never has the origin on it. The origin
is below the “T” fragment on the two matching pieces nearer to the tensile
stressed surface. Sometimes it is handy to keep the “T” fragments to confirm
the other fragments do fit together.
4-16
General Examination
Cushioning material
Higher strength bars have two or more fractures and it must be deduced which
was primary. Secondary fractures are caused by reverberations and stress
reflections after the first fracture has occurred.12,13 Secondary breaks
frequently occur at the loading points. If there are two fractures and one is
4-17
Fractography of Ceramics and Glasses
located at an inner loading point and the other is in the inner gage section, the
latter is probably the primary fracture with the first origin. If the two breaks are
close to the two inner loading points, then the exact locations on the tensile
surface should be examined closely. Often one fracture will be right under a
loading point, whereas the other may be close, but not right at the loading
point. The latter is usually the primary fracture. If both are exactly under the
two inner points, it may be difficult to deduce which was first. There is also the
possibility the specimen might have been loaded unevenly causing stress
concentrations at both sites. If this occurs for more than a few specimens in a
set, then the fixture should be checked as well as the specimen’s parallelism.
An unusually large flaw outside the gage section can trigger a fracture with
crack plane tilted towards an inner loading pin. Specimens with such low
strength fractures usually do not have secondary breaks. Fractures outside the
inner gage section in properly aligned specimens and fixtures are usually
completely valid and a warning that unusually large flaws are present. They
should not be cavalierly dismissed as invalid tests.
(a) (b)
Figure 4.11 A twisted fracture surface in a misaligned alumina bend bar in a
fixture that did not articulate properly. (a) shows the fracture surface, and (b)
the tensile surface of both halves. The principal stress direction (solid black
arrow) was 14 degrees off the expected direction (dotted arrows). Twisted or
angled fracture planes should prompt the tester to check the fixture articulation
and specimen alignment procedures.
4-18
General Examination
low
strength
medium
strength
high
strength
4-19
Fractography of Ceramics and Glasses
The single ball loading schemes have the drawback that only a tiny spot
directly under the loading point experiences the maximum stress. The
concentrated forces of the ball distort the stress field in the vicinity. Fracture
almost always starts directly below the loading ball.
The ball-on-three balls or piston-on-three balls schemes are better suited for
uneven specimens, but the stress state is not very symmetric. Cracks often
align themselves to run between the supporting balls. Excessively thick test
specimens should not be used, since breaking forces are so high that contact
cone cracks can be initiated by the middle loading ball. These disturb the
stress state and, in the worst case, the cone cracks will propagate through the
specimen thickness and reach the tensile side and cause unintended fracture.
This is an invalid test.
4-20
General Examination
(a)
(b)
Outer
rim
Tensile surface
Figure 4.13 A fractured glass ring-on-ring tested disk. (a) shows the overall
breakage pattern. Fracture started on the rim (large arrow) in response to the
uniaxial hoop stress σ (small arrows), propagated towards the middle, and
branched in response due to the greater biaxial stresses in the middle. (b)
shows the fracture surface. The origin is a grinding crack on the outer rim
(large arrow).
4-21
Fractography of Ceramics and Glasses
It is often helpful to look at contact marks or scuff marks where the specimen
was loaded, whether it was in tension, uniaxial or biaxial bending. Such
surface marks are one type of witness marks:
Test specimens are often broken in batches of ten, thirty, or more. Which
specimens should be examined first? The pragmatic answer is that the weakest
ones should be examined first, since they are the easiest to interpret. The
experience gained from these easy specimens facilitates examination of the
more complex fractured specimens, since the fractographer will be better
prepared to recognize primary fractures and distinguish them from secondary
breaks.
Component fractures have the complication that the state of stress or the cause
of failure may be unknown. The general fragmentation pattern can provide
powerful clues. The fractographer first should obtain whatever ancillary
information is possible about the component. What exactly was the material?
What were the circumstances of the fracture? Was it an isolated case or part of
a series? Had some change been made to the material or the loading
conditions? Was there a noise (e.g., shattering sounds or water hammer)?
Were there other parts or evidence found with the fractured material (impact
debris)? Was there a witness? Is the witness honest? Does the witness have a
good memory? Background information such as this may be crucial for the
broader failure analysis. Nevertheless, the fractographer must keep an open
mind, and let the pieces tell their story.
4-22
General Examination
a b
c d
Figure 4.14 Dr. J. Quinn with a window fracture problem at NIST. Evidence
collected at the scene included foreign material attached to the impact site as
shown in (c). The crack patterns indicated impact from the exterior. That, plus
the height of the impact above the ground, the tuft of fur, the time of the year
(November, the rutting season) and similar occurrences elsewhere suggested
the cause of failure shown in (d). The deer saw its reflection in the window and
attacked it!
4-23
Fractography of Ceramics and Glasses
An overall sketch or photo will help the analyst show other fractographers,
engineers, managers, or clients the overall context of the fracture origin or
fracture pattern. Is the fracture from the same location as other parts, is the
pattern of breakage the same, or has a new failure mode been triggered? Much
time and accuracy may be lost in orally explaining or drawing sketches from
memory the fracture patterns days or weeks later. A handy overall photo of the
part can save time and reassure others that the correct fracture origin photos
indeed have been taken. A ruler, magnification marker, or a common object
such as a coin should be placed next to the component when a photo is taken.
Many close-up images of fracture features in the technical literature are
unconvincing or may in fact have nothing to do with a bona fide fracture origin.
4-24
General Examination
(a)
(b)
4-25
Fractography of Ceramics and Glasses
Some common fracture patterns are shown in the following examples. The
reader is referred to the bibliography for more information. Fractographers
who are dealing with specific component failures are strongly advised to search
the literature for prior examples and documentation of similar fracture patterns.
Appendix B of this guide is a compilation of fractographic case studies.
Crack branching creates patterns shown in (a). The branch angle depends upon
the stress state as shown previously in section 4.3. Even with biaxial loading
and branch angles near 180o, a close examination of the branching point will
show which way the crack was moving as it split. Hence the determination of
the crack propagation direction is easy. Intersecting cracks as shown in (b) are
also easy to interpret. The first crack passes through the plate undisturbed. The
second crack approaches and is stopped at the intersection since it is unable to
traverse the previously cleaved material. Intersections are commonly at 90o
since the second crack moves at right angles to tensile stresses and tensile
stresses cannot be carried across the previously cleaved crack. Examination of
the fracture surfaces (to be discussed in the next chapter) will also corroborate
the interpretation, since the fracture surface of the first crack is planar and
undisturbed though the intersection. Wallner lines and hackle lines for the first
crack plane are continuous and connected on each side of the intersection.
4-26
General Examination
The scenario shown in Figure 4.16c is less common, but occurs in cases where
an initial crack does not completely cleave the plate. This can easily occur with
bending fractures wherein the crack leads on one side that is in tension, but
does not necessarily go all the way through on the compression side. The
window in Figure 4.14b has examples. In such cases, the remnant ligament of
unbroken material provides a path for a crossing crack to traverse the first
crack. The key feature to look for is a disturbance in the fracture surface
features. The fracture surface markings (Wallner lines and hackle) on the first
plane are continuous on either side of the intersection. On the other hand, the
second crack has a pronounced change in the markings on its crack plane, often
a jog, or a hook around and up to complete the fracture just on the other side of
the first crack. Sometimes two cracks seem to cross exactly over each other at
exactly 90o, but a very close inspection of the exact intersection will show that
one (the second crack) has a slight offset or step where it intersects the other.
4-27
Fractography of Ceramics and Glasses
Plates and windows can exhibit a wide variety of fracture patterns. Some
common modes are shown in Figure 4.17. An excellent article on the fracture
of flat glass is by Shinkai, reference 3. His article shows subtle variations in
the fracture patterns depending upon whether the plate edges are freely
supported or are built in. Variations also occur with plate thickness. The extent
of residual stress in tempered glass plates may also be estimated from the size
of the diced fragments as will be discussed in 7.2.2. Heat strengthened glasses
also have compressive residual stresses such that the glass is twice as strong as
an annealed plate, but the pieces do not dice into small fragments. Heat
strengthened pieces break into triangular shards like annealed plates.
Blunt objects may create a cone crack that penetrates partially or completely
through the plate. Such a flaw may or may not necessarily constitute a failure,
since the plate is essentially intact. At higher velocities, radial cracks may be
generated from the impact site. Radial cracks also may be triggered by impact
of sharp objects.
4-28
General Examination
Figure 4.17 Plate and window fracture patterns. Figure 4.14b shows an
example of a center impact plate with secondary ring cracks
4-29
Fractography of Ceramics and Glasses
In other cases, the bending forces from the impact may cause the crack (once it
is away from the immediate impact site) to run on the opposite side of the plate
from the direction of impact. At even higher velocities, many radial cracks fan
outward from the impact site. The continued loading of the plate causes the
radially-fractured segments to bend inward, causing them to break in bending
leading to circumferential secondary cracking. In these secondary fractures, the
maximum tension is on the impacted side of the plate. These rings around the
impact often have offsets at the radial cracks, confirming that the radial cracks
occurred first. Figures 4.14b and 4.17 show examples. Sometimes a secondary
ring crack may step across a radial crack if the latter has not completely severed
the plate. Very blunt objects may not necessarily create cracking at the impact
side. Bending forces can create tension opposite the impact side and can
trigger crack growth from flaws on the opposite side. In some instances, the
very blunt object can generate bending forces in the plate that triggers a crack
from a flaw at the plate edge. A crack runs to the impact site, and then radiates
and branches repeatedly outward. In analyzing such fracture sequences, it is
often very useful to combine the overall crack pattern analysis with an
examination of the fracture surfaces, as described in the next chapter. The
fracture surface examination reveals whether the crack was running uniformly
through the thickness or was leading on the inside or outside plate surfaces.
The latter information can establish whether the plate was in bending, and if so,
which side was in tension.
4-30
General Examination
at the origins, and may instead have cone cracks that penetrate from the
surface.
Finding the origin fragments may seem hopeless for extensively fractured
tempered plates if the fragments have fallen out of a frame, but the hexagonal
(or pentagonal or heptagonal) fragments are distinctive and can be found by
simply sifting through the rubble. With persistence and a little luck, the patient
fractographer may be rewarded. Professor J. Varner has singled out origin
pieces from a bucket full of swept up fragments. In some instances, tempered
plates do not break up into uniform fragments. The existence of atypical
fragments is often a sign that the temper was non-uniform either by design
(e.g., Shinkai, Ref. 3, p. 288) or not (e.g., Fréchette, Ref. 1, case 10.18), or
relieved by high temperature exposure (e.g., Fréchette, Ref. 1, case 10.22).
a b
c d
Figure 4.18 Tempered glass fractures in the author’s son’s car. Flying debris
from a tornado in College Park, Maryland in September, 2001 caused the
impact fractures (a, b, d). The arrows show the actual origin in one window
and (c) shows some variations of shape of typical matching origin pieces. The
fracture surface of the origin piece is shown in Figure 5.19b
4-31
Fractography of Ceramics and Glasses
A center-heated plate is one which is heated in the middle while the rim is
cooler. Differential strains put the middle of the plate into compression and the
rim into tension. The plates fracture as shown in Figures 4.19 and 4.20. The
crack starts from an edge origin in tension and initially propagates at 90o to the
edge, but then changes to a meandering wavy pattern as it approaches the
initially biaxial compression stressed middle portions of the plate. The waves
are often periodic as shown in (b). The waviness is an example of Hopf
Figure 4.19 Fractures in center-heated plates. (a) and (b) are low stress
fractures (less than 10 MPa {1,500 psi} in glass), and (c) is a higher stress
fracture such that branching occurred.
(a) (b)
Figure 4.20 Thermal fractures with origin marked by arrows. (a) shows an
alumina furnace baffle plate. Fracture started from grinding cracks associated
with a hole. (b) shows glass slides broken by simply heating them on a hot
plate.
4-32
General Examination
bifurcation problem that can occur in nonlinear systems. 23,24 Figure 4.20
shows a thermal failure in an alumina plate. Case 6 in Chapter 10 shows
another example in a SiC furnace plate. Fréchette1 noted that meandering
cracks are not proof of thermal stresses and they can also form in impact cases
in the final stages of breakup.
Thermal stress fractures often have localized tensile stresses that start fracture,
but diminish away from the origin site. Branching may be minimal and the
fracture surfaces may be relatively featureless and flat in regions away from the
origin. Wallner lines or hackle lines (discussed in the next chapter) are rare. On
the other hand, there may be multiple arrest lines attesting to stepwise
propagation of the crack. Scarps may be present on the initial parts of the
fracture surface if the part was thermally shocked by a cooler fluid.
The relatively featureless fracture surfaces mean that the crack is moving
slowly when the wavy pattern forms. The crack alters the stress distribution and
it in turn alters the crack path. This nonlinear interaction causes the crack to
curve back and forth.
4-33
Fractography of Ceramics and Glasses
a b c d e
f g h i j
Figure 4.21 Bottle fracture patterns. (a-c) sidewall fractures due to internal
pressure at progressively greater pressures; (d) internal pressure fracture
starting from the base; (e) is a thermal fracture from sudden cooling of the
base; (f) is an impact fracture on the right side (labeled 1) with a hinge
fracture on the side (labeled 2); (g-i) show water hammer fractures and (j)
shows a fracture from diametral rim clamping.
4-34
General Examination
Sometimes impact damage sites may be found around the base, either on the
inside or the outside. Impact damage sites on the inside of drinking glasses or
beer mugs are often due to careless dropping of eating utensils into the glasses.
External impact fractures on the side walls often have a starburst pattern at the
impact site as shown in Figure 4.21f and 4.22. The principal crack system does
not necessarily start at the impact site, however. The force of impact can cause
the sidewall of the vessel to flex outwards creating a bending stress in the side
of the vessel. This can trigger so-called “hinge fractures” 45o to the side of the
impact site. Wallner lines indicate that fracture starts on the outside surface of
the hinge fracture sites. A network of cracks from the impact site can start on
either the inside or outside of the impact site, depending upon the sharpness of
the impactor. The impact site may create the initial fractures, but in many cases
the hinge origins fracture first. They then send a leader crack over to the
impact site that triggers or exacerbates the fragmentation at the original impact
site. Figure 4.22 shows an example of this in a glass tube.
Filled glass containers dropped on their side trigger fracture from origins on the
inside surface.
4-35
Fractography of Ceramics and Glasses
A B A C B
(a) (b)
C
B
(c)
Figure 4.22 Small glass tube loaded diametrally at opposing points A (only
the front is shown). Fracture started at the side hinge point B which generated
an array of radiating cracks. One of these is the “leader crack” C (arrows in b
and c) which went to the loading site A where it branched repeatedly and
fanned out yet again. (Tube courtesy of W. Haller.)
4-36
General Examination
(a)
Label
removed Origin
and vial
reoriented
Fracture Outer
Inner wall Outer Inner wall surface surface
walls
Origin
(e) (f)
Origin
Figure 4.23 Pressurized thin-walled vessels. (a) shows that the side walls are
stressed in biaxial tension. Hoop stress is double the axial stress and therefore
the first fracture is usually vertical as shown in the medicinal vial shown in
(b)–(f). (e) shows the matching fracture surfaces at the origin. There is a
distinct fracture mirror centered on the origin. (f) shows the origin was an arc
shaped crack in the outer wall, a clear indication of blunt contact damage.
4-37
Fractography of Ceramics and Glasses
Imagine making a vertical cut through the middle of the bottle with height L,
radius r, and wall thickness t, under pressure P. This is a common practice for
mechanical engineers to do when they construct “free body diagrams” to
analyze stress problems. A part is cut in an imaginary sense. The external
forces and moments acting on one half of the component are balanced with
internal stresses in the walls. For the bottle, the forces from the sideways
expanding pressures that act on the arc of the inside surface are balanced
against the constraining forces in the side walls. These are the sum of the hoop
stresses in the two side walls that have cross-sectional area 2Lt:
𝜋𝜋�
2 (4.2)
� 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑟𝑟𝑃𝑃𝑃𝑃𝑃𝑃 = 2𝑟𝑟𝑃𝑃𝑃𝑃 = 2𝑃𝑃𝜋𝜋𝜎𝜎ℎ𝑜𝑜𝑜𝑜𝑜𝑜
−𝜋𝜋�
2
𝑃𝑃𝑟𝑟
𝜎𝜎ℎ𝑜𝑜𝑜𝑜𝑜𝑜 =
𝜋𝜋 (4.3)
For the axial stress (vertical direction), imagine splitting the vessel in a
horizontal plane midway along its height. The force created by the sum of the
pressures on the bottom circular surface of area πr2 is balanced by the sum of
the vertical axial stresses in the vessel side walls which have a total cross
sectional area of 2πrt:
4-38
General Examination
τ σ2
σ1
(a) (b)
4-39
Fractography of Ceramics and Glasses
4.15 Chipping
Concentrated loads near an edge can chip off a portion of the body. Figure
4.25 shows some examples. Chipping can either be a primary mode of fracture,
or it can be a secondary nuisance. For example, it is one of the leading causes
of failure of all-ceramic dental restorations such as crowns or bridges, and
hardmetal (e.g., tungsten carbide) or ceramic cutting tools. On the other hand,
it may be a nuisance if there is so much chipping that it interferes with
interpretation of a fracture from some other cause. Thus, it is important to be
careful when fitting pieces together during an initial examination.
The force to cause fracture depends upon the shape of the object that applies
the contacting force, the distance from the edge, the angle of the applied force
P, the angle of the edge, and the material’s fracture toughness. In general, the
greater the distance away from an edge, the greater the force that is necessary
to create a chip. Modern use of this test for engineering materials was started
in 1986 at the National Physical Laboratory in United Kingdom by Almond and
McCormick. 28 A remarkable finding was that the shape of the chip was
invariant with material. Edge chipping is a field of growing attention, and there
is a growing literature on it. Quite interestingly, Charles de Freminville’s
seminal 1907 treatise on fracture of brittle materials included excellent
illustrations (Fig. 5) of edge chips in glass. 29 Several papers on edge chipping
were given in a session at the 2000 Alfred Conference on Fractography of
Ceramics and Glasses. 30 Edge chipping also has been a topic in the lithic
fracture literature and flint knappers have a good practical knowledge on the
topic (e.g., Refs. 11, 31).
The exact force-distance trend depends upon the type of indenter (sharp or
blunt). In a laboratory setting, sharp diamond indenters (Rockwell, Vickers,
Knoop) or sharp conical tools are used, but in practice any contact type loading
can create a chip at an edge. Lab-scale indenters initiate cracks around the
contact site which then grow to form a chip. On the other hand, some loadings
with blunt contactors near an edge cause a pre-existing crack to propagate
parallel to the specimen side and to form a long flake. The top of the flake,
where the force is applied, is called the “platform” in the lithics literature. The
fracture mechanics of chip formation and how long slender flakes may be
formed by stable crack extension are fascinating. 31,32 The long slender flakes
can be used as cutting tools. Cotterell makes the case that the ability to form
such long slender stone tools was a critical step in the evolution of mankind.31
4-40
General Examination
(a) (b)
(c) (d)
(e) (f)
Figure 4.25 Examples of chip fractures. (a) is a schematic that shows how a
chip is formed and the telltale shape. (b) shows a top view of a chip that is just
about to pop off a lithium disilicate foundation dental glass-ceramic.
The loading point is obvious. (c) shows a row of edge chips created in
porcelain under controlled conditions in the laboratory. (d) is a close-up. (e)
and (f) show edge chips in an ancient Roman (circa 300 AD) 22 cm square x
3.4 cm thick hypocaust stone from Trier, Germany. (b,c,d, courtesy J. Quinn)
4-41
Fractography of Ceramics and Glasses
The shape of chips varies with the angle of the applied load as shown in Figure
4.26. 33 This has practical significance, since the shape of the chip can tell a
fractographer the direction of the force that caused the chip. Since the first
edition of this book was written, I have followed up on the edge chipping work
of my departed spouse Dr. Janet Quinn. We have published a number of papers
on edge chipping of dental ceramic and composite restorative materials and
structural engineering ceramics and we have a new quantitative model.34,35,36
Early work emphasized a linear relationship between force and distance the
(a)
(b)
Figure 4.26 The angle of the applied load affects the shape of edge chips. (a)
is a schematic that shows the patterns and (b) shows front views of two chips in
glass (at different loads and distances from the edge) made by a conical
indenter tool angled away from the edge at 105o. (after J. Quinn and R.
Mohan, Ref. 33).
4-42
General Examination
force is applied away from the edge, but new results indicate nonlinear trends
can result from the partitioning of indenter energy into competing deformation
and fracture modes in the material. Figures 4.27 and 4.28 show some of the
new chips.
(a)
(b) (c)
Figure 4.27 Edge chipping. (a) shows the test machine with a 120o sharp
diamond scriber making a chip in a dental stone material. (b) and (c) show a
ceramic ball indenter chipping brittle denture material tooth specimens.
(a) (b)
(c) (d)
Figure 4.28 Edge chips formed in (a) dental lithium disilicate glass ceramic;
(b) dental 3Y-TZP zirconia; (c) dense alumina; and (d) laminated
alumina/alumina-zirconia ceramic. The chips were stained with a green felt tip
pen to make the fracture surfaces easier to see in these translucent materials.
4-43
Fractography of Ceramics and Glasses
Edge chips are very common as secondary fractures on broken ceramic or glass
fragments. As secondary fractures, they may have had nothing to do with the
original cause of fracture. They easily occur if fragments bump into each other
or impact other objects during breakage or subsequent handling. If there is any
doubt as to whether an edge chip is a secondary or a primary fracture, compare
the matching fracture halves. If the chip is only on one half, then it is a
secondary fracture. It is quite common, especially with glasses, to have one
fracture half rub against the edge of the matching fracture half, causing chips in
the latter. Undesirable secondary fracture chips can occur in bend bar strength
specimens and other test coupons if the pieces fly around and impact the
fixtures after the first break. Often the tensile edge of a bend bar will hit first,
and the genuine fracture origin may be chipped off! Soft cushioning material
should be placed in the fixtures under the bars as discussed in section 4.6.
Some additional means to prevent the fragments from flying about should also
be used (e.g., cotton or rolled paper or tape pieces).
The author used this approach to aid the interpreting of broken large (≈1 m)
sintered porous fused-silica missile radomes in the mid-1980s. This is Case 7
discussed in Chapter 10. Only a few broken components were available and
an initial examination of the weak material with its rough fracture surfaces was
unsuccessful. Examination of bend bar and double torsion fracture toughness
specimens gave the author more experience in interpreting crack propagation
directions. Optimum examination conditions for detecting hackle and Wallner
lines were discovered. This know-how was then applied to the full size
components, with the result that fracture surfaces which initially confounded
the author became tractable on the reexamination. The origins were found.
4-44
General Examination
al. 37 and Richter 38 have documented two superb examples of this approach for
ceramic hip joint balls. Morrell et al.37 supplemented their component testing
with biaxial disk and bend bar fracture testing.
B
(a)
A
(c) Rubber
insert
(b)
Zinc phosphate
cement
origin margin
Figure 4.29 Alumina molar crown (clinical 33) which fractured after 6 years
in the patient’s mouth. (a) shows an occlusal surface (top) view. (b) shows a
side view and the fracture surface of piece A. Careful examination showed the
origin was on the margin on the left, and the fracture ran up through the crown
and to the other side, splitting it in half. (c) shows the laboratory test set up
which produced nearly identical fracture patterns. The epoxy base simulated
the dentin of the reduced human tooth. (courtesy M. Øilo)
4-45
Fractography of Ceramics and Glasses
Another dental example, but on an actual human tooth, was by Dr. J. Quinn. 40
She mimicked a fracture of an upper right canine tooth that broke at the root
several days after the patient had heart surgery. It was thought that blunt force
damage during intubation by an anesthesiologist had caused damage that led to
the fracture. Fractographic analysis showed that fracture started from the
lingual (tongue) side and caused a curved fracture plane from cantilever
loading. Dr. Quinn obtained a comparable tooth from an extraction (from
another patient) and loaded it in cantilever loading in a laboratory setting. It
broke in an almost identical manner. This, plus tooth decay in the original
tooth, evidence of step wise crack growth, and evidence of bruxism, showed
the tooth was damaged before the surgery. The anesthesiologist was not at
fault. Nevertheless, it was recommended that mouth guards (commonly used
by bruxers when sleeping) ought to be used during intubation for surgery as
well.
Stress solutions may be straightforward for simple shapes and loadings, but
many components have complex shapes and asymmetrical loadings. Closed-
form solutions for the stress state may not be available. Finite element analysis
(FEA) is a valuable tool for such problems. FEA models partition a body into
small elements and uses computer analysis to obtain the stresses and strains in
each using the theory of elasticity. In principle, the entire stress distribution in
a body can be analyzed, provided that the loading conditions are accurately
simulated. Software for FEA is becoming increasingly easy to use, but some
skill is needed in setting up the problem, applying a good mesh, and identifying
the correct boundary conditions. Richter utilized FEA results to help interpret
the ceramic hip ball fractures.38
FEA analysis can be a case of “garbage in and garbage out.” Modelers may be
so enamored of the models with their assumed loading conditions and antici-
pated fracture modes that they are unreceptive to obvious evidence that a
component is not fracturing “the way it is supposed to break.” A good
4-46
General Examination
modeler should be flexible and not afraid to modify or adapt the simulation
based on the fractographic findings. Reference 41 reviews several case studies
and compared model predictions to actual fracture modes.
Thermal stresses are created by differential strains created when portions of the
body try to stretch or contract, but are constrained by other portions of the body
at different temperatures. Thermal stresses may either be steady state (invariant
with time) or time dependent. Gradual or linear spatial temperature gradients
usually do not generate thermal stresses. Two types of thermal stress cases
occur: (a) components with stresses arising from overall temperature gradients
between portions of the body, and (b), surface and interior stresses caused by
through-thickness temperature gradients.
4-47
Fractography of Ceramics and Glasses
Thermal stresses and strains occur during transient cool down or heat up
events. Sudden temperature changes are called “thermal shock.” Thermal
shock can cause strains between portions of a body or through-thickness
stresses. Ceramics and glasses are most susceptible to sudden cool down
thermal stresses, since tensile stresses are created at the surface. The surface
layer of the part cools quickly and tries to contract, but the hotter interior
portion prevents the contraction thereby creating tensile strains and stresses in
the outer layer. Compressive stresses are generated in the interior. Suddenly
cooling the surface with a fluid such as water can generate substantial tensile
stresses and fracture since the fluid promotes rapid heat transfer from the
surface. Some procedures, such as laboratory quenching of hot bend bars into
water, create sharp stress gradients with a biaxial stress state at the surface.
This leads to the formation of a two-dimensional network of surface cracks that
may not necessarily penetrate completely into the interior. Examples are
shown in Figure 4-30. Shao et al. 42 showed some images showing how
cracking patterns in alumina plates changed with increasing temperature shock.
At low shock, a few looping cracks formed. With greater shock, a dense
cellular structure formed. With the maximum shock, a cellular structure with
enveloping larger cracks appeared.
(a) (b)
Figure 4.30 Thermal shock cooldown cracks in alumina bend bars (a), and a
disk (b). The cracks are highlighted by red dye penetrant. (courtesy of R.
Danzer and T. Lube, Mountain Univ. Leoben)
4-48
General Examination
Impacts may create localized damage sites that can become the origin of
fracture if the impactor is small or sharp, but can also cause fracture by
imposing bending stresses on a structure such that fracture initiates on the side
opposite the impact site or elsewhere in the body. Section 4.13 shows both
scenarios. Sharp and blunt impacting objects create different types of localized
damage origins as shown in sections 6.7.4 and 6.7.5. When an impact or
contact fracture is suspected, the surface should be carefully examined for
“witness marks” as discussed in section 4.6.4.
Oxidation and corrosion can weaken a part and leave it susceptible to fracture.
Telltale surface reactions are often readily apparent on the surface in the form
of numerous pits, discoloration, and surface scale. Sometimes the damage is
confined to small localized pits, bubbles, or blisters as shown in Figure 6.26. In
some difficult to interpret cases, the reactions are localized along grain
boundaries and can only be discerned with the SEM. Chemical analysis can
show cation migration to or from a reaction site. It can also indicate
contamination or reaction with the environment or furnace linings or insulation.
Richerson 43 shows several examples of corrosion-oxidation damaged silicon
nitrides.
Residual stresses can arise from many sources. They can be intentional as in
tempered or heat-strengthened glass or unintentional as in ceramics sintered in
furnaces with thermal gradients. Residual stresses can exist at different scales
within a structure. It is convenient to differentiate these as: bulk, through the
thickness, localized, or microstructural residual stresses.
There may be bulk residual stresses in a part whereby one portion or side is
straining against other portions due to differential contractions during cooling.
They can even cause spontaneous fractures in the absence of externally applied
stresses. For example, there are reported cases of improperly-sintered porcelain
toilets that fractured spontaneously leading to catastrophic house flooding.
Residual stresses may also exist through the thickness of a part, as in heat-
strengthened or tempered glass plates. Residual stresses may be very localized,
such as around a sharp impact or contact damage site (e.g., a sharp particle
impact site or Vickers indentation). Shallow residual stresses are also common
4-49
Fractography of Ceramics and Glasses
on machined and ground surfaces. They typically are compressive within a few
micrometers of the outer surface and change to tensile further beneath the
surface. As such they can affect origins located on the surface. Finally,
residual stresses may exist at a microstructural level, such as at the grain
boundaries between grains or between two phases in polycrystalline materials,
or between an inclusion and the matrix. The inclusion may have very different
thermal contraction or elastic properties than the matrix. Thus, it is simplistic
to say: “the piece had residual stresses.” One must be more specific and
recognize that there may be multiple types of residual stress in a part. Further-
more, the existence of a residual stress may not necessarily be harmful. Some
are compressive and some are tensile. Where an origin is located and the type
of residual stress it is exposed to are important factors to consider. Crack
propagation and branching may be dramatically affected as a crack moves from
one type of stress field into another (e.g., compressive to tensile). Residual
stresses may be highly variable within a part. Reference 44 is an interesting
new paper on the variability of residual stresses in air-jet tempered glass plates.
Reference 45 has other examples of nonuniformly-tempered glass including
one illustration of a large automobile windshield. Uneven fragmentation and
crack patterns are telltale signs of nonuniform tempering.
The fractographer may be tipped off to the existence of residual stresses by the
overall fracture patterns as shown in this chapter. A sudden change in crack
propagation direction may be another sign. They can also be indicated by
abrupt, seemingly spontaneous fractures when a component is being cut or
machined. They may also be signaled by Wallner line markings on the fracture
surfaces as shown in the next chapter. Distortions in the size and shape of a
fracture mirror can be evidence of residual stresses. Analysis of flaw sizes can
also be used to estimate residual stresses as shown in section 7.14.
4-50
General Examination
Delayed fracture can occur due to slow crack growth from preexistent flaws.
The loading conditions and stresses may initially be insufficient to cause
breakage, but a flaw may slowly grow when it is under load until it reaches a
critical condition and then triggers breakage. The rate of crack growth is very
sensitive to the stresses and stress intensities acting on a flaw, and growth can
be very slow or rather fast. For constant stress, cracks accelerate with time in
accordance with fracture mechanics analyses shown in section 7.10. There
may be little or no warning of the imminent failure since the extent the crack
grows to become critical may be very small. Sometimes the slow crack
growth leaves telltale markings on a fracture surface as shown in section 5.9.
4-51
Fractography of Ceramics and Glasses
σ σ
σt
(a) (b)
σt
σ σ
(c)
4-52
General Examination
prone to these problems, and often fractures occur due to end splitting as shown
in Figure 4.31a. Such results are invalid. Dumbbell tapered specimens are
preferred, as shown in Figure 4.31b,c. A properly-tested specimen fracture is
shown in Fig 4.31c. The extraordinary fragmentation of the gage section is
notable. There is no point in looking for a single fracture origin. The goal of a
fractographic analysis is more to verify that a test was done correctly, and to
identify the mechanism of compression fracture as described in the next
paragraphs.
Figure 4-32c shows several alternative modes of fracture that do not necessarily
require initial flaws. Microcracks can be generated in a microstructure by
several mechanisms. 53,54 Grain boundary microcracks (on the left in Figure 4-
32c) can be initiated due to elastic property mismatches between the adjoining
grains that have different orientations. The elastic moduli and crystal constants
can be dramatically different on different orientations, especially in non-cubic
materials. Furthermore, strong residual stresses can exist along grain bound-
aries due to differential shrinkage between adjoining grains with different
orientations. The middle of Figure 4-32c shows two grains with different
orientations and moduli and, as vertical force is applied, the lateral strains can
4-53
Fractography of Ceramics and Glasses
be very different, causing the nucleation of wedging cracks in one or the other
grain. On the right of the Figure 4-32c, twins or dislocations within grains can
initiate microcracks. Usually there are not enough independent slip systems
active at low temperatures to lead to generalized plasticity, but the very large
compression stresses can create sufficient dislocation activity and buildups to
form microcracks within grains or at grain boundaries.
Chapter 4 References
4-54
General Examination
16. J. A. Salem and L. Powers, “Guidelines for the Testing of Plates,” Ceram. Eng.
Sci. Proc., 24 [4] (2003) 357-364.
17. D. Maccauley, Cathedral: The Story of Its Construction, Houghton Mifflin, New
York, 1981.
18. D. Maccauley, City: A Story of Roman Planning and Construction, Houghton
Mifflin, New York, 1974.
19. D. G. Holloway, “The Fracture Behavior of Glass,” Glass Technol., 27 [4] (1986)
120 -133.
20. T. A. Michalske and E. R. Fuller, Jr., “Closure and Repropagation of Healed
Cracks in Silicate Glasses,” J. Am. Ceram. Soc., 68 [11] (1985) 586-590.
21. S. M. Wiederhorn and P. R. Townsend, “Crack Healing in Glass,” J. Am. Ceram.
Soc., 53 [9] (1970) 486 – 490.
22. B. Stavrinidis and D. G. Holloway, “Crack Healing in Glass,” Phys. Chem.
Glasses, 24 [1] (1983) 19-25.
23. A. Yuse and M. Sano, “Instabilities of Quasi-static Crack Patterns in Quenched
Glass Plates,” Physica D, 108 (1997) 365-378.
24. A. Yuse and M. Sano, “Transition Between Crack Patterns in Quenched Glass
Plates,” Nature, 362, 25 March 1993, 329-331.
25. J. B. Kepple and J. S. Wasylak, “Fracture of Glass Containers,” pp. 207 – 252 in
Fractography of Glass, ed. R. C. Bradt and R. E. Tressler, Plenum, NY, 1994.
26. F. W. Preston, “Bottle Breakage – Causes and Types of Fractures,” Bull. Am.
Ceram. Soc., 18 [2] (1939) 35 -60.
27. T. A. Michalske, “Quantitative Fracture Surface Analysis,” pp. 652 – 662 in
Engineered Materials Handbook Ceramics and Glasses, Volume 4, ed. S.
Schneider, ASM Int., Metals Park, OH 1991.
28. E. A. Almond and N. J. McCormick, “Constant-Geometry Edge-Flaking of Brittle
Materials,” Nature, 321 [6065] (1986) 53-55.
29. Ch De Freminville, “Caractères Des Vibrations Accompagnant Le Choc Dé
L’Examen Des Cassures,” Review de Mettalugie, 4 (1907) 833-884.
30. Fractography of Glasses and Ceramics IV, eds. J. R. Varner and G. D. Quinn,
Ceramic Transactions, Vol. 122, Amer. Ceram. Society, Westerville, OH, 2001.
31. B. Cotterall, Fracture and Life, Imperial College Press, London, 2010.
32. M. D. Thouless, A. G. Evans, M. F. Ashby, and J. W. Hutchinson, “The Edge
Cracking and Spalling of Brittle Plates,” Acta Metall., 35 [6] (1987) 1333-1341.
33. J. B. Quinn and R. Mohan, “Geometry of Edge Chips Formed at Different
Angles,” Ceram. Eng. Sci. Proc., 26 [2] (2005) 85 – 92.
34. G. D. Quinn and J. B. Quinn, “A New Analysis of the Edge Chipping Resistance
of Brittle Materials,” Ceram. Eng. Sci. Proc., 35 [1] (2015) 209 -224.
35. G. D. Quinn, A. A. Giuseppetti, and K. H. Hoffman, “Chipping Fracture
Resistance of Dental CAD/CAM Restorative Materials: Part 1, Procedures and
Results,” Dent. Mater., 30 (2014) e99 – e111.
36. G. D. Quinn, A. A. Giuseppetti, and K. H. Hoffman, “Chipping Fracture
Resistance of Dental CAD/CAM Restorative Materials: Part 2, Phenomenological
Model and the Effect of Indenter Type,” Dent. Mater., 30 (2014) e112 - e123.
4-55
Fractography of Ceramics and Glasses
4-56
Fracture Surface Examination
Fracture surfaces have a wealth of information that may help determine the
cause of fracture and even the fracture origin itself. The fracture surfaces
should have an initial examination to look for any signs of contamination, and
they should be cleaned as discussed in section 3.1 if necessary. There are two
important cases where it may be wise to examine fracture surfaces before
cleaning. First, surface contaminants may provide clues about the fracture
event and may not be merely contamination picked up afterwards from
handling. Figure 10.26e in Chapter 10 is a good example where discoloration
at the origin was an important clue to understanding the initial flaw in a broken
dental crown. Second, fine debris on a fracture surface may be a clue that
cyclic fatigue may have occurred, if the debris is crushed material that was
dislodged from the microstructure. This is covered in section 5.9 on page 5-84.
The origin may range from a type and location anticipated by the fractographer
(as in a broken bend bar) to an unexpected type from an unexpected location or
cause (as in a broken component). With familiar laboratory strength test
specimens, the fractographer may proceed directly to the primary fracture
surface and quickly locate the origin area. Usually the fractographer seeks a
fracture mirror, a key feature centered on the origin. Figure 5.1 shows two
examples. The origins are often very small and magnification by a microscope
is necessary. Stereoptical microscopes are ideal for initial fracture surface
examinations. Sometimes this examination may be all that is needed for a full
characterization of the origin. In other instances, the flaw at the origin is so
small that scanning electron microscopy is needed.
5-1
Fractography of Ceramics and Glasses
The greater the stress in the part at fracture, the more the stored energy, and the
richer the fracture markings. Weak parts with low stored energy are often
difficult to interpret. The part breaks into two pieces with relatively featureless
fracture surfaces. Very porous or coarse-grained materials also can mask
fracture surface fractographic markings. On the other hand, very strong
Compression curl
(a)
Fracture
mirror
Origin
Compression curl
(b)
5-2
Fracture Surface Examination
components such as some glass optical fibers may have so much stored elastic
energy at fracture that excessive fragmentation occurs and the origin cannot be
found.
“Conchoidal fracture” and “cleavage fracture” are terms commonly used in the
geological community. Cleavage fractures are smooth and planar fractures
such as those on preferred breakage planes in single crystals. Conchoidal
fractures do not follow natural planes of separation and often are curved, in a
manner reminiscent of curved seashells. Conchoidal is derived from the Greek
word for a mussel shell. Glass and many ceramic fractures are conchoidal.
“Joint” is a common term in the geological community and means a fracture
surface. Joints are defined as a fracture or crack in a rock mass along which no
appreciable movement has occurred. “Parting” is another geological term
describing flat fractures in a mineral, but is not necessarily along preferred
cleavage planes. Parting is sometimes associated with twinning.
The fracture mirror is the region where a crack radiates outwards from a flaw at
5-3
Fractography of Ceramics and Glasses
(a)
(b)
(c)
Figure 5.2 Three views of a fracture mirror in fused silica glass rod broken in
bending at a high stress level (96.2 MPa, 13.9 ksi). (a) shows the entire
fracture surface. (b) is a close-up of the mirror, and (c) is the same region but
with the illumination adjusted. Maximum tension was at bottom center in view
(a). The mirror is asymmetric due to the bending stress field.
5-4
Fracture Surface Examination
(a)
(b)
mist mirror
hackle
Figure 5.3 Two views of a fracture mirror in a moderate strength (76.2 MPa,
11.0 ksi) fused silica glass rod broken in bending. The mirror is incomplete
since it was large relative to the cross section size. The advancing crack
propagated into a decreasing stress field in the interior and did not have
sufficient energy to form the mist and hackle in a complete circle. If this rod
had been loaded in direct tension instead of bending, the mirror would have
been fully formed and semicircular. The incomplete mirror in the direction of
the interior is a telltale sign of bending.
5-5
Fractography of Ceramics and Glasses
the fracture origin. The crack accelerates from near zero velocity to terminal
velocity within the mirror region. Using ultrasonic stress wave fractography on
soda lime silica specimens, Richter and Kerkhof 2 and Field 3 showed that a
crack accelerates to between one-half to the full terminal velocity (≈ 1500 m/s)
by the time the mirror-mist markings formed. This remarkable acceleration
over a very short distance occurs in microseconds. Smekal 4 arrived at the same
conclusion from his analysis of the Wallner lines (described later in this
chapter) inside fracture mirrors.
The terminal velocity in glasses ranges from 700 m/s to 2500 m/s depending
upon composition and elastic properties. Terminal velocities are greater for
ceramics. Table 5.1 lists some values. The terminal velocity is about 0.5 to 0.6
times the Raleigh surface wave velocity in an elastic body, or about 0.38 times
the longitudinal (dilatational) stress wave velocity. The Raleigh wave velocity
is less than the longitudinal wave velocity, vl = √E/ρ, where E is the elastic
modulus and ρ is the density. The terminal velocity of the crack is limited by
the speed that elastic energy is transferred to the crack tip. In principle, the
crack can travel as fast as the Raleigh wave speed, but in practice the local
deviations of the crack front that create mist and hackle limit the crack speed.
Some atypical shock loading cases can drive a crack faster than the normal
terminal velocity provided that the crack does not develop mist, hackle, or
branching. Maximum crack velocities in single crystals whereby mist and
hackle formation are suppressed can reach a greater fraction of the Raleigh
wave speed than glass and polycrystalline ceramics. In some ballistic impact
conditions, a damage wave of cracking in glass can proceed at velocities 80 %
to 90 % of the longitudinal wave speed. 5,7 High speed photography revealed
that individual crack segments traveled no more than the conventional terminal
velocity of 1480 m/s, but that secondary cracks nucleated right behind the
shock wave.
If the stresses acting on the advancing crack decrease, the crack can decelerate
and even stop. This is not unusual in thermal shock cases where stress
gradients can be large.
5-6
Fracture Surface Examination
Maximum Crack
Material Ref.
Velocity (m/s)
Soda lime glass 1510 8
Soda lime glass 1580 3
Soda lime glass 1540 12
Soda lime glass 1600 9
Soda lime glass 1500 11
Soda lime glass 1460 7
Various flint glasses 750 - 1400 8
Heavy flint glasses 700 - 1095 9
Mirror glass 1520 9
Borosilicate crown glass (BK-7) 1677 9
Borosilicate crown glass (BK-7) 1800 4
Fused silica 2100 4
Fused silica 2500 11
Fused silica 2155 8
Reaction bonded Si3N4 (AME) 2950 10
Hot pressed Si3N4 (HS 130) 3500 10
KCl, single crystal 700 3
LiF, single crystal 1700 3
MgO, polycrystalline (85% dense, 15 µm) 1800 9
MgO, single crystal, cleavage {100} 5100 3
Diamond, cleavage {111} 7200 3
Al2O3, single crystal sapphire, cleavage on {10Ī0} 4500 3
Al2O3, polycrystalline (Lucalox 15 µm) 2000 10
Tungsten 2200 3
Tool Steel 1700 10
Resin Plastic 420 8
PMMA 770 10
Table 5.1 Terminal crack velocities from Field (3), Smekal (4), Winkler et al.
(7), Schardin et al. (8), Kerkhof (9), Congleton and Denton (10), Chaudhri
( 11), and Barstow and Edgarton (12)
The underlying principal that accounts for the micro and macro branching has
been attributed to the crack reaching a critical velocity, 13 a critical energy
level, 14 a critical stress intensity, 15,16,17,18 or a critical strain intensity. 19 A
velocity criterion for crack branching was discounted by data shown by
Congleton and Petch.16 Richter and Kerkhof,2 Field,3 and Döll 20 used
ultrasonic fractography to show that cracks approached terminal velocity before
the formation of the mist boundary and there was no pronounced change in
velocity when the mirror formed. Kirchner et al.17,18 showed convincing
evidence that a stress intensity criterion accounts for the mirror sizes and
5-7
Fractography of Ceramics and Glasses
Careful electron microscopy and atomic force microscopy now have shown that
the formation of the mirror is a gradual progression of very localized crack path
deviations from the main plane. Figures 5.4 a-e show superb transmission
electron microscope images of various regions in a glass fracture mirror
collected by Beauchamp. 22,23 Poncelet 24 showed comparable images in 1958,
but Gölz may have been the earliest with extraordinary electron microscope
photos published in 1943. 25 Rounded ridges that are elongated in the direction
of crack propagation gradually coarsen in amplitude. At some point, they
develop slight hackle steps where over running and under running fingerlike
crack segments link. This point, where nano-micro scale hackle lines begin to
form, could be an important transition point. Close examination of
Beauchamp’s Figure 5.4 b shows they are not present inside the mirror, but
they are just visible in his Figure 5.4c near the mirror-mist boundary.
The mirror markings can be explained with reference to Figures 5.2, 5.3b, and
5.4. The explanation below closely follows Fréchette’s1 general discussion as
well as Beauchamp’s22 convincing analysis. As the crack accelerates away
from the origin, micro portions of the crack front begin to twist slightly or tilt
up and down out of the main fracture plane. These local deviations occur as
consequence of the stress field in front of a fast crack having maxima that are
out of plane, unlike the case for a static or slowly moving crack.13 The
momentary tilting or twisting does not persist for very long since crack plane
5-8
Fracture Surface Examination
(a)
(b)
5-9
Fractography of Ceramics and Glasses
(c)
Mist
(d)
Figure 5.4 continued (c) is a region still inside mirror but close to the mist
boundary. The ridges are longer and wider. (d) is in the mist, just beyond the
mirror/mist boundary. The features are similar to the prior figure, except that
they are now large enough to be detected optically. Close examination of (c)
and (d) reveals faint but distinct twist hackle lines in some of the large features,
attesting to some localized lateral crack growth.
5-10
Fracture Surface Examination
Hackle
(e)
Figure 5.4 continued (e) shows features in the hackle zone. The features are
now larger than 10 µm, but are similar in character to those in the mist and
mirror zones.
5-11
Fractography of Ceramics and Glasses
could be the impetus to form a band of hackle along the crack front and hence
the mist-hackle boundary.
Fréchette1 defined hackle and mist hackle. This Guide adopts his definition for
the former, but makes a slight modification to the latter:
It is implied in these definitions that hackle lines are physical steps on the
fracture surface, as opposed to undulations. Mist is fairly obvious in glasses,
but is difficult or impossible to discern in ceramics due to the microstructure.
Fréchette’s definition of mist hackle originally also included the words “and
coarsening up to the stage at which the crack bifurcates” which suggests it
encompassed all mirror markings including hackle right up to the point of crack
branching. It seems reasonable to distinguish the mist region from hackle
region for practical reasons that will become evident when ceramic mirrors are
discussed below. Therefore this Guide adds the following term:
Velocity hackle: Hackle markings formed on the surface of the crack close to
its terminal velocity, observable as discrete elongated steps aligned in the
direction of cracking.
Mirrors are very important since they not only draw attention to the fracture
origin, but their size may be used to estimate the stress in the part at the
moment of fracture. For a long time it was common practice to interpret the
mirror as having distinct boundaries between the mirror and mist and between
the mist and hackle. Extensive empirical work showed that the relationship
between the mirror size and the stress at the instant of fracture is:
5-12
Fracture Surface Examination
σ R=A (5-1)
where σ is the tensile stress at the origin at the instant of fracture, R is the
mirror radius, and A is a constant, known as the “mirror constant” and is
considered by many to be a material property constant. It has units of
stress√length, which are the same units as stress intensity which is discussed in
detail in section 7.5.1. The mirror radius may either be the mirror-mist
boundary or the mist-hackle boundary for glasses. There are different A’s for
each boundary. For polycrystalline ceramics there usually is simply a single
mirror-hackle boundary. Equation 5-1 is discussed in more detail in Chapter 7,
Quantitative Analysis and in Appendix D (where guidance is given on how to
actually make the measurement), but for the moment we need only consider
some general ramifications.
Mist and hackle are sometimes triggered earlier than they would otherwise
form by material flaws or microstructural features. Figures 5.5a-c shows
examples. Strong Wallner lines, described later in this chapter, may also
trigger mist formation in fracture mirrors such as shown in Figures 5.6. Figures
5.2 and 5.3 shown earlier also hint at this possibility. Beauchamp has described
cases like these where perturbations of the crack front can trigger early mist
and hackle 27 and long, persistent “hackle streaks.”
5-13
Fractography of Ceramics and Glasses
(a)
(b)
SC
(c)
dcp
Figure 5.5 Early mist and hackle streaks associated with defects. (a) shows
mist and hackle triggered by a surface crack at site “A” in a soda lime glass
specimen at 65 MPa. The origin was well to the right, beyond the photo edge.
(b) shows early mist and hackle triggered by a large bubble “B” in a 70 MPa
strength specimen. The origin is a surface crack “SC.” (c) shows mist behind
inclusions in obsidian. Gull wings (described later in this chapter) are also
evident. (a and b courtesy R. Rice, c is courtesy A. Tsirk.)
5-14
Fracture Surface Examination
Wallner lines
(a) origin
Early mist
(b)
Figure 5.6 Early mist and hackle. (a) and the enlargement (b) show an
example of early formation of mist. The Wallner lines in (a) suggest the origin
is off to the left in the image. The crack radiating outwards encounters other
cracks in the surface shown by the arrows in (b). A strong primary Wallner
line is generated and it triggers mist earlier than normal during mirror
formation.
5-15
Fractography of Ceramics and Glasses
(c)
(a) (b)
Figure 5.7 Mixed-mode loading creates fracture mirrors without mist but with
a systematic array of “lance hackle.” (a) shows a normal mirror from pure
opening mode I loading. (b) and (c) show mirrors from mixed opening/shear
modes I/II, (c is courtesy J. Mecholsky).
Gopalakrishnan and Mecholsky et al. 28,29 showed that mist does not form
around glass fracture mirrors if the loading has some in-plane shear (Mode II
loading) acting on the crack at the same time it is under opening Mode I
loading (see chapter 7). Hackle markings are very distinct in such cases as
shown in Figure 5.7. The boundary is comprised of “lances” or “lance hackle”
that are very similar to twist hackle. These steps on the fracture surface
probably all go up (or go down) systematically as one goes around the crack
front periphery, unlike the case of Mode I loading, where hackle steps are
random or alternating up and down out of the fracture surface. The size of the
mirror is not altered by the mixed-mode loading. The mirror/mist boundary
for pure Mode I loading and the onset of lances in mixed Mode I/II loadings are
comparable in size. The practical significance of this difference in appearance
is that if a glass fracture mirror is comprised of lances with no mist, then the
loading was not pure tension, Mode I, but mixed mode with some shear.
Sommer 30 has shown similar fracture surfaces with lance hackle and no mist in
glass rods loaded in combined tension and torsion (i.e., mixed Mode I and
Mode III loading).
5-16
Fracture Surface Examination
forms the mirror boundary. The mode of crack propagation, whether it is trans-
or intergranular, also affects the mirror markings. Transgranular fracture is a
mode whereby the crack propagates directly through grains on cleavage planes,
whereas intergranular fracture is a mode whereby the crack takes paths along
grain boundaries between grains. These modes will be discussed later in
section 5.8. Mirrors may be easier to see in materials that start out with
transgranular fracture, but then develop increasing fractions of intergranular
fracture. On the other hand, it may be more difficult to define mirrors when the
fracture mode is intergranular from the origin right through the entire mirror.
For these reasons it is appropriate to add the qualifier “relatively” as in
relatively smooth when describing the mirror region since there is an inherent
roughness from the microstructure even in the area immediately surrounding
the origin. Rice observed that mirrors are more detectable in strong specimens
and are less clear in weak specimens. 31 Mirrors may not be detectable in coarse
grained materials or those with more than 10 % porosity. Mist can only be
detected in fine-grained ceramics or those that fracture transgranularly. 32
(a) (b)
(c)
5-17
Fractography of Ceramics and Glasses
(a) (b)
(c)
5-18
Fracture Surface Examination
(a)
(b)
5-19
Fractography of Ceramics and Glasses
5-20
Fracture Surface Examination
(a)
(b)
Figure 5.11 Multiple mirrors from grinding damage in a silicon nitride bend
bar. (b) is a close-up of (a). Grinding damage caused fracture in these
examples and the origins in each mirror were nearly identical. (Ref. 35)
5-21
Fractography of Ceramics and Glasses
Mirror
(a)
Polished surface
(b)
Figure 5.12 Bent mirror in a BK-7 borosilicate crown glass ring-on-ring disk
strength specimen. The origin was a Vickers indentation flaw (large white
arrows) with multiple radial and median cracks. (a) and the close-up (b) show
the fracture surface with the specimen tilted back to show the outer polished
surface where the indentation (white arrows) started fracture. Fracture started
from Vickers indentation radial cracks oriented in different directions. A
hackle line extends from the origin straight up through the mirror. Notice how
main crack branching occurs (small white arrows) soon after the mirror
boundaries formed in this equibiaxial-loaded specimen. Notice also that the
branching angle initially is small then flares out to its final configuration.
5-22
Fracture Surface Examination
may persist up into the mirror or even further. These steps are one type of
“hackle” as described later in section 5.3. They usually point back to the origin
and are an additional aid to locating the origin flaw.
Parts with mirrors centered on an interior origin in part are usually circular.
Oblong or elliptical mirrors in the interior are less common. Figure 5.9 shows
three internal fracture origins and mirrors in ceramic test specimens. In one
case (Figure 5.9c), the part was tested in flexure, but the mirror was relatively
small and showed no obvious distortion or elongation.
On the other hand, parts with surface origins often have flared or elongated
mirrors, especially if the part was stressed in flexure. The bending loading
mode usually creates a stress gradient with a maximum tension at the surface,
zero stress at the middle (“the neutral axis”), and compression on the opposite
surface as shown in Figure 4.18. As the crack accelerates from the surface
origin, it traverses through a diminishing stress field towards the interior. This
slows down the crack and elongates the mirror shape towards the interior as
shown in Figures 5.10, 5.13, and 5.14. It can even eliminate the mirror
markings altogether if the mirror is large relative to the specimen cross-section
size. The stress gradient does not affect the markings along the tensile surface.
There often is a small cusp on either side of the mirror pointing back to the
origin in glasses. The cusp is accounted for by fracture mechanics
considerations. (A crack that connects to a free surface experiences a slightly
greater stress intensity factor right at the surface.) The small cusp is
inconsequential for most analysis, except when fracture mirror sizes are
measured as discussed in chapter 7 and Appendix D.
5-23
Fractography of Ceramics and Glasses
Figure 5.13 Mirror shapes around surface origins. (a) is common for tension
specimens, or flexure specimens or plates in bending if the mirror is small
relative to the thickness. (b) and (c) show mirrors that are incomplete into the
interior. This occurs when the mirror is large in a part loaded in bending.
Figure 5.14 Fracture mirrors in four glass rods broken in flexure. The
maximum tensile stress was on the bottom in each case. Fracture mirrors on
the rod sides had pronounced flair or were incomplete due to the stress
gradient.
5-24
The elongations in Figures 5.13 and 5.14 are common. Flexural stress
gradients always cause elongations towards the interior. On the other hand,
there are several important cases whereby elongations along the outer surface
of the part occur as shown in Figure 5.15. Elongated flaws may generate
elongated mirrors and sometimes even mirrors with side protuberances. Figure
5.16 shows two possible scenarios for the formation of such elongated mirrors.
Machining/grinding cracks and polishing scratches cause such markings.35,36
Figures 5.17 and 5.18 show examples.
Figure 5.15 Mirror elongations along the surface can occur if the origin is a
scratch or a long parallel grinding crack. (a) shows a single elongated crack
at the origin, (b) shows a grinding zipper crack (see section 6.7.6), and (c)
shows a zipper crack that generated lobes on the sides of the mirror.
(a) (b)
Figure 5.16 Two scenarios that could account for mirror elongations along
the surface from long flaws. (a) shows how a crack could unzip or propagate
quickly along the surface. (b) shows that criticality could be reached at several
points on the flaw periphery and fracture commences accordingly.
5-25
5-26
Fracture Surface Examination
(a)
(b)
5-27
Fractography of Ceramics and Glasses
Residual surface stresses may also alter mirror shapes. The shape distortion
depends upon the magnitude, sign, and depth of the residual stresses. Figure
5.19 shows a typical parabolic residual stress distribution in glass from thermal
tempering. 37 Stresses are strongly compressive at the surface, decrease to zero
about 21% of the thickness in from each surface (assuming the two surfaces are
cooled at the same rates), and change to tensile in the interior. The
compressive stress at the surface is twice the magnitude of the interior tensile
stress. Figure 5.20 shows mirrors in a heat-strengthened glass consumerwares.
For a surface origin, the compressive surface residual stresses suppress mirror
boundary markings along the outer surface. Compensating residual internal
tensile stresses shorten the mirror dimensions into the interior. A more detailed
discussion of discussion of mirror shapes in tempered plates is in Appendix D
and Figure D.8 in particular. Diced pieces from fully-tempered glass show mist
and velocity hackle in the interior in Figure 5.21.
σt, max
Tension
0
Stress
Compression
σc, max
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5
5-28
Fracture Surface Examination
(a)
(b)
Figure 5.20 Residual stresses alter the mirror shape. These images show
thermally-strengthened glass consumerwares. The piece in (a) shows a
complete suppression of mirror boundary markings along the lower right
surface, but they are present to the left of the origin. This indicates the piece
was thermally-strengthened with emphasis to protect the rim. The mirror is
well-developed into the depth since residual tensile stresses there help form the
mist and hackle. (b) shows a piece with mirror markings into the interior, but
not along the bottom surface. The crack branched from the mirror sides in the
interior of the piece then hooked back around to sever the fragment as shown
by the arrows. The origin is one or a series of chatter cracks (black arrows)
from contact damage on the outer surface.
5-29
Fractography of Ceramics and Glasses
(a)
(b)
Figure 5.21 Mist and velocity hackle in tempered glass. These photos are of
the same auto window pieces in Figures 4.18b and d. (a) shows the crack
reached terminal velocity in the interior where the tensile stresses were
greatest. The black arrow shows the direction of crack propagation and the
onset of mist. Each diced piece usually has this pattern since cracks slow at a
branch then accelerate to terminal velocity again. Crack propagation was
retarded at the outer surfaces due to the compression. Final breakage through
to the surface has vertical twist hackle (white dashed ring). The arc shaped
lines are secondary Wallner lines described later in this chapter. (b) shows the
exact single impact origin (arrow) and mirror for the window.
5-30
Fracture Surface Examination
5.3 Hackle
In addition to the mist hackle and velocity hackle (described on page 5-12 in
the mirror section), that are self-initiated by the crack at terminal velocity, there
are other types of hackle that are very useful for ascertaining the direction of
crack propagation. These markings enable the fractographer to trace crack
propagation back to an origin. Hackle lines are sometimes called “lances” and
less commonly “striations,” but this latter term is not recommended. Hackle
can be generated from irregularities in the origin flaw. A good example is
“grinding crack hackle” shown in section 6.7.6 in the chapter on origins.
Coarse microstructural hackle: Large broad hackle lines that form from non-
specific sources, but most likely variations in the microstructure.
Ceramics often develop coarse hackle even in the absence of mirrors or other
well-defined fracture surface markings. They occur in low-strength or porous
ceramics and often are the only marking on a fracture surface that indicate the
direction of crack propagation. The microstructure or some geometric
irregularity causes portions of the crack to advance on non-coplanar regions
separated by rounded ridges as opposed to sharp steps. Figures 5.22 and 5.23
show examples. “Fine microstructural hackle” is described on page 5-36.
5-31
Fractography of Ceramics and Glasses
(a)
(b)
5-32
Fracture Surface Examination
Wake hackle: A hackle mark extending from a singularity at the crack front in
the direction of cracking.
Wake hackle in glazes or veneers often is the only recognizable fracture surface
feature in kitchenware, electrical insulators, and some dental ceramics. The
core ceramic materials in these instances are often coarse-grained, porous, or
weak, and few fracture surface markings are evident. Sometimes in even the
most difficult materials, tiny porosity will aid the fractographer by serving as a
source of wake hackle. A superb example was shown by Ko, reference 38.
dcp
Figure 5.24 Wake hackle, the fractographers “weather vane” for crack
propagation. This shows wake hackle from bubbles in glass. The direction of
crack propagation (dcp) is shown by the arrow. Many bubbles are beneath the
surface in this view.
5-33
Fractography of Ceramics and Glasses
dcp
(b)
Figure 5.25 Wake hackle. (a) shows an array of wake hackle from pores in
an Empress II dental crown veneer. The direction of crack propagation is
marked by the arrow. Notice that some hackle is triggered even though the
crack did not cleave the pore (small white arrow). (b) shows an example of
transgranular fracture from a large grain origin in polycrystalline AlON.
Wake hackle was triggered by grain boundaries and tiny pores (small arrows).
5-34
Fracture Surface Examination
(a)
dcp
(b) (c)
dcp
Figure 5.26 Wake hackle (a) shows wake hackle from inclusions in a lead
zirconium titanate bend bar. They point back to the origin on the bottom. The
tensile surface is on the bottom and a compression curl is on the top. The
origin is a semielliptical surface crack (arrow) generated by a Knoop
indentation. (b) and (c) show parabolic double wake hackle tails triggered by
inclusions in obsidian glass. (b,c courtesy A. Tsirk)
5-35
Fractography of Ceramics and Glasses
Irregularities such as pores or inclusions are not the only sources of wake
hackle in ceramics. Grain boundaries and local crack path redirection as a
crack propagates from one grain to another with a different crystallographic
orientation, or local-density or other microstructural variations all can trigger
“fine microstructural hackle:”
Fine microstructural hackle: Small hackle lines of the order of size of the
microstructure that occur as a crack interacts with the local microstructure.
Twist hackle: Hackle that separates portions of the crack surface, each of
which has rotated from the original crack plane in response to a lateral
rotation or twist in the axis of principal tension. (Fréchette, Ref 1.)
Twist hackle are very telltale markings as illustrated in Figures 5.27 and 5.28.
They are also known as delta patterns, river deltas, river patterns, lances, and
sometimes even in the geological community as striations. (Striations in this
instance are not to be confused with fatigue striations, a well-known metal-
urgical fracture surface marking). The roughly parallel segments point in the
direction of local crack propagation. Twist hackle can be generated by the
primary crack as it travels directly through the body, especially as it goes
around corners or geometric irregularities, or when stress conditions changed.
Twist hackle can also occur from final breakage between local crack segments,
or through to the side of a specimen that had had strong thermal or stress
gradients. Twist hackle shows the local direction of crack propagation. Plates
in bending often have curved cracks that run quickly on the tension side, but do
not quite break through to the opposite surface. The crack later snaps through
to the opposite surface leaving twist hackle markings as shown in Figure 5.29a.
Double torsion fracture mechanics specimens often have such markings.
Figures 5.21a and 28b show a similar circumstance whereby final breakthrough
of tempered glass has twist hackle near both outer surfaces.
5-36
Fracture Surface Examination
(a)
Crack
propagation
direction
End on view
(b) ( c)
Figure 5.27 Twist hackle. (a) shows how twist hackle is formed. A crack
propagating from left to right runs perpendicular to the initial stress field as
shown on the left. The axis of principal stress then tilts as shown on the top
middle. The crack is unable to rotate all at once in response to the new stress
direction, so it breaks into small, unconnected segments. The steps are “hackle
lines” or “lances.” Lateral breakthroughs occur between the micro segments
as shown in (b) and (c). Note in (c) how the micro segments can link either by a
top segment breaking though to a bottom segment or vice versa. On any given
lance, the link up can alternate between either segment creating a barbed
appearance. Sometimes an incomplete segment can breakthrough long after the
main fracture has occurred, generating faint tinkling sounds and creating very
sharp needle like fragments that fall free from the fracture surface. The small
hackle lines gradually merge into coarser hackle lines, and so on, until the new
crack plane is aligned perpendicular to the new tensile stress axis. Usually
crack propagation is in the direction of fine to coarse hackle.
5-37
Fractography of Ceramics and Glasses
(a)
dcp
(b)
Figure 5.28 Twist hackle. (a) and (b) show close-ups of twist hackle in glass.
The crack was running in the direction of the arrow in (a). (b courtesy J.
Varner)
5-38
Fracture Surface Examination
(a)
dcp
(b)
Figure 5.29 Twist hackle. (a) shows a glass microscope slide that was broken
in bending with a little superimposed torsion. The main crack fracture from
left to right, leading on the bottom surface as attested to by the profusion of
mist and secondary Wallner lines and an occasional primary Wallner line
(described in the next section). Final break-through to the top surface (white
arrow) created the twist hackle along the top. (b) shows a similar pattern but
with a tempered glass fragment. .
5-39
Fractography of Ceramics and Glasses
(a)
dcp
side side
groove groove
(b)
side
groove
Figure 5.30 Twist hackle in grains. (a) and (b) show fracture surfaces of
chemical vapor deposition (CVD) silicon carbide micro-tensile specimens.
CVD materials often have coarse, columnar grains. Such grains run from top
to bottom in this view. Twist hackle in these grains in both images show that
fracture propagated from left to right as shown by the arrows. Fracture
occurred from cracks located at the bottom of the grooves on the left side in
both cases.
5-40
Fracture Surface Examination
Sommer30 analyzed the formation conditions of twist hackle and concluded that
in glasses a stress redirection of 3.3o is needed to initiate the twist hackle.
Beauchamp27 observed hackle with only 1o of twist. Torsional stresses
superimposed on tension stresses can create twist hackle markings as shown in
Figure 5.31 which shows a rod loaded in mixed mode I, opening mode, and
mode III, out-of-plane shear, as described in section 7.1. An interesting paper 40
reported on how a continuum phase-field approach could be used to model
twist hackle formation and propagation. Very small initial perturbations along
the initial crack front would form parallel “daughter crack segments” that
gradually coalesce, as the main crack plane rotated.
5-41
Fractography of Ceramics and Glasses
Shear hackle: A particular form of twist hackle that occurs in the lazy loops
This definition may seem peculiar, but once the fractographer has seen an
example such as Figure 5.32, it becomes clear what Fréchette had in mind.1 He
said: “Far from the fracture origin, it is common to see the crack turn through
90o and develop a cupped surface, inclined at 45o to the free surface, from
whose centerline a spray of twist hackle emerges. It has little significance in
reconstructing a failure event in industry, but it is useful in problems of rock
fractures.”
dcp
Figure 5.32 Shear hackle in a glass plate. The crack was perpendicular to
the glass surface and was slowing as it moved in the direction of the white
arrow. It then turned 90o parallel to the surface and finally broke through to
the surface at about 45o, producing the spectacular fan-shaped shear hackle.
Shear hackle is a variant of twist hackle.
5-42
Fracture Surface Examination
Corner hackle: A fan like array of hackle lines created when a crack curves in
a plate or goes around a corner of a component.
Figure 5.33 shows two examples. The fan like array spreads outwards in the
direction of crack propagation.
(a) (b)
Figure 5.33 Two types of corner hackle. (a) shows a fan like pattern that
occurs when a crack that is leading on the bottom surface turns to another
direction. (b) shows a fan-like pattern that occurs when a crack goes around a
corner of a container or vessel.
Step hackle: A type of twist hackle in the form of a single line that is an arc
shaped line on a fracture surface of a plate broken in bending and twisting.
Step hackle is a line that is a step on the fracture surface such as shown in
Figure 5.34. The line is parallel to the local direction of crack propagation.
Unlike most twist hackle, it is an isolated line. The arc curvature of a step
hackle line is opposite to the curvature of arrest lines or Wallner lines that may
exist on the same fracture surface as shown in Figures 5.34b and c. The
fracture surface is usually tilted or curved and not perpendicular to the plate
outer surfaces (end-on view Figure 5.34a.). The step can initiate by itself if the
tilt is strong enough, or may be triggered by other features such as bubbles,
inclusions or mist. Step hackle is shown here for glasses, but can occur as
“cleavage step hackle” as described in Chapter 8 on single crystals.
5-43
Fractography of Ceramics and Glasses
(a) dcp
twist hackle
(c) Wallner
lines step hackle line Wallner
lines
Figure 5.34 Step hackle schematic (a). (b) and (c) show a piece of a large
(1.5 m by 0.75 m) glass mirror that fell off a wall in the author’s home. The
mirror broke into over 100 pieces. Many fragments had evidence of bending
and twisting. Notice the opposite arcs of the step hackle line and the Wallner
lines (thin dashed lines in a).
5-44
Fracture Surface Examination
5.4.1 Introduction
Also known as “ripples” or “rib marks,” Wallner lines were named for Helmut
Wallner who first explained how they formed in 1939. 41 He showed beautiful
images (similar to Figures 5.2 and 5.3 of this Guide) of fracture mirrors in
flexurally tested glass rods with the lines within the mirrors. He presented a
splendid sketch showing how an expanding crack front was overtaken by an
expanding elastic wave. The Wallner line is the locus of the intersection of
the elastic wave and the crack front. The elastic wave momentarily causes the
crack to ripple out of plane like a wave on a pond surface. The shape of the
line depends upon the respective shapes and speeds of the elastic wave and the
crack front, as well as the direction the wave approaches the crack front.
Wallner lines are invaluable in determining the direction of crack propagation.
As will be shown, they usually do not have the exact same shape as the crack
front, but they are usually curved (bowed) in the direction of crack propagation.
They are best understood by the examples that follow.
Wallner lines are very shallow hillocks on a fracture surface. They look like a
thin shadow band and they tend to shift slightly as illumination is adjusted.
This is in marked contrast to crack arrest lines that are sharp and do not appear
to move as light is adjusted. Wallner lines are very easy to see on fracture
surfaces once the lighting is adjusted, but they are very difficult to see on the
scanning electron microscope, since they produce almost no contrast and are
very shallow.
There are several types of Wallner lines depending upon the source of the
elastic wave. Fréchette1 thoughtfully categorized these as primary, secondary,
and tertiary. This Guide follows his scheme and adopts his definitions. His
figures showing the evolution of the loci are instructive, but have been revised
and expanded into separate illustrations in this Guide to make them easier for
the beginner to understand. Keep in mind that a crack front through a plate
may be straight or have different shapes according the stress state as shown in
Figure 5.35.
5-45
Fractography of Ceramics and Glasses
(a)
(b)
(c)
(d)
Figure 5.35 Crack front variations due to stress distributions through the
thickness. (a) shows a crack front in a uniform tensile loaded plate. (b) a
plate with greater tensile stress on the bottom. (c) a plate in bending such that
the bottom is in tension and the top in compression. (d) a tempered plate with
tensile stresses in the interior and compressive stresses at the surface.
5-46
Fracture Surface Examination
Unless the crack is moving very slowly (and the elastic wave traverses the
entire crack front very quickly), primary Wallner lines do not show the exact
crack front profile. Nevertheless, they are curved in the same direction and can
help the fractographer deduce which way a curved crack was running. It is
fairly easy to show that straight crack fronts generate straight Wallner lines, but
they will tilt at a different angle than the original crack front depending upon
the crack velocity.
dcp
5-47
Fractography of Ceramics and Glasses
front
elastic pulse is
generated here
radiating
Figure 5.37 Primary Wallner lines. This schematic series shows the
progressive advance of the crack at successive time intervals. The elastic wave
that forms the Wallner line (thick line) is generated when the advancing crack
hits a surface irregularity “O”. The crack leads on the bottom surface, but is
moving at constant velocity in this example.
5-48
Fracture Surface Examination
dcp
dcp
(a) (b)
Figure 5.39 Gull wings are a variant of primary Wallner lines. (b) is a close-
up of (a). Most of these pores also have wake hackle as well. These images
are from obsidian with crack propagation from bottom to top as shown by the
arrows. The wing angles can be used to measure crack velocity (section 7.9).
(specimen courtesy A. Tsirk)
5-49
Fractography of Ceramics and Glasses
Wallner lines are less easy to see in ceramics. Figures 5.10b and 5.38 show
examples. In porous, granular, or coarse-grained ceramics, large broad ridges
that are Wallner lines may be the only recognizable marking on the fracture
surface. If so, the ridge may have been caused by microstructural irregularities
or surface flaws on one side of the origin or to a geometric irregularity (e.g., an
inside corner in a formed-ware, or a ridge on a plate).
Figure 5.39 shows “gull wings,” a variant of primary Wallner lines. They are
usually formed by a stress pulse created as the crack encounters pores,
inclusions, or other irregularities. There even are instances where they may be
triggered by secondary fracturing behind the primary crack front, and hence
will be detached from obvious irregularities. 42
Unlike the primary Wallner lines that form as a result of a crack encountering
an external feature, secondary Wallner lines occur from features the crack
generates itself. Mist and hackle are disturbances created by crack running at
or near terminal velocity. The mist and hackle generate copious secondary
Wallner lines. Figure 5.40 shows their formation sequence and Figure 5.41
shows examples. It is very common for both primary and secondary Wallner
lines to be present. They can be distinguished by their different curvatures.
Secondary Wallner lines are more hooked (like a fish hook) than primary
Wallner lines. As was the case with primary Wallner lines, secondary Wallner
lines do not show the exact crack front profile. Even so, it is helpful to deduce
from their curvature which way a crack was running.
5-50
Fracture Surface Examination
t=0
crack front dcp
t=1
radiating
Wallner line elastic wave
t=2
t=3
Final
Wallner line
Figure 5.40 Secondary Wallner lines are formed by the crack itself. This
schematic series shows the progressive advance of the crack and the elastic
pulse that forms the secondary Wallner line (dark line) in a plate with a stress
gradient. Notice the hook final shape, a consequence of the stress gradient
caused by the variation in the crack velocity from a maximum on the bottom to
close to zero at the top.
5-51
Fractography of Ceramics and Glasses
(a)
(b)
(c)
dcp
5-52
Fracture Surface Examination
Tertiary Wallner lines: Wallner lines caused by elastic pulses generated from
outside the crack front. (Fréchette, Ref. 1)
Vibration or shock from a source other than the crack itself can create waves
that intersect with and cause momentary deviations in the crack front.
Depending upon the source of the elastic pulse relative to the crack plane, the
interaction may generate Wallner lines that may be very close to the shape of
the crack front, or it may produce an arc formed by a progression of
intersections as in the examples shown above. For example, if the crack is
moving relatively slowly, the fast elastic wave will overtake all portions of the
crack front nearly simultaneously, producing a Wallner line “snapshot” of the
crack front.
If the pulse source is located at a site normal to the fracture plane, the pulse
may reach all portions of the crack front simultaneously also producing a
Figure 5.42 Tertiary Wallner lines in a large chunk of glass that was
hammered into pieces. There are multiple impact sites marked by arrows, the
most prominent of which is marked by the larger white arrow. Notice how the
tertiary Wallner lines are concentric about the impact sites. This was one of
Professor Fréchette’s demonstration pieces.
5-53
Fractography of Ceramics and Glasses
“snapshot” of the crack front. The impacted glass shown in Figure 5.42 is a
good example. Notice how the tertiary Wallner lines are concentric about the
impact sites. The impact site not only generated the cracks, but the elastic
pulses as well. (In contrast note how the primary and secondary Wallner lines
in Figures 5.3 and 5.41a cross over each other on top of the origin.) Impact
typically results in a ringing within the specimen that may be short lived, but
will be long enough to endure for the period a crack traverses through a body.
On the other hand, a sudden stress relief may create a brief vibration when a
crack pops in or propagates. This may be accompanied by an audible snap, but
only generates faint elastic pulses so that only a few tertiary Wallner lines form
near the origin. Fréchette1 noted that window panes that are cracked by thermal
stresses may have only a few tertiary Wallner lines near the origin and are
featureless elsewhere.
A schematic case of a pulse overtaking a crack from the rear is shown in Figure
5.43. The reader is reminded that Wallner lines do not always create a
“snapshot” of the crack front shape. Tertiary Wallner lines are not always
formed by pulses arriving from a source so conveniently located as in test
pieces studied with ultrasonic fractography. They may come from any
direction including from sources that are out of the crack plane.
5-54
Fracture Surface Examination
overtaking
5-55
Fractography of Ceramics and Glasses
(a)
(b)
5-56
Fracture Surface Examination
500 µm
(a)
dcp b
dcp dcp
a
(b)
100 µm
dcp
5-57
Fractography of Ceramics and Glasses
Elastic waves can reflect off free surfaces creating a lambda shaped line on the
fracture surface as shown in Figure 5.46. These are typically formed by
primary or secondary Wallner lines. Crack velocity can be calculated from the
spacing of the Wallner lines as discussed in section 7.9 and Figure 7.31.
Propagating
t
crack front
Figure 5.46 Wallner lambda line mark. The schematic shows the advancing
crack (thinner line) encountering a discontinuity at point O. An elastic pulse
creates a radiating elastic wave. The intersection of the wave and the
advancing crack front creates a primary Wallner line that runs upward to the
top surface. When the elastic wave reflects from the opposite surface, the
Wallner line continues back downward.
Arrest Line: A sharp line on the fracture surface defining the crack front
shape of an arrested, or momentarily-hesitated crack. Resumed crack
propagation occurs under a more or less altered stress configuration.
5-58
Fracture Surface Examination
Arrest lines are different than Wallner lines in two key respects: arrest lines are
sharp, and, unlike most Wallner profiles, arrest lines show the crack front
profile at an instant in time. Sometimes, if the crack repropagation occurs in a
slightly different direction, twist hackle lines will form that lead away from the
arrest line.
Michalske showed that cracks that arrested after slow crack growth sometimes
require extra loading to repropagate after aging. 43 The crack tip blunts slightly
during the aging, and the crack has to resharpen before repropagation. This
process occurs with slight variations along the crack front so that when the
crack repropagates there are slight humps and valleys along the crack front and
tiny hackle lines leading away from the arrest line.
5-59
Fractography of Ceramics and Glasses
arrest line
(a)
Wallner lines
(b)
(c)
Figure 5.47 Crack arrest line examples. (a) shows a glass microscope slide
broken in bending and the arrest line is on the right (arrow). The lines to the
left are primary Wallner lines. Wallner lines are not as sharp as arrest lines.
(b) shows an arrest line in a glass slide that broke from thermal stresses. (c)
shows concentric arrest lines from repeated cyclic loading of a glass plate in
bending. Compare the sharp lines in this photo to the wavy tertiary Wallner
lines in Figure 5.44b. (c is courtesy D. Green and is the same as Fig. 5.66).
5-60
Fracture Surface Examination
5.6 Scarps
Scarps shown in Figure 5.48 are subtle curved lines on a fracture surface
caused by interaction of a propagating crack and a liquid or a reactive
environment. The liquid can act to retard a crack front or can cause it to
accelerate, sometimes in a discontinuous fashion. The scarps mark sudden
transitions in crack velocities that can occur due to interactions of the crack
with liquid. They usually correspond to transitions in the crack tip
environment, including changing from wet to dry, dry to wet, humid air to dry
air, etc. They even can occur in totally wet environments if the crack
accelerates so fast that the viscous liquid cannot keep pace and instead cavitates
behind the crack tip. Scarps are usually only detected in glasses and single
crystals. Scarps could exist in polycrystalline ceramics, but the inherent
roughness from the microstructure masks their presence.
The practical significance of scarps is that they are evidence that fracture
occurred in the presence of a fluid, most commonly water. They also can help
confirm the local direction of crack propagation.
Fréchette has identified several types of scarps and accounted for their
formation.1 Scarps can vary from as simple as a single faint line across the
fracture surface perpendicular to the direction of crack propagation (cavitation
scarp), to a complex network of curves such as shown in Figure 5.48. Figure
5.48b of a scarp in single crystal sapphire is, to the best of my knowledge, the
only documented example of a scarp in a ceramic material.
Fréchette called some of these “Sierra scarps” if they are pointed in the
direction of crack propagation as shown here as Figure 5.48c (and also in
Frechette’s book, Figures 2.28 and 2.29) due to their similarity to his favorite
mountain range. They form when water (in the form of spaced fingerlets as in
a Taylor instability) overtakes a decelerating dry crack, causing local advances
forward by the crack front.
5-61
Fractography of Ceramics and Glasses
(a) (b)
dcp
(c) (d)
(e) (f)
dry
wet
Figure 5.48 Examples of scarps. The arrows show the direction of crack
propagation. (a) is a fractured glass medicinal vial. The liquid was the
medicine within the vial. (b) is from a single crystal sapphire hemispherical
dome broken by water quench thermal shock. This differential interference
contrast optical photo shows the scarps have variability in height. (c) shows
Sierra scarps in glass caused by water “fingerlets” overtaking the advancing
crack (courtesy E. Beauchamp) (d) - (f) are obsidian flakes formed by
knapping. (courtesy A. Tsirk) The faint vertical lines within the scarps in (d)
are either “streaks” or “subcavitation hackle,” a precursor to complete
cavitation. The parallel bands in (c), (e) and (f) are tertiary Wallner lines
intentionally introduced by vibrations from an engraver (c), or a hand palm
sander (e,f) to study the crack velocities. In (f) water that was available from
the right side facilitated crack advance causing the scarp.
5-62
Fracture Surface Examination
Scarps are very subtle and can be very difficult to see. They are easily
overlooked. They are best seen in an optical microscope with the specimen
tilted to create a mirror-like reflection from the fracture surface. The specimen
or the lighting must be tilted back and forth until one catches just the right
reflection. They usually exist in only a portion of the specimen and usually
where the crack is moving at a slow to moderate speed.
Tsirk 45,46 has studied a variety of scarps in knapped obsidian arrowheads and
knives formed by pressure flaking. Some of the scarps were generated in
antiquity when they chipped during usage. Tsirk has postulated that the shape
of scarps may depend upon the nature of the fluid and possibly the viscosity.
Scarps that have been observed associated with blood are different from those
formed with water or saliva.45,46 The reader is referred to Fréchette’s book1 and
References 43-46,47,48 for details on scarps.
glue metal
σ
(a) (b)
Figure 5.49 Glue chipping cracks. (a) show a side view with glue applied to
the surface of the glass of ceramic. Contraction of the glue causes a crack to
enter the ceramic or glass at about 45o to the surface. (b) shows a comparable
geometry with a metal bonded to a ceramic or glass. If the metal contracts
more than the ceramic, edge cracks can be created.
5-63
(a) (b)
fracture surface
dcp
(c)
fracture surface
twist hackle
Figure 5.50 Examples of glass glue chips. (a) shows an Alfred University
demo plate covered with glue chips. (b) is a close-up of isolated glue chips on
the surface. (c) shows a side view of a broken plate that shows glue chips on
the bottom that affect crack propagation. (d) shows large glass disk that
fractured from a glue chip caused by an epoxy bond to a metal part.
5-64
Fracture Surface Examination
Glass damage
(a) (b)
glass glass
alumina
substrate
100100
µm µm 100 µm
(c) (d)
Figure 5.51 Glue chipping damage in a thick-film resistor that cracked during
thermal cycling. (a) shows the component in its epoxy encapsulant. (b) shows
the exposed electrical circuitry. Scallop cracks (white-colored reflective
patches) formed in an inner glass encapsulating layer. (c) and (d) show two
glue chips in the glass. The arrows show the direction of crack propagation.
(all photos and interpretation courtesy R. Tandon and S. Monroe)
5-65
Fractography of Ceramics and Glasses
Glue chips were the origins of fracture in spectacular failures of very large
windows in the John Hancock skyscraper in Boston in the early to mid-1970s.49
The windows were 25 mm – 32 mm (1 inch to 1.25 inch) double-glazed glass
panels with two 6.35 mm (0.25 inch) thick individual panels bonded with a lead
spacer between them. A copper coating was flame sprayed to the inside rims of
the glass panels. The glass panels were then soldered to the square lead spacer
around the rim. Differential contraction of the lead versus the glass created
incomplete glue chips in cold weather. (A silver reflective coating on the
inside of the outer glass panel may also have been a factor.) Subsequent
mechanical and thermal loadings caused the cracks to propagate, and the
windows eventually fractured. The author and his future spouse, who were
undergraduate students at Northeastern University, looked at the building in the
1970s as we came out of engineering classes. At any time, hundreds of glass
panels were replaced temporarily by plywood sheets. Eventually, all the
windows in the skyscraper were replaced. The above interpretation of the
cause of the John Hancock window failures is based on my private
communications with Prof. Fréchette, Mr. Leighton Orr, and information in
Reference 49.
5-66
Fracture Surface Examination
(a)
(b)
5-67
Fractography of Ceramics and Glasses
(a)
(b)
5-68
Fracture Surface Examination
Many glasses, oxide ceramics, and nonoxide ceramics with an oxide sintering
aid are susceptible to environmentally-assisted slow crack growth (SCG). The
expression “slow crack growth” usually is used in the context of stable crack
extension due to combined effects of stress and environment. (It is also
commonly used to describe high temperature stable crack extension as
described later in section 5.9.3.) Water, either in liquid or as a vapor, is the
most common reactive species that causes slow crack growth, although other
small polar molecules such as hydrazine (N2H4), ammonia (NH3), and methanol
(CH3OH) can enhance crack growth in silicates. 50,51 Ceramics such as alumina
and magnesium fluoride are also susceptible to environmentally-enhanced
crack growth. Acetonitrile (CH3CN) can attack magnesium fluoridate, but not
alumina or silica.50,51 Slow crack growth requires both a stress to open the
crack faces and a reactive species to chemically react with the strained atomic
bonds at the crack tip. Hence, it is sometime called “stress corrosion cracking.”
Figure 5.54 illustrates the leading theory for how environmentally-assisted slow
crack growth occurs in glasses. Slow crack growth can lead to weakening
and/or time dependent fracture. Slow crack growth is also sometimes called
“static fatigue” if the loading is steady or constant. If the stable crack growth
occurs while loading increases at a constant rate, as in common strength tests, it
is sometimes referred to as “dynamic fatigue.”
All glasses are susceptible to slow crack growth. Many other ceramics are
too, particularly if they are bonded by a glassy boundary phase, no matter how
thin. Covalently-bonded ceramics are not susceptible to SCG at room temper-
5-69
Fractography of Ceramics and Glasses
ature, unless there is an oxide boundary phase. Section 7.10 has more quali-
tative information about the mechanics of slow crack growth and Table 7.2 lists
materials that are or are not susceptible.
Si
Si Si O
H
O H
(a) O H O
H O H
H
Si O
Si
Si
stress
(b)
H2O
Figure 5.54 Stress corrosion cracking can cause slow crack growth in glasses.
The polar water molecule, either in liquid or gaseous form, attacks strained
silicate bonds. (after Michalske and Freiman50 and Michalske, Bunker, and
Freiman51)
5-70
Fracture Surface Examination
Slow crack growth halo: A region or band around a flaw in a ceramic caused
by stable crack extension. The band is caused by changes in the mode of crack
propagation.
SCG halos can often be detected with both optical and SEM microscopy as
shown in Figure 5.55 for the case of a high-purity alumina. 52 Usually the
intergranular zone appears darker when viewed with an optical microscope
since the rougher fracture surface scatters light in different directions. In
contrast, the slow crack growth zone looks lighter in the electron microscope
possibly due to slight charging effects on edges of the intergranularly-fractured
grains. Mecholsky 53 showed superb pictures of intergranular slow crack
growth (from constant stress-to-fracture tests, “static fatigue”) versus
transgranular fast fracture in magnesium fluoride. Morrell presented a case of a
fractured alumina hip ball with an unusual circular intergranular zone of slow
crack growth that appeared shiny in the optical microscope. 54
mixed mixed
transgranular / intergranular
intergranular
intergranular
mixed
mixed
(a) (b)
Figure 5.55 Knoop semicircular flaws in two specimens with SCG “halos” in
a sintered 99.9 % alumina. (a) is an optical image of a 29 N indentation flaw
(Courtesy J. Swab), and (b) is a SEM image of a 49 N flaw. The precracks are
mixed transgranular-intergranular. The halo is mostly intergranular. The fast
fracture is mixed transgranular-intergranular. The crack growth occurred in
only a few seconds during an ordinary strength test. Notice that the rougher
intergranular surface appears dark in the optical photo since it scatters light,
whereas it appears light-colored in the SEM.
5-71
Fractography of Ceramics and Glasses
(a)
(b)
The usual remedies to eliminate SCG in laboratory strength testing are to test
either at a very fast loading rate or in an inert atmosphere. The SCG halos
disappear when testing is done in inert atmosphere.52
5-72
Fracture Surface Examination
In glasses a subtle partial elliptical line that sometimes appears on the fracture
surface provides a strong clue that SCG has enlarged a flaw as shown in
Figures 5.45a and 5.56. 55,56 The part elliptical line is a slight step that forms as
a result of the intersection of the unstable crack that wraps around the portion
of the crack that has not gone unstable.
Stable crack extension can occur, even in the absence of environmental effects,
if a flaw has a strong local residual tensile stresses. Indentation crack flaws
made by either Vickers or Knoop indentation are good examples. Damage
zones underneath the indentation can provide the tensile stresses that are the
driving force for crack extension. As the crack extends away from damage
zone and the crack periphery gets larger, the effect of the tensile stresses
diminish and the crack slows down and arrests in the absence of externally-
applied stresses. Alternatively, in fast fracture tests where the specimen is
loaded to fracture, there can be momentary stable crack extension of the flaw
prior to rupture. Figure 5.57 shows an example of stable crack extension at
room temperature due to local residual stresses in a hot-pressed silicon nitride.
Rising R-curves, which will be discussed in more detail in chapter 7, can also
lead to stable crack extension at room temperature. There is only a small
fractographic literature on this effect, in part due to the fact that the amount of
stable crack extension prior to a flaw going unstable is often very small.
(Stable crack extension from local residual stresses or due to R-curve effects is
usually not termed “slow crack growth” in the literature.)
Figure 5.57 Knoop flaw (dark semiellipse) in silicon nitride showing stable
crack extension (white arrows). The local residual stresses underneath the
indentation provided the impetus for the extension. Traces of the very shallow,
elongated Knoop indentation can be seen on the bottom surface.
5-73
Fractography of Ceramics and Glasses
5-74
Fracture Surface Examination
(a) (b)
(c)
5-75
Fractography of Ceramics and Glasses
(a) (b)
(c) (d)
5-76
Fracture Surface Examination
Creep can cause fracture at higher temperatures and lower stresses. There is so
much boundary phase ductility that sharp flaws are blunted. Fracture is due to
the nucleation and growth of cavities, microcrack formation, and their
coalescence. Creep fracture surfaces often look like the slow crack growth
zones shown above, but with growth zones that extend across a major fraction
of the fracture surface. Final rupture frequently occurs from creep crack zones
connected to specimen corners. Tensile surfaces often reveal extensive
tortuous microcrack patterns. A key difference between slow crack growth and
creep fractures is that the former can occur with negligible bulk deformation,
whereas the latter inevitably have significant permanent deformations with
strains of 1% or more. Figure 5.61 shows several examples.
stress
stress
(a) (b)
(c)
Figure 5.61 Bend specimens tested under creep fracture conditions. (a) shows
the tensile surface of a hot-pressed silicon nitride specimen. Notice the
extraordinarily rough fracture surface on the right side of (a). (b) shows the
tensile surface of a siliconized silicon carbide specimen. Note that the
microcracking is perpendicular to the axis of tensile stresses (double arrow).
(c) shows a side view of a bend bar tested in creep fracture conditions.
5-77
Fractography of Ceramics and Glasses
Figure 5.62 illustrates the differences between slow crack growth and creep
fracture. Glassy boundary phases such as shown in Figure 5.63 usually
account for both processes.
Slow crack growth occurs from preexistent flaws that grow until they reach a
critical condition. Fracture toughness at elevated temperature is often greater
than at room temperature due to the onset of limited plasticity. Hence, critical
flaw sizes can actually be much larger at elevated temperature than at room
temperature. Specimens that fail from slow crack growth have negligible (<
0.2 %) permanent strain.
Creep fracture on the other hand usually does not involve the initial flaws. It
blunts them. Voids and microcracks are nucleated, then coalesce, and then
grow leading to rupture. Creep fracture specimens have considerable
permanent strain at fracture. Failure is more a matter of accumulation of
damage. The intergranular crack growth zone on creep fractured bend bars
often extends over 33 % or more of the fracture surface.
(a) (b)
Figure 5.62 Creep fracture and slow crack growth usually are intergranular
at high temperature. In creep fracture (a), cavitation and microcrack
nucleation occur throughout the body. In slow crack growth (b), the processes
are confined to the crack tip region.
5-78
Fracture Surface Examination
(a) (b)
(c)
5-79
Fractography of Ceramics and Glasses
Figure 5.64 Fracture mechanism map for hot-pressed silicon nitride from
Reference 58. This map is for flexurally-loaded specimens.
5-80
Fracture Surface Examination
Figure 5.64 is a fracture mechanism map that puts these pheneomena into
context. 58 It is for a commercial silicon nitride hot-pressed with about 1 mass
% magnesium oxide. The sintering aid reacted with silica on the surface of the
starting powders and formed a very thin silicate boundary phase glass.
Depending upon the stress and temperature, flexural specimens fractured from
any one of several mechanisms. At low temperature, material and grinding
flaws controlled strength. At moderate temperatures in an oxidizing
environment, flaws either remained unchanged and controled strength, or they
healed or blunted causing some strengthening. On the other hand, oxidation
sometimes created pit flaws that weakened the body. At 1000oC and greater,
integranular slow crack growth caused flaw growth and eventual fracture in
times shown in hours on the dashed lines. At low stresses and high
temperature, creep fracture controlled lifetime.
With ceramics and glasses, if lines like these are observed, they are more likely
to be simple arrest lines from iterative loading such as shown in Figure 5.66
which is from Reference 59. The lines are visible due to slight variations in the
crack plane between loading cycles. Incremental crack growth during cyclic
loadings might be simply due to the accumulated effect of the slow crack
growth phenomena described in section 5.9.1. Evans showed another good
example in glass in Reference 60.
5-81
Fractography of Ceramics and Glasses
(c)
(e)
Figure 5.65 Classic metal fatigue striations in (a-d,) a steel railroad coupler
pin (see also Fig. 3.21), and (e) an aluminum bicycle bracket. An arc line does
not necessarily correspond to one loading cycle.
Figure 5.66 Concentric arrest lines in a plate of soda lime silica that was
cyclically loaded in flexure. The lines are due to slight variations in the crack
growth planes on each loading cycle. (courtesy of D. Green, Penn State
University, same as Figure 5.47c)
5-82
Fracture Surface Examination
(a) (b)
(c)
Figure 5.67 Bend bar that was tested to fracture in cyclic fatigue loading.
The material was a yttria and alumina additive sintered silicon nitride with an
elongated grain structure for enhanced fracture toughness. (a) shows the bend
bar fracture surfaces with the origin marked by the arrow. (b) shows a typical
fast fracture portion of the fracture surface. (c) shows the origin region and
fine debris (arrows) created by fatigue cycling. (courtesy T. Lube and R.
Danzer)
5-83
Fractography of Ceramics and Glasses
One subtle, but effective sign of genuine fatigue damage cyclic, is small debris
particles on the fracture surfaces of polycrystalline ceramics such as shown in
Figure 5.67, which are from Reference 61. The debris is small and detectable
only in the SEM. The backscatter mode may help detect such debris. The fine
debris can occur if the loading entails cyclic tension and compression, (e.g., the
R factor, which is the ratio of the maximum stress to the minimum stress, is
negative.) Dislodged or fractured grains near the crack tip may be crushed
between the crack faces if the crack faces go into compression. The debris may
or may not be present if the crack is loaded in cyclic tension-tension (e.g., the R
factor is zero or positive). One suspects that too zealous a cleaning of a
fracture surfaces beforehand might be harmful! On the other hand, it is
possible that the crushing may help the debris adhere to the fracture surface and
a light cleaning will not necessarily remove it.
Some component fracture analyses are complex, and an origin site may not be
immediately obvious. In such cases, the entire fracture surface may be
examined with the goal of observing the local directions of crack propagation.
Wake hackle, gull wings, Wallner lines or arrest lines as described in this
chapter provide such clues. A map of the overall fracture surface can be
constructed in the form of a large montage, around which key local closeup
photos of the crack propagation directions can be marked. Little arrows labeled
“dcp” can then be put on the overall image. The localized direction of crack
propagation can be “backtracked” to an origin site that might not have been
immediately obvious. A good example is Figure 5.68 for an alumina dental
crown which split into two pieces in a patient’s mouth. It was a relatively low
energy fracture that created only two pieces. The core ceramic microstructure
was not easy to interpret, but there were enough markings in it and the outer
veneer porcelain layer that it became obvious that fracture started at a margin.
Ko also showed marvelous montage examples for an alumina that broke from
fatigue loading.37 There were many tiny wake hackle lines from microporosity
in an otherwise rough fracture surface that he used to back track to the origin.
5-84
Fracture Surface Examination
A fracture map like this also has great value in communicating a failure
analysis. It illustrates how much work has gone into an anlysis, and it can also
provide context. It can be much more effective in convincing a client or reader
than merely showing a handful of closeup photos and relying on the client to
trust the fractograpgher’s judgement.
5-85
Fractography of Ceramics and Glasses
Chapter 5 References
5-86
Fracture Surface Examination
5-87
Fractography of Ceramics and Glasses
36. G. D. Quinn, L. K, Ives, and S. Jahanmir, “On the Nature of Machining Cracks in
Ground Ceramics: Part I: SRBSN Strengths and Fractographic Analysis,” Mach.
Sci. Technol., 9 (2005) 169 - 210.
37. J. Barsom, “Fracture of Tempered Glass,” J. Am. Ceram. Soc. 51 [2] (1968) 75 –
78.
38. H. Ko, “Fracture Behavior of Sintered Alumina Under Rotary Bending and Static
Fatigue,” pp. 15 – 29 in Fracture Mechanics of Ceramics 12, eds. R. C. Bradt, D.
P. H. Hasselman, D. Munz, M. Sakai, V. Ya Shevchenko, Plenum, NY, 1996.
39. F. Kerkhoff and E. Sommer, “Das Bild der Bruchfläche (von Gläsern),” in
Handbuch der Mikroscopie der Technik, Vol. IV, Mikroskopie der Silicate,
Umschau Verlag, Frankfurt am Main, Germany, 1964.
40. A. J. Pons and A. Karma, “Helical Crack-Front Instability in Mixed-mode
Fracture,” Nature, 464 (4 March 2010) 85 - 89
41. H. Wallner, “Line Structures on Fracture Surfaces,” Zeitschrift fur Physik, 114
(1939) 368 – 378.
42. A. Tsirk, “Formation and Utility of a Class of Anomalous Wallner Lines on
Obsidian,” pp. 57 -69 in Fractography of Glasses and Ceramics, Advances in
Ceramics, Vol. 22, eds., J. Varner, and V. Fréchette, American Ceramic Society,
Westerville, OH, 1988.
43. T. A. Michalske, “Fractography of Stress Corrosion Cracking in Glass,” pp. 111 –
142 in Fractography of Glass, eds., R. C. Bradt and R. E. Tressler, Plenum, NY,
1994.
44. T. A. Michalske and V. D. Fréchette, “Dynamic Effects of Liquids on Crack
Growth Leading to Catastrophic Failure in Glass,” J. Am. Ceram. Soc., 63 [11-12]
(1980) 603 – 609.
45. A. Tsirk, “An Exploration of Liquid-Induced Fracture Markings,” pp. 87 – 101 in
Fractography of Glasses and Ceramics, IV, Ceramic Transactions, Vol. 122, eds.,
J. Varner, and G. D. Quinn, American Ceramic Society, Westerville, OH, 2001.
46. A. Tsirk, Fractures in Knapping, Archaeopress, Oxford, UK, 2014.
47. V. D. Fréchette, “Markings Associated with the Presence of H2O During
Cracking,” pp. 71 – 76 in Fractography of Glasses and Ceramics, Advances in
Ceramics, Vol. 22, eds., J. Varner, and V. Fréchette, American Ceramic Society,
Westerville, OH, 1988.
48. T. A. Michalske, J. R. Varner, and V. D. Fréchette, “Growth of Cracks Partly
Filled With Water,” pp. 639 – 649 in Fracture Mechanics of Ceramics, Vol. 4, eds.
R. C. Bradt, D. P. H. Hasselman, and F. F. Lange, Plenum, NY, 1978.
49. V. D. Fréchette and M. Donovan, “Some Effects of The “Glue-Chipping” Process
on Strength,” pp. 407 –411 in Fractography of Glasses and Ceramics, II, Ceramic
Transactions, Vol. 17, eds., V. Fréchette, and J. Varner, American Ceramic
Society, Westerville, OH, 1988.
50. T. A. Michalske and S. W. Freiman, “A Molecular Mechanism for Stress
Corrosion in Vitreous Silica,” J. Amer. Ceram. Soc., 66 [4] (1983) 284-288.
51. T. A Michalske, B. C. Bunker, and S. W. Freiman, “Stress Corrosion of Ionic and
Mixed Ionic/Covalent Solids,” J. Amer. Ceram. Soc., 69 [10] (1986) 721 – 724.
5-88
Fracture Surface Examination
52. J. J. Swab and G. D. Quinn, “Effect of Precrack ‘Halos’ on KIc Determined by the
Surface Crack in Flexure Method,” J. Am. Ceram. Soc., 81 [9] (1998) 2261 - 2268.
53. J. J. Mecholsky, Jr., “Intergranular Slow Crack Growth in MgF2,” J. Am. Ceram.
Soc., 64 [9] (1981) 563-566.
54. R. Morrell, “Dealing with Component Failures,” pp. 353 -368 in Fractography of
Glasses and Ceramics V, eds., J. R. Varner, G.D. Quinn, M. Wightman, Ceramic
Transactions, Vol. 199, American Ceramic Society, Westerville, OH, 2007.
55. T. A. Michalske, “Fractography of Slow Fracture in Glass,” pp. 121 – 136 in
Fractography of Ceramic and Metal Failures, ASTM Special Technical
Publication 827, eds., J. J. Mecholsky, Jr. and S. R. Powell, Jr., ASTM, West
Conshohocken, PA, 1984.
56. T. A. Michalske, V. D. Fréchette, and R. Hudson, “Dynamic Effects of Liquids on
Surface Crack Extension in Glass,” pp. 1091 – 1097 in Advances in Fracture
Research, Vol. 2, ed., D. Francis, Pergamon, NY, 1981.
57. G. D. Quinn and W. R. Braue, “Fracture Mechanism Maps for Advanced Structural
Ceramics, Part 2, Sintered Silicon Nitride,” J. Mat. Sci., 25 (1990) 4377 – 4392.
58. G. D. Quinn, “Fracture Mechanism Maps for Advanced Structural Ceramics, Part I,
Methodology and Hot-Pressed Silicon Nitride,” J. Mat. Sci, 25 (1990) 4361 - 4376.
59. P. J. Dwivedi and D. J. Green, “Determination of Subcritical Crack Growth
Parameters by In Situ Observation of Indentation Cracks,” J. Amer. Ceram. Soc.,
78 [8] (1995) 2122-2128.
60. A. G. Evans, “Engineering Property Requirements for High Performance
Ceramics,” Mat. Sci. and Eng., 71 (1985) 3 – 21.
61. T. Lube, I Khader , A. Kailer, U. Degenhardt, and K. Berroth, “Ermüdungsbruch in
Siliziumnitrid,” Prakt. Met. Sonderband, 46 (2014) 353-358.
5-89
Origins of Fracture
6. Origins of Fracture
6.1 Origins, Flaws, and Defects
The origin is not only a location, but also an object. In the parlance of the
engineer or scientist, the irregularities that initiate fracture are termed flaws or
defects. The fractographer should use the terms “defect” and “flaw” with
discretion. Scientists and engineers understand that virtually all brittle
materials are imperfect and contain irregularities that can behave as flaws, but
nontechnical people may misunderstand these terms. That the material
contains flaws or defects does not necessarily mean that the material has been
prepared improperly or is somehow faulty. Conversely, there are instances
where defects or flaws do indicate defective material. The author usually avoids
using the word “defect” because of its similarity to the word “defective.”
“Flaw” seems a little less severe and sounds to many as a more technical term.
If there is any doubt, then the fractographer may fall back upon the most
innocuous term: “fracture origin.” Ceramists, engineers, and researchers
occasionally use the term “Griffith flaw:”
There are hundreds if not thousands of papers on flaws in ceramics and glasses.
Only a handful are cited in this chapter, but many other key papers and books
are listed in Appendix A, in the sections: “Origins in Ceramics” and “Origins
in Glass.” The contributions of Mr. R. Rice, Prof. K. Uematsu, and Dr. F.
Lange are particularly valuable. Starting in the 1990, Profs. Keizo Uematsu
and Satoshi Tanaka at Nagaoka University did fine work to characterize flaws
in ceramics, starting from the earliest processing stages and monitoring flaw
evolution during subsequent processing. They confirmed the link between the
flaw size and the strength distributions. This topic is discussed further in
Weibull Analysis section 7.15 of this Guide.
6-1
Fractography of Ceramics and Glasses
Fracture
surface
Figure 6.1 Schematics that show the four possible locations of a volume-
distributed flaw.
scale the strengths of laboratory test data to predict component strengths via the
Weibull theory. 1,2,3,4,5 Strength scaling and reliability predictions may be
dramatically different for area and volume scaling. Glasses usually, but not
always, break from surface-distributed flaws. Ceramics can fracture from
surface or internal flaws.
One aspect about the spatial distribution of flaws needs clarification as shown
in Figure 6.1. A single volume-distributed flaw may be located in the volume
(in the bulk), at the surface, near the surface, or at an edge. For example, an
inclusion originally may have been in the interior of a ceramic plate, but when
a test bar was cut out, the inclusion ended up on the surface of the test bar.
Surface-distributed flaws can only exist at the surface or at an edge. While the
above statements may seem obvious, there are two reasons to keep the
distinction clear. Firstly, if a flaw is found at a surface location, it does not
necessarily mean it is a surface type flaw. It could be a volume type flaw
located at the surface. Some causal analysts simply look at the locations of the
fracture origins. If they are located at the surface they simply interpret the
origin as a surface type flaw. This is often wrong, particularly if the test data
set is from bend bars or biaxially-loaded plates. It is a common misunder-
standing that these two configurations always break from surface flaws.
6-2
Origins of Fracture
Origins indeed are often located at the surface of these bending configurations,
but if reasonable precautions are taken to control the grinding damage, then
they are quite likely to break from volume-distributed flaws. The following
pages show many such examples. Secondly, even if the data are not intended
for design, a common mistake is to assume that all surface-located flaws are
machining damage.
d d2
d1
Fracture surface
ℓ ℓ
Specimen surface
Figure 6.2 Near surface (NS) origins are no deeper (ℓ) than the flaw diameter
or major axis length.
One might ask at what point is a volume type flaw “near surface.” A review of
the fracture mechanics stress intensity shape factors for flaws located at various
distances in proximity to a surface suggests the criterion shown in Figure 6.2.
At distances closer than one diameter, the stress intensity shape factor for such
flaws increase by more than a few percent. Sometimes the characterization of
location is important for service-performance issues or fracture mechanics
analyses of the flaws. For example, some near-surface located origins may be
more susceptible to time-dependent crack growth than equivalent volume-
located origins. A classic example of this for time-dependent fractures in
sintered silicon carbide is Case Study 3 in Chapter 10 of this Guide. Pores
located at or near the surface behaved differently than identical pores located in
the bulk. Near-surface-located origins may also be likely to link up with
surface machining or impact damage, or to extend subcritically to the surface
prior to catastrophic fracture.
6-3
Fractography of Ceramics and Glasses
(a) (b)
Figure 6.3 Wherever possible, describe flaws by what they are, not how they
appear. (a) shows a white spot at the origin in a reaction-bonded silicon
nitride bend bar. The SEM image of the origin in (b) shows the origin is a
pore. FS denotes fracture surface and T denotes tensile surface.
6-4
Origins of Fracture
(a) (b)
(c)
6-5
Fractography of Ceramics and Glasses
Fracture surface
(a)
External surface
(b)
Figure 6.5 The exterior surface should be checked if the flaw is located on the
surface. (a) shows an irregular, deep grinding flaw with a distinctly different
ground-surface trace at the origin in a reaction-bonded silicon nitride
component. It is very likely that a large “renegade” abrasive grit in the
grinding wheel caused this damage. (b) shows the fracture origin of a fused
silica rod broken in flexure. This tilted-back view shows a scratch on the
surface (arrow) coincident with the origin.
6-6
Origins of Fracture
Pores in glass or porcelain are often very smooth and round, and may contain
trapped gas. As such they may be called “bubbles.” Their effect upon strength
is very different. Since they are very smooth, they usually act as simple stress
concentrators and not as sharp flaws. This distinction is important as discussed
in detail in Chapter 7. A material may have large internal pores that are
relatively harmless, and they may not necessarily act as fracture origins. On
the other hand, they can be extremely deleterious if they contact an exterior
surface or each other, as illustrated in Figure 6.7. Bubbles that are close to an
outer surface are also prone to interact with grinding or machining damage that
can link with the bubble, turning an otherwise harmless bubble into an
extremely serious flaw. A good example of this was shown as Figure 5.5 in the
previous chapter. Figures 6.8 and 6.9 show other examples of bubble origins.
Reference 7 has more information about the feldspathic porcelain strengths and
bubble/pore fracture origins. This dichotomy in behavior poses a dilemma to
design engineers working with flaw sizes and Weibull statistics. A volume-
distributed bubble can be either harmless or a dominant flaw depending upon
where it is located in a piece.
6-7
Fractography of Ceramics and Glasses
(a) (b)
(c) (d)
(e) (f)
Figure 6.6 Examples of pore fracture origins in ceramics. (a) shows a SEM
image of a pore in a 99.9% alumina bend bar. Electron charging created the
bright spot. (b) shows a pore in an aluminum oxynitride bend bar that is within
5 µm of the surface. This is a “near surface” location example. The
transgranular fracture near the pore helps the pore stand out very clearly. (c) is
an irregular pore in a sintered alpha silicon carbide bend bar. (d) is a pore in a
porcelain electrical insulator (courtesy J. Taylor and R. Rice). (e) is a
cylindrical pore in a silicon nitride probably from a fiber or hair burn out. (f) is
a pore in a yttria-stabilized tetragonal zirconia polycrystal (3Y-TZP) (f is
courtesy J. Quinn).
6-8
Origins of Fracture
(a) (d)
Exterior surface
(b) (e)
Exterior surface
(c) (f)
Exterior surface
(g)
Figure 6.7 Bubbles are smooth pores and may be relatively harmless or
extremely deleterious depending upon location. (a) shows a bubble just
“kissing” the exterior surface, creating very sharp cusps from which cracks
will likely pop in and grow. (b) shows a near-surface bubble which can easily
link up with grinding or other surface damage. (c) shows an interior bubble
that is relatively harmless. Similar behavior can occur with two bubbles in
proximity to each other. (d) and (e) are likely to form a strength-controlling
flaw, whereas configuration (f) is relatively harmless. (g) is an example of (a)
in alumina. (courtesy T. Bernthaler)
6-9
Fractography of Ceramics and Glasses
(a)
(c)
(b)
Figure 6.8 Bubble origin in a mock-up of a dental crown. Dental stone is cast
into a cup shape that is loaded in a vise with an inner rubber stopper. When
the vise is tightened, the stopper expands and the “crown” splits in half. Blue
tape holds the fragments together. Hoop tensile stresses in the wall triggered
fracture from a bubble touching the inner surface (c). Grazing illumination
accentuates hackle lines. Fracture started from the cusp marked with a black
arrow. These mock-up crowns simulate real crown fractures.
(a) (b)
(c)
6-10
Origins of Fracture
20 µm
200 µm
(a) (b)
(c)
20 µm
Figure 6.10 Internal pore-bubbles in ceramics and glass ceramics. (a) and
(b) show an internal spherical pore fracture origin in a cordierite glass
ceramic bend bar (315 MPa). The pore was from a bubble in the original
glass. The material crystalized around the bubble during the ceraming step.
The arrows mark the hackle boundary of the fracture mirror. (courtesy D.
Lewis, III) (c) shows a large internal seeded-pore origin in alumina.
(courtesy T. Bernthaler)
Pores may often be very irregular and jagged. One way these may occur is by
the formation of pressing defects when preparing the green body. Spray-dry
powder agglomerates may not be eliminated during final pressureless sintering.
Figure 6.11 shows examples. Hot pressing or hot isopressing would likely
eliminate such flaws. These flaws are sometimes called: “pressing flaws” or
“blank pressing flaws.”
6-11
Fractography of Ceramics and Glasses
(a) (b)
(c) (d)
(e) (f)
Figure 6.11 Irregular pore origins. (a) through (d) were formed during cold
pressing of alumina spray-dried powders. (courtesy T. Bernthaler).
(e) and (f) are in a sintered 3Y TZP zirconia, (courtesy S. Scherrer).
6-12
Origins of Fracture
(a) (b)
(c) (d)
(e) (f)
6-13
Fractography of Ceramics and Glasses
(a) (b)
(c ) (d)
Figure 6.13 Examples of porous seam fracture origins. (a) and (b) show the
origin in a fine-grained sintered 99.9% alumina. (c) and (d) show the origin in
matching halves of another fine-grained sintered 99.9 % alumina bend bar.
6-14
Origins of Fracture
6.6.4 Agglomerates
(a) (b)
(c)
6-15
Fractography of Ceramics and Glasses
(a)
(b) (c)
6-16
Origins of Fracture
6.6.5 Inclusions
Inclusions are also volume-distributed flaws that are a foreign body with a
composition different than the normal composition of the glass or ceramic.
They are often easy to detect due to color or reflectivity differences compared
to the matrix. Sometimes they are round and equiaxed, other times they are
odd shaped or fragmented. Figures 6.16 and 6.17 shows some examples.
(a) (b)
(c) (d)
(e) (f)
6-17
Fractography of Ceramics and Glasses
(a) (b)
(c) (d)
6-18
Origins of Fracture
Not all inclusions behave the same. Their effect depends upon whether their
elastic and thermal properties match or mismatch those of the matrix. 15
Inclusions can cause cracking in the matrix, may crack themselves, or may
detach and pull away from the matrix creating a void like flaw. Figure 6.18
from a paper by Evans15 illustrates the possibilities.
6-19
Fractography of Ceramics and Glasses
(a) (b)
(c) (d)
(e)
6-20
Origins of Fracture
(a) (b)
(c) (d)
6-21
Fractography of Ceramics and Glasses
(a) (b)
(c)
Large grains can occur due to local exaggerated grain growth. They may form
due to a slight compositional or density variations during sintering. The
compositional variability can be from an impurity or a sintering aid. Large
grains can be a single or a cluster of grains having a size significantly greater
than the range of the normal grain size distribution. They can often be seen
with the optical microscope due to a difference in reflectivity compared to the
matrix. Examples are in Figure 6.22. Rice’s book 16 on grain size and
mechanical properties of ceramics has a good discussion of large grains as
fracture origins on pages 13 – 16.
6-22
Origins of Fracture
(a) (b)
(c) (d)
(e) (f)
Figure 6.22 Examples of large grain fracture origins. (a,b) show a cluster of
large grains at a chamfer in a 99.9% alumina bend bar. (c,d) show an over-
sized grain in a silicon nitride. The material had small interlocking elongated
grains intended to enhance fracture toughness. Notice how these are legitimate
material flaws, and the chamfer preparation was completely satisfactory.
(e, f) show examples of acicular large grains in sintered alpha silicon carbide.
6-23
Fractography of Ceramics and Glasses
Large quartz grains can act as strength-limiting flaws in porcelains. Quartz has
several polymorphs and displacive transformations during cooldown from
processing temperatures can cause cracking around and within the large quartz
grains as shown in Figure 6.23. This is referred to as the β - α quartz inversion.
If the quartz grains are much larger than the normal microstructure, they may
stand out clearly on a fracture surface. If the quartz grains are similar in size to
the normal microstructure, then the origin flaw may blend into the rough
fracture surface and be difficult to discern. Good examples have been shown
by Gee and Morrell, 17 Carlstrom and Carlsson, 18 Southan 19 (who had superb
high-quality glossy color-prints), and Banda and Messer. 20
(a)
(b)
(c)
Figure 6.23 Cracking around and through quartz grains. (a)-(c) are from a
porcelain mortar that was broken and are courtesy R. Morrell. (a) shows a
polished and etched microstructural view. (b) shows an SEM image of one
fracture origin (circled). (c) is a close-up SEM image showing microcracking
around quartz particles. (d) – (f) are from porcelain dental crowns, courtesy of
D. Southan.19 The thin-section polarized-light color image (d) reveals a
reaction zone and microcracking around the quartz grains.
6-24
Origins of Fracture
Grain boundaries may be vulnerable areas and, if the grain size is large enough,
the boundaries can act as strength limiting flaws. Such flaws are very difficult
to unequivocally identify as origins since the fracture path may be intergranular
and follow other grain boundaries thereby creating a very rough fracture
surface. There are many papers on the effect of grain boundary cracking upon
strength, but few actually show a grain boundary origin. Davidge and Tappin 21
showed grain boundary origins in alumina.
Sometimes the grain boundary cracking occurs from internal strains set up from
differential contractions in anisotropic, noncubic ceramic grains. In other
cases, impurities or sintering aids can gravitate to grain boundaries and weaken
them. For example, Salem22 found tiny alumina particles at grain boundaries of
coarse-grained MgAl2O4 cubic spinel transparent plates. These so weakened
the material that jagged crack paths formed and debris separated from the
material upon fast fracture. Fracture origins were either large grains or the
grain boundaries in disk strength specimen. Swab et al.23 also studied a coarse-
grained spinel which had irregular crack surfaces and even had granular debris
generated during fracture. Lithium and fluoride from the sintering aid
concentrated at the grain boundaries and dramatically weakened the material.23
I examined one large (40 cm square) plate made of this material and there was
so much grain boundary cracking that green dye used to stain the fracture
surface bled profusely into the bulk. Grains literately fell out of the fracture
surface while I was doing this.
6-25
Fractography of Ceramics and Glasses
(a) (b)
(c) (d)
(e)
6-26
Origins of Fracture
(a) (b)
(c) (d)
Figure 6.25 Examples of surface void fracture origins. (a) and (b) are in as-
fired bend bars of injection-molded sintered silicon nitride. (c) and (d) show a
surface void in an as-fired silicon nitride component with yttria and alumina
sintering aids. (d) is tilted so that the outer surface is visible. EDX analysis
showed an excess concentration of aluminum at the site, so the flaw also has
characteristics of a compositional inhomogeneity. (a, b are courtesy A. Pasto;
c,d are courtesy T. Lube, W. Harrer, and R. Danzer).
6-27
Fractography of Ceramics and Glasses
Pits are surface cavities that may form as a result of a reaction between a
ceramic or glass and the usage environment. Corrosion or oxidation can create
pits. Examples are shown in Figure 6.26. Richerson 24 shows several good
examples of oxidation and corrosion pits in silicon nitride. Pits have chemistry
or structural differences that make them distinguishable from pores or surface
voids. Sometimes the damage can be confined to grain boundaries and the
flawed regions can be difficult to discern on a fracture surface.
(a) (b)
(c) (d)
Figure 6.26 Examples of pit fracture origins. (a) – (c) are in oxidized hot-
pressed silicon nitride. (b) is a close-up of (a). (d) is a reaction-bonded
silicon nitride.
6-28
Origins of Fracture
6-29
Fractography of Ceramics and Glasses
Side surface
scratch
(a) (b)
tensile surface
(c)
fracture surface
Figure 6.27 Examples of scratch origins with the scratch at an angle to the
tensile stress. The cracks penetrate much deeper than the scratch groove
depth. Scratches can create damage that may be mistaken for grinding cracks,
so it is wise to look at the orientation of the scratch relative to the grinding
striations. Scratches will be at irregular angles and are often isolated. (a) is
in a SiAlON bend bar with a scratch near the chamfer. (b) is a side view of a
reaction-bonded silicon nitride bend bar. Scratches often require illumination
at just the right angle to be seen. (c) shows a scratch on an alumina bend bar
surface (white arrows) and the continuation of it into the interior that acted as
the fracture origin in a bend bar (black arrows) (c is courtesy J. Swab).
6-30
Origins of Fracture
Scratch/dig specifications are used with visual inspections of the outer surfaces
of glass or transparent ceramic materials. The older MIL-PRF-13830B (1997)
scratch specification is based on a visual comparison to reference pieces. The
number is sometimes connected to the width of the scratch surface trace, but
the literature and web sites are contradictory about the actual values. A #40
scratch sometimes is listed as a 4 µm wide scratch and in other cases as a 40
µm wide scratch. The dig specification is more consistent, and is the diameter
of a spot such as is caused by grinding, impact, or contact damage (see 6.7.4
and 6.7.5 below). Unfortunately, the outer surface features often have no
correlation to the extent and size of cracks that may, or may not, lie underneath.
(a) (b)
(c) (d)
Figure 6.28 Two examples of scratch crack fracture origins with the tensile
stress normal to the scratch. (a) and (b) show the origin in a polished BK-7
borosilicate crown glass disk broken in ring-on-ring flexure. The origin is a
very long, but smooth 7.3 µm deep crack. Figure 5.15 shows the fracture
mirror. (c) and (d) show matching SEM and optical images of an origin in a
sapphire dome that broke in a high-temperature wind tunnel test rig. There
were no traces of the scratches on the outer polished surface in either case.
6-31
Fractography of Ceramics and Glasses
When scratches are suspected, it is usually a good idea to look for other
scratches elsewhere on the exterior surface of the part. Figure 6.29 shows an
example where scratches controlled strength in an engine part.
(a) (b)
(c)
6-32
Origins of Fracture
(a) (b)
(c) (d)
Figure 6.30 Examles of sharp particle imact origin. in glass. The direction of
impact is shown by the arrows. All of the impacts have created penetrating
median – radial cracks. (b) and (d) have pronounced lateral cracks as well. In
each case the impact site has been pulverized or crushed, unlike blunt impact
sites.(a-d courtesy J. Varner).
6-33
Blunt impacts or blunt contact loadings produce round Hertzian cone cracks as
illustrated in Figure 6.31. These may be complete or partial rings as shown in
Figures 3.57 (a C-crack in a ball bearing), and 6.32, 6.33 – 6.35. The cone
crack initiates as a ring just outside the footprint of the two contacting bodies as
shown in Figure 6.31a. The size of the contact area depends upon the load,
geometries, and elastic properties of the two materials. A simple rule of thumb
is that the contacting object is almost always several times larger than the
observed ring size on the contacting surface as shown in Figure 6.31. Of
course, if the blunt object penetrates through the plate and makes a large hole,
then the impactor may be comparable in size to the hole. When contact
damage is detected, the exterior surface should be examined for witness marks
at the contact site, as well as other nearby impacts.
Blunt contact cracks often leave little trace on an outer surface. A pretest
inspection for “digs” may not find contact damage cracks. The contact cracks
might be revealed prior to fracture by very careful examination with adjustment
of lighting angles. After fracture, they are easier to detect since they leave a
small telltale arc on the fracture surface plane at the origin site.
Partial cones
behind contact
Impact
Impact
Oblique 55o
250 m/s
150 m/s
skids
(d) (e)
Figure 6.31 Blunt contact or blunt impact cone crack profiles. These are side
views. Notice the blunt impactor is larger than the size of the contact circle
and first ring to form at the impact site.
6-34
Origins of Fracture
Ring cracks initially pop in normal to the surface but then turn and propagate
into the depth with an included angle of 125o to 135o for quasi static loading 1 in
glasses as shown in Figure 6.31a. The included cone angle decreases
markedly with increasing impact velocity as shown in Figure 6.31d. It may be
possible to estimate the projectile velocity from the included cone angle.25,26,27
For example, at 250 m/s, 0.8 mm to 1 mm diameter steel balls created 60o to
80o cone cracks in glass25 and 70o to 90o cracks in silicon carbide.27 During
quasi static or impact loadings, as load increases the footprint area may expand
and generate multiple concentric ring and cone cracks. Kirchner and Gruver 28
have described interesting variations of cone cracks from impact, whereby
cracks can nucleate on one side, then branch, and then even spiral around the
contact site.
If a ball that is loaded perpendicular to the surface is slid over the surface at
quasi static rates, the cone axis tilts such that the rear of the cone dips down
deeper, and the front portion of the cone disappears (Figure 6.31c). 29 Partial
cone cracks are the result. This is not surprising since the forward motion of
the ball creates an additional tensile stress behind the ball, but adds
compressive stresses ahead of the ball.
Conditions are different if the contactor is loaded (or strikes at) at non-
perpendicular angles. Full cones form, but they tilt so that the backside
approaches the surface while the front side penetrates deeper into the surface as
shown in Figures 6.31b,e. 30
Most blunt contact loadings produce a cone crack that penetrates partially or
completely through a plate or body as shown in Figure 4.17. These may not
cause component breakage and may merely be a cosmetic fault unless the
structure is intended to seal the environment. Subsequent loading in tension or
flexure may cause new crack extension from the cone crack as shown in
Figures 6.32 (d) and (e). Many structures such as dental crowns may have
numerous harmless cone cracks in the veneers. They are only revealed when a
fracture that commenced elsewhere passes through the cone (Figure 6.32f).
Impacts at moderate or high energy create radial cracks and bending stresses
that can propagate cracks far beyond the impact site and any cone cracks,
thereby causing component fracture. Figure 6.34b shows an example.
1
Stationary or with slow motion.
6-35
Fractography of Ceramics and Glasses
tensile surface
fracture surface
(a) (b)
outer surface
fracture
surface
outer surface
(c) (d)
outer surface fracture core ceramic fracture surface
surface
veneer
outer surface
(e) (f)
Figure 6.32 Examples of blunt impact or contact crack damage fracture
origins. (a) shows the tensile surface of a silicon nitride bend bar with partial
crack rings that were created by a ball-tipped micrometer. Such rings were
fracture origins in some other bend bars. (b) is a cone crack fracture origin
from a blunt contact or impact on a rectangular bend bar. (c) shows a blunt
contact crack from a loading roller on a cylindrical rod glass flexure specimen.
Notice that the ring is incomplete. The cone crack can act as a strength-
controlling flaw if loaded in tension as shown in (d) and (e). (f) shows a cone
crack in a dental bilayer crown. The cone crack, which initiated from contact
on the veneer layer, was harmless. Fracture initiated elsewhere and a crack
simply passed though the cone crack.
6-36
Origins of Fracture
(a) (b)
(c) (d)
(e) (f)
Figure 6.33 Blunt impact origins in electrical insulator porcelains. (a) shows
the installed insulators. (b) and (c) are blunt impact damage from “BB” gun
projectiles. (d) is damage from a sling shot projectile. (e) and (f) are impact
origins from contact with support structures. Note the partially exposed cone in
(f) (arrow). (Photos courtesy of J. J. Taylor and R. W. Rice)
6-37
Fractography of Ceramics and Glasses
(a)
(b)
Figure 6.34 Blunt impact origins in glass plates. (a) shows a mild – moderate
impact with cone cracks. The plate did not break into pieces. (b) is a severe
impact that was sufficient to create cone and radial cracks. The latter caused
plate fracture. (Specimens courtesy of J. Varner and V. Fréchette, Alfred
University)
6-38
Origins of Fracture
dye penetration. It was a partial cone crack. They are often formed by ceramic
balls impacting each other! Manufacturers wisely started to package ceramic
balls more carefully once the problem was recognized.
500 µm 500 µm
500 µm
abrasion damage
syringe outer surface syringe outer surface
(d) (e)
Figure 6.35 Blunt contact damage in pharmaceutical glass. (a) and (b) are a
vial. The pictorial (a) schematic shows a possible impact direction and is a
useful aid for explaining a fracture event. (c) - (d) are three glass syringes.
When arc-shaped contact cracks propagate to fracture, they often have hackle
lines that emanate from the flaw. Notice how these tilt in opposite directions
on the flaw sides indicating that fracture radiated outwards. Such hackle at an
origin site suggests “eyelashes.” Notice the witness mark of abrasion damage
in (d). (Photos courtesy of F. Maurer)
6-39
Fractography of Ceramics and Glasses
Surface grinding procedures use grinding wheels as shown in Figure 6.36. The
grinding process entails material removal, often by means of microcrack and
microchip formation. By its very nature, cracking is introduced into the
surface. Initial grinding is done with coarse-grit wheels and modest depths of
cut in order to facilitate material removal. Intermediate- and final-stage
grinding is done with finer-grit wheels and smaller depths of cut to reach final
part dimensions and surface finish requirements, but also to remove prior
damage while introducing a minimum amount of new damage.
6-40
Origins of Fracture
6-41
The differences between the orthogonal and parallel cracks give rise to strength
dependence with orientation (if grinding cracks are the controlling flaws).
Only one striation groove is shown in the schematic and it should be
remembered that the ground surface actually contains a complex overlapping
network of such cracks and grooves.
With some experience it is not too difficult to find parallel grinding crack flaws
in dense materials, such as glass, single crystals, and either very coarse-grained
or extremely fine-grained ceramics. Detection of grinding cracks in porous,
heterogeneous, or mid grain-sized materials is very difficult, however, since the
machining crack blends in to the background microstructure. This is especially
true for the short orthogonal cracks. They are hard to find in nearly all
materials. Figure 6.37 shows some examples of orthogonal cracks, which
sometimes are described in the literature as “longitudinal-grinding flaws.”
(That expression means: “the flaws that are created by longitudinal grinding”
with respect to the specimen’s long axis.) Additional examples are shown in
some of the papers by Rice and Mecholsky listed in the Bibliography,
Appendix A.
Parallel machining cracks are much easier to detect and often can be found with
routine optical microscopy, but only if low-angle grazing illumination is used.
Figure 6.38 shows examples. Figure 6.39 shows a schematic of these parallel
flaws and their telltale signs. Figure 6.40 shows how “V” machining cracks
form in ground round parts. Several of these grinding crack flaws exhibit
another form of hackle on fracture surfaces:
Grinding crack hackle is sometime also known as “shark’s teeth.” The size of
grinding cracks, and hence their effect on strength, depends upon material
properties such as hardness, elastic modulus, and fracture resistance or fracture
toughness. It also depends on the grinding wheel and machine conditions such
as abrasive type, grit size and concentration, depth of cut, wheel speed,
grinding forces and machine horsepower. The crack size also depends upon
the material’s grain size. Deep grinding cracks can form in very coarse-grained
ceramics since the single crystal fracture toughness is less than the
polycrystalline fracture toughness. It is beyond the scope of this guide to delve
into this topic other than to say that abrasive grit size is the dominant parameter
that controls crack size for typical surface grinding procedures. Figure 6.40
shows the effect of grit size upon parallel crack depth in a commercial silicon
nitride. 32
6-42
Origins of Fracture
(a)
(b)
6-43
Fractography of Ceramics and Glasses
(a)
(b) (c)
(d) (e)
Figure 6.38 Examples of parallel grinding cracks. (a) and (b) show optical
images of a “zipper crack” in a sintered silicon nitride bend bar. (d) and (e)
show SEM images of a crack in a silicon nitride rod.
6-44
Origins of Fracture
(f)
(g)
Figure 6.38 (cont’d) Examples of parallel grinding cracks. (f) shows back-to-
back halves of a silicon nitride bend bar with coarse grinding (150 grit) cracks.
A zipper crack is at the origin (large white arrow). The small black arrows
show that there also is a machining crack “skin zone” along the entire surface.
(g) is an alumina bend bar with low-angle grazing illumination from the side
revealing a zipper crack.
6-45
Fractography of Ceramics and Glasses
specimen surface.
from the links or overlaps of the flaws and extend up into the fracture
mirror. These tails may be tilted to the left or right and help confirm that
fracture originated in the central region of the set. The short tails are
machining cracks (or a scratch) and are often easier to see with an
optical microscope with low angle lighting than with a scanning electron
cracks. Larger tails than in (c ) are created. In severe cases, the tail
may extend all the way to the mirror boundary. The fracture mirror may
be elongated. This origin is common in transversely ground or
occurs in the fracture mirror. One or two (shown) tails extend well up
into the fracture mirror. The machining direction is not quite perpen-
dicular to the specimen length and uniaxial stress axis due to grinding
The origin is a deep machining crack that extends along the entire
surface. The origin is often bumpy since the origin is comprised of offset
parallel cracks. Thin bands of uniform depth extend along the specimen
surface on either side of the fracture mirror. The bands have the same
which may be in the thin bands are tilted away from the origin. This
origin type is common in coarse ground surfaces.
6-46
Origins of Fracture
(a) (b)
Ground or lapped
surface
(c) (d)
Figure 6.40 Round parts may have “V” machining cracks formed from small
parallel crack segments a, b, c on different planes as shown in (a) and (b) for a
silicon nitride rod. The steps between the individual crack segments form two
or more telltale hackle lines, which extend up into a mirror. Examination of
the ground surface shows striations (grinding grooves) that have a small angle
relative to the rod axis. V crack flaws are easy to see even at very low
magnification with low-angle grazing illumination.
6-47
Fractography of Ceramics and Glasses
Figure 6.41 The depth of grinding cracks as a function of grinding wheel grit
in a silicon nitride. The notes show the strength-controlling flaws. Ref. 32
The average grit size may not necessarily be the controlling parameter.
Instances of renegade abrasive grits in a grinding wheel that caused atypical
damage have been reported.32 Figure 6.5a earlier in this chapter and Figure
6.42 below show an example of damage from a renegade grit in a silicon
nitride. Calculations based on the wheel and table speeds and the specimen and
wheel diameters confirm that a single grit can cause such long striations.
Striation here means a grinding groove.
Although it is widely believed that materials with rising R-curves form smaller
grinding cracks than materials with flat R-curves, there is evidence that just the
opposite is true. 33 Materials with rising R-curves are more difficult to grind,
but for a given wheel speed and depth of cut, greater grinding forces are
generated to overcome the removal resistance. The greater forces cause cracks
to penetrate deeper than in materials with low fracture toughness or flat R-
curves that cut or grind more freely.
6-48
Origins of Fracture
4 mm
(a) (b)
(c) (d)
6-49
Fractography of Ceramics and Glasses
Machining cracks often link up with each other or with other materials flaws as
shown in Figure 6.43. We tend to focus only on the ones that we see on a
fracture surface, and it should be borne in mind that the surface actually
contains a complex network of cracks as shown in Figure 6.44.
Figure 6.43 Machining cracks can link with each other or with material flaws.
6-50
Origins of Fracture
ground
ground
SEM view
6-51
Fractography of Ceramics and Glasses
(a) (b)
(c)
Figure 6.46 Fracture surface of a feldspathic porcelain bend bar. (a) shows
the entire surface of the 3 mm x 4 mm bar with a gold coating. The origin is on
the right side wall (arrow). (b) and (c) show SEM images of the damage along
the right side wall. Notice the slightly greater roughness of the fracture surface
along the right side as compared to the fracture mirror centered on it.
Careful examination of the exterior surfaces of the porcelain bars revealed that
the bottom tensile surface had been polished adequately to minimize residual
grinding damage. The tensile surface probably had been initially ground in a
longitudinal direction. The side walls on the other hand had residual grinding
damage from transverse grinding. Although the side walls were also polished,
it was not done deep enough and the damage was not completely removed.
Most porcelains have a rough fracture surface due to the microstructure, and
such distinct fracture mirrors and grinding damage may be obscured. This
particular porcelain, Vita Mark 2, is an exception due to its fine-grain size.
6-52
Origins of Fracture
6.7.7 Chips
Blunt or sharp impacts or contact loadings near an edge can chip off a piece as
shown in Chapter 4. The chip can in turn become a strength limiting flaw on
subsequent loadings. Figure 6.47 shows examples.
(a) (b)
(c) (d)
Figure 6.47 Examples of chips that acted as origins. (a) is an edge chip that
acted as the flaw in a bend bar. This view shows the reassembled halves from
the side. The top arrow shows the impact point that first created the chip.
When broken in flexure, the fracture initiation site was found to be at the small
black arrow. Notice the continuity of the chip shape on each side of the final
fracture. This confirms that the chip was in the bar to begin with, and was not
a secondary fracture. (b) is an edge chip in a sintered silicon carbide bend bar
chamfer. (c) and (d) show chips on bend bar chamfers in hot-pressed silicon
nitride. Close examination of the chamfers revealed that the grinding was not
done longitudinally, but at an angle causing the chips. Notice the striation
marked by the white arrow on the chamfer in (d).
6-53
Cracks that form in the green state or during firing are quite variable and
troublesome. Firing stresses can be very irregular, may be transient, and may
depend upon thermal gradients in the kiln or furnace. Firing or shrinkage crack
examples are shown in Figures 6.48 - 6.52.
Cracks that meander, are segmented, and/or are granular on the fracture
surface may have formed in the green state or during firing. Since the body is
not fully dense, the fracture surfaces are very granular and do not have the
characteristic fracture surface markings. Close examination of interior crack
surfaces may reveal thermal rounding of the particles and grains.
(a) (b)
tensile surface
tensile surface
(c) (d)
Figure 6.48 Examples of processing cracks. (a) shows the tensile surface and
(b) the fracture surface of two different silicon nitride bend bars which had an
unstable boundary phase associated with yttrium silicates. (c) shows a bend
bar that had a moderate strength despite a huge processing crack running
perpendicular to the tensile axis. The origin was at the arrow. (d) shows
another example of a firing crack in a bend bar. The origin is marked by the
arrow.
6-54
Origins of Fracture
(a) (b)
(c)
separated surface
(d) (e)
Figure 6.49 Processing cracks in dental ceramics. (a) and (b) show one tooth
of a three-unit alumina dental bridge. (a) shows a frontal (buccal) view of the
end unit which fractured at the narrow connection to the next unit (small white
arrow). A sintering crack is evident on the face of the crown (large white
arrow). It did not cause fracture in this instance because it was in a low stress
area. (b) shows a close-up of the processing cracks. (c) – (e) show a broken
zirconia crown. A large piece has separated off a cusp on the crown top
(occlusal) surface (c) and (d). The apparent “fracture surface” is in fact a
huge processing crack. Portions of the crown separated during firing and what
is seen are exposed spray-dry agglomerates. The manufacturer claimed the
fracture was due to patient overload or faulty installation by the dentist. This
was wrong, since the CAD/CAM blank had the processing crack in it. (c - e
are courtesy Dr. S. Scherrer and the ETH, the University of Zurich.)
6-55
Fractography of Ceramics and Glasses
(b)
(c)
(a)
(d) (e)
6-56
Origins of Fracture
(a) (b)
On the other hand, cracks that are sharp and planar (not rough and granular)
may have popped in during or after cool down from the firing temperatures.
Such cracks may have fracture surface markings such as hackle or arrest lines.
The microstructure will look fully dense with no roughness or rounding.
Figure 6.52 shows such an example. Sometimes these are termed “thermal
cracks.”
Varner and Fréchette 36 identified dunt cracks in whitewares. These are cracks
that from during cool down when quartz particles undergo a sharp change in
volume (due to the β - α phase change) when matrix glass is viscoelastic. The
dunt crack appears as an atypically shiny transgranular fracture.
6-57
Fractography of Ceramics and Glasses
Starting point
firing crack
(a)
final fracture
(b) (c)
Figure 6.52. A sharp firing crack in a dental prosthesis. This is a large single
piece zirconia five-unit “telescoping denture,” a single-piece insert for a
patient with substantial tooth and jaw bone loss. Two implants held it in place
at the left end and the fourth tooth in this view. The cantilevered fifth molar
tooth broke off after only three months. (a) shows the four surviving units and
(b) shows the end tooth that broke off. (c) shows the same piece with a gold
coating to make optical examination easier. The origin of the final fracture
was the large firing crack that propagated almost two-thirds through the
massive zirconia cross section. This sharp, smooth and probably very-tight
crack was unnoticed during fabrication and installation in the patient. It
started from irregularities in the veneer and core area marked by the yellow
arrows. Final fracture occurred on a different plane and in stages as attested
to by multiple arrest lines. One normally does not expect fracture in such a
massive zirconia piece, but the cantilever loading on the end piece which had
an enormous initial crack caused it to break. (courtesy M. Karl, U. Lohbauer,
J. R. Kelly) References 37 and 38.
6-58
Origins of Fracture
The Knoop indenter produces a much more controlled “cleaner” crack than the
Vickers indenter. The Knoop indentation procedure for creating controlled
flaws is described in more detail in section 7.12. When used to measure
fracture toughness with bars in bending, the method is known as the surface
crack in flexure (SCF) method. Figures 3.26, 3.56, 5.23, 5.55, 5.57, and 6.53
show examples of Knoop surface flaws that were fracture origins in bend bars.
(a) (b)
Figure 6.53 Examples of nice, clean Knoop indentation flaws. (a) shows a
precrack in a sintered alpha silicon carbide bend bar. (b) shows a BK-7
borosilicate crown glass bend bar. (c)-(h) are on the next page
6-59
Fractography of Ceramics and Glasses
(c) (d)
(e) (f)
(g) (h)
Figure 6.53 continued. (c) and (d) are optical and SEM images of cracks in
a sintered reaction-bonded silicon nitride. (e)and (f) are an e.max CAD dental
lithium dislicate glass ceramic with optical green dye(e) and SEM (f) images
of the same crack. (g) is a dental porcelain designed to be fused to zirconia,
and (h) is a Vita Enamic ceramic network, resin-infiltrated composite.
6-60
Origins of Fracture
Examples of glass origins have been shown previously in this chapter: nickel
sulfide inclusion in Figure 6.17a, scratches in Figures 6.5b, polishing-grinding
cracks in Figure 6.28a,b, impact and contact cracks in Figure 6.30, 6.32 and
6.35, and artificial flaws in Figure 6.53. Other flaw types that are common in
glasses are shown below. Varner 39 and Fréchette 40 have described the
character of flaws in glass.
Flaws from the melt may come in several forms. These flaws are inclusions,
pores, or compositional inhomogeneities, but have been given more specific
names by the glass community.39,40,41,42,43 Devitrification inclusions or
devitrification stones (Figure 6.54) can form if the glass crystallizes. Stones
are grains of unreacted raw material or refractory material broken off the tank
or pot walls. Devitrification stones and stones have completely different
chemistries and causes, so they should not be confused. Gas bubble origins are
termed blisters if large, and seeds if small. Striae are streaks of
inhomogeneous, transparent glass. Their refractive index is different than the
glass bulk. Very intense striae are termed cords and may appear as ropes or
strands such as shown in Figure 6.56.
6-61
Fractography of Ceramics and Glasses
(a)
(b)
(c)
6-62
Origins of Fracture
cord
(a) (b)
stone and cord
(c)
Figure 6.56 Flaws in a borosilicate glass cookware. (a) – (c) show a chord.
(crossed polarizers highlighted the internal strains in views (a) and (b). (c) is a
close-up of the cord with a mix of transmitted and reflected illumination. The
striae - stringers are marked by arrows. (d) shows a close-up of a stone with
striae - stringers. (e) is a seed (a bubble). (f) is an odd internal hairpin flaw.
Preston42 described heat sleeks, which are burnished streaks of glass where a
metallic polishing or grinding wheel makes direct rubbing contact with a glass
surface if the loose abrasive and water are inadvertently squeezed out. The
glass locally reaches a high temperature and develops a number of hair-like
fissures entering the glass normally to the surface and penetrating only a short
distance. Surface contact damage flaws were previously presented in section
6.7.4. Sharp contact damage sites are pecks.
6-63
Fractography of Ceramics and Glasses
(a) (b)
(c)
(d)
Figure 6.57 Chill check cracks on the outer surface of unbroken glass plates.
(a) and (b) show the same chill check. (c) and (d) show two close-up views of
another chill check. Note how different illumination can dramatically alter the
appearance and apparent extent of the flaw. Much more of the cracking is
evident in (d). (Specimens courtesy V. Fréchette and J. Varner)
6-64
Origins of Fracture
Figure 6.57 shows chill checks, a pernicious flaw that is often very difficult to
detect. A chill check is a small wavy and tight crack that is introduced into a
hot glass surface by localized contact with a cooler object. Thermal stresses
create the crack, but since the contact and thermal stresses are localized, the
cracks are small. Glasses with high thermal expansion coefficients are most
susceptible. The cracks often are difficult to detect in an unbroken component
and require lighting at just at the right angle. Sometimes transmitted
illumination is effective. If chill checks are suspected, then the piece or the
illumination source should be manipulated a variety of ways and angles so as to
try to get a momentary reflection from the crack. Chill checks form quickly
and close back up very quickly. They may close so tightly (less than the
wavelength of light) that they are not discernable. Nevertheless, some portions
of the crack may have captured small amounts of gasses or water molecules
that create slight opening pockets that are just barely detectable. These are
often brightly colored, since they refract the light and may only be detectable
from certain viewing angles. Once a momentary reflection is detected, the
fractographer can zero in on the flaw site. Often the full extent of the crack is
not visible because the crack is so tight. Fréchette40 observed that chill checks
are hard to characterize since mechanical effects can also occur at the contact
site. If a glass object breaks from a chill check during subsequent use and
stressing, then it is likely the origin site will be irregular and may not be
recognized as a chill check without examination of the outer surface. This
underscores an earlier recommendation: if an origin on a fracture surface is
located at the surface of the component, then it is prudent to examine the
exterior surface of the part as well as the fracture surface. Varner 44 has
compared features of chill checks and impact origins. Chill checks are curved
and not flat, often have no visible outer surface damage, and, once the flaw is
exposed on a fracture surface, have few markings on the flaw itself other than
some twist hackle. Impact cracks are semicircular and are straighter, and the
outer surface has chipping and lateral cracks. Once exposed on a fracture
surface, the impact flaw exhibits tertiary Wallner lines.
Polishing flaws from a loose abrasive rolled across a surface while under load
can leave a variety of damage including indentations, scratches, Hertzian full or
partial ring cracks and gouges according to Fréchette.40 The crack depths are
typically four to five times larger than the particle diameter. They often are
very elongated along the surface. Polishing, lapping or grinding glass can
create are sleeks which are fine, scratch like, smooth-boundary imperfections
usually caused by a foreign particle during polishing. 45
Chatter marks such as shown in Figures 6.58 and 6.59 often are clues to contact
cracks underneath. The periodicity is a manifestation of vibration or a stick-
6-65
Fractography of Ceramics and Glasses
slip interaction when a contacting object slides over the surface. Fracture often
starts from the first or last chatter crack in a series. Abrasion tracks, which
often occur around the bottom rim or sides of a glass vessel, are also vulnerable
sites for localized contact damage as shown in Figure 6.58. Polishing, lapping
or grinding glass can create miniature crack patterns of chatter cracks.
Preston45 and Ghering and Turnbull 46 showed numerous examples and
correlated them with the forces, the surface lubrication conditions, and the
abrasive or contactor types. Preston showed that a blunt contactor or ball
dragged across a surface creates small partial cones cracks behind the
contactor. The cracks are created by tensile stresses behind the ball or
contactor as it moves and the cracks are bowed in the direction of the contactor
movement such as shown in Figure 6.36. In sharp contrast, a rolling ball
creates arc cracks that bow in the opposite direction. They arch towards the
direction from which the ball came.45
(a) (b)
Figure 6.58 Chatter cracks from an object that contacted and slid across the
surface of a heat-strengthened soda lime glass kitchenware. (a) shows the outer
surface. The black arrow points to the fracture surface and the origin site. (b)
shows the specimen tilted to show the origin crack on the fracture surface
(large black arrow) and the other chatter cracks on the outer surface (white
arrows).
6-66
Origins of Fracture
(a)
(b)
Figure 6.59 Very tiny chatter cracks from a small object that contacted and
slid across the surface of a heat-strengthened soda lime glass kitchenware. (a)
shows the outer surface with an assortment of common scratches and edge
chips, some of which are secondary edge chips after the main fracture. The
arrow points to the exact origin site which is seen to be one of a series of
curved chatter cracks. (b) shows the fracture surface. The multiple secondary
edge chips may confuse interpretation, but the Wallner lines and a faint pop in
arrest line are centered on the chatter crack (arrow).
6-67
Fractography of Ceramics and Glasses
(a)
(b)
Figure 6.60 Surface abrasion can create subsurface cracks. (a) shows a piece
from a broken heat-strengthened kitchenware tilted back to show the fracture
surface and the elongated mirror and the outer surface with the abrasion track
(large black arrow). Incidental secondary handling edge chips on the left
(smaller black arrow) came from handling after fracture. (b) is the fracture
surface showing an elongated mirror and the origin (white arrow). The origin
region is dark due to a slight curvature of the fracture mirror surface.
6-68
Origins of Fracture
The author once had a surprising experience with very bright retroreflective
flaws when using a stereoptical microscope with dental highly-filled resin-
matrix bend bars. Slight adjustments to the lighting angle eliminated the intense
reflections. Some of the fracture origins were glassy inclusions or
agglomerates of glassy filler particles that strongly reflected the incoming light.
This was just as disconcerting and surprising as when one shines a flashlight at
the retro-reflective eyes of a cat. Retroreflective glass beads are used with
roadway marker paints to improve their visibility. Regrettably, the author did
not take photos at the time of examination.
(a) (b)
Figure 6.61 Unusual fractures. (a) shows a vein of silicon that was the
strength limiting flaw in a siliconized SiC bend bar. This probably was caused
by silicon infiltration of a seam or crack in the green body. (b) shows the
fracture surface of a silicon nitride that had an unstable yttria silicate
boundary phase. Phase changes distorted the bar shape from its original
rectangular shape. Fracture occurred from the crack marked by arrows.
T and C denote the tension and compression sides of the bend bar.
6-69
Fractography of Ceramics and Glasses
5 µm
(a) (b)
ceramic ceramic
clay
clay
(c) (d)
(e) (f)
6-70
Origins of Fracture
6.10 Contaminants
Every care should be taken to avoid contaminating the fracture surface. This is
especially true during preparation of the specimens for SEM examination.
During an optical examination, cleaning is not too difficult, but it is a different
story if the specimen is being examined in the SEM. Once a specimen has been
gold coated and is in the SEM, it can be disappointing to realize the flaw is
masked by contamination or that a photo of the origin has distracting debris in
the vicinity. Extraction, recleaning, and preparation are time consuming and
frustrating. Sometimes contaminants can be recognized if they appear to sit up
on top of the fracture surface and are obviously not part of the material. This
criterion is by no means perfect, however, as many contaminants get into the
fracture surface crevices and blend in diabolically well. As noted previously in
Chapter 3, modeling clays are especially pernicious since they are fine, blend in
well on the fracture surface, adhere extremely well, and are next to impossible
to clean off. An EDX analysis of clay contamination on a fracture surface
revealed aluminum, silicon, and titanium. Figures 6.63 – 6.64 illustrates some
common contaminants and how they look in the SEM.
ceramic
veneer
(a) (b)
origin
veneer
origin
origin
clean outer surface
Plaque over outer surface
(c) (d)
Figure 6.63 Dental plaque is a common contaminant on ceramic restorations.
(a) shows the broken middle (“pontic”) piece of a lithium disilicate bridge that
broke after 7 weeks in vivo. (b) shows the overall fracture surface. (c) and (d)
show the origin site before and after cleaning to removal plaque. Plaque can
be removed with bleach as described on page 3-6. See Figure 6.74 for
additional images of this bridge.
6-71
Fractography of Ceramics and Glasses
(a) (b)
(c) (d)
(e) (f)
Figure 6.64 Contaminants in 3Y-TZP bend bars. (a) and (b) show a clean
fracture with a well-defined fracture mirror and the common flaw in this
material: a pore or porous region origin. (c) and (d) show clay smeared over
the fracture surface and fracture mirror in another specimen. Notice the
parallel smear traces (arrows) from lower left to upper right in (d). The clay is
not evident in the low magnification image of the mirror, but it covers the
origin in (d). It is possible the wrong end of the bend bar initially was put into
mounting clay! (e) and (f) show a mirror with an unknown contaminant
covering the origin (arrows). (all courtesy of J. Quinn)
6-72
Origins of Fracture
In other cases two small flaws can link to form a strength limiting flaw. A plus
sign may be used to indicate that these origin types linked together to limit the
strength of the ceramic. For example, P + MD indicates a pore combined with
machining damage. Origins can be difficult to characterize if they have mixed
attributes. For example, porous regions often have large pores associated with
them (Figure 6.66e). References 50, 51, and 52 also discuss flaw linking.
6-73
Fractography of Ceramics and Glasses
(a)
(b) (c)
(d) (e)
Figure 6.66 Combined (hybrid) flaw fracture origins (a) shows a flaw in a
sintered 99.9 % alumina that can be described as primarily a pore, but with
some large grains (P & LG). (b) is a pore/large grain (P & LG) combination
in a sintered alpha silicon carbide bend bar. Note the good quality of the
chamfer. Edge damage was not a factor in this case. (c) is a combination of
an agglomerate and a large grain (A & LG) in sintered silicon carbide. (d) is
a surface grinding crack that linked with a pore in a sintered silicon nitride (P
+ MD). (e) shows a flaw in a sintered alumina could be described as either a
porous region or a pore.
6-74
Origins of Fracture
Figure 6.67 Careful processing may eliminate severe flaws such as inclusions
or gross pores. In the absence of such flaws, the material may fail from
“baseline microstructure” origins.
(a) (b)
Figure 6.68 Baseline microstructure flaw in a sintered aluminum oxynitride
four-point flexure bar. (a) shows a single grain that was the fracture origin.
The origin in this case may actually be the grain boundary around the grain.
(b) shows a polished section view of the microstructure.
6-75
Fractography of Ceramics and Glasses
Sharp points are extremely vulnerable to crack pop-in at very low stresses
during fabrication or subsequent handling. Once the part is put under stress,
these pop-in cracks can propagate to fracture. There may not necessarily be a
material flaw as a starter. Sometimes an arrest line will be detected at the end
of the pop-in crack, but often the final plane of fracture is identical to the initial
pop-in crack plane, and the latter will not stand out. Figures 6.69a-d show
exceptionally thin cusps between etch pits in theta strength specimens. 53
(a) (b)
(c) (d)
Figure 6.69 Small cracks may pop-in from sharp points or cusps. The insert
of (a) shows a miniature (300 µm) silicon theta strength-test specimen. A force
applied to the top of the ring causes the middle web to stretch and break.
(a) shows the fracture mirror around the surface pits fracture origin. The pits
are caused by the deep reactive ion-etched (DRIE) processing. Sharp cusps
between the pits are an easy path for tiny micro-cracks to pop-in and link.
(c) and (d) show another theta specimen. The insert in (c) shows the fragment
and the origin location. The hackle lines in (d) indicate fracture started at a
single sharp cusp between two pits. The cusp is less than 50 nm in size and
may be the smallest fracture origin ever recorded.
6-76
Origins of Fracture
origin
Too
thin !
(a) (b)
Figure 6.70 A “feather tip” margin in an alumina dental crown that broke in
a patient’s mouth (“clinical 22”). The margin is only 30 µm wide and was
extremely vulnerable! Hoop stresses caused fracture from some of the tiny
flaws at the margin tip. (crown courtesy of M. Øilo)
Figures 6.69 shows a sharp “feather-tip margin” origin in a dental crown. The
margin of this crown is much too thin. This type of margin is used by some
dentists since less reduction of the tooth is needed and the crown blends into
the gum line very well. Such practice is unwise for all-ceramic crowns. Figure
10.25g shows another crown with a margin pop-in crack at a feather tip.
Not all sharp point/cusp origins are in small pieces. The author once found
such an origin in a ~ 1 m large SiC honeycomb structure for a space mirror.
Flaws may not necessarily be oriented so that they are perpendicular to the
principle tensile stress. Only a portion of a flaw may be exposed on a fracture
surface as shown in Figure 6.71. The cone cracks of Figure 6.32 are examples.
fracture surface
tensile surface
Figure 6.71 The flaw may not be fully-exposed on the fracture surface.
6-77
Fractography of Ceramics and Glasses
Glazes and veneers are glassy and have low fracture toughness and resistance
to fracture. A crack can form in the vulnerable glaze and then propagate into
the remainder of the component. Figures 6.72 – 6.74 show some examples for
three clinical dental fractures. Excessive mismatch in the thermal expansion
coefficients of the glaze and the body can cause cracking, such as shown in Fig.
6.72. Glaze or veneer cracking formed by tension is called “crazing.” Some-
times excessive porosity can occur at a boundary, and can be the strength-
limiting flaw as shown in Figure 6.73.
(a) (b)
Figure 6.72 Glazes can be preferred sites of fracture initiation. This was an
in vivo broken three-unit zirconia bridge (courtesy T. Kosmac) with a crazed
veneer probably from a mismatch of the veneer type to the ceramic. Fracture
started on the plane on the left insert marked by the yellow arrow.
origin
Veneer
(a) palatal
(b)
zirconia
buccal core
200 µm
margin
Figure 6.73 Glaze interfaces can be preferred sites of fracture initiation. This
is an in vivo broken six-unit zirconia bridge with a poor design. The origin is
the pore cluster at the interface between the veneer and core apex. (images
courtesy U. Lohbauer and S. Scherrer)
6-78
Origins of Fracture
Krohn et al. 54 showed pore and pigment aggregate fracture origins in enamel
layers that were applied to float glass. Kobayahsi et al.55 showed excellent
images of fracture origins that were a series of bubbles at the boundaries
between an alumina porcelain and a glaze.
some
plaque here
(a) (b)
plaque was
removed
(c) (d)
Figure 6.74 A gold-coated lithium disilicate glass ceramic bridge pontic unit
(courtesy S. Scherrer). It broke in vivo on the palatal side of a connector. The
origin was a sharp contact damage crack in the vulnerable veneer. Notice the
arrest lines marked by arrows in (c), the helpful wake hackle lines from the
bubbles in the veneer seen in (c) and (d), and even the concentric tertiary
Wallner lines typical of impact in (d). See also Figure 6.63.
6-79
Even if a fracture origin site can be found, there may not be an obvious flaw at
the origin. Subtle porous regions or grinding damage crack at the surface can
possibly be present but not detectable. This dilemma is more likely to occur
with coarse-grained ceramics, porous ceramics, and heterogeneous ceramics,
wherein a flaw could “hide” against the background microstructure.
(a) (b)
(c)
6-80
Origins of Fracture
(a)
(b)
(c)
6-81
Fractography of Ceramics and Glasses
Fracture origins of the types shown in Figure 6.75 are extremely difficult to
diagnose on fracture surfaces, since the features at the origin look exactly like
the normal microstructure. Figure 6.76 is an example. Polished specimen
microstructural analysis helped with the interpretation as shown in Figure 6.77
The origin may take on the character of porous regions as shown in Figure
6.12. The literature is sparse on such flaws, no doubt because engineers and
scientists are loath to show fractographs of origins that fail to show a distinct
flaw!
Figure 6.78 and previous Figures 6.4, 6.67, and 6.77 show examples of how
polished microstructural sections can provide complimentary information about
flaw types and their distributions. Polished section preparation is a staple for
metals failure analysis, usually to ensure that the microstructure is as expected
and the correct phases are present.
6-82
Origins of Fracture
(a) (b)
(c)
(d)
6-83
Fractography of Ceramics and Glasses
It may suffice to identify the nature of the flaw, but a better characterization
includes at least a rough estimate of its size. A 100 µm inclusion may not have
the same source or behavior as a 20 µm inclusion. The size need not be
measured precisely if the characterization is intended to describe the general
nature of the origin. On the other hand, there are some instances when a good
estimate is needed, such as when it is intended to do a fracture mechanics
analysis of the flaw as described in the next chapter. This may be to confirm
that the fractographer has found the correct flaw. So, for example, if a
fractographer has tentatively identified a 2 µm feature as an origin, but fracture
mechanics suggests the flaw should be 50 µm in size, then the fractographer
ought to reevaluate his assessment.
For each component, or laboratory-tested specimen, record the origin type, the
location in the particular specimen, and the flaw size. If a fracture mirror is
present, it may also be helpful to measure its size, especially if the fracture is in
a component. Mirror size measurements in laboratory strength test specimens
also may be done so that mirror constants can be estimated and published for
new materials. Figure 6.79 shows images that follow the recommendations in
ASTM C 1322.49
6-84
Origins of Fracture
Figure 6.79 Multiple photos should be taken showing whole fracture surface,
the mirror region and the origin. The right side shows an example for an
inclusion in a silicon nitride rod flexural strength specimen.
6-85
Fractography of Ceramics and Glasses
Component failure analyses often have some overall photographs of the entire
structure and close-ups of various critical features. Much time and confusion
can be saved if the successive images are presented in an orderly, systematic
way. It is always wise to start with an overall image of the structure and then
add images at progressively higher magnification at key locations.
When one jumps from one magnification to a higher magnification, later it may
be difficult to correlate the two images. In the author’s experience,
magnification jumps of about 5 are the most one should use.
At each step the relationship of the images to each other should be clear.
Figure 5-68 in the previous chapter and Figure 6-81 on the next page show
good examples for two dental crown failures.
6-86
Origins of Fracture
(a) (b)
origin
(c)
dcp
2 mm
(d)
(e)
Figure 6.81 Failure of a first upper-left molar crown. (a) shows a view onto
the occlusal surface (the crown top) and the chipped porcelain veneer on the
side of the crown (arrow). The porcelain veneer was inadequately supported
by the alumina core as shown in the x-ray image (b). (c) – (e) show
progressive close-ups of the fracture surface on an epoxy resin replica.
This is the same crown as shown in Figure 3.34 on replica techniques.
(All courtesy S. Scherrer, Univ. Geneva)
6-87
Fractography of Ceramics and Glasses
Chapter 6 References
6-88
Origins of Fracture
6-89
Fractography of Ceramics and Glasses
6-90
Origins of Fracture
6-91
Quantitative Analysis
7. Quantitative Analysis
7.1 Introduction
Quantitative analysis can aid the fractographic analysis and extend its value. It
may be used to confirm that the right flaw has been identified as a fracture
origin. It can make good estimates of the stresses in the part at failure. Four
methods to estimate stress are shown in this chapter: fragmentation analysis,
branching distance analysis, fracture mirror size analysis, and origin size
analysis. Some analyses are empirical and others are based on the powerful
tool of fracture mechanics. One goal of this chapter is to convey a working
knowledge of fracture mechanics so that it can be used to solve practical
problems. Table 1 includes some useful conversion factors.
length
1 in * = 25.4 mm
1 micrometer (micron) = 1 x 10-6 m
.001 inch * = 25.4 micrometers
force
1 kgf (kilogram force)* = 9.80 N (Newtons )
stress or pressure
1 ksi (1,000 psi) * = 6.895 MPa (Megapascal)
1 Pa = 1 N/m2
1 MPa = 1 MN/m2 = 1 x 106 Pa
1 MPa = 145.04 psi = 0.14504 ksi *
1 GPa = 1000 MPa = 1 x 109 Pa
example: 100 ksi = 690 MPa
stress intensity
1 ksi √inch * = 1.10 MPa√m
1 ksi in
example: 5 MPa m x
1.10 MPa m = 4.55 ksi in
energy
1 erg = .0000001 (1 x 10-7) J (Joule)
1 ft-lbf *(foot pound force) = 1.355 J
energy/area
erg/cm2 = .001 J/m2
Length on Photos
1 mm = 1 micrometer on a 1000X photo
7-1
Fractography of Ceramics and Glasses
A general assessment of stress levels can be made from the fracture pattern.
Low stressed parts break into only two pieces. Fréchette1 and Orr 2 and have
suggested that 10 MPa (1,500 psi) is a threshold stress level for branching and
fragmentation in glass as illustrated in Figure 7.1. Additional fragmentation
occurs with increasing stress in the part, or increasing stored elastic energy.
Very highly-stressed parts fragment to such a degree that laypersons are apt to
describe the fracture as an “explosion.” Retrieval of the two fragments with
the fracture origin may be problematic.
Fréchette and his students 3 have shown fragmentation correlations with internal
pressure of bottles as shown in Figure 7.2. They did a similar analysis for
impact velocity for impacted bottles. 4 While their work may have been
specific to certain bottle types and shapes, similar testing could be done with
alternative shapes and sizes to aid in the failure analysis of new service failures.
7-2
Quantitative Analysis
showed examples of such test specimens and their breakage patterns. Fragmen-
tation in Figure 7.3 was quantified by the number of radial cracks that reached
the outer rim. This was an arbitrary criterion. Morrell showed a very similar
graph for the number of fragments as a function of stress in ring-on-ring tests of
high purity alumina. 6 Shand presented a similar graph counting radial cracks
for 6 cm square window glass plates broken by impact in the middle. 7 He
pointed out that by such counting, one could obtain a stress at failure estimate
even if the impacted site was lost or destroyed. More recently, Wang8 did
stress analyses and counts of the number of pieces in glass plate coupons of
different sizes when pressure loaded to fracture. Graphs such as these are very
specific to specimen or component size, shape, and material.
Stress (MPa)
0 50 100 150 200 250 300 350
60
50
40
20
.36 L (12 oz)
10
7-3
Fractography of Ceramics and Glasses
Figure 7.3 was actually used to analyze an unexpected fracture. A similar glass
disk was impacted by a hailstone that was intended to weaken the disk by
creating surface damage that could be assessed in a subsequent strength test.
The hailstone unexpectedly fractured the disk. The fragments were recovered
and reassembled and the origin determined to be an ordinary polishing scratch
on the surface opposite the hail impacted side. The hailstone diameter (25 mm)
was similar to the inner ring diameter in the strength test set. From the number
of radial cracks, the stress at failure was estimated. The estimate matched the
dynamic stress estimated from an impact stress analysis, thereby confirming
that the impact indeed had imparted sufficient energy to flex the disk to fracture
from the ordinary surface flaw.
110
100
90
N, Number of Radial Cracks
80
70
60
50
40
30
0
0 5 10 15 20 25 30
Inner Ring Stress (ksi)
Figure 7.3 The number of radial cracks that reach the outer rim as a function
of fracture stress for 76 mm diameter by 5.4 mm thick borosilicate crown glass
disk broken in ring-on-ring flexure. Data for this graph included only fractures
that initiated inside the inner loading ring.
7-4
Quantitative Analysis
More often, the fractographer simply makes a qualitative stress assessment with
a cursory examination. This is often sufficient for a diagnosis. For example, if
a fracture occurs at an unexpectedly low stress, the fractographer might search
for an unexpectedly large flaw in the component or specimen. A handling or
preparation flaw may have weakened it to an unexpected level.
McMaster et al. 9 state that the most common method of inferring stress levels
in tempered glass is by breaking it. They and Shinkai 10 cite work by
Akeyoshi, 11 wherein 10 cm square tempered plates of various thickness were
broken by sharp impact. The number of fragments in a 5 cm square region
correlated to the internal tension temper stress as shown in Figure 7.4.
Figure 7.4 The interior tension stress in tempered soda lime glass plates may
be estimated from the fragment density when the plate is broken. Each line
corresponds to the plate thickness shown. (After Akeyoshi et al., Ref. 11)
7-5
Fractography of Ceramics and Glasses
Figure 7.5 The interior tension stress in tempered soda lime sheet and plate
glass may be estimated from the average fragment mass. Center tension stress
is plotted versus mass normalized by the plate thickness, t. Data for five
different thickness plates and sheets from 3.2 mm to 9.5 mm (1/8 in to 3/8 in)
fall on the same curve. (After Barsom, ref. 13)
For a particular thickness plate, the data suggests there is a minimum temper
stress to cause dicing. Blank 12 presented some interesting fragmentation work
on tempered soda lime silica plates, 300 mm by 300 mm in size and with
thicknesses varying from 3.2 to 7.2 mm. The average fragment size (area) in
mm2 dramatically decreased with increasing surface compression
strengthening.
Barsom13 showed that interior tension stress σct was related to the fragment
mass:
1
m 4
σ ct = constant (7.1)
t
where m is the fragment weight (mass), t is the plate thickness. The constant
has units of stress • (mass/thickness)1/4.
7-6
Quantitative Analysis
It should be borne in mind that these relationships and trends are altered if there
are significant externally applied stresses superimposed on the temper stresses.
The fractographer should be on the lookout for irregular breakage patterns and
fragment sizes that are signs of nonuniform temper stresses. Fréchette1 showed
two examples. In one case a glass cylinder was not tempered uniformly during
the fabrication process (his case 10.18). A second case (his 10.22) featured a
high-powered lamp that cracked spontaneously as a result of heating in service.
The fragments were atypically large in the fracture origin area, but elsewhere
the normal dicing fragmentation pattern was observed. Fréchette concluded
that the intense in-service heating caused the middle of the lamp to lose its
temper. Blank12 discussed the use of tempered glass in front automobile
windshields in Germany and showed a fascinating drawing of a fractured
windshield that had dramatically uneven fragment sizes and crack patterns due
to nonuniform heat transfer during tempering. It is commonly assumed that the
residual stresses are symmetric through the thickness as shown in Figure 5.19,
but Menčík 14 illustrated the effects of asymmetric tempering of plates on the
final stress distributions. Different residual compression stresses could exist on
the two plate surfaces. There may even be a net tensile stress on one surface.
The stress at fracture is related to the branching distance away from the origin
by:
σ Rb = A (7.2)
b
where σ is the stress, Rb is the branching radius, and Ab is a constant called the
fracture branching constant. It has units of stress⋅√length. These units are the
same as for stress intensity which is described in section 7.5. The relationship
between Ab and fracture toughness is discussed in 7.6. A similar relationship
applies to mirror sizes as will be discussed in section 7.4. Rearranging shows
that the branching distance is inversely dependent upon the square of the stress:
2
A (7.3)
Rb = b
σ
7-7
Fractography of Ceramics and Glasses
The branching angle can at least be a qualitative indicator of the stress state as
shown in Figure 4.4 in chapter 4. The branch angle is actually variable and
curves at the exact branching location. It is recommended that the branching
angle be measured close to but somewhat away from the splitting juncture, at a
point where the branching angle has settled into a more or less constant value.
7-8
Quantitative Analysis
(a)
(b)
7-9
Fractography of Ceramics and Glasses
The fracture mirror size is related to the stress at the origin. Orr’s empirical
mirror size relationship is:
σ R=A (7.4)
where σ is the tensile stress at the origin at the instant of fracture, R is the
mirror radius, and A is the “mirror constant” with units of stress intensity
(MPa√m or ksi√in) and is considered by most as a material property. (A
version of this equation with a residual stress term is discussed later in this
chapter.) This relationship is similar to, but not quite the same, as that for
stress intensity or fracture toughness as will be discussed in section 7.6 below.
Figure 7.7 shows more details of a mirror centered on a surface flaw. This
rendition also depicts the prospect that the initial flaw might grow somewhat
from a to ac before going critical.
This author reviewed the evolution of fracture mirror size analysis in a 2006
paper. 18 Leighton Orr of the Pittsburg Plate Glass (PPG) Company was the
first to systematically use equation 7.4 with the square root dependence of
mirror size to solve glass fracture problems as far back as the 1940s.2,19,20
ac
a
Ri
Ro
Rb
Ab Ao Ai KIc
Figure 7.7 Fracture mirror schematic. Ri is the inner or mirror-mist
boundary, Ro is the mist-hackle boundary, and Rb is the branching distance.
The fracture toughness KIc (discussed in section 7.5.1), the mirror constants Ai
and Ao and the branching constant Ab correspond to each of these boundaries.
All have units of stress √length.
7-10
Quantitative Analysis
The practical significance of equation 7.4 is that one can calculate the origin
stress if a fracture mirror size is measured and a fracture mirror constant is
available. The calculated stress is the net tensile stress acting on the flaw and
the region around the flaw. It may include several stress sources including
mechanical, thermal, and residual stresses. It is possible to discern separate
mirror-mist and mist-hackle boundaries in glasses and each has a corresponding
mirror constant A. The most common notation is to refer to the mirror-mist
boundary as the inner mirror boundary and the corresponding mirror constant is
designated Ai. The mist-hackle boundary is referred to as the outer mirror and
the corresponding mirror constant is designated Ao. The mirror-mist boundary
is usually not perceivable in polycrystalline ceramics and only the mirror-
hackle boundary is usually measured.
Equation 7.4 is valid even for large mirrors if loading is in direct tension. If a
part is in flexural loading, deviations from the trend can occur as discussed in
Appendix D.
7-11
Fractography of Ceramics and Glasses
(described later in section 7.5) gives the best fit to data and the exact shape of
mirrors in various stress fields. A crucial observation is that the stress intensity
model fully accounts for the small inward facing cusp at specimen surfaces
(Figures 5.2 and 7.7), whereas the other models do not. Kirchner et al. 29,30
argued that a more fundamental material parameter might be KIB, the stress
intensity factor at branching, rather than the A’s in equation 7.4 since the latter
do not take into account the free surface, geometry factors, and non-uniform
stress gradients over the crack surface. There is validity to their argument, but
in practice the ease of use of 7.4 has led to its widespread adoption. (The size
and shape predictions based on A or KIB match very closely for the limiting
cases of small mirrors in tension specimens. This is also true for most of the
periphery of small semicircular mirrors centered on surface flaws in strong
tension or flexure specimens.)
The accuracy of the stress estimate varies depending upon the material, the
experience of the fractographer, the microscopy and illumination used, and
component geometry and stress gradient effects. Stress estimates may be
accurate to within ≈ 10 % with glasses if they form ideal fracture mirrors.
Kirchner and Gruver 31 also showed that the relationship holds for elevated
temperature fractures in ceramics. The mirror constant was invariant from
room temperature to 1400 OC for a hot-pressed alumina with a 1 µm grain size.
This is a remarkable finding since the alumina was fired at 1475 OC. Kerper
and Scuderi 32 also showed that mirror constants were unaffected by elevated
temperature for eight glasses. Shinkai 33 showed the mirror-mist constant, Ai,
varied little with temperature for three glasses. Ai varied little with temperature
for the Pyrex, but for the soda lime silica float glass and the 96% silica (Vicor),
the Ai values at -150oC were greater (0.03 – 0.04) than at room temperature.
The trends were comparable to similar increases of the fracture surface energy,
γf, with decreasing temperature. Shinkai33 also did experiments in various
environments. There was negligible environmental effect and the Ai’s
measured at room temperate in air and dry N2 gas were comparable. The lack
of an environmental sensitivity is not surprising since mirror boundaries form
when the crack is travelling at or near terminal velocity.
Crack branching constants Ab have sometimes been reported, but some caution
is advised, however, as discussed on page 7-7. There is evidence to suggest
that Ab depends upon the loading mode (uniaxial or biaxial) and the size of the
mirror relative to a part thickness if the part breaks in bending. The three
boundaries correspond to an expanding crack front, and hence, Ai < Ao < Ab.
Appendix C has a comprehensive compilation of fracture mirror and branching
constants. The table shows there is some variability in the published mirror
7-12
Quantitative Analysis
constants, even for model materials such as fused silica or soda lime silica.
Some of this is due to the aforementioned metrology and judgment issues, but
much of it is due to genuine material-to-material variability. Thus, one should
not expect all silicon nitrides or aluminas to have the same mirror constants.
Microstructural differences can change the mirror constants considerably.
The word “boundary” must be used with some caution. It is now clear from a
number of studies that there probably is not a discrete or sudden transition point
on the fracture surface corresponding to a mirror boundary. The mode of
viewing and magnification affects the judgment of the boundaries location.
The higher the magnification the smaller the mirror seems to be, since fine
detail that was washed out or not resolvable at lower power can be discerned at
higher power. The mirror-mist boundary probably corresponds to surface
roughness features that are of the order of 0.1 µm to 0.2 µm, which is a fraction
of the wavelength of light and at the threshold of observable features with an
optical microscope. Section D.3 in Appendix D has a discussion of several
studies that have used various surface roughness measurement methods to
measure mirrors, but these methods have yielded different results.
So, although there is some variability in how different observers judge and
measure fracture mirror sizes, most fractographers agree that there is a fracture
mirror that can be measured. Size estimates can vary depending upon the type
of microscope used, the mode of illumination, the objective power of the lens,
and the judgment criteria of the fractographer.
This section gives a brief outline of how to measure fracture mirror sizes. Very
detailed guidelines are presented in Appendix D. The procedures in this Guide
were adopted as an ASTM standard34 in 2007.
The mirror-mist boundary in glasses is the periphery where one can discern
the onset of mist. This boundary corresponds to Ai, the inner mirror constant.
7-13
Fractography of Ceramics and Glasses
The mist-hackle boundary in glasses is the periphery where one can discern
the onset of systematic hackle. This boundary corresponds to Ao, the outer
mirror constant.
2R
(a) (b)
Figure 7.8 Example of a mirror in a ceramic material that is difficult to
precisely measure. This is one of the specimens of the 3Y-TZP for the data in
Figure 7.11 below. (a) shows the uncoated fracture surface of a 2.8 mm thick
bend bar with low angle, grazing illumination (486 MPa). (b) shows an
interpretation for a mirror-hackle boundary whereby new hackle is detected
(small white arrows) as compared to the roughness inside the mirror. The
mirror is slightly elongated into the interior due to the stress gradient.
Judgments are best made while viewing the surface through a microscope.
7-14
Quantitative Analysis
Figure 7.9 shows the two most common graph types that are used to show
stress - mirror size data. The merits of each approach are discussed in
Appendix D. This Guide recommends the use of stress versus 1/√R, Fig. 7.9a.
The linear graph of stress versus 1/√R in Figure 7.9a has the advantage that the
fracture mirror constant is the slope of the line, and residual stresses can be
detected by non-zero intercepts as described later.
The log stress versus log R graph in Figure 7.9b facilitates plotting a broad
range of stress and mirror sizes. The mirror constant corresponds to the stress
necessary to create a mirror of unit size, since (from eq. 4):
log σa = log A - ½log (R). If R = 1, then log σa = log A, and σa = A.
•
•• • log
σa •• A σa -½
•
σ = A/√1
large small
small large
mirrors mirrors mirrors mirrors
R=1
log R Log R = 0
(a) (b)
Figure 7.9 Stress and mirror size graphs. (a) shows the preferred graph of
applied stress, σa, versus inverse square root of mirror radius. The mirror
constant is the slope of the line. (b) shows log σa versus log radius as
commonly used in the older literature.
Two examples of mirror data are shown in Figures 7.10 and 7.11 on the next
page. The glass data in Figure 7.10 is extraordinary since the mirror size range
is almost a factor of 100. The 3Y-TZP zirconia data in Figure 7.11 shows that
measurements on uncoated and gold coated specimens concurred. This was
gratifying since the uncoated specimens were difficult to interpret due to
material translucency and difficulty in judging a boundary. Not surprisingly,
the data scatter was greater for the uncoated specimens. More examples of
marked fracture mirrors are shown in Appendix D which also has detail how
residual stresses, stress gradients, the origin flaws, and microstructure can
affect mirror sizes and shapes.
7-15
Slope ≈ -0.5
Ai = 2.08 ± .02 MPa√m (1,891 psi √in)
n = 259 rods
Figure 7.10 A superb example of fracture mirror data for glass rods of many
sizes and strengths. Two hundred and fifty-nine Pyrex 7740 rods varying in
size by almost a factor of ten (4.1 mm to 38.1 mm) were broken in three-point
flexure by Kerper and Scuderi at the National Bureau of Standards in 1964.32
900
800
700
+ uncoated disks
Fracture Stress (MPa)
Ao = 9.7 MPa√m
600
± 0.2 MPa√m
500
L gold coated disks
Ao = 9.6 MPa√m
± 0.2 MPa√m
400
300
' bars Ao = 8.6 MPa√m
± 0.2 MPa√m
200
100
0
0 10 20 30 40 50 60 70 80 90
7-16
Quantitative Analysis
Figure 7.12 A force P acts upon the object. Fracture occurs at a critical
force that depends upon the material properties and the object’s size. (b) stress
is the force normalized by the cross sectional area. (c) discontinuities such as
a thorough-hole concentrate stress and magnify it in the vicinity of the
discontinuity. (d) sharp discontinuities such as cracks intensify the stress field
in front of the discontinuity.
7-17
Fractography of Ceramics and Glasses
2Eγ f
σf = (7.5)
πc
where σf is the fracture stress (the nominal stress in the plate), E is the elastic
modulus, and γf is the fracture energy that is the energy to create unit surface
area. In this Guide, flaw size is denoted as either “a” or “c” following
conventional practice in the fracture mechanics literature. Usually the
dimension is the radius or half width if the flaw is in the middle of a plate as
shown in Figure 7.13a. If the crack is on the side of the plate, then a or c are
the full crack length. The critical feature of this relationship is that strength is
inversely proportional to the square root of flaw size. The larger the flaw, the
weaker is the structure. Griffith’s formula was derived on the premise that
crack propagation occurs if the elastic strain energy release in a body is greater
than the creation of surface energy due to the new surfaces if the crack were to
extend. His formula sets a necessary condition for a decrease in overall energy
of the system.
7-18
Quantitative Analysis
σa
(a) (b)
σa
Figure 7.13 Through crack in a plate loaded in tension. (a) Long slender
elliptical cracks of this type were envisioned by Griffith as flaws. If the crack
tips are atomically sharp, fracture mechanics developed by Irwin38 in the 1950s
must be used to model the stress state in the immediate vicinity for the crack tip
as shown in (b). The dashed line shows σy as a function of x directly in front of
the crack tip at y = 0. The stress intensity KI quantifies the magnitude of the
stress field acting to open up the crack tip. KI depends upon the far field stress,
σa and the geometry of the crack and the structure. Fracture occurs when the
intensity of the stress field reaches a limiting value, KIc, known as the “critical
fracture toughness, Mode I.”
7-19
Fractography of Ceramics and Glasses
(a) (b)
(c)
flat
Figure 7.15 Lines of stress or “flow lines of stress” in a body. They cannot
be transmitted through the irregularity such as a bubble or a pore. The flaws
create a stress concentration on the sides where the flow lines converge.
Sharper irregularities, such as the flattened wide flaw on in (b) create even
greater stress concentrations. In the limit, for a flat sharp crack, the flow lines
of stress kink severely, and fracture mechanics must be used as opposed to a
simple stress concentration factor. (c) shows four flaws that have similar size.
The left two are not particularly severe flaws despite their size, but the right
two are more deleterious, especially if the rim-cracked pore (“Rings of
Saturn”) flaw has a large rim crack, c. The flat penny-shaped flaw (shown
schematically) on the right is the severest.
7-20
Quantitative Analysis
KI = Y σ c (7.6)
and fracture occurs when:
K Ic = Y σ f c (7.7)
Y is the stress intensity shape factor, a dimensionless parameter that takes into
account the shape of the crack and the loading conditions, c is the crack size,
and σf is the stress at fracture. In this Guide, Y combines all geometric terms
and π or √π terms. Examples of Y are shown below. Fracture occurs when any
combination of Y, σ, or c leads to the critical condition. It should be
emphasized that equations 7.6 and 7.7 only apply to small flaws loaded by
remote tensile stresses, also known as “far-field stresses.” Most textbooks on
fracture mechanics show illustrations such as Figures 7.12 and 7.13, whereby
the stress field is remotely applied (not near the crack) and σa is the nominal
stress if no crack were present. The flaws may be embedded in the bulk or may
be on the specimen surface. Cracks may be loaded in other component shapes
7-21
Fractography of Ceramics and Glasses
or loading configurations that do not create far-field stresses. For example, the
local stresses from the plastic deformation zone underneath a Vickers indent-
ation (that act on a median crack) are not far-field stresses, nor are the localized
contact stresses where a sphere presses on a ceramic or glass surface. Further-
more, equations 7.6 and 7.7 may be inapplicable once cracks become very large
relative to a component size. For example, a bend bar with a crack that reaches
half way through the bar thickness has a much more complicated equation
relating force, crack size, and beam dimensions to KI. The point here is that if
one finds a fracture origin crack in a broken ceramic or glass object, and is
tempted to apply equation 7.7, decide first whether the stresses really are
remotely applied and are genuinely far field. Do not apply equations 7.6 or 7.7
if the stresses are localized, or if the part dimensions are so small that the flow
lines of stress around the flaw are not well established as in Figure 7.15.
Note that the stress and flaw size dependencies are the same as the Griffith
equation. Indeed, the energy and stress intensity approaches are consistent, and
for plane strain (page 1.5) loading conditions:
2Eγ f
K Ic = (7.8)
(1−ν 2 )
where E is the elastic modulus, γf is the fracture energy, and ν is Poisson’s
ratio. (Note: sometimes the reader may find this equation without the 1-ν2
term. That is the appropriate form for plane stress conditions, but nearly all
practical problems for ceramics and glasses are plane stress conditions. In any
case, since ν is typically between 0.15 and .25 for ceramics and glasses, ν2 is
small and the denominator is close to 1.)
Sometimes the fracture mechanics literature uses a related term, GIc, the critical
strain energy release rate. For plane strain conditions, it is related to KIc by:
EGIc
K Ic = (7.9)
(1−ν 2 )
and GIc = 2γf . The ceramics and glass technical literature of the 1960s and
early 1970s used analyses based on γf and the Griffith equation 7.5, but few
flaws have a simple through-plate elliptical shape. Fracture mechanics analysis
based upon stress intensity KI is more versatile and can model more complex-
shaped cracks. It became more common in the 1970s and is dominant today.
Fracture toughness, KIc, is an important material property that characterizes a
material’s intrinsic resistance to fracture. It has units of stress √length and in SI
units is expressed either as MPa√m or MN/m1.5. (The English system of units
7-22
Quantitative Analysis
7-23
Fractography of Ceramics and Glasses
for fracture toughness is ksi√in and the conversion factor to the SI units is only
1.1 as listed in Table 7.1) Fracture energies are usually expressed as erg/cm2 or
as J/m2. Table 7.2 shows some approximate KIc values. It is much better,
however, to use a value for a specific material, since fracture toughness
depends strongly upon composition and microstructure. Figure 7.16 shows
some crack-microstructure interactions that can affect crack propagation and
hence the fracture toughness.
Glass and
fine grained
crack ceramics
Microcracking
ceramics
Crack branching
Crack deflection
Crack bridging
Second phase
whisker reinforcement
Transformation
toughening
Continuous fiber
reinforcement
7-24
Quantitative Analysis
Table 7.2 includes two columns that require explanation. The third column is
for R-curve behavior whereby fracture resistance increases as a crack extends.
This behavior is caused by several of the crack microstructure interactions
shown in Figure 7.16 such as transformation toughening around a crack tip or
bridging behind the crack tip. The fourth column in Table 7.2 is for “slow
crack growth” which is a phenomenon whereby cracks grow stably at stress
intensities less than KIc. Both of these phenomena are dependent upon the
chemistry and microstructure. It is convenient to categorize materials
according to Table 7.3. For now, the discussion will focus on the simplest
combination in the top left box: a material that does not have slow crack
growth and which has a flat R-curve. Such a material has a single value for
fracture resistance, KIc. Many ceramics (e.g., solid-state sintered silicon
carbide and many fine-grained materials) and all glasses if tested under inert
atmosphere conditions fit this category. Slow crack growth and R-curve effects
are discussed later in this chapter.
Table 7.3 The fracture resistance of ceramics and glasses may be categorized by
whether the material does or does not exhibit slow crack growth (SCG) or rising R-
curve behavior. (After Fuller, private communication.)
7-25
Fractography of Ceramics and Glasses
Figure 7.17b shows a different loading configuration on the same flaw. At first
glance it looks similar, but it is not. Opposing forces P pull on the center of the
crack faces. This is not a far-field stress and equation 7.7 is inapplicable. KI
has a completely different form as shown in the figure. Indeed, the larger the
crack size, a, the less the stress intensity! This can easily be understood by
(a) (b)
Figure 7.17 The stress intensity KI for penny-shaped circular cracks for two
loading configurations. (The three-dimensional axes make the flaws appear
elliptical in this view.) (a) shows a far-field tensile stress σ, and (b) shows
opposed forces P pulling directly on the middle of the crack. KI increases with
crack size in (a), but decreases with crack size in case (b).
7-26
Quantitative Analysis
considering that the effect of the force P is distributed along the flaw’s
perimeter. The larger the crack radius, a, the greater the perimeter and the less
the force per unit length of crack perimeter. (This latter case, or more
precisely the case for a semicircular surface flaw, is commonly used to
represent the effect of concentrated residual indentation stresses from a Vickers
indentation on a median crack flaw.)
a a
2c
2W t
(a) (b)
Ydepth = 1.509
t or h
a a bending
2c 2c
Ysurface = 1.252
(c) (d)
7-27
Fractography of Ceramics and Glasses
The flaw shape affects the Y factor. Y varies around the flaw periphery. A
flaw goes critical when one portion of the flaw reaches KIc, so the maximum Y
is of primary concern. The author has seen a number of instances where
localized fractographic markings confirmed that fracture indeed had started at
the portion of the flaw periphery that had the maximum Y. Figure 7.19 shows
maximum shape factors for a variety of elliptical and semi elliptical flaws
subjected to a far field uniform tensile stress. In each instance, the flaw size for
equation 7.6 is the depth for surface flaws, or the half minor axis length for
internal cracks.
Note that for comparably sized internal or surface flaws, the surface flaw has a
greater Y factor. This is because it has less supporting material in the vicinity
to share the forces. The crack opening displacements on a surface connected
flaw are also greater.
The model cracks shown in Figures 7.18 and 7.19 may be used with equation
7.7 for small flaws that cause fracture in ceramics and glasses. The equation
has four variables: the far-field tensile stress at fracture, σf ; the flaw size, a or
c; the shape factor, Y; and the fracture toughness, KIc. This simple equation
may be used in three different ways:
(1) Flaw size may be estimated if Y, σf, and KIc are known. This approach is
very useful in laboratory testing, whereby specimens are broken under
controlled conditions, and break loads and specimen and fixture
geometries are used to compute failure stresses.
(2) The failure stress, σf, can be estimated if the flaw size and shape can be
measured fractographically, and KIc is known. This approach is useful in
estimating stresses in fractured components.
(3) Fracture toughness, KIc, may be computed if the flaw size and shape, and
the failure stress are known.
Examples will be shown below, but first additional information about the flaw
shape and stress gradients must be considered so that appropriate Y values may
be estimated. The reader should be forewarned that there is no consistency in
the fracture mechanics and materials science literatures. Crack sizes are
usually described by a, b, c, 2c, or other symbol, but sometimes these same
symbols are used differently and sometimes they are even used for the
component dimensions. Sometimes the Y factor includes π or a √π term, but
often it does not. (One reference even uses Y as the inverse of how it is used
everywhere else.) In this Guide, and in ASTM standards C 1322 41 for
fractography and C 1421 42 for fracture toughness, all geometric terms for the
stress intensity shape factor (including π or √π) are combined into Y.
7-28
Quantitative Analysis
Uniform
tension
Figure 7.19 The stress intensity shape factors Y for surface and internal
round and elliptical for internal (volume) cracks and surface cracks for far-
field tension stresses. Minimum Y factors for the surface cracks are shown in
parenthesis. The Y factors for the surface flaws may also be used for parts
loaded in bending if the flaw depth a is small relative to the specimen thickness
(a/t is small).
7-29
Fractography of Ceramics and Glasses
Ceramics and glasses often have surface origins and the stress intensity factor
solutions of Newman and Raju40 for both tension and bending stress fields are
very helpful. Their versatile solutions may be applied to tension, bend bar, and
biaxial disk specimens, and indeed any component shape wherein a remote
stress field causes fracture. With reference to Figure 7.18 for dimensions for
surface flaws in a plate or beam in bending, the Ydepth and Ysurface for locations
at deepest part of the crack and at the surface are:
M
Ydepth = H 2 π
(7.10)
M
Ysurface = S H 1 π
Q
(7.11)
24
a 0.89 a
2 4
0.5 - 1.
+ 14. 1.- a
a
M = 1.13 - 0.09 + − 0.54 + +
c a t a c t
0.2+ 0.65+
c c
(7.12)
a a
H 1 = 1. -
0.34 + 0.11
(7.13)
c
t
a a a
0.75
a
a
1.5 2
H 2 = 1.
-
1.22 + 0.12
+
0.55 - 1.05 + 0.47
(7.14)
c
t
c
c
t
1.65
a (7.15)
Φ
=
Q = 1.+ 1.464 for a/c ≤
1
c
a a
2
(7.16)
S =
1.1
+
0.35
t
c
7-30
Quantitative Analysis
(Equations 7.10 and 7.11 do not include an additional term fw in the original
Newman-Raju analysis, since for nearly all cases for ceramics and glasses, the
flaw width 2c is small relative to the component width, and fw is 1.0.) These
polynomials appear intimidating at first glance, but are easily programmed into
hand calculators or spreadsheets. Notice how many of the dimensions appear
as ratios, e.g., a/c, the flaw ellipticity ratio; or a/t, the ratio of the flaw depth to
the plate thickness). The (1-(a/c))24 term for M in eq. 7.12 is genuine and not a
typographical mistake. The Newman–Raju Y factors have been widely used in
many textbooks and are included in several ASTM standards including C 1421
for fracture toughness of ceramics42 and E 740 for metals, 43 as well as ISO
standard 18756. 44
The above equations are for surface cracks in bending stress fields. Newman
and Raju listed alternative formulas for direct uniform tension loadings.40 It is
not necessary to list them here. Equations 7.10 – 7.16 may be easily used for
cracks in uniform tension loaded parts by the simple expedient of inputting an
arbitrary large value (e.g., 1000) for the part thickness, t.
So for example, if a bend bar is broken and it has a surface crack with a depth
(a) of 40. x 10-6 m; width (2c) of 144. x 10-6 m; and specimen thickness (t) of
2.998 x 10-3 m; then M = 1.080, H1 = 0.995, H2 = 0.983, √Q = 1.247, and S =
0.820. Then, Ydepth = 1.509, Ysurface = 1.252. These are shown in Figure 7.18d.
In practice, the more elliptical the flaw, the more likely the maximum Y is at
the flaw depth (unless the crack penetrates deeply into the specimen interior
and into the stress gradient.)
Using the larger value for Y and the flexural strength which was 453 MPa,
equation 7.7 gives KIc = 1.509 x 453 MPa √40 x 10-6 = 4.32 MPa√m. In this
case, the parameters are used to estimate the material’s fracture toughness.
As noted on page 7.28, equation 7.7 for KIc can be used two other ways. If
instead, the fracture toughness and flexural strength were known, the flaw size
could have been estimated. Similarly, if the crack size had been measured and
the fracture toughness was known, then the stress of 453 MPa could have been
estimated.
7-31
Fractography of Ceramics and Glasses
Figures 7.20, 7.21, and 7.21 show three examples of how the calculation can be
used to confirm that the right flaw has been identified. Sometimes the
calculated and measured flaw sizes agree very closely, but there are a number
of reasons why they may differ as discussed below in section 7.8. ASTM
standard C 132240 states that if the calculated and measured flaw sizes do not
agree within a factor of two or three, then the origin should be reexamined to
verify that the correct feature has been identified.
The surface crack in the BK-7 borosilicate crown glass disk shown in Figure
6.28a,b and 7.22 was 7.3 µm deep and the stress at fracture was 146 MPa. For
a long surface crack, Y is 1.99. The calculated fracture toughness from these
values is 0.78 MPa√m, which is very close to published values for this
particular glass.
7-32
Quantitative Analysis
(a)
(b)
10 µm
Figure 7.21 Fracture origin in a single crystal silicon wafer. (a) shows the
fracture mirror which has a very faceted shape typical of some single crystals.
The origin is a grinding crack shown by the white arrows in (b). The specimen
was tilted back to show both the fracture and outer ground surfaces. A 3000 X
scanning electron microscope photo showed that the crack was 2 micrometers
deep, and when combined with a 282 MPa fracture stress, and a Y factor of
1.99, gives KIc = 0.79 MPa√m, in reasonable good agreement with published
values for fracture on the {110} plane.
7-33
Fractography of Ceramics and Glasses
(a)
(b) (c)
Figure 7.22 Fracture origin crack for the borosilicate crown glass biaxial
disk specimen that was broken in inert conditions as shown in Figure 5.17. (a)
is an SEM photo of the distorted mirror. The arrows show the long polishing
scratch. . (b) and (c) show two portions of the polishing scratch which was 7.3
mm deep (arrows). Using Y = 1.99 and the breaking stress of 146 MPa, gives
KIc = 0.78 MPa√m. The optical image (see Figure 5.17) showed a slight bulge
in the flaw at the center of the mirror suggesting the flaw may have grown
slightly before going critical. This growth is not evident in the SEM images.
7-34
Quantitative Analysis
Figure 7.23 may be useful for those who do not wish to program a calculator or
spreadsheet. Y values may be picked off the graph for cracks in bending
problems for given a/t and a/c ratios. Notice it demonstrates how Y decreases
as a flaw penetrates into the stress gradient (larger a/t). (Y values for small
surface flaws in uniform tension fields may be obtained by reading the graph
value for an a/t ratio of zero.)
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
2.0 2.0
Bending compression side
t, plate thickness
1.9 1.9
tension side a
1.8 2c2c 1.8
1.5
c/a = 10 1.5
c/a = 6
1.4 c/a = 4 1.4
c/a = 2
1.3 c/a = 1.4 1.3
Ymax
1.2 c/a = 1 is at the surface 1.2
1.1
Ymax
1.1
is at the surface
c/a = 1 semicircle
1.0 1.0
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Figure 7.23 The maximum stress intensity shape factor Y for semielliptical
surface cracks in a bending stress field in a plate or bar of thickness t. In the
limit as the flaw is very small relative to the plate thickness, a/t → 0, the
solutions are the same as for the uniform tension case. The kink in the loci
corresponds to where the maximum Y changes from the surface to the deepest
point in the interior or vice versa.
7-35
Fractography of Ceramics and Glasses
Occasionally the fractographer may find a flaw that is deeper than a semicircle,
i.e., a/c > 1. Fett devised modified Newman-Raju formulas that permits one to
compute the Yd and Ys factors in such cases for either tension or flexural
loadings. The conscientious reader may consult a paper by Fett 45 for the
modified formulas. Rather than listing all the revised formulas or showing
charts or tables, some approximations are provided here.
In the cases with semi-ellipses that are deeper than a semicircle, the maximum
Y is at the surface, Ys. For cracks that are shallow relative to the plate
thickness (a/t ≤ 0.2) in tension or bending, the surface Ys factor decreases
linearly from that for a semicircle by only 10% as a/c goes from 1 (the
semicircle) to 2 (a very elongated crack into the thickness).
The stress intensity shape factor Yd for the deepest part of the flaw also
decreases approximately linearly, but by greater amounts. Yd decreases by
33 % in both tension and bending as a/c goes from 1 to 2. If the flaw is larger
relative to the plate thickness (e.g., a/t = 0.2), Yd decreases by 33% and 45% in
tension and bending, respectively, as the a/c ratio goes from 1 to 2. In other
words, as a very deep semi-ellipse penetrates into the stress gradient in bending
(a/t = 0.2, or 40% of the way to the neutral axis), Yd diminishes rapidly.
A few words about the accuracy of the Newman-Raju stress intensity factors Y
are in order here. For most fractographic analyses, an accuracy of a few
percent is perfectly adequate. Developed by finite element analysis almost 40
years ago, the Newman-Raju factors are widely used in textbooks, fracture
mechanics and fractography standards around the world. Fett has reviewed46,47
them carefully and concluded that for most common configurations, they are
accurate to within 3%. Fett noted that a solution of Isida et al., 48 obtained by
means of a modified body-force method exhibits superior accuracy with a
maximum deviation of less than 1% for Yd.
Even more accurate and versatile solutions have been developed recently by
Stobl et al. 49. For most common semi-elliptical cracks their Yd solutions
concur with Newman and Raju’s within a few percent, but the newer solutions
cover a broader range of Poisson’s ratios and cover surface cracks that have
peripheries that do not reach the specimen surface at exactly 90o. Strobl et al.’s
cracks are sections of a semiellipse and in some cases they are better models.
7-36
Quantitative Analysis
Bansal reviewed the stress intensity shape factors for semi-elliptical surface
flaws in uniform tension 50 and arrived at a simple approximation:
2.0 4
𝐾𝐾𝐼𝐼 = 1.68 𝜎𝜎 √𝐴𝐴 7.17
where A is the area of the flaw. Since the area of a semi-ellipse is πac/2:
7-37
Fractography of Ceramics and Glasses
K i = 1.15 σ 4
A (7.20)
where A is the area of the flaw. For example, if the flaw is semicircular with
radius a, this simplifies to
K I = 1.29 σ a (7.21)
that is nearly the same as equation 7.19. Numerical analysis showed that the
fourth root of area equation 7.20 is an adequate approximation for stress
intensity factor solution for such diverse shapes as partial circles, triangles,
rectangles and semi ellipses. Equation 7.20 is also for the case of uniform
tension. It may be used for small cracks in bending specimens, provided that
the crack does not penetrate very deeply into the specimen, a/t < 0.05. The area
approximation should not be used for very long shallow flaws c/a > 5.
7-38
Quantitative Analysis
Figure 7.25 An irregular planar shaped surface crack in a tension stress field.
(After Murakami, ref. 50.)
Real flaws are not always flat and planar. Figure 7.26a shows a sphere that is
not unlike many pores in ceramics and glasses. Sometimes round pores are
modeled by a penny-shaped crack, Figure 7.26c. This is a gross over-
simplification. Indeed, if the pore is round and crack free, it does not act to
intensify the stress, but simply concentrates stress around the flaw sides. The
stress concentration factor for an interior sphere in a part in uniform tension is
only k = 2. Smooth round pores require much greater force to cause fracture
than the penny-shaped flaw of the same diameter. Round pores may be quite
docile and not act as strength-limiting flaws. Figure 7.26b shows another
variation. This rim-cracked pore (“Rings of Saturn”) flaw is a combined stress
concentration and fracture mechanics problem. For small rim cracks, the pore
acts as a concentrator of stresses, and the rim crack size, c, may be regarded as
the crack size in a fracture mechanics calculation. For a very large rim crack,
the pore’s effect is diminished, and the penny shaped crack is a very good
approximation. In practice, it is very difficult to detect a rim crack around a
pore, much less measure it. It is not unreasonable to expect that rim cracks of
one or two grain diameters extend from the pore and in fact many pores in
ceramics do have local microstructural irregularities or microporosity in the
vicinity. Only very rarely have fractographers attempted to precisely model
7-39
Fractography of Ceramics and Glasses
flat
Figure 7.26 Flaws may be three-dimensional. A flat penny shaped crack (c)
is a two-dimensional approximation.
KIc and the fracture mirror and branching constants are related to σ√a or σ√R
and hence have the same set of units: MPa√m (or ksi·√in). As Figure 7.7
illustrates, KIc pertains to conditions at the flaw. The A values correspond to
events away from the origin as the crack propagates and reaches terminal
velocity. Kirchner and Conway30 showed that mirror shapes can be predicted
on the basis of stress intensity at the crack front as it expands outwards from
the origin, so it is not unreasonable to expect that KIc and the A’s are related.
If they are related, then in principle one can estimate KIc from A or vice versa.
Mecholsky et al. 52 tabulated A’s and KIc’s for over two dozen ceramics and
7-40
Quantitative Analysis
glasses and showed there was an approximate correlation. On average the ratio
of Ao to KIc was 20/7 = 2.86, although there was considerable scatter. Bansal
and Duckworth 53,54 carefully measured Ao to KIc ratios of 2.2 to 2.4 for a glass
ceramic and a sintered alumina, and 2.3 for float glass and a hot-pressed
alumina. Subsequent tabulations 55 on more contemporary materials, many of
which have R-curve behavior, have given ratios as low as 1.5 to 2.0.
Kirchner et al. 56 showed that the branching constant Ab varied by over a factor
of 2 with the degree of transformation toughening in a calcia partially-
stabilized zirconia. The greater the toughening, the greater the suppression of
branching to greater distances; i.e., the Ab/KIc ratio increased with increased
toughness.
In a dissenting view, Bradt and colleagues 57,58 have suggested that A is not
really a material constant and noted the discrepancies in A/KIc ratios. In
summary, variability in the quality and accuracy of the KIc and A data, plus
genuine microstructural effects, probably mean there is no unique A/KIc ratio.
A simple but very useful observation is that mirror and flaw sizes are related.
Krohn and Hasselman 59 showed that the mirror size to flaw size ratio (R/c) is
about 10 for glasses. Mecholsky et al. 52,60 obtained ratios of 10 to 11.6 using
the mirror-mist (inner mirror) boundary in glasses. They estimated the ratio to
be about 13 for the mist-hackle (outer) boundary to flaw size ratio for glasses.
Ratios for ceramics are more problematic and depend upon the microstructure
and the ability to judge a specific boundary. Mecholsky et al.51 estimated the
ratio of the hackle boundary (outer mirror) to flaw size was 13 for polycryst-
alline ceramics and single crystals, but varied from as low as 8 to as large as
40!
Rice 61 pointed out that the bluntness of the flaw may alter the mirror to flaw
size ratio. For example, pores often act as blunt flaws and it requires extra
energy or stress to overcome the bluntness. The excess energy causes a mirror
to form at shorter distances from the origin than if the flaw were sharp. Thus
the R/c ratio is smaller.
Slow crack growth can also alter the apparent ratios.60 If the mirror to flaw
size ratio (R/c) is much greater than the values listed above, then the flaw
should be checked for evidence that it grew stably prior to fracture.
7-41
Fractography of Ceramics and Glasses
Residual stresses in the vicinity of the flaw or the mirror also can change the
ratios. Unexpected mirror to flaw size ratios may be evidence of residual
stresses acting on the flaw or the whole mirror.
The mirror to flaw size ratio (R/c) may also be affected by the presence of
mixed Mode I/II loading. Gopalakrishnan and Mecholsky 62, and Rice 63 have
shown that increasing Mode II loading on a crack decreases R/c.
The R/c ratio is also affected if a material has a rising R curve. The larger the
fracture toughness, the lower the R/c ratio.61,64 In rising R-curve materials, the
ratio may even depend upon how large the initial crack was and how far it
might have stably grown before it went unstable. (Stable crack growth and
crack instability is discussed more in sections 7.10 and 7.11 later in this
chapter.) In R-curve materials, the larger the initial crack, the smaller the R/c
ratio. Gopalakrishnan and Mecholsky’s 2013 paper62 even goes so far as to
analyze the combined effects of R-curve and mixed Mode I/II behavior on R/c
ratios. (Note, their discussions are on the ratio, c/R, unlike most other work
which focuses on the R/c ratio.)
Ccalc < Cmeas Ccalc > Cmeas Ccalc > Cmeas or Ccalc < Cmeas
Crack blunting Stable crack extension -SCG High temperature SCG
Use of 2-D model Stable crack extension – R Multiple flaws nest or
curve interact
Stress gradients Flaw causes a local fracture Residual stresses
toughness degradation
Crack healing Flaw is within a single grain Flaw is truncated on the
fracture surface
Flaw links with other flaws or Flaw shape irregularity
the surface
Flaw elastic properties are
different than the matrix
Faulty fracture toughness
data
Table 7.4 Factors that can cause a measured flaw size cmeas to differ from the
calculated ccalc size. (Quinn and Swab, Ref. 65 and 66.)
7-42
Quantitative Analysis
The cases where ccalc < cmeas are considered first. A blunt flaw of actual size c
requires a greater stress to cause fracture than a comparably-sized sharp flaw.
Some flaws are blunt to begin with. Other initially-sharp flaws may have been
blunted or even partially healed by oxidation or other environmental effects.
Glass cracks can partially heal in vacuum or inert atmospheres, especially if
they form during dynamic events and are unloaded quickly. Inserting the
greater stress to break healed or blunted cracks into equation 7.22 leads to a
smaller ccalc than the actual flaw size c. Assuming that flaws are 2-D penny
shaped or are flat ellipses when they really are 3-D flaws, usually leads to Y
estimates that are too large, and this also leads to smaller ccalc than the actual
flaw size c. Unexpected stress gradients can cause σf to be overestimated.
That in turn will again cause ccalc to be too small. The stress at the origin site
should be used in equation 7.22.
Next are the cases where ccalc > cmeas. Slow crack growth and rising R-curve
behavior can cause stable crack extension. Using the actual breaking stress in
equations 7.7 or 7.22 gives a correctly calculated flaw size, but if the
fractographer detects and measures only the initial flaw size, then ccalc > cmeas.
In this case it is the fractographic size measurement that is off. Alternatively, if
stable crack extension due to R-curve behavior does occur, but the
fractographer uses a long crack (plateau) value of fracture toughness, then the
fracture toughness is overestimated and again ccalc > cmeas . R-curve behavior is
discussed in more detail in section 7.11. Inclusions or chemical inhomo-
geneities can be particularly deleterious if they degrade the fracture toughness
of the surrounding matrix. If the fractographer uses the baseline KIc value in
7-43
Fractography of Ceramics and Glasses
equation 7.22, this also leads to ccalc > cmeas. In this case, the calculated size is
wrong and the observed crack size is correct. The fracture toughness of a
single crystal, or a single grain, is almost always less than that of a polycrystal.
If a flaw resides within a single grain, but a polycrystalline KIc value is used in
equation 7.22, then ccalc will be too large. Flaws may link up with other flaws or
a surface weakening the specimen. If the linkage is undetected by the
fractographer, then his cmeas will be too small compared to the calculated and
actual critical flaw size.
Some factors can cause the measured size versus calculated size discrepancy to
go either way. Flaws are often assumed to exist in isolation, but as Figure
7.27b shows they can interact causing either an increase or a decrease in the Y
factors that act on them. It all depends upon the geometries and arrangements.
This is an instance where the “flow lines of stress” concept can aid
interpretation. If flaws are lined up in a row parallel to the stress, the flaws will
shield each other. If they are staggered or lined up sideways, they can intensify
the KI fields and the Y factors. Flaws may be truncated and only a portion
detected on the fracture surface. Undetected residual stresses can throw off the
calculated flaw size as well. Faulty fracture toughness data can also throw off
the correlations.
The point here is that fracture mechanics is a valuable tool to aid fractographic
analysis. It can help confirm that the fractographer has found the correct origin
flaw. A finding that the calculated and measured sizes are appreciably different
should prompt further consideration and analysis. What factors caused the
discrepancy? Is the discrepancy an isolated outcome or part of a pattern?
Figure 7.27 Factors that cause calculated and measured crack sizes to differ
(continued on next page)
7-44
Quantitative Analysis
7-45
Wallner lines can be used to estimate crack velocities, whether they are created
incidentally during the fracture event or deliberately by ultrasonic fractography.
Wallner lines may not be evident in very slow fractures in glass or in ceramics
due to the latter’s rough fracture surfaces.
E (7.23)
vt =
2 ρ (1+ ν )
where E is the elastic modulus, ρ is the density, and ν is Poisson’s ratio.
Richter66,67 studied cracks running at terminal velocity at stress intensities
above KIc.
7-46
Quantitative Analysis
to produce the gull wing shape. The included angle (2β) between the wing
segments may be used to calculate crack velocities. Crack velocity is quite
simply:
vc = vt cos β (7.24)
where vc and vt are the velocities of the crack front and the transverse wave.
Slowly-moving cracks do not advance very far in a time interval ∆t, and the
elastic pulse quickly overtakes the entire front. Thus, the “V” shape is very flat
and, if it assumed that the included angle 2β is 180o – 2o = 178o, then the
slowest velocity that can be measured in a glass with vt = 3460 m/s is vc,min = 60
m/s. Conversely, if the crack is advancing at the terminal velocity of 60 % of
the transverse wave velocity, then the minimum included angle 2β is 106o.
Tsirk 71 identified hyperbolic “V” shaped Wallner lines in obsidians, created by
an elastic pulse generated by secondary fracturing behind the main crack front.
vt vt
3
2 Crack
2β 1 front
vc
0
3 2 1
Wave front
dcp (b)
(a)
v c ∆t
cos β =
vc ∆t β vt ∆t
v t ∆t
(c)
v c = v t cos β
Figure 7.28 Crack velocity may be estimated from Wallner line analysis. Gull
wing lines occur when a straight crack front passes a point discontinuity such
as a bubble or inclusion. (a) shows gull wings in knapped obsidian (specimen
courtesy A. Tsirk). The direction of crack propagation is from bottom to top.
The included angle (2β) is used to estimate crack velocity. Slow-moving cracks
have flat (horizontal) wings. Fast-moving cracks create wings with a “V”
shape. (b) and (c) show the geometries and simple trigonometry used to
estimate crack velocity.
7-47
Fractography of Ceramics and Glasses
Wallner lines such as those inside a mirror shown in Figure 7.29 may be
analyzed as shown in Figure 7.30. 14,27,72,73
Wallner
line
Radiating
wave
(a) Radiating
crack
(b) (c)
Figure 7.29 Crack velocities can be estimated from Wallner line analysis. (a)
shows a crack front that radiates outward from the origin at velocity vc (thin
circular lines) and reaches irregularity A on the left. An elastic pulse is
generated at A and it radiates a transverse (shear) wave outward from A
(dashed lines) with a faster speed vt than the crack front. The numbers label
the crack and elastic pulse fronts at the successive times starting at 0, the
instant the crack reaches A. The intersections for the specific times are marked
by dots, but of course this is a continuous process and the solid line shows the
ensuing Wallner line. The crack front is assumed to be moving at constant
velocity vc = 0.5 vt in this case. If the crack were accelerating, the Wallner
line would be less hooked in the beginning and would extend more towards the
upper right. (b) and (c) show Wallner lines inside mirrors in glass rods.
7-48
Quantitative Analysis
v t ∆t
cos β =
v c ∆t
sin γ =
origin
(a) (b)
Figure 7.30 Analysis of Wallner lines to determine the crack velocity if a crack
radiates outwards from an origin. (a) shows a case where the location of the
Wallner line elastic wave pulse (site A) and the direction of the crack extension
are known. (b) shows the case where the precise crack growth direction is not
known, but the origins (A1 and A2) of the Wallner lines are known.
In the case where the exact direction of crack propagation (radiating at speed vc
from the origin) and the initiation site of the Wallner line (A) are known:
vc sin γ (7.25)
=
v t cos β
Crack velocity can be determined even if the crack propagation direction is not
known. Two intersecting Wallner lines may be analyzed as shown in Figure
7.30b:
vc sin ϕ
= (7.26)
vt cos β1 + cos β 2 + 2cos β1 cos β 2 cos ϕ
2 2
7-49
Fractography of Ceramics and Glasses
Elastic waves interactions from slowly-moving cracks run across the crack
front fairly quickly and thus the Wallner line corresponds closely with the crack
front shape. The minimum velocity that can be estimated by Wallner line
analysis was estimated by Mencík as vc/vt = 0.175, using equation 7.27 with ϕ
=2o, corresponding to nearly flat crossing Wallner lines. For glass with vt =
3460 m/s, then vc,min = 60 m/s. So in general, quantitative Wallner line analysis
for crack velocities is most suitable for fast moving cracks.
The spacing between the legs of Wallner lambda lines, discussed in section
5.4.5 and shown in Figure 7.31, can easily be used to estimate a crack velocity.
In the time that the crack advances approximately a distance a/2, the elastic
wave has traveled across the thickness once. For a straight crack front:
a
Vc = V s (7.28)
2 ( 2)
a
2
+t 2
The formula can also be used to approximate crack velocities for slightly
curved crack fronts as well.
Propagating
t
crack front
Figure 7.31 The spacing between the lambda line legs can be used to estimate
crack velocity.
7-50
Quantitative Analysis
Cracks may grow stably when loaded at stresses and stress intensities less than that
necessary for the flaw to become unstable. As discussed in sections 5.9.1 and 5.9.3,
glasses, oxide ceramics, and ceramics with glassy boundary phase are susceptible.
The mechanism of slow crack growth at ambient temperature is a stress corrosion
phenomenon whereby stressed cracks that are open to the environment are attacked
by water (or another polar molecule.) The water may be in the form of liquid or
gaseous molecules. The water molecules attack the strained silicate or oxide bonds
at the crack tip causing them to rupture which in turn leads to stable crack extension.
Other fluids that promote slow crack growth are listed in section 5.9.1. At elevated
temperature, oxidation attack, grain boundary softening, or other mechanisms can
lead to slow crack growth (section 5.9.3).
Crack velocity depends strongly upon the stress intensity, KI. The larger KI, the
faster the crack grows. Slight changes in it have dramatic effects upon crack
velocity. As a crack size increases, it increases KI in accordance with equation 7.6.
If stress and the crack shape are invariant, the crack will gradually accelerate. Data
are often graphed on log - log axes as shown in Figure 7.32a. Straight lines imply a
power law dependence of velocity upon stress intensity:
V = A KI
N (7.29)
Figure 7.32b shows how slow crack growth affects the lifetime of specimens
loaded at constant stress levels below the fast fracture strength. This is referred
to as “static fatigue” or “stress rupture” testing. The decrease in log strength
varies with log time to the –1/N power.
Slow crack growth can even affect outcomes in ordinary strength tests as
shown in Figure 7.32c. The slower the test is run, the weaker are the specimens
since the crack can grow a surprising amount during the time of loading.
7-51
Fractography of Ceramics and Glasses
(a) (b)
(c)
Figure 7.32 Slow crack growth manifests itself in different ways. (a) shows a
crack velocity versus stress intensity (K-V) graph showing three regions.
Usually such data is plotted on log-log axes. Fracture mechanics type tests are
used to collect such data. (b) shows a graph of log stress – log time to failure,
also known as a “static fatigue” or “stress rupture” plot wherein a constant
stress is applied until a piece breaks. (c) shows a graph of strength as a
function of stressing rate also with log - log axes, also known as a “dynamic
fatigue” plot. The slow crack growth exponent N plays a prominent role in
each instance.
7-52
Quantitative Analysis
Crack-fluid interactions can cause abrupt changes in crack growth that can
create markings such as scarps on fracture surfaces as shown in section 5.6.
For example, an accelerating crack may outrun the ability of a fluid to keep up
with the crack tip and this may cause cavitation and a scarp. Conversely, a
decelerating crack may slow down enough that water can catch up to the crack
causing a scarp. Sometimes a jump in the crack from region I behavior to
region II behavior creates a scarp.
The Newman-Raju stress intensity shape factors also explain one commonly
observed trend: surface cracks that grow stably evolve shape and become
semielliptical. Figure 7.33 illustrates the trend for a beam or plate in bending
and Figure 5.44b an example. The stress intensity shape factor Y varies around
the crack front. For a semicircular flaw, Y is greatest at the point where the
crack intersects the outer surface, point B. At the internal point A, there is
more material on either side of the crack front that can share the load, and
hence the stress intensity shape factor is about 13 % less. Thus, the crack
initially grows faster along the outer surface, but then, as the crack shape
evolves into a semiellipse, the stress intensity evens itself out around the
periphery. Newman and Raju40 showed examples of this for fatigue crack
growth in metals, both for bending and direct tension stress fields. The
semielliptical shapes differ in the two cases, depending upon the size of the
crack relative to the specimen thickness. The identical trends have also been
observed in ceramics and glasses.67,74,75,76,77,78 Small surface cracks in large,
uniform, direct-tension loaded specimens or plates will assume a constant semi
elliptical shape with an aspect ratio a/c of about 0.83. Cracks in bending stress
fields will also gravitate to this shape if they are small relative to the thickness,
but if they penetrate into the stress gradient, they will become more elongated
sideways as they grow and the semiellipse ratio a/c will decrease as shown in
Figures 7.33b and c.
7-53
Fractography of Ceramics and Glasses
(a)
2cc
2co
(b)
B• ao
A• ac
t
(c)
(d)
Figure 7.33 Crack shape often changes with growth. (a) shows an initially
semicircular crack of size ao that changes to a semi elliptical shape as it grows
in a bending stress field in a plate of thickness t. (b) shows concentric arrest
lines from cyclic loading (same as Figure 5.47c) (c) shows tertiary Wallner
lines from a rapidly moving crack (same as Figure 5.44b); and (d) shows the
evolution of crack shapes in glass bend specimens converges to a shallow
semiellipse, irrespective of the starting shape (hollow circles). (b courtesy D.
Green. c is courtesy H. Richter. d is after Fett et al., Ref. 74)
7-54
Quantitative Analysis
Figure 7.34 illustrates the general concepts. Flat R-curve materials such as
glasses and many fine-grained ceramics behave as shown in Figure 7.34a.
Fracture toughness is independent of the crack size and has a specific value:
KIc. When the combination of stress, flaw shape, and size reach the critical
condition, unstable crack extension usually occurs. If there are substantial
tensile stresses throughout the body, it will break. If, on the other hand, the
stresses decrease either temporally as in thermal shock cases or spatially as in
localized contact loading cases, the crack initially may be unstable, but then
slows down and may even stop.
Phenomena such as these can lead to behavior shown in Figure 7.34b. Starting
with an initial crack size, co, if the stress intensity Ka from an applied stress
reaches the level of Ko, the crack begins to propagate. As it grows, Ka increases
in proportion to √c if the stress is held constant as per eq. 7.6. The material’s
fracture resistance KR may increase at a greater rate thereby retarding or
arresting further crack propagation. Further propagation requires additional
applied stress to increase Ka. The R-curve typically reaches a plateau such that
further toughening is not possible. Fracture may occur before the plateau is
reached if:
7-55
Fractography of Ceramics and Glasses
(a)
K
(b) - Ko
(c)
Figure 7.34 R-curves in ceramics. (a) shows the case of a material with a
constant, or flat R-curve. (b) shows a general schematic of a rising R-curve.
Crack extension leads to a greater applied stress intensity Ka, but the
material’s fracture resistance curve KR may also increase. (c) shows Ka versus
√c, in which case the Ka plots as a straight line with slope proportional to
applied stress. Fracture occurs at ccrit where the rate of increase of applied Ka
is greater than the slope of the KR curve. Initial flaws larger than ccrit may not
exhibit any stable crack extension and may immediately go critical.
7-56
Quantitative Analysis
(7.30)
Ka = KR
and
dK a dK R
>
dc dc (7.31)
In other words fracture occurs when the applied Ka reaches the level of the R-
curve and the rate of increase in Ka with further extension is greater than the
rate of toughening. To better show the relative increases, it is common to plot
fracture resistance with K2 versus c, or K versus √c as shown in Figure 7.34c.
In this fashion, the increase in Ka with crack size appears as a straight line with
slopes that increase with σ. This Ka line can be more easily compared to the KR
trend. Instability occurs at c = ccrit, a value well before the plateau is reached,
after only a small amount of crack growth has occurred.
There is no universal R-curve for a material. 79 The R-curve depends upon the
starting size of the crack relative to the microstructure and its prior propagation
history. It also depends upon the size, shape, and mode of loading of the
specimen. R-curves for large cracks are not the same as those for small cracks.
In other words, data from large crack fracture mechanics specimens are usually
not applicable to natural small flaws. Large fracture mechanics specimens
usually are effective in measuring plateau toughness values. A double-
cantilever beam specimen has a dramatically different crack opening
displacement than a notched beam or a surface flaw in a bend bar. The number
of bridges behind the crack tip bridges depends upon the crack opening
displacement and the greater the opening, the fewer the intact bridges. For
example, a 20 µm grinding-induced crack may have already interacted with the
microstructure and formed bridges or transformed some material during the
crack formation. Figure 7.35 shows two possible examples. In contrast, a 20
µm round pore may not have activated any prior toughening.
7-57
Fractography of Ceramics and Glasses
(a)
(b)
7-58
Quantitative Analysis
Many studies have used Vickers indentation artificial flaws, but they often do
not behave like genuine flaws as discussed in the next section. The R-curves
that are generated are not relevant to natural flaws or are simply wrong. For
example, one study showed that without proper care, the usual analysis with
Vickers indentation flaws produced the implausible result that a soda lime
silica glass has a rising R curve. 82 (The problem was that the standard
indentation strength model fails to account for changes in crack shape,
especially as cracks penetrate into strong flexural stress gradients.) The most
credible R-curve data to date (that is applicable to strength limiting flaws) has
been that collected by careful microscopic examination of crack growth from
natural flaws in controlled loading experiments.77,78,83,84
The instability point depends upon the starting crack size and the slope of the
R-curve at small crack sizes as shown in Figure 7.34c. Indeed, the overall
plateau value may be inconsequential in many cases. A particular crack may
grow only a small amount before going unstable79,80,85 or it may immediately be
unstable. Fett and Munz’s mathematical analysis is quite revealing.85 They
showed that R-curve behavior led to negligible strength enhancements in an
alumina that had bridging grains, quite simply because the natural flaws did not
have a chance to grow very far before going unstable. Their analytical and
experimental results may easily be understood by the simple realization that in
strong aluminas the flaws are only 5 to 20 times larger than the average grain
size. The stable crack extension may involve only a few grains before the flaw
goes critical and it is impossible to generate many behind-the-crack tip bridges.
Perceptible strengthening would only occur in very weak specimens (σ < 100
MPa) with large initial flaws. Fett and Munz also showed that the R-curve
behavior in their aluminas had negligible effect on the Weibull distribution85
contradicting predictions of dramatic improvements in Weibull modulus based
on indentation mechanics analysis. 86
7-59
Fractography of Ceramics and Glasses
The fractographer should keep the above factors in mind. As noted above, R-
curve behavior does not leave any intrinsic telltale markings on the fracture
surface. R curve behavior can throw off flaw size calculations if they are based
on a single point value for fracture toughness in equations 7.7 or 7.22. In
principle, the fractographer may be able to find a region of stable extension
around a flaw, but in practice this may be difficult especially if the toughening
phenomena and the microstructure produce a rough fracture surface. One
indication of potential R-curve behavior may be a systematic ccalc ≠ cmeas trend.
As noted previously, if the fractographer detects and measures only the initial
flaw size, then ccalc > cmeas. In this case it is the fractographic size measurement
that is off. Alternatively, if stable crack extension due to R-curve behavior
does occur, but the fractographer uses a long crack (plateau) value of fracture
toughness, then the fracture toughness is overestimated and the calculated crack
size is wrong, and again ccalc > cmeas.
7-60
Quantitative Analysis
Boron carbide
(a)
Lateral
Crack
Median
Crack
Figure 7.36 Indentation generated damage sites. (a) and (b) show common
crack patterns and examples around Vickers and Knoop indentations. The
author prefers the Knoop indenter since it makes a simple crack system.
7-61
Fractography of Ceramics and Glasses
(a) (b)
(c)
Figure 7.37 The surface crack in flexure method uses a Knoop indenter to
create a semi elliptical surface crack in a bend bar (a). (b) shows a section
view of the indentation and the median crack below it. The residual stress
damage zone is removed by polishing or hand grinding leaving a controlled
surface flaw in the test piece. (c) shows a Knoop crack in a silicon nitride
(arrows).
7-62
Quantitative Analysis
Controlled flaws for fracture mechanics experiments should be made with the
Knoop indenter. 88,89 The flaw geometry is simpler and more easily controlled.
Loads of 19.6 N to 294 N are used to create a semi-elliptical surface cracks in
ceramic bend bars (Figure 7.37). Loads of 19.6 N to 29.4 N are used for
glasses. The indentation residual stress damage zone and any lateral cracks are
removed by polishing and then the test piece broken in flexure. Fractographic
techniques are used to find and measure the flaw size. Other precrack
examples are shown in Figures 3.26, 5.55, 5.57, and 6.53.
Vickers indentation procedures are widely used since they usually do not
require fractographic examination. This supposed advantage leads to severe
drawbacks with the method, however. In 2006 94 and 2007 95 the author and
Prof. R. Bradt wrote about the limitations of Vickers indentations for fracture
mechanics studies. Vickers generated flaws are much more complex than
Knoop median cracks, and since the Vickers flaws are tested without removal
of the residual stresses and damage zones, all sorts of complications arise.
Rather than a single median crack, Vickers indentations have multiple median,
radial, and lateral cracks. The cracks also have a complex stress residual stress
damage associated with them. 96,97 Rather than envision these Vickers
indentation sites as model flaws with discrete median cracks with a modest
residual stress pulling on the faces, it is actually more appropriate to regard
them as Vickers damage zones. A semicircular median crack is typically
assumed to pop in underneath the indentation, although the crack type and
formation sequence is quite complicated and varies dramatically with
material. 98,99,100 The residual stress damage zone is approximated by a point
force acting to open the crack. The stress intensity factor KIr acting on the flaw
7-63
Fractography of Ceramics and Glasses
If a test piece with a Vickers flaw is broken, KIr combines with the stress
intensity created by a far field tensile stress to produce a total KI:
P (7.33)
K I = Yσ a c + χ
c1.5
The first term on the right dominates for large cracks. The second term is
significant for small crack sizes, but rapidly diminishes as the crack extends
away from the residual stress damage zone. The net effect of the two terms is
to cause a crack to extend stably somewhat prior to catastrophic propagation.
The extent of stable propagation has been estimated to be 2.5 times the original
flaw size when the testing is done in inert environment and all the assumptions
in the analysis are upheld.97
It is clear from above that the local residual stresses dramatically alter the local
stress intensity field, and that a simple application of equations 7.6 (or eq. 7.33
without the second term on the right) based on only the far-field tensile
stresses, will be faulty. Typically ccalc will be larger than cmeas, if the latter is the
original flaw size and stable crack extension is not detected.
By the time the crack propagates to the extent that it is forming the mirror
boundary markings, the effect of the localized residual stress term is
insignificant.101 Hence, the mirror size is unaffected by the indention residual
stresses and the stress-mirror size relationship equation 7.4 is unaffected. The
mirror to initial flaw size ratio is affected, however, and can be larger than
values for annealed flaws.101 Indentation residual stresses do not affect
branching distances or the fragmentation patterns.
7-64
Quantitative Analysis
The observant reader will note that Vickers indentation flaws do not resemble
the majority of origins shown in Chapter 6. Notwithstanding claims to the
contrary, it is not at all clear whether indentation damage sites simulate genuine
flaws in ceramics and glasses. They may model contact damage sites and/or
machining cracks, but Figure 6.30 in the previous chapter shows that actual
contact damage sites are often more irregular, with significant amounts of
material removed by lateral cracks and spalls. The contact-generated residual
stresses are often partially or completely relieved. Service impact conditions
are rarely as controlled as those in an indentation process, where an ideal
diamond slowly applies load perpendicular to the surface.
7-65
Fractography of Ceramics and Glasses
Area = π D
2
(7.34)
4
and
Perimeter = π D (7.35)
so that:
Perimeter ∝ Area (7.36)
This is the case for a perimeter line that is unwavering and follows the circle
rim exactly. If on the other hand, the perimeter line had perturbations and
wiggles, the perimeter length is greater. If the perimeter line is magnified and
studied in greater detail, it is possible that each wiggle or perturbation itself
may be seen to have yet smaller wiggles creating additional length and so on.
That is the nature of fractal dimensions. The greater the magnification used,
and the finer the measuring stick used to measure the lengths, the longer is the
overall perimeter length. This leads to the somewhat unsatisfying outcome that
the perimeter length actually varies with magnification and the size of the
measuring scale. The perimeter of a fractal shape around an area can be
characterized by the following relationship:
L = L0 s 1−D (7.37)
Two ways to measure the fractal dimension of a fracture surface are the
fracture profile technique and the slit island analysis.103 The fracture profile
technique simply makes a vertical cross section through a fracture surface. The
profile with all its ups and downs is analyzed. In the slit island technique, a
horizontal cross section is made through a fracture surface to reveal islands
corresponding to the peaks of hills and ridges and their wiggly outlines. This
process is repeated and measurements remade much like in a classical
metallographic serial section analysis. Both procedures can be tedious, but
computer analysis and image analysis software can simplify this task. Confocal
7-66
Quantitative Analysis
Fractal analysis has usually been used to correlate fracture toughness with the
fractal dimensions. The processes that enhance fracture toughness often
increase the roughness of a fracture surface. Passoja, Mecholsky and
colleagues have pioneered correlations of this type for ceramics.102,103,106,107
For example, they have shown that:
( )
K Ic = K o + A D *
1
2 (7.38)
Mecholsky and Freiman107 showed that variations in the mirror to flaw size
ratio correlated with the fractal dimension of the fracture surface well outside
the mirror:
R 1
∝ * (7.39)
c D
where R is either the mirror-mist, mist-hackle, or branching distance, and c is
the flaw size. In other words, the very processes that contribute to formation
of roughness at the mirror boundary are related to those that create roughness
elsewhere in areas remote from the origin.
At the beginning of this section it was noted that fractal analysis is a tool to
characterize fracture surfaces. Most fractal analyses have focused on using the
fractal dimension as a materials science tool to study the microstructure or
material properties. Additional work showing practical applications is needed.
7-67
Fractography of Ceramics and Glasses
7.14.1 Introduction
For example, residual stresses from grinding vary with direction, whether
parallel to or perpendicular to the grinding direction. They are strongly
compressive right at the surface, but rapidly decrease and become tensile a
short distance (5 µm to 20 µm) below the surface. A grinding crack tip may be
in tension or in compression depending upon how deep it is. The net effect of
the stress gradient on the flaw as a whole is a complex fracture mechanics
problem, but analytical attempts have been made to determine the effect on
flaws with mixed success. The reader is referred to the work of Holstein et
al. 108 for a good example. It is often difficult to determine whether the net
effect is tensile or compressive. The weak point in these analyses is the lack of
adequate quantitative information about the residual stress spatial profile.
The usual approach in dealing with residual stress problems is to rely on the
principle of superposition. That is, the effect of stresses on a location or flaw in
a body is additive, whether the stresses are mechanical, transient thermal, or
residual. To a first approximation, residual stresses, σr, simply add to or
subtract from the applied stresses acting on the flaw from known external
sources, σa, so that the net stress acting on a site or a flaw is:
σ net = σ a + σ r (7.40)
7-68
Quantitative Analysis
reduce the stresses or forces necessary to cause fracture from a particular flaw.
Compressive residual stresses acting on a flaw must be overcome and require
greater applied stresses. (The usual convention is that compressive stresses are
negative.) For the case of a flaw loaded with a far-field stress σa with a
residual stress σr that is constant in the vicinity of the flaw:
K Ic = Y σ net a (7.41)
Rearranging:
K Ic (7.42)
σr = −σa
Y a
If a convenient flaw is found that has a shape that is conducive to a fracture
mechanics analyses and KIc, σa, and the flaw size, a, are known, then equation
7.42 can be used to estimate σr.
Residual stresses can cause systematic deviations in the stress - mirror size
trends presented previously. Trend deviations are shown in Figure 7.38. The
easiest interpretation is in Figure 7.38a, whereby a nonzero intercept indicates
the magnitude and sign of the residual stress. A compressive residual stress
shifts the line upward since a greater applied stress, σa, is necessary to cause
fracture. If a single mirror is measured and the externally applied stress, σa, is
(a) (b)
σr
log slope < - ½
σa A σa -½
σr
σ = A / √1
σr
1 R=1
log R log R = 0
R
Figure 7.38 Systematic deviations in the fracture mirror size trends for
applied stress, σa, as a function of mirror radius are indicative of residual
stresses. (a) shows that residual stresses create nonzero intercepts. (b) shows
that residual stresses alter the slope.
7-69
Fractography of Ceramics and Glasses
known, then:
(σ a + σ r ) R=A (7.43)
and A
σr = −σa (7.44)
R
A good example of the application of this approach of measuring mirror-mist
and mist—hackle boundaries on glass rods with cladding is in a paper by
Mecholsky and Drexhage. 109 Similarly, the branching distances were used by
Conway and Mecholsky 110 to estimate residual stresses in tempered glass
plates. Additional details by this method are in Appendix D.
7-70
Quantitative Analysis
Ceramic and glass strengths depend upon the size and shape of the specimens
and the mode of loading. Specimen strengths vary due to variations in the size,
severity, location, and density of flaws. Strength variability is usually
analyzed in accordance with the Weibull distribution, 113 which is based on the
premise that the weakest link in a body controls strength. A strength test
produces two key bits of information: a strength datum and a fracture origin
flaw. The flaws are just as important as the strength. “Flaw management” is
an important strategy for process control and reliability improvements.
7-71
Fractography of Ceramics and Glasses
where Pf is the probability of fracture, σa is the applied stress, and m and σθ are
constants called the Weibull modulus and the characteristic strength of the
specimen, respectively. Strength data is typically ranked from weakest to
strongest and then plotted on a special set of axes intended to linearize the data
as shown in Figures 7.39 and 7.40.
With Weibull analyses, the maximum nominal stress in the body is used.
Stress is not adjusted for location. If the data fits the Weibull distribution, then
it should fit on a line with the slope equal to the Weibull modulus, m, and the
characteristic strength σθ corresponding to the strength of the specific test
specimen configuration at the 63.2 % level. Figure 7.39 and 7.40 show two
actual data sets with 100% fractographic analysis. The goodness of fit to the
Weibull function can be assessed by eyeball by comparing the fitted line to the
data. Both data graphs are fractographically-labeled and have not only the
individual strength values, but also the flaw identities.
Figure 7.39 for the alumina with silicon carbide whisker reinforcements shows
that the strength-controlling flaws were almost always large alumina grains.
The whiskers that were added to enhance KIc had chemistries or impurities that
caused grain growth in the matrix. The good news was that one flaw type
primarily controlled strength and the Weibull distribution fit quite well.
In contrast, Figure 7.40 shows a SiAlON material that had many different
concurrent flaw types and a very low Weibull modulus. 114 Brittle materials
design with such a material would be problematic, since each flaw type has its
own distribution. Strength scaling with size would be complicated. The tester
might have wondered why the curve had so many wiggles. The fractographic
analysis makes it clear why the single Weibull line does not fit very well.
7-72
Quantitative Analysis
99.9
99 Sialon
95 ASTM C1322
-96
90
Stress Identity
80
Probability of Failure (%)
699 ?
σθ = 602 688 Machining
63.2
687 Inclusion
675 ?
672 Porous Region
50 664 Handling
641 Pore
mu = 7.9 636 Inclusion
618 Pore
614 Agglomerate
602 Porous Seam
20 595
570
Handling
Inclusion
569 Pore
554 ?
10 545
545
Inclusion
Pore
540 Agglomerate
527 Agglom./Pore
5
519 Porous Region
518 Porous Seam
517 Pore(linear)
513 Agglom./Pore
506 Aggolm./Pore
505 Handling
2 483
482
Handling
Porous Seam
466 Pore
457 Agglom./Pore
1 450 Pore
that flaw size distributions could have the large flaw size tails matched by an
inverse power law distribution as shown in Figure 7.41. Using this inverse
power law distribution and applying classical fracture mechanics (equation
7.7), and weakest-link theory, they were able to mathematically derive the
Weibull distribution. The Weibull modulus is directly related the flaw size
distribution exponent. The Weibull distribution is now based on a solid
theoretical footing. Danzer has considered more general distributions, but
concluded that if the frequency of flaws follows the inverse power law, then the
material may be deemed a “Weibull material.117 In the 1990s and 2000s, a
number of studies confirmed the large flaw size distribution tails can indeed be
matched by the inverse power law. The reader is referred to papers by Prof.
Uematsu in the Origins in Ceramics section of Appendix A for many examples.
Quinn and Morrell’s paper 118 has a useful review of a number of successful
applications using Weibull analysis and it lists all of the assumptions necessary
for Weibull strength scaling with size.
7-73
Fractography of Ceramics and Glasses
(a)
(b) m = 2n’ -2
Figure 7.41 The Weibull distribution can be derived from the flaw size
distribution shown in (a). H’(c) is the probability for occurrence of a flaw of
size c. n’ and α are constants. The largest flaws control strength and that
portion of the distribution can be modeled by an inverse power law with
exponent n’. (b) shows the Weibull distribution for strength. The Weibull
modulus m is equal to 2n’ – 2.
Average bulk properties such as density, surface finish, or grain size often do
not correlate with strength. Strength often depends upon infrequent or aberrant
microstructural features, namely flaws, and not on the average microstructure.
In some cases, the aberrant features do scale with the average microstructure,
so there may be a correlation between strength and average bulk microstructure
or surface finish. One important aspect of the Weibull model is that the larger
the structure or test piece is, the more apt it is to contain a severe flaw. Hence,
strength usually varies inversely with component size: the larger the comp-
onent, the weaker it is likely to be. Strengths can be correlated or scaled by
7-74
Quantitative Analysis
Strength scaling analyses become very complex if more than one flaw type
(e.g., pores and inclusions; or pores, inclusions, and machining damage)
controls strength such as shown in Figure 7.40. Various flaw types may
suddenly appear or disappear as strength controlling flaws as component sizes
are scaled up or down. See the Quinn and Morrell review115 for more
information on these fascinating topics. Sung and Nicholson published results
on a good study 119 where fractography was effectively used to identify five
different flaw types in a 4 mass % yttria-stabilized, tetragonal polycrystal
material: fiber burn-out pores, agglomerates, iron inclusions, alumina
inclusions, and pores. By careful process control and “flaw management,” they
were able to eliminate two flaw types, and control the remaining three. Small
pores, small alumina inclusions, and small agglomerates became the primary
strength-controlling flaws.
The author is often asked whether the Weibull distribution is the best for
modeling strength variations, and my answer is always yes, but with some
caveats. The Weibull distribution is the only one that the author knows of that
permits size scaling of strength. Indeed, the ultimate proof that the Weibull
distribution is valid is not whether one set of data can be plotted linearly on a
Weibull graph, but whether size scaling is successful. In the dental materials
community, confusion about the validity of the Weibull model prompted Dr. J.
Quinn and the author to write a review article 120 on this topic. Everything in
the review is applicable to other ceramics and glasses as well.
7-75
Fractography of Ceramics and Glasses
suitable. It was better than the log normal, the normal, or the gamma
distributions. The author congratulated them on their study and offered to do
some fractography. Imagine the challenge! It was learned that they broke
even more and had reached 11,700 tests. The Weibull distribution still held
true. Regrettably, they did not save any specimens for the author to look at.
What a pity!
Chapter 7 References
7-76
Quantitative Analysis
7-77
Fractography of Ceramics and Glasses
39. SRM’s 2100, 2100A, and 2100B, “Fracture Toughness of Ceramics,” National
Institute for Standards and Technology, Gaithersburg, MD, 1999.
40. J. C. Newman and I. S. Raju, “An Empirical Stress Intensity Factor Equation for
the Surface Crack,” Eng. Fract. Mech., 15 [1-2] (1981) 185-192.
41. ASTM C 1322-96, “Standard Practice for Fractography and Characterization of
Fracture Origins in Advanced Ceramics,” Annual Book of Standards, Vol. 15.01,
ASTM, West Conshohocken, PA, 1996.
42. ASTM C1421-99, “Standard Test Method for the Determination of Fracture
Toughness of Advanced Ceramics,” ibid, 2000.
43. ASTM E 740-88, “Standard Practice for Fracture Testing with Surface-Crack
Tension Specimens,” Annual Book of Standards, Vol. 3.01, ASTM, West
Conshohocken, PA, 1988.
44. ISO 18756, Fine Ceramics (Advanced Ceramics, Advanced Technical Ceramics) –
Determination of Fracture Toughness of Monolithic Ceramics at Room Temperature
by the Surface Crack in Flexure (SCF) Method, International Organization for
Standards, Geneva, SW, 2003.
45. T. Fett, “An Extension of the Newman-Raju Formula,” Int. J. Fract., 33 (1987)
R47-R50.
46. T. Fett, “Stress Intensity Factors for Semi-elliptical Surface Cracks in a Plate
Under Tension Based on the Isida’s solution,” ibid, 48 (1991) 139 – 151.
47. T. Fett and D. Munz, Stress Intensity Factors and Weight Functions,
Computational Mechanics Publ., Southampton, UK and Boston, MA, USA, 1997.
48. M. Isida, H. Noguchi, and T. Yoshida, “Tension and Bending of Finite Thickness
Plates with a Semi-elliptical Surface Crack,” Int. J. Fract., 26 (1984) 157-188.
49. S. Strobl, P. Supancic, T. Lube, and R. Danzer,” Surface Crack in Tension or
Bending – A Reassessment of the Newman and Raju Formula in Respect to
Fracture Toughness Measurements in Brittle Materials,” J. Eur. Cer. Soc., 32
(2012) 1491 – 1501.
50. G. K. Bansal, “Effect of Flaw Shape on Strength of Ceramics,” J. Am. Ceram.
Soc., 59 [1-2] (1976) 87-88.
51. Y. Murakami, Stress Intensity Factors Handbook, Vols. 1 and 2, ed., Pergamon
Press, NY, 1986. (Case 9.49)
52. J. J. Mecholsky, Jr., S. W. Freiman, and R. W. Rice, “Fracture Surface Analysis
of Ceramics,” J. Mat. Sci., 11 (1976) 1310–1319.
53. G. K. Bansal and W. H. Duckworth, “Fracture Stress as Related to Origin and
Fracture Mirror Sizes,” J. Am. Ceram. Soc., 60 [7–8] (1977) 304 – 310.
54. G. K. Bansal, “On Fracture Mirror Formation in Glass and Polycrystalline
Ceramics,” Phil Mag., 35 [4] (1977) 935- 944.
55. S. R. Choi and J. P. Gyekenyesi, “Crack Branching and Fracture Mirror Data of
Glasses and Advanced Ceramics,” NASA Technical Report TM 1998-206536,
NASA Glenn Research Center, Cleveland, OH,1998.
56. H. P. Kirchner, R. M. Gruver, M. V. Swain, and R. C. Garvie, “Crack Branching
in Transformation-Toughened Zirconia,” J. Am. Ceram. Soc., 64 [9] (1981) 529-
533.
7-78
Quantitative Analysis
7-79
Fractography of Ceramics and Glasses
7-80
Quantitative Analysis
89. G. D. Quinn, R. J. Gettings, and J. J. Kübler, “Fractography and the Surface Crack in
Flexure (SCF) Method for Evaluating Fracture Toughness of Ceramics,” pp. 107 -
144 in Fractography of Glasses and Ceramics III, Ceramic Transactions, Vol. 64,
eds., J. R. Varner, V. D. Fréchette, and G. D. Quinn, ACS, Westerville, OH, 1996.
90. G. Quinn, R. Gettings, and J. Kübler, “Fracture Toughness by the Surface Crack in
Flexure (SCF) Method: Results of the VAMAS Round Robin,” Cer. Eng. Sci. Proc.,
15 [5] (1994) 846-855.
91. G. D. Quinn, K. Xu, J. A, Salem, and J. J. Swab, “SRM 2100: the World’s First
Fracture Toughness Reference Material,” pp. 499 – 530 in Fracture Mechanics of
Glasses and Ceramics, Vol. 14, eds., R. C. Bradt, D. Munz, M. Sakai, and K. W.
White, Klewer/Plenum, NY, 2005.
92. I. Bar-On, G. D. Quinn, J. Salem, and M. J. Jenkins, “Fracture Toughness Standard
Test Method C 1421-99 for Advanced Ceramics,” pp. 315-335 in Fatigue and
Fracture Mechanics, Vol. 32, ASTM STP 1406, ed. R. Chona, ASTM Int., West
Conshohocken, PA, 2001.
93. EN ISO 18756:2005, “Advanced Ceramics, Advanced Technical Ceramics);
Determination of Fracture Toughness of Monolithic Ceramics at Room Temperature
by the Surface Crack in Flexure (SCF) Method,” European Committee for
Standards, Brussels.
94. G. D. Quinn, “Fracture Toughness of Ceramics by the Vickers Indentation Crack
Length Method: A Critical Review,” Cer. Eng. Sci. Proc., 27 [3] (2006) 45 – 62.
95. G. D. Quinn and R. C. Bradt, “On the Vickers Indentation Fracture Toughness
Test,” J. Am. Ceram. Soc., 90 [3] (2007) 673 – 680
96. B. R. Lawn and D. B. Marshall, “Indentation Fractography,” pp. 1 –36 in
Fractography of Glass, eds., R. Bradt and R. Tressler, Plenum, NY, 1994.
97. P. Chantikul, G. R. Anstis, B. R. Lawn and D. B. Marshall, “A Critical Evaluation
of Indentation Techniques for Measuring Fracture Toughness: II Strength
Method,” J. Am. Ceram. Soc., 64 [9] (1981) 539 – 543.
98. R. F. Cook and G. M. Pharr, “Direct Observation and Analysis of Indentation
Cracking in Glasses and Ceramics,” ibid, 73 [4] (1990) 787 – 817.
99. T. Wilantewicz, “Crack Initiation Behavior of Optical Glasses from Vickers
Indentations,” Ph. D. Thesis, Alfred University, November, 2005.
100. T. Wilantewicz and J. R. Varner, “Vickers Indentation Fracture in Optical Glass
Compositions,” pp. 131 -151 in Fractography of Glasses and Ceramics, Vol. 5,
eds. J. R. Varner, G. D. Quinn, M. Wightman, American Ceramic Society,
Westerville, OH, 2007.
101. D. B. Marshall, B. R. Lawn, and J. J. Mecholsky, “Effect of Residual Stresses on
Mirror/Flaw Size Relations,” J. Am. Ceram. Soc, 63 [5-6] (1980) 358 – 360.
102. D. E. Passoja, “Fundamental Relationships Between Energy and Geometry in
Fracture,” pp. 101 – 126 in Fractography of Glasses and Ceramics, eds., J. R.
Varner and V. D. Fréchette, Advances in Ceramics, Vol. 22, American Ceramic
Society, Westerville, OH, 1988.
103. J. J. Mecholsky, T. J. Mackin, and D. E. Passoja, “Self-Similar Crack Propagation
in Brittle Materials,” ibid, pp. 127 – 134.
7-81
Fractography of Ceramics and Glasses
7-82
Single Crystals
8. Single Crystals
8.1 General
Fractographic procedures for single crystals are similar to those for glasses and
polycrystalline ceramics, but there are some nuances and differences. Fracture
surfaces may have regions with conchoidal fracture, Wallner lines, and twist
hackle. Other regions may have dramatic faceting. Fracture resistance often
varies significantly between the different crystallographic planes, causing
jagged fracture patterns as cracks radically change directions onto preferred
cleavage planes. Fracture surfaces may be very difficult to interpret. Some
examples of the unusual markings are shown in Figure 8.1.
(a) (b
(c) (d)
Figure 8.1 Examples of single crystal fractures. (a) and (b) show fracture
mirrors on the {100} type planes in MgO, but with two different tensile surface
orientations. (c) and (d) show a cubic zirconia bend bar with curved linear
lance like markings emanating from the origin area and the grinding crack
origin. (courtesy R. Rice)
8-1
Fractography of Ceramics and Glasses
The overall fracture plane may not necessarily be perpendicular to the direction
of principal normal stress, since a preferred cleavage plane may be at an
irregular angle to that stress. Sometimes the overall fracture extension may be
approximately perpendicular to the maximum principal stresses, but local
fracture proceeds in a zigzag fashion on other planes. This crack redirection is
different than conventional crack branching in glasses or polycrystalline
ceramics. Branching is much less common in single crystals. If the maximum
principal stress is perpendicular to a preferred cleavage plane, the fracture
surface may be very flat and featureless with little or no hackle, and no
branching at all.
Fracture mirrors may be present, but may have odd shapes and markings.
There may be no mist or hackle on preferred cleavage planes. Mirror sizes
may be extremely difficult or impractical to measure. Cleavage hackle steps
may be confused with Wallner lines. Twinning may cause fracture.
Bradt et al. 1,2 have reviewed the various criteria for cleavage and concluded
that fracture toughness is the best indicator of the preferred cleavage plane.
The fracture toughness can vary dramatically on different planes. The energy
to propagate cracks along preferred cleavage planes is less than that for less-
favored planes. Nevertheless, there is contradictory information in the
literature on preferred cleavage planes in some crystals, even for such well
studied materials as silicon, alumina (sapphire), and magnesium aluminate
spinel. Some of contradictory information surely is due to experimental error
or specific aspects of the test method. The direction of crack advance on a
plane is as important as the plane itself. The speed of crack advance can even
determine which fracture plane is preferred. 3 Although the double cantilever
beam method and the constant applied moment variation as shown in Figure
8.2 are often used, results are not strictly valid if the crack zigzags down the
guiding groove or is tilted to the intended fracture plane.
8-2
Single Crystals
the propagation plane. Bradt et al. 6,7,8 have used the SCF method on alumina
and spinel. Figure 8.3 shows a Knoop precrack in sapphire. One unresolved
issue with the SCF method for single crystals is whether or not indentation
residual stresses vary with plane and orientation. Akimune and Bradt 9 showed
that there is considerable hardness anisotropy even on one plane (100) in
spinel, but negligible variability on the (111).
Double Cantilever
Beam
Figure 8.2 Test methods for single crystal fracture toughness determination.
Plane KIc
Reference Fracture characteristics
Type (MPa√m)
Chen and Flat and smooth fractures.
{111} 0.82
Leipold (4) This is the preferred cleavage plane.
Large but asymmetric cathedral mirrors
Chen and centered on the flaws.
{110} 0.90
Leipold (4) Propagation switched to {111} planes
and zigzagged at the mirror boundaries.
Chen and A very rough fracture surface.
{100} 0.95
Leipold (4) Cracks switched to {111} very quickly.
Xin, Hsia, and Residual indentation stresses intact:
{110} 0.95
Lange (5) Cracks propagated on {110}.
Xin, Hsia, and Residual indentation stresses removed
Lange(5) {110} -
by annealing. Crack switched to {111}.
Table 8.1 Fracture toughness of single crystal silicon from Knoop SCF tests.
8-3
Fractography of Ceramics and Glasses
(a)
(b)
Figure 8.3 Relatively smooth fracture surface (a) of a surface crack in flexure
(SCF) sapphire test specimen with a 49 N Knoop indentation as a starter flaw
(large arrow). The bar tensile surface was the a-plane. Fracture occurred on
the m-plane. (b) is a close-up which shows grinding cracks along the surface
(small white arrows), but the indentation crack (large white arrow) was
dominant. The long curved line leading from the origin to the upper right is a
“cleavage step hackle” line. (courtesy R. Krause).
8-4
Single Crystals
Confocal microscopes may not be very effective due to the specular reflections.
Gold coating of a transparent single crystal may aid in interpretation and may
make the fracture surface easier to view, but if the surface is very flat, the gold
coating may make it behave like a mirror, which will pose problems for
inspection and photography.
8-5
Fractography of Ceramics and Glasses
(a)
(b)
(c)
8-6
Single Crystals
Fracture surface
Polished surface
(a) (b)
(c)
8-7
Fractography of Ceramics and Glasses
Figure 8.6 Schematics showing (a) a Wallner line and twist hackle in a glass
plate loaded in bending and (b) “cleavage step hackle” in a single crystal plate
loaded in bending. In both cases the crack was leading on the bottom surface.
Single crystal fracture planes often have small “cleavage step hackle” that may
be confused with Wallner lines. Figure 8.6 shows the differences.
Cleavage step hackle: A form of hackle in single crystals caused by the crack
running on two very close parallel planes. It is in the form of one or more
single lines, parallel to the local direction of crack propagation.
Fracture surfaces and fracture mirrors may have extraordinary shapes and
facets, but usually have some degree of recognizable symmetry that will help
the fractographer identify an origin site. Mirror shapes may vary dramatically
within a crystal depending upon the fracture plane. To illustrate these
fascinating variations, fracture markings for single crystal cubic spinel are
shown in Figures 8.7 through 8.14. These figures follow the recommendations
of this Guide and ASTM C 1322. 11 They show a picture of the whole fracture
8-8
Single Crystals
surface, the fracture mirror, and a close up of the origin. The fractographic
findings are summarized in Table 8.2. These images are from Rice and
colleagues3,12,13,14 who studied fracture in this material using double cantilever
beam and flexural strength specimens. They found a remarkable variety of
fracture surfaces with features such as “cathedral mirrors,” “gull wing mirrors,”
“skewed cathedral mirrors,” and “whisker lances.” Cathedral and gull wing
mirrors have also been observed in other cubic crystals such as silicon (Figure
8.4) and fully-stabilized zirconia. Gull wing mirrors look like cathedral mirrors
that are tilted 90o on the fracture surface. Bear in mind that there may be
threshold stresses or stress intensities for the formation of markings on certain
planes. Thus, the flat fractures shown in figures 8.8 and 8.9 may indicate
fracture on a preferred cleavage plane, or that the stresses were too low to
generate mirror markings. The preferred cleavage plane in spinel is the {100}
which has a fracture toughness of 0.9 to 1.2 MPa√m depending upon the
method of measurement.3,6,7,12,13,14 The fracture toughness of the {110} is not
much greater at 1.0 to 1.5 MPa√m. Consequently, fracture usually runs on
{100} or {110} type planes. Fracture is not preferred on the {111} and the
fracture toughness is much greater (1.9 MPa√m or more). 3,6,7,12,13,14
Table 8.2 Fracture markings in MgAl2O4 spinel bend bars. The brackets {100}
denote the family of (100) planes and the angles <100> denote the family of
[100] directions. (After Wu, McKinney, and Rice, Ref. 3)
8-9
Fractography of Ceramics and Glasses
(a)
Bend bar orientation
(b)
(c)
Figure 8.7 Spinel bend bar with a {100} tensile surface and <100> tensile
axis designed to cause fracture on a {100} plane. (185 MPa) (a) shows the
whole fracture surface and origin (arrow). (b) shows the “cathedral mirror”
and the arc-ribs that gradually get progressively severer until a branch or
major crack redirection occurs (white arrow). (c) shows the grinding crack
fracture origin. Fracture starts on the {100} but shifts or branches to the {110}
just beyond the sides of the mirror. (Case A, Table 8.2) (Courtesy R. Rice)
8-10
Single Crystals
(a)
(b)
(c)
Figure 8.8 Spinel bend bar with a {110} tensile surface and <100> tensile
axis designed to cause fracture on the {100} cleavage plane. (170 MPa) (a)
shows the whole fracture surface which is flat except for the cleavage steps on
the side and the cantilever curl. Other similar specimens had no cleavage
steps. (b) shows the origin site and (c) shows the grinding crack fracture
origin. (Case D, Table 8.2) (Courtesy R. Rice)
8-11
Fractography of Ceramics and Glasses
(a)
(b)
Figure 8.9 Spinel bend bar with a {100} tensile surface and <110> tensile
axis designed to cause fracture on a {110} type plane. (169 MPa) (a) shows
the whole fracture surface which is flat and featureless except for the cantilever
curl. The arrow shows the origin. (b) shows the grinding crack fracture
origin. Fracture stays on a {110} type plane. (Case B, table 8.2) (Courtesy R.
Rice)
8-12
Single Crystals
(a)
(b)
(c)
Figure 8.10 Spinel bend bar with a {110} tensile surface and <110> tensile
axis designed to cause fracture on a {110} plane. (283 MPa) (a) shows the
whole fracture surface, (b) shows the mirror with “whisker lances” radiating
outward, and (c) shows the grinding crack fracture origin. (Case F1, table 8.2)
(Courtesy R. Rice)
8-13
Fractography of Ceramics and Glasses
(a)
(b)
(c)
Figure 8.11 A second example of a spinel bend bar with a {110} tensile
surface and <110> tensile axis. (172 MPa) The fracture surface (a) is very
rough as in the previous figure. The mirror (b) is somewhat more symmetric
than in the previous figure. This mirror was shown in an earlier publication
(Figure 29 of Ref. 13), but with a typographical mistake identifying it as
{110}<100>. (Case F2, table 8.2). (Courtesy R. Rice)
8-14
Single Crystals
(a)
(b)
(c)
Figure 8.12 Spinel bend bar with a {110} tensile surface and <110> tensile
axis designed to cause fracture on a {110} plane. (361 MPa) (a) shows the
whole fracture surface, (b) shows the “gull wing mirror,” and (c) shows the
grinding crack fracture origin. (Case C, table 8.2) (Courtesy R. Rice)
8-15
Fractography of Ceramics and Glasses
(a)
(b)
(c)
Figure 8.13 Spinel bend bar with a {111} tensile surface and <110> tensile
axis designed to cause fracture on a {110} plane. (unknown stress) (a) shows
the whole fracture surface and the major crack redirection on the mirror sides.
(b) shows the “skewed cathedral mirror,” and (c) shows the grinding crack
fracture origin from the matching fracture piece. (Case G1, table 8.2)
(Courtesy R. Rice)
8-16
Single Crystals
(a)
(b)
(c)
Figure 8.14 Another example of a spinel bend bar with a {111} tensile surface
and <110> tensile axis designed to cause fracture on a {110} plane. (261
MPa) (a) shows the whole fracture surface, (b) shows a “skewed mirror”
with “whisker lances” and (c) shows the grinding crack fracture origin (black
arrows) which is at a slight angle to the tensile stress axis and the fracture
surface). (Case G2, table 8.2) (Courtesy R. Rice)
8-17
Fractography of Ceramics and Glasses
(a)
(b)
(c)
Figure 8.15 Spinel bend bar with a {110} tensile surface and <111> tensile
axis designed to cause fracture on a {111} plane. (325 MPa) (a) shows the
whole fracture surface and the major crack redirection on the top half, (b)
shows the “gull wing mirror,” and (c) shows the grinding crack fracture
origin. (Case E, table 8.2) (Courtesy R. Rice)
8-18
Single Crystals
Another distinctive shape seen in some crystals are long hackle lines that
extend vertically from the sides of an origin suggesting a “batman mirror”
shape as shown in Figure 8.16.
(a)
(b)
Figure 8.16 “Batman mirror” in a sapphire bar. (a) shows matching halves
of the bend bar with a tensile surface parallel to the a-plane and bend bar sides
oriented so as to cause fracture on the r-plane. The tensile direction was the r
axis. The origin was grinding damage. (Specimen courtesy R. Krause.)
8-19
Fractography of Ceramics and Glasses
Scarps can occur in single crystals and are indicative of the presence of liquid
during fracture. Figure 5.48b shows an example of scarps in sapphire.
Fracture surface
Tensile
surface
10 µm
(a) (b)
Figure 8.17 Fracture origin of the silicon wafer shown in Figure 8.4. (282
MPa) (a) shows the origin is a grinding crack that extends only 2 µm below
the tensile surface and is associated with a grinding striation (groove) that
seems to be deeper and has caused greater cracking than other nearby
striations. The grinding direction is slightly different than the final plane of
fracture, which caused the fracture surface to have a jog at the origin as shown
in the optical view of the tensile surface (b). The fractographer should be alert
to subtle features like the jog, since they are important clues.
8-20
Single Crystals
8.5 Origins
Single crystals are very sensitive to surface contact damage of any type. The
low fracture toughness and high stiffness means that tiny strength-limiting
cracks can easily pop into the crystal. There are no grain boundaries or other
microstructural features to impede or redirect surface initiated cracks.
Grinding or polishing flaws are common fracture origins. Flexure and tensile
specimens are very sensitive to edge flaws and special care must be taken with
the edge beveling or rounding procedures. Figure 8.17 shows the grinding
crack origin of the silicon wafer shown earlier in Figure 8.4.
(a) (b)
Fracture surface
(c)
Polished surface
Figure 8.18 Polishing or grinding crack origin in a sapphire dome. (a) shows
the entire fractured structure. The origin is marked by the black arrow. (b)
shows an SEM photo of the piece with the origin marked by the arrow.
Fracture followed the axis of a polishing crack for a short distance, but then
quickly shifted to a different plane. (c) is a SEM close-up of the origin. The
piece has been tilted back to show both the fracture surface and the polished
surface. There is no trace of damage on the polished surface.
8-21
Fractography of Ceramics and Glasses
8-22
Single Crystals
← <110> →
• {100}
(a) (b)
{110} plane
(c) (d)
Figure 8.20 Tiny silicon theta test specimens. The ring shown in (a) is
compressed on its rim on the top and bottom in a nanoindenter causing the 7.5
µm tall by 100 µm thick web section in the middle to stretch in uniaxial tension.
(b) shows a fractured specimen on a strip. (c) shows that the web fractures
usually occurred on {111} planes which are the cleavage planes. A few
initiated on {110} planes but then changed to {111} planes. Origin flaws are
shown in the next figure.
8-23
Fractography of Ceramics and Glasses
(a) (b)
(c) (d)
(e) (f)
Figure 8.21 Fracture origins in single crystal silicon theta specimens. (b) is a
close-up of (a). The “cathedral mirrors” are very clear. The origins are etch
pits from the DRIE processing. The cusps between them are very sharp and
probably initiated cracks. Figure 6.69 shows other sharp cusp origins.
8-24
Single Crystals
8-25
Fractography of Ceramics and Glasses
(a) (b)
(c)
8-26
Single Crystals
Figures 8.24 and 8.25 show two fractured c-axis sapphire tubes with inter-
secting twins that created fracture origin cracks. The tubes were 38 mm in
diameter and 400 mm long and fractured during two runs of a prototype plasma
asher. This is a device used to dry-remove photoresist from large 300 mm
diameter silicon wafers during manufacture of integrated circuits. The tubes
contained oxygen and nitrogen gasses which were energized into a plasma state
by a 2.5 kW microwave chamber which was cycled on and off in 60 s intervals.
The sapphire tubes reached temperatures of 600 oC to 700 oC. One tube broke
during a gas purge step three seconds after the shutdown of the microwave
energy. The second tube broke two minutes after a shutdown. The tubes
should have had lifetimes in excess of 400,000 cycles The breakage during
cool down suggested thermal stresses had caused fracture. The cracks curved
and wiggled at the tube ends, indicating the cracks slowed down and arrested.
Some of the cracks crossed over each other, indicating that primary cracks did
not run all the way through the wall thickness. During optical microscopic
examination of one of the fracture surfaces, using common reflected light, a
chance reflection off one surface revealed thin parallel reflective lines that ran
counter to the fracture surface faceting as shown in Figure 8.24b. This was a
sign of twinning and the next step was to view the tubes between crossed
polarizers (as discussed in sections 3.8 and 3.25) with the results shown in
Figure 8.25. Multiple twins in the hot region triggered crack formation at twin
intersections. The twins were oriented at 32o to the sapphire tube axis
corresponding to classic r-plane twinning in sapphire, which can be caused by
rather low compression stresses at temperatures as low as 300 o to 500 oC.22
Fracture of these tubes occurred in two stages. First, axial constraint from a
new type of seal at the tube ends prevented the free expansion of the tubes
during the heating portion of the operating cycle. The earlier model tubes that
did not have breakages used a different seal design that allowed expansion.
The constraint in the new design created compression stresses. These created
the intersecting twins that generated cracks. The cracks grew in stages and
propagated by cool down thermal stresses until the tube ruptured. Although
initially there was some concern that the sapphire tubes were faulty, they were
fine. The problem was solved when the seal design was changed to allow for
expansion of the sapphire tubes.
8-27
Fractography of Ceramics and Glasses
(a)
Tube 1 Tube
twins
(b)
Cleavage
steps
8-28
Single Crystals
(a)
Tube 1
Tube 2
(b)
Figure 8.25 Twin bands became very apparent when the sapphire tubes were
illuminated by light from crossed polarizers. Cracks initiated at twin band
intersections (white arrows) and then were propagated by cool down thermal
stresses. (Photos courtesy R. Pingree)
8-29
Fractography of Ceramics and Glasses
Earlier in this chapter, it was shown in the spinel examples Figs. 8.8 and 8.9
that, if a preferred cleavage pane is perpendicular to the maximum principal
tensile stress, the fracture surface will be flat and smooth and at right angles to
the tensile stress. If the preferred cleavage plane is not perpendicular to the
maximum principal tensile stress, then the fracture plane may be at an inclined
angle to the tensile stress, such as in Figure 8.23 for sapphire, and the zig-zag
planes of Fig. 8.10, 8.11 and 8.13 for spinel. These illustrate exceptions to the
law of normal crack propagation discussed in section 4.3
In some extreme cases, the loading geometries and stresses may be such that
fracture is forced to occur on an unfavorable plane, such as the basal plane in a
sapphire tube shown in Figure 8.26. The fracture surface is extremely jagged
with many zig-zags. Other examples of basal plane fracture in sapphire are
shown in references 8, 20, and 26.
Good estimates of the fracture toughness KIc may be obtained from fracture
mechanics analyses of flaws. For example, the long uniform 2 µm deep surface
crack in the silicon wafer shown in Figure 8.17a combined with a breaking
stress of 282 MPa and a shape factor Y = 1.99 produced an estimate of KIc =
1.99 (282) √2 x 10-6 = 0.79 MPa√m in good agreement with values in table 8.1.
Rice showed several good examples13,21 and Newcomb and Tressler20 were able
to measure fracture toughness on the basal plane of sapphire by this approach,
when other methods usually fail due to the difficulty of controlling fracture on
this plane. Many groups have effectively used Knoop indentation surface crack
in flexure tests to study fracture behavior and measure fracture toughness of
single crystals since the small semi-elliptical flaws mimic the behavior of other
flaws.
8-30
Single Crystals
(a)
(b)
(c)
Figure 8.26 Example of basal plane fracture in a c-axis sapphire tube that
fractured from severe thermal shock (a). The origin is at the tube mid length
(large arrow) and is perpendicular to the tube axis. The thermal stresses acted
in the axial direction. (b) and (c) show the mirror and origin on the outer tube
surface.
8-31
Fractography of Ceramics and Glasses
Rice has written several outstanding, richly illustrated papers on single crystal
fracture and fractography. Reference 13 is a book length article with over
thirty figures showing single crystal fracture surfaces in MgO, MgAl2O4, Si,
SiC, Al2O3, ZrO2, CaF2, and W. Reference 14 has six pages of illustrations and
a discussion of fracture markings in various single crystal ceramics.
Chapter 8 References
8-32
Single Crystals
8-33
Ceramic and Glass Composites
The composite ideally should have a controlled microstructure such that the
reinforcing phase is uniformly distributed. This is routinely achieved with
glass ceramics, but is more problematic with fiber or particulate-reinforced
ceramics. Sometimes the reinforcing agent forms clumps or aggregates that
can act as strength-limiting flaws as shown in Figure 9.1. Flaws of this type
may be broadly described as compositional inhomogeneities using the logic in
chapter 6, but more specific descriptors are probably better (fiber clump, fiber
tangle, platelet cluster, etc.).
9-1
Fractography of Ceramics and Glasses
Second phase aggregates can also inhibit local sintering causing pockets of
porosity that also act as strength limiting flaws. Self-reinforced materials that
rely on interconnected elongated-grain microstructures may develop
excessively large grains that control strength as shown in Figures 6.22c and d.
Very often reinforcing agents such as SiC whiskers can create local chemistry
imbalances in the matrix leading to exaggerated grain growth. SiC whisker-
reinforced alumina for cutting tools is an example. These large grains then can
act as flaws. Although few would argue against reinforcing ceramics to
enhance fracture toughness, this approach sometimes comes with a price. The
very agents used to reinforce the material can create flaws that limit the
material’s strength.
9-2
Ceramic and Glass Composites
Figure 9.2 A CVD SiC (SCS-6) filament-reinforced Si3N4 bend bar tilted back
at an angle so as to view the fracture and tensile surfaces. A series of
branching cracks radiate outwards from one origin site (arrow).
(a) (b)
Figure 9.3 Fracture surface of CVD SiC filament reinforced Si3N4 bend bar.
(Ref. 1) (a) shows that there was not very much fiber pullout in this case.
Wake hackle in the matrix behind the filaments are telltale indicators of the
local direction of crack propagation through the matrix and lead back to the
origin, which is a porous seam in the matrix shown by the white arrows in (a)
and (b).
9-3
Fractography of Ceramics and Glasses
reinforced and the size and shape of the component and the loading conditions.
Although bend testing is reliable for creating tensile fractures in monolithic
ceramics and glasses, composites are more susceptible to alternative failure
modes when loaded in bending. Figures 9.4 and 9.5 shows tensile stress
fractures with good pull-out in a fiber-reinforced epoxy and a ceramic,
respectively.
(a)
(b)
Figure 9.4 Fracture surface of an E-glass fiber reinforced epoxy matrix one-
dimensional composite. The arrow marks the origin in (b). (courtesy of S.
Scherrer)
9-4
Ceramic and Glass Composites
Figure 9.5 SEM image of the fracture surface of a SiC (Nicalon) ceramic
fiber – zirconia matrix composite. The arrows mark fracture origins and
mirrors in several fibers. (Ref. 2, courtesy of R. Rice)
9-5
Fractography of Ceramics and Glasses
The usual desired mode of fracture is one with a very rough fracture surface
with ample fiber pullout. Observation that portions of a composite structure
have fractured in a planar fashion with minimal pull out is often a sign that
some deterioration in the composite has occurred or that the fiber-matrix bond
is too strong. For example, Glass et al. 9 analyzed a spectacular rupture of a
large fiberglass–epoxy tank that stored pressurized sulphuric acid. Large
portions of the tank had the classic rough fracture fractures with extensive fiber
pullout, but the origin region had flat planar fracture zones. The cause was
traced to spilled acid on the exterior surface that penetrated matrix microcracks
and reached and degraded the reinforcing E-glass fibers. Fracture mirror
analysis showed the fibers broke at only 20 % to 40 % of the baseline fiber
strengths. Progressive fracture from the exterior occurred over a period of time
as outer fibers broke, matrix microcracks opened up, and more acid penetrated
and damaged more fibers until the vessel ruptured.
9-6
Ceramic and Glass Composites
They are cured in situ with a “blue” ultraviolet light to polymerize the material.
There are some similarities, but many important differences in the fractography
of dental composites versus glass ceramics and ceramics. The lower elastic
modulus of resin composites and often lower strengths mean they have much
less stored elastic energy when fracture occurs. Thus, there will be less
fragmentation and branching. Fracture surfaces will have some of the same
markings described in this book, but sometimes are described in the polymer
literature with different nomenclature. The reader is referred to Anne Roulin-
Moloney’s excellent book 14 on the fractography of polymers and composites
for more information about resin-matrix composites in general. Much of the
fractography is aimed at observing filler-matrix interfaces and what cracks do
at the interfaces as opposed to finding origin flaws.
Fracture mirrors may exist, and are described as smooth areas on the fracture
surface at the origin. With highly-filled composites, the interpretation is the
same as described in this book: the mirror is a smooth zone surrounding the
fracture origin flaw. By the time the rough hackle lines form, the crack has
reached terminal velocities (e.g., 800 m/sec) that are much slower than in
glasses and ceramics. Figure 9.6 shows fracture surfaces of one such highly-
filled resin matrix composite from Reference15. It has a mean strength of 145
MPa, a Weibull modulus of about 10, and a fracture resistance of 0.9 MPa√m.
to 1.3 MPa√m indicative of a material with a slight R-curve. These materials
have very small glass filler particles and are translucent. Inclusions and
agglomerates are easy to detect in uncoated bend bars, either with reflected
illumination or sometimes more effectively with transmitted illumination.
Once the specimens have been examined in the natural state, they can be gold
coated. The fracture surfaces will be much easier to interpret.
9-7
Fractography of Ceramics and Glasses
(a) (b)
(c) (d)
(e) (f)
Figure 9.6 Filled resin-matrix composite fractography. (a) shows the fracture
mirror in an uncoated Paradigm MZ 100 bend bar, with transmitted light from
under the bar, (b) shows the same view, but with reflected light after gold
coating. (c) - (e) show SEM images of the mirror and the flaw, a calcium-rich
inclusion. (f) shows the original blanks and a machined bend bar. (a-e
courtesy J. Quinn)
9-8
Ceramic and Glass Composites
On the other hand, mirrors in unfilled resins14,16 might look similar, but they are
in fact a region of slow crack extension that occurs prior to the crack going
critical. The “mirror” is in fact the critical crack. So one must be careful in
interpreting “fracture mirror constants” in the polymers field. They may not be
the same as those in the ceramic and glass field.
Chapter 9 References
9-9
Fractography of Ceramics and Glasses
9-10
Case Studies
for failure analysis. Many additional cases are tabulated in Appendix B and
Case 3 Conflicting Carbide Data (Silicon carbide flaws and slow crack
growth)
Case 4 Vulnerable Vials (Broken medicinal vials)
Case 5 Damaged Dome (Coarse-grained aluminum oxynitride IR dome)
Case 6 Suffering Setter Plate (Silicon carbide furnace plate)
Case 7 Ruptured Radomes (Fused silica missile nosecones)
Case 8 Maligned machinists (Bend bars made by different machine shops)
Case 9 Modeler’s Match (Fracture origins in MEMS scale micro tensile
specimens)
Case 10 Fractious Fractographers (A VAMAS fractography round robin)
Case 11 Perilous Prostheses (Four ceramic dental crowns)
Case 12 Mangled Margins (Twenty alumina dental crowns )
Case 13 Ruined Refractors (Single crystal sapphire solar lenses)
Figure 10.1 shows a model gas turbine rotor made of hot-pressed silicon nitride
that was designed to fail. It was one of twelve such model structures made by
the Ford Motor Company’s scientific research laboratory in the late 1970s to
mid-1980s. 1,2 It was part of an ambitious endeavor to incorporate advanced
ceramics into automotive gas turbine engines. Considerable effort was
expended on improving ceramic materials, developing reliability codes,
generating data bases, fabricating parts, and running them in test rigs. Full
scale engine testing was expensive and risky, so a small project was set up to
verify the ceramic design and reliability codes by using a realistic model rotor.
This simulated rotor was mounted in a hot-spin rig shown in Figure 10.2 and
rotated at 50,000 rpm while heated by hot gasses to a peak rim temperature of
1260 oC (2300 oF). These conditions were representative of the gas turbine
operating conditions and the speed, mass, and shape of the part were designed
so that some of the rotors would fracture in a time-dependent manner, from
slow crack growth of preexisting flaws. The goal was to correlate failure times
to predictions based on life prediction computer models.
10-1
Fractography of Ceramics and Glasses
Twelve model rotors were fabricated and tested. The rotors were designed to
have a high probability of survival on loading (> 95 %), but to probably fail
within 25 h. The design analysis assumed that failure would occur due to slow
crack growth of preexisting volume-distributed flaws. The rotors were
intended to break in the thin web where the maximum principal steady state
stress was 131 MPa (19 ksi) and the peak temperature was 1240 oC (2260 oF).
Figure 10.1. A Ford silicon nitride 95 mm diameter model gas turbine rotor
that survived 25 h intact. The outer rim is thicker than a real rotor in order to
simulate the mass of the vanes that would be attached in a real rotor.
ceramic spacer
ceramic rotor
Figure 10.2 The Ford hot spin test rig (Ref. 2).
10-2
Case Studies
Six broke in a time dependent manner as intended. Two rotors broke on initial
startup. Four were run outs and did not fail in the allotted 25 hour times and
were treated as censored outcomes.
The rotors were made of a state-of-the-art hot-pressed silicon nitride a that was
carefully machined to final dimensions. The particular grade was one of the
most thoroughly-analyzed structural ceramics of all time and was eventually
used as the world’s first reference material for the property fracture toughness. 3
Eventually enough data was available that a comprehensive fracture
mechanism map was constructed, 4 but that was after the conclusion of the Ford
study. The finite element and heat transfer models used the best data available
and temperature-dependent material properties. A substantial amount of
laboratory test coupon data were available including flexural and tensile
strengths, elastic properties, and slow crack growth parameters from three types
of tests: variable stressing rate strength tests (dynamic fatigue), crack velocity -
fracture mechanics tests (double torsion), and stress rupture (flexural and
tensile).
Figure 10.3 shows the reliability as a function of time. The three types of slow
crack growth data gave very divergent predictions. The six data points show
the six rotors that failed at times from 0.2 h to 18.6 h. The predictions made
a
Grade NC 132, Norton Co., now St. Gobain Advanced Ceramics, Worcester, Ma.
10-3
Fractography of Ceramics and Glasses
with the laboratory stress rupture data gave the best correlation to the actual
lifetimes. The Ford authors concluded that stress rupture data was the best for
predicting reliability versus time.1,2
Fractographic analysis revealed that fracture did not occur quite as anticipated.
The test rig was constructed in such a manner that most of the fragments were
retrievable. The longest (18.6 h) and third longest (13.8 h) rotors were sent to
the author for fractographic evaluation. It was felt that these would have large
slow crack growth zones that would be easy to find. One virtue of NC 132 is
that it was a very fine-grained fully-dense material with minimal second phase.
Rotor spacer
g
A h
A
g g
A h
Primary
h origin
Figure 10.4 Reconstructed model rotor and spacer. Fracture initiated either
at location “A” on the left side of the bore of the rotor or at two locations “g”
or “h” on the opposite side. Fracture started at “h.” Fracture “g” occurred
later. The lower schematic shows the overall fracture pattern in the rotor.
10-4
Case Studies
Fracture markings were quite clear and the local direction of crack propagation
could be assessed in every single piece. One rotor (18.6 h) was reconstructed
piece-by-piece over the course of one week culminating in the assembly shown
in Figure 10.4. It was very much a puzzle-solving exercise. Many fragments
from the rim were missing, but they did not matter. The figure also shows that
the ceramic spacer part had also fractured and its pieces were commingled with
the rotor, thereby complicating the matter. Nevertheless, the central and web
portions of the primary rotor were almost completely reassembled. After
completion, the rotor was taken apart again and a map of crack propagation
directions was made. The interpretation was clear. Primary fracture
commenced at the bore or the curvic coupling teeth and cracks branched and
ran out to the rim as shown in the schematic in Figure 10.4. Every single
fragment was examined in a futile search for a web origin with a slow crack
growth zone, but none were found. All primary fractures started from the bore
or coupling area. There were three origins from the inner part of the rotor.
Origin “A” was at the exact edge of the bore, which although chamfered,
nonetheless had surface grinding cracks (Figure 10.5a-c). It could not be
determined whether slow crack growth had enlarged the grinding cracks. The
other two origins “g” and “h” were located on the other side of the bore and
were from the curvic coupling teeth machined into the rotor (figure 10.5d,e.
These were part of the attachment scheme. One origin, “g”, was a parallel
machining crack (aligned parallel to the axis of grinding) in the tooth, and the
other, “h”, was an impact – contact crack also on a tooth.
Thus, fracture started on one side of the rotor in the bore area at either “g” or
“h,” then branched into two main crack(s) that ran out to the rim. This opened
up the disk on that side. The unbalanced forces triggered a rupture on the
opposite side at site “A.” Origin “A” had a well-defined small fracture mirror
allowing a stress estimate of a 630 MPa (92 ksi) which is much greater than the
model estimated bore stress of ≈172 MPa (25 ksi). This, plus the violent
branching at that site, confirms that site “A” was a secondary fracture. Primary
fracture occurred on the other side of the bore at site “g” or site “h,” causing
the rotor to go completely out of balance. Origin “h” was determined to be the
first break based on crack intersection patterns. The flaw was a large contact
damage crack with a large fracture mirror. The local stress was estimated to be
500 MPa (72 ksi) from the flaw size and approximately 580 MPa (84 ksi) from
the fracture mirror. Origin “g” was determined to have fractured after site “h”,
since the crack from “g” stopped when it encountered the prior crack from “h.”
At these locations the computer models indicated there were modest bore
stresses (≈138 MPa, 20 ksi) and temperatures (1150 oC, 2200 oF). These did
not match the fractographic estimates.
10-5
Fractography of Ceramics and Glasses
origin A
(a) (b)
Branch
origin A
(c)
Fracture
Mirror
origin h
(d) (e)
Figure 10.5 Fracture surfaces showing a fracture mirror and the origin site
A, which is a surface crack from grinding. (b) and (c) highlight the fracture
mirror and grinding crack origin at site “A.” The primary fracture origin was
contact crack damage at site “h” on the curvic coupling teeth as shown in (d)
and (e). Origin “h” is marked by the larger arrow on the right. Notice how
the radiating hackle lines converge on it.
10-6
Case Studies
This exercise was a good example of one of the author’s laws of fractography:
“The first one is the hardest.” While at first glance, reassembly of a burst rotor
may appear to be a formidable task, it was merely time consuming. The second
rotor took less time to analyze. It also had fracture initiation sites at the bore or
at the teeth. Years later, the author became aware of comparable work done
on model silicon nitride rotors at Daimler-Benz’s research center in Stuttgart. 5
They also reconstructed burst silicon nitride rotors and used fractography to
find that grinding cracks in the bores sometimes were fracture origins.
machining cracks and contact cracks were found and Weibull area
The U. S. Army has off and on over the past twenty-five years investigated
ceramic liners to improve gun barrel life and reduce mass. The low density,
high compression strength, refractoriness, and erosion and wear resistance of
ceramics could be advantageous. Figures 10.6 and 10.7 show drawings for a
50-caliber machine gun breech with a ceramic liner from a project conducted
for the U. S. Army in the 1980s. 6 One design placed the ceramic into
compression by shrink fitting a steel sleeve around the ceramic. The steel
sleeve was heated and placed over the cool ceramic tube. As the steel cooled, it
contracted and put the ceramic liner into axial, radial, and hoop compression.
The dimensions and temperature differentials were chosen so that the ceramic
liner was always in compression, even when firing a projectile. Several
ceramics were tried, but most testing was on a sintered α-SiC. Figure 10.7
shows the maximum axial stresses, σx, in the three components
10-7
Fractography of Ceramics and Glasses
Breech Muzzle →
Figure 10.6 Schematic of the breech end of the 50 caliber (12.7 mm) gun
barrel (Ref. 6).
Steel jacket
Steel sleeve
σx
(a)
Ceramic liner
670 MPa
(b) X
10-8
Case Studies
of the assembly from combining the residual shrink fit and projectile firing
stresses. The maximum axial stress σx was 670 MPa in compression in the
ceramic liner mid-portion. This stress tapered to zero at each end. The axial
stress maximums were only 112 MPa tensile in the steel sleeve and 154 MPa
tensile in the steel jacket. The maximum radial stress in the ceramic sleeve
was 345 MPa compression, and the maximum hoop stress, σθ, was 590 MPa
compression during projectile firing. The environment is severe, but if the
ceramic was always under compression, perhaps it would not fracture.
So if the shrink fitting created compressive stresses, where did the tensile
stresses come from? The plane of the fractured surfaces and also of the initial
semi-elliptical contact cracks indicated that the tensile stresses were axial: σx.
Figure 10.7 shows that the axial residual compressive stresses did taper off
towards the tube ends. The most likely sources of tensile stresses are dynamic
stress waves generated during the firing. Even if these are initially
compressive, they can change phase and become tensile when the stress wave
reflects off of the end faces. Furthermore, the elastic wave velocity and
impedance of the silicon carbide and the steel sleeve are not matched, and
stress waves propagated axially at different rates. Hence, a dynamic tensile
stress could develop at the steel/ceramic interface due to mismatch of the
transient elastic strains.
The contact stress cracks often were periodic around the rim. Although the
parts were machined to tight tolerances, it is likely that slight variations in the
mating surfaces led to an uneven fit and stress concentration sites that triggered
the contact cracks.
So, in this case, the ceramic was designed to always be in compression, but
fractography showed otherwise. Some design and modeling improvements
were suggested. The tolerances and surface specifications for the mating parts
10-9
Fractography of Ceramics and Glasses
could be changed. The elastic properties of the ceramic and the confining
sleeve could be matched better. More sophisticated stress models could
examine the transient stress states in the assembly.
(a) (b)
(c)
(c)
Figure 10.8 Fractured α-SiC gun barrel. (a) is a schematic sketch. (b)
shows the fracture surface of one of the ring shaped fragments. Contact cracks
are spaced periodically around the outer rim and appear as shadows in this
view (arrows). The bore diameter is 12.7 mm. (c) shows a close-up of one of
the contact cracks.
10-10
Case Studies
This case is not about a specific component, but about the risk of treating all
flaws as if they behave the same. As part of a program on characterization of
structural ceramics for heat engine ceramics in the 1980s, the author conducted
extensive stress rupture testing of pressureless-sintered α-SiC. b 7,8,9 Stress
rupture testing entails applying a constant stress to a specimen, at a level below
that needed to cause fast fracture, and measuring the time to failure. Slow
crack growth can cause flaw enlargement and time-dependent fracture.
Sometimes such tests are termed creep rupture if the loading conditions are
such that bulk creep deformation leads to accumulated damage and fracture.
The author’s data7,8,9 at 1300 oC in air was consistent with data published by
another team. 10 Both teams detected time-dependent fractures due to
intergranular slow crack growth of preexisting flaws as shown in Figure 10.9.
Figure 10.9 Fracture origin in a 1300 oC stress rupture SiC specimen. The
origin is a large intergranular slow crack growth region that may have started
from the side of or beneath the large pore.
b
Hexoloy SA, Carborundum, Co., Niagara Falls, NY. Now St. Gobain Advanced
Ceramics.
10-11
A much different story emerged at 1200 oC. This author recorded time-
dependent fractures such as shown in Figure 10.10 with origin flaws as shown
in Figure 10.11. The other team did not observe a single time-dependent
fractures at all in their experiments. Why did the two groups have such
contrasting outcomes at 1200 oC? The answer was that they used Knoop
indented, artificially-flawed specimens, whereas the author used as-machined
specimens that allowed the material to fail from whichever of its flaws it
“chose.”
Figure 10.10 Stress rupture curve at 1200 oC in air for sintered α-SiC.7-9
Every time-dependent fracture was from a pore or porous zone and only if it
was surface connected. The hollow points with arrows to the left are breaks on
loading and are consistent with the fast fracture strength which was about 350
MPa. The slow crack growth exponent N was 40.8.
(a) (b)
10-12
Case Studies
On the other hand, every one of the 1200 oC time-dependent fracture origins
was a pore or porous region connected to the outer surface as shown in Figure
10.11. These flaws did not have slow crack growth markings. Evidently these
flaws were susceptible to very localized stress-corrosion crack extension (from
oxidation) that sharpened or locally extended tiny microcracks on the pore
periphery. These flaws did not need much change in flaw size or severity for
them to go critical, since the failures occurred at stresses (250 to 350 MPa) very
close to the fast fracture strength (350 MPa). Figure 10.12 summarizes the
fractographic findings.
10-13
The other testing team did not detect the 1200 oC stress corrosion mechanism
since they used Knoop artificial flaws in all of their bend specimens.10
Artificial flaws often are effective tools for studying fracture, but in this
instance they produced misleading results. At 1300 oC and above, the Knoop
flaws did grow by slow crack growth. They did not grow by slow crack growth
or by stress corrosion at 1200 oC. The Knoop flaws were not susceptible to the
same failure mechanism as the pores and porous zones in the as-machined
specimens. The Knoop flaws reduced the fast fracture strength of the
specimens to less than 200 MPa. Their stress rupture specimens were loaded at
stresses from 100 to 200 MPa, a stress well below that necessary to activate the
pore flaw stress corrosion mechanism.
This case demonstrates that it is often best to let a material reveal what type of
flaws it is apt to fail from, rather than rely on artificial indentation flaws.
10-14
Case Studies
Close-up view of
the neck
Aluminum cap
Glass vial
(a) (b)
(c)
Figure 10.13 Rim cracking in glass medicinal vials. (a) and (b) show blunt
contact cracking damage (arrows). (c) shows a top view of the fracture surface
of a vial with a severed cap.
10-15
This case illustrates what can be accomplished with optical microscopy alone,
even on very difficult ceramics. A coarse-grained aluminum oxynitride missile
sensor dome broke during a thermal stress proof test. It broke into several
pieces, and the branching patterns directed attention to an area near the base as
shown in Figure 10.14a. One might expect the origin to be at the dome rim,
possibly at a bevel, but that was not the case.
The roughness of the microstructure hid nearly all fracture markings as shown
in 10.14b. Careful adjustment of low-angle grazing lighting at higher
magnifications revealed local hackle lines and twist hackle within the
transgranularly-fractured grains as shown in Figure 10.14c and d. It was
possible to read these lines back to the inside surface of the dome to a location
well away from the rim. Close-up examination in Figure 10.15 revealed the
origin site had zipper machining crack hackle lines within one or two grains,
indicating the origin was grinding or scratch damage that was not completely
removed by polishing.
(a) (b)
(d)
(c)
Figure 10.14. AlON infrared missile dome. (a) shows how the branch on the
right leads away from the origin area. (b) shows the fracture surface with the
coarse microstructure.
10-16
Case Studies
Unfortunately, the dome was not available for SEM confirmation or further
examination of the inside surface for traces of a scratch at the origin site. The
polished surfaces in general did have an assortment of polishing flaws as shown
in the insert. Similar grinding/polishing flaws inside single grains of aluminum
oxynitride have been reported by Swab et al. 12 A similar approach may be
used to find origins in coarse-grained polycrystalline alumina (PCA) lamps.
Polished surface
Figure 10.15 Wake and twist hackle within cleaved AlON grains lead back to
the inside surface and to tiny zipper machining crack hackle lines within one or
two large grains. This image was taken at the maximum magnification (125X)
of the stereoptical microscope that was used for the examination. The insert
shows the outer polished surface with polishing scratches and a well-ground
bevel-chamfer.
10-17
A large silicon carbide furnace setter plate fractured during a furnace run. Was
the material faulty or was there some other cause? Figure 10.16 shows one half
of the broken plate. It has the wavy fracture that is a telltale sign of a center-
heated plate fracture as shown in Section 4.12. Figure 10.17 shows the fracture
surface revealing that fracture initiated from grinding damage at a hole near the
rim. No material flaws of consequence were detected. The mirror size was
about 9.3 mm. Using an Ao value of 11 MPa√m for a dense hot-pressed SiC
(from Appendix C), the stress at fracture was estimated to be only 115 MPa.
The material was completely satisfactory, but the machining, grinding, and
attachments were not. This failure analysis required only a one minute visual
inspection.
(a)
(b) (c)
Figure 10.16 Fractured SiC furnace setter plate (a). The curvy pattern is
typical of a center-heated plate fracture as shown in Chapter 4. The arrow
marks the origin which is a hole through the plate. The machined grooves
were chipped and had ragged edges as shown in (b). The plate was 36 cm
round and 1.9 cm thick. (c) shows the origin hole from the bottom surface.
There is spall and evidence of a chemical reaction with a mounting bolt or
washer. (plate courtesy of B. Mikijelj)
10-18
Case Studies
(a)
(b)
Figure 10.17 Fracture surface of the SiC setter plate. A fracture mirror at the
origin (arrow in (a)) attests to a moderate stress level. Hoop tensile stresses on
the rim were generated by thermal strains. Thermal expansion of the hot plate
center was resisted by the cooler rim. As the crack propagated towards the
plate middle it slowed since it reached hot regions that were originally in
compression. The fracture surface became relatively featureless and the crack
meandered. The hole through the plate at the origin (b) had chipping and
grinding damage that weakened the plate.
10-19
First generation U. S. Army Patriot air defense missiles in the early 1980’s
utilized fused-silica radomes (Figure 10.18). Radomes are nosecones that are
radar transmissive. The radome was prepared by slip casting and then sintering
silica particles to greater than 85% density. This porous structure had a low but
acceptable strength and, like an insulating firebrick, was more effective in
arresting cracks than if the radome were a fully-dense glassy body. Although
sintered fused silica is weak, it was selected since it had a low coefficient of
thermal expansion and thus would be less susceptible to thermal stresses from
aerodynamic heating. Selected radomes were deliberately loaded to fracture or
proof tested as part of the engineering development program. This was done by
attaching a sling to the side of the radome and pulling it laterally several times
in different directions. Unexpected fractures occasionally occurred. Most
fractures occurred in the ceramic where it was joined to a threaded fiber-epoxy
thread ring attachment part (Figure 10.19a,b). Fractography was difficult in the
porous weak ceramic since it had a rough fracture surface, but a number of
fractures were successfully diagnosed. Figure 10.19 shows an instance where
atypical grinding damage created strength limiting cracks. Better control of the
surface grinding eliminated the problem. Figure 10.20 shows a different
example wherein excessive shrinkage of the epoxy bond caused debonding.
This created a stress concentration site in the ceramic that initiated fracture.
The remedy in this case was to adjust the bond-cure thermal cycle.
(a) (b)
Figure 10.18 U. S. Army first generation Patriot air defense missile. The
radome was 1.1 m tall with a 40 cm diameter. (b is courtesy Raytheon)
10-20
Case Studies
Fused
silica
Epoxy
bond
Polymer
spacer
(a) (b)
origin crack
bond
(d)
Figure 10.19 Sintered fused silica radome that fractured during a proof test.
(a) is a top view with the origin region marked by the arrow. The ceramic is
bonded on the inside to a fiber epoxy attachment ring. (b) is a schematic
showing the bottom of the radome where it was attached to the thread ring.
Fractures in the ceramic typically occurred near the top of the thread ring. (c)
shows a grinding crack at the origin area (arrows). (d) shows a grinding
pattern on the inside ceramic surface.
10-21
(a) (b)
(c)
Fused silica
Debonded
region
(d)
Epoxy bond
Figure 10.20 Sintered fused silica radome fracture. (a) shows the origin site
on the inside wall. (b) is a close-up showing a fracture mirror centered on a
feature on the inside wall of the ceramic at the top of the bond line. (c) is a
composite of two SEM photos of the origin area. The origin is less obvious, but
hackle lines point back (white arrows) to the circled area. (d) shows there is
no significant material flaw at the origin, but the ceramic and the epoxy are
debonded at the origin.
10-22
Case Studies
This case was part of an investigation whether different machine shops could
prepare common bend bars to a set of specifications, with minimal damage and
at a reasonable cost. The work described here was done in the 1980s in support
of the first standard test method for flexural strength of high performance
ceramics in the United States, MIL STD 1942(MR). 13 The standard, which was
adopted by the U.S. Army in 1983, served as a basis for American Society for
Testing and Materials standard C 1161 in 1990 14 and (along with elements
from analogous European and Japanese standards) was used as the basis for
Inter-national Organization for Standards ISO 14704 in 2000. 15
Standardization has led to dramatic improvements in the quality of data and
significant cost savings. Bend bar costs decreased as machine shops prepared
specimens to a common, reliable procedure.
It has long been known that grinding may introduce strength-limiting damage.
A side topic investigated in the round robin was whether different machine
shops could meet the grinding specifications and prepare damage-free B type
bend bars (3 mm x 4 mm x 50 mm). Requests for price quotations were sent to
eight machine shops to prepare 20 specimens from the sintered alumina
according to the MIL STD 1942. The specifications were for a multi-step
grinding process ranging from coarse initial grinding to final finishing. Finer
grinding wheels with reduced depths of cut were used at each step with the
objective of removing prior damage and minimizing final residual damage.
One shop did not bid. Two others bid $101 and $112 per bar and were not
contracted since these prices were exorbitant. In 1984 – 1985 the typical price
for bend bars from experienced shops were in the $12 to $25 range. Plates
from the common batch were sent to each of the remaining five shops which
bid from $15 to $50 per bar. The bars that they machined were then broken in
four-point flexure on 20 mm x 40 mm spans. The surfaces and edges of all
bars were inspected beforehand to ascertain whether the bars met specifications
or if there were any signs of machining damage. The results are shown in
Table 10.1. The initial visual examination showed that the five venders met all
of the specifications for the most part. Deviations were usually on isolated
10-23
Table 10.1 Flexural strengths of sintered 99.9% alumina from test sets
prepared by different machine shops. The flaws that were strength limiting are
listed in their order of frequency in each set.
10-24
Case Studies
Porosity was the most common flaw type, but it manifested itself differently in
the plates. In some plates the flaws were primarily round discrete pores. In
other plates, the flaws were equiaxed regions of microporosity, porous seams,
or porosity associated with inclusions. The plates were prepared from the same
powder lot by the identical procedure and to all external appearances were
identical. Only when the specimens were fractured was the true flaw character
revealed.
Fractographic analysis was difficult, but productive. 19,20 The CVD SiC had a
coarse microstructure that created a very rough fracture surface that masked
common fracture markings. Large SiC grains affected the crack propagation
across the fracture surface, causing significant crack redirection and severe
roughness as the propagating crack sought out preferred cleavage planes. The
most helpful features for pinpointing the origin were twist hackle lines on
cleaved grains and occasional large hackle lines that showed the local direction
of crack propagation. These markings led back to the origins.
10-25
Notched
Straight
Curved
1 mm
(a) (b)
(c)
100 µm
Figure 10.21 Miniature 3.1 mm long SiC tensile test specimens. (a) shows
curved, straight and small notched gage sections. The notch is too small to see
in this view. (b) shows a fractured curved gage length specimen (c) shows the
fracture surface of a straight section specimen. The arrow shows the origin is
a 25 µm x 50 µm large grain at the root of an etch groove. (photos a and b,
courtesy of W. Sharpe)
The fracture origins in all three specimen types usually were a combination of a
deep etch groove that combined with a large, favorably-oriented columnar
grain. Cracks popped-in on a preferred cleavage plane in such grains. Thus,
most critical flaws were a hybrid surface-volume type flaw: a sidewall groove
and a large grain as shown in Figure 10.21c. The fractographic results
supported the Weibull area scaling for comparing the straight and curved
10-26
Case Studies
specimens, and the best correlation was obtained if only the etched sidewall
areas were used in the calculation.
The reason the Weibull strength scaling did not work for the intentionally-
notched specimens became obvious as shown in Figure 10.22. The notches
were quite small (15µm to 25 µm radius). They concentrated stress to a very
small region around the notches. Hence, the Weibull effective volumes or
areas were tiny, and comparable to the size of some of the single grains.
Hence, one cannot assume there is a well-distributed set of flaws scattered
throughout the stressed volume, as is assumed in conventional Weibull
analysis. Furthermore, an analysis based on continuum mechanics assuming
the material was isotropic and homogeneous is questionable.
notch notch
(a) (b)
(c) (d)
Figure 10.22 A broken notched miniature SiC tensile specimen. (a) shows the
whole specimen. (b) shows a close-up of the gage section notches. Trans-
illumination through the green single crystal SiC grains created the light
patches. They are individual grains. (c) and (d) show the fracture surface.
Fracture started at the bottom of the round notch (arrows), from a etch groove
linked with a single long and large columnar SiC grain that ran from the top to
the bottom.
10-27
Can different fractographers agree on fracture origin analysis? Can they reach
similar conclusions? How do they exercise their craft?
This author and Dr. J. Swab of the U.S. Army Materials Technology
Laboratory (now with U. S. Army Research Lab, Aberdeen, MD) conducted a
major international round robin on characterization of fracture origins of
advanced ceramics in 1994 - 1995. Seventeen laboratories participated. The
participants had experience that ranged from zero to thirty-five years. The
project was coordinated under the auspices of the Versailles Advanced
Materials and Standards (VAMAS) program. This work evaluated a new
(1992) set of guidelines for finding and characterizing fracture origins in
ceramics: “Fractography and Characterization of Fracture Origins in Advanced
Structural Ceramics,” Military Handbook MIL HDBK 790 22 which was a
predecessor to ASTM Standard Practice C 1322, 23 the world’s first
fractography standard for characterizing fracture origins.
Topic 1 of the VAMAS round robin was a photo interpretation exercise with
focus on detection and characterization of grinding damage in bend bars.
Photos of one specimen each of silicon nitride, a zirconia-alumina composite,
and an alumina were furnished. Six photos of each specimen were provided
showing for each half, a low magnification overall shot, a picture of the
fracture mirror area, and a close-up of the origin area. Results were mixed.
There was consensus that the origins were machining cracks, but there was
considerable variability in how they were marked and measured. Some
participants disagreed and felt the origins were not grinding damage. Although
a short paragraph on each specimen with strength and fracture toughness
information was furnished along with the photos, very few of the participants
used this information to estimate a critical crack size. Participants were also
asked to mark and measure the fracture mirror sizes on the photos. The
organizers were surprised at the scatter in mirror size estimates. Evidently
many participants were not familiar with measuring fracture mirrors.
10-28
Case Studies
Topic 2 was an actual examination of six bend bar strength specimens. Each
participant received one example of the specimens listed in Table 10.2. Most
were carefully prepared (no small feat) by Dr. Swab. Again, results varied.
Most participants had no trouble identifying the pore as the fracture origin in
specimen #3. Many had difficulty with the others. For example, the scratch
that caused cracking damage in specimen #2 was sometimes identified as
machining damage or even a pore or large grain. Figure 6.25c shows one of
these scratched specimens. The scratch should have been obvious. Dr. Swab
had made it with a diamond indenter and it stood out very clearly on the tension
surface of the bend bar. Evidently some participants did not bother looking at
the external surface and only looked at the fracture surface.
In the other specimens, participants looked at only one half. Sometimes this
was satisfactory, but in just as many cases observers looked at the less clear
piece.
Many participants evidently did not read the brief four to six sentences of
information furnished with each specimen, including treatments if any and the
breaking stress and fracture toughness. For example, the surface pit origin
specimens #4 had been exposed to an oxidizing environment at high
temperatures. Some labeled the origin a pore and ignored the surface condition
and reaction zones around the pits. Only a few participants estimated flaw
sizes from fracture mechanics and compared them to their fractographically-
measured sizes for verification.
Specimen
Material Origin Flaw
type
1 Hot-pressed alumina with silicon Large grain
carbide whiskers
2 Sintered 99.9% alumina Handling damage (scratch)
3 Sintered 3 mol% yttria stabilized Pore
zirconia (Y-TZP)
4 Reaction bonded silicon carbide Surface Pit
5 Hot isopressed silicon nitride Machining damage
6 Sintered titanium diboride Porous seam or porous region
Table 10.2 Bend bar specimen types distributed in Topic 2 of the VAMAS
round robin.
10-29
Fractography of Ceramics and Glasses
The entire exercise was a learning experience for the organizers. Some
participants did very well, but many had difficulty. Much was learned by
studying the causes of the misinterpretations. Even for such mundane objects
as bend bars, there was considerable variability in how the participants
analyzed the photos, the specimens, and the ancillary information provided.
Some simply looked at the fracture surfaces and ignored everything else. The
organizers came to realize that fractographic analysis (and failure analysis in
general) is a process whereby an expert integrates all information including, but
not limited to, fracture surface examination. This realization led the author and
Dr. Swab to prepare Figures 1.1 (the fractographer as detective) and 1.2 (the
fractographic analysis puzzle) of this Guide. The organizers also came to the
realization that they had a huge advantage compared to the participants. The
organizers could look at many specimens of a type, and were not limited to a
solitary example. The participants only had one. Furthermore, there is no
substitute for actually looking at the actual specimen under a microscope, as
opposed to looking at two-dimensional photographs as in Topic 1. Several of
the author’s fractographic Rules of Thumb listed in the next chapter stem from
this exercise.
c
Dentsply Int., York, PA.
10-30
Case Studies
“foundation.” The crowns also have a glassy exterior veneer layer added for
cosmetic purposes. Brittle, highly-filled resin matrix composites with ceramic
or glass fillers are common crown materials as well, but do not need a veneer
layer. The fracture resistance and durability of the new ceramic and composite
crowns are a primary concern. Despite the sanguine claims of the manufact-
urers, premature fractures do occur, much to the consternation of the patient
and his or her dentist.
Progress in the field has been typified by a trial and error approach. The usual
approach has been to statistically analyze the rates of failures and infer or
guess causes of failures. There are conflicting explanations as to why the
crowns do or do not fail. It is not even clear what properties are desired and
what are the best laboratory-scale tests to evaluate these properties. Until the
late 1990s, postmortem failure analyses were quite rare.
One key reason for this was that dental ceramic crown fractures are in fact
among the most challenging problems to solve. Many dental ceramics have
coarse-grained and/or porous microstructures. They often fracture at low stress
and classic fractographic markings are often masked by the roughness of the
fracture surface. High-density aluminas and 3Y-TZP zirconias are exceptions
and are easier to interpret. Another complication is that stress states in the
crowns are transient and uneven. Fracture mirrors are almost never observed in
a clinical fracture since fracture stresses are low, or they are larger than the
crown wall sizes. Another difficulty is that key fragments are often missing,
either due to loss in the mouth or damage during crown extraction. Crowns
usually accumulate damage from multiple events at multiple sites, creating
complex, conflicting fracture networks. For example, the author has often
detected Hertzian cone crack damage sites on crown occlusal and facial
surfaces, but has only seen a few instances where the cone acted as an actual
crown fracture origin. The Hertzian cones crack damage sites usually are
dormant. Contact damage does occur and cause clinical failures, but it almost
always in due to edge chipping.
d
Originally developed by Coors Biomedical, Lakewood, Co.. Available later from
Ceramco Inc., East Windsor, NJ.
10-31
following three examples, gold coating has made the veneer look dark in the
optical images and the core material lighter. The origin was located at the
margin (or bottom) of the crown and the vertical-split nature of the fracture
suggests that hoop stresses around the bottom of the crown, where the crown
Half B origin
Half B
Half A
Half A Missing triangular piece
(a) (b)
(c)
Figure 10.24 a-c Fractured alumina-spinel molar crown. (a) is a top occlusal
surface view showing the two halves held together. Half A is fully- and half B
is partially-gold coated. (b) shows the interior looking up into the crown. The
missing triangular piece was caused by secondary breakage. The origin was
on the margin, at the bottom of the crown. (c) shows an optical photo of the
fracture surface with arrows showing the local crack propagation direction.
The dark outer regions are the gold-coated glassy veneer. The lighter gray
portions are the gold-coated core ceramic material. The letters d, e, f identify
regions shown in images (d) – (f) on the next page. (images courtesy J. Quinn)
10-32
Case Studies
veneer
core veneer
(d) (e)
(f)
ceramic glass
core veneer
Figure 10.24 d-f (continued). (d) wake hackle from tiny bubbles in the veneer
glass. This location is marked on the left of view (c) and indicates the crack
moved right to left (arrow). (e) shows wake hackle and gull wings from
bubbles in the veneer on the right side of the crown. At this location the crack
was running upwards from the margin. Careful scrutiny of the adjacent core
ceramic material showed that it also had wake hackle generated by pores in the
ceramic pointing in the same direction. (f) is an optical image from the origin
area at the margin. There is no obvious flaw.
10-33
was the thinnest, caused the breakage. A specific material flaw could not be
identified. This is not unusual with dental crown fractures: an exact site is
identified, but a specific flaw is not obvious. It is very likely that an initial
radial crack has popped-in at some point, and then final fracture occurred on
the same plane. Sometimes the final fracture will be on a slightly different
plane, and an arrest line from the pop-in will be apparent as evident in the next
example given below. By the time the crack propagated over to the opposite
side of the crown, the two halves hinged apart in bending, creating the
compression curl.
Figure 10.25 shows a second example, a similar case from Ref. 24, but for a
Procera 99.9% purity alumina. e It also split vertically, evidently in response to
a hoop stress. The fractographic markings again lead to a margin origin site. A
specific single material flaw is not evident. In the first edition of this Guide,
the author stated that the origin site coincided with the end of the cement bond
to the core material, on the inside of the crown. The author reevaluated the
crown and now believes the origin was in fact a barely-discernable semicircular
crack in the core material at the end of the margin as shown in Figure 10.28f+g.
The origin crack was on almost exactly the same plane as the final fracture. It
was not discernable in SEM photos. From its size (179 µm radius), and using
Occlusal surface
Compression
curl
Compression Corner
curl hackle
Half A Half A
Margin
origin
(a) (b)
Figure 10.25 Procera alumina molar crown. The crown split vertically into
two halves. The fracture surfaces have been gold coated. (b) is a close-up of
the compression curl on the side away from the origin site. The veneer glass is
dark and the core ceramic is a lighter shade. The arrows show the dcp from:
corner hackle around the inside corner, the compression curl, and from veneer
wake hackle. (c) – (g) are on the next page.
e
Nobel Biocare, Stockholm, Sweden.
10-34
Case Studies
KIc of 4.0 MPa√m for dense fine-grained alumina, and a Y of 1.4, the hoop
stress at fracture was estimated to be 213 MPa. The margin in this example
was well-prepared and over 1 mm thick.
(c) (d)
Figure 10-25 continued. (c) and (d) are examples of wake hackle from pores
in the veneer. (c) is an optical image, and (d) is an SEM image. (e)-(g) show
optical images of the margin. The origin is a subtle margin crack as marked
clearly in (f) and (g). This origin crack was so flat and coplanar with the final
fracture surface, that it did not stand out well in SEM images. (all images
courtesy of J. Quinn and S. Scherrer)
10-35
Cantilever
curl
Origin
(a) (b)
veneer core
dcp
(c) (d)
veneer
As received Cleaned
and
magnified
edge chip
(e) (f)
Figure 10.26 Procera alumina molar crown that broke from a margin chip.
(a-f) are optical images without any coatings. (a) is a clinical view with the
origin marked. (b) is one half of the fractured crown. (c) shows a fanlike array
of corner hackle (between the two arrows which show the propagation
direction) in both the veneer and core emanating from the top inside corner. (d)
is from near the margin on the left side and shows tiny wake hackle from
veneer pores. The crack ran from bottom to top. (e) and (f) show the origin
was an edge chip from a force applied on the bottom of the margin aimed
upwards (image a courtesy of S. Scherrer).
10-36
Case Studies
The third crown, also a Procera alumina, is shown in Figures 10.26 and 10.27.
It also has a margin origin site that led to crack propagation by hoop stresses.
Images 10.26 a-f are all optical images of the crown without coatings. As
such, some of the details were “washed out” and did not photograph well due to
the material’s translucency and light scattering. Nevertheless, this crown
failure was diagnosed solely on the basis of an optical examination with a
stereo optical microscope. Subsequent SEM examination produced the sharper
images shown in Figures 10.27. A full fractographic montage for this crown
was shown previously as Figure 5.68.
Corner hackle around the inside corners of the core, a compression curl, and
wake hackle again led back to a margin initiated fracture origin. A well-
defined edge chip on the veneer was the origin. The edge chip was discolored
suggesting that it had been present for quite some time. The direction of the
edge chip was puzzling as well. It appeared to have been from a force directed
upwards on the crown. Such a loading is not likely in the mouth since the
margin is protected by the gum line and jaw. The chip was probably created
during fabrication or installation.
veneer core
(a) (b)
Figure 10.27 SEM images for the Figure 10.26 Procera crown that broke
from a margin chip (arrows). The final fracture that split the crown started at
the bottom at the margin, from part of the margin chip, and ran as shown by
the black arrows. (images courtesy of S. Scherrer)
f
Ivoclar, Schaan, Liechtenstein.
10-37
fracture plane was vertical and perpendicular to the dental arch. The two
halves fitted closely together. A few small chips were missing, probably from
secondary fracture, and were inconsequential to the analysis. The lingual
surface (the side that faces the tongue) had some surface damage and also an
unusually thick veneer. There were several shallow depressions (≈ 1 mm2
square area) as well as numerous conchoidal chip fractures, small pits and
gouges at mid crown height. The shallow depressions had tool marks
indicating they were from adjustments to the lingual surface by the dentist once
the crown had been cemented in place and opposing tooth contact checked.
This adjustment was only partially successful in relieving contact from the
opposing tooth as evidenced by the numerous pits, chips and gouges. This
location was eventually determined to be near the fracture origin region.
The facial surface (the surface that faces outwards) was in good condition and
had little damage or evidence of abuse. The unusually thin veneer had a few
secondary spall chips and one harmless Hertzian cone crack that did not
penetrate into the core material. The chips were noted on one fragment but not
on the other, indicating that they were secondary fractures that occurred after
the crown had broken.
Occlusal surface
Facial
surface
Margin
(a) (b)
Figure 10.28 Fractured Empress II glass ceramic anterior (#7) incisor crown.
The insert (a) shows a facial view of the cleaved crown. (b) is a view looking
up into the interior with the two halves together. The origin is located between
the two red dots on the inside surface.
10-38
Margin
Case Studies
(a)
(b)
Figure 10.29 The fracture surfaces of the two halves. (a) shows an optical
photo of the uncoated halves. (b) shows gold-coated surfaces. Two different
coaters were used causing the different coloration. The gold coating of the
glassy veneer reflects more. The veneer thickness is very uneven and out of
normal guidelines. It is very thick on the lingual side and very thin on the
facial side. The arrows mark the origin site located at the thinnest part of the
core glass ceramic material.
10-39
Fractography of Ceramics and Glasses
(a)
core veneer
(b)
Figure 10.30 Details of the incisor crown fracture. (a) is an SEM image wake
hackle in the glassy veneer from a region near the top (occlusal) surface of the
A half. The wake hackle shows that the crack was running in the direction of
the arrow in this region. All the wake hackle in the entire crown showed
fracture ran from the inside to the outside. (b) is an optical image of the origin
which originated from a contact crack on the inside of the crown at the thinnest
point of the core.
10-40
Case Studies
(a)
(b)
Figure 10.31 SEM images of the origin. Notice all the wake hackle in the
veneer aiming away from the interior. The crack ran from the inside to the
outside. The inside surface is uneven and there are mold or tool marks in the
vicinity. This region is directly opposite a region of extensive localized contact
damage on the outer veneer surface.
10-41
There was negligible damage on the occlusal (top) surface. Some residual
porous powdery material was detected on the inside surface of the crown and
was the remnants of the cement. Tool grinding marks and/or casting
impression marks were detected on the inside of the crown.
This was a difficult analysis. Initial examinations had only limited success, but
with repeated careful scrutiny of the fragments using different examination
techniques (optical, optical green dye coated, optical gold coated, and SEM)
enabled the author and J. Quinn to find the origin region. Maps of crack
propagation direction were constructed, primarily through observations of the
wake hackle markings in the veneer. Fracture initiated in the core material
(where it was very thin) from an internal surface Hertzian cone crack. This
location was in the same vicinity where external lingual surface damage has
also been detected. Fracture then radiated upwards and downwards in response
to hoop stresses causing the crown to split into two halves.
Only after repeated examination was it realized that both fracture halves
showed matching damage features characteristic of blunt contact damage at the
exact same location on the inside (cementatious) surface of the crown. The
critical observation occurred when stereo SEM paired images were viewed of
this area, and the arc-shaped crack that is typical of blunt contact damage was
revealed. It was further observed that the inside crown surface was uneven and
had raised hills, ridges, and depressions.
In summary, the Empress II incisor crown had uneven veneer and core wall
thickness. Fracture initiated from contact damage on the inside of the crown,
probably due to an uneven fit or improper placement. Opposing tooth contact
in the same vicinity on the crown exterior surface (as evidenced by shallow pit
and chip damage) created the stresses in the origin area. Surface grinding
adjustments made to the same area suggested a fitting problem.
These four crowns broke from hoop stresses and split vertically, perpendicular
to the dental arch. Specific gross material flaws were not be detected in any of
the four.
10-42
Case Studies
Recent work that typifies progress in the dental fractography field is presented
here. The failures shown above in Case 11 were isolated cases. Most failure
analyses have been on such isolated examples or on batches of laboratory-
tested (in vitro) crowns, which may or may not be accurate representations of in
vivo failures. Indeed, many “crunch the crown” tests with loading by steel or
tungsten carbide balls on the occlusal surfaces are ridiculous and clinically
irrelevant.
The cases presented here are the first study to identify a single leading cause of
fracture, with high specificity, for a large number of cases in one particular
restoration system. Prof. M. Oilo at the University of Bergen, Norway has
collected dozens of clinical fractures and is organizing them by type. As
discussed in Chapter 4, she has also devised a laboratory set-up that causes
crowns to fracture into the same patterns as clinical fractures. 30,31 The author
and she identified the fracture origins in twenty-two crowns of a set of twenty-
seven retrieved crowns, all from one material system. 32 The crowns were all
high-alumina crowns prepared by CAD/CAM machining. They fractured in
vivo from times as short as days to years. Twenty-one of the twenty-two
crowns (95%) broke from one cause: hoop stresses on the margins that split the
crowns, usually at the area of shortest wall length in the approximal area, with
the fracture origins being irregularities in the margin including chips, grinding
cracks and veneer irregularities. Figures 10.32 - 10.34, show several of the
fractured crowns.
One common element in all these crowns was irregular grinding damage on
feather margins that were as thin as 100 µm (0.1 mm). Compare these margins
to the nice, thicker ones in Figures 10.24 and 10.25. It is unwise to make thin
ceramic margins since they are vulnerable to all sorts of machining and
handling damage and chipping.
10-43
veneer
(b)
Figure 10.32 Alumina central incisor that broke in vivo after 1 year. Fracture
started at the margin from cracking in both the veneer and core. The margin is
too thin, and even worse, the core portion is exceptionally thin. The origin is
chipping damage at the margin. (clinical case 2)
veneer core
(a) (c)
(b)
10-44
Case Studies
veneer core
(a) (b)
Figure 10.34 Alumina molar #37 that broke in vivo at 7 months in vivo. It
split in half from the mesial to distal side. Fracture started at the margin from
cracks and chipping in the veneer which extends to the very tip and even onto
the inside surface as evident in (a) Notice the tiny wake hackle lines in the
veneer in (a) that show the origin was at the margin. (a) also shows chipping
elsewhere on the margin. (clinical case 9).
(b)
Figure 10.35 Alumina molar that broke in vivo at 5.5 year in vivo. It split into
two pieces, but additional fractures were made during extraction. The initial
split was from mesial to distal side. The origin is at the margin of piece C.
The SEM image in (c) shows excess veneer that has come into the inside
surface and tiny wake hackle lines from the veneer bubbles(marked by the
arrow showing the dcp, and the exceedingly small core thickness at the margin
tip. Ceramic margins should never be made this thin. (clinical case 12).
10-45
This case is of two large elaborate components made of single crystal sapphire.
Sapphire is a harsh mistress. She has alluring physical and mechanical
properties such as good transmission of optical and infrared light, strengths of
1000 MPa or more, and good high temperature properties. She also has
vulnerabilities, such as high variability in strength, sensitivity to contact and
machining damage, and a shocking weakness at 300 oC - 500 o C due to
compression-induced twinning.
NASA is laying the groundwork for spacecraft of the future that could harness
solar energy. One concept is to use a large primary mirror solar collector that
focuses light to a secondary lens collector. The collector focuses and
concentrates energy by total internal reflection into a heat exchanger. 33 Two of
these full-scale (30 cm long by 9.5 mm diameter) c-axis sapphire lenses were
tested in the giant solar thermal vacuum facility that simulates outer space at
the Glenn Research Center. They fractured during high-temperature trials.
This fractographic analysis was done by Dr. J. Salem and myself.33
Figure 10.36 shows one of the fractured lenses, which has changed to a rose-
brown color from solarization at the 1300o C peak temperature of the test. The
fracture surface had a mix of cleavage and conchoidal fracture and some crack
bifurcations. The fracture origin was on the lens face. The origin shown in
Figure 10.37 was a large, partial Hertzian cone crack that extended from one of
two partially polished impact divots or “bruises” on the polished surface. From
the size of the ring crack, it was deduced that the impact object had a diameter
greater than 0.5 mm. Thermal stresses subsequently caused the lens to fracture.
This structure also had twins, but they were incidental and not a cause of
fracture. Figure 3.59b in the chapter on equipment shows one twin in a
polariscope.
Figure 10.38 shows the origin in the second broken refractor. It also broke
from an origin on the curved incident surface, again from thermal stresses, but
the origin was a scratch. Unusual “micro twins” were found near the origin
site, but they were not associated with the fracture.
Both of these refractors, and some of the earlier IR domes shown in this Guide
typify the vulnerability of sapphire to residual grinding cracks, polishing and
handling damage. The thermal gradients created both tensile and compression
stresses, and it is very likely that with additional high-temperature exposure and
cycling, more twins would have formed with the possibility they would have
intersected at some point leading to crack formation and eventually fracture.
10-46
Case Studies
(a)
(b)
10-47
(a) (b)
Figure 10.37 The origin of the first refractor was a blunt contact damage
crack shown on the fracture surface (a) and on the lens face in (b).
fracture plane
polished c
surface
origin
scratch
Fracture
(c) origin (d)
Figure 10.38 The second refractor lens also broke from damage on the
rounded surface, but from a scratch. (a) shows the fracture surface. (b) shows
the scratch on the outer polished surface. It curves at the origin site. With
careful lighting in (c) and (d), “micro-twins” near the origin site are revealed.
10-48
Case Studies
Chapter 10 References
10-49
16. G. D. Quinn, “Fractography and the Army Flexure Standard,” pp. 319 - 334 in
Fractography of Glasses and Ceramics, eds., J. Varner and V. Fréchette, American
Ceramic Society, Westerville Ohio, (1988).
17. G. D. Quinn, “Flexure Strength of Advanced Structural Ceramics: A Round Robin,”
J. Am. Ceram. Soc., 73 [8] (1990) 2374 - 2384.
18. W. N. Sharpe, Jr., O. Jadaan, N. N. Nemeth, G. M. Beheim, and G. D. Quinn,
“Fracture Strength of Silicon Carbide Microspecimens,” J. Microelectromech-
anical Systems, 14 [5] (2005) 903 - 913.
19. G. D. Quinn, W. Sharpe, Jr., G. Beheim, N. Nemeth, and O. Jadaan, “Fracture
Origins in Miniature Silicon Carbide Structures,” pp. 62 - 69 in Fractography of
Advanced Ceramics, II, eds., J. Dusza, R. Danzer, and R. Morrell, TransTech
Publ., Zurich, 2005.
20. G. D. Quinn, “Design and Reliability of Ceramics, Do Modelers, Designers and
Fractographers See the Same World?” Ceram. Eng. and Sci. Proc., 26 [2] (2005)
239 - 252.
21. J. J. Swab and G. D. Quinn, “The VAMAS Fractography Round Robin: A Piece of
the Fractography Puzzle,” pp. 55 -70 in Fractography of Glasses and Ceramics III,
Ceramic Transactions, Vol. 64, eds., J. R. Varner, V. D. Fréchette, and G. D. Quinn,
ACS, Westerville, OH, 1996.
22. MIL HBK 790, “Fractography and Characterization of Fracture Origins in Advanced
Structural Ceramics,” U. S. Army Materials Technology Laboratory, Watertown, MA,
June, 1992.
23. ASTM C 1322-96, “Standard Practice for Fractography and Characterization of
Fracture Origins in Advanced Ceramics,” Annual Book of Standards, Vol. 15.01,
ASTM Int., West Conshohocken, PA.
24. J. B. Quinn, G. D. Quinn, J. R. Kelly, and S. S. Scherrer, “Fractographic Analyses
of Three Ceramic Whole Crown Restoration Failures,” Dent. Mater., 21 (2005)
920 - 929.
25. S. S. Scherrer, J. B. Quinn, G. D. Quinn, and J. R. Kelly, “Failure Analysis of
Ceramic Clinical Cases Using Qualitative Fractography,” Int. J. Prosthodont., 19
[2] (2006) 151 - 158.
26. J. B. Quinn, G. D. Quinn, J. R. Kelly, and S. S. Scherrer, “Useful Tools for Dental
Failure Investigation,” pp. 36 - 53 in Proc. Conf. on Scientific Insights and Dental
Ceramics and Photopolymer Networks, ed. D. C. Starrett, Trans. Academy of
Dental Materials, Vol. 16, 2004.
27. J. R. Kelly, “Fractography of Dental Ceramics,” pp. 241 – 251 in Fractography of
Glasses and Ceramics, IV, eds., J. R. Varner and G. D. Quinn, American Ceramic
Society, Westerville, OH, 2001.
28. J. B. Quinn, “Failure Analysis of a Broken Tooth,” J. Fail. Anal. Prev., 4 [1]
(2004) 41 - 46.
29. J. Y. Thompson, K. J. Anusavice, A. Naman, and H. F. Morris,” Fracture Surface
Characterization of Clinically Failed All-Ceramic Crowns,” J. Dent. Res., 73 [12]
(1994) 1824 – 1832.
30. M. Øilo, A. K. Kvam, J. E. Tibballs, and N. Gjerdet, “Clinically Relevant Fracture
Testing of All-Ceramic Crowns,” Dent. Mater., 29 (2013) 815-823.
10-50
Case Studies
10-51
Conclusions
11. Conclusions
This Guide is intended to introduce engineers and
scientists to the science of fractographic analysis of
brittle materials as a tool to solve fracture
problems. This tool may be applied to a broad
range of practical problems including component
failure analyses, materials processing refinements,
routine materials characterization, laboratory scale
mechanical testing, and reliability and design.
Fractography is the examination of fracture
surfaces, the examination of the general crack
patterns, other part surfaces, other specimens, and a
review of processing, exposure, and testing
conditions. Much like the master detective, the
fractographer starts with an open mind, integrates
and analyzes the available information and relates Figure 11.1
it to similar episodes from his own experience and The fractographer as a
to what is published in the literature. The detective.
fractographer applies deductive logic to arrive at a
conclusion.
11-1
Fractography of Ceramics and Glasses
8. It’s better to let the material tell you why it failed than the other way
around. Theoretical modelers cause the most trouble. They already know why
something should fail since it is in their computer model. Indentation
mechanics people come next. They have more interest in indentations than in
the real flaws in a material.
11-2
Conclusions
Finally, the astute reader will have noted that fracture surfaces can be objects of
beauty. Figure 11.2 is one such example. This figure, from an unpublished
study by Dr. J. Quinn of dental porcelains with varying amounts of crystal-
lization and leucite content, illustrates many of the features described in this
book.
11-3
Appendix A Bibliography
APPENDIX A BIBLIOGRAPHY
Fractography, Vol. 12 ASM Handbook, ASM Int., Materials Park, OH, 1987.
The seminal work for metals fractography. Expensive. Includes a large atlas of
metal failures, but only a few examples for ceramics or glasses. Includes excellent
chapters on the history of fractography, metal models of fracture, photographic
equipment and techniques, optical and electron microscopy that are directly relevant
to ceramics and glasses. Caution: Some cleaning and preparation techniques such
as surface coatings, replicating tapes, replicating tape stripping, and aggressive
detergent cleaning prescribed for metals are not recommended for ceramic fracture
surfaces.
Failure Analysis and Prevention, Vol. 11, ASM Handbook, ASM Int., Materials Park,
OH, 2002.
Includes sections on fractography, and fractographic equipment and techniques, but
deals more with the broader issues of failure analysis: why a component fails to
perform its intended function. Fracture is but one of many failure modes. Includes
sections on failure analysis philosophies and approaches. Has one brief chapter on
ceramics. The fracture mechanics chapter is tailored towards metals fracture and not
ceramic or glass fracture. The quantitate fractography chapter by Gokahle is
excellent and has good information about quantitative microcopy and fractals and
their limitations.
A-1
Fractography of Ceramics and Glasses
Fractography of Glasses and Ceramics, Advances in Ceramics, Vol. 22, J. Varner, and
V. Fréchette, eds., American Ceramic Society, Westerville, OH, 1988.
Proceedings of the first of a quadrennial-quinquennial conference series by the same
name held at Alfred University in 1986. Sections on: fundamental phenomena, high-
temperature fracture, fractography and fracture mechanics, fractography in materials
development and testing, and component failures.
Fractography of Glasses and Ceramics, III, Ceramic Transactions, Vol. 64, J. Varner,
V. Fréchette, and G. D. Quinn, eds., American Ceramic Society, Westerville, OH,
1996.
Twenty-five papers from the third Alfred University conference, held on campus in
1995.
Fractography of Glasses and Ceramics, IV, Ceramic Transactions, Vol. 122, J. Varner,
and G. D. Quinn, eds., American Ceramic Society, Westerville, OH, 2001.
Thirty-two papers from the fourth Alfred University conference, held on campus in
2000. Special session on edge chipping.
A-2
Appendix A Bibliography
Fractography of Glasses and Ceramics, VI, Ceramic Transactions, Vol. 230, J. Varner
and M. Wightman, eds., American Ceramic Society and Wiley, Westerville, OH, 2012.
Twenty-five papers from the sixth Alfred university conference, held in 2011 this
time in Jacksonville, FL. Notable papers by Quinn on the history of fractography;
Glaesemann et al., on fractography in the development of ion-exchanged cover
glasses (e.g., Gorilla glass); Gross on crack evolution in alkali aluminosilicate glasses
for cover glass; Bradt on an energy criterion for mirror formation; Hecht-Mijic on
dental implants; Maurer on a glass pharmaceutical syringe fracture.
Fractography of Advanced Ceramics, ed., J. Dusza, Trans Tech Publ., Zurich, 2002.
First (2001) of a new conference series in Stará Lesná, Slovakia. Thirty-six papers,
many of which are on mechanical testing and have no fractography. Excellent papers
including a bioceramics fracture paper by H. Richter and a paper by R. Morrell and J.
Kübler on the background of the then new CEN prEN 843-6 fractography standard.
Fractography of Ceramic and Metal Failures, J. Mecholsky, Jr., and S. Powell, Jr.,
eds., ASTM STP 827, ASTM, Philadelphia, PA, 1984.
Eight papers on ceramics from a 1982 symposium at ASTM headquarters in
Philadelphia. Includes an outstanding, comprehensive, and well-illustrated review
paper by Rice, and papers by Pantano and Kelso, and Healy and Mecholsky (cited
below).
A-3
Fractography of Ceramics and Glasses
Microscopy Techniques
A-4
Appendix A Bibliography
R. C. Bradt, “The Fractography and Crack Patterns of Broken Glass,” J. Fail. Anal.
Prev., 11 (2011) 79-96.
A-5
Fractography of Ceramics and Glasses
J. J. Swab and G. D. Quinn, “The VAMAS Fractography Round Robin: A Piece of the
Fractography Puzzle,” pp. 55 - 70 in Fractography of Glasses and Ceramics III, eds. J.
Varner and G. D. Quinn, American Ceramic Society, Westerville, OH, Ceramic
Transactions, Vol. 64, (1996).
Seventeen laboratories analyzed photos of grinding cracks. They also analyzed
actual specimens with different flaws. Results varied. Much was learned by the
organizers from this exercise. How do fractographers sift through and use the clues?
Origins in Ceramics
H. Kirchner, R. Gruver, and W. Sotter, “Characteristics of Flaws at Fracture Origins
and Fracture Stress—Flaw Size Relations in Various Ceramics,” Mater. Sci. Eng., 22
(1976) 147–156.
A concise but useful report on strength-limiting origins in alumina, silicon nitride,
and silicon carbide with a detailed tabulation of different types of origins. Emphasis
is on porosity, large grains, and machining origins. An important observation (Fig.
1b) is that origins in the center of fracture mirrors may intersect the fracture surface
at an angle and a true view of the origin may not be seen.
A-6
Appendix A Bibliography
A-7
Fractography of Ceramics and Glasses
A-8
Appendix A Bibliography
J. E. Ritter, Jr., and R. W. Davidge, “Strength and Its Variability in Ceramics with
Particular Reference to Alumina,” J. Am. Ceram. Soc., 67 [6] (1984) 432-437.
A good study of strength variability in alumina with flaws deliberately introduced by
adding glucose or corn starch to starting powders, or by grit blasting surfaces. Flaw
linking was observed.
M. Hangl, “Processing Related Defects in Green and Sintered Ceramics,” pp. 133-156
in Fractography of Glasses and Ceramics, IV, Ceramic Transactions, Vol. 122, J.
Varner, and G. D. Quinn, eds., American Ceramic Society, Westerville, OH, 2001.
An excellent article about spray-dried agglomerates in varistor grade sintered zinc
oxide. Many photographs show the agglomerates from their initial state through
green body and final sintered stages. Agglomerates can be hollow or donut shaped.
The final fracture surface will reveal the spray-dried pattern if one looks carefully.
A-9
Fractography of Ceramics and Glasses
A-10
Appendix A Bibliography
An exceptional study that showed strength varied with the season of the year. The
variation was traced to subtle flaw variations in the powder compacts that were
caused by differences in the flowability of the spray-dried powders and their
deformability in winter and summer due to temperature and humidity differences in
the processing environment!
Pores and Their Relevance to the Strength in Alumina Ceramics,” J. Mater. Res., 14 [8]
(1999) 3370-3374.
New South Wales, Sydney, Australia, and the University of Aalen, Germany, April
2011
A remarkable study under the guidance of Profs. Mark Hoffman and Gerhard
Schneider, comparing flaw size distributions measured with automated image
analysis on polished sections and strength distributions and fractographic analysis in
alumina, alumina with seeded defects, dense and porous silicon carbide, and an
alumina-zirconia composite. A variety of flaw types were identified and
characterized. The size distributions matched an inverse power law function.
Origins in Glass
J. R. Varner, “The Practical Strength of Glass,” pp. 389 – 406 in Strength of Inorganic
Glass, ed. C. Kurkjian, Plenum, NY, 1986.
Fracture Mirrors
J. J. Mecholsky, Jr., S. W. Freiman, and R. W. Rice, “Fracture Surface Analysis of
Ceramics,” J. Mat. Sci., 11 (1976) 1310–1319.
A detailed correlation of origin size, fracture mirror sizes and characterization, and
fracture mechanics parameters for single and polycrystalline ceramics. A table of
mirror constants is given for a range of ceramics, and it is demonstrated that the outer
A-11
mirror (hackle) to origin size ratio is about 13 to 1. The inner mirror (mist) ratio is
between 6 to 1 and 10 to 1.
A-12
Appendix A Bibliography
G. D. Quinn and J. J. Swab, “Fractography and Estimates of Fracture Origin Size from
Fracture Mechanics,” Ceram. Eng. Sci. Proc., 17 [3] (1996) 51–58.
Fracture mechanics should be used routinely in fractographic analyses to verify that
the correct feature has been identified as the fracture origin. This paper reviews the
factors that may cause calculated and measured flaw sizes to differ.
H. Tada, P. C. Paris, and G. R. Irwin, The Stress Analysis of Cracks Handbook, 3rd
edition, ASM International, Metals Park, OH, 2000.
An update of their 1973 collection of stress intensity factors for various cracks under
different loading conditions.
T. Fett, and D. Munz, Stress Intensity Factors and Weight Functions, Wessex Institute
of Technology, Southhampton, UK, 1997.
A collection of stress intensity factors for various cracks under different loading
conditions.
A-13
F. Baratta, “Stress Intensity Factor Estimates for a Peripherally Cracked Spherical Void
and a Hemispherical Surface Pit,” J. Am. Ceram. Soc., 61 [11-12] (1978) 490-492.
D. J. Green, “Stress Intensity Factor Estimates for Annular Cracks at Spherical Voids,”
J. Am. Ceram. Soc., 63 [5-6] (1980) 342-344.
T. Fett, “Stress Intensity Factors and Weight Functions for a Void with an Annular
Crack,” Int. J. Fract., 67 (1994) R41-R47.
A-14
S. T. Gulati, R. Akcakaya, and J. Varner, “Fracture Behavior of Tin vs. Air Side of
Glass,” pp. 317 – 325 in Fractography of Glasses and Ceramics, IV, eds., J. R. Varner
and G. D. Quinn, American Ceramic Society, Westerville, OH, 2001.
T. Cleary and S. Gulati, “Influence of Glass Score and Seam Orientation on Edge
Strength of Multi-Layered Glass Articles,” pp. 327 – 341 in Fractography of Glasses
and Ceramics, IV, eds. J. R. Varner and G. D. Quinn, American Ceramic Society,
Westerville, OH, 2001.
Fracture patterns in laminated safety glass (sandwich of two glass plates and a
central plastic sheet).
B-1
Fractography of Ceramics and Glasses
R. A. Allaire and T. Ono, “Fracture Analysis of the Glass Scoring Process,” pp. 467 –
471 in Fractography of Glasses and Ceramics, IV, eds. J. R. Varner and G. D. Quinn,
American Ceramic Society, Westerville, OH, 2001.
Fracture patterns on scored glass plates.
B-2
Appendix B Case Studies
J. E. Marion, “Fracture of Solid State Laser Slabs,” J. Appl. Phys., 60 [1] (1986) 69 –
77.
B-3
Fractography of Ceramics and Glasses
F. W. Preston, “Bottle Breakage – Causes and Types of Fractures,” Am. Ceram. Soc.
Bull., 18 [2] (1939) 35 – 60.
An early paper, but with an exceptional treatment of a variety of failure causes.
R. E. Mould, “The Behavior of Glass bottles Under Impact,” J. Amer. Ceram. Soc., 35
[9] (1952) 230 – 235.
A well illustrated paper on controlled breakages of bottles. This article (and
Dimmick’s that followed it) proved that hinge origins away from the impact site
were sources of breakage in some cases.
C. Colin and J. L. Heitz, “Failure Analysis: Glass Bottle Breakage,” Verre, 16 [4]
(2010) 33-36.
Sparkling wine bottle broke spontaneously in boxes in a warehouse. Glass-to-glass
contact damage from improper packing caused cracks that grew to failure.
Fluorescent lamps
D. Johnson, “Arc Induced Fractures from Lampmaking Test Equipment,” pp. 517 –
525 in Fractography of Glasses and Ceramics, III, Ceramic Transactions, Vol. 64, J.
R. Varner, V. D., Fréchette, and G. D. Quinn, eds., American Ceramic Society,
Westerville, OH, 1996.
B-4
Appendix B Case Studies
B-5
Fractography of Ceramics and Glasses
Ceramic Seals
B-6
Appendix B Case Studies
B-7
Fractography of Ceramics and Glasses
D. G. Christie and H. Kolsky, “The Fractures Produced in Glass and Plastics by the
Passage of Stress Waves,” J. Soc. Glass Technol., 36 (1952) 65 - 73.
Y. Wang and M. Hadfield, “The Influence of Ring Crack Location on the Rolling
Contact Fatigue Failure of Lubricated Silicon Nitride: Experimental Studies,” Wear,
243 (2000) 157 – 166.
B-8
Appendix B Case Studies
MEMS structures
S. J. Glass, D. A. LaVan, T. E. Buchheit, and K. Jackson, “Strength Testing and
Fractography of MEMS Materials,” pp. 227 –240 in Fractography of Glasses and
Ceramics, IV, eds. J. R. Varner and G. D. Quinn, American Ceramic Society,
Westerville, OH, 2001.
Theta Specimens
G. D. Quinn, “Fractographic Analysis of Very Small Theta Specimens,” pp. 201-208 in
Fractography of Advanced Ceramics III, eds. J. Dusza, R. Danzer, R. Morrell, and G.
D. Quinn, Trans Tech Publ., Zurich, 2009.
B-9
Fractography of Ceramics and Glasses
R. Danzer, M. Lengauer, D. Rubesa, and W. Harrer, “Silicon Nitride Tools for Hot
Rolling of High-alloyed Steel and Superalloy Wires,” pp. 43 – 54 in Fractography of
Advanced Ceramics III, eds. J. Dusza, R. Danzer, R. Morrell, and G. D. Quinn, Trans
Tech Publ. Zurich, 2009.
B-10
Electronic Ceramics
Silicon wafers
R. E. Moore, P. G. Hansen, W. Carty, and J. W. Ha, “Fracture of Thermally Shocked
Silicon Disks,” pp. 351 – 361 in Fractography of Glasses and Ceramics, eds. J. D.
Varner and V. D. Fréchette, Advances in Ceramics, Vol. 22, American Ceramic
Society, Westerville, OH, 1988.
B-11
ZnO varistors
Z. Wang, P. Supancic, F. Aldrian, A. Schriener and R Danzer, “Strength and
Fractography of High Power Varistors,” pp. 358 - 365 in Fractography of Advanced
Ceramics II, eds. J. Dusza, R. Danzer and R. Morrell, Transtech, Zurich, 2005.
Bioceramics
B-12
crack growth. The second case was an in vitro fracture that broke from cracks in
the bore created by contact with a Co-Cr stem.
J. B. Quinn, “Failure Analysis of a Broken Tooth,” J. Fail. Anal. Prev., 4 [1] (2004) 41
– 46.
A patient’s natural tooth fractured days after open-heart surgery. Was the
laryngoscopy by the anesthesiologist done correctly?
B-13
B-14
B-15
R.C. Bradt and R. L. Martens, “Shattering Glassware,” Bull. Amer. Ceram. Soc., 91 [7]
(2012) 33-38.
Consumerwares. Thermal shock fractures of heat-strengthened soda lime silica
versus borosilicate glass.
B-16
B-17
Fibers – Nonoptical
Fiber Composites
B-18
Diamond coatings
H. A. Hoff, K. A. Snail, A. A. Morrish, and J. E. Butler, “Fractography and Fracture
Mechanics of Combustion Grown Diamond Thin Films,” pp. 25 – 54 in Fractography
of Glasses and Ceramics II, eds. V. D. Fréchette and J. R. Varner, Ceramic
Transactions, Vol. 17, American Ceramic Society, Westerville, OH, 1991.
Whiteware glazes
V. D. Fréchette, Failure Analysis of Brittle Materials, Advances in Ceramics, Vol. 28,
American Ceramic Society, Westerville, OH, 1990, pp. 94 – 101.
Internal pressure, water hammer, thermal shock, and impact fractures.
B-19
Jade
D. J. Rowcliffe and V. Frühauf, “The Fracture of Jade,” J. Mat. Sci., 12 (1977) 35 – 42.
Gemstones
J. E. Field, “Brittle Fracture: Its Study and Application,” Contemp. Phys., 12 [1] (1971)
1 - 31.
Rocks, Lithic Materials
B-20
A. I. Younes and T. Engelder, “Fringe Cracks: Key Structures for the Interpretation of
the Progressive Alleghanian Deformation of the Appalachian Plateau,” Geolog. Soc. of
Amer. Bull., 111 [2] (1999) 219 – 239.
B-21
The following table includes all the published values of mirror and branching
constants that the author could find up to late 2015 when the second edition of
this Guide was in preparation. The original reference is listed for each entry.
σ R = A (C.1)
Mirror-mist (or inner mirror) Ai and mist-hackle (or outer mirror) Ao constants
are listed for glass and glass ceramics. Mist is difficult or impossible to discern
in most ceramics and the mirror constant Ao that is listed corresponds to the
mirror-hackle boundary.
Entries where a modified version of equation C.1 was used are noted by a
footnote.
No judgment is made of the veracity of the listed values. Some of the values
may be inaccurate. Even for ideal materials such as fused silica and soda lime
silica there are unsettling variations in reported values. The variations
probably are caused by:
One is tempted to conclude that most of the discrepancies are due to viewer
judgment, but this cannot account for the all of the variability. The mirror
C-1
Fractography of Ceramics and Glasses
constant is only sensitive to the square root of the measured radii. In other
words, a 10% systematic error in radii measurements causes only a 5% error in
the mirror constant.
All values in the table are listed with the same number of significant figures as
in the original reference, but only up to a limit of three significant figures.
Uncertainties (± one standard deviation) are listed when available from the
original reference. Multiple entries in a cell denote estimates by different
microscopy techniques or analysis. For polycrystalline ceramics, the mirror
constants taken from the reference sources are assumed to be for the mist-
hackle boundary unless otherwise stated.
If the user cannot find the mirror constant for a particular material, he or she
may consider values for comparable materials within the same class. For
example, if a particular glass is not listed, then use the value for a glass of
similar composition. For polycrystalline ceramics, microstructure plays an
important role in determining a mirror constant. Conscientious readers should
check the references listed for more details on a particular material.
It is hoped that the adoption of the Guidelines in Appendix D, and the adoption
of ASTM C 1678, Standard Practice for Fractographic Analysis of Fracture
Mirror Sizes in Ceramics and Glasses in 2007, will lead to improved
consistency and accuracy in future mirror size constant determinations.
.
C-2
Appendix C Mirror Constants
Mist-Hackle or Branching
Mirror-Mist
Mirror-Hackle Ab
Material Technique Ai Reference
Ao (MPa·
(MPa· m )
(MPa· m ) m)
Glasses
Flint (Kimble R6 soda
Flexure (Rods) 2.0 Kirchner, 8
lime)
Flint (Kimble R6 soda
Flexure (Rods) 1.9 Kirchner, 9
lime)
Flint (Kimble R6 soda
Flexure (Rods) 2.3 Kirchner, 10
lime)
Soda-Lime Silicate – Flexure (biaxial ring-on
2.09 Orr, 5
window glass ring, large)
Soda-Lime Silica
(window, sheet, and Flexure (Bars or laths) 2.05 Shand, 4
plate)
Soda-Lime Silicate – Flexure (biaxial ring-on-
2.09 Barsom, 62
float, plate, sheet ring and bend bars)
Soda-Lime Silicate – Pressurized windows,
1.96 Reed, 39
window glass large
Soda-Lime Silicate Flexure (Bars, 3-point) 1.74 Bansal, 14
Soda-Lime Silicate A
Flexure (Bars – large)
– plate glass 1.86 ± 0.66 Kerper, Scuderi
Room Temperature to
Soda-Lime Silicate B 1.82 ± 0.91 6
Strain Point
– plate glass
Soda-Lime Silica 1.80 ± 0.15 * 2.42 ± 0.16 *
Flexure (Bars) Ball, 40
Float 1.81 ± 0.25 2.29 ± 0.24
Soda-Lime Silica Tension (Rods) 1.89 ± 0.06 2.04 ± 0.06 Johnson,
Float (G.E.C. – X8) Flexure (Bars) 2.09 Holloway, 1
Soda-Lime Silica Flexure (Bars) 1.92 2.21 Mecholsky,
Float Flexure–Delayed failure 18,21
(Bars) 2.0 ± 0.1 2.2 ± 0.1 Mecholsky, 21
Soda-Lime Silica Flexure 1.8 2.0 2.3 Mecholsky, 22
Flexure (Bars, large and
Soda Lime Silica Float small, 3-point and 4- 2.06 ± 0.07 2.29 ± 0.19 Duckworth, 16
point)
Congleton,
Soda-Lime Silica Tension (Plates) 1.2 – 1.6
Petch, 2
3.54 ±
Flexure (Bars)
0.64 Choi, 33
Soda-Lime Silica Flexure (Biaxial ring-on-
ring plates) 1.81 ± 0.28
Soda-Lime Silica Tension 1.9 Clark, 34
Soda-Lime Silica Pressurized Tube 2.0 Aoki, 35
Flexure (Biaxial ring-on-
ring disks) 2.28 –
Soda-Lime Silica 1.82 - 1.94 2.03 – 2.13 Marshall, 19
3 environments, 2.42
Vickers indented
Flexure (Biaxial, 2.1 ± 0.1
Soda-Lime Silica Shetty, 15
pressure on ring disks) NZ
(10 ± 4)
Soda-Lime Silica Flexure (rods) 1.8 Terao, 63
1.85 ± 0.21
Soda-Lime Silica Flexure (rods) 3.20 ⊕ Varner, 68, 70
2.82 ⊕
Flexure (3-point rods) 3.15 ± 0.11
Soda-Lime Silica Flexure (4-point rods) 2.74 ± 0.15 Abdel-Latif, 49
Tension (Fibers) 1.72 ± 0.28
Soda-Lime Silica Flexure (3-point, laths) (2.9 ***) Levengood, 64
Soda-Lime Silica Flexure, plates, 2 sizes 1.47 NZ (19.5) 2.88 NZ (18.) Gaume. 73
1.92 NZ (7.0) 3.1 NZ (5.7)
C-3
Fractography of Ceramics and Glasses
Mist-Hackle or Branching
Mirror-Mist
Mirror-Hackle Ab
Material Technique Ai Reference
Ao (MPa·
(MPa· m )
(MPa· m ) m)
1.80 ± 0.07
(air, 20oC)
2.08 ± 0.04
(air, -150OC)
Soda-Lime Silica Flexure (4-point, laths) Shinkai, 69
1.89 ± 0.05
(water, 17oC)
1.92 ± 0.04
(dry N2,24oC)
Borosilicate A
Flexure (Bars – large;
(P 3235) 1.98 ± 0.46 Kerper, Scuderi,
Room Temperature to
Borosilicate B 2.04 ± 0.75 6
Strain Point)
(C 7740)
Borosilicate Flexure (Rods, many Kerper, Scuderi,
2.08 ± 0.02
(C 7740) diameters) 7
Borosilicate Mecholsky,
Flexure (Bars) 1.87 ± 0.3 2.10
(C 7740) 17,18
Borosilicate Flexure (Bars) and
1.9 ± 0.3 Mecholsky, 20
(C 7740) Biaxial disks
Borosilicate
Flexure (Rods) 2.2, 2.35 Shand, 3
(C 7740)
1.98 ± 0.09
(air, 20oC)
2.08 ± 0.11
Borosilicate (air, -150OC)
Flexure (4-point, laths) Shinkai, 69
Pyrex 2.02 ± 0.08
(water, 17oC)
1.98 ± 0.12
(dry N2,24oC)
Flexure and Tension
Borosilicate 1.9* Alarcón et al. 56
(Rods)
2.28 ±0.03
Borosilicate crown Flexure (Biaxial, ring on 1.98 ± 0.02 NZ
2.11 ±0.03 NZ
NZ Quinn, 25
(Schott BK-7) ring disks)* 2.3
Aluminosilicate
Flexure (Bars) 2.14 2.40 Mecholsky, 18
(C 1723)
Aluminosilicate A
Flexure (Bars –large)
(P 6695) 2.31 ± 0.76 Kerper, Scuderi,
Room Temperature to
Aluminosilicate B 2.34 ± 0.97 6
Strain Point
(C 1723)
Alkali-borosilicate Tension (Fibers) 1.33 * Jaras et al., 12
Alkaline-earth
Flexure (Biaxial ring-on-
boro aluminosilicate 2.07 ± 0.01 Gulati et al., 57
ring)
(C1737)
Barium silicate
NR (Flexure Bars?) 1.3 1.5 Mecholsky, 22
3BaO-5SiO2
E glass
CaO-Al2O3-B2O3 Tension (Fibers) 1.47 * Jaras et al., 12
alumina borosilicate
Lead silicate Johnson,
Tension (Rods) 1.71 ± 0.06
(G.E.C. L1) Holloway, 1
Lead silicate Flexure (Bars) 1.61 1.78 Mecholsky, 18
Leaded silicate,
Flexure (Rods) 1.4 ± 0.1 1.6 ± (0,1) Salem et al., 77
6.7 g/cm3
Lithium silicate
Flexure Bars (3-point) 2.2 2.6 Mecholsky, 22
Li2O-2SiO2
Zinc silicate Flexure Bars 2.1 2.6 3.2 Mecholsky, 22
Zirconia silicate
Tension (Fibers) 2.37 * Jaras et al., 12
(Cem-FIL AR)
Yttrium-Alumino-Silica
Oxyntride Flexure (Bars) 1.5 ± 0.2 Coon, 48
(2Y-Al-Si-O-N)
Fused Silica Mecholsky, 18,
Flexure (Bars) 2.23 2.42 2.7
(C 7940) 22
C-4
Mist-Hackle or Branching
Mirror-Mist
Mirror-Hackle Ab
Material Technique Ai Reference
Ao (MPa·
(MPa· m )
(MPa· m ) m)
Fused Silica
Flexure (Bars – large;
(C 7940) Kerper, Scuderi,
Room Temperature to 1.89 ± 0.51
6
Strain Point)
Fused silica Johnson,
Tension (Rods) 2.33 ± 0.06
(Vitreosil) Holloway, 1
Fused silica Flexure (Rods) 2.20 ± 0.33 Choi, 33
Flexure and Tension
Fused silica 1.2 NZ Alarcón et al. 56
(Rods)
Fused silica fibers Tension 2.10 Choi, 33
Fused silica clad
Tension 1.96 ± 0.13 Baker, 29
fibers
Fused silica fibers Tension 1.83 Castilone, 30
Tension (Fibers)
Fused silica fibers, 2.2 ± 0.5
Flexure (Bars)
bars, disks 2.3 ± 0.5 Mecholsky, 20
Flexure (Biaxial, piston
on 3 balls) 2.4 ± 0.3
Fused silica fibers Tension 2.224 Chandan, 37
Leached High Silica
Flexure (Bars) 0.91 1.19 Mecholsky, 18
(C 7930)
Flexure (Bars – large;
96% Silica 1.84 ± 0.65 Kerper, Scuderi,
Room Temperature to
(C 7900) 6
Strain Point)
1.77 ± 0.06
(air, 20oC)
2.14 ± 0.06
96% Silica (air, -150OC)
Flexure (4-point, laths) Shinkai, 69
Vycor 1.80 ± 0.05
(water, 17oC)
1.83 ± 0.07
(dry N2, 24oC)
Mecholsky,
Glassy Carbon Flexure (Bars) 1.17 1.67
17,18
Bullock, Kaae,
Glassy Carbon Flexure (Bars) 2.1
61
Chalcogenide 1.58
Flexure (Fibers, 2 point
As-S-Se and 1.77 ± 0.14 NZ Quinn et al., 75
bending)
Ge-AS-Se-Te ( - 66.4 ± 45.5)
As2Se3 chalcogenide
glass, untreated 0.69 0.77
Tension (Fibers) Hulderman, 45
As2Se3 chalcogenide 0.35 irradiated 0.38 irradiated
glass, UV irradiated
As2S3 Flexure (Bars) 0.56 0.65 Mecholsky, 18
Ge33As12Se55 Flexure (Bars) 0.55 0.65 Mecholsky, 18
0.3PbSe -
Flexure (Bars) 0.48 0.55 Mecholsky, 18
0.7Ge1.5As0.5Se3
ZBLA (halide glass) Flexure (Bars) 0.8 Mecholsky, 22
Lead Borate glass
30PbO-70B2O3 (mol) 1.7 ± .05
40PbO-60B2O3 1.45 ± .05 Shinkai,
Flexure (Bars)
50PbO-50B2O3 1.15 ± .05 Ishikawa, 60 **
60PbO-40B2O3 0.85 ± .01
70PbO-30B2O3 0.65 ±.01
Glass Ceramics
Pyroceram 9608 Adams +
NR 2.8
(Li, Mg, Al silicate) DeMartino, 27
Pyroceram 9607 Adams +
NR 2.1 NZ
(Li, Mg, Zn, Al silicate) DeMartino, 27
Pyroceram 9606
(Cordierite, Mg, Al Flexure 3.6 6.5 Mecholsky, 17
silicate)
C-5
Fractography of Ceramics and Glasses
Mist-Hackle or Branching
Mirror-Mist
Mirror-Hackle Ab
Material Technique Ai Reference
Ao (MPa·
(MPa· m )
(MPa· m ) m)
Pyroceram 9606
(Cordierite, Mg, Al
Flexure (Bars) 6.5 Lewis, 44
silicate)
Pyroceram 9606
Flexure (Bars, 3-point
(Cordierite, Mg, Al 5.7 Bansal, 13,14
and 4-point)
silicate)
Pyroceram 9606 Flexure (Bars) and
(Cordierite, Mg, Al Flexure (Biaxial, piston 6.3 Mecholsky, 20
silicate) on 3 balls)
Pyroceram 9606 Flexure (Biaxial,
3.1 ± 0.2
(Cordierite, Mg, Al pressure on ring, and NZ Shetty, 15
(84 ± 6)
silicate) ball on ring)
Pyroceram 9606
(Cordierite, Mg, Al Flexure (Rods) 4.8 Shand, 3
silicate)
Li2O-SiO2 (NPL glass
Flexure (Bars) 3.3, 3.8 4.5, 5.4 Mecholsky, 17
ceramic, 2 grades)
Alumina
β-Al2O3 Flexure ~ 6.5 Mecholsky, 17
Hot-pressed Flexure (Rods, 4-point)
10.4 Kirchner, 11
(1 µm, Cer. Fin.)) 25 C to 1400 C
Flexure (Rods, 4-point)
Hot-pressed 10.3 Kirchner, 8,9
Flexure-Delayed
(99+% pure, Cer. Fin.) 9.9 Kirchner, 8
Fracture (Rods)
Hot-pressed
Flexure (Rods-3-point) 9.1 Kirchner, 10
(99+% pure,Cer. Fin.)
Hot-pressed
Flexure 5.2 12 Mecholsky, 17
(99+% pure)
Hot pressed Flexure (Rods) 10.4 Mecholsky, 11
Hot-pressed Flexure (4-point, 2
9.8 Bansal, 14
(Avco, 1-2 µm) sizes)
Sintered Tension (Plates) 7.3 Congleton,
(Lucalox) Petch, 2
C-6
Mist-Hackle or Branching
Mirror-Mist
Mirror-Hackle Ab
Material Technique Ai Reference
Ao (MPa·
(MPa· m )
(MPa· m ) m)
Sintered (96%)
Flexure (Rods) 8.5 Kirchner, 9
(Alsimag 614)
Flexure (Rods)
Sintered (96%) 8.3
Flexure-Delayed Kirchner, 8
(Alsimag 614) 8.9
Fracture (Rods)
Sintered (96%) Flexure (Bars, 3-point
9.0 Bansal, 13,14
(Alsimag 614) and 4-point, 2 sizes)
Sintered (96%)
Flexure (Rods) 9.1 Kirchner, 10
(Alsimag 614)
Sintered (96%)
Flexure (Bars) 13.1 Mecholsky, 17
(Alsimag 614)
Flexure (Bars),
Flexure (Biaxial ball on
Sintered (96%) 7.4 ± 0.6?
ring), 7.6 ± 0.5 Choi, 33
(Alsimag 614)
Flexure (Biaxial ring on
7.2 ± 0.7?
ring)
Sintered 96% Flexure (Biaxial, 4.0 ± 0.3
NZ (164 ± Shetty, 15
(Alsimag 614) pressure on ring disks)
11)
Hot isopressed
Flexure (Bars, 4-point) 7.7 Hecht-Mijic, 76
Biolox forte
Hot isopressed
Biolox delta Flexure (Bars, 4-point) 9.5 Hecht-Mijic, 76
ZTA
Silicon Carbide
Sintered SiC
Flexure (Bars) 5.39 Quinn, 23
(Hexoloy SA)
Sintered SiC Flexure (Biaxial, ring on
6.3 ± 0.5 Salem, 31
(Hexoloy SA) ring plates)
Sintered SiC Flexure (Biaxial, ring-on-
5.5 ± 0.3 Choi, 33
(Hexoloy SA) ring plates)
Sintered SiC
C-ring 5.50 ± 0.94 8.20 ± 0.54 10.5 ± 1.0 Conway et al. 67
(Hexoloy SA)
Sintered SiC
Tension (Rods) and
toughened 7.0 7.0? Srinivasan, 42
Flexure (Bars)
(Hexoloy SX)
Sintered with Al
C-ring 5.67 ± 0.60 8.16 ± 0.54 11.6 ± 0.8 Conway et al. 67
(Rioceram SiC)
Sintered Flexure (Bars, optical,
6.1, 6.8 Woodtli, 36
(Carolt S) SEM)
Hot-pressed SiC Flexure (Rods) 11.4
(NC-203) Flexure-Delayed Kirchner, 8
Fracture (Rods) 11.9
Hot-pressed SiC
Flexure (Rods) 11.5 Kirchner, 9
(NC-203)
Hot-pressed SiC
Flexure (Rods) 10.8 Kirchner, 10
(ACE)
Siliconized SiC (KT) Flexure 10.7 Mecholsky, 17
Reaction bonded SiC
C-ring 4.11 ± 0.31 5.22 ± 0.23 6.4 ± 0.4 Conway et al. 67
(Coors SCRB205)
Silicon Nitride
Sintered Reaction
Flexure (Rods) 8.47 ± 0.07
Bonded Quinn, 26
Flexure (Bars) 7.79 ± 0.02
(Ceralloy 147-31N)
Sintered (SSN-500
Flexure (Bars) 5.8 Quinn, 23
yttria/alumina)
Sintered Flexure (Biaxial, ring on 8.1 ± 2.4 Choi, 33
(SN 220) ring disks)
C-7
Fractography of Ceramics and Glasses
Mist-Hackle or Branching
Mirror-Mist
Mirror-Hackle Ab
Material Technique Ai Reference
Ao (MPa·
(MPa· m )
(MPa· m ) m)
Sintered Flexure (Biaxial, ring on
10.9 ± 2.7 Choi, 33
(AS 44) ring disks)
Hot-pressed
Flexure (Bars) 7.83 Quinn, 23
(Ceralloy 147A)
Hot-pressed
Flexure (Rods) 9.2 Kirchner, 9
(NC-132)
Flexure (Rods)
Hot pressed 8.9
Flexure-Delayed Kirchner, 8
(NC-132) 9.2
Fracture (Rods)
Hot-pressed
Flexure (Rods) 14.3 Kirchner, 10
(NC-132)
Flexure (Bars, 1100OC) 9.4 ± 1.2
Hot-pressed
Flexure (Biaxial ring on Choi, 33
(NC-132)
ring, 1100OC) 7.9 ± 2.1?
Hot-pressed
Flexure 18.1 Mecholsky, 17
(HS-130)
Hot-pressed
Flexure (Rods) 9.1 Kirchner, 8
(HS-130)
Hot-isopressed
Flexure (Bars) 5.9 ± 0.14 Choi, 32,33
(NT 154)
Hot-isopressed +
Flexure (Bars) 6.6 ± 0.11 Choi, 32,33
30vol% SiC whiskers
Flexure (Biaxial ring-on- 10.3?
Hot-isopressed
ring) Choi, 33
(GN-10)
Tension (Rods) 11.8 ± 1.4
Reaction Bonded
Flexure (Bars) 3.89 Messier, 24
(NC 350)
Reaction Bonded
Flexure (Bars) 3.19 Larsen, 28
(NC 350)
Reaction Bonded
Flexure (Rods) 4.2 Kirchner, 8
(AME A25B)
Zirconia
Ytttria stabilized
Flexure (Bars) 9.95 Morrell, 43
(3Y-TZP)
Flexure (Bars)
Ytttria stabilized 8.6 ± 0.23 Quinn, 50
Flexure (Biaxial piston
(3Y-TZP) 9.6 ± 0.23
on 3 ball)
Yttria stabilized
Flexure (Bars) 10.7 Quinn, 58
(3Y-TZP, Lava)
Yttria stabilized Flexure (Biaxial, ring on
11.5 ± 1.5 Choi, 33
(3.5Y-TZP) ring disks)
Calcia partially
stabilized PSZ
As fired - quenched Flexure (Bars, 3-point) 9.9
As fired 15.5
Under aged, 1300oC 19.2 Kirchner et al, 80
Peak aged 22.1
Peak aged 25.3
Over aged -
Zircar
(Alfred-Union Carbide,
Flexure (Bars) 15.2 Mecholsky, 17
0.4 µm, 5-6 mol% Y
TZP)
Zyttrite (AFML, 10
Flexure (Bars) 7.4 Mecholsky, 17
µm, cubic)
Single crystals
Ammonium
diphosphate,
Flexure (Bars) 0.5 Mecholsky, 17
single crystal
C-8
Mist-Hackle or Branching
Mirror-Mist
Mirror-Hackle Ab
Material Technique Ai Reference
Ao (MPa·
(MPa· m )
(MPa· m ) m)
MgAl2AlO4 Spinel,
Flexure (Bars) 2.6 Mecholsky, 17
single crystal
Quartz
(112� 0) plane with
Tension (Plates) 1.64 Ball, Payne, 78
tension in [0001]
direction
MgO ? 4.3 Ball, Payne, 78
Sapphire (average of
Flexure 6.1 Mecholsky 17
several planes)
Sapphire
(Tyco filaments, Tension 5.5
Abdel-Latif, 38
c-axis parallel to fiber Flexure 10.0
axis)
Sapphire
(Ruby rods, c axis Flexure (Rods) 3.3 Abdel-Latif, 38
~60o off rod axis)
Sapphire
Filament, c-axis Tension, 800 to 1000oC 2.4 3.2 Rice, 51
parallel to fiber axis
7.3 ± 0.6
Sapphire, a plane Flexure (Bars, 4-point) 7.1 ± 0.3 Salem, 70
NZ
(11)
Sapphire ? 7.3 Ball, Payne 78
Sapphire
o
filament, c-axis Tension, 800 to 1000 C 2.4 3.2 Rice, 51
parallel to fiber axis
2.01 ± .14
{110} plane
<110> dir. Tsai and
Silicon, single crystal Flexure (Bars)
1.61 ± .14 Mecholsky, 72
{110} plane
<100> dir.
Silicon, single crystal ? 5.8 Ball, Payne, 78
Other
3BaO-SiO2 Flexure (Bars) 3.9 6.0 Mecholsky, 17
B4C hot-pressed Flexure (bars) 4.8 9.27 Mecholsky, 17
BaTiO2 (2 grades) Flexure (Bars) 5.0, 5.4 Mecholsky, 17
Chromium ? 4.5 Ball, Payne, 78
Mullite Flexure 6.1 Mecholsky, 17
MgO Flexure 9.6 Mecholsky, 17
Congleton,
MgO Tension (Plates) 4.3
Petch, 2
MgO, single crystal Flexure (Bars) 5 Mecholsky, 17
MgF2 (Kodak) Flexure (Bars) 1.8 3.1 Mecholsky, 17
MgF2 (Kodak, Flexure (Bars) and
4.4 Mecholsky, 20
IRTRAN 1) Biaxial Disks
MgAl2AlO4 Spinel Flexure (Bars) 4.0 7.8 Mecholsky, 17
MgAl2AlO4 Spinel,
Flexure (Bars) 2.6 Mecholsky, 17
single crystal
PMMA ? 8.5 * 8.5 * Ball, Payne, 78
Porcelain, aluminous,
electrical insulator Flexure, (Rods, 3-point) ~ 3.5 Carlström, 74
grades
PZT Flexure 1.7 3.7 Mecholsky, 17
Graphite (POCO) Flexure 3.32 Mecholsky, 17
SrZrO3 Flexure (Bars) 4.4 6.0 Mecholsky, 17
Steatite (Mg silicate Flexure (Rods) 4.8 Kirchner, 8,9
insulator, DC -144) 4.5 Kirchner, 10
C-9
Fractography of Ceramics and Glasses
Mist-Hackle or Branching
Mirror-Mist
Mirror-Hackle Ab
Material Technique Ai Reference
Ao (MPa·
(MPa· m )
(MPa· m ) m)
WC-Co 24.4 ± 5
8 wt % (H8 grade) 41.0 ± 2
12 wt % (H12) 36.8 ± 2
10 wt % (G10)
51.7 ± 10
15 wt % (G15)
Flexure (Bars, 4-point) 75.0 ± 3 Luyckx, 41
30 wt % (E30)
10 wt % (J10) 29.6 ± 2
15 wt % (WCS15) 58.8 ± 4
20 wt % (WCS20) 71.9 ± 2
25 wt % (WCS25) 86.4 ± 1
Flexure 8.2 NZ
(Biaxial, disks ⊗ (1154)
WC-Co 6% Ni
pressure-on-ring (POR) Salem, et al, 79
(Kennametal)
14. NZ ⊗
POR + ring-on-ring (329)
WC-Co
Flexure 3.6 NZ ⊗
2% Ni, 1% Co Salem, et al, 79
(Bars?- 4-point) (493)
(Cercom)
WC (no Co) Flexure 10 ⊗ Swab, 46
Yttrium Aluminum
Mezeix and
Garnet, polycrystal. Flexure (Bars) 2.15 ± 0.08
Green, 66
(2.2 µm)
Yttrium Aluminum
Mezeix and
Garnet, single crystal Flexure (Bars) 2.20 ± 0.06
Green, 66
(111)
Zircon Porcelain
Flexure (Rods) 4.0 Kirchner, 8,9
(Alsimag 475)
ZnSe Flexure (Bars) 1.7 Mecholsky, 17
Flexure (Biaxial, ring-on-
ZnSe (44 µm) 0.8, 1.0 # Salem, 65
ring disks)
Fibers
Alkali-borosilicate Tension (Fibers) 1.33 NZ Jaras et al., 12
Chalcogenide 1.58
Flexure (Fibers, 2 point
As-S-Se and 1.77 ± 0.14 NZ Quinn et al., 75
bending)
Ge-AS-Se-Te (-66.4 ± 45.5)
Chalcogenide 0.69 0.77 Hulderman et al.,
Tension (Fibers) (0.35 if UV (0.38 if UV
As2Se3 irradiated) irradiated)
76
E glass
CaO-Al2O3-B2O3 Tension (Fibers) 1.47 NZ Jaras et al., 12
alumina borosilicate
Fused silica fibers Tension 2.10 Choi, 33
Fused silica clad
Tension 1.96 ± 0.13 Baker, 29
fibers
Fused silica fibers Tension 1.83 Castilone, 30
Fused silica fibers Tension (Fibers) 2.2 ± 0.5 Mecholsky, 20
Fused silica fibers Tension 2.224 Chandan, 37
Si-Ti-C-O Fibers in 3-D SiC matrix
2.50 ± 0.09 Davies, 53
(Tyranno LoxM) fiber composite
Si-C-O fibers
(Nicalon)
Tension ~2 Sawyer, 54
Si-C-N-O fibers
Si-N-C-O fibers
Fibers in 2-D CVI SiC Eckel + Bradt,
Si-C-O, (Nicalon) 2.51
matrix composite 55
Sapphire
filament, c-axis Tension, 800 to 1000oC 2.4 3.2 Rice, 51
parallel to fiber axis
Zirconia silicate
Tension (Fibers) 2.37 NZ Jaras et al., 12
(Cem-FIL AR)
C-10
NR Not reported
NZ A non-zero intercept was detected on the graph of stress versus inverse square root radius. The intercept
value in MPa is listed in parenthesis. Mirror or branching constants calculated with non-zero intercepts
are usually different than those calculated with intercepts through the ordinate. Consult the original
reference for more information.
* It is not clear whether Ball and Payne, ref. 78 listed this as Ao for PMMA, or Ab. Their other listed glass
and polycrystalline ceramic Ab values are actually Ao values from Kirchner or Johnson and Holloway.
** Additional mirror constants for the lead borate glasses are available at liquid nitrogen temperature (-196
o
C) in the original reference.
*** Mostly likely an overestimate since stresses were not adjusted for failure location in the 3-point rod
specimens.
# Alternative analyses (besides a graph of stress versus inverse square root radius) are in the original
reference.
? These values (from the original references) for Ab cannot be correct since they are less than or equal Ao.
⊗ Large internal residual stresses have been reported in WC-Co materials, often lead to large non-zero
intercepts. Consult the original reference for guidance.
⊕ Tempered soda lime silica rods
REFERENCES
1. J. W. Johnson, and D. G. Holloway, “On the Shape and Size of the Fracture Zones
on Glass Fracture Surfaces,” Phil. Mag., 14 [130] (1966) 731 - 743.
2. J. Congleton, and N. J. Petch, “Crack Branching,” Phil. Mag., 16 [142] (1967) 749
- 760.
3. E. B. Shand, “Breaking Stress of Glass Determined from Dimensions of Fracture
Mirrors,” J. Amer. Ceram. Soc., 42 [10] (1959) 474 - 477.
4. E. B. Shand, “Breaking Stresses of Glass Determined from Fracture Surfaces,” The
Glass Industry, April 1967, 90-194.
5. L. Orr, “Practical Analysis of Fractures in Glass Windows,” Materials Research
and Standards, 12 [1] (1972) 21 - 23.
6. M. J. Kerper, and T. G. Scuderi, “Modulus of Rupture of Glass in Relation to
Fracture Pattern,” Bull. Amer. Ceram. Soc., 43 [9] (1964) 622 - 625.
7. M. J. Kerper, and T. G. Scuderi, “Relation of Fracture Stress to the Fracture
Pattern for Glass Rods of Various Diameters,” Bull. Am. Ceram. Soc., 45 [12]
(1966) 1065 - 1066.
8. H. P. Kirchner, R. M. Gruver, and W. A. Sotter, “Fracture Stress-Mirror Size
Relations for Polycrystalline Ceramics,” Phil. Mag., 33 [5] (1976) 775 - 780.
9. H. P. Kirchner, R. M. Gruver, and W. A. Sotter, “Use of Fracture Mirrors to
Interpret Impact Fracture in Brittle Materials,” J. Am. Ceram. Soc., 58 [5–6]
(1975)188 - 191.
10. H. P. Kirchner, and R. M. Gruver, “Fracture Mirrors in Polycrystalline Ceramics
and Glass,” pp. 309 – 321 in Fracture Mechanics of Ceramics 1, eds. R. C. Bradt,
D. P. H. Hasselman, and F. F. Lange, Plenum, NY, 1973.
11. H. P. Kirchner, and R. M. Gruver, “Fracture Mirrors in Alumina Ceramics,” Phil.
Mag., 27 [6] (1973)1433 - 1446.
12. A. C. Jaras, B. J. Norman, and S. C. Simmons, “The Measurement of Glass Fibre
Strength in Composites from Studies of Their Fracture Surfaces,” J. Mat. Sci., 18
(1983) 2459 - 2465.
C-11
Fractography of Ceramics and Glasses
C-12
Appendix C Mirror Constants
C-13
Fractography of Ceramics and Glasses
44. D. Lewis, III, “Fracture Strength and Mirror Size in a Commercial Glass-
Ceramic,” J. Am. Ceram. Soc., 64 [2] (1961) 82 - 86.
45. F. Hulderman, J. S. Sanghera, and J. D. Mackenzie, “The Effect of UV Radiation
on the Mechanical Strength of As2Se3 Glass Fibers,” J. Non-Cryst. Sol., 127 (1991)
312 - 322.
46. J. J. Swab, J. C. LaSalvia, and W. A. Gooch, “Characterization of WC-Based
Armor Ceramics,” unpubl. priv. com., U.S. Army Research Laboratory, Aberdeen
Proving Ground, MD, 21005.
47. J. R. Kelly, “Fractography of Dental Ceramics,” pp. 241-251 in Fractography of
Glasses and Ceramics, IV, eds. J. R. Varner and G. D. Quinn, Ceramic
Transactions Vol. 122, American Ceramic Society, Westerville, OH 43081, 2001.
48. D. N. Coon, “Mirror/Mist Radius Constant for Y-Al-Si-O-N Glass,” J. Mat. Sci.
Lettrs., 21 (2002) 1121-1122.
49. A. I. A. Abdel-Latif, R. C. Bradt, and R. E. Tressler, “Dynamics of Fracture Mirror
Boundary Formation in Glass,” Int. J. Fract., 13 [3] (1977) 349 - 359.
50. G. D. Quinn, J. Eichler, U. Eisele, and J. Rödel, “Fracture Mirrors in a Nanoscale
3Y-TZP,” J. Am. Ceram. Soc., 87 [3] (2004) 513 - 516.
51. R. W. Rice, “Corroboration and Extension of Analysis of c-Axis Sapphire
Filament Fractures from Pores,” J. Mat. Sci. Lttrs., 16 (1997) 202 - 205.
52. H. Fischer, J. Gröbel, and R. Marx, “Fraktograpische Schadensanalyse an
Dentalkeramik,” Deutsche Zahnarztl. Zeitschrifft, 55 [10] (2000) 667 - 670.
53. I. J. Davies, and T. Ishikawa, “Mirror Constant for Tyranno Silicon-Titanium-
Carbon-Oxygen Fibers Measured in Situ in a Three-Dimensional Woven Silicon
Carbide/Silicon Carbide Composite,” J. Am. Ceram. Soc., 85 [3] (2002) 691 -693.
54. L. C. Sawyer, M. Jamieson, D. Brikowski, M. I. Haider, and R. T. Chen, “Strength,
Structure, and Fracture Properties of Ceramic Fibers Produced from Polymeric
Precursors: I: Base-Line Studies,” J. Amer. Ceram. Soc., 70 [11] (1987) 798 - 810.
55. A. J. Eckel, and R. C. Bradt, “Strength Distribution of Reinforcing Fibers in a
Nicalon Fiber/Chemically Vapor Infiltrated Silicon Carbide Matrix Composite,” J.
Am. Ceram. Soc., 72 [3] (1989) 455 - 458.
56. O. E. Alarcón, R. E. Medrano, and P. P. Gillis, “Fracture of Glass in Tensile and
Bending Tests,” Metall. Mat. Trans A,, 25A, May (1994) 961 - 967.
57. S. T. Gulati, J. F. Bayne, W. R. Powell, and J. D. Helfinstine, “Mirror Constant for
AMLCD Glass,” Bull. Am. Ceram. Soc., 83 [5] (2004) 14 - 15.
58. J. B. Quinn, D. Cheng, R. Rusin, and D. Suttor, “Fractographic Analysis and
Material Properties of a Dental Zirconia,” International Association for Dental
Research (IADR) paper, 0560, Baltimore, MD, March, 2005.
59. J. B. Quinn and G. D. Quinn, “Materials Properties and Fractography of an Indirect
Dental Resin Composite,” Dent. Mater., 26 (2010) 589-599.
60. N. Shinkai, and H. Ishikawa, “Crack Branching of Binary PbO-B2O3 Glasses,” J.
Noncryst. Solids., 52 [1-3] (1982) 385 - 394.
61. R. E. Bullock, and J. L. Kaae, “Size Effect on the Strength of Glassy Carbon,” J.
Mat. Sci., 14 (1979) 920 - 930.
62. J. M. Barsom, “Fracture of Tempered Glass,” J. Am. Ceram. Soc., 51 [2] (1968) 75
- 78.
C-14
Appendix C Mirror Constants
C-15
Appendix D Guidelines for Fracture Mirrors
D.1 Introduction
“The position assigned to the boundary between mirror and mist zones
depends upon the illumination and the magnification at which the fracture
is examined, even within the range of the optical microscope. ….
However, under given conditions a reproducible position for the boundary
can be assigned.”
Since many of the early mirror measurements were made while viewing
through the optical microscope, it is safe to say that the first perception of a
mirror boundary was where the surface roughness was a fraction of the
wavelength of light (0.39 µm - 0.80 µm). Threshold levels of detectability
have been estimated to be as small as a few tens of nanometers to as much as
0.25 µm. Attempts to make objective determinations of the mirror boundary by
quantitative surface roughness characterization are discussed in section D.3
below, but a single threshold roughness value cannot be specified.
σ R=A (D.1)
D-1
Fractography of Ceramics and Glasses
where σ is the stress at the origin, R is the mirror radius, and A is the
appropriate mirror constant as described below. Leighton Orr recognized as
early as 1945 that stress varied inversely with the square-root dependence of
the mirror radius, so the author now refers to equation D.1 as “Orr’s equation.”
The mirror-mist boundary in glasses is the periphery where one can discern
the onset of mist. This boundary corresponds to Ai, the inner mirror constant.
The mist-hackle boundary in glasses is the periphery where one can discern
the onset of systematic hackle. This boundary corresponds to Ao, the outer
mirror constant.
If the mirror is being measured for a component failure analysis, and if the
mirror constant A is known, follow steps 1 through 9 in the next section and
compute the stress at the origin in accordance with equation D.1. Note that this
origin stress may or may not necessarily be the maximum stress in the part.
Examples of how to judge the boundaries are shown in Figures D.1 through
D.6. Low power images provide an overview. Matching unmarked and
marked images are shown so that the reader can make a judgment for
comparison. The boundaries were assessed while looking through a
microscope and the digital images in Figures D.1 to D.6 were marked. Mirrors
D-2
Appendix D Guidelines for Fracture Mirrors
(a)
hackle
mist
(b)
mirror
Figure D.1 Fused silica rod broken in flexure (122 MPa). The origin is a
surface flaw located at the bottom of the specimen (the insert in a) where the
stress was a maximum. A stress correction is unnecessary. The mirror-mist
boundary is small relative to the cross section size and is approximated by a
circle in (b). The inward cusps at the surface are ignored as discussed in
section D.3. The mist – hackle boundary is slightly elongated towards the top.
Close examination of the fracture surface in the vicinity of the flaw (not shown)
showed that fracture started from the deepest part of the flaw. (c) and (d) on
the next page show SEM images of the same mirror.
D-3
Fractography of Ceramics and Glasses
(c)
mist - hackle
(d)
mirror - mist
Figure D.1 continued (c) is an SEM image of the same mirror and at the
same magnification as (a) and (b). The mist is indistinct in the SEM image. (d)
is a composite of two SEM images showing the transition from mirror to mist to
hackle. The locations of the boundaries as assessed by optical microscopy (as
shown in (b))are marked by dashed lines on the SEM images.
D-4
Appendix D Guidelines for Fracture Mirrors
(a) (b)
(c)
(d)
Figure D.2 Silicon carbide tension strength specimen (371 MPa) with a
mirror centered on a compositional inhomogeneity flaw shown in Figure
6.20a,b. Note how clear the mirror is in the low power images. The mirror
boundary (arrows in d) is where systematic new hackle forms and there is an
obvious difference compared to the roughness inside the mirror region.
D-5
Fractography of Ceramics and Glasses
(a) (b)
(c)
(d)
2R
Figure D.3 Silicon nitride bend bar with a Knoop surface crack (449 MPa,
Ceralloy 147-31N, Refs. 6,7). The mirror is incomplete into the stress
gradient, but the mirror sides can be used to construct boundary arcs in (d).
Radii are measured in the direction of constant stress along the bottom.
D-6
Appendix D Guidelines for Fracture Mirrors
(a)
(b)
2R
D-7
Fractography of Ceramics and Glasses
(a) (b)
(c)
(d)
D-8
Appendix D Guidelines for Fracture Mirrors
(a) (b)
(c) (d)
Figure D.6 Fracture mirror in another 3Y-TZP bend bar. (3M ESPE Lava,
1059 MPa maximum stress, 1024 MPa at the origin). (a) and (b) are stereo
optical microscope images. (c) and (d) are SEM images. Images (b) and (c)
are at the same magnifications. The flaw is a pore. Fine hackle lines in the
mirror run right to the origin. They are created by the flaw itself and the
microstructure. The mirror boundary is where systematic new hackle forms
and there is an obvious difference compared to the roughness inside the mirror
region. This is a very difficult example. One interpretation is shown in the
lower magnification SEM image. (SEM images (c) and (d) courtesy J. Quinn)
D-9
Fractography of Ceramics and Glasses
Use the following steps to measure mirror sizes. These steps are repeated, but
with detailed notes, clarifications, and illustrations in section D.3.
4. Use a magnification such that the fracture mirror area occupies about
75 % to 90 % of the width of the field of view for glasses, and
approximately 33 % to 67 % of the width of the field of view for
ceramics. An additional, lower-power image may be helpful for ceramics.
5. Measure the mirror size while viewing the fracture surface with an
optical microscope whenever possible.
Use either calibrated reticules in the eyepieces or traversing stages with
micrometer-positioning heads. Alternatively, measurements may be made
on digital images on a high-resolution computer monitor, preferably while
the fracture surface can be simultaneously viewed through the microscope
eyepieces in order to aid judgment.
Measurements from photos may be used as a last resort if the steps above
cannot be followed. This may be necessary for very small specimens or
very strong specimens with tiny mirrors and a scanning electron
microscope must be used to photograph the mirror.
Measurements from other devices may be used provided that the criterion
used for identifying the mirror boundary is carefully documented.
D-10
7. Exercise caution when mirrors are large relative to the specimen cross-
section size.
8. Show at least one photo with arrows or lines marking the mirror size.
Use some judgment in the regression analysis since fracture mirror data
frequently has moderate scatter. If the data does not appear to fit a trend
that has a zero intercept, regress the data with a non-zero intercept and
equation D.3. Report the intercept if it deviates significantly (> 10 MPa)
from zero. Investigate possible residual stresses or specimen size or shape
issues if the intercept deviates significantly from zero.
A
σa = − σ r (D.3)
R
where σa is the stress at the origin from known applied stresses, σr is the
residual stress at the origin location. A negative σr is a compressive
residual stress and a positive σr is tensile, i.e., a positive intercept on the
graph corresponds to a compressive residual stress and a negative intercept
corresponds to a residual tensile stress.
D-11
12. Mirrors sizes should be collected over a broad range of sizes and
fracture stresses if possible. Data from different specimen types and
sizes may be combined.
(a) (b)
Figure D.7 Plot of applied stress σa (at the origin) versus 1/√R. (a) shows the
trend for residual stress-free parts. (b) shows it for parts with residual
stresses. Compressive residual stresses move the locus up with a positive
intercept σr, but with the same slope. Tensile residual stresses shift the data
downwards with a negative intercept (not shown).
D-12
smaller than those obtained with conventional viewing modes. During a review
process in 2010, paragraph 6.1.3 of standard ASTM C 16782 was revised to
allow DIC mode for glasses, provided that the polarizing elements are used in a
consistent manner in the grey mode (with the lambda plate for color control
removed).
Dark-field illumination may be used with glasses, but dark-field images may
lose a little resolution with glasses and radii may be slightly larger as a result.
It is very effective with highly-reflective mirror surfaces in single crystals.
D-13
features. Hull 11,12 and Wünsche et al. 13 used atomic force microscopy to show
that the mist region in glass or brittle epoxy has a roughness of as small as a
few tens of nanometers, which is much less than the wavelength of visible light
(390 nm to 800 nm). The scanned regions were quite small, however, and both
groups noted that if the AFM scans a small region between hackle or river line
steps, then the measured roughness is much less than if the latter are included.
Hull pointed out that the greater undulations from Wallner lines need to be
factored out when evaluating the intrinsic mist roughness. His study showed
that roughness increased continuously and there were no dramatic jumps in
roughness at the boundaries, but the rate of change of roughness did change
significantly at the mirror-mist boundary. Surface roughness measurements
should be taken perpendicular to the direction of crack propagation since high-
resolution transmission microscope images 14,15,16 of the mist and hackle show
the surface features are elongated in the direction of crack propagation.
The author is not aware of any systematic studies comparing fracture mirror
sizes as a function of the viewing mode or microscope type.
The requirement poses a small problem if the mirrors are examined with stereo
binocular microscopes. These have two different tilted optical paths. If
viewing with both eyes in a stereomicroscope, the specimen should be flat and
D-14
facing directly upwards. The observer’s brain will interpret the image as
though he is facing it directly.
4. Use a magnification such that the fracture mirror area occupies about
75 % to 90 % of the width of the field of view for glasses, and approx.-
imately 33 % to 67 % of the width of the field of view for ceramics. An
additional, lower-power image may be helpful for ceramics.
Observers usually mark the mirror boundaries closer to the origin at greater
magnifications than they would at lower magnifications. This is because mist or
micro hackle markings are easier to see at distances closer to the origin at high
magnification. This is particularly the case with glasses. Conversely, at very
low magnifications, much detail is lost and observers typically overestimate the
mirror size.
D-15
regions outside the mirror. “Stepping back” and using the 33 % to 67 % rule
should help an observer better detect the topography differences. Images
recorded at these magnifications are also more convincing when shown to other
fractographers or engineers. Even lower magnification images may also be
made to aid interpretation such as shown in Figures D-2b, D-3b, and D-6b.
The images should not be more than 5 times different in magnification;
otherwise it is difficult to correlate features in one image to another.
The word systematic requires some elaboration. Mirror boundary hackle lines
are velocity hackle lines created after the radiating crack reaches terminal
velocity. Premature, isolated hackle can in some instances be generated well
within a mirror, however. It should be disregarded when judging the mirror
boundary. Wake hackle from an isolated obstacle inside the mirror (such as a
large grain or agglomerate) can trigger early “premature” hackle lines such as
shown in Figure 5.5 in the main text. Steps in scratches or grinding flaws can
trigger hackle lines that emanate from the origin itself.
5. Measure the mirror size while viewing the fracture surface with an
optical microscope whenever possible.
A compound optical microscope is best for glasses. A stereoptical microscope
is best for ceramics. Use either calibrated reticules in the eyepieces or
traversing stages with micrometer-positioning heads. For routine
measurements for ceramics, the author uses a stereoptical binocular microscope
with a traversing X-Y stage with digital micrometer heads that read out to
0.001 mm (1 µm). Alternatively, measurements may be made on digital
images projected onto a high resolution computer monitor preferably while the
fracture surface can be simultaneously viewed through the microscope
eyepieces in order to aid judgment.
D-16
Measurements from photos may be necessary for very small specimens or very
strong specimens with tiny mirrors, such as in fibers or microelectro-
mechanical system (MEMS) devices. Scanning electron microscope images
may be used. Again, the fractographer should take an overall framing
photograph or image shot in accordance with the 75 % to 90 % rule for glasses
and 33 % to 67 % rule for ceramics. Higher and lower magnification images
may be used to help aid in interpretation.
Mirror size measurements from photographs are usually less accurate. They
frequently overestimate mirror sizes unless conditions are carefully optimized
to accentuate contrast and topographic detail. Two-dimensional photographic
renditions of a three-dimensional fracture surface usually lose much of the
topographic detail discernable by the eye with a compound optical or
stereomicroscope.
Mirror size measurements made on computer monitor screens are also subject
to inaccuracies, also because they are two-dimensional renditions of a three-
dimensional fracture surface. Video cameras should not be used to capture
mirror images since they lack adequate resolution. High-resolution cameras
and monitors are beginning to match the capabilities and accuracy of an
observer peering through the optical microscope.
The center of the mirror may not necessarily be the center of the flaw at the
origin. Careful inspection of tiny localized fracture surface markings (Wallner
lines and micro hackle lines) right around the origin may reveal that fracture
started at one spot on a flaw periphery. For example, fracture from grinding or
impact surface cracks in glass often starts from the deepest point of the flaw
and not at the specimen outer surface. Figure D.1 shows an example. Large
pores often trigger unstable fracture from one side. An example is the pore in
Figure D.6d, where fracture seems to have started on the left side of this
D-17
Rd Rd
Rd
D-18
glass ceramic
(a) (b)
Figure D.10 Grinding cracks and scratches can cause mirror elongations
along the surface, even in bend bars with stress gradients. (a) shows a
schematic of such a mirror with the mist-hackle boundary marked in glass, and
(b) shows a comparable image in a polycrystalline ceramic. It has some
intrinsic microstructural roughness inside the mirror and the mirror-hackle
boundary is marked. Use an average radius: Ravg = {(R1 + R2 + Rd}/3.
On the other hand, mirror shapes are commonly affected by stress gradients in a
plate or a beam. Mirror radii are elongated in the direction of decreasing stress.
Examples are shown in Figure D.9b and c. In such cases, measure the mirror
radius along the tensile surface where the stress is constant. Do not measure
the mirror radii into the gradient. Even with this precaution, there is
considerable evidence that the data begins to depart from the stress-mirror size
curves and the relationship in equation D.1 when the sizes approach the plate,
beam, or rod thickness. For mirrors radii larger than the plate thickness, data
points fall above the trend on a log stress – log mirror size plot which means
that mirrors are larger than they would otherwise be if the part were loaded in
uniform tension.
A trend for mirrors to elongate the opposite way, along the external surface of a
specimen, was detected by the author in recent work on the fractographic
analysis of grinding flaws in structural ceramics.6,7 Long surface cracks often
caused mirrors to have perceptible deviations from a semicircular shape as
shown in Figure D.10. In such cases, measure an average radius:
D-19
Rd
Rh Rh R2
R2
R1
R1
R1 + R2 R1 + R2
(a) (b)
(c) (d)
Figure D.11 Fracture mirrors in rods tested in flexure. The load was applied
in a vertical direction and the maximum tensile stresses are at the bottom
center. Fractures started at flaws part way up the sides of the rods causing the
mirrors to have unequal radii. Schematics (a and b) are similar to two glass
rods (c and d). The rod shown in (c) was sufficiently strong that a nearly
semicircular mirror formed, but with unequal radii due to the stress gradient.
Use R = Rh if the origin and mirror center is distinct. Otherwise use Ravg = (R1
+ R2 + Rd) / 3 if the mirror is nearly semicircular. Use Ravg = (R1 + R2) / 2 if
the mirror is elongated into the interior and Rd is large. (d) shows a weaker
glass rod. Use R = Rh if the origin and mirror center are distinct, otherwise
use Ravg = (R1 + R2) / 2.
D-20
For origins located in the interior of a rod broken in flexure, only use the radii
in the direction of constant stress. As a check, and only if the mirror is a
complete ellipse, one may measure two orthogonal mirror diameters and
compute an average diameter and halve this to obtain an average radius.
There is one important detail about mirror sizes that warrants discussion.
Mirrors located on a specimen external surface have small cusps at the
intersection with the outer surface as shown in Figure D.12a. Cusps are often
detected in glass mirrors, but they are rarely if ever discerned in polycrystalline
ceramic mirrors. The small cusp is a consequence of fracture mechanics. A
small element of material near the tip of a crack at the specimen exterior
surface experiences greater stress intensity than a similar element buried in the
interior whereby neighboring elements can “share the load.” Kirchner et al.4,5
discussed the shapes of fracture mirrors that intersect outer surfaces and they
showed that the local enhancement of the stress intensity KI accounts for the
cusps. The slightly-greater stress intensity at the surface triggers the mirror
markings a bit sooner than for interior elements. The usual convention, and the
one adopted in this Guide, is to truncate the cusp. Extend the semicircular (or
other mirror shape) arcs as shown in Figure D.12c. Another reason to be
wary of measurements right along the surface is that surface roughness,
machining damage, or other surface irregularities may trigger mist or hackle
formation a bit earlier than in the interior. Others have noted that
measurements taken right on an exterior surface are slightly different than those
D-21
(a)
(b)
cusp
(c)
Figure D.12 Fracture mirror in a fused silica rod (a), illustrating the cusps in
the mirror near the outer surface (b). Mirror measurements should not
include the inward bend of the mirror and may be made as shown in (c).
D-22
taken into the interior. Even Shand 18 recommended in 1959 that size
measurements be taken 0.1 mm (.004 in) beneath the exterior surface.
Mecholsky and Freiman 19 in 1979 recommended ignoring the cusp at the
surface on the basis of fracture mechanics considerations: “In measuring the
mirror-mist and mist-hackle boundaries, these should be projected to the tensile
surface to compete a circular arc, since there is curvature at the surface due to
free surface effects.”
A dilemma occurs when mirrors are large in plates or beams broken in flexure
as shown in Figure D.9c. Cusps cannot be detected. In this case the only
plausible way to measure a mirror radius is directly along the surface. In such
cases, the general warning of step 7 applies.
It is worth reiterating that equations D.1 and D.2 are empirical. Kirchner and
colleagues4,5 showed a more rigorous fracture mechanics analysis (based on a
critical stress intensity for branching) completely explained the mirror shapes
and distortions including the cusp mirrors in various stress fields. Despite the
merits of their rigorous fracture mechanics analysis, there is significant
practical advantage to using the simpler mirror size - stress relationships D.1
and D.2 even if some rigorousness is sacrificed. In addition, for small mirror
sizes relative to the specimen dimensions, the simpler approach is almost as
rigorous.
Figure D.13 Surface residual stresses also may alter a mirror shape. σa is
the applied stress to cause fracture and σf is the fracture stress in an annealed
plate. (a) shows a surface mirror in an annealed plate. (b) shows the mirror
shape in a plate with surface compression stresses that decrease into the
interior and become tensile. (c) shows a mirror in a plate with surface tensile
stresses that diminish into the interior and become compressive.
D-23
Residual stresses alter mirror shapes. If the mirror is very small relative to the
stress gradient, the mirror shape may remain circular or semicircular if along
the surface. On the other hand, if the mirror is larger or the stress gradient is
steep, then the gradient alters the mirror shape as shown in Figure D.13. Figure
D.13a shows an annealed plate that requires an applied stress of σa = σf to
cause fracture. Figure D.13b shows the case where the same plate has
residual surface compression stress σr = σc from ion exchange or thermal
tempering, so that an applied stress to cause fracture is σa = σf + σc. In other
words, the applied stress must be increased to overcome the residual surface
stress. Nevertheless, the net stress at the surface at the moment of fracture is
σ = σa - σc = (σf + σc) - σc = σf, the same stress as in the annealed plate. Hence
the mirror radii along the surface are unchanged compared to the annealed
plate. In contrast, in the direction into the interior, tensile stresses combine
with the applied tensile stress to cause the mirror markings to form sooner, at a
shorter radius into the interior than in the annealed plate. In this example, the
mirror shape is flattened to a semiellipse. Mirror radii should be measured only
along the surface (or just beneath the surface if there is a cusp) in these cases.
Figure D.13c shows that surface residual tensile stresses have the opposite
effect: mirror radii are elongated into the interior. Mirror radii again should
only be measured along the surface, since again the net stress to cause fracture
is σ = σf.
There are two possible paths for data analysis if there are residual stresses:
7. Exercise caution when mirrors are large relative to the specimen cross-
section size.
At some point, one can expect departures from of the stress - mirror size
relationship. The point where the departure occurs depends upon the loading
geometry and the stress state.
D-24
Pronounced deviations occur once the mirror size approaches or is greater than
the component thickness in plate or beam bending fractures. Experimentally
measured radii are much greater than predicted by equation D.1. Shand
recommended that the maximum mirror size should be no more than 15% of
the rod diameter (or 30% of the distance to the centroid) for
flexure tests.18,20 Kirchner and Conway5 warned about limitations in the
fracture mechanics models for mirror radii exceeding 20 % of a rod diameter
tested in flexure. On the other hand, Castilone et al.17 had success with mirrors
that were almost as large as the fiber diameters for fibers tested in tension.
There is merit to measuring and recording mirrors even if they are large relative
to a cross section size. The data may have value for use with analyzing genuine
component fractures.
Mecholsky and Freiman19 warned that systematic deviations from the mirror
size relationship occur at large mirror sizes but also at very small sizes, the
latter due to internal stress effects, e.g., from thermal expansion anisotropy of
grains in ceramics.
8 Show at least one photo with arrows or lines marking the mirror size.
This simple step will help clear up a lot of the doubt and confusion about what
an investigator has actually measured.
10. Use the stress at the origin site (additional steps to determine A).
If the specimen was broken in controlled conditions where the stress
distribution was known (e.g., beams, rods, or plates in flexural loadings) correct
the stress for location in specimens with stress gradients.
D-25
stresses that acted on and in the immediate vicinity of the flaw origin. While
this may seem obvious, it is probable that some analysts in the past have
erroneously used nominal stresses in a specimen or component rather than the
actual stress that was acting upon the mirror region in a body. In contrast,
some researchers have correlated the stress at every site along the mirror
periphery with the mirror radius at that periphery site, but this complex process
is not practical on a routine basis.
The nominal flexural strength is the maximum stress at the outer fiber or tensile
surface of the bar. The stress correction for locations beneath the surface is
y h
y′
(a) (b)
(c)
y
y
y′
Figure D.14 The flexural stress gradient is shown in (a) and (b) for a
rectangular beam. (c) shows how the origin location may be measured from a
rod neutral axis.
D-26
simple if the material is linearly elastic. The stress decreases from a maximum
at the outer tensile surface to zero at the middle “neutral axis” which
corresponds to the centroid in a square, rectangular, or circular cross-section
beam specimen as shown in Figure D.14. At the origin,
where y is the distance from the centroid and h is the beam thickness or rod
diameter. Alternatively, in terms of the distance from the tensile surface, y’:
In similar fashion, if a fracture origin is not directly under the middle loading
point in three-point bending, or if the origin is outside the inner gage length in
four-point bending, then the stress should be correspondingly adjusted to the
stress that actually acted on the origin site. Corrections for breaks outside the
gage section or away from the middle loading point are also linear. The stress
is reduced in proportion to the distance from the fracture plane to the outer
loading points. This correction requires knowledge of the loading point
locations. This is an important reason why loading points should be marked on
any test piece prior to its fracture. For three-point loading:
where x is the distance from the outer loading point to the fracture plane and L
is the total three-point span. For four-point flexure,
where x is the distance from an outer loading point to the fracture plane, and a
is the distance from the outer loading point to an inner loading point. If an
origin in a flexural beam is subsurface and outside the inner span, then both
corrections should be applied. If the origin is on the tensile surface, within the
inner span length, and the mirror size is small and semicircular, then no stress
correction is needed.
If the origin is on the tensile surface in the inner loading circle, and the mirror
size is small and semicircular, then no stress correction is needed. Fracture
mirrors from volume flaws located beneath the tensile surface should be
corrected in the same manner as above for the uniaxial bend test methods.
D-27
Corrections for mirrors located outside the inner loading ring also should be
corrected for location. The corrections are not as simple in this case, however.
In the annulus outside the inner loading ring, the radial and circumferential
(hoop) stresses decrease out to the specimen rim but at different rates. 21, 22
Radial stresses decrease faster, and are zero at the outer rim, and are therefore
less likely to induce fracture. Fracture planes and mirrors will usually be
perpendicular to the hoop stresses outside the inner ring, so use the hoop stress
in the analysis.
On the other hand, hoop stresses decrease more slowly and are non-zero at the
outer circumference of the specimen. So although the outer hoop stresses are
only 10 % to 50 % of the maximum stress in the inner circle (depending upon
the disk and fixture geometries, and especially the disk thickness), large
grinding or handling flaws on the specimen rim can cause fracture.
The same comments apply to piston on three-balls, ball on three balls, or even
pressure loaded plates in bending. Consult appropriate references as needed
for the stress distributions in these configurations.
If the mirror constant for the material is known and a mirror size is measured,
the stress at the origin site can be calculated. The peak stress or the nominal
stress in the part may be different, however. So as a rudimentary example,
imagine a bend bar that broke from an internal origin site, but at an unknown
stress. The fractographer uses the mirror size and the mirror constant to
estimate the stress at the origin site. The nominal strength of the beam (the
maximum outer fiber stress) or the “flexural strength” is greater, however. In
this case, the stress adjustment should use σ, A, h, x, y or y’ to compute the
greater nominal stress, σmax.
D-28
(a) (b)
Figure D.7 (duplicate ) Plot of applied stress, σa, versus 1/√R, the preferred
regression method. (a) shows the trend for parts with no residual stress. (b)
shows the effects of residual stresses. Compressive residual stresses move the
line up with a positive intercept σr, but with the same slope. Tensile residual
stresses shift the data downwards.
Use some judgment in the regression analysis since fracture mirror data usually
has moderate scatter. If the data does not appear to fit a trend that has a zero
intercept, regress the data with a non-zero intercept as shown in Figure D.7b.
Again use some judgment in the interpretation, since a strict linear regression
fit may produce implausible outcomes, particularly if the data is collected over
a limited range of mirror sizes and stresses. Report the intercept if it deviates
significantly (> 10 MPa) from zero. Investigate possible residual stresses or
specimen size or shape issues in such cases.
Consistent units should be used. That is to say, if the stress axis is MN/m2 or
MPa, then the abscissa (horizontal axis) should be 1/√R where radius is in units
D-29
The mirror constant as a slope is easily visualized. Deviations from the trend
usually cause a nonzero intercept, which may be conveniently interpreted as an
effective residual stress. If residual stresses σr are present in addition to the
externally applied stress, σa, then the net stress acting on the origin site is: σa +
σr. The fracture and the mirror markings respond to the actual net stress, σnet:
A (D.10)
σ net = (σ a + σ r ) =
R
A
σa = −σr (D.11)
R
An intercept below the origin corresponds to a net tensile residual stress. A
positive intercept corresponds to residual compressive stress since the usual
sign convention is for compressive stresses to have a negative sign.
Some caution is advised since residual stresses are often nonuniform. The
residual stress estimated from the intercept is an effective residual stress, which
in reality may vary in magnitude through the mirror region. If the mirror is in a
heat-strengthened or tempered piece (where stress may be constant along the
surface, but change dramatically through the thickness) the mirrors should only
be measured along the surface or just underneath to avoid the cusp. Residual
stresses from an indentation or impact site are very local to the origin and may
have very little effect on a mirror size.
Although most researchers have felt that the regressed lines should go through
the origin in annealed test pieces, there is evidence by J. Quinn 23 that a small
but measurable intercept may exist in even annealed materials. The intrinsic
intercept was evaluated as 10 MPa (1,500 psi) for glass, a value that
interestingly concurs with Orr’s 24 estimate of the minimum stress necessary for
branching in glass.
D-30
log
σa -½
(a)
σ = A / √1
large
mirrors
R=1
log R Log R = 0
log
slope < - ½
σa
-½
σr
(b)
σ = A / √1
R=1
log R Log R = 0
Figure D.15 Plot of log σa versus log R for residual stress-free parts (a) and
parts with residual stress (b). Compressive residual stresses move the locus
upwards, but with a different slope and intercept. Tensile residual stresses
move the loci below the baseline curve (not shown).
D-31
where σa is the stress at the origin estimated from known applied stresses. R is
the mirror radius in the direction of constant stress. A’ is a modified mirror
constant. Report the slope m and the alternative constant A’. Again use some
judgment in the interpretation, since a strict linear regression fit may produce
implausible outcomes, particularly if the data is collected over a limited range
of mirror sizes and stresses. Investigate possible residual stresses or specimen
size or shape issues if m deviates significantly from the value –½.
If stresses are in units of MN/m2 (MPa) and the mirror size is measured in
meters, then the mirror constant A has units of MN/m1.5 or MPa√m. If the
mirror size is 1 m, then log R = 0. Then log σ = log A and hence, σ = A. Thus,
the mirror constant A corresponds to the value of stress that would create a
mirror of size 1 m.
Use consistent units with this approach for the same reasons mentioned in
section 11a. If the stresses are in MPa, then the abscissa (horizontal axis)
should be with radii in units of meters or micrometers. (If the stress units are
psi or ksi, the mirror radii should be in inches.)
Since most actual mirrors that are measured are usually much smaller than unit
size, it is apparent from Figure D.15 that the mirror constant (or the stress at R
= 1) lies somewhat beyond the range of data usually collected. This method of
showing the results and calculating a mirror constant was common in the older
technical literature and is occasionally still found today. Deviations from the
linear relationship on a log - log plot occur when residual stresses are present
but unaccounted for (since the plotted stress may not be the true stress at the
origin), or when the mirror size becomes large relative to the component size,
or when there are stress gradients. The residual stress deviations cause the line
to have a slope other than -½ as shown in Figure D.15b. Attempts to compute
the residual stresses may then be made by guessing values of the residual
stresses σr, replotting the data with a vertical axis of (σa - σr) , and checking the
goodness of fit of a line of slope -½. This is a fairly cumbersome process and
the alternative procedure in section 11a and Figure D.7b may be simpler and
more effective.
If a single mirror is measured, and the externally applied stress, σa, and the
mirror constant A are both known, then:
(σ a + σ r ) R=A (D.14)
and A
σr = −σa (D.15)
R
D-32
Mecholsky and colleagues 25,26 have shown excellent examples how residual
surface stresses in tempered or clad glasses may be estimated from equation
D.8 and fracture mechanics analysis.
Some researchers may have evaluated test pieces with residual stresses, and
then force-fitted regression lines through the data with zero intercepts in the
former scheme, or lines of slope -½ through the data in the latter scheme. This
probably contributed to variability in published mirror constants.
Figure D-16 shows both graph types for a common set of data from the author’s
work on the effects of machining on the strength of silicon nitride.6,7 A large
number of rod- or bar-shaped specimens were tested. They had various grind-
ing treatments. The mirror constants from the two analyses are quite similar,
although there is a small difference (9%) between the constants for the rods and
bars.
(a) (b)
Figure D-16 Fracture minor data for silicon nitride.6,7 The round blue
symbols are for 60 rods and the square red symbols are for 80 bars.
D-33
Data from flexurally-loaded specimens frequently deviate from the trends when
mirror sizes are large and are a significant fraction (50 % or greater) of the
cross section size. Upward deviations from the log stress - log radius graphs
have been noted in a number of studies (e.g., Shand,20,27 Orr24). The
experimentally measured mirrors are larger than they otherwise would be in a
uniform tension stress field. Regression lines chase the upward deviations from
the trend and dramatically alter the estimate of the mirror constant. On the
other hand, with the σ versus 1/√R graph, large or oversized mirror data points
are closer to the origin and have less influence on the regression line, and hence
have less effect upon the slope.
Regression analysis with the σ versus 1/√R approach minimizes the deviations
of σ from the fitted line. Regression analysis for the log σ versus log R
approach minimizes deviations of log σ from the fitted line.
In the 1950s and 1960s many researchers plotted log stress versus log radius
probably because they were not confident of the theoretical justification for the
-½ slope. They let the exponent vary on the log-log plots and discussed the
differences, if any, relative to -½ power. Differences were usually due to
residual stresses or overly-large mirrors relative to the specimen size.
Gradually the case for the σ√R relationship solidified and more researchers
began to plot stress versus 1/√R.
In summary, the linear stress versus inverse square root radius approach is
superior and is recommended by this Guide. Analyses are simple and intuitive.
The mirror constant is the slope of the regressed line. Residual stresses may be
interpreted from non-zero intercepts. The uncertainty of the slope can be
estimated from routine analyses available in many statistical software packages.
Data deviations due to large mirrors in flexure specimens have less effect upon
the regression process and the mirror constant estimates. The alternative
method (log stress – log radius) may be useful in some cases for displaying data
with an unusually large range of mirror sizes and stresses. 28
12. Mirrors sizes should be collected over a broad range of sizes and
fracture stresses if possible. Data from different specimen types and sizes
may be combined.
This is a fairly obvious conclusion in light of the discussion in the previous
paragraph. Superb examples are shown by Kerper and Scuderi28, for
borosilicate glass rods with diameters that varied by a factor of ten, and by
Mecholsky and Rice 29 for various sized fused silica rods, disks, and fibers.
D-34
They then listed some values for a few glasses and ceramics that did in fact
vary as much as 20 % to 30 %. For example, the soda-lime glass values varied
by 23 %. Hopefully, adoption of the guidelines in this Guide and ASTM C
1678 will improve the consistency of future data to within 10%.
D-35
Another interesting finding from the literature review is how few authors have
shown good fracture mirror photos. Even fewer have marked them. Perhaps
the authors were not sure or were hesitant to show an interpretation for fear of
criticism. Hopefully, step 8 of this Guide (which requires marked photos) and
the examples shown in this Guide will help future authors improve their
reporting. One is also struck by the fact that nearly all the mirrors shown in the
literature, even in the classical papers, are not exactly semicircular or circular,
despite all the schematics that imply that they are. Fractographers should not
be alarmed if their mirrors are not perfect.
Many of the steps in this Guide have already been proposed. Shand recom-
mended that stresses be corrected for the origin location,20 that radii be
measured beneath the surface to avoid surface effects,18 that low-angle grazing
illumination be used.20 He also warned about deviations from the trends if the
mirror sizes were too large relative to the component thickness.20 Shand also
said that mist could not be discerned in glass ceramics.20 Morrell et al. 30
agonized over the interpretation of mirrors in Y-TZP zirconia (such as shown in
Figures D.4 – D.6), but settled on a set of specific criteria. Optical microscopy
with a stereo optical microscope at a fixed magnification was used, with
grazing incidence illumination. No reflective coating was applied to the
surface. The specimen sides were masked to block transmitted-light scatter.
Matching fracture halves were mounted together to aid the interpretation. The
best set of recommendations predating this Guide were crafted by Mecholsky
and Freiman.19 Six of their recommendations match steps in this Guide: (a)
optical microscopy is preferred over scanning electron microscopy whenever
possible, (b) suitable magnifications should be used, (c) mirror boundary arcs
should be projected to the outer surface to complete a circular arc to eliminate
the surface cusps, (d) lighting should be varied to obtain optimum contrast, (e)
radii should be measured in directions of constant stress and not into gradients,
and (f) trend deviations occur for mirrors that are large relative to part
thickness.
D-36
Appendix D References
1. G. D. Quinn, “Guidelines for Measuring Fracture Mirrors,” pp. 163 – 190 in:
Fractography of Glasses and Ceramics, V, eds., J. R. Varner, G. D. Quinn, M.
Wightman, American Ceramic Society, Westerville, OH, 2007.
2. ASTM C 1678, “Standard Practice for Fractographic Analysis of Fracture Mirror
Sizes in Ceramics and Glasses,” Annual Book of Standards, Vol. 15.01 ASTM Int.
West Conshohocken, PA, 2007.
3. J. W. Johnson and D. G. Holloway, “On the Shape and Size of the Fracture Zones
on Glass Fracture Surfaces,” Phil. Mag., 14 (1966) 731 - 743.
4. H. P. Kirchner and J. W. Kirchner, “Fracture Mechanics of Fracture Mirrors,” J.
Am. Ceram. Soc., 62 [3-4] (1979) 198 - 202.
5. H. P. Kirchner and J. C. Conway, Jr., “Criteria for Crack Branching in Cylindrical
Rods: I, Tension and II, Flexure,” J. Am. Ceram. Soc., 70 [6] (1987) 413 - 425.
6. G. D. Quinn, L. K, Ives, and S. Jahanmir, “On the Fractographic Analysis of
Machining Cracks in Ground Ceramics: A Case Study on Silicon Nitride,” SP 996,
NIST, Gaithersburg, MD, May, 2003.
7. G. D. Quinn, L. K. Ives, S. Jahanmir, and P. Koshy, “Fractographic Analysis of
Machining Cracks in Silicon Nitride Rods and Bars,” pp. 343 - 365 in
Fractography of Glasses and Ceramics IV, Ceramic Transactions, Vol. 122, eds.,
J. R. Varner and G. D. Quinn, American Ceramic Society, Westerville, OH, 2001.
8. G. D. Quinn, J. Eichler, U. Eisele, and J. Rödel, “Fracture Mirrors in a Nanoscale
3Y-TZP,” J. Am. Ceram. Soc., 87 [3] (2004) 513 - 516.
9. W. H. Duckworth, D. K. Shetty, and A. R. Rosenfeld, “Influence of Stress
Gradients on the Relationship Between Fracture Stress and Mirror Size for Float
Glass,” Glass Technol., 24 [5] (1983) 263 - 273.
10. D. M. Kuluwansa, L. C. Jensen, S. C. Langford, and J. T. Dickinson, “Scanning
Tunneling Microscope Observations of the Mirror Region of Silicate Glass
Fracture Surfaces,” J. Mater. Res., 9 [2] (1994) 476 – 485.
11. D. Hull, “Influence of Stress Intensity and Crack Speed on Fracture Surface
Topography: Mirror to Mist Transition,” J. Mat. Sci., 31 (1996) 1829 - 1841.
12. D. Hull, Chapter 5, in Fractography, Observing, Measuring and Interpreting
Fracture Surface Topography, Cambridge Univ. Press., Cambridge, 1999.
13. C. Wünsche, E. Rädlein, and G.H. Frischat, “Morphology of Silica and
Borosilicate Glass Fracture Surfaces by Atomic Force Microscopy,” Glastech. Ber.
Glass Sci. Technol., 72 (1999) [2] 9 - 54.
14. E. Gölz, “Übermikroskopische Feinstructuren an Glasbruchflächen,” Zeitschrift für
Physik, 120 (1943) 773 - 777.
15. E. K. Beauchamp, “Mechanisms of Hackle Formation and Crack Branching,” pp.
409 - 446 in Fractography of Glasses and Ceramics III, eds., Varner, Fréchette,
Quinn, Cer. Trans., Vol. 64, 1996, American Ceramic Society, Westerville, OH.
16. E. K. Beauchamp, “Fracture Branching and Dicing in Stressed Glass,” Sandia
Research Report SC-RR-70-766, January 1971, Sandia National Laboratory,
Albuquerque, NM.
D-37
D-38
Index
INDEX
A
Abrasion 6-39, 6-66 – 6.69
Acid, aqua regia 3-26
Acid storage tank 9-6
Agglomerate, see origins
Alumina 4-18, 4-32, 4-45, 5-19, 5-68, 5-71, 6-7,
6-8, 6-11, 6-12, 6-14, 6-23, 6-30, 6-70,
6-77
10-30 –10-37, 10-43 – 10-45
Alumina, β 6-16
Alumina furnace plate 4-32
Alumina crown 4-45, 10-30 – 10-37, 10-43 – 10-45
Alumina single crystal, see sapphire
Alumina, whisker reinforced 7-71
Aluminum oxynitride 5-34, 5-67, 5-74, 6-8, 6-13, 6-75, 10-16 –
10-17
Aqua regia 3-26
Arrest lines 5-58 - 5-60
Atomic Force Microscope 3-60
Automobile window 4-31, 4-50, 7-7
B
Ball bearings 3-68, 6-39
Ball mill 6-60
Baseline microstructural flaws 6-75
Batman mirror 8-19
Bend bar 4-14 – 4-18, 10-19 – 10-25
Bevels, see chamfers
Biaxial disk 3-23, 4-5, 4-9, 4-19 – 4-21, 5-72, 6-29,
6-31, 7-4
Biaxial stress 1-3, 1-4, 4-19 – 4-21
Bifurcation
Black light 3-55
Blanchard grinding 6-40
Bleach 3-6
Blisters, see origins
Blood 5-63
Boron carbide 4-50, 7-61
Bottles 4-7, 4-38, 7-3, 10-14, 10-15
Boundary phases, ceramics 5-78, 5-79
I-1
Fractography of Ceramics and Glasses
C
C crack 3-68, 6-35
Camera 3-7 – 3-11, 3-14 – 3-16
Camera, single lens reflex 3-8
Camera stand 3-7
Camera mount, F-mount or C-mount 3-14
Cantilever curl 4-15 – 4-17, 5-2
Capacitor, multilayer 3-23
Cathedral mirror 8-9, 8-10, 8-16, 8-24
Cavitation, scarp 5-61, 5-62 7-53
Cell phone 3-10, 3-11
Cellulose acetate 3-36 - 3-38, 3-42
Center heated plate 4-32, 4-33
Chain of events 1-10
Chamfers 6-16, 6-23, 6-40, 6-53
Characteristic strength 7-72
Chatter marks, see origins
Chill checks, see origins
Chips, chipping, see edge chips
Clay 3-5, 6-70, 6-71
Cleaning coatings off 3-26, 3-71
Cleaning specimens 3-5, 3-6, 3-71, 5-1
Cleaning, dental plaque 3-6
Cleaning, green dye 3-29
Cleaning, removal of gold coatings 3-26
Cleaner, ultrasonic 3-71
Cleavage 5-3, 8-1, 8-2, 8-5
Cleavage step hackle 8-8
Coatings 3-26 – 3.29
Combination flaw 6-73, 6-74
Component fracture patterns 4-22, 4-44, 4-45
Composites 6-20, 7-24, 7-72, 9-1 – 9-11
Compositional inhomogeneity, see origins
Compound optical microscope 3-32 – 3-35
I-2
Index
D
Damage wave 5-6
Dandruff 6-69
Dark field illumination 3-20, 3-24, 3-32, 8-5, 8-6
Debonding 10-20, 10-21
Debris on fracture surface 5-83, 5-84
Deep Reactive Ion Etching (DRIE) 6-76, 8-23, 8-24, 10-25 – 10-27
Deer 4-23
Defect, see also flaws and origins 6-1
Delaminations 6-57
Delta markings 5-3, 5-36
Dental ceramics, crowns, bridges 3-38, 3-40, 3-41, 10-30 – 10-45
Alumina 3-41, 4-45, 6-55, 6-58, 6-87, 10-30–10-37,
10-43 – 10-45
I-3
Fractography of Ceramics and Glasses
E
E-glass composite 9-4, 9-6
Edges 6-1, 6-2, 6-40, 6-53
Edge chips 4-40 – 4-44, 10-36, 10-37, 10-43 – 10-45
Elastic waves 5-6, 5-47 – 5-58, 7-46 – 7-50
I-4
Index
F
Failure analysis 1-6 – 1-8
Far field stress 7-21
Fatigue crack 3-24, 5-81 – 5-84
Fiber composite 9-1 – 9-6
Fiberglass- epoxy composite 9-4, 9-5
Fiber reinforcement 7-24, 9-1, 9-2
File formats, digital images 3-8, 3-9
Finite element analysis 1-5, 4-46, 7-18
Firing cracks 4-10, 6-54 – 6-56, 6-58
Flaw 6-1
Flaw bluntness 7-20, 7-39 – 7-41
Flaw linkage 6-50, 6-73, 6-74, 6-80 – 6-82, 7-44, 7-45
Flaw management 6-75, 7-75
Flaw shielding 7-44, 7-45
Flaw size 6-3, 6-75, 7-18, 7-19
Flaw size distribution 7-73, 7-74
Flaw truncation 6-77, 7-42, 7-44, 7-45
Flexural strength 4-14 – 4-18, 5-20
Flexural stress gradient D-18, D-19, D-26, D-27
Flow lines of stress 7-20, 7-21, 7-44
Fluorescent dye penetration 3-67 – 3-68
Focus stacking 3-17
Ford Rotor 10-1 – 10-7
Ford Turbocharger 6-56
Fractal analysis 7-65 – 7-67
Fractography, definition 1-1
Fractography, Laws of 11-1 – 11-3
Fracture, conchoidal 5-3
Fracture energy 7-18, 7-22
Fracture map 5-80, 5-81
Fracture mirror, definition 5-3
Fracture mirror 5-3 – 5-30, Appendices C and D,
see also mirror
I-5
Fractography of Ceramics and Glasses
G
Gas turbine rotor 10-1 – 10-7
Glass
Borosilicate, Pyrex 7-16
Borosilicate crown 4-5, 4-21, 6-31, 6-59, 7-4, 7-34, 7-61
Fused silica 5-2, 5-4, 5-5, 6-6, D-3, D-22
Fused silica, sintered 10-20 – 10-22
Soda lime 4-23, 6-18, 6-57 – 6-64, 7-3, 7-5, 7-6,
7-9
Glass bottles, see bottles
Glass ceramic 6-7, 6-11, 6-60
Glass ceramic, dental 3-29, 6-60
Glass disk, see biaxial disk
Glass rod, flexure 1-1, 4-15, D-3, D-20, D-22, D-26
Glass tube 4-36
Glass windows, see windows
Glazes, veneers 5-33, 5-34
Gloves, cotton 3-2
Glue, specimen reconstruction 4-2, 4-3
Glue chips 5-63 - 5-66
Grazing illumination 3-12, 3-22
Green pen, dyes 3-28, 3-29, 10-42
Griffith flaw 6-1, 7-18, 7-19
Griffith criterion, equation 7-18, 7-19
Grinding cracks, see origins, grinding cracks
Grinding damage, glass disks 4-20, 4-21
Gull wings 5-14, 5-49, 5-50, 7-46, 7-47
Gull wing mirrors 8-9, 8-15, 8-18
Gun Barrel 10-7 – 10-10
H
Hackle 5-10 – 5-16, 5-31 - 5-44
cleavage step 8-8
corner 5-43
I-6
Index
definition 5-12
coarse microstructural 5-31, 5-32
grinding crack 6-40 – 6-52
fine microstructural 5-36, 5-41
mist hackle 5-12
shear hackle 5-42
step hackle 5-43, 5-44
twist hackle 5-36 – 5-41
velocity hackle, definition 5-12
wake hackle 5-33 – 5-36, 5-49
Hailstone 7-4
Hair 3-23, 6-69
Halo, slow crack growth 5-71
Hand magnifier 3-2
Handling damage 6-29, 6-36, 6-39, 8-7
Healed cracks 4-28
Heat sleeks 6-63
Hertzian cone cracks 6-34 – 6-40, 10-10, 10-31, 10-47, 10-48
High speed photography 3-64
Hinge fracture 4-34 – 4-36
History of brittle materials fractography 1-8, 1-9
Holders 3-4, 3-5
Hoop stress 4-33, 4-37 – 4-38
Hopf bifurcation 4-32
Human enamel 4-4, 4-46
Human skin contamination 6-70
Hybrid flaws 6-73, 6-74
Hydrogen peroxide 3-29
I
Illumination 3-12, 3-20 – 3-26
Impact 4-29, 4-35, 4-49
Impact origins, see origins impact
Impact, tertiary Wallner lines 5-53
Inclusion, see origins-inclusion
Incomplete bonding 6-16
Indentation fracture mechanics 7-25, 7-61 – 7-65
Insect 3-37, 6-69
Insulator, electric 6-8
Intrinsic flaws 6-3, 6-4
Intergranular fracture 5-17, 5-66 - 5-70, 5-75, 5-79
Intersecting cracks 4-26, 4-27
I-7
Fractography of Ceramics and Glasses
J
Jeweler’s Loupe 3-3, 3-4
Joint 5-3
JPEG format 3-8, 3-9
K
Kitchenware 6-63, 6-66, 6-68
Knoop indentation crack, see origins
K-V diagram (slow crack growth) 7-51 – 7-53
L
Laboratory fractures 4-11, 4-44
Lambda lines 5-58
Lamp fracture 4-11, 7-7
Lances, see also hackle 5-16, 5-36, 8-13, 8-17
Law of normal crack propagation 4-3
Lead zirconium titanate (PZT) 5-35, 6-57
Leader crack 4-35, 4-36
Lint 6-8, 6-69
Lithic fractures 5-3, 5-62
Lithic fractures - Obsidian 5-35, 5-49, 5-62
Litigation 1-8
Longitudinal grinding 6-40, 6-41
Loupes 3-3, 3-4
Low-angle grazing illumination 3-12, 3-22, 5-19, 6-45, 8-5
M
Machine gun barrel 10-7 – 10-10
Machine shops 10-23 – 10-25
Machining cracks, see origins
Machinist loupe 3-3, 3-4
Macrofractography camera stand 3-7
Magnesium fluoride 5-71
Magnesium oxide 8-1
Magnesium aluminate spinel, see spinel
Magnifying glass 3-2
Mandelbrot relationship 7-66
Margin damage, dental crowns 10-32 – 10-37, 10-43 – 10-45
Maximum distortion energy criterion 1-5
I-8
Index
I-9
Fractography of Ceramics and Glasses
N
Near surface 6-3
Neutral axis, bending 4-14 – 4-16
Nickel sulfide 6-18
Nomarski illumination 5.62, D-12, D-13
Nomenclature 1-8, 1-9, 5-1, 5-3, 6-1
O
Obsidian 5-14, 5-35, 5-49, 5-62
Optical comparator 3-3
Optical profilometer 3-60 – 3-62
Origin 6-1
Abrasion tracks 6-66 – 6-48
Agglomerate 3-51, 6-6, 6-13, 9-1
Ball mills 6-5, 6-69
Baseline microstructure 6-75
Blank pressing flaws 6-11
Blisters, glass 6-61
Bubbles 6-7, 6-9, 6-10, 6-11
Chatter marks, cracks 5-29, 6-65 – 6-68
Chill checks 4-28, 4-48, 6-64, 6-65
Chips 6-53
Contact cracks 3-42, 3-68, 6-34 – 6-39
Cords, glass 6-61
Compositional inhomogeneity 3-51, 6-21, 6-22
Cusps, geometric irregularities 6-76
Dandruff 6-69
Delaminations 6-58
Devitrification stones, inclusions 6-61
Dunt cracks 6-57
Etch groove 10-26, 10-27
Etch pits 8-24
Fatigue crack 3-24
Feces 6-69
Fiber 6-8
Fiber clump 9-1
Firing cracks 6-54 – 6-56, 6-58
Geometric sharp points 6-76
I-10
Index
I-11
Fractography of Ceramics and Glasses
Vickers indentation crack 5-22, 6-59, 6-33, 7-25, 7-25, 7-59, 7-61 –
7-65
Whisker clump 9-1
Origin, fracture definition 4-1
Orthogonal machining crack, see origins
Oxidation 4-49, 6-28
P
Parallel machining crack, see origins
Parfocal 3-12, 3-20
Parting 5-3
Patriot missile 10-20
Pattern recognition 1-3
Pecks 6-33, 6-63
Penny shape cracks 7-26, 7-29, 7-40
Pharmaceutical vials and syringes 4-37, 5-62, 6-39, 10-14, 10-15
Phase instability 6-69
Photos, Photography 3-7 – 3-11
Pits, see origins
Plaque, dental 3-6
Plasma asher 8-27 – 8-29
Plane strain 1-5
Plane stress 1-5
Plaque, dental 3-6, 6-71
Plate fracture patterns 4-28 – 4-30
Plate fracture patterns, thermal 4-32, 4-33
Plate, whiteware 3-42
Polariscope 3-70, 3-71
Polarizer 3-25, 3-26, 8-5, 8-26 – 8-29,
Polished microstructural section 6-5, 6-75, 6-82, 6-83
Polishing surface flaws, see origins
Polycrystalline alumina (PCA) 10-17
Porcelain 3-37, 3-42, 3-66, 4-41, 6-8, 6-10, 6-24,
6-37, 6-52
Porcelain electrical insulator 3-37, 6-8, 6-37
Polyvinylchloride (PVC), replicas 3-38
Polyvinylsiloxane (PVS) replicas 3-38, 3-41
Pore, see origins
Porous region, see origins
Porous seam, see origins
Pressing flaws 6-11
Pressure flaking 5-63
I-12
Index
Q
Quartz grains, Quartz inversion 6-24, 6-57
Quasi static loading 6-35
R
R-curve 5-73, 6-48, 7-23, 7-25, 7-55 – 7-60
Radome, fused silica 10-20 – 10-22
Rayleigh wave velocity 5-5
RAW format 3-8
Reliability 10-1 – 10-7
Renegade abrasive grit 6-48, 6-49
Replicas 3-35 – 3-42, 6-82
Residual stress 4-49 – 4-50
Compression induced cracking 5-63
Estimates of magnitude 7-68– 7-70, D-30 – D-34
Grinding 6-40, 7-68
Heat strengthened glass 5-28, 5-29
Indentation 6-59, 6-60, 7-63 – 7-65, 7-70
Mirror shapes, effect on 5-28, D-18 – D-24
Stable crack extension from 5-73
Tempered glass 4-28 – 4-31, 4-49, 5-30
Resin-matrix composite, dental 6-20, 6-21, 6-69, 9-6 – 9-8
Retroreflective flaws 6-69
Rib marks 5-45, 5-58
Rings of Saturn 6-7, 7-20, 7-39, 7-40
Ripples 5-45, 5-56
River deltas 5-3, 5-36
Rock fracture, see lithic fracture
Roman brick 4-41
Rotor, gas turbine 10-1 – 10-7
Round robin, flexural strength 10-23 – 10-25
Round robin, fractography 10-28 – 10-30
Round robin, machining 10-23 – 10-25
S
Saliva 5-63
Sapphire, single crystal alumina 3-65, 6-31, 8-4, 8-7, 8-19, 8-21, 8-22,
I-13
Fractography of Ceramics and Glasses
I-14
Index
Zirconia 8-1
Sketches 4-24 – 4.26
Sleeks 6-65
Slow crack growth 4-51, 5-69 – 5-76, 7-23, 7-25, 7-51 –
7-54, 10-1 – 10-5, 10-11 – 10-14
Slow crack growth exponent, N 7-51, 7-52
Soda Lime silica, see glass
Spinel, magnesium aluminum oxide 6-25, 6-26, 8-9 - 8-18, 8-25
Spray-dried particles 6-5, 6-15, 6-16, 6-55
Stable crack extension 5-69 - 5-76, 7-42, 7-43, 7-51 – 7-60, 10-11
Staining 3-26– 3-29, 3-66 – 3-70
Standards
ASTM C 1322, fractography 6-73, 6-84, 7-28, 7-32, 8-9, 10-28
I-15
Fractography of Ceramics and Glasses
T
Teeth 4-4, 4-46
Tempered glass, heat strengthened glass 3-10, 4-11, 4-29 – 4-31, 6-18, 6-66, 6-68,
7-5 - 7-9
10-46 – 10-48
I-16
Index
U
Ultrasonic cleaner 3-71
Ultrasonic fractography 3-64, 5-56, 5-57
Uniaxial stress 1-4
USB microscope 3-19
V
VAMAS 10-28
Vanadium inclusion 6-17
Velocity hackle 5-11, 5-12, 5-31
Velocity, terminal 5-6, 5-7
Veneer, dental 6-36, 6-71, 6-77 – 6-79, 6-87
Vials, medicinal 4-37, 5-62, 6-39,10-14, 10-15
Vicinal illumination 3-22, 3-23
Vickers indentation flaws, see origins
Video camera 3-15
V machining crack, see origins
Von Mises failure criterion 1-5
W
Wake hackle 5-33 – 5-36, 5-49, 9-3, 10-16, 10-17,
10-33
10-35, 10-36, 10-45
Wallner lines 5-39– 5-58, 7-46 – 7-50, 8-8
Gull wings 5-14, 5-49, 5-50
Lambda 5-58
Primary 5-45 – 5-47
Secondary 5-50 – 5-52
Tertiary 5-53 – 5-55, 8.22
Watch bracelet 6-56
Watchmaker’s loupe 3-3
Water hammer 4-34, 4-35
Wave velocity, see elastic waves
Weibull distribution 6-86, 7-71 – 7.76
Weibull modulus 7-59, 7-71 – 7-76
Whisker lances, single crystal 8-9, 8-13, 8-17
Whisker reinforcement 7-24, 9-1, 9-2
Wing cracks 4-53
Window failure 4-23 – 4-33
Window failure, glue chips 5-66
Window patterns 4-28 – 4-33
I-17
Fractography of Ceramics and Glasses
X
X-ray topography 3-65, 8-5, 8-7
X-Y stage, traversing 3-31, 3-34
Z
Zinc oxide, varistor 6-16
Zipper crack, see origins, machining
Zirconia, cubic single crystal 8-1
Zirconia, magnesia stabilized 5-31, 6-25, 6-26, 6-81, 6-82, 7-23, 7-57,
7-58
Zirconia, yttria stabilized 5-17, 5-19, 6-8, 6-13, 6-18, 6-56, 6-58,
6-72, 7-14, 7-15, 7-16, 7-60, D-7 – D-9
I-18