ASTM - STP436 Electron Fractography

ELECTRON
FRACTOGRAPHY
A symposium
presented at the
Seventieth Annual Meeting
AMERICAN SOCIETY FOR
TESTING AND MATERIALS
Boston, Mass., 25-30 June, 1967
ASTM SPECIAL TECHN!CAL PUBLICATION NO. 436
List price $11.00; 20 per cent discount to members
@ published by the
AMERICAN SOCIETY FOR TESTING AND MATERIALS
1916 Race Street, Philadelphia, Pa. 19103
BY AMERICAN SOCIETY FOR TESTIN~ AND MATERIALS 1968
Library of Congress Catalog Card Number: 68-15547
NOTE
The Society is not responsible, as a body,
for the statements and opinions
advanced in this publication.
Printed in Rahway, N.J.

July, 1968
Foreword
The Symposium on Electron Fractography was presented during

the Seventieth Annual Meeting of the Society held in Boston, Mass.,
25-30 June 1967. The symposium was sponsored by Committee E-24
on Fracture Testing of Metals. The symposium chairman was C. D.
Beachem, Naval Research Laboratory.
Related
ASTM Publications
Techniques of Electron Microscopy Diffraction

Microprobe Analysis, STP 372 (1965), $3.75
Fracture Toughness Testing and Its Applications,

STP 381 (1965), $19.50
Advances in Electron Metallography, Vol. 6,

STP 396 (1966), $7.00
Fatigue Crack Propagation, STP 415 (1967), $30.00

Contents
Introduction . . . . . . . . . . . . . . . . . . . . . 1
Electron F r a c t o g r a p h y - Tools and T e c h n i q u e s - J. L. MCCALL 3

A Fractographic Analysis of the Relationships Between Fracture
Toughness and Surface Topography in Ultrahigh-Strength Steels
--W. A. S P I T Z I G , G. E, P E L L I S S I E R , C. D. B E A C H E M , M. H I L L , A N D
W . R. W A R K E . . . . . . . . . . . . . . . . . . 13
Discussion . . . . . . . . . . . . . . . . . . . . 30
Cleavage Phenomena and Topographic F e a t u r e s - H . C. BURGHARD,JR.,

A N D N. S. S T O L O F F . . . . . . . . . . . . . . . . 32
Fracture by Microscopic Plastic Deformation Processes-c. O.

B E A C H E M A N D D. A. M E Y N . . . . . . . . . . . . . 59
Application of Electron Fractography to Fatigue S t u d i e s - J . c. MC MIL-

L A N A N D R. W . H E R T Z B E R G . . . . . . . . . . . . . 89
Environmental Effects on Fracture M o r p h o l o g y - N . A. NIELSEN 124
Special Fractographic Techniques for Failure A n a l y s i s - B . V. WroTE-

SON, A. PHILLIPS, V. KERLINS,AND R. A. RAWE . . . . . . . 151
Engineering Applications of F r a c t o g r a p h y - A . j. BROTHERS AND S.

YUKAWA . . . . . . . . . . . . . . . . . . . . 179
Fractography and Microstructure of Aluminum Alloys 7075-T651 and

7075-T7351--M. S. H U N T E R A N D J. C. M C M I L L A N . . . . . 196
Techniques for Electron Microscopic F r a c t o g r a p h y - w . R. WARKE,
N. A. N I E L S E N , R. W . H E R T Z B E R G , M. S. H U N T E R , A N D M. H I L L 212
STP436-EB/Jul. 1968
Introduction
The purpose of this symposium is to describe the present "state-of-

the-art" of electron fractography, and report the results of activity
sponsored by Subcommittee II on Fractography of ASTM Committee
E-24 on Fracture Testing of Metals. It is hoped that future symposia
on this subject will benefit from contributions from invited authors
from laboratories throughout the world where excellence of electron
fractographic usage in research and failure analyses has been proved.
As the papers in the volume indicate, a great deal has been learned
about the mechanisms of crack propagation, and the correlation of fine-
scale fracture surface features with macroscopic stress and environ-
ment conditions.
This volume, however, does not cover all of the fractography sub-
jects presently being studied, because tfiis tool is versatile and is being
used to complement other research tools in a variety of disciplines.
Neither have the subjects discussed herein been fully explored. There-
fore, this volume should serve as a progress report, and one should ex-
pect future symposia in electron fractography to include significant
new applications of the tool and further refinements of existing theory
and techniques.
C. D. Beachem
Head, Micro-Mechanical Metallurgury Section,
Physical Metallurgical Branch, Metallurgical Div.,
Naval Research Laboratory, Washington, D.C.;
symposium chairman.
Copyright~ 1968 by ASTM International www.astm.org

J. L. M c C a l l a
Electron Fractography Tools and

Techniques
R E F E R E N C E : McCall, J. L., " E l e c t r o n F r a c t o g r a p h y - Tools a n d Tech-

n i q u e s , " Electron Fractography, AS T M S TP 436, American Society for Test-
ing and Materials, 1968, pp. 3-16.
ABSTRACT: Electron fractography has recently become a valuable tool for

studying the topographic features of fractures, and from it much new theoreti-
cal and applied information regarding the micromechanisms of fractures has
been obtained. Prior to the use of electron microscopes, the unaided eye, opti-
cal microscopes, and metallographic cross sections were used to study frac-
tures. These methods, however, have distinct limitations such as low resolution
and small depth of field. Both of these limitations have been overcome with the
electron microscope. Other tools are now being applied to fracture studies to
complement the electron microscope. These include deep-focusing micro-
scopes, electron microprobe analyzers, and scanning and reflection electron
microscopes. The combination of all these tools has provided much new infor-
mation, but it must be realized that electron microscopy as a tool for studying
fractures is still essentially in its infancy, and it offers tremendous opportunities
which, as yet, have not been fully explored.
KEY WORDS: fractography, fracture, election microscopy
Valuable information has long been known to exist in the fracture

surfaces of materials, and, through the years, various approaches have
been employed to obtain and interpret this information. One of the
earliest accounts of fractographic studies is the report of the French
scientist, DeReaumur, who studied fractured facets of steel by micros-
copy in 1722 [1]. 2 In the latter part of the 19th Century, Martens [2],
generally considered as the one who originated metallography, per-
formed some limited fracture studies of metals in his work; however,
he is reported to have been somewhat discouraged because of the small
depth of focus of his microscope, which allowed him to see clearly
only a small portion of the fracture at any one time. Around the turn of
the century, Ewing and Humfrey [3] and Ewing and Rosenhain [4-6]
1 Division chief, Structure of Metals, Battelle Memorial Institute, Columbus Labora-
tories, Columbus, Ohio.
"~The italic numbers in brackets refer to the list of references appended to this paper.
3
Copyright s 1968 by A S T M International www.astm.org
4 ELECTRONFRACTOGRAPHY
were able to provide some significant information regarding the be-

havior of metals from studies of fractures. For example, they showed
that the crystalline appearance of certain fatigue fractures was not re-
lated to the grain size of the material but rather to the type of fracture.
Furthermore, they were the first to show that metals remained crystal-
line when deformed and to describe deformation markings which we
now know as slip lines.
Except for some studies by Howe [7], most of the microscopic
studies of metals in the early 1900's were limited to examinations of
polished sections. Somewhat later, a number of investigators recog-
nized that the properties of steels could be correlated with the coarse-
ness or fineness of the fracture surface. For example, it was found that
fractured specimens could bemused as a means of evaluating the grain
size of a metal. In 1927, Arpi [8] developed a set of standard fractures
which were believed to cover the full range from coarse to fine grain
size. Some time later, Shepherd [9], working in this country, developed
a similar set of standards for evaluating grain sizes of hardened steels.
His method is still in somewhat limited use today.
The most significant and detailed studies of fractures were performed
approximately 20 years ago by Zappfe and Clogg [I ]. For their work,
they developed techniques for using the optical microscope to study
fractures. Although they were bothered by the relatively small depth of
focus of the optical microscope, they were able to orient the facets of a
fracture with the axis of the microscope such that examinations could
be made at a relatively high magnification. As a result of their work, the
term "fractography" was coined. Most of their work was done on
brittle fractures, from which they described in considerable detail the
appearance and crystallography of cleavage facets.
Since Zappfe and Clogg's work, except for routine examinations of
the fractures of service failures, another lull occurred in fractographic
studies until the advent of the application of electron microscopy to
fracture studies [10,11]. With this new tool, there has been a resur-
gence of effort in research studies concerning the micromechanisms of
fracture, and, as a result, much new information regarding fracture has
been developed.
Use o f Optical Microscopes
Examination of Fracture Surfaces

Prior to the development of the electron microscope as a tool for
studying fracture, fractographers were limited essentially to the use of
their unaided eyes and to light microscopes. With the unaided eye, it is
possible to group fractures into rather broad categories, such as fibrous
and shear. In many instances, it is possible to positively recognize
McCALL ON TOOLS AND TECHNIQUES 5
certain types of fractures by certain characteristic markings. For ex-

ample, fatigue failures are characterized by their normally flat ap-
pearance and by the presence of stop rings or beach marks. Sometimes,
the origin of fracture can be determined by examinations with the un-
aided eye.
The optical microscope has made it possible to distinguish the frac-
ture mechanisms more precisely and to study such features as indi-
vidual grain facets, cleavage facets, fatigue striations, and fracture
origins. However, the optical microscope is severely limited for this
type of study by its restricted depth of field. The maximum depth of
field (defined as the vertical displacement of the object that can be
tolerated without loss of focus) [12] of optical microscopes in use today
is approximately 0.2/~m. This means that features on a fracture that
are displaced vertically by more than this amount cannot be observed
simultaneously.
In an attempt to overcome the small-depth-of-field problem,
McLachlin [13] has recently developed a microscope that allows rather
high-magnification photography of extremely rough specimens.
Figure 1 is a photograph of a commercial model of this instrument.
This instrument, called a deep-field photographic microscope, uses a
very thin beam of light to illuminate the specimen. The light beam is at
a constant distance from the microscope objective and is at the focal
plane. While the photograph is being made, the specimen is moved at a
constant rate through the beam of light. Since only the illuminated por-
tions of the specimen will be recorded on the photographic film, and
FIG. 1 - D e e p field photographic microscope developed by McLachlin [13].

since all of the illuminated portions will be in focus, the entire photo-
graph will also be in focus. Figure 2 is an example of the type of result
that can be obtained from this microscope. If the fracture shown in
Fig. 2 had been photographed at the same magnification (xl00)
through a normal microscope, only a small portion of the surface
would have been in focus because of the roughness of the surface.
When this instrument is fully developed, it should have wide applica-
tion in the field of fractography.
Examination of Metallographic Cross Sections

Because of the limitations of optical microscopes, a common tech-
nique for studying fractures has been to prepare and examine metallo-
graphic cross sections through the fractured faces. Often the two
halves of the fracture are matched and mounted together. To preserve
the edge the fracture is often plated prior to mounting. The mounted
fractures can be examined by optical microscopy, or, when details of
the fine structure of interest, electron microscopy can be used. Certain
types of significant information can be obtained from cross-sectional
analyses; in fact, prior to the use of electron fractography, this was
about the only method of determining whether a fracture occurred
intergranularly or transgranularly. One drawback of this type of
examination is that it presents only a one-dimensional view of the
fractured surface. As a result, only a limited amount of the fracture is
seen at any one time. Furthermore, the cross section often makes it
FIG. 2 - Fracture o f a steel specimen photographed with a deep-field photographic

microscope (XlO0).
difficult to relate the features seen in the microscope to those macro-

scopic features of the failed part. Of course, this type of examination is
destructive to the fracture.
Use of Transmission Electron Microscopes

Approximately 10 years ago [10,I1] electron microscopes were
first used to study fractures of materials. Electron microscopic studies
of fractures have overcome many of the limitations of the other means
of fracture examination [14-I 7]. Aside from the abilities of electron
microscopes to permit high magnification and high-resolution examina-
tions of materials, they also have much higher depths of field than do
optical microscopes for the magnifications employed.
The Replication Technique

Transmission electron microscopes, however, are not capable of
examining a fracture directly. Rather, it is necessary to employ a
replica of the fractured surface, much the same as that used for micro-
structural studies.
Before a fracture can be replicated, the fracture surface must be
clean and undisturbed. It can never be emphasized too strongly that,
from the time of fracture until the time of replication, the fracture must
be handled very carefully if maximum information is to be obtained
from it. People always seem to have an uncontrollable impulse to see if
two halves of a fracture fit exactly together. Only occasionally is such
information significant, but the process of mating the fractures can
severely damage the microtopographic features of a fracture to the ex-
tent that its usefulness for electron fractographic study may be limited.
In addition to mechanical damage, the surface can be destroyed, or at
least the examination hindered, by chemical and atmospheric attack.
Acceptable methods of coating and protecting fractures have been
developed [15],3 and these should be used when fractographic work is
anticipated. In fact, it is recommended that these procedures be used
whenever possible when a fracture occurs, since unanticipated fracto-
graphic work may be required. Often a fracture is intentionally ex-
posed to a media which may be corrosive, such as certain types of
crack-detection fluids. In these instances, the effects of these media on
the fracture should be thoroughly understood before fractographic
work is to be performed.
For fractures produced intentionally in the laboratory, it is normally
possible to exercise the required precautions of handling and surface
preservation. For failures that occur in service, however, this is seldom
possible; fractures often become severely oxidized, corroded, me-
:~See p. 212.
chanically abraded, or in other ways damaged. As a result, their use-

fulness for fractographic studies is limited. A number of acceptable
procedures for cleaning fractures have been developed, the best of
which is probably continuous replication of the surface with plastic
tape. This procedure removes surface dirt, loosely adhering oxide
scales, and some types of surface films. Other techniques, such as
ultrasonic cleaning, inhibited-acid treatments, and oxide-removal
processes, have been developed3 and can be used when the plastic-
tape replication procedure is not satisfactory.
There are several methods for replicating a fracture for electron
fractography. These are described in detail in a later paper of this
symposium 3 and therefore will only be mentioned here. The prime re-
quirement of any replica is that it duplicates faithfully the finest fea-
tures of the fracture. Numerous replication procedures have been de-
veloped; however, the three most common for electron fractography
are the plastic tape, the direct carbon, and the oxide. Each of these has
particular features which makes them suitable for certain applications.
For example, the plastic tape, which is the most widely used, has the
distinct advantage of being nondestructive to the specimen. Further-
more, replicas of this type can be made practically anywhere with a
minimum of equipment. For example, if a failure should occur in a large
part in the field, it is possible, without moving the part, to prepare a
plastic replica of it. This replica can then be returned to the laboratory
for shadowing, final preparation, and examination. Replicas of this
type can be examined by both optical and electron microscopy, which
provides an added advantage. Figure 3 shows a technician preparing a
FIG. 3-Plastic-tape replication o f a service failure.

FIG. 4 - S h a d o w i n g a plastic-tape replica in a vacuum evaporator.
plastic tape replica of the fracture of a rather large service failure.

After the plastic replica is prepared, it is normally shadowed with some
metal by vapor evaporation as shown in Fig. 4.
Direct carbon replicas have the advantage of better resolution than
the plastic tape and can be particularly advantageous in research.
Their greatest limitation is that only one replica can be made from a
fracture since the specimen is destroyed during the removal of the
replica. Therefore, the fracture is not available later for metallographic
examination or other types of study. The oxide-replication technique,
which has been developed for only a limited number of materials, pro-
vides other information which at times can be useful.
It should be recognized that introduction of the replica raises the
possibility that the structure seen does not truly duplicate the features
of the fracture. Features present on the replica but not the fracture are
called artifacts. Several types of artifacts are encountered in the
replicas, and these can originate either from the replica or from the
specimen. Even when the most careful procedure is used by an ex-
perienced electron microscopist, artifacts still occur. Fortunately, the
types of artifacts that are frequently encountered in electron fracto-
graphic work can be recognized readily and have been described
[18,19].
In addition to normal examinations of replicas by electron micros-
copy, there are various specialized methods of examining replicas
which have been developed and which can provide additional informa-
tion. Foremost of these is stereoscopic viewing. This is accomplished
by taking pairs of photographs of a single area of a replica at two
FIG. 5-Examination of stereo-pairs in a special viewer to observe the three-dimen-

sional characteristics of a fracture.
slightly different angles in the electron microscope. When these pairs

are viewed in a special stereo viewer, as shown in Fig. 5, the three-di-
mensional characteristics of a fracture can be observed and studied.
This type of viewing is particularly useful for studies of fractures, since
the true topographic appearance of the fracture can be seen readily.
Furthermore, the topographic structure of a fracture can be quanti-
tatively analyzed from the stereo pairs by use of a parallax bar. This
type of information has proved useful in both theoretical and applied
studies concerning the micromechanisms of certain types of fractures.
Other specialized methods of replications may be used for fracto-
graphic work, such as replication of a precise area in a fracture and
replication and examination of identical areas of matching halves of a
fracture. Many of these methods are described in a later paper of this
symposium?
Other Methods o f E x a m i n a t i o n
The Electron Microprobe Analyzer

Although optical and transmission electron microscopes are the
principal tools of a fractographer, other instruments are presently
being used to supplement these. The electron microprobe analyzer is
probably the most widely developed and used so far. Since its develop-
ment about 15 years ago [20], the microprobe has found wide applica-
tion in all fields of materials research, including fractography. It has
the unique capability of quantitatively analyzing areas as small as
1 /xm in diameter for their chemical composition. Figure 6 shows a
commercial electron microprobe instrument. The instrument makes

use of a finely focused beam of high-energy electrons which impinge on
a selected point of the specimen. When the electron beam is stopped by
the specimen, a portion of its energy is converted to X-radiation
having wave lengths characteristic of the elements contained in the
analyzed area of the specimen. The wave lengths of the emitted X-rays
can be analyzed by spectrometers, and, since the wave lengths of the
X-rays from each element are peculiar to that element, the identity of
the elements can be determined from measurement of the wave lengths.
Quantitative measurements can be made by measuring the intensities
of the emitted characteristic X-radiations and comparing them with
intensities from standard specimens.
The instrument has been used in conjunction with fractographic
work in a number of ways. For instance, quite often the path of fracture
of a material is dictated by nonmetallic inclusions or other microstruc-
tural features of a material. This has been particularly demonstrated
for fatigue fractures [17]. In many instances, these features are con-
tained on the surface of the fracture, and their identification would be
desirable. Also, many fractures are initiated on inclusions or other
types of impurity defects. It has also been shown that each of the
dimples of ductile fractures is initiated on inclusions. Since most of
these inclusions are microscopic, identification of them in situ is pos-
sible only by electron-probe analysis.
Many electron microprobes are capable of performing beam-
scanning types of analysis, where the beam is electronically deflected
so that it scans a raster on the specimen. While the beam scans, the
FIG, 6 - C o m m e r c i a l electron microprobe analyzer.

FIG. 7 - C o m m e r c i a l scanning electron microscope.
signal from a spectrometer is fed into a cathode-ray tube that scans

synchronously with the electron beam. The resultant image that forms
on the cathode-ray tube shows the elemental distribution in the
specimen. Banerjee [21] and Rice and Racus [22] have shown how this
mode of electron-microprobe analysis can be used to view fractures
and identify constituents.
The Scanning Electron Microscope

Another instrument which is beginning to be employed for studies
of fractures is the scanning electron microscope. 4 A commercial model
of this instrument is shown in Fig. 7. This instrument makes use of a
beam-scanning mode of operation similar to that in electron micro-
probe analysis. A fine electron beam scans the surface of the specimen
in a two-dimensional raster. The back-scattered electrons are analyzed
with a scintillation counter, and the signal from the counter is fed into a
cathode-ray tube which is set to scan synchronously with the electron
beam. As a result, a point-by-point image of the specimen is displayed
on the cathode-ray tube. This instrument has an extremely la4'ge depth
of focus and is therefore quite applicable to fracture studies. The reso-
lution of this instrument depends on several factors, but l:esolutions of
better than 300 ,~ have been achieved. Figures 8 and 9 are photo-
graphs of fractures taken with a scanning electron microscope.
Figure 84 is a photograph of a fracture of magnesium oxide and Fig. 95
4 Cox, B., private communication.
5 Wh~tcomb, P. B., private communication.
FIG. 8-Scanning electron micrograph of a fracture of magnesium oxide (see foot-

note 4) (x4000).
is of a tensile fracture in a soft iron wire. These figures show the type
of result that can be obtained with this instrument. As with transmis-
sion electron microscopy, stereo pairs of a fracture can be obtained and
studied either qualitatively or quantitatively. To obtain the pairs, the
fracture is tilted in the scanning electron microscope between photo-
graphs.
FIG. 9-Scanning electron micrograph of a ductile fracture in an iron wire (see foot-
note 5) (x260).
Reflection Electron Microscopy

Reflection electron microscopy has also been used as a method of
studying fractures. It overcomes the large disadvantage of transmission
electron microscopes in that a fracture can be examined directly;
preparation of a replica step is unnecessary. This type of examination
can be made on most commercial electron microscopes by the addition
of certain accessories. The image obtained is similar in many respects
to that seen with oblique illumination in the optical microscope. The
image is formed by placing the specimen on the side of the electron
beam in an electron-microscope column. The beam is focused so that
it strikes the specimen at a glancing angle. The back-scattered elec-
trons that are emitted are recorded on a fluorescent screen from the
same low angle. Unfortunately, this leads to a severe foreshortening of
the image, and normally only a small area of the specimen can be
focused at any one time. The greatest disadvantage of the reflection
technique is its low resolution. This is primarily due to inelastic scatter-
ing of the electron beam. For this reason, although it might seem like an
ideal method for studying fractures, it has seen only limited applica-
tion to date.
Information Obtained from Electron Fractography

One might inquire what information electron fractographic research
has provided in the relatively few years that it has been employed.
The answer will be found in part by this symposium and is further
demonstrated by the literature and widespread use of the technique.
The micromechanisms of ductile fracture, that is, the initiation,
growth, and coalescence of microvoids, has been confirmed by elec-
tron fractography, and correlations between void size and such things
as stress and material cleanliness have been developed. Also, correla-
tions between void shapes and state of stress have been developed.
Several new models to explain the mechanisms of fatigue fracture have
evolved, and correlations between fatigue striations, load cycles,
striation spacing, and loading conditions have been developed. Also,
rather conclusive experimental evidence regarding initiation mecha-
nisms of fatigue fracture has been acquired from electron fractographic
studies. In brittle fracture, an explanation has been offered for the
cleavage patterns that occur on fractures, and the form of the patterns
has been used successfully to determine such things as fracture direc-
tions and initiation points. The influence of the microstructure of a ma-
terial on its fracture has been demonstrated for many different fracture
modes.
In addition to a research tool, electron fractography also has de-
veloped into an extremely valuable tool for the analyses of service
failures. On many occasions, it has been shown by electron fractog-

raphy that the interpretation of the cause of a failure based only on
some of the more conventional techniques can be in error.
Other information regarding fracture processes that has been derived
from electron fractographic examinations will be discussed in this sym-
posium. It should be realized, however, that the use of electron micros-
copy as a tool for studying fracture is still essentially in its infancy and
there are tremendous opportunities for performing significant research
with it.
References
[1] Zappfe, C. A. and Clogg, M., "Fractography-A New Tool for Metallurgical Re-
search," Transactions, American Society for Metals, Vol. 34, 1945, p. 71.
[2] Martens. A., "The Microscopical Examination of Iron," Z. Ver. dent. Ing., Vol.
21, 1878, p. 11.
[3] Ewing, J. A. and Humfrey, J. C. W., "The Fracture of Metals Under Repeated
Alternations of Stress," Transactions, Royal Society, London, Vol. 200A, 1902,
pp. 241-250.
[4] Ewing, J. A. and Rosenhein, W., "Experiments in Micro-Metallurgy: Effects of
Strain Preliminary Notice," Proceedings, Royal Society, London, Vol. 65, 1899,
pp. 85-90.
[5] Ewing, J. A. and Rosenhein, W., "The Crystalline Structure of Metals," Transac-
tions, Royal Society, London, VoL 193A, 1899, pp. 353-375.
[6] Ewing, J. A. and Rosenhein, W., "The Crystalline Structure of Metals," Transac-
tions, Royal Society, London, Vol. 195A, 1900, pp. 279-302.
[7] Howe, H. M., Metallography of Steel and Cast Iron, McGraw-Hill, New York,
1916.
[8] Arpi, R., "The Fracture Test as Used for Tool Steel," Metallurgia, Vol. 11, 1935,
p. 123.
[9] Shepherd, B. F., "The P-F Characteristic of Steel," Transactions, American
Society for Metals, Vol. 22 (1934), p. 979.
[10] Plateau, J., Henry, G., and Crussard, C., "Quelques nouvelles applications de la
microfractographie," Revue de MetaUurgie, March 1957, pp. 200-216.
[/I] Crussard, C. et al, ' % Study of Impact Tests and the Mechanism of Brittle Frac-
ture," Journal, Iron and Steel Institute, Vol. 183, June 1956, p. 146.
[12] Brandon, D. G., Modern Techniques in Metallography, Van Nostrand, London,
1966.
[13] McLachlan, D., "'Extreme Focal Depth in Microscopy," Applied Optics, Vol. 3,
1964, pp. 1009-1013.
[14] Phillips, A., Kerlins, V., and Whiteson, B. V., "Electron Fractography Handbook,"
Technical Report ML-TDR-64-416, Air Force Materials Laboratory, Research
and Technology Division, Air Force Systems Command, Wright-Patterson Air
Force Base, Ohio, Jan. 1965.
[15] Warke, W. R. and McCall, J. L., "Fractography Using the Electron Microscope,"
Technical Report No. W3-2-65, American Society for Metals, presented at the
1965 Western Metal Congress, 22-26 Feb. 1965, Los Angeles, Calif.
[16] Beachem, C. D. and Pelloux, R. M. N., "Electron Fractography-A Tool for the
Study of Micromechanisms of Fracturing Processes," Fracture Toughness Test-
ing and Its Applications, ASTM STP 381, American Society for Testing and Ma-
terials, 1965.
[17] Pelloux, R. M. N., "The Analysis of Fracture Surfaces by Electron Microscopy,"
Metals Engineering Quarterly, Vol. 5, Nov. 1965, p. 26.
[18] Beachem, C. D., "The Interpretation of Electron Microscope Fractographics,"
Report 6360, Naval Research Laboratory, Jan. 1966.
[19] Beachem, C. D. and Dahlberg, E. P., "Fractography Part XV. Some Artifacts
Possible with the Two-Stage Plastic Carbon Replication Technique," Memoran-
dum Report 1457, Naval Research Laboratory, Sept. 1963.
[20] Castaing, R., thesis, University of Paris, 1951.
[21] Banerjee, B. R., Bingle, W. D., and Blake, N. S., "Scanning Electron-Probe Micro-
analysis in the Metals Industry," Journal of Metals, Oct. 1963, pp. 769-773.
[22] Rice, R. W. and Racus, R. G., "Electron-Probe Examinations of Fracture Sur-
faces," Journal of Applied Physics, Vol. 38, Jan. 1967, pp. 422-423.
W. A. Spitzig, 1 G. E. Pellissier, 1 C. D. B e a c h e m , 2
A. J. Brothers, ~ M. Hill, 4 a n d W. R. Warke ~
A Fractographic Analysis of the

Relationships Between Fracture
Toughness and Surface Topography in
Ultrahigh-Strength Steels
REFERENCE: Spitzig, W. A., Pellissier, G. E., Beachem, C. D., Brothers,

A. J., Hill, M., and Warke, W. R., " A F r a e t o g r a p h i c Analysis of the Rela-
tionships Between Fracture Toughness a n d Surface T o p o g r a p h y in
U l t r a h i g h - S t r e n g t h S t e e l s , " Electron Fractography, ASTM STP 436,
American Society for Testing and Materials, 1968, pp. 17-31.
ABSTRACT: Fracture surfaces of fatigue precracked, notched tension test

specimens of experimental oil quenched and tempered (800 F) 0.45C-Ni-Cr-Mo
steels of different residual sulfur contents, and of two lgNi maraging steels (250
and 300 grades) were investigated, using electron fractography, by a task group
composed of various members of Subcommittee 11 on Fractography of ASTM
Committee E-24 on Fracture Testing of Metals. Sulfur levels of 0.008, 0.016,
0.025, and 0.049 weight per cent in the 0.45C-Ni-Cr-Mo steels corresponded
to fracture toughness values (Kle) of 65.3, 55.6, 51.0, and 42.8 ksiX/in-, re-
spectively. The 250- and 300-grade maraging steels exhibited KI~ values of 68
and 42 ksi lx/FnT., respectively. The investigation was carried out to determine
whether the differences in fracture toughness could be related to changes in
fracture surface topography.
The differences in fracture toughness among the quenched-and-tempered
0.45 carbon alloy steels of different residual sulfur contents could be accounted
for by differences in the frequency (or spacing) of the sulfide inclusions. These
particles seemed to have nucleated microvoids in advance of the crack tip,
which coalesced and resulted in dimple rupture. The particle spacing in each
steel was equivalent to a process zone size based on the Krafft model. A similar
correlation was not found for the two grades of 18Ni maraging steel having dif-
ferent toughness levels, possibly because the mechanism of fracture is more
complex than for the 0.45 carbon alloy steels. Another fractographic feature,
' Senior research metallurgist and senior research consultant, respectively, Applied
Research Laboratory, U.S. Steel Corp., Monroeville, Pa.
z Head, Micromechanical Metallurgy Section, Metallurgy Div., U.S. Naval Research
Laboratory, Washington, D.C. Personal member ASTM.
3Research metallurgist, Metallurgy Applied Research, Materials and Processes
Laboratory, Schenectady, N.Y.
4 Research metallurgist, Research Center, Republic Steel Corp., Cleveland, Ohio.
5 Instructor, Department of Metallurgy, Illinois Institute of Technology, Chicago, I11.
17
namely, the extent of a "stretched" zone between the fatigue-cracked and over-
load regions of the fractures, did appear to show some correlation with the K~e
levels for both classes of steels, and may be related to the size of the process or
plastic zone at the crack tip. This feature, and its significance in terms of plane-
strain fracture toughness, is being more extensively investigated by Subcom-
mittee II of E-24.
KEY WORDS: fractography, fracture toughness, alloy steels, maraging steels,

quenching (cooling), tempering
The important influence of microstructure and attendant strength

level, and of certain impurity elements, on the fracture toughness of
ultrahigh-strength steels has become generally recognized. For some
medium-carbon, quenched-and-tempered alloy steels, it has been found
that there is a strong dependence of plane-strain fracture toughness
(G,c or K,c) on the microstructure and strength developed by temper-
ing at different temperatures [1]. 6 Furthermore, for a given micro-
structure and strength level, the fracture toughness increased with a de-
crease in the combined amounts of the impurity elements, sulfur and
phosphorus, in these steels [1]. Also, the sensitivity of fatigue-crack
growth rates in such steels to changes in testing atmosphere appeared
to be related to fracture toughness and microstructure [2,3]. On the
other hand, for the ultrahigh-strength 18Ni maraging steels, the rather
large difference in fracture toughness between the 250 ksi yield
strength grade and the 300 ksi yield strength grade could not be readily
related to differences in microstructure [4]. Nevertheless, the sensi-
tivity of fatigue-crack growth rate in these steels to changes in testing
atmosphere did appear to be related to differences in fracture tough-
ness and fracture surface topography [3,5].
To gain further insight into the influence of impurities on the fracture
toughness of ultrahigh-strength steels, and to ascertain the micro-
mechanisms of fracture in relation to fracture toughness, a task group
of Subcommittee II on Fractography, of ASTM Committee E-24,
made a fractographic investigation of two types of ultrahigh-strength
steels. The specific objective was to determine, on an interlaboratory
basis, whether differences in fracture toughness among a series of
medium-carbon, quenched-and-tempered, alloy steels having different
sulfur contents, and between the 250-grade and the 300-grade 18Ni
maraging steels, could be related to differences in fracture surface
topography.
Materials and Experimental Work
A series of 0.45C-2Ni-I.5Cr-0.4Mo steels (Table 1) having dif-
ferent levels of sulfur (0.008, 0.016, 0.025, and 0.049 per cent, re-
6The italic numbers in brackets refer to the list of references appended to this paper.
5PITZlG ET AL ON FRACTURE TOUGHNESS AND SURFACE TOPOGRAPHY 19
TABLE I-Chemical compositions of the steels investigated.
0.45C-NI-CR-Mo STEELS
Steel C Mn P S Si Ni Cr Mo AI N O
A ...... 0.43 0.27 0.007 0.008 0.24 2.05 1.48 0.43 0.025 0.007 0.002
B ...... 0,45 0.26 0.007 0.016 0.25 2.04 1.49 0.44 0.025 0,009 0.001
C ...... 0.46 0.25 0.005 0.025 0.24 2.04 1.48 0.43 0.1)27 0,009 0.001
D ...... 0.46 0.24 0.007 0.049 0.22 2.04 1.54 0.44 0.023 0.008 0.001
MARAG[NG STEELS
Steel C Mn P S Si Ni Mo Co Ti Al
250-grade ...... 0.020 0.05 0.006 0.005 0.09 18.35 5.32 7.18 0.32 0.04
300-grade ...... 0.030 0.021 0.003 0.010 0.06 18.53 4.64 8.89 0.69 0.15
NOTE--All compositions in weight per cent.
spectively) were especially prepared from electrolytically refined iron

and high-purity alloying additions, by a combination of vacuum
induction melting and vacuum consumable electrode remelting. The
chemical compositions of the four steels were carefully controlled, so
that they were nearly identical, except for sulfur content; these steels
were remarkably free of nonmetallic inclusions other than sulfide
particles. Selected-area electron diffraction analysis of particles ex-
tracted from fracture surfaces showed that the inclusions were manga-
nese sulfide (MnS). The identification was confirmed by electron-beam
X-ray microprobe analysis of the extracted particles. The specimens
used for fractographic examination were the broken halves of fatigue-
cracked, notched round tension test specimens (1.00 in. in diameter)
which were convenient to test, and which provided valid plane-strain
fracture toughness values (K~,.). The 1.00-in.-diameter notched round
specimens were 6 in. long and contained a machined circumferential
notch with a flank angle of 45 deg. The depth of the machined notch
was 0.115 in. The heat treatment of the four steels consisted of
normalizing at 1650 F for 45 min, austenitization at 1550 F for45 rain,
oil quenching, and tempering at 800 F for 1 h. The tempering treatment
was selected, on the basis of earlier work [I], to provide a microstruc-
ture in which considerable spheroidization of the grain boundary
cementite (FeaC), and recovery of the matrix, had taken place. The
specimens were finish-machined and fatigue-cracked in rotating bend-
ing (to provide a fatigue crack of about 0.04 in. deep) after heat treat-
ment, and tested at room temperature. The rate of loading corre-
sponded to a rate of increase of the nominal net section stress of 10~
psi/min. The K~ values were computed from the notch tensile strengths
(o-,) using the relationship
Kl~ = 0.233o-,X/~D
where D = major specimen diameter. The values are given in Table 2,
together with the smooth tensile properties. The K~ values are
averages of three specimens for the 0.45C-Ni-Cr-Mo steels and of four
specimens for the maraging steels. The deviation from the average
value of Klc was less than ___2.5 ksiV]~, for all the steels.
T A B L E 2 - Mechanical properties o f steels investigated.
0 . 4 5 C - N I - C R - M o STEELS
Sulfur
Level, Yield Strength Tensile Elonga- Reduc-
weight (0.2% off- Strength, tion in tion of Klc
Steel % set), ksi ksi 2 in., % Area, % ksi
A ...... 0.008 211.0 237.6 11.8 48.2 65.3

B ...... 0.016 211.5 237.8 10.0 40.1 55.6
C ...... 0.025 214.2 237.7 11.5 44.4 51.0
D ...... 0,049 217.8 236.0 10.0 41.7 42.8
MARAGING STEELS
Yield Strength Tensile Elonga- Reduc-

(0.2% off- Strength, tion in tion of Kr~
Steel set), ksi ksi 2 in., % Area, % ksi
250-grade ...... 259 267 11.0 44.5 68 ~ 68 b

300-grade ...... 285 293 6.5 31.5 52 ~' 42 ~
a N o t c h e d bend tests (believed to be more accurate).

0 N o t c h e d tension tests,
The two 18Ni maraging steels (250 grade and 300 grade) were com-
mercial steels that were vacuum consumable electrode remelted. Their
compositions are given in Table 1. The heat treatment of these steels
consisted of solution annealing at 1600 F for 1 h, air cooling, and aging
at 900 F for 3 h. The 1-in.-diameter notched round tension test speci-
mens that were used to obtain K~c values, and for subsequent fracto-
graphic investigation, were finish-machined and fatigue-cracked in ro-
tating bending after heat treatment. The Kxc values, and smooth tensile
properties are listed in Table 2. Additional Klc values were obtained for
these same steels from fatigue-cracked notch bend test specimens, and
these values also are given in Table 2. The values obtained for the 250-
SPITZIG ET At ON FRACTURETOUGHNESS AND SURFACE TOPOGRAPHY 21
grade steel by the two methods were identical, but the bend test value
for the 300-grade steel was significantly higher than that obtained from
the notch tension test; in the latter case, the bend test value is believed
to be more accurate, because of possible nonaxiality of loading of this
rather brittle material in the notch tension test. However, fractographic
examinations were made only on the notch tension test specimens.
The replicas for electron fractographic examination of all steels were
prepared by a two-stage plastic-carbon technique, using chromium as
the replica shadowing material.
Results and Discussion

The macroscopic appearance of the fracture surface of the 0.45C
low-alloy steel containing 0.008 per cent sulfur is shown in Fig. 1. The
fracture surfaces of the other steels of different sulfur contents looked
FIG. 1 - F r a c t u r e surface o f 0.45C-Ni-Cr-Mo steel with 0.008 weight per cent sulfur.
about the same. Figure 2 shows the macroscopic appearance of the

fracture surface of the 18Ni(250) maraging steel notch tension test
specimen; the appearance of the 18Ni(300) maraging steel fracture sur-
face was very similar.
Although fractographs were prepared from many areas across the
fracture surface of each steel specimen, only those taken from a region
near the onset of plane-strain fracture instability (A in Figs. 1 and 2),
immediately adjacent to the fatigue-crack, are discussed in this report.
This region should be of most direct significance in any correlation of
fracture surface topography with fracture toughness and microstruc-
ture.
In general, the results of fractographic examination of the series of
0.45C low-alloy steel specimens showed that all of the fracture sur-
faces were completely dimpled, and exhibited about the same degree of
microscopic roughness, as evaluated by stereo-microscopic measure-
FIG. 2 - F r a c t u r e surface of 250-grade mara~,,ing steel.

SPITZIG ET AL ON FRACTURE TOUGHNESS AND SURFACE TOPOGRAPHY 23
FIG. 3 - Fracture surface of O.45C-Ni-Cr-Mo steel with 0.008 pet" cent sulfm" show-
ing area adjacent to.fhtigue crack.
ments. Figures 3 and 4 are fractographs taken from regions adjacent to

the fatigue starter crack of the steels containing 0.008 and 0.049 per
cent sulfur, respectively. The fractographs of the other two steels of
intermediate sulfur contents were very similar to these. The only dif-
ference observed among the fractographs of the four steels was an in-
crease in number, and a corresponding decrease in spacing, of the
rather large, sulfide-nucleated dimples (A in Fig. 4) with increasing
sulfur content. The smaller dimples in Figs. 3 and 4 als0 appeared to
have been nucleated by smaller particles, which resembled the
cementite (Fe:~C) particles in the tempered microstructure of these
steels.
Recent work [6] has shown that the average spacing of the sulfide
inclusions in these steels is equivalent to the process zone size pre-
dicted by the Krafft model for the onset of plane-strain fracture in-
FIG. 4 - F r a c t u r e surface o f O.45C-Ni-Cr-Mo steel with 0.049 per cent sulfur show-
ing area adjacent to fatigue crack.
stability [7]. According to this model, the plane-strain fracture tough-

ness (Kze) is related to the plastic flow properties of a material by
Kic = En 2~V/2~T
where E = Young's modulus, n = strain hardening exponent, and dr =
process zone size. The process zone is a region ahead of the crack in
which tensile instability occurs, resulting in the onset of unstable frac-
turing; it has been shown to be invariant for a given material [7]. F o r
these 0.45C low-alloy steels of different sulfur contents, it seems likely
that the process zone size, and therefore K~c, may be controlled, in
part, by the coalescence of voids that are nucleated by the sulfide
particles.
H o w e v e r , another microscopic feature of the fracture surface was
observed that might be related to the fracture toughness levels of these
steels, namely, a "stretched" region between the fatigue starter crack

and the main overload fracture [8]. Such a region in the fracture surface
of the steel containing 0.008 per cent sulfur is shown in Fig. 5. A region
of this kind could not be found in the steel containing 0.049 per cent
sulfur; either it did not exist or was too narrow to be detected. The
fracture surfaces of the two steels of intermediate sulfur contents were
not examined specifically for this feature. It is possible that this
stretched zone corresponds to a region of crack extension through the
plastic zone that existed at the tip of the fatigue crack just prior to the
onset of the rapid, main fracturing process; the stretching might be in-
volved in the separation process, or perhaps it represents localized
plastic flow in the crack surfaces resulting from the bending that ac-
companies the crack opening. In either case, it is reasonable to think
FIG. 5 - Fracture surface o f O.45C-Ni-Cr-Mo steel with 0.008 per cent sulfur show-
ing area between fatigue crack and overload fracture regions.
that the extent of this stretched region may be related to plastic zone
size, and that it increases with increase in plane-strain fracture tough-
ness and plastic zone size. In this connection, it is interesting to note
that the width of the stretched zone shown in Fig. 5 is about 6 t2m,
which is the size of the process zone for this material computed from
the Krafft model using values of 2.9 • l04 ksi, 0.06 and 65.3 ksi~TnT.
for E, n and K~c, respectively [6]. The significance of this microscopic
fracture surface feature is being explored more extensively by a newly
formed Task Group of Subcommittee II on Fractography of ASTM
Committee E-24.
The fracture surfaces of the 250-grade and of the 300-grade 18Ni
maraging steels exhibited a wide range of dimple size throughout the
fracture surfaces, as shown in the typical fractographs of Figs. 6 and 7,
respectively. Although many inclusions appeared in the fracture sur-
FIG. 6--Fracture surface of 250-grade maraging steel showing area adjacent to

fatigue crack.
SPITZlG ET AL ON FRACTURE TOUGHNESS AND SURFACE TOPOGRAPHY 27
FIG. 7-Fracture surface of 300-grade maraging steel showing area adjacent to

fatigue crack.
faces (for example, A in Fig. 9), there was no obvious correlation be-
tween the inclusion frequency or spacing, dimple size, and the meas-
ured values of plane-strain fracture toughness (K,c). Stretched zones
between fatigue cracked and overload fracture regions were observed
in both steels, as shown in Figs. 8 and 9, and the width of the stretched
zone was greater in the 250-grade steel having the greater plane-strain
fracture toughness. However, for these steels, the measured widths of
the stretched zones were more than an order of magnitude smaller
than the process zone sizes calculated from mechanical property data
for similar steels [3], on the basis of the present Krafft model [7].
Thus, the most readily measured microscopic features of the fracture
surfaces of these two maraging steels (dimple size, or inclusion spac-
ing, and width of stretched zone) have not provided a very satisfactory
explanation for the measured differences in their plane-strain fracture
FIG. 8- Fracture surface of 250-grade maraging steel showing area between fatigue
crack and overload fracture regions.
toughness (KIe values). Possibly, the microscopic fracture mechanism

in these steels is more complex than for the medium-carbon quenched-
and-tempered (800 F) alloy steels previously discussed, in that a
tendency toward intergranular fracture in the maraging steels may re-
duce the importance of the role of microvoid formation and coales-
cence in the fracture process, and thus the significance of dimple size
(or inclusion spacing) as the primary indicator of fracture toughness
level. Likewise, the more complicated fracture mechanism may require
some modification of the present Krafft model in order to permit cal-
culation of a more realistic process zone size for this class of steels.
Summary
This investigation revealed no differences in the type of fracture sur-
face topograPhY of the oil quenched and tempered (800 F) 0.45C-Ni-
Cr-Mo steels of different residual sulfur levels. In general, all the frac-
ture surfaces were completely dimpled, and exhibited about the same
level of surface roughness. There was, however, an increase in num-
ber, and decrease in spacing, of larger dimples with increasing sulfur
content and decreasing plane-strain fracture toughness. This suggested
that the larger dimples had been nucleated by sulfide particles, and that
this mechanism was effective in reducing fracture toughness at the
higher sulfur levels. For each steel of different sulfur content, the sul-
fidebparticle spacing was equivalent to the process zone size calculated
from the measured K~c values and strain hardening exponent (n), using
the Krafft model. Another fractographic feature of possible significance
in relation to fracture toughness was noted, namely, a "stretched" zone
between the fatigue cracked and overload regions of the fracture sur-
face. The width of this stretched zone appeared to increase with in-
creasing plane-strain fracture toughness.
A wide range of dimple size was observed throughout the fracture
FIG, 9 - F r a c t u r e surface of 3OO-grade maraging steel showing area between fatigue

crack and overload fracture regions.
surfaces o f the two 18Ni maraging steels. T h e r e was no o b v i o u s corre-

lation b e t w e e n the inclusion f r e q u e n c y or spacing, dimple size, and the
level o f plane-strain fracture t o u g h n e s s o f these steels. A l t h o u g h the
extent o f the stretched z o n e b e t w e e n fatigue c r a c k e d and o v e r l o a d
fracture regions was greater in the 250-grade steel o f greater toughness,
its width was an o r d e r of magnitude smaller than the p r o c e s s z o n e size
calculated f r o m m e c h a n i c a l p r o p e r t y data on similar steels, using the
Krafft model. T h u s , the fractographic findings did not a c c o u n t for the
difference in fracture t o u g h n e s s o f these steels v e r y satisfactorily, pos-
sibly b e c a u s e o f a m o r e c o m p l e x m e c h a n i s m o f m i c r o s c o p i c rupture,
involving both m i c r o v o i d f o r m a t i o n and c o a l e s c e n c e and intergranular
rupture, in the maraging steels.
References
[f] Baker, A. J., Lauta, F. J., and Wei, R. P., "Relationships Between Microstructure
and Toughness in Quenched and Tempered Ultrahigh-Strength Steels," Structure
and Properties of Ultrahigh-Strength Steels, ASTM STP 370, American Society
for Testing and Materials, 1965, pp. 3-29.
[2] Spitzig, W. A. and Wei, R. P., "A Fractographic Investigation of the Effect of En-
vironment on Fatigue-Crack Propagation in an Ultrahigh-Strength Steel," Trans-
actions, American Society for Metals, Vol. 60, 1967, pp. 279-288.
[3] Wei, R. P., Talda, P. M., and Li, Che-Yu, "Fatigue-Crack Propagation in Some
Ultrahigh-Strength Steels," Fatigue Crack Propagation, ASTM STP 415, American
Society for Testing and Materials, 1967, pp. 460-485.
[4] Baker, A. J. and Reisdorf, B. G., "Strengthening Mechanisms of the 18 Percent
Nickel Maraging Steels," 4th Maraging Steel Project Review, Technical Docu-
mentary Report ML-TDR-64-225, Air Force Materials Laboratory, 1964.
[5] Spitzig, W. A., Talda, P. M., and Wei, R. P., "Fatigue-Crack Propagation and
Fractographic Analysis in 18Ni(250) Maraging Steel in Argon and Hydrogen En-
vironments," presented at The National Symposium on Fracture Mechanics at
Lehigh University, Bethlehem, Pa., June 1967, to be published in Engineering Frac-
ture Mechanics, 1968.
[6] Birkle, A. J., Wei, R. P., and Pellissier, G. E., "Analysis of Plane-Strain Fracture in
a Series of 0.45C-Ni-Cr-Mo Steels with Different Sulfur Contents," Transactions,
American Society for Metals, Vol. 59, 1966, pp. 981-990.
[7] Krafft, J. M., "Correlation of Plane Strain Crack Toughness with Strain Hardening
Characteristics of a Low, a Medium, and a High Strength Steel," Applied Materials
Research, Vol. 3, 1964, pp. 88-101.
[8] Beachem, C. D. and Meyn, D. A., "Illustrated Glossary of Fractographic Terms,"
Section 2, U.S. Naval Research Laboratory Memorandum Report 1547, June 1964.
DISCUSSION
W. R. W a r k e ~ (written d i s c u s s i o n ) - I n the report, the " s t r e t c h e d

z o n e " is interpreted in terms of " c r a c k e x t e n s i o n t h r o u g h the plastic
z o n e that existed at the tip o f the fatigue c r a c k j u s t prior to the o n s e t o f
the rapid, main fracturing p r o c e s s . " M y recollection o f the e x a m i n a t i o n
Instructor, Illinois Institute of Technology, Chicago, Ill.
DISCUSSION ON FRACTURE TOUGHNESS AND SURFACE TOPOGRAPHY 31
of these stretched zones by stereoscopic electron fractography is that

the stretching involved blunting of the crack tip as well as extension.
The extent of modification of the crack tip through stretching was ap-
proximately the same in the vertical direction (blunting) as in the hori-
zontal direction (extension). This observation is in keeping with current
views of deformation at the tip of a propagating fatigue crack during the
increasing portion of the load c y c l e r ,'3'4 Loading of a fatigue precracked
specimen to failure is basically the same as a quarter of a fatigue cycle
so similarity of the crack tip deformation process would be expected. 1
wonder if Mr. Spitzig would care to comment on the influence the ob-
served blunting would have on the fracture mechanics interpretation
of the significance of the stretched zones.
Author's closure- Additional work 5,6 has shown that there definitely
appears to be a correlation between the extent of the stretched zone
and the process zone size. However, the actual physical significance of
the stretched zone is not obvious at present, an'd, therefore, it is not
possible to evaluate the effect of blunting or other plastic deformation
processes on the significance of this region for fracture mechanics in-
terpretation.
2 Jacoby, G., "Application of Microfractography to the Study of Crack Propagation
Under Fatigue Stresses," A G A R D Report 541, 1966.
:~Laird, C., "The Influence of Metallurgical Structures on the Mechanism of Fatigue
Crack Propagation," Fatigue Crack Propagation, ASTM STP 415, American Society
4 McMillan, J. C. and Pelloux, R. M. N., "Fatigue Crack Propagation Under Pro-
grammed and Random Loads," Boeing Scientific Research Laboratories Document
D1-82-0558, Fatigue Crack Propagation, ASTM STP 415, American Society for Test-
5 Spitzig, W. A., "Plane-Strain Fracture in 0.45C-Ni-Cr-Mo Steel," submitted to
Transactions, American Society for Metals.
Spitzig, W. A., "Correlation Between Microstructural Features and Plane-Strain
Fracture Toughness in an Ultrahigh-Strength Steel," submitted for presentation at the
71st Annual Meeting of ASTM, San Francisco, Calif., June 1968.
H. C. Burghard, Jr. 1, and N. S. S t o l o f f 2
Cleavage Phenomena and Topographic

Features
R E F E R E N C E : Burghard, H. C., Jr., and Stoloff, N. S., "Cleavage P h e n o m -

ena and Topographic F e a t u r e s , " Electron Fractography, ASTM STP 436,
ABSTRACT: Cleavage fractures have been widely studied in body-centered-

cubic and certain hexagonal-close-packed metals and alloys as well as ionic
crystals. It is shown that the theoretical treatments due to Stroh, Cottrell, and
Petch have successfully accounted for the effects of yield strength, grain size,
and mode of deformation on crack nucleation and propagation. The results of
these studies are summarized, and their application to the prediction of the
ductile-to-brittle transition temperature is described. In addition, theoretical
studies which have led to the prediction of cleavage planes and to an explana-
tion of the influence of dispersed particles are reviewed.
Fractographic studies have established topographic features characteristic
of cleavage fracture; namely, steps and river patterns, cleavage tongues, and
herringbone patterns. Examples of these features are presented, and factors
associated with their formation are described. It is shown that steps and fiver
patterns are formed by secondary cleavage, slip, or plastic fracture and are in-
fluenced by crystallographic imperfections and microstructural discontinuities.
Tongues and herringbone patterns are shown to be associated with deformation
twins formed ahead of the advancing crack. In addition, a discussion of "cleav-
age-like" features observed in quenched and tempered steels, face-centered-
cubic materials, and in certain environmentally induced fractures is presented.
KEY WORDS: cleavage (crystallographic), fracture, fractography, electron

microscopy, brittle fracture, crack nucleation, crack propagation
Fracture of solids may occur by direct separation along low-index

crystallographic planes by the process known as cleavage. Cleavage
phenomena have been known and studied for many years; it is only in
the past 10 to 15 years that the methods of dislocation theory have
been applied to consideration of the mechanism of cleavage fracture.
The past decade has also witnessed the wide application of fracto-
graphic techniques to reveal fine details of the fracture process which
1 Senior research engineer, Department of Materials Engineering, Southwest Research
Institute, San Antonio, Tex.
Associate professor, Department of Materials Engineering, Rensselaer Polytechnic
Institute, Troy, N.Y. Personal member ASTM.
32
Copyright* 1968 by ASTM International www.astm.org
BURGHARD AND STOLOFF ON CLEAVAGE PHENOMENA 33
may be used to develop realistic models of cleavage fracture. This

paper discusses the mechanism of cleavage crack nucleation and
propagation in solids, shows that fractographic observations are com-
patible with the proposed models, and finally discusses some charac-
teristics of cleavage phenomena that only recently have been observed.
In addition, attention will be directed to certain cases in which
topographic features resembling those associated with cleavage have
been observed in materials generally considered to be immune to
cleavage.
The conditions under which the cleavage mechanism is operative
vary with the particular material and the type of loading. In metals and
metallic alloys, cleavage is generally limited to the body-centered-
cubic (bcc) and hexagonal-close-packed (hcp) materials. In tension,
cleavage may occur after significant amounts of plastic deformation (as
measured over the entire specimen), while cleavage fracture induced
by impact loading frequently occurs with little or no plastic deforma-
tion.
There are several materials which undergo a ductile-to-brittle transi-
tion in simple tension but do not fail by cleavage at low temperatures.
Examples are cadmium [1]3, magnesium [2], and Ti-0.25Oz alloys [3].
In addition, there are many examples of low-energy failures in high-
strength steels and aluminum alloys in which there is no tendency to
cleavage failure. These examples are listed here to emphasize the fact
that cleavage and brittleness are not synonymous. Brittle failures can
occur by mechanisms other than cleavage, and cleavage may be in-
volved in the fracture of certain materials which exhibit relatively high
ductility in simple tension. These factors and the recent advances in
the understanding of fracture processes emphasize the importance of
precise definitions of the terms used to describe and classify fracture.
In this regard, Burghard and Davidson [4] have suggested that the
specification of three characteristics is necessary for an unambiguous
classification of fracture, that is, the basic mechanism involved (plastic
fracture, cleavage, or fatigue), fracture mode (transgranular or inter-
granular), and the macroscopic aspects (ductile or brittle). With
reference to the particular basic mechanism of interest here, cleavage
may be defined as a low-energy, rapid fracture process producing a
fracture surface coincident with low-index crystallographic planes.
Theoretical Considerations
Mechanisms of Cleavage Crack Nucleation

Cleavage cracks in nominally single-phase, polycrystalline metals
and alloys originate at blocked slip or twin bands. The obstacle may be
The italic numbers in brackets refer to the list of references appended to this paper.
a grain or subgrain boundary, another twin, or an inclusion particle.

Cottrell [5] has proposed that intersecting slip bands also may con-
stitute additional obstacles which form sites for crack nucleation. This
model is generally referred to as the "double pileup model." Cracks
originating at intersecting slip bands are frequently observed in
Fe-3Si and in ionic crystals. A crack may form along the slip plane, as
in zinc, or at some large angle to the slip plane, as in iron.
Direct observation of crack formation at blocked slip bands has been
provided by experiments on ionic solids and certain metals. Ku and
Johnston [6] have used birefringence of transmitted polarized light to
reveal stress concentrations, and eventually crack nucleation, at
blocked slip bands in magnesium oxide (MgO) bicrystals (see Fig. 1).
Similarly, crack formation at blocked slip bands in silver chloride
(AgCI) immersed in an aqueous sodium chloride (NaC1) solution has
been noted by Westwood et al [7]. In the case of metals, Wronski and
Fourdeux [8] have observed cracks to form along arrays of blocked dis-
locations on { 110} slip planes in thin foils of polycrystalline tungsten.
Nucleation of cleavage cracks at twin intersections has been noted
in many metals; a typical example for zinc is shown in Fig. 2. Cleavage
crack formation along slip planes which are blocked by tilt boundaries
is a common characteristic of hcp metals and alloys and is illustrated
in Fig. 3. If the alloy is distinctly two-phase, an additional mode of
crack formation is provided by the possibility of the second phase de-
veloping a crack which can extend into the matrix. Cleavage fractures
originating at grain boundary inclusions as small as one micron in
diameter have been observed in Fe-0.005C alloys [9] (see Fig. 4). In
pearlitic steels, cracks originate in the cementite lamellae and then
propagate into the adjacent ferrite. In more complex alloys, there may
be a variety of possible nucleation sites, any of which could give rise to
cleavage fracture.
The dislocation models of cleavage fracture at blocked slip or twin
bands were developed for iron and steels, and have been applied to a
variety of alloys with relatively simple microstructures. Perhaps the
most successful model to date is that presented by Stroh [10], Cottrell
[5], and Petch [11]. This model is based on the concept that a shear dis-
placement arising from dislocation pileups gives rise to stress concen-
trations sufficient to nucleate a crack.
Cottrell [5] calculated the fracture stress, o>, of a perfect crystal in
the presence of an isolated wedge of n coalesced dislocations of Bur-
gers vector b:
(2~,)
where 3' is the surface energy associated with the crack. The wedge
may be formed, in the case of iron, by a/2{ 111 ) dislocations gliding on
FIG. 1 -Stress concentrations generated at the tips of blocked glide bands in mag-
nesium oxide as revealed by birefringence. After Ku and Johnston [6]. (•
FIG. 2 - C l e a v a g e crack nucleated at the intersection of twin bands in a zinc single

crystal compressed 9.6 per cent at -196 C (•
36 ELECTRON FRACTOGRAPHY
F I G . 3 - Cleavage crack nucleated at a tilt boundary in Cd-3.7Mg alloy tested in com-

pression at - 1 9 6 C. After Stoloff and Gensamer [1]. (x250).
F I G . 4 - C l e a v a g e cracks nucleated at grain boundary carbide particles in Fe-

0.005C. After McMahon and Cohen [9]. (x200).
intersecting {110} planes, and interacting to form sessile (100) dis-

locations through the reaction:
[111]+ [ T I l l = a[001] ......................... (2)

a 2
Since this theory has been widely used for several years, and the de-
tails of the derivation are provided in many textbooks on mechanical
behavior, we shall present only the final result of this model. It was
calculated that cleavage fracture should occur at the yield stress, o-y,
when the following condition is satisfied:
o-vk~d v2 >1 f i G y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (3)
The yield stress, o-u, is related to the grain diameter, d, through the
following relation:
0% = 0-4 + k u d -112. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4)
Therefore, ky is a measure of the contribution of grain boundaries to the

strength of a polycrystal, and/3 is a constant expressing the ratio of
shear to normal stress in the test specimens. G is the shear modulus. In
Eq 4, 0-4 is associated with the lattice friction which resists dislocation
motion.
Substituting Eq 4 into Eq 3, an equivalent fracture criterion is ob-
tained in terms of o-~ and k~.
(o-~d 1/2 § ky)ku ~ /3G 7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . (5)
This equation correctly predicts that bcc and hcp materials with high
friction stress, high grain boundary resistance to slip propagation, or
large grain size tend to be susceptible to cleavage fracture.
Ductile-to-Brittle Transition
The analysis outlined in Eqs 1 through 5 also predicts a ductile-to-

brittle transition at constant test temperature with increasing grain
size. The transition occurs at a critical grain size d * ( T ) which varies
with test temperature, T. The more usually studied shear-to-cleavage
transition occurring at low temperatures in steels also has been ra-
tionalized by identifying o-~ as the most temperature-dependent term
in Eq 4. For example, Petch [12] has shown that, if:
o4 = B e - c r . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (6)
where B and C are numerical constants established by experiment,
then the relation between transition temperature, To, and metallurgical
variables, obtained by substituting Eq 6 into Eq 5 is
1
Te = ~ In [ ( B k ~ d * ' / z ) / ( f l G y - ku-)] .................... (7)
The relation between transition temperature and grain size reduces

to:
Tc = D I n (d .1/2)................................ (8)
where D is a constant.
This equation has been experimentally verified for several ma-
terials, and data obtained for an iron-base alloy are shown in Fig. 5 [13].
Johnston et al [14] have recently suggested that hcp materials which
exhibit few slip systems (for example, zinc), as well as bcc metals
which slip on several systems but in which cross-slip is difficult, will
tend to have a high value of k~, and will therefore be more susceptible
to cleavage fracture than materials with many degrees of freedom to
slip. Therefore, iron-silicon alloys, which deform primarily by { 110}
slip, with little cross-slip onto other planes, tend to be more brittle
than unalloyed iron, which slips also on {l 12} and {123} planes, with
copious cross-slip among the planes. Moreover, in many alloys, the
number of active slip systems and ease of cross-slip tend to decrease
with decreasing temperature, constituting an additional reason for an
increase in the tendency for cleavage with decreasing temperature. As
a result of these factors, one may expect a more complex relation be-
tween transition temperature and microstructural parameters than is
suggested by Eq 8.
500
U
O
400
:E
i,Ll
I.-- 0
z 300 -
O
I---
z
e~
I--
I I I l I
20 30 40 60 80 100
GRAIN SIZE, MICRONS
FIG. 5 - E f f e c t of grain size on the ductile-to-brittle transition temperature in Fe-

49Co-2V. After Jordan [13].
Crack Propagation
The question of whether or not a cleavage crack, once nucleated, can
grow catastrophically has been the subject of much analysis. Clearly,
in certain steels, crack propagation requires a higher stress than crack
nucleation. In such cases, Hahn et al [15] have noted stable micro-
cracks of a length corresponding to one grain diameter. These cracks
are nucleated at the yield stress, propagate through a single grain, and
then are held up by grain boundaries. On the other hand, for pure
single-phase solids tested over a wide range of temperature, crack
nucleation was noted to be the critical event, and no microcracks are
observed [14].
The CottreU-Petch model does not explicitly take into account the
fact that cleavage crack growth will be more difficult than nucleation
only if the effective fracture surface energy, ye (which includes the
work of plastic deformation), is increased appreciably above the true
elastic energy, ys, appropriate to the nucleation stage. Neither Cottrell
nor Petch has taken into account the possibility of increase in surface
energy with crack length. Rather, the surface energy has been treated
as a disposable parameter which generally is calculated from Eqs 3
and 5 to be at least an order of magnitude greater than the true elastic
surface energy. Another disadvantage of the Cottrell-Petch approach is
concerned with the lack of means to take into account the role of pre-
cipitates in crack nucleation. Finally, there is quantitative disagreement
between fracture surface energies calculated from the theory and di-
rectly measured true surface energies. Hahn and Rosenfield [16] have
suggested that the excessively large calculated surface energy is not a
consequence of plastic relaxation due to cross-slip, as suggested by
Johnston et al [14], but rather originated from an incorrect estimate of
the number of coalesced dislocations and the effective shear stress
appropriate to the Cottrell double-pileup model. In spite of these reser-
vations, this theory stands as the only framework with which to discuss
the influence of metallurgical variables on cleavage fracture.
tnfluence of Dispersed Particles

When cleavage in low-carbon iron is initiated within carbide particles
located at grain boundaries, the probable cause is pileup of disloca-
tions in the adjacent ferrite grains. If the particle-matrix interface is
weak, the interface tends to crack first, but the size of the crack is
limited by the size of the particle. The crack can propagate into the
matrix only if the Griffith criterion (for plane strain) is satisfied:
O'F >I { (2Ey)/[Trc(1 -- v2)]}'~z ........................ (9)
where c, the crack length, is proportional to the particle radius. This
condition predicts a critical particle size for propagation of a crack

through the matrix. The double pileup model of Hahn and Rosenfield
[16] predicts that fine, uniform dispersions of hard particles can
strengthen without impairing ductility and can even lower the ductile-
to-brittle transition temperature. This prediction is based on the
premise that the particles decrease the maximum pileup length from
the grain diameter to the order of the interparticle spacing, and there-
fore the stress concentration is reduced.
Examples of instances where fine, hard particles have enhanced
strength without increasing the tendency towards cleavage are as
follows: AgC1-A1203 [I 7], 4340 steel (200 ,~ cementite particles) [18],
and W-ThO2 alloys [19]. However, in the discussion of the work on
W-ThO2 alloys, Rosenfield and Hahn point out that there is no un-
ambiguous experimental evidence that the improved ductility arises
from smaller dislocation arrays. Since coherent precipitates, which
can be cut by dislocations, strengthen the matrix without decreasing
pileup length, alloys containing this type of particle should be no more
resistant to cleavage than the matrix.
Prediction o f Cleavage Planes

Since one of the most fundamental characteristics of cleavage frac-
ture is that the crack propagates on a low-index crystallographic plane,
it is of interest to predict the cleavage plane from atomic binding prin-
ciples. Gilman [20] has examined the problem, basing his calculations
on the concept that the cleavage plane is the plane of minimum surface
energy. By assuming a sinusoidal attractive stress field between two
atomic surfaces, and equating the surface energy to the work done in
separating the two surfaces, Gilman developed the following relation
between surface energy y and lattice constants:
yhet -= (Elyo)(ao/rc)2............................. (10)
where E is the elastic stiffness perpendicular to the designated plane
hkl, Yo is the separation between these planes, and a0 is the diameter of
atoms in the cleavage surface, measured parallel to the surface. This
calculation correctly predicted the cleavage plane (plane of lowest
energy) in 16 crystalline solids but failed to predict the primary
cleavage plane of beryllium, iron, and tungsten. The latter two ele-
ments were predicted to cleave on { 110} planes, in contrast to the
{ 100} cleavage generally observed. The reason for the discrepancy
may lie in the energy density of broken bonds, which is greater for
{110} than {100} cleavage. It is interesting to note, however, that
fracture along {110} slip planes has been observed in molyhdenum
[21], Fe-3Si [22], tungsten [8] and other bcc metals, suggesting that
there is a greater tendency for { 110} cleavage in the bcc structure than
BURGHARD AND $TOLOFF ON CLEAVAGE PHENOMENA 41
has been generally recognized. The specific cleavage planes for various
classes of solids are listed in Table 1.
TABLE 1- Cleavage planes in solids.
Crystal Primary Secondary

Structure Material Cleavage Plane Cleavage Plane
bcc .................... Fe 001 .....

Cb 001 110
W 001 110
Mo 001 110
Fe-3Si 001 110
hcp ................... Zn 0001 .....
Cd 0001 .....
Mg 10]-1 1124
aTi-0.502 1232 10]-0, 1230
Diamond ............ Si 111 .....
Ge 111 .....
Ionic solids ......... NaCI 100 110
KCI 100 110
LiF 100 110
Zinc blende ......... ZnS 1I0 .....
Fluorite .............. CuF2 111 .....
UO2 111 .....
ThO2 111 .....
Topography of Cleavage Fracture

Consideration of the factors discussed in the previous section
emphasizes that the unique, distinguishing feature of cleavage fracture
is the definite association of the fracture path with particular, low-index
crystallographic planes. This crystallographic dependence results in
characteristic topographic features, both on a macro- and microscopic
scale. T h e fracture surfaces created by cleavage in polycrystalline ma-
terials are generally made up of a multitude of flat facets, each repre-
senting the particular cleavage plane for a given grain. It is these flat
facets, and their contribution to the high reflectivity of cleavage frac-
ture surfaces, that have long served as a basis for classification of
fracture from macroscopic examinations.
Based on the single factor of a definite crystallographic dependence,
cleavage fracture would be expected to produce perfectly fiat, feature-
less facets. Real materials, however, exhibit a high degree of sub-
microscopic imperfections, and metallic crystals are subject to a
definite, though limited, amount of plastic deformation in the im-
mediate vicinity of a cleavage crack. T h e latter point is confirmed
by observations of regions of high dislocation density ahead of the tip
of a cleavage crack [23], Fig. 6, and in certain cases by the presence
FIG. 6-Plastic deformation in immediate vicinity of stopped cleavage crack in

Fe-3Si. After Tetelman [23]. (xlSO).
of slip markings on cleaved surfaces [24]. These factors, coupled with

microstructural features such as second-phase particles, inclusions,
grain boundaries and subgrain boundaries, influence the propagation of
a cleavage crack and result in the formation of characteristic fine-scale
surface features within individual cleavage facets.
Cleavage Steps in River Patterns

As a cleavage crack propagates through a crystal, it is most often
broken into a set of parallel cracks by interaction with imperfections
and microstructural features. Thus, gross crack propagation may be the
net result of the simultaneous propagation of individual crack seg-
ments on sets of parallel planes. As the individual segments approach
one another (and possibly overlap), the segments join by fracture of
the connecting ligament, producing steps in the fracture surface. These
steps are generally observed to converge in the direction of local crack
propagation, either cancelling or reinforcing each other to produce the
familiar fiver patterns on individual facets.
Comprehensive investigations of the nature of steps observed in
cleavage fractures have been carried out by Berry [25] and Low [26].
These investigations established that the steps within a single cleavage
facet may be attributed to one or more of the following factors: inter-
section of screw dislocations with the cleavage plane, secondary
cleavage, shear, secondary fracture on a twin-matrix interface or, in
the case where considerable overlap of two crack segments occurs,

deformation and necking-down of the interconnecting ligament. Since
a number of processes may be involved in the formation of cleavage
steps, facets exhibiting' wide variations in the appearance of steps and
river patterns are observed. Examples illustrating the extremes in ap-
pearance of steps and river patterns are given in Fig. 7.4 Steps formed
by secondary cleavage, shear, or twin-interface separation may be ex-
pected to appear as distinctly resolvable subfacets (Fig. 7a), while
those associated with the formation of flaps or extensive local deforma-
tion would appear as heavy lines with less resolvable detail (Fig. 7b).
Since steps and river patterns result from the division of a crack
into parallel segments, they may be expected to originate at regions of
mismatch (grain boundaries or subboundaries). Figure 8 illustrates a
case where new river patterns initiated at the point where the crack
crossed a boundary. In cases where the crack crosses a low-angle
boundary, an increase in step density occurs. Propagation of a cleavage
crack across a high-angle grain boundary usually requires the initiation
of a new crack in the second grain, resulting in the formation of new
sets of rivers or steps.
The microstructure of the material also exerts a marked influence on
the appearance of steps and river patterns formed in cleavage fracture.
Low [26] has reported that in a 0.5 per cent carbon steel with a
pearlitic or upper bainitic structure, the fracture path follows the
cleavage plane of the ferrite through several pearlite or bainite colonies.
For the same steel with a tempered lower bainitic or tempered mar-
tensitic structure, many small cleavage facets were noted. In the latter
case, the facets were reported to be related to the bainite or martensite
needle size. Irvine and Pickering [27] have studied the topography of
impact fractures in low-carbon bainitic steels heat treated to various
strength levels. In this study, it was noted that, at the higher strength
levels, the fracture surfaces exhibited smaller river patterns and more
irregular tearing than in the structures corresponding to the lower
strength levels. Some alteration of the direction of crack propagation at
bainitic boundaries was noted, and more marked changes were ob-
served at prior austenite boundaries.
Recent studies utilizing electron fractographic techniques have pro-
vided numerous additional examples of steps and river patterns in
cleavage fractures [28-30]. These examples have shown that, while
differences in fine-scale features may exist, steps and river patterns are
a characteristic feature of cleavage fracture, and specific information
related to this mechanism may be gained from such features.
As mentioned earlier, steps and river patterns converge in the direc-
4 In Figs. 7, 9, 10, and 12, the block arrow in one corner indicates the macroscopic
crack propagation direction.
(a) Molybdenum: --320 F. Two-stage plastic-carbon replica (x5000).

FIG. 7-Cleavage facets observed in low-temperature impact fractures of molyb-
denum and steel.
(b) A 302B Steel: -98 F. Long arrow indicates local direction of crack propagation.
After Burghard and Davidson [4]. Two-stage plastic-carbon replica (x5000).
FIG. 7-Continued
FIG. 8 - River patterns on cleavage surface o f ordered Fe-49Co-2 V. After Johnston

et al [14]. (x30,O00).
tion of crack propagation within individual facets. Thus, these mark-

ings serve as a reliable indication of the local or microscopic direction
of crack propagation. In numerous studies, it has been observed that
the microscopic direction of crack propagation does not necessarily
correspond to the macroscopic direction. Individual cleavage facets
may show directions of propagation markedly different from the
macroscopic direction as illustrated in Fig. 9. These observations indi-
cate that the propagation of a cleavage crack in a polycrystalline aggre-
gate occurs by simultaneous or near simultaneous fracture of several
FIG. 9 - F r a c t u r e surface topography of A 302B steel impact specimens broken at

-98 F. Arrows indicate local direction of crack propagation in individual facets. After
Burghard and Davidson [4]. Two-stage plastic-carbon replica (x2500).
individual grains, each with an independent direction of propagation,

and a joining of the individual cracks to form a general, macroscopic
crack front.
Tongues and Herringbone Patterns

In certain metals and alloys, fine-scale fracture surface features
generally described as "tongues" have been observed to be uniquely
associated with cleavage fracture. These features consist of protrusions
from or intrusions into the general fracture plane and are made up of
flat facets inclined to the cleavage plane and terminating at a sharp step.
Cleavage tongues were observed in impact fractures of a pearlite-free
steel as early as 1956 [31]. Examples of cleavage tongues, observed in
an impact fracture of a ferritic stainless steel, are shown in Fig. 10. It
should be noted that these two examples represent observations on
fractures occurring at widely different temperatures (70 and 264 F). In
this case, the occurrence of tongues was more predominant in the 70 F
fracture than in the specimen broken at 264 F. Berry [25] suggested
that the inclined facets of tongues represented fracture along {112}
planes corresponding to the interface between the matrix and the defor-
mation formed ahead of the advancing crack. In a written discussion of
Berry's work, Crussard et al also presented examples of well-defined
cleavage tongues in Fe-3Si crystals and concluded, from observations
of shadowing effects, that these features were associated with the
twin-matrix interface [25]. Cleavage on twin-matrix interfaces has also
been observed in cadmium tested in liquid gallium [32], 5 magnesium
[33], and Fe-3Si [22]. More recently, Beachem [34] has observed dis-
tinct tongues in cleaved iron and has established that on two mating
fracture surfaces, protruding tongues on one surface are matched by
corresponding depressions on the opposite surface. In addition, the
orientation of individual tongues relative to crystallographic directions
was established by etch-pitting fracture surfaces [35]. These observa-
tions provide definite evidence that tongues are formed when a cleav-
age crack (propagating along an { 100) plane) is diverted along the
{112} twin-matrix interface of an intersecting twin while the main
crack propagates around the twin. The step bounding the tongue on
three sides is formed by deformation and fracture of the ligament be-
tween the two local cracks. This final stage results in the rejoining of
the diverted crack segment with the main crack which continues to
propagate on the main cleavage plane. To date, observations of well h
defined cleavage tongues have been reported in the fracture of
Fe-3Si [25], iron [34], low-carbon steel [31], chromium-base alloys
[36], and in Type 430 stainless steel (see Fig. 10).
In certain particular cases of cleavage fracture, topographic features
referred to as "herringbone patterns" have been observed. These
features are characterized by two groups of lines and narrow inclined
facets (sometimes including a few well-defined, isolated tongues)
oriented at 90 deg to one another and separated by a narrow, flat strip
bisecting the 90-deg angle. Cleavage herringbones, nearly identical in
appearance, have been observed in tungsten [37], chromium-base
alloys [36], Fe-25Cr-3A1 alloy [38], and in Type 430 stainless steels. A
5 C a d m i u m does not cleave w h e n tested in air, but liquid metal e n v i r o n m e n t s induce
cleavage on slip planes and twin interfaces.
FIG. 10-Cleavage tong,ues in impact fractures o f Type 430 stainless steel.
49
,50 ELECTRONFRACTOGRAPHY
typical cleavage herringbone, observed in the last case cited above, is

shown in Fig. 11. The relative orientation of the various components in
herringbone patterns has led Beachem to conclude that the narrow cen-
tral strip corresponds to cleavage crack propagation on one of the
{ 100} planes in one of the (100) directions and that the fracture on
each side is formed by {100}(110} and {112}(110) twin interface
cleavage mixed with considerable plastic fracture [38].
The cleavage tongues and herringbone patterns described above
constitute closely related topographic features in that both are
attributed to the interaction of an advancing cleavage crack with
deformation twins. As such, these particular features serve as evidence
of the nature of plastic deformation accompanying cleavage fracture
in certain materials. Because of their dependence on the presence of
deformation twins, it is reasonable to expect that both features would
be observed in fractures in a single metal or alloy. This expectation is
borne out in the observations of impact fractures of Type 430 stainless
steel, see Figs. 10 and 11. However, no cases of herringbone patterns
in cleaved iron or well-defined tongues in tungsten have been published
to date.
FIG. 11 -Herringbone pattern in impact fracture of Type 430 stainless steel, Two-
stage plastic-carbon replica (x2500).
BURGHARD AND $TOLOFF ON CLEAVAGE PHENOMENA 51
Mixed Fracture
The foregoing discussion has considered cleavage fracture inde-
pendently without regard to the possibility of simultaneous involve-
ment of two distinct fracture mechanisms in a single fracture. Al-
though there are cases in which fracture occurs completely by cleav-
age, intimate mixtures of plastic fracture and cleavage are frequently
observed. This latter situation is particularly true of fractures in ma-
terials exhibiting a ductile-brittle transition. In impact fractures of low-
alloy steels, tested in the transition temperature range, topographic
features associated with both cleavage and plastic fracture have been
observed in the central, flat portion of the fracture surfaces [27,31,39].
In each case, the degree of plastic fracture, as indicated by the rela-
tive area of dimpled structure, increased with increasing temperature.
Similar observations have been made on impact fractures in Type 430
stainless steel, two examples of which are shown in Fig. 12. This
fractographic evidence of the simultaneous involvement of the two dis-
(a) Test temperature 72 F. Note patch of dimples (oblique arrows), facet with river
pattern (horizontal arrows) and cleavage tongues (vertical arrows). Two-stage plastic-
carbon replica (•
FIG. 12-Regions of mixed fracture in Type 430 stainless steel impact specimens.
(b) Test temperature 90 F. Note large dimples and stretched surface (arrows) ad-
jacent to cleavage facet with river patterns. Two-stage plastic-carbon replica (x2500).
FIG. 1 2 - C o n t i n u e d
tinct fracture mechanisms supports the conclusion drawn earlier that

the interdependence of transition temperature and microstructural
parameters in polycrystalline materials may be expected to be more
complex than the relations derived from simplified models (see Eqs 7
and 8).
Quasi-Cleavage
The fracture surfaces of quenched and tempered steels often exhibit
flat facets resembling those observed in cleavage. These facets are con-
siderably larger than the microstructural features of tempered mar-
tensite and are generally transgranular with respect to prior austenite
grains. Because of the highly complex microstructure involved in
these Cases, no definite relationship has been established between the
orientation of these facets and any specific crystallographic planes. As
a result, the fracture process producing these particular topographic
features cannot be considered as true cleavage. In view of the above
factors, the term "quasi-cleavage" was coined to describe this particu-
lar class of topography in quenched and tempered steels until such
time as the mechanisms involved in their formation were definitely es-

tablished. A general definition for quasi-cleavage has been given by
Phillips et al [30] as: "A fracture mode resembling cleavage in that it
produces planar or nearly planar facets but differing from cleavage in
that the facets are not known to be parallel to cleavage planes."
Beachem [40] has noted that quasi-cleavage facets in quenched and
tempered steels usually exhibit river patterns and tongues similar to
those observed in true cleavage but that, in contrast to true cleavage,
the fracture origin is usually located within the boundaries of the facet.
In addition, these facets often exhibit tear ridges intermixed with the
river patterns and tongues. These latter features are not observed in
true cleavage and serve to distinguish the two classes of fracture sur-
face topography.
Beachem has also suggested that the growth and coalescence of com-
pletely submerged cracks (in or near the same plane) could result in
features corresponding to those observed in quenched and tempered
steels (see Ref 40). No further work related to the nature of the
mechanisms involved in this fracture process has been published to
date. Thus, it seems that, until further investigations are carried out,
the formation of quasi-cleavage facets in quenched and tempered steels
should be considered to result from an undefined modification of the
cleavage mechanism associated with the complex microstructural
characteristics of these materials.
In general, cleavage cracks are not observed in single phase face-
centered-cubic (fcc) alloys, presumably because of the multiplicity of
slip systems, high dislocation mobility and ease of cross-slip. How-
ever, in certain cases, fiat facets with fine-scale features resembling
those of cleavage facets have been observed. Similar features have also
been noted in certain hcp materials normally considered immune to
cleavage. Although these facets exhibit steps and partial river patterns,
and are frequently feather-shaped, no characteristic appearance com-
mon to a particular material or to particular fracture conditions has
been established. The only specific aspects of these observations are
that the facets are definitely transgranular, are not associated with any
significant degree of plastic fracture, and are observed both in static
and fatigue fractures.
This particular type of fracture surface topography has been ob-
served most frequently in cases involving environmentally induced (or
enhanced) fracture. Vennett and Ansell [41] have cited examples in
tensile fractures of Type 304 stainless steel tested in high-pressure
hydrogen, Fig. 13. Similar features have also been observed in stress-
corrosion failure of titanium alloys tested in aqueous environments
[42]. In the latter case, X-ray diffraction measurements established
that the facets were oriented at a small angle (varying from 8 to 20
FIG. 1 3 - Fracture surface topography of Type 304 smooth-bar tension specimen ex-
posed 24 h and pulled to fracture in 10,000 psi hydrogen. After Vennett and Ansell
[41]. Two-stage plastic-carbon replica (x5300).
deg) to the basal plane. Other examples of "cleavage-like" features

have been noted in stress corrosion failures of Ti-6A1-4V alloys tested
in methyl alcohol [43] and in titanium-aluminum alloys fractured by
low-cycle fatigue crack propagation followed by tensile overload [44].
Because of the uncertainty in the identification of the crystallo-
graphic orientation of the facets exhibiting cleavage-like features in
fcc materials and in other particular cases where cleavage fracture is
unexpected, the observed features cannot be associated with true
cleavage fracture. As a result, the term quasi-cleavage has also been
employed in the description of these features. The use of this term in
these cases is justified only in the sense of the general definition and
meaning of quasi-cleavage (that is, "quasi" meaning seemingly or "as
if"). It should be emphasized that all fracture surface regions described
as quasi-cleavage do not exhibit the specific features categorized for
the particular case of fracture in quenched and tempered steels. In
view of origin and definition of quasi-cleavage, particular care should
be exercised in the use of the term, and every effort should be made to
restrict its use to cases where the meaning is clear. Certainly the term
should not be used, in any case, to suggest or imply a specific fracture
mechanism.
Summary
Cleavage fracture of solids has been the subject of considerable
study over the past 10 to 15 years. Significant advances in the theo-
retical aspects of this fracture mechanism have been made through the
application of dislocation theory to cleavage crack nucleation and
propagation. In addition, a more complete description of the micro-
mechanical aspects of cleavage has resulted from observations of the
fine-scale topographic features of fracture surfaces.
Theoretical considerations of cleavage fracture have been based on
models which attribute crack nucleation to blocked slip bands. The
common obstacles to slip include grain boundaries, other slip bands or
twins, and second-phase particles. In general, the theoretical models
consider these microstructural features as barriers to dislocation mo-
tion which lead to dislocation pileups. Cleavage crack initiation is con-
sidered to occur as the result of stress concentrations associated with
these dislocation pileups. Consideration of these models has led to the
statement of an equivalent fracture criterion expressed in terms of the
friction stresses resisting dislocation motion and the contribution of
grain boundaries and other microstructural discontinuities to the
strength of polycrystals. This fracture criterion correctly predicts the
susceptibility of certain bcc and hcp materials to cleavage fracture.
Applications of this fracture criterion to materials exhibiting a ductile-
to-brittle transition also led to a relationship between transition tem-
perature and grain size consistent with experimental observations. In
addition, current fracture theories predict effects of variables such as
grain size, size and distribution of precipitate particles, and yield
stress on the susceptibility of a material to cleavage fracture.
The theoretical models of cleavage fracture have not explicitly taken
into account the fact that, under certain conditions, crack growth may
be more difficult th~n crack nucleation. It appears that the principal
problem associated with a theoretical description of crack propagation
is that of accurately accounting for the energy associated with plastic
deformation in the vicinity of the crack tip.
In addition to considerations of cleavage crack nucleation and
propagation, some attention has been directed toward the prediction of
characteristic cleavage planes from atomic binding principles. Calcula-
tions based on the concept that a cleavage plane is the plane of mini-
mum surface energy have correctly predicted the cleavage planes in
the majority of cases to which the theory has been applied.
Fractographic studies have established three classes of fine-scale
topographic features characteristic of cleavage fracture, that is, fiver
patterns, cleavage tongues, and herringbone patterns. Each of these
features has been studied in detail, and the factors affecting their for-
mation are fairly well understood. These topographic features provide

direct information related to the nucleation and propagation of cleavage
cracks. Specifically, the fine-scale topography frequently provides
direct information as to the origin of fracture within a facet, the local
direction of crack propagation, and the general influence of micro-
structural features on the fracture process.
Cleavage tongues and herringbone patterns have been shown to be
associated with deformation twins formed ahead of an advancing crack,
and certain aspects of river patterns have been attributed to plastic
deformation occurring during fracture. Thus, topographic features pro-
vide a qualitative indication of the amount of plastic deformation
associated with cleavage fracture. Fractographic studies have also
shown fracture can occur by simultaneous involvement of cleavage
and microvoid coalescence.
Topographic features resembling those of cleavage have been ob-
served in numerous cases where a definite relationship between the
fracture surface and low-index crystallographic planes cannot be estab-
lished. The term quasi-cleavage has been introduced as a temporary
description of this particular class of topography in quenched and
tempered steels. Cleavage-like features have also been observed in the
fracture of certain fcc materials and in other cases where cleavage
fracture is unexpected. These observations are usually associated
with fractures induced or enhanced by environmental conditions (for
example, stress corrosion and environmentally enhanced fatigue). Be-
cause of the uncertainty in the identification of the crystallographic
orientation of facets in this latter group of fractures, the term quasi-
cleavage has also been employed in the description of these features.
It is emphasized that the term quasi-cleavage was introduced strictly
for the purpose of describing certain topographic features which are
not fully understood and that the term should not be used to imply a
specific fracture mechanism.
It may be concluded that certain basic aspects of cleavage fracture
have been thoroughly described from a theoretical point of view and
that fractographic analysis has established definite characteristics
associated with true cleavage. The observation of fine-scale topo-
graphic features associated with plastic deformation suggests that
further application of fractographic techniques may provide more
specific information related to the role of plastic deformation in cleav-
age fracture. Additional systematic studies directed toward correla-
tion of cleavage fracture topography with microstructural features are
necessary for a more complete understanding of the cleavage mecha-
nism. It is also evident that additional work is necessary to establish
the mechanisms involved in the formation of topographic features
described as quasi-cleavage.
References
[1] Stoloff, N. S. and Gensamer, M., "Deformation and Fracture of Polycrystalline
Cadmium," Transactions, American Institute of Mining, Metallurgical, and
Petroleum Engineers, Vol. 227, 1963, p. 70.
[2] Toaz, M. W. and Ripling, E. J., "Flow and Fracture Characteristics of Binary
Wrought Magnesium-Lithium Alloys," Journal, Institute of Metals, Vol. 85, 1956,
p. 137.
[3] Amateau, M. F., Burrier, H. I., Jr., and Ebert, L. J., "Brittle Fracture in Alpha-
Titanium," Transactions, American Society for Metals, Vol. 59, 1966, p. 921.
[4] Burghard, H. C. and Davidson, D. L., "Fracture Mechanisms and Fracture Sur-
face Topography," Proceedings of the First International Conference on Fracture,
Japanese Society for Strength and Fracture of Materials, 1966, Vol. 2, p. 571.
[5] Cottrell, A. H., "Theory of Brittle Fracture in Steel and Similar Metals," Trans-
actions, American Institute of Mining, Metallurgical, and Petroleum Engineers,
Vol. 212, 1958, p. 192.
[6] Ku, R. C. and Johnston, T. L., "Fracture Strength of MgO Bicrystals," Philosophi-
cal Magazine, Vol. 9, 1964, p. 231.
[7] Westwood, A. R. C., Goldheim, D. L., and Pugh, E. N., "Fracture Behavior of
AgC1 in Aqueous NaCI," Acta Metallurgica, Vol. 13, 1965, p. 695.
[8] Wronski, A. and Fourdeux, A., "Slip-Induced Cleavage in Polycrystalline Tung-
sten," Journal of Less Common Metals, Vol. 6, 1964, p. 413.
[9] McMahon, C. J., Jr., and Cohen, M., "Initiation of Cleavage in Polycrystalline
Iron," Acta Metallurgica, Vol. 13, 1965, p. 591.
[10] Stroh, A. N., "The Formation of Cracks as a Result of Plastic Flow," Proceedings,
The Royal Society, Vol. 223A, 1954, p. 404.
[11] Petch, N. J., "The Ductile to Brittle Transition in the Fracture of Alpha-Iron,"
Philosophical Magazine, Vol. 3, 1958, p. 1089.
[12] Petch, N. J., "The Ductile-Cleavage Transition in Alpha Iron," Fracture, Wiley,
New York, 1959, p. 54.
[13] Jordan, K. R., Ph.D. thesis, Rensselaer Polytechnic Institute, 1967.
[14] Johnston, T. L., Davies, R. G., and Stoloff, N. S., "Slip Character and the Ductile
to Brittle Transition of Single Phase Solids," Philosophical Magazine, Vol. 12,
1965, p. 305.
[15] Hahn, G. T. et al, "Initiation of Cleavage Microcracks in Iron and Steel," Fracture,
Wiley, New York, 1959, p. 91.
[16] Hahn, G. T. and Rosenfield, A. R., "A Modified Double Pile-Up Treatment of the
Influence of Grain Size and Dispersed Particles on Brittle Fracture," Acta Metal-
lurgica, Vol. 14, 1966, p. 1815.
[17] Johnston, T. L., Stokes, R. J., and Li, C. H., "The Fracture Behavior of Silver
Chloride-Alumina Composites," Transactions, American Institute of Mining,
Metallurgical, and Petroleum Engineers, Vol. 221, 1961, p. 792.
[18] Keh, A. S., Leslie, W. C., and Sponseller, D., "Interactions of Dislocations and
Precipitates in Some Iron-Base Alloys," Precipitation in Iron Base Alloys, Gordon
and Breach, New York, 1966, p. 281.
[19] Rosenfield, A. R. and Hahn, G. T., "Effects of Second Phase Particles on Duc-
tility," Proceedings of Conference on Physical Basis of Yield and Fracture, Insti-
tute of Physics and Physical Society, Conference Series No. 1, 1966, p. 47.
[20] Gilman, J. J., "Cleavage, Ductility and Tenacity in Crystals," Fracture, Wiley,
New York, 1959, p. 193.
[21] Guiu, F., "Twinning and Cleavage Fracture in Molybdenum Single Crystals,"
Journal of Less Common Metals, Vol. 12, 1965, p. 77.
[22] Hull, D., "Twinning and Fracture of Single Crystals of 3% Silicon Iron," Acta
Metallurgica, Vol. 8, 1960, p. 11.
[23] Tetelman, A. S., "The Ductile-Brittle Transition in Iron-3% Silicon Monocrys-
tals," Acta Metallurgica, Vol. 12, 1964, p. 324.
[24] Rosenbaum, H. S., "Non-Basal Slip and Twin Accommodation in Zinc Crystals,"
Acta Metallurgica, Vol. 9, 1961, p. 742.
[25] Berry, J. M., "Cleavage Step Formation in Brittle Fracture Propagation," Transac-
tions, American Society for Metals, Vol. 51, 1959, p. 556.
[26] Low, J. R., "A Review of the Microstructural Aspects of Cleavage Fracture,"
Fracture, Wiley, New York, 1959, p. 68.
[27] Irvine, K. J. and Picketing, F. B., "The Impact Properties of Low Carbon Bainitic
Steels," Journal, Iron and Steel Institute, VoL 20, 1963, p. 518.
Study of Micromechanisms of Fracturing Processes," Fracture Toughness Test-
ing and Its Applications, ASTM STP 381, American Society for Testing and Ma-
terials, 1965, p. 210.
[29] Pelloux, R. M. N. and 'McMillan, J. C., "The Analysis of Fracture Surfaces by
Electron Microscopy," Proceedings of the First International Conference on Frac-
ture, Japanese Society for Strength and Fracture of Materials, Vol. 2, 1966, p. 547.
ML-TDR-64-416, Air Force Materials Laboratory, Wright-Patterson Air Force
Base, Ohio, 1965.
[31] Crussard, C. et al, "A Study of Impact Tests and the Mechanism of Brittle Frac-
ture," Journal, Iron and Steel Institute, Vol. 183, June 1956, p. 146.
[32] Stoloff, N. S. and Johnston, T. k., "Crack Propagation in a Liquid Metal Environ-
ment," Acta Metallurgica, Vol. 11, 1963, p. 251.
[33] Reed-Hill, R. E. and Robertson, W. D., "The Crystallographic Characteristics of
Fracture in Magnesium Single Crystals," Acta Metallurgica, Vol. 5, 1957, p. 728.
[34] Beachem, C. D., "The Formation of Cleavage Tongues in Iron," Report of NRL
Progress, Naval Research Laboratory, Feb. 1966, p. 19.
[35] Beachem, C. D., "1966 ASTM Photographic Exhibit," Materials Research and
Standards, Vol. 6, 1966, p. 519.
[36] Gilbert, A., "Influence of Alloying Elements on Mechanisms of Fracture in
Chromium," Second Annual Report (to Pratt & Whitney Aircraft Corp.).
[37] Dahlberg, E. P., "Characteristics of Twinning on Fracture Surfaces of Single and
Polycrystalline Tungsten," Proceedings of Fifth International Congress for Elec-
tron Microscopy, Academic Press, 1962, Paper J-5.
[38] Beachem, C. D., "The Crystallography of Herringbone Fractures in an Iron-
Chromium-Aluminum Alloy," Report of NRL Progress, Naval Research Labora-
tory, Jan. 1967, p. 28.
[39] Burghard, H. C., Jr., and Norris, E. B., "The Effect of Neutron Irradiation on the
Fracture Characteristics of Pressure-Vessel Steels," ASME Paper No. 67-MET-1,
American Society of Mechanical Engineers, 1967.
[40] Beachem, C. D., "Electron Fractographic Studies of Mechanical Processes in
Metals," Transactions, American Society of Mechanical Engineers, Series D,
Vol. 87, 1965, p. 299.
[41] Vennett, R. M. and Ansell, G. S., "Effect of High-Pressure Hydrogen upon
Tensile Properties and Fracture Behavior of 304L Stainless Steel," Transactions,
American Society for Metals, Vol. 60, 1967 (to be published).
[42] Meyn, D. A., "A Study of the Crystallographic Orientation of Cleavage Facets
Produced by Stress-Corrosion Cracking of Ti-7 AI-2 Nb-1 Ta in Water," Report
of NRL Progress, Naval Research Laboratory, Aug. 1965, p. 21.
[43] Meyn, D. A., Dahlberg, E. P., and Beachem, C. D., "Analysis of Stress-Corrosion
Cracking of Ti-6 AI-4 V Fuel Tank Material in Methyl Alcohol," NRL Memoran-
dum Report 1744, Naval Research Laboratory, Jan. 1967.
[44] Judy, R. W., Jr., et al, "Low Cycle Fatigue-Crack Propagation and Fractographic
Investigation of Ti-7 AI-2 Cb-I Ta and Ti-6 AI-4 V in Air and Aqueous Environ-
ments," Transactions, American Society for Metals, Vol. 59, t966, p. 195.
C. D. B e a c h e m 1 a n d D. A. M e y n 2
Fracture by Microscopic Plastic

Deformation Processes
REFERENCE: Beachem, C. D. and Meyn, D. A., " F r a c t u r e by Micro-

scopic Plastic Deformation Processes," Electron Fractography, ASTM
STP 436, American Society for Testing and Materials, 1968, pp. 59-88.
ABSTRACT: The various fracture surface features formed by fine-scale

plastic flow processes are discussed, and the mechanisms by which they are
formed are discussed where they are known.
Glide plane decohesion, as it has been used in the literature, is shown to be
misleading, and a more descriptive term, "stretching," is used in its place. The
use of "ductile cleavage" to describe the same process is also misleading.
Since brief, descriptive, nonmisleading terminology is necessary for im-
mediate communication, several terms have evolved to describe the surfaces
created by unknown processes. Glide plane decohesion, serpentine glide,
ripples, stretching, and three kinds of microvoid coalescence are shown to be
parts of a spectrum of plastic rupture.
KEY WORDS: electron microscopy, fractography, surface features, plastic

flow, metals, fracture mechanisms
T h e m e t a l s a n d a l l o y s w h i c h are tough o w e this p r o p e r t y to their

a b i l i t y to d e f o r m p l a s t i c a l l y a n d a b s o r b e n e r g y b e f o r e o r during frac-
turing. T h e p r e s e n t p a p e r d e s c r i b e s t h e m e c h a n i s m s r e s p o n s i b l e for
several fracture modes involving extensive plastic deformation, insofar
as t h e s e m e c h a n i s m s h a v e b e e n w o r k e d out. All o f t h e s e m o d e s i n v o l v e
the s a m e f u n d a m e n t a l p r o c e s s , glide on o n e o r m o r e sets o f c r y s t a l l o -
g r a p h i c p l a n e s , b u t the r e s u l t i n g f r a c t u r e t o p o g r a p h y is different, and
t h e r e f o r e different n a m e s are u s e d for the v a r i o u s f r a c t u r e a p p e a r a n c e s
e v e n t h o u g h the b a s i c p r o c e s s in e a c h is the same.
T h e n a m i n g o f f r a c t u r e t o p o g r a p h i c f e a t u r e s often reflects the cur-
rent c o n c e p t s o f the m a n n e r in which the f e a t u r e s a r e g e n e r a t e d . T h e
f e a t u r e s o b s e r v e d at the m a g n i f i c a t i o n s u s e d in e l e c t r o n m i c r o s c o p y
a r e p r o d u c e d b y local s t r e s s e s which, b e c a u s e o f i n c l u s i o n s , v o i d s ,
~Head, Micro-Mechanical Metallurgy Section, Physical Metallurgy Branch, Metal-

lurgy Div., Naval Research Laboratory, Washington, D.C. Personal member ASTM_
"-'Physical metallurgist, Micro-Mechanical Metallurgy Section, Physical Metallurgy
Branch, Metallurgy Div., Naval Research Laboratory, Washington, D.C.
59
Copyrights 1968 by ASTM International www.astm.org
anisotropic grains in polycrystalline metals, etc., may be very different

from the macro stresses imposed on the specimen. One is often forced,
therefore, to infer not only the microfracture process but also the local
state of stress which produced the observed feature.
F i n e Scale M e c h a n i s m s
Glide Plane Decohesion

Single crystals of pure metal may, when plastically deformed, glide
apart along slip planes. When these crystals are large, slip along these
planes is readily visible to the unaided eye. This process, if continued
on the same plane, leads eventually to the specimen parting along a
glide plane. This mode of separation, whether on a macroscopic or
microscopic scale, and whether or not carried to completion, is appro-
priately called glide plane decohesion when the glide surfaces can be
identified as such. It is an important process in the fracturing of many
alloys of practical usefulness.
If a metal specimen separates by glide plane decohesion, particu-
larly if this is along only a few members of one set of crystallographic
planes, the fracture surface consists largely of facets of one or at most a
few specific lattice planes, and in this respect the fracture surface might
be considered akin to cleavage, which also occurs on only one or at
most a few crystallographic planes. It is understandable then why the
process has been called "ductile cleavage" (ductile because extensive
glide precedes fracture). This is considered a less suitable term than
glide plane decohesion.
The term "ductile rupture" has been used in the past for the process
which, in this paper, has been termed microvoid coalescence. Although
some of the most ductile metals and alloys fracture by this mode, so do
some of the alloys having very low engineering ductility. The term
ductile rupture might suggest to the mechanical metallurgist greater
toughness than a given alloy may possess, and, accordingly, ductile
rupture might best be used to describe macroscopic ductile fracture,
such as occurs in pure gold or ultra-pure aluminum.
The use of the term glide plane decohesion is misleading in cases
where the featureless surfaces under consideration are created by
unidentified processes of extreme plastic deformation operating on a
scale much smaller than the size of the smooth surface. The term
stretching is now being widely used to specify this type of process, as
discussed in the next section.
Serpentine Glide
Particularly in polycrystalline metals where the deformation of a
given grain must accommodate to the deformation of its neighboring
BEACHEM AND MEYN ON MICROSCOPIC PLASTIC DEFORMATION PROCESSES 61
grains, gliding may occur on many sets of intersecting planes and in

an irregular manner giving rise to a pattern which has come to be
known as serpentine glide. Serpentine glide surfaces, then, are formed
by (partial) glide plane decohesion on several sets of planes by un-
specified mechanisms, giving a particular interwoven or plaited ap-
pearance on the fracture surface. This will be illustrated in several sub-
sequent electron fractographs.
The sketch in Fig. 1 shows the idealized stress pattern at a free sur-
face at the root of a notch or crack (or any other free surface) in a
stressed specimen. Figure 2 is a further consideration of the stresses
at a free surface. Here we are viewing the stressed element along the
intersection of the local planes of maximum shear stress, that is, along
the direction of o-2. Though glide need not occur on planes of maximum
shear stress, localized glide is shown here along one family of planes
parallel to the planes of maximum shear stress producing steps which
may also be seen in an actual case in Fig. 3a. This specimen was an
ingot iron wire, 1/8 in. diameter, notched with a file perpendicular to
the axis of the wire. The root of the notch was caused to open by bend-
ing the wire as shown in the sketch in Fig. 3b. The features indicated at
A and B in Fig. 3a are serpentine glide, and such surfaces may be seen
in stereo viewing to be stepped as in Fig. 2. The light surfaces marked
FIG. 1-Stress configuration at a free surface where the local maximum principal
stress is vertical. The shaded planes are the local planes of maximum shear stress.
O"
A. BEFORE GLIDE
cr I
B.. AFTER GLIDE ON A SERIES OF

PARALLEL PLANES THAT ARE
FAVORABLY ORIENTED FOR GLIDE
F I G . 2-Simplified general model for the creation of a rupture or fracture surface by

a simple glide process. Though glide by a slip process is shown, glide by twinning may
also produce new free surfaces. Arrows indicate the new (rupture, fi'acture) surfaces,
created by glide plane decohesion.
by B's (Fig. 3a) are the new free surfaces created by plastic flow on
glide p l a n e s - glide plane decohesion. If further glide takes place either
on other planes in the same set or on planes in other sets, the markings
tend to be smoothed out, becoming first ripples [1]3 (as seen at C in
Fig. 3a) and then featureless as the glide steps become too small to be
shown by the fidelity of present-day replicas. These events are
sketched in Fig. 4 and are apparent in Fig. 3a when one realizes that
the sharp step-like markings of serpentine glide are at the bottom of
the trough where plastic flow is not extensive and that the ripples are
smoother near the file marks where additional glide has taken place.
Such glide markings as are visible in Fig. 3a, and others throughout
this report, can in general be formed by a combination of such ele-
mentary slip mechanisms as pencil or wavy slip [2-4], irregular slip,
cross slip, and multiple slip. Two of these are illustrated in Fig. 5. It is
likely that multiple slip will operate independently of the other mecha-
nisms and increase the general complexity of the surface appearance.
T h e italic n u m b e r s in brackets refer to the list of references appended to this paper.
FIG, 3a-Serpentine glide surface in annealed Armco iron. This glide was produced
by slip when the root o f a notch in this wire specimen was opened up by bending at room
temperature. The dark surfaces (A) apparently were a single surface before slip on a
second series of parallel planes produced the light parallel surfaces (B). Examples o f
ripples may be seen between the arrows (C). Palladium-shadowed direct-carbon replica
(•
FILE MARKS
FIG. 3 b - S k e t c h o f wire specimen from which replica shown in Fig. 3a was taken.
G ,O PL N O OOH S,O .
SERPENTINE GLIDE SURFACES
DISTINCT GLIDE SURFACES

~.~.______ PRODUCEDtN IIBII ARE
SMOOTHED OUT INTO
~ " * - - - . . WAVES OR RIPPLES,
C
"%~MORE
/ - -
RECENT GLIDE
L~. PLANEDECOHESION
D
/~.. ._.
~ , ' ~
) ~ M O S T
kL-~
DISTINCT GLIDE SURFACES
PRODUCED ,N "8" A R E , O W
SMOOT.-NO MARKINGS
V,S,BLE ,N CARSON
REPLICA
RECENT GLIDE
PLANEDECOHESO
IN
~ DISTINCTGLIDESURFACES
PRODUCED IN "C" ARE NOW
~ . SMOOTHEDOUT INTO RIDGES.
FIG. 4 - T h e production o f smooth surfaces by plastic flow. Surfaces definitely pro-
duced by glide plane decohesion are smoothed by severe plastic deformation-stretch-
i n g - until no markings are visible on ordinary replicas. Elongation is vertical, and in-
creases from top to bottom. Crack propagation direction is from right to left.
64
FIG. 5a-Irregular slip. Slip intersects the surface in irregular fashion, producing
irregular slip steps (cross-hatched). This is very similar to some features actually seen
in Fig. 3a.
FIG. 5 b - P e n c i l slip, also called wavy slip or wavy glide in c~-iron by Barrett [2].
Cross-slipping screw or multiplanar edge dislocations can create this appearance. Re-
sulting slip step is cross-hatched. From Cottrell [3].
Stretching
When glide produces new surfaces in heavily cold-worked metal,

the glide planes are poorly defined and the glide steps may be so poorly
delineated (at least in replication processes now available for fracture
surfaces) that the new surface may appear featureless. The term which
has been used for the processes which generate such surface areas is
stretching, which is then glide plane decohesion on a scale so fine that
the glide marks are no longer seen.
An experiment which illustrates the fact that these smooth feature-
less surfaces are formed by stretching rather than by glide plane de-
cohesion or ductile cleavage is illustrated in Figs. 6 and 7. A piece of
FIG. 6 - S t r e t c h i n g at the root o f a notch in a 99.999 p e r cent pure copper wire

specimen. Stretching occurred in directions o f the arrows after the palladium was de-
posited, a n d before the carbon was deposited. D a r k areas are palladium, light areas are
newly stretched. Direct-carbon replica (x6000).
FIG. 7-Extensive stretching in 99.999 per cent copper after the deposition o f palla-
dium, Direct-carbon replica (x9000).
annealed 99.999 per cent pure copper wire was notched, and the wire
was bent to open the root of the notch thus introducing severe cold
work into the material at the notch root. After palladium was evapo-
rated into the notch, the wire was bent further, and carbon was
evaporated into the notch. This direct carbon replica was then freed
from the specimen and studied. Figure 6 shows a portion of the root of
the notch which was only slightly stretched in the direction of the ar-
rows after the palladium was evaporated. Figure 7 shows a region
where additional stretching was more extensive on one side than on the
other. The right-hand side of the picture shows such extreme stretch-
ing that one cannot tell in which direction it was stretched.
Another example of the stretching process is shown in Fig. 8. In
this as-cast Ti-2A1-2Zr-2V-1Mo alloy, a sheet of an unknown con-
stituent is seen to be broken up into segments which have moved apart
by stretching of the bulk material in the directions shown by the ar-
rows.
Stretching is defined as the production of smooth featureless free
surfaces by extremely complex plastic flow mechanisms acting on too
fine a scale to produce characteristic surface traces in present-day
replicas. Thus stretching, which is by no means an accurate description
of the basic plastic flow mechanisms, is a much more descriptive and
much less misleading term than glide plane decohesion or ductile
cleavage for this form of deformation.
Microvoid Coalescence
Small precipitate particles, undissolved carbides, sulfides, silicates,
oxides, or other brittle or weak constituent particles are generally
present in large numbers in all but the purest of metals. When the metal
around them is plastically deformed these particles often create internal
free surfaces by either cleaving or separating from the adjacent metal
along their interfaces.
Under plastic strain such free surfaces nucleate voids which grow
by plastic rupture processes similar to those at the root of the opening
notch in Fig. 3. With extreme plastic growth of voids, which fre-
quently grow one or two orders of magnitude larger than the free sur-
faces that initiate them, steps or ripples are visible only rarely in
present-day replicas. This is particularly true in the high-strength
alloys that have complex fine-scale microstructures. The spherical or
rounded shape of the voids prohibits simple deformation: glide must
occur in a highly complex manner in the severely worked material at
the surface of the void during the growth of a spherical or rounded in-
ternal void (hundreds or thousands of per cent elongation-if one cares
to visualize plastic deformation on this scale in this manner).
The voids which are nucleated at these metallurgical singularities
FIG. 8-Stretching in as-cast Ti-2AI-2Zr-2V-IMo. The flat constituent particles

have been expanded in two directions (arrows) on a relatively fiat area as a result of
stretching of the underlying metal. Surrounding metal is much more severely stretched.
Two-stage carbon replica (•
grow under plastic strain by the glide processes which have been
termed "stretching" above. (Growth by diffusion of vacancies is too
slow a process.) A void may grow until it impinges on a neighboring
void under one of three stress configurations as depicted in Fig. 9.
When voids impinge and the last remaining metal separates under nor-
mal stress, the newly opened surface can be shown to consist of a cup-
like depression on both fracture faces which has come to be called an
equiaxed dimple (or sometimes a cupule), and the process has come to
be called normal dimpled rupture.
If, however, the voids grow and coalesce under shear stresses,
as in the second series of Fig. 9, they are elongated in the shearing
direction. The surface of the shear fracture shows the elongated
o"1
t.~o" 2
(o)
o'!
(b)
o,t
(r ~ L~r2 ~ %
[<<<l
d
FIG. 9- Three observed basic modes for the coalescence of microvoids. For each
mode the sketches show from left to right: material stressed almost to the point o f local
rupture," local rupture; and the directional sense of dimples on the rupture surfaces. The
three coalescence modes are called (a) normal dimpled rupture; (b) shear rupture; and
(c) tearing.
dimples with one end continuous and the other end stretched out until
that end appears open. Observations on precisely matched fracture
sites have confirmed that one set of shear dimples, as these are called,
points toward the origin of the shear fracture, and the set on the other
fracture surface points in the opposite direction.
If voids grow and coalesce under a strain gradient, as in the third se-
quence of Fig. 9, they are also elongated, and when observed on only
one fracture surface they are presently indistinguishable from shear
dimples. But the dimples on the matching surface in this case are
"pointing" in the same direction as on the other half, and on both
fracture surfaces the dimples point toward the fracture origin. These
have been termed tear dimples since they have been generated by a
tearing action.
It will be seen that these three forms of surfaces created by micro-
void coalescence- equiaxed, shear, and tear dimples- may sometimes
appear close together in what appears macroscopically to be a surface
of a single fracture mode. This is attributed to the metallurgical
heterogeneities mentioned earlier or to transients in the stress pattern.
Fractographs Further Illustrating Plastic Rupture Surfaces

Metals that are soft and sufficiently free from precipitate particles,
inclusions, etc., tend to favor failure by glide plane decohesion and its
extensions, serpentine glide and stretching. Figures 3a, 6, 7, and 10
show fracture surfaces created by this mode. The serpentine glide sur-
faces might be mistaken for fatigue fracture surface markings by the
unwary doing a failure analysis. The occasional crossing of the glide
markings, however, may often be used to identify these as serpentine
glide surfaces rather than fatigue fracture surfaces.
Figure 3a (arrows) also shows isolated dimples which nucleated near
the stretching surface, grew to the surface, opened, and stretched out
in the surface. In Fig. 3a the stretching of the dimple in the surface is
so extensive that its two separated ends are not easily recognized as
mates. Figures 10 and 11 show increasing numbers of dimples in
stretched surfaces, while Figs. 12 and 13 are composed entirely of
dimples. Thus a full spectrum is seen, ranging from no dimples to all
dimples on a stretched surface. Figure 12 clearly shows serpentine
glide in many of the dimples.
Surfaces formed by the tearing mode of microvoid coalescence are
illustrated in Figs. 12 and 13. (Figures 10, 11, and 13 show both ends of
oval dimples in a stretch surface where it is obvious that the dimples
grow by an "opening" mode and are therefore by definition tear
dimples.) Most of the tear dimples in Fig. 13 could be identified only by
checking the directional sense of the dimples on the matching fracture
surface.
F I G . lO-Ripples (lower left quarter of fractograph), stretching, and oval tear

dimples (arrows) in Armco iron. Two-stage palladium-shadowed carbon replica
(x9000).
FIG. 11 - S t r e t c h e d regions, shear dimples, and oval tear dimples (arrows) in A I S I

304 stainless steel. Palladium-shadowed direct-carbon replica (x6000).
FIG. 1 2 - E x a m p l e s o f serpentine glide in dimples in tough pitch copper. Specimen

alligatored due to residual stresses from rolling. An equiaxed dimple is shown by the
horizontal arrow. Tear dimples are shown between slanted arrows. Palladium-shadowed
direct-carbon replica (•
FIG. 1 3 - T e a r dimples formed during the early stages of internal .fracture of a

hardened and tempered AISI 4340 smooth tension test specimen broken in tension.
Oval tear dimples are shown by arrows. Palladium-shadowed two-stage carbon replica
(•
The tear dimples shown in Fig. 13 were observed in a replica from

the center of a cup-and-cone tensile failure surface in a 282,000 UTS
AISI 4340 smooth bar specimen. The two halves of the specimen are
shown in Fig. 14 while the fracture initiation region is shown in Fig. 15
at a high magnification. The fracture started at a small diameter (but
deep) "pinhole" by radial tearing. Tearing proceeded about one tenth
the specimen diameter before it changed to mixed tearing, shear rup-
ture, and normal dimpled rupture.
Examples of surfaces formed by normal dimpled rupture are seen in
Figs. 16 and 17. Normal dimpled rupture is a mode often operative in
running "brittle fractures" of high-strength steels and aluminum alloys.
This fracture mode is almost always accompanied by shearing or tear-
ing in neighboring microscopic regions.
Examples of surfaces formed by shear rupture are seen in Figs. 18 to
20. Precisely matched shear dimples are shown in Figs. 19 and 20 with
FIG. 1 4 - M a c r o s c o p i c view o f A I S l 4340 cup-and-cone fracture surface from which

the fractograph in Fig. 13 was made. Fracture initiation region may be seen just above
the center o f the fracture surface (x4).
FIG. 15-Closer view of "pinhole" fracture initiation region. Tearing was found in
the central region bounded by the arrows (•
the matching dimples marked with arrows. In the shear dimple area of
Figs. 19 and 20 there is an area of quasi-cleavage [5], and it may be
noted that matching shear dimples adjacent to the quasi-cleavage are
not of the same size or length. This effect extends for several microns
up the shear lips. The sketch in Fig. 21 explains why the matching
dimples are not of equal length. This effect has been observed also in
AISI 4340 steel fracture toughness specimens, where the boundary
between the flat fracture and the shear lips was well delineated. The
effect is difficult to recognize in fracture faces of smooth tensile speci-
mens and in fracture toughness specimens when the flat fracture is
composed of dimples. Both test conditions give rise to large plastic
flow, producing less sharply defined boundaries.
Tear dimples form frequently when voids coalesce with larger free
surfaces. Low magnification photographs of a rubber-band model are
shown in Fig. 22 for comparison with the oval tear dimples in Figs. 10,
11, and 13, and an explanatory sketch of the process is shown in Fig.
23. The model is shown only to indicate, in a simple manner, the
stretching of this type dimple after the void has coalesced with the sur-
face. Successive replicas from progressively stretched dimples at the
bottom of notches in iron wire have confirmed this process in metal.
Another repeatedly observed feature of void coalescence and
stretching is shown in Fig. 24. Sharp fatigue cracks are frequently used
as stress raisers in crack propagation fracture toughness evaluations.
FIG. 16-Equiaxed dimples in the AMS 6434. Two-stage palladium-shadowed car-

bon replica (•
FIC;. 17-Equiaxed dimples in a hardened-and-tempered low-alloy steel specimen.

Palladium-shadowed two-stage carbon replica (•
FIG. 18-Shear rupture dimples in A ISI 304 stainless steel "scab" formed by hyper-
velocity impact. Palladium-shadowed direct-carbon replica (•
FIG. 1 9 - S h e a r rupture dimples on two sides of a quasi-cleavage facet. The arrows

indicate shear dimples that have mates in Fig. 20. Material is hardened-and-tempered
410 stainless steel. Palladium-shadowed two-stage carbon replica (•
FIG. 2 0 - F r a c t u r e surface which matches that shown in Fig. 19. Same replicating
conditions (x3500).
QUASI-CLEAVAGE
CRACK \
,O
A \
B
A \
B/ A
SHORTDIMPLES
LONGDIMPLES
LONGDIMPLES ~ ~
SHORT DIMPLES- ~
FIG. 21 - A simplified model o f the formation o f the shear rupture dimples shown in
Figs. 19 and 20. A slightly opened quasi-cleavag'e crack is shown at top, andsuccessive
stages of shear rupture are illustrated in the lower sketches. The macroscopic principal
stresses are shown at the top. Arrows on either side of the crack indicate shear direc-
tions. Dimples are stretched more at points B than at points A, particularly after the
voids have started to coalesce with the advancing, shear rupture crack.
FIG, 2 2 - A rubber model demonstrating how a tear dimple stretches into an oval
shape after the void coalesces with another free surface. The model is a rubber band,
sliced part-way through the thickness with a razor blade, and then stretched.
Invariably, if the specimen does not cleave, stretching will occur at the
tip of the crack when it is opened up in testing, as indicated by Fig. 24.
Initial failure by void coalescence cannot occur because (1) a third
stress component at the free surface is lacking, or (2) the material at
the tip of the crack is not sufficiently deformed to grow voids. After a
finite amount of growth by stretching, the crack usually intersects in-
creasing numbers of voids as it grows through increasingly strained
material, and tear dimples are formed as in Fig. 24. As the crack grows
further-through material contained in an increasingly larger plastic
z o n e - p a t c h e s of voids open up and coalesce ahead of the crack, and
the fracture is seen to increasingly consist of random mixtures of tear
dimples, shear dimples, and equiaxed dimples, as seen in Fig. 24.
As shown in Ref 6 the length of shear dimples depends upon the
value of the ratio of normal (perpendicular to the shear plane) strain to
shear strain in the material surrounding the voids during growth and
coalescence of the voids. As the value of the ratio decreases, the
FIG, 2 3 - Sketch o f the coalescence o f a void with a larger free surface. The arrows
indicate the direction of principal strain. The fracture surface is seen in (A) and the
corresponding cross sections o f the void are seen in (B). Plastic deformation increases
from left to right.
FIG. 24-Stretching, glide plane decohesion and tearing at the tip of a fatigue crack.
The fatigue crack was introduced into this 7075-T6 aluminum alloy specimen as a
stress raiser from the monotonic test which later produced the plastic rupture. Palla-
dium-shadowed two-stage carbon replica (x6000).
dimples are observed to become very l o n g - five to ten times as long as

they are wide (Fig. 25). Another effect of reducing the ratio of normal
strain to shear strain is that when the specimen does shear apart at tiny
local points, the two rupture surfaces may rub against one another as
they separate. Where void coalescence markings end and burnishing
markings begin has not yet been established, but both will consist
mainly of long parallel lines; neither of the markings indicates glide
FIG. 2 5 - L o n g shear rupture dimples in an OFHC copper torsion rupture surface.

Two-stage palladium-shadowed carbon replica (•
F | G . 2 6 - S h e a r rupture dimples in 99.999 per cent copper. These dimples were

formed by lightly rubbing two mechanically polished surfuces together, causing friction
welding and subsequent rupture. Two-stage palladium-shadowed carbon replica
(•
plane decohesion. To illustrate the complexity of distinguishing be-

tween initial rupture surfaces and burnishing markings, Fig. 26 shows
rupture dimples where two mechanically polished pure copper surfaces
were lightly rubbed together. These dimples were formed during rup-
ture after local friction-welding occurred.
Summary
Fine-scale plastic flow processes are seen to be basically the same
regardless of the features produced. Early stages of plastic deformation
in soft crystals produce stepped free surfaces by glide plane deco-
hesion. If these steps form an interwoven pattern, the process is
called serpentine glide. If the steps are further smoothed out, the sur-
face features are called ripples. If the process is carried far enough,
smooth featureless surfaces are formed. T h e s e featureless surfaces
grow under further deformation by an unidentified mechanism (or
mechanisms) called stretching.
Depending upon whether or not internal microvoids are opened, and
upon their number, the stretched surfaces may or may not contain
dimples, and the number of dimples will depend upon the number of
microvoids intersected by the stretching surface. If a large number of
closely spaced microvoids is present, the fracture may progress by
the microvoid coalescence process, rather than by stretching. T h r e e
distinctly different types of microvoid coalescence mechanisms have
been o b s e r v e d - n o r m a l dimpled rupture, shear rupture, and tearing.
Tearing dimples are found usually at the beginnings of fracture where
fracture initiates at a sharp crack or internal flaw. A macroscopically
flat "brittle" fracture surface may contain scattered patches of the three
different types of fracture.
T h e mechanisms of stretching operat.e on a scale usually much
smaller than the dimple size, and no difference is suspected between
the mechanisms underlying the growth of dimples and stretched sur-
faces.
Acknowledgments
T h e authors extend their thanks to B. F. Brown and E. P. Dahlberg
of the Naval Research L a b o r a t o r y and to several members of A S T M
Subcommittee II of Committee E-24 for their interest, help, and con-
structive criticism during the preparation of this paper.
References
[1] Rogers, H. C., "Tensile Fracture of Ductile Metals," Transactions, American
Institute of Mining, Metallurgical, and Petroleum Engineers, Vol. 218, July 1960, p.
498.
[2] Barrett, C. S., Structure of Metals, McGraw-Hill, New York, 1952, p. 340.
[3] Crussard, C. et al, "A Comparison Between Ductile and Fatigue Fractures,"
Swampscott Conference on Fracture, Wiley, New York, 1959, pp. 19-21.
[4] Cottrell, A. H., Dislocations and Plastic Flow in Crystals, Oxford University
Press, London, 1953, p. 4.
[5] Beachem, C. D., "Electron Fractographic Studies of Mechanical Fracture Processes
in Metals," Transactions, American Society of Mechanical Engineers, Journal of
Basic Engineering, Vol. 87, Series D, No. 2, June 1965, p. 299.
[6] Beachem, C. D., "An Electron Fractographic Study of the Influence of Plastic
Strain Conditions Upon Ductile Rupture Processes in Metals," Transactions,
American Society for Metals, Vol. 56, No. 3, Sept. 1963, pp. 318-326.
J. C. McMillan t and R. W. Hertzberg ~
Application of Electron Fractography

to Fatigue Studies
R E F E R E N C E : McMillan, J. C. and Hertzberg, R. W., "Application of

Electron Fractography to Fatigue Studies," Electron Fractography,
AS TM S TP 436, American Society for Testing and Materials, 1968, pp. 89-123.
ABSTRACT: A review of the application of fractography to fatigue studies

is presented that covers fatigue fracture appearance, the occurrence of stria-
tions, and qualitative and quantitative aspects of the fracture surface micro-
details. Recent experimental results in three areas are also discussed: (1) the
effect of prior loading history on subsequent crack propagation, (2) the smallest
observable crack growth increments, and (3) the effect of environment on fa-
tigue crack growth.
KEY WORDS: electron microscopy, fractography, fatigue (materials), alumi-

num alloys, corrosion fatigue, fracture mechanics
Fatigue crack extension occurs in either three or four stages, depend-

ing on the applied stress conditions, specimen geometry, and the mech-
anical properties of the material. Stage one, the initiation stage, usually
extends only over a small percentage of the fracture surface. If no
nucleation site is readily available, reverse slip of dislocations on a
slip plane subjected to a high resolved shear stress will lead to crack
formation on a slip plane at an angle to the applied stress direction.
After a relatively short distance, this crack will shift into a position
leading to the plane strain (stage two) and plane stress (stage three)
propagation stages. With a low relative stress level, the initiation stage
may represent a significant portion of the total fatigue life of the com-
ponent, but the total percentage of the fracture surface in stage one
will usually be much smaller. In most engineering structures and com-
ponents there are a sufficient number of nucleation sites to cause a
fatigue-nucleated crack to enter the propagation stages after relatively
few loading cycles. At the other extreme, stage four represents final
fracture in the last quarter cycle. Since this stage is beyond the scope
1 Research engineer, Commercial Airplane Div., The Boeing Co., Renton, Wash.
Assistant professor, Department of Metallurgy and Materials Science, Lehigh Uni-
versity, Bethlehem, Pa.
89
Copyright s 1968 by ASTM International www.astm.org

of interest of this paper, attention will be focused on stages two and

three, the propagation stages. Therefore, succeeding remarks will draw
attention to the surface characteristics of the propagation stages in
fatigue.
Macroscopic Fracture Surface Appearance
A macroscopic examination of service and laboratory failures by
fatigue loading reveals distinct fracture surface characteristics. The
fracture surface is generally flat, indicating the absence of any appre-
ciable amount of gross plastic deformation in the overall component.
The surface can be either normal or inclined at 45 deg to the applied
stress direction; the latter condition is more prevalent in thin plates
and sheet material. The gross orientation of the fracture surface de-
pends on the specimen geometry, material properties, and the stress
conditions during the crack propagation sequence. This is readily
seen by consideration of the concepts of fracture mechanics [ l - 3 ] ?
When the stress-intensity factor, a measure of the stress amplification
at the crack tip of the advancing crack front, is small, the resulting
plastic zone at the crack tip is also small. When the thickness of a
component or test panel is large compared with the size of the crack-
tip plastic zone, the fracture will propagate on a plane normal to the
stress direction under plane strain conditions. When the size of the
plastic zone becomes large compared with sheet thickness before the
incidence of stage four (final fracture), the fracture surface will shift
to a plane at 45 deg to the sheet thickness and stress direction, and
stage three propagation will occur, reflecting plane stress conditions.
When the initial size of the plastic zone is very large compared with
sheet thickness, stage three propagation will begin immediately and
preclude stage two growth.
In the case of service failures occurring over long periods of time,
the fracture surface contains lines referred to in the literature as
"clam shell" markings, arrest lines, and "beach markings." These
markings have been attributed to periods of crack extension followed
by crack arrest. During the arrest periods, the fracture surface is
presumed to be oxidized or corroded; consequently, the alternate
periods of crack growth and arrest result in the formation of a fracture
surface containing concentric rings of varying colors, thereby indicat-
ing the intermittent nature of the fatigue cracking process. In a re-
cent review article, Jacoby [4] cited examples of clam shell markings
and associated color changes on the fracture surfaces of specimens
subjected to programmed loading. The explanation for these markings
is not clear at the moment, but it does indicate that the arrest mark-
ings generally reveal perturbations in the loading pattern during the
T h e italic n u m b e r s in brackets refer to the list of references appended to this paper.
McMILLAN AND HERTZBERG ON APPLICATION TO FATIGUE STUDIES 91
life of the component. By contrast, when simple sinusoidal loading
patterns are imposed on laboratory specimens, no evidence of such
markings is found.
Microscopic Fracture Surface Appearance

A detailed examination of the clam shell markings on service frac-
tures and the fracture surfaces of laboratory specimens subjected to
simple loading patterns has revealed that they are composed of many
smaller parallel lines referred to as fatigue striations (Fig. 1). (In all
FIG. 1 - O p t i c a l f r a c t o g r a p h of fatigue striations in an aluminum alloy.

the photographs the large arrow indicates the general direction of

crack propagation.) Observation of this fine structure, often orders of
magnitude smaller than the clam shell markings, has prompted exten-
sive examination of fatigue specimens by several fractographic tech-
niques.
Zappfe and Worden were the first to report patches of parallel mark-
ings on the fractured surfaces of fatigue specimens [5]. During the
early 1960's, work continued to establish more clearly the details of
these striations and to determine the extent to which they were found
in many classes of engineering materials. It was during this period
that the techniques of electron fractography were perfected and em-
ployed in this investigation. Under Forsyth and Ryder [6-9] and
Crussard and co-workers [I0,11] much progress was made. It was
found that fatigue striations appeared on the fracture surfaces of
faced-centered-cubic, body-centered-cubic and hexagonally-closely-
packed metals and even in amorphous polymer materials [12]. Of
extreme importance was the determination of a possible correlation
between the number of fatigue striations found on the fracture surface
and the number of loading cycles. It was not until Forsyth and Ryder
[6] performed their critical experiment that this point was clarified.
They proved that each load excursion was responsible for one stria-
tion and that the size (that is, distance between markings) depended
strongly on the stress amplitude. Thus, it is appropriate to emphasize
that there is a clear distinction between the macroscopically observed
clam shell markings, which represent periods of growth during which
thousands of loading cycles may have occurred, and the striations,
which represent the extension of the crack front during one load excur-
sion. There may be thousands or even tens of thousands of striations
within a single clam shell marking.
Qualitative Aspects of Fracture Surface Microdetails

As discussed earlier, the propagation of a fatigue crack can occur
either as stage two plane strain propagation or as stage three plane
stress propagation. An examination of the fracture surfaces resulting
from these two modes reveals that the two fracture processes are some-
what different. In the plane strain region, the fracture surface contains
many areas of fatigue striations. In some of these the striations are
well defined, while in others they are only roughly outlined. Two gen-
eral observations about fracture surfaces in the plane strain region in
metals are that adjacent packets of striations are often at large angles
to each other (Fig. 2) and that there are regions where no striations are
observed. The significance and interpretation of these observations
will be discussed later. In the plane stress region, the fracture surface
contains areas of elongated dimples resulting from the resolved shear
FIG. 2-Large angle between adjacent packets of fatigue striations in an aluminum

alloy.
stresses on the 45-deg fracture planes, far fewer indications of fatigue

striations, and abraded or rubbed areas that obliterate many of the de-
tails of the fracture surface (Fig. 3).
Striation Morphology
Fatigue striations can take many forms. They may be clearly or
poorly delineated, as described earlier, and may take on many shapes.
Forsyth, for example, has made a distinction between ductile and brit-
tle striations [8,9]. Examples are shown in Fig. 4. It is not absolutely
clear why these morphologies are different, but they are often associ-
ated with the test environment during crack propagation. Fatigue stri-
ations tend to be much flatter and take on a brittle appearance when
formed in an aggressive environment, but tend to be generally more
0
z
(a) Elongated dimple formation. (b) Rubbing marks.

FIG. 3 - Fractographs of plane stress fatigue fracture surface of 2024-T3 aluminum alloy.
FIG, 4 - E l e c t r o n fractographs o f J~ltigue striations in an aluminum alloy.

ductile when formed in an inert environment. The effects of environ-

ments and alloy composition on fatigue crack propagation rates and
striation morphology will be discussed in a later section. While stri-
ation formation may be affected by environment, drastic changes have
been observed in the appearance of fatigue striations after exposure to
oxidizing or corroding atmospheres. Thompson and Mulhearn [13]
observed that fatigue striations were completely obliterated as a result
of exposure to a high-temperature, oxidizing environment. Striations
have also been noted to be both continuous and noncontinuous, clearly
and poorly defined, and straight and smoothly curved in different
materials.
From metallographic sections and electron fractographic examina-
tion, three basic interpretations of the morphology of fatigue stria-
tions have evolved. The striations are considered to be undulations
on the fracture surface with (1) peak-to-peak and valley-to-valley
matching of the two mating surfaces, (2) matching crevices separating
flat facets, or (3) peak-to-valley matching. Based on these interpreta-
tions of striation morphology, different mechanisms have been pro-
posed for striation formation. These mechanisms have recently been
reviewed [14], so a detailed discussion will not be given at this time.
One aspect of the models will be reviewed, however. The earlier mod-
els, proposed by Forsyth and Ryder [7] and Laird and Smith [15], did
not relate striation formation with crystallographic considerations, but
this was proposed by Stubbington [16], Jacoby [4], and Hertzberg
[I 7]. More recently, Laird [14] considered some of the details of the
deformation process at the crack tip and the role of crystallographic
effects on striation morphology. He argued that striation morphology
would be more symmetrical if the individual grains were favorably
oriented with respect to the stress pattern at the tip of the crack. This
evaluation reflects agreement with Stubbington, Jacoby, and Hertz-
berg as to the importance of crystallographic considerations, although
not with their proposed mechanisms for striation formation.
The effect of crystallography on striation formation can be supported
by much circumstantial evidence. For example, if the slip planes are
favorably oriented with respect to the stress pattern at the advancing
crack front, then a clearly defined striation could be formed. A similar
argument relating to slip plane orientation could explain the formation
of poorly defined striations or none at all. Large angle differences found
between adjacent packets of striations could be explained in terms of
the misorientation of adjacent grains, or the crystallographic orienta-
tion change associated with a twin boundary. McEvily and Boettner
[18] have explained the difference in the curvature, of striations ob-
served in several alloys in terms of the stacking-fault energy of the
material, a crystal structure property: striations become straighter
and the crack growth rate decreases with decreasing stacking-fault

energy [18,19].
It has been argued that it is not correct to base striation formation
mechanisms on crystallographic considerations, since striations are
clearly observed in amorphous polymer materials [12]. There is no in-
consistency in this approach, however. A striation will form as a re-
sult of the stress pattern at the crack tip. Because a polymer material
is amorphous, there is no restriction on the manner in which the poly-
mer may deform. Thus, it will deform in accordance with the prevailing
stress pattern. In metals, where there are a restricted number of pos-
sible slip planes and directions, a striation will be formed when the slip
systems are favorably oriented with respect to the stress pattern at the
crack tip. If the slip systems are not favorably oriented, a poorly de-
fined striation or no striation will result. The observation of striation-
free regions on the fatigue fracture surfaces of metals and of 100 per
cent striation formation on polymer fatigue fracture surfaces supports
the aforementioned argument.
Quantitative Aspects of Fracture Surface Microdetails
More quantitative information has been obtained from the measure-
ment of fatigue striations than from any other fracture surface detail.
Since the striation represents the position of the crack front at the end
of every loading cycle, it can be used to measure the fatigue crack
propagation rate at a given position along the fracture surface. The use-
fulness of this observation depends on the correlation of the growth
rate measurements with some quantity or quantities that reflect the
prevailing stress conditions at the tip of the advancing crack front.
Pelloux [20], Schijve [3], and Hoeppner et al [21] have found that stri-
ation spacing increases with increasing crack length under constant-
load range conditions. As mentioned earlier, the experiment by For-
syth and Ryder [6] showed that striation spacing was a function of the
load amplitude. Since the stress-intensity factor reflects both stress
level and crack length in the description of the stress conditions at the
crack tip, it was logical to determine whether this factor could more
completely correlate striation spacings. Paris and Erdogan [22,23]
clearly showed that the macroscopically measured crack propagation
rates were strongly related to the stress-intensity factor range at the
crack tip. Extending this approach to fractographic observations,
Hertzberg and Paris [2], Carmen and Shuler [24] and McMillan [25]
showed that as the stress-intensity factor increased, the spacing of the
fatigue striations also increased. Furthermore, there was a general cor-
relation between the macroscopically determined growth rate as meas-
ured with a traveling microscope and the microscopic growth rate as
measured by the striation spacings. More recently, McMillan and
Pelloux [26] have demonstrated that striation spacing is affected by

both maximum and range values of the stress-intensity factor.
In an attempt to determine the rate of crack propagation by measur-
ing elongated dimple widths during stage three propagation, Hertz-
berg [15] found that the width of the dimples was a constant value even
though the stress-intensity factor increased. Since the macroscopic
growth rate increased by over two orders of magnitude while the dim-
ple width remained constant, it was assumed that microscopic growth
was produced by an increasing number of dimples being formed during
each loading cycle.
Though the stress-intensity factor has been proved to affect crack
growth rates, there are many factors that also must be considered,
for example, the roles of mean and maximum stress level, cyclic fre-
quency, and test environment on crack propagation rates. Once these
roles are fully understood, it will become possible to determine by
postfracture analysis the stress conditions in a fatigue-fractured com-
ponent by measurement of the fatigue striations.
Effect of Loading History

It has been known for some time that load history can affect subse-
quent fatigue behavior. With respect to crack propagation studies,
Hudson and Hardrath [27] and Schijve [28] have observed that the
macroscopic crack growth rate appears to decrease with the imposi-
tion of peak loads in the loading sequence. This retardation of crack
velocity has been associated with a favorable residual stress pattern at
the crack tip. However, it has been reported by Jacoby [4], Hertz-
berg [17], and McMillan and Pelloux [26] that striations formed im-
mediately after the peak loads were larger than those formed prior to
the peak loads, even though the applied stress-intensity conditions
were the same (Fig. 5). This behavior tended to decrease with increas-
ing distance from the crack position where the peak loads were ap-
plied [15]. It is thought that this behavior might be related to strain
hardening or to some kind of conditioning of the material ahead of the
crack front. The observed macroscopic crack arrest and microscopic
crack acceleration is significant and shall be considered in a later sec-
tion in the light of new experimental observations.
Recent Experimental Observations

In this section, quantitative and qualitative electron fractography
results from three recent fatigue crack propagation studies are dis-
cussed: (1) the effect of prior loading history on subsequent crack
propagation, (2) the smallest observable growth increments, and (3)
the effect of environment on fatigue crack growth.
McMILLAN AND HERTZBERGON APPLICATION TO FATIGUE STUDIES 99
FIG. 5 -Fractograph of 7075-T6 aluminum alloy revealing increased striation spac-

ing ar peak load cycles.
Effect of Prior Loads on Subsequent Crack Propagation

These experiments were designed to investigate the effect on subse-
quent crack growth rates of superimposing periodic peak loads on uni-
form cyclic loads. The peak loads are termed positive when the super-
imposed load is greater than the maximum load of the uniform cyclic
loads and negative when the superimposed load is less than the mini-
mum load of the uniform cyclic loads. Schijve et al [29] previously
examined similar load histories and observed that the introduction of a
positive peak load can have a very favorable influence on the fatigue
life because crack extension following the peak load is arrested or
severely retarded. The retardation was attributed to the develop-
ment of compressive residual stresses in the plastic zone ahead of
the crack that resulted from the peak tensile load. If the positive peak
load was followed by a negative peak, the retardation, while still ap-
parent, was significantly reduced. In that instance, the effect of the
negative peak presumably was to remove part of the residual com-
pressive stress field.
Hardrath [30], in a related series of tests, showed that under uni-
form cyclic loading a change in maximum stress from o-1 to o-2 resulted
in acceleration of crack growth when 0-1 < o-z, and arrest or retardation
when o-1 > o-5. Hardrath also suggested that compressive residual
stresses near the crack tip accounted for this behavior. Rice [31] con-
sidered several models and argued that blunting of the crack tip by
large deformations was a more plausible explanation for the retarda-
tion of growth.
We felt that electron fractography could provide additional informa-
tion concerning crack arrest or retardation, and specimens from two
test programs with peak loads were examined. In the first set of tests,
a positive and a negative peak were introduced after every 200 uni-
form load cycles. In one loading program, designated P 14, the nega-
tive peak preceded the positive peak and in the other, designated P 15,
the order was reversed. The loading sequences are shown in Fig. 6.
(Programs P 1 through P 12 were discussed in a previous paper [26].)
The test specimens were 9 by 24 by 0.160-in. center-notched aluminum
2024-T3 fracture panels with an initial crack length (2a) of approxi-
mately 3/4 in. The mechanical properties of the materials tested with
these program loads are listed in Table 1. The program loads were
introduced by punched tape digital programming, and crack lengths
versus crack growth rates are shown in Fig. 7.
The macroscopically observed growth rate was faster when the neg-
TABLE 1 - M e c h a n i c a l properties of materials in test programs

(transverse grain direction).
Reduction
Elongation, of Area,
Test Program Material Ft,, psi Ftu, psi % %
P 14 and P 15 ................ 2024-T3 70 100 45 600 20 27

(peak loads) ................ 2024-T3 70 400 45 700 20
Environmental tests ......... 2024-T3 71 200 49 400 21 ~8

2024-T6 70 000 60 600 10 26
7075-T6 88 000 81 000 14 22
7178-T6 90 200 85 300 12 22
(from R e f 40) ....... 7000 series 0 C u 55 200 48 000 14.2 ...

7000 series 0 . 9 C u 58 400 46 600 16.5 ...
PROGRAM P14
201
ag (KSI)
201
TIME
PROGRAM P15 201

II
2 4 6 8 lg4 196 198 200 12
14
(~g (KSI)
10
2 4 6 192 194 196 198 200
201
TIME
FIG. 6-Oscillograph recordings for the applied loads of programs P 14 and P 15.
ative peak followed the positive peak (program P 15). On the premise
that crack extension occurs only on the load rise portion of the load
cycle, we expected that one larger-than-normal growth increment,
corresponding to the half cycle 201-201, would be associated with the
peak load in program P 14, while two larger growth increments, cor-
responding to the half cycles 200-201 and 201-1, would be observed
in program P 15. Characteristic electron fractographs of the fracture
topographies resulting from these two programs are shown in Figs.
8 and 9. As expected, program P 14 shows one large growth increment,
designated A in Fig. 8, while program P 15 shows two large growth in-
crements, designated B and C in Fig. 9. The magnitude of the stress
rise (AK) is the same for 200-201 and 201-1 in program P 15. The
growth increment is not only controlled by the magnitude of the load
excursions, but is also affected by the mean stress, as evidenced by the
difference in striation size of B and C in program P 15. The uniform
load cycles in each program caused growth, since it was possible to
count 201 separate growth increments for each program.
3,000'-
s
/
2,000 - g
/
/
/ /
,/ /
1,000 - P
, /
8O0 /
CRACK o/
/
A'
GROWTH
RATE,
600 - , /
d2ald SPECTRUM
(~II~/SPECTRUM) 7/ /"
400-
/./ .P,4
/ oP15
/ 9
0
200 -
O~ I I I I I I I I
0.5 1.0 2.0 3.0 4.0
CRACK LENGTH, 2a (IN.)
FIG. 7 - C r a c k growth rates f o r programs P 14 and P 15.
The two fractographs (Figs. 8 and 9) are of the same magnification

and were obtained at the same crack lengths. The growth increments
associated with the peaks are approximately the same size; that is,
increment A is approximately equal to B plus C, and cannot account
for the observed difference in macroscopic growth rates. The major
portion of the growth rate difference is due to the difference in propa-
gation rate resulting from the uniform load cycles. This difference
does not appear to be related to crack-tip blunting, since the growth
increments due to the uniform load cycles for both programs are larger
immediately following the peaks than they are immediately preceding
the peaks. This suggests that programs P 14 and P 15 introduce dif-
ferent amounts of plastic damage or different states of residual com-
pressive stress in the material so as to alter crack growth rate. A com-
plete analysis of these observations is presently impossible, since the
peak load plastic zones overlap, and the effects of this interaction are
not understood.
In the second set of tests to determine the effect of prior load history,
9 by 24 by 0.160-in. center-notched aluminum 2024-T3 fracture
FIG. 8 - F r a c t u r e topography resulting from application of program P 14.

FIG. 9 - F r a c t u r e topography resulting from application of program P 15.

M,-MILLAN AND HERTZBERG ON APPLICATION TO FATIGUE STUDIES 105
24, 000
GROSS
AREA
STRESS
(PSI)
12, 000
4, 000
TIME
F I G . 1 0 - Oscillograph recording of load-time history for application of a single peak

load.
panels with an initial crack length (2a) of approximately 1/2 in. were
subjected to a sinusoidal loading pattern with positive peak loads in-
troduced at specific crack lengths. A fixed-range load controller was
used to introduce the uniform loads, while the peak overload was in-
troduced by changing the load limits for one cycle. An oscillograph re-
cording is shown in Fig. 10. The time delays before and after the peak
were necessary for setting and resetting load limits. Only three peaks
per specimen were introduced at crack lengths (2a) of approximately
0.6, 1.2, and 2.4 in. to ensure that no interaction occurred between the
plastic zones for each peak. Estimates of the plane stress plastic zone
size based on Irwin's analysis [32] are tabulated in Table 2.
T A B L E 2-Estimates of plastic zone size.
Plane
o-~ (max), ~ (min), Stress,
2a ksi ksi K," k s i ~ 2r~b (in.)
0.6 ............ 12 4 11.6 0.02

0.6 ............ 18 4 17.4 0.046
0.6 ............ 24 4 23.2 0.08
1.2 ............ 12 4 16.4 0.04

1.2 ............ 18 4 24.6 0.09
1.2 ............ 24 4 32.7 0.16
2.4 ............ 12 4 23.3 0.08

2.4 ............ 18 4 35.0 0.18
2.4 ............ 24 4 46.6 0.32
K = o-~ (max) X / ~ .
~ 2ru = 1 ( ~ - y plane stress.
] 06 ELECTRON FRACTOGRAPHY
The crack length versus cycles data are shown in Fig. 11. The data
are from the baseline case (maximum gross area stress of 12 ksi but
with no peak loads) and from tests where peak loads of 18 and 24 ksi
were introduced. Very pronounced crack retardation occurred with the
24-ksi peaks. It was not possible to tell by macroscopic measurement
whether complete arrest had occurred (particularly over the whole
crack front) or whether growth from the 12-ksi uniform loadings was
still occurring, but at a much slower rate. At the end of each retarda-
tion period, crack growth again progressed at about the same rate as
in the baseline specimen. With the 18-ksi peak loads, the retardation
effects were substantially reduced.
The magnitude of the retardation depended on the magnitude of the
peak loads. When the peak loads were 11/2times the uniform maximum
cyclic load, each retardation period was less than 10 per cent of the
baseline life. However, when the peak loads were twice the uniform
maximum cyclic load, each retardation period was equivalent to the
total life of the baseline case.
These specimens were examined fractographically in the region of
the peak load application to assess the magnitude and type of crack
growth associated with the maximum load and its effect on subsequent
growth at lower loads. In Fig. 12, the effect of the 18-ksi peak load at
2a = 1.2 in. is shown. The characteristic striation spacing 0.003 in.
before the peak application is shown in Fig. 12a, and the extent of the
growth due to the 18-ksi peak, designated P, is shown in Fig. 12b.
There is little or no difference in striation spacings before and im-
mediately after the peak load application. However, as the crack propa-
gates away from the peak, the striation spacings become progressively
.... UNIFORM LOADS (12 KSI)

_ _ . m UNIFORM LOADS + 18-KSI PEAK LOADS
~UNIFORM LOADS + 24-KSI PEAK LOADS
o APPLICATION POINTS FOR PEAK LOADS
7.0
6.0 II9
I
CRACK
5.0 '1
I
LENGTH, 4. 0
2a (INCHES)3.0
2.0
1 . 0 ~
0 i I I I I I
100 200 300 400 500 600
CYCLES (THOUSANDS)
FIG. 11- C r a c k retardation due to application o f peak loads.
FIG. 12-Effect o f 18-ksi p e a k l o a d at 2a = 1.2 in.

FIG. 12-Continued
smaller, reach a minimum, and then become progressively larger until

they are approximately the same size as those in the area preceding the
peak (Figs. 12b, c, and d). The significant finding is that the maximum
retardation occurs not at the peak load application point, but later, as
the crack propagates through the peak load plastic zone.
The clear definition of the peak load and subsequent striations im-
plies that there was no microscopic crack arrest. Propagation contin-
ued with the next smaller load-amplitude cycle and the macroscopically
observed effects were due to a retardation of growth rates. However,
this situation was not as clear-cut when the peak stress intensity was
higher, as shown in Fig. 13. Again, in Fig. 13a, the point of applica-
tion of the peak load, designated S, is readily apparent. However, the
total extent of the peak load growth increment is not clearly defined
and is probably beyond the boundary of this photomicrograph. Exten-
sive deformation, as evidenced by the deformation bands, designated
D, and some tearing of the matrix, designated T, accompanied these
peak load applications. Moreover, the fracture surface immediately
following the peak load was badly damaged by fretting and abrasion
(Figs. 13b, c, and d), indicating contact between the two fracture
halves. Owing to the damage, striations are not readily visible in the
areas immediately following the peak, Fig. 13b. However, as the crack
propagates away from the peak, the extent of the surface damage de-
creases and striations are observed in protected regions, designated
F, in Figs. 13c and d. These striations are extremely fine and not yet
well defined even 0.050 in. past the peak.
In analyzing the results of the peak load tests, three factors should
be considered: (1) mechanical factors, (2) plastic zone conditions, and
(3) residual stress distributions. It is obvious in Figs. 12 and 13 that
large plastic deformations accompany the application of the positive
peak loads and that the extent of the deformations is a function of their
magnitude. Blunting of the crack tip would be expected. However,
when the load is lowered, reversed plastic flow occurs that resharpens
the crack tip. The crack extension model discussed by Schijve [33]
demonstrates how crack-tip resharpening may occur. The fact that stri-
ation spacings immediately following the peak load are the same as
those immediately preceding the peak (Fig. 12) suggests the effective-
ness of this resharpening process.
When the peak load is applied, a large volume of material yields
plastically; the remainder of the specimen cross section is strained
elastically. As the load falls, a smaller volume of material yields
through reversed flow. At the same time, the remainder of the mono-
tonic plastic zone experiences elastic compressive stresses due to the
relaxation of elastic tensile strains in the surrounding material. The
magnitude of these compressive stresses is a function of the plastic
] 10 ELECTRONFRACTOGRAPHY
FIG. 13--Effect o f 24-ksi peak load at 2a = 1.2 in. The photomicrographs are a)~-
ranged to show the actual spatial relationships.
McMILLAN AND HERTZBERG ON APPLICATION TO FATIGUE STUDIES 11 1
volume and the extent of the elastic strain recovery. Subsequent posi-
tive excursions of smaller amplitude will be insufficient to completely
offset the surrounding compressive stress field established by the peak
load. Based on the fractographic evidence, the macroscopically ob-
served crack arrest due to occasional peak loads can be attributed to
the microscopic crack retardation due to the compressive residual
stress zone in front of the crack.
Smallest Observable Growth Increments

Acknowledging the importance of quantitative measurements of
fatigue striations, the Electron Fractography Subcommittee has begun
a study of the lower limit of striation size. In this program, it is pro-
posed to determine, within the limits of replication technique, how slow
fatigue cracks can propagate by the striation-formation growth mech-
anism. Based on the strong relation of striation spacings to the stress-
intensity conditions at the tip of the moving crack, these studies would
determine the tentative lower limits of stress-intensity conditions be-
low which striation-type growth might not occur. Such information
would be valuable in design considerations and would be useful in
numerical integration procedures of total fatigue life by furnishing
carefully measured, lower limit growth rate values. These studies
have thus far shown that striations spaced approximately 100 A apart
have been found on the fracture surfaces of aluminum alloys. Based
on the unit slip distance for aluminum, this suggests a dislocation mo-
tion of 30 to 40 Burgers vector units of translation. If it is assumed that
the striation formation process is related to some multiple of disloca-
tion translation, the minimum striation spacing would be expected to
be not much less than 50 to 100 A. In this connection, Linder [34] has
reported macroscopically determined crack growth rates in 7075-T6
aluminum alloy of the order of 2 x 10-1~ in./cycle. Since this rate is an
extension of only hundredths of an angstrom per cycle, not possible
for dislocation processes, it implies that the growth rate was discon-
tinuous across the crack front.
The experimental difficulties inherent in the direct measurement of
striation spacings less than 100 ,~ are formidable, and the measure-
ments themselves are subject to some controversy. It would seem de-
sirable to investigate indirect measurement to determine these values.
The introduction of periodic peak loads offers one possible tech-
nique for indirect measurement. In programs P 14 and P 15, we ob-
served that every load cycle produced a growth increment at the larger
crack lengths. There is reason to assume that each uniform load cycle
would also produce growth at progressively shorter crack lengths until
growth was mechanically impossible. A fractograph at crack length
2a = 0.81 in. (very near the origin) for program P 14 is shown in Fig.
FIG. 14-Spectrum spacing for P 14 at 2a = 0.81 in. is 54 ixin. One spectrum is

designated S.
14. The periodic repetition is readily apparent, with the peak loads
separated by growth bands attributed to the 200 uniform load cycles.
The crack extension per spectrum is measured at 54/zin./spectrum and
the 200 uniform load cycles account for approximately 45/xin. of this
growth. The growth increment per uniform load cycle, then, is less
than 60 ~. Attempts at direct observation of striations within these
bands have so far been unsuccessful.
Environmental Effects
Haigh [35] established 50 years ago that the choice of test environ-
ment could have an appreciable effect on the fatigue lives of certain
brasses and bronzes. Since this pioneering work, corrosion fatigue
studies have been extended to include many materials and many test
environments. Environmental effects were reported initially in terms of
total fatigue life. However, fatigue life can be divided into at least four
parts: (1) precrack initiation damage and crack initiation, (2) plane
strain stable crack propagation, (3) plane stress stable crack propaga-
tion, and (4) final fracture. More recently, workers have investigated
these life periods separately, and it appears that a major portion of the
effect occurs in the stable crack propagation stages. The exact mechan-
isms by which the environment influences crack growth have not been
Mr AND HERTZBERG ON APPLICATION TO FATIGUE STUDIES 113
firmly established, but the total effect appears to be a function of fre-

quency, stress range, and stress intensity.
It has been established that water vapor and various aqueous en-
vironments [36-38] can reduce the fatigue life and accelerate the crack
growth rate of certain aluminum alloys. In one of the few fractographic
examinations of corrosion fatigue specimens, Stubbington and For-
syth [39] compared the fracture appearances of specimens of an
AI-7.5Zn-2.5Mg alloy tested in air and in an aqueous 3 per cent
sodium chloride (NaCI) solution. They observed that the specimen
tested in air had the longer life and exhibited only ductile type stria-
tions on the fracture surface, whereas the specimen tested in the salt
solution exhibited brittle type striations.
The purpose of our test program was to determine the influence of
various environments on the fatigue crack propagation behavior of
several high-strength aluminum alloys. A fractographic analysis was
conducted in support of this program to determine whether observed
growth rate differences could be correlated with fracture surface fea-
tures. The specimens, 9 by 24-in. center-notched crack growth panels,
were tested in sinusoidal tension-tension loading at 120 cpm. The
standard material thickness was 0.160 in., but 0.040 and 0.50-in.
materials were also tested.
The initial phase of the program was conducted to determine the
relative effects of five environments on 7178-T6 aluminum alloy.
Desiccated air was chosen as the standard environment. The remain-
ing environments were: (1) high humidity air (relative humidity 90 per
cent at 70 F), (2) distilled water, (3) aqueous 3.5 per cent NaC1 solu-
tion, and (4) hydraulic fluid (Skydrol).
The results of this series of tests are shown in Fig. 15. The slowest
fatigue crack growth rate occurred in the standard dry air environment.
The fastest growth rates occurred in wet air and distilled water; the
growth rate was slightly slower in the salt solution. The growth rates
in Skydrol were lower than those in water-containing environments.
The environmental effect was greatest at the shortest crack length
(smallest stress intensities) and appeared divergent for extrapolations
to lower stress-intensity values. The growth rates in the initial stages
differed by as much as a factor of five. This information has particular
significance when it is realized that the major portion of crack growth
life is expended in the area where crack extension per cycle is ex-
tremely small, that is, less than 10 /xin./cycle. If these extrapolated
rates are truly divergent, then life changes of one order of magnitude
or greater could be expected. At the higher stress-intensity factor end
of the curve, the rates in all environments tended to converge, indicat-
ing that the mechanical aspects of fatigue crack growth were overshad-
owing any environmental effects.
•MAX : 12,000PSI FREQUENCY: 12OCPM /

(~MIN = 6,000PSI /o
1,000 o DRYAIR ./~ "
eWETAIR ~" - _
~. DISTILLEDWATER / so~
9 AQUEOUS 3.5~ NaCl SOLUTION Z /
CRACK
GROWTH
RATE,
d2a/dN
(1~IN./CYCLE)
IO0
lO i , I I I I I
0.5 1.0 2.0 3.0 4.0
CRACKLENGTH,2a (IN.)
FIG. 15-Effects o f environments on crack growth rate o f 7178-T6 aluminum alloy.
To shed some light on the environmental effects, the fracture sur-

faces of these specimens were examined by electron fractographic
techniques. Topographies typical of specimens tested in dry air and
wet air are shown in Fig. 16. In the dry-air specimen, finely spaced
ductile striations cover nearly 100 per cent of the fracture surface.
In contrast, the wet-air specimen (and the distilled water specimen)
exhibits only isolated patches of ductile striations, while the re-
mainder of the surface has a brittle, cleavage-like appearance. How-
ever, crack arrest markings are sometimes observed; this appears to
be a form of the brittle striations described by Forsyth [8]. Corrosion
product or corrosion damage is not apparent on these surfaces. On
the other hand, the fracture surfaces of the specimens tested in salt
solution and Skydrol both showed corrosion damage, but it was appar-
ent that ductile and brittle striations were present. However, we ob-
served that the proportion of ductile striations was significantly larger
on the specimens tested in Skydrol.
The gross changes in crack growth rates in dry air and wet air are
associated with the change in fracture mode from ductile to brittle
striation formation. This implies that, for the same stress condi-
tions, brittle striation growth is more rapid than ductile striation
growth. It is not clear at this time what differences exist in the mechan-
ism of formation of ductile and brittle striations. It should be noted
(a) Dry air.

(b) Wet air.
FIG. 1 6 - Typical fracture topographies o f 7 1 7 8 - T 6 aluminum alloy specimens tested
at 2a = 0.6 in. in wet and dry air.
that fatigue specimens tested in laboratory air and specimens with ser-
vice-induced fatigue cracks mainly show brittle striations in 7178-T6
aluminum alloy.
The second phase of the test program compared the fatigue crack
propagation behavior of several commonly used high-strength alumi-
num alloys in two environmental test conditions. Again, dry air was
chosen as the standard environment, and wet air was used for compari-.
son. The materials compared were 7178-T6, 7075-T6, 2024-T3, and
2024-T6. The mechanical properties of these materials are tabulated
in Table 1. The results of the environmental tests are summarized in
Fig. 17.
The two 7000 series aluminum alloys showed a significant difference
in crack growth rates in the two test conditions. Both alloys had es-
sentially the same rate in dry air, but 7178-T6 was noticeably more af-
fected by wet air. The 2024 alloy in both the room temperature and
artificially aged conditions showed no difference in growth rate as a
function of humidity.
Significant differences in fracture surface appearance were also
noted. In dry air, all four materials showed extensive ductile striation
development. Brittle striations predominated in the two 7000 series
1,000 DRY WET / ~

AIR AIR
o 9 7178-T6
9 ....... 7075--T6
,, 9 ............2024-T3
o 9 .... 2024-T6
~MAX = 12'000PSI / - -/ y
CRACK oA/tlN = 6, 000PSI " "
GROWTH
RATE, FREQUENCY= 120CPM /.,as"
d2aldN .s"
(F I N./CYCLE] /0
/ /"
S ~ ,
Jo f
/
lO0 - / ./' J / ,=
/ _.s" _@" / ~,."
/ _..," .r / .....
/ .s'"
s~
_L;r 6 .,*9..-"~
,#
//"
9 ,,'_oS ~
/ I
, 4 ..9176
. ....
oS j ~149
m," ~,o,O~176176149176
9 /tI5. "
11 I I o ~ ] I
1.0 2.0 3.0 4.0
CRACKLENGTH,2a (IN.)
F I G . 1 7 - C r a c k growth rate o f several aluminum alloys as a function o f environ-
ment.
McMILLAN AND HERTZBERG ON APPLICATION TO FATIGUE STUDIES 1 17
alloys when tested in wet air. In the 2024 alloy, however, brittle stri-
ations were not observed for either aging treatment in wet air.
Again, the pronounced changes in crack growth rates with environ-
ment are related to the change in fracture mode. However, it should be
noted that the rate may change without a concurrent change in fracture
mode. This was observed by Hyatt and Quist [40] while testing two
7000 series experimental alloys in dry and wet air. The chemistry of
the two alloys differed only in that one contained 0.9 per cent copper,
while the other contained no copper. The change in copper content
exerted only a small effect on the mechanical properties of these al-
loys as shown by the mechanical properties given in Table 1. The crack
growth rate data are shown in Fig. 18. The addition of 0.9 per cent
200
DRY WET
AIR AIR
150
o 9 SA-19 0.9%Cu C
o 9 SA-20
i00
50
CRACK
GROWTH
RATE,
d2aldN
(t~IN./CYCLE)
10
I I I I I I
6 7 8 9 10 15
STRESS-INTENSITYFACTOR.K(KSI ~ )
FIG. 18-Effect of environment on crack growth rate o f two experimental 7000
series aluminum alloys as a function o f copper content (from R e f 40).
copper slows the growth rate and, while not eliminating the environ-
mental effect, substantially reduces its severity.
In the subsequent fractographic examination, Fig. 19, the material
without copper exhibited the expected topography, that is, ductile
striations in dry air and brittle striations in wet air. The surpris-
ing result was that the specimen containing copper exhibited duc-
FIG. 19-Typical fracture topographies of two experimental 7000 series aluminum

alloys tested in wet and in dry air (from R e f 40).
tile striations in both environments, even though a positive environ-

mental effect was observed in the crack growth rate data. On the basis
of this observation it is not possible to make a general statement con-
cerning the relation between a positive environmental effect and a
transition in striation morphology from ductile-to-brittle type.
When the fracture surfaces of the aluminum specimens tested in
different environments are examined optically, the fracture mode
changes are readily apparent. The brittle striations occur typically on
flat, highly reflective facets whose orientations are nearly normal to
the tensile axis. Around these flat facets are patches of ductile stria-
tions on the inclines. When 100 per cent ductile striations are ob-
served, the flat reflective facets do not occur, and flat fracture planes
normal to the tensile axis are not common. It is also observed that
when brittle and ductile striations are present in adjacent areas, the
crack front leads and the striation spacing is larger in the brittle stri-
ation area. The ductile striations trail behind, as shown in Fig. 20.
Changes in fracture appearance as a function of environment are not
limited to aluminum alloys or to aqueous environments. Spitzig [41]
reported in acceleration of fatigue crack growth in an 18Ni (250)
maraging steel tested in a dry hydrogen atmosphere. The cracking
mode was transgranular, with the resulting structure appearing
F I G. 2 0 - O p t i c a l fractograph of brittle and ductile striations in adjacent areas of an

aluminum alloy.
very similar to the brittle striations in aluminum alloys. When this

same material was tested in dry or humid argon or humid hydrogen,
the cracking mode was characterized by ductile striations.
Another case where the fracture appearance changed as a function
of environment was reported by Dahlberg [42]. The material tested
was AISI 4340 steel, heat treated to very high strengths (approxi-
mately 260 to 280 ksi ultimate tensile strength). The environments
chosen were dry and wet air. The crack growth rate in dry air was sig-
nificantly slower than in wet air. The fracture path in dry air was
transgranular, with some striations visible, whereas in wet air the frac-
ture path was intergranular, along the prior austenite grain boundaries.
Summary
The resolution and depth of field of the electron microscope allow
us to examine fatigue fracture surfaces in minute detail. Concurrent
with improvements in the interpretation of fractographic observations,
electron fractography has evolved from producing purely qualitative
results to the point where quantitative information may be derived.
Different surface characteristics of fatigue fracture surfaces may be
observed, depending on the level of magnification. With little visual
enlargement, many service failure fatigue surfaces show "clam shell"
markings that indicate that fatigue crack propagation occurred in fi-
nite stages separated by crack arrests. Electron fractographic examina-
tion at high magnification reveals the clam shell markings to be com-
posed of many fine striations representing crack growth increments for
each loading cycle. By careful measurement of these striations, it has
been possible to relate fatigue crack propagation rates with crack-tip
stress-intensity conditions. Changes in striation appearance and orien-
tation on the fracture surface have been associated with crystallo-
graphic considerations.
The principal findings with respect to the recent experimental results
are as follows:
1. When periodic positive and negative peaks are superimposed on
a uniform loading history, the sequence of applying the peak load has
a pronounced effect on the subsequent crack growth rate.
2. When occasional peak loads are superimposed on a uniform load-
ing history so that the peak load plastic zones do not overlap:
(a) Crack retardation results.
(b) Crack-tip blunting occurs, but the crack resharpens on the load
decrease portion of the peak load cycle and crack propagation contin-
ues with the next load cycle.
(c) The cracking rate immediately before and immediately after
the peak load application point is approximately equal.
(d) Crack retardation occurs in the plastic zone ahead of the crack
and the extent of the retardation is governed by the magnitude of the

peak load.
3. The effects of environment on crack growth rate in several com-
mercial aluminum alloys are:
(a) In 7178-T6, crack growth rate is slowest in dry air and progres-
sively faster in Skydrol, aqueous 3.5 per cent NaC1 solution, distilled
water, and wet air. The growth rates differed by as much as a factor
of five.
(b) In both 7178-T6 and 7075-T6 when tested in dry and wet air,
a pronounced acceleration of growth rate was observed in wet air.
Alloys 2024-T3 and 2024-T6 showed no difference in growth rate in
these two environments.
(c) The effect of environment was most pronounced at the lowest
stress-intensity values.
(d) Changes in cracking rate at a constant stress intensity were usu-
ally associated with changes in fracture mode, with the faster rates
favoring a more brittle mode.
/t c k n o w l e d g m e n t s
The authors wish to thank W. E. Anderson, C. S. Carter, and
R. M. N. Pelloux for their helpful suggestions and comments concern-
ing the text. One of the authors (R. W. Hertzberg) is grateful for sup-
port from the National Science Foundation under Grant GK-1225.
The assistance of T. E. Kane in performing the fatigue tests and R. R.
Boyer in conducting the fractographic analyses is gratefully acknowl-
edged.
References
[1] Irwin, G. R., "Fracture Mode Transition for a Crack Traversing a Plate," Journal
of Basic Engineering, Transactions, American Society of Mechanical Engineers,
Series D, Vol. 82, No. 2, June 1960, p. 417.
[2] Hertzberg, R. W. and Paris, P. C., "Application of Electron Fractography and
Fracture Mechanics to Fatigue Crack Propagation," presented at International
Conference on Fracture, Sendal, Japan, Sept. 1965.
[3] Schijve, J., "Analysis of the Fatigue Phenomenon in Aluminum Alloys," NLR
Technical Report M2122, Amsterdam, The Netherlands, July 1964.
[4] Jacoby, G., "Fractographic Methods in Fatigue Research," Experimental Mech-
anics, Vol. 5, No. 3, March 1965, p. 65.
[5] Zappfe, C. A. and Worden, C., "Fractographic Registrations of Fatigue," Trans-
actions, American Society for Metals, Vol. 43, 1951, p. 958.
[6] Forsyth, P. J. E. and Ryder, D. A., "Fatigue Fracture," Aircraft Engineering,
Vol. 32, No. 374, 1960, p. 96.
[7] Forsyth, P. J. E. and Ryder, D. A., "Some Results of the Examination of Alumi-
num Alloy Specimen Fracture Surfaces," Metallurgia, Vol. 63, 1961, p. 117.
[8] Forsyth, P. J. E., "A Two Stage Process of Fatigue Crack Growth," Proceedings
of the Crack Propagation Symposium, The College of Aeronautics, Cranfield, Eng-
land, Vol. I, Sept. 1961, p. 76.
[9] Forsyth, P. J. E., "Fatigue Damage and Crack Growth in Aluminum Alloys,"
Acta Metallurgica, Vol. 11, July 1963, p. 703.
[I0] Crussard, C., "A Study of Impact Tests and the Mechanism of Brittle Fracture,"
Journal of the Iron and Steel Institute, Vol. 183, June 1956, p. 146.
[11] Crussard, C. et al, "A Comparison of Ductile and Fatigue Fractures," Fracture
(Swampscott Conference), Wiley, New York, 1959.
[12] McEvily, A. J., Jr., Boettner, R. C., and Johnston, T. L., "On the Formation and
Growth of Fatigue Cracks in Polymers," Fatigue-A n Interdisciplinary Approach,
Syracuse University Press, Syracuse, N.Y., 1964, p. 95.
[13] Thompson, K. R. L. and Mulhearn, T. O., "The Obliteration of Fatigue Fracture
Striations by Corrosion," The Journal of the Australian Institute of Metals, Vol.
10, No. 3, Aug. 1965, p. 303.
[14] Laird, C., "The Influence of Metallurgical Structures on the Mechanism of Fatigue
Crack Propagation," Fatigue Crack Propagation, ASTM STP 415, American
[15] Laird, C. and Smith, G. C., "Crack Propagation in High Stress Fatigue," Philo-
sophical Magazine, Vol. 7, 1962, p. 847,
[16] Stubbington, C. A., "Some Observations on Air and Corrosion Fatigue of an
Aluminum-7.5% Zinc-2.5% Magnesium Alloy," Metallurgia, Vol. 68, 1963, p.
109.
[I 7] Hertzberg, R. W., "Fatigue Fracture Surface Appearance," Fatigue Crack Propa-
gation, ASTM STP 415, American Society for Testing and Materials, 1967, pp.
205-225.
[18] McEvily, A. J., Jr., and Boettner, R. C., "On Fatigue Crack Propagation in F. C. C.
Metals," Acta Metallurgica, Vol. 1 I, No. 7, 1963, p. 703.
[19] Miller, G. A., "Fatigue Crack Growth in Some Copper Base Alloys," Ph.D.
dissertation, Massachusetts Institute of Technology, 1965.
[20] Pelloux, R. M. N., "Fractographic Analysis of the Influence of Constituent Par-
ticles on Fatigue Crack Propagation in Aluminum Alloys," Transactions, Ameri-
can Society for Metals, Vol. 57, No. 2, 1964, p. 511.
[21] Hoeppner, D. W., et al, "Fatigue Behavior of Materials for the SST Electron
Fractographic Studies," Battelle Report for Contract NAS 1-4460, 30 July 1965.
[22] Paris, P. C., "The Growth of Cracks Due to Variations in goad," Ph.D. disserta-
tion, Lehigh University, Sept. 1962.
[23] Paris, P. C. and Erdogan, F., "A Critical Analysis of Crack Propagation Laws,"
Journal of Basic Engineering, Transactions, American Society of Mechanical
Engineers, Series D, Vol. 85, No. 4, Dec. 1963, p. 528.
[24] Carmen, C. and Schuler, M., "Low Cycle Fatigue Properties of 18 Ni-Co-Mo250
Maraging Steel," presented at American Society for Testing and Materials Sub-
committee Meeting on Electron.Fractography, Schenectady, N.Y., Sept. 1964.
[25] McMillan, J. C., "Fractographic Analysis of Fatigue Crack Propagation," pre-
sented at American Society for Testing and Materials Subcommittee Meeting on
Electron Fractography, Lehigh University, Oct. 1965.
[26] McMillan, J. C. and Pelloux, R. M. N., "Fatigue Crack Propagation Under Pro-
grammed and Random Loads," Fatigue Crack Propagation, ASTM STP 415,
[27] Hudson, C. M. and Hardrath, H. F., "Investigation of the Effects of Variable-
Amplitude koadings on Fatigue Crack Propagation Patterns," NASA Technical
Note D-1803, National Aeronautics and Space Administration, Aug. 1963.
[28] Schijve, J., "The Significance of Fatigue Cracks in the Micro-Range and Macro-
Range," Fatigue Crack Propagation, ASTM STP 415, American Society for Test-
[29] Sehijve, J., Broek, D., and de Rijk, P., "Fatigue Crack Propagation Under
Variable-Amplitude Loading," N LR Report NLS-RN-M. 2094, Naval Research
Laboratory, 1961.
[30] Hardrath, H. F., "Cumulative Damage," Fatigue-An lnterdisciplinaryApproach,
Syracuse University Press, Syracuse, N.Y., 1964, p. 345.
[31] Rice, J. R., "The Mechanics of Crack Tip Deformation and Extension by Fatigue,"
Fatigue Crack Propagation, ASTM STP 415, American Society for Testing and
Materials, 1967, pp. 247-311.
[32] Irwin, G. R., "Plastic Zone Near a Crack and Fracture Toughness," Seventh
Sagamore Ordnance Materials Research Conference, Aug. 1960.
[33] Schijve, J., discussion of "Fatigue Crack Propagation Under Programmed and
Random Loads" (see Ref 26).
[34] Linder, B,, "Extremely Slow Fatigue Crack Growth Rates in 7075-T6 Aluminum
Alloy," M.S. thesis, Lehigh University, Bethlehem, Pa., June 1965.
[35] Haigh, B. P., "Experiments on the Fatigue of Brasses," Journal of the Institute of
Metals, Vol. 18, 1917, p. 55.
[36] Broom, T. and Nicholson, A., "Atmospheric Corrosion-Fatigue of Age-Hardened
Aluminum Alloys," Journal of the Institute of Metals, Vol. 89, Feb. 1961, p. 183.
[37] Bennett, J. A., "Changes in the influence of Atmospheric Humidity During
Fatigue of an Aluminum Alloy," Journal of Research, National Bureau of Stand-
ards, Vol. 68C, 1964, p. 91.
[38] Bradshaw, F. J. and Wheeler, C., "The Effect of Environment on Fatigue and
Crack Growth in Aluminum and Some Aluminum Alloys," Applied Materials Re-
search, April [.966.
[39] Stubbington, C. A. and Forsyth, P. J. E., "Some Corrosion Fatigue Observations
on a High-Purity Aluminum-Zinc-Magnesium Alloy and Commercial D.T.D. 683
Alloy," Journal of the Institute of Metals, Vol. 90, 1961-62, p. 347.
[40] Hyatt, M. V. and Quist, W. E., "The Influence of Heat Treatment and Composi-
tion on the Stress Corrosion, Fatigue, and Fracture Properties of 7075 Aluminum
Alloys," presented at the AFML Fiftieth Anniversary Corrosion of Military and
Aerospace Equipment Technical Conference, 23-25 May 1967.
[41] Spitzig, W. A., "Effect of Hydrogen Environment on Fatigue Crack Propagation in
18 Ni(250) Maraging Steel," presented at the Electron Fractography Subcom-
mittee Meeting, Westinghouse Research and Development Center, Pittsburgh, Pa.,
March 1967.
[42] Dahlberg, E. P., "Fatigue Crack Propagation in High-Strength 4340 Steel in Humid
Air," Transactions, American Society for Metals, Vol. 58, 1965, p. 46.
N. A. Nielsen I
Environmental Effects on
Fracture Morphology
R E F E R E N C E : Nielsen, N. A., " E n v i r o n m e n t a l Effects o n Fracture

M o r p h o l o g y , " Electron Fractography, ASTM STP 436, American Society
ABSTRACT: The utility of fractography can be greatly increased if the tech-

nique can be applied on a broader basis to provide unambiguous interpretation
of fracture surface morphology (and thereby, identification of the fracture
mode) resulting from specific environmental exposure of stressed metals and
alloys.
This paper reviews the present status of electron fractography in this area
of expanded application. High-strength steels, martensitic and austenitic alloys,
and titanium and zirconium alloys are considered with respect to the effects
chemical environment, including high energy irradiation, induces in the frac-
ture characteristics of the alloys.
KEY WORDS: embrittlement, fractography, electron microscopy, fracture,

stress corrosion, radiation effects, stainless steel, ferritic alloys, titanium, zir-
conium
With the acceptance of electron fractography as a unique and im-

portant tool in the observation and analysis of the micromechanisms
of "simple" fracture processes in metals, new extensions and applica-
tions of the techniques are now being found in more complex fracture
processes where special environmental conditions have a controlling
effect. Environmental conditions are here arbitrarily defined as the
"chemical" environment (gaseous or liquid) as well as exposure to high
energy radiation. When fracture occurs in these environments, it is be-
cause the specific environmental conditions often act synergistically
with the applied or residual stresses to produce failure which, in the
absence of the specific environment, would not occur. Examples of this
can be found in the phenomena of stress-corrosion cracking, stress-
accelerated intergranular corrosion, hydrogen embrittlement, liquid
metal embrittlement, corrosion fatigue, etc.
1 Research fellow, Engineering Materials Laboratory, Engineering Department, E. I.
du Pont de Nemours & Co., Inc., Wilmington, Det.
12,1
NIELSEN ON ENVIRONMENTAL EFFECTS ON FRACTURE MORPHOLOGY 125
Because the application of electron fractography to all of these areas

is relatively recent, interpretation of the information in the fractographs
is difficult and often incomplete and problematical. The effects of cor-
rosion, the solution of gases, or the effects of irradiation damage super-
posed with a static or dynamic stress state on the metal or alloy
generally produce fractures of complex and varying surface morphol-
ogies. This is particularly true for stress-corrosion fractures where
fractography provides more information than can yet be reliably
interpreted and agreed upon.
The purpose of this paper is not to provide a comprehensive "state-
of-the-art" review of the subject but to indicate by chosen examples
how electron fractography is being used in several of the previously
mentioned areas in the attempt to provide new information on environ-
ment-induced fracture processes in alloys. The writer's primary
utilization of the technique has been in the subject of stress-corrosion
cracking of stainless steels, and this application will be emphasized.
When other systems are mentioned, the contributions of other indi-
viduals will be acknowledged.
Stress Corrosion and Hydrogen Embrittlement
High Strength Steels

Premature failure often occurs in these materials through the em-
brittling effect of hydrogen or by a stress-corrosion mechanism in
which hydrogen evolution may or may not play a secondary role. Be-
cause the fractures in both types of failure are primarily intergranular,
one of the earliest hopes for electron fractography has been that this
technique will be able to provide microtopographical information
which can reliably and uniquely distinguish a failure caused purely by
hydrogen embrittlement from one in which corrosion is involved.
Specific research on this subject has been carried out by Phillips et al
and is summarized in Section Six of the Electron Fractography Hand-
book [1]. 2 They conclude that in AISI 4340, 4330M, and D64C steels
"no exclusive single feature precisely identified either fracture type."
It appears, however, that stress-corrosion fractures show a combina-
tion of features differing in degree from those observed in hydrogen em-
brittlement which may allow a differentiation to be made. On the basis
of their work, to which it is recommended the reader refer, it is obvious
that careful technique and observation must be employed in this
particular application of fractography. Knowledge of the exposure and
service history of the steel is also necessary to make a considered
judgment of the cause of failure.
lntergranular fracture can be produced by other mechanisms in high
2 The italic numbers in brackets refer to the list of references appended to this paper.
strength steels as Warke and Elsea [2] have pointed out. Quench
cracks, for example, are quite similar in their fracture surfaces to those
produced by hydrogen embrittlement and stress corrosion. More in-
formation than the fractographic data are presently needed to dif-
ferentiate these.
Tiner and Gilpin [3] have used tritiated water solutions in a study of
the microprocesses in stress corrosion of Type 4340 steel to elucidate
the role of hydrogen in the failure mechanism.
Martensitic Stainless Steels
The martensitic stainless steels when heat treated to very high
strength levels are subject to failure by stress corrosion in relatively
mild corrosive environments. For example, a USS 12MoV stainless
steel (12 per cent chromium, 1 per cent molybdenum with nickel and
vanadium) was reported by Phelps and Mears [4] to fail intergranularly
in 10 min when exposed as a stressed U-bend specimen to 1.0 M
neutral sodium chloride (NaC1). The standard commercial martensitic
alloys do not exhibit this extreme susceptibility, but nevertheless their
behavior is similar. Inasmuch as they can also be made to fail by hydro-
gen embrittlement when cathodically polarized (with applied cathodic
currents sufficiently high to stop all corrosion attack), the martensitic
alloys constitute another class of steels on which electron fractography
may cast further light in differentiating these two fracture mecha-
nisms.
Austenitic Stainless Steels

Another potentially fruitful material for this specific study is cold-
worked AISI 301 and 302 stainless steels. Phelps and Mears [4]
demonstrated that Type 302 stainless, cold-worked 75 per cent could
be made to fail either by stress-corrosion cracking (with applied anodic
current) or by hydrogen embrittlement (with applied cathodic current)
in a 3 per cent NaCI solution saturated with hydrogen sulfide (H2S).
Matsushima et al [5] have discussed the stress corrosion and hydro-
gen cracking of 17-7 stainless steel (Type 301). They concluded that
"contrary to reactions observed in pure phase alloys, stainless .steels
containing mixtures of austenite and ferrite may fail either by hydrogen
cracking or by stress corrosion, depending on the environment."
The writer has also confirmed that severely cold-rolled Type 301
stainless can be made to fail by either mechanism. Tested as stressed
U-bend specimens Type 301 sheet cold-reduced 70 per cent fails
rapidly by stress-corrosion cracking in boiling 42 per cent magnesium
chloride (MgC12) (in less than 1 h). Failures occurred overnight at room
temperature in 10 per cent ferric chloride (FeCI~) as well as in a
MgCI2-Na~2S slurry. When exposed to the latter environment and
simultaneously polarized with a strong cathodic current (0.1 A cm-~),

the stressed Type 301 stainless steel will crack as a result of hydrogen
embrittlement.
Examples of electron fractographs made of these specimens appear
in Figs. 1-4. It is obvious from these that the environment in which
fracture occurred had an important influence on the resulting fracture-
surface morphology. These fractographs were selected to illustrate
what the main points of typical difference appear to be. (However, it
must also be pointed out that in each set of micrographs made for each
fracture system, there is enough variation in topographical structure
that the examination of many areas is required for complete apprecia-
tion of the range of fracture morphology that can form in one system.)
Cold-rolling of Type 301 stainless steel develops a banded ferritic-
austenitic structure. The stress-corrosion fracture produced in boiling
42 per cent MgCL shows both elongated "blocky" strata and the ir-
regular topography more typical of stress corrosion in austenite (Fig.
1). Further observations, including the use of oxide replicas are re-
F I G . 1 -Stress-corrosion fracture o f Type 301 stainless steel in boiling 42 per cent

MgCl 2 (•
FIG. 2-Stress-corrosion fracture o f Type 301 stainless steel in 10 per cent FeCl~,
room temperature (•
quired to classify the fracture features characterizing each phase in this

alloy. (It is noted that Douglass et al [6] have made fractographs of
Type 310 stainless steel cold rolled 50 per cent and tested to failure in
boiling 42 per cent MgCI2. They reported that the austenite-martensite
interface appeared to be the preferential fracture path and indicated
that their work was the first observation of stress-induced martensite
in Type 310 stainless steel.)
The stress-corrosion fracture produced in 10 per cent FeCI3 shows
considerable etching and surface roughening which is to be expected of
FeCI3, an acidic pitting corrodent (Fig. 2). (The alteration of fracture
surfaces by subsequent reaction with their chemical environment oc-
curs in many other systems as well. It must be considered and
evaluated for each fractographic analysis of an environmentally in-
duced alloy failure.) The fractures, however, which developed on ex-
posure of the cold-worked and stressed Type 301 alloy to MgC12-

Na~2S, were entirely different in character. The path of cracking was
partially intergranular; the fracture surfaces showed elongated tear
dimples, the size of which varied with change in direction of the frac-
ture in following a local intergranular path. Again the ferrite and
austenite cannot be reliably differentiated in their topography (Fig. 3).
Lastly, failure induced in stressed Type 301 stainless steel by
cathodic charging of hydrogen develops more of a dual-structured
topography (Fig. 4). Areas of quasi-cleavage containing ridges of small
equi-axed rupture dimples were observed. Local areas of large equi-
axed dimples were also present (in what is believed to be austenite)
indicating that differentiation of austenitic and ferritic (martensitic)
areas by electron fractography may be accomplished in this or other
duplex ct-y alloys when failure occurs by a hydrogen embrittlement
mechanism.
In summary, it appears from these preliminary studies that electron
fractography will provide further information on the several fracture
F I G . 3 - Stress-corrosion fracture of Type 301 stainless steel in MgCI2 + Na2S, room

temperature (•
FIG. 4-Hydrogen-embrittlementfracture of Type 301 stainless steel (x9200).
mechanisms which can cause failure of the hardenable (by heat treat-
ment or cold rolling) grades of ferritic and austenitic stainless steel.
The differentiation and classification of fractographic features which
characterize failures in various stress-corrosion cracking environ-
ments, however, will require a more intensive and organized research
study than anyone has yet made in this field.
Stress-Corrosion Cracking of Austenitic Stainless Steels

The susceptibility of this class of steels tO transgranular stress-corro-
sion cracking (SCC) in solutions containing chloride ions has en-
gendered corrosion research on a world-wide basis. The objective of
the work in many, laboratories is to determine the mechanism of this
form of "brittle" fracture in the normally, inherently ductile austenitic
structure.
Stress-corrosion cracking takes the form of a low strain failure.
There is often little surface appearance of corrosion having taken
place. A failed specimen does not show evidence of plastic deforma-
tion. There are no macroscopic dimensional changes measurable, and
NIELSEN ON ENVIRONMENTAL EFFECTSON FRACTUREMORPHOLOGY 131
the path of cracking is mostly transcrystalline, hence, the designa-

tion "brittle fracture."
In the 1959 Pittsburgh Conference on Physical Metallurgy of Stress-
Corrosion Fracture, the present author [7] described an application of
fractography in research on the micromechanisms of SCC in austenitic
stainless steels. In restatement of the advantages of fractography in
this case, it can provide information on: (1) the ductile-brittle charac-
teristics of the topography, (2) fracture-face corrosion and precipitated
corrosion products in the crack system, and (3) the role of metallic or
nonmetallic phases which may embrittle the alloys by providing pref-
erential paths for fracture.
That study and a later one (Nielsen [8]) in which electron metallog-
raphy was employed to examine the early stages of corrosion attack by
chloride ions on stressed stainless steel surfaces, along with their tran-
sition into recognizable microcracks which lead to ultimate brittle
fracture, have confirmed the utility and value of fractography in stress-
corrosion research. If there is a weakness or limitation in the practice
of fractography, it exists in the present inability to interpret completely,
and with 100 per cent confidence, all of the information which the
fractograph (particularly the electron fractograph) presents. It is the
writer's conviction that inherent in the proper interpretation of electron
fractographs are many of the answers to the now puzzling questions of
crack propagation mechanism in SCC of austenitic stainless steels.
Figure 5 is a photomicrograph of an SCC fracture in Type 316 stain-
less steel. It shows surface morphology that is commonly observed in
these cases, namely, a feathery, dendrite-like surface topography.
Figure 6 is a second example of this taken from a failed section of
Type 316 stainless tubing. In this case the photomicrograph has been
made from the carbon replica of the fracture. It is included here to illus-
trate a minor problem in applying electron fractography to fracture
surface analysis. The difficulty is that of correlating surface topography
observable by light microscopy with the same surface features, in
replica form, examined in the electron microscope. The dendritic-like
structure in Fig. 6 shows up clearly because it is all within the depth of
field of the metallograph and properly oriented to reflect light back to
the observer. It stands out from the rest of the fracture surface for
these reasons. When the same area is examined in the electron micro-
scope, all of the fracture surface is now within the depth of field of the
microscope, and the outlines of the "dendrite" are lost. Many of the
features seen in the photomicrograph are no longer readily identifiable
in the electron micrograph. Examination based solely on electron
fractography would characterize the surfaces shown in Fig. 7 as cleav-
age faces or perhaps quasi-cleavage. What this indicates, of course, is
that the actual metallurgical surface, or if not convenient, the replica of
FIG. 5 - "Feather" morphology in fracture by SCC of Type 316 stainless steel tube
(x750).
FIG. 6 - " F e a t h e r " morphology in carbon replica of fracture by SCC of Type 316
stainless steel tube (x750).
FIG. 7 - E l e c t r o n micrograph of "feather" area of Fig. 7 (•
the fracture should also be examined on a metallograph in order to de-

velop a maximum of descriptive information on the fracture surface
morphology.
In other SCC failures of austenitic stainless steels (particularly
tubing in which residual tensile stresses from the forming operation
are present), the fracture mode is not always entirely transcrystalline
cracking. A small percentage of the fracture surface shows reflective
facets which represent grain surfaces exposed by an intergranular
corrosion cracking. However, some highly reflective surfaces may be
true cleavage surfaces which formed when fracture followed a prefer-
ential path along a pseudo-martensite-austenite interface in Type 304
steel. The exact mechanistic interpretation for the formation of frac-
ture areas such as shown in Fig. 8 is uncertain. These photomicro-
graphs are of matching areas in the two fracture surfaces. They
possess surface features ("cleavage steps") suggesting that their mode
FIG. 8-Matching "cleavage facets" in fracture surfaces of Type 304 steel failed in
boiling 42 per cent MgCl2 (x750).
of formation was mechanical rather than electrochemical- yet electro-

chemical corrosion must necessarily occur in SCC of these alloys. The
surface detail at x750 does not indicate, however, any corrosion attack
sufficiently intense to have altered or obscured the matching topog-
raphy. In this case and in future work precise electron fractographs
must be examined to determine the true degree of matching detail and
whether superficial corrosion has in fact occurred on these surfaces.
Such a study remains to be made to answer the point.
The stainless steels and refractory metals and alloys in general are
particularly amenable to electron fractographic study through the use
of oxide replicas. These possess the best dimensional stability of all
surface replicas. Being the actual specimen surface thermally con-
verted to a thin oxide film, this form of replica can provide the truest
topographical information (Mahla and Nielsen [9]). Figure 9 is an ex-
ample of oxide replication of a Type 304 stainless steel SCC fracture in
which mixed modes of crack propagation are present. The central area
shows a region of intergranular decohesion surrounded by areas of
transgranular fracture. The smooth topography of the grain surface
contrasts sharply with the topographic structure of the transgranular
fracture. The latter frequently shows "river markings" similar to
FIG. 9 - Mixed modes of fracture produced by SCC of Type 304 steel (oxide replica)
(xlO,O00).
cleavage structure commonly found in brittle fractures where the en-

vironment has had no effect on the fracture process. The striations that
are also present in the electron fractograph are best interpreted as
crack-arrest lines. Another example of these is shown in Fig. 10.
Striations of this nature have been observed in other low strain SCC
systems. They have been reported to occur in stress-corrosion cracked
aluminum alloys and copper alloys. They will probably be found in
other systems when electron fractography is applied to the study of
stress-corrosion failures.
If the striations are crack-arrest lines, SCC cannot be considered as
a confirmed form of purely electrochemical corrosion. The evidence
points to their origin in a two-stage crack propagation process where
short steps of corrosion alternate with steps of crack propagation by
mechanical fracture. This periodic process, which appears to be
crystallographically oriented, produces striations having a sawtooth
profile and running at right angles to the direction of crack propagation.
The dark patches present in the electron fractographs of Figs. 9 and
10 are corrosion-product films randomly extracted with the oxide
replica from the original fracture surfaces. That such corrosion
products play a role in the SCC mechanism has been theorized by
Nielsen [7] and Pickering et al [10].
The observation of striations in stress-corrosion fracture and their
FIG. 1 0 - Striated fracture surface produced by SCC of Type 304 steel (oxide replica)
(•
significance in terms of failure mechanism is being given increased

attention. Nielsen [7] and Logan [11] have commented on striations
markings in SCC of stainless steel. McEvily and Bond [12] have dis-
cussed striation formation in alpha brass and A1-5.5Zn-2.5Mg alloys
during SCC. Their interpretation is in terms of a mechanism of
periodic rupture of anodic films or tarnish layers. The common con-
clusion that investigation of these striation markings has developed is
that SCC is not solely an electrochemical mechanism but does involve
steps of mechanical fracture separating the stages of crack advance by
electrochemical corrosion.
Stress-Corrosion Cracking of Titanium Alloys

Most titanium alloys are now known to be susceptible to hot salt
stress-corrosion cracking or to accelerated failure or both when ex-
posed under stress in a fatigue precracked condition to a variety of
environments at ambient temperature. These environments include
water, methanol, nitrogen tetroxide, and many aqueous and non-
aqueous solutions and liquids. The mushrooming interest and activity
stems from 1965 when Brown [13] reported at the national ASTM
meeting that titanium will crack when exposed as a stressed and pre-
cracked cantilever beam to 3 per cent sodium chloride solution.
Hot salt stress-corroslon cracking has been known since 1957 [14]
and was discussed in a symposium held at the ASTM Pacific Area Na-
tional Meeting in Seattle, Wash., in 1965 [15]. Several of the studies
reported at this Symposium utilized electron fractography to provide
information on stress-corrosion crack propagation. Avery and Turley
[16] identified both intergranular and transgranular propagation in
5A1-2.5Sn alloy. This alloy was much less resistant to SCC than
4AI-3Mo-lV alloy which cracked primarily along an intergranular
path. The sites of crack initiation were grain boundaries and slip
planes emergent in the surface exposed to the hot (600 to 800 F) syn-
thetic sea salt.
One problem with replication of hot salt SCC fractures is the
presence of corrosion products and adherent oxide films on the frac-
ture surfaces. Martin [17], Hatch et al [18], and Rideout et al [I 9] men-
tion this point. Even so, distinguishing intergranular fracture from
transgranular fracture does not present special difficulty.
As examples of stress-corrosion cracking of titanium alloys at room
temperature, two examples are given in this paper, both through the
courtesy of C. D. Beachem of the U.S. Naval Research Laboratory.
Figure 11 is an electron fractograph of a 6AI-4V titanium alloy which
in test under sustained load in methanol (the specimen was not pre-
cracked) developed a stress-corrosion crack along a machining mark.
The fracture surface is characterized by elongated areas of cleavage
showing shallow river-pattern markings. Alternating and interspersed
bands and regions of dimpled rupture are also present. Meyn et al [20]
suggested that the latter areas which are characteristic of microvoid
coalescence may correspond to the ductile failure of ligaments remain-
ing between the cleavage areas. Similar features, but with somewhat
less cleavage, were observed in a similar specimen broken in distilled
water. The respective KI values for these stress-corrosion failures were
15ksi~and30ksi lX/~.
Another example of stress-corrosion cracking of a titanium alloy
(7AI-2Cb-ITa) in distilled water was reported by Meyn [21]. In this
all martensitic, a-phase alloy, the cleavage surfaces were sufficiently
large (alloy grain size was 1/16 in. diameter) to permit the use of Laue
back reflection X-ray diffraction techniques to determine the crystal-
lographic orientation of the cleavage planes. The cleavage facets
shown in Fig. 12 are inclined at an angle of 16 deg +_ 2 deg from the
(0001)a plane. They are high index planes, contrary to the normal ex-
pectation of a low index cleavage plane, but there are two (or more)
similar crystal planes along which the crack can propagate in a given
grain.
FIG. 1 1 - S C C in Ti-6A-4V alloy exposed to methanol [21] (crack initiated at ma-

chining mark on right) (X6000).
FIG. 1 2 - SCC in Ti-7A I-2Cb-I Ta alloy exposed to distilled water [21 ] (•
The reader interested in the subject of ambient temperature stress-

corrosion cracking of titanium alloys is referred to the bibliography by
Dahlberg [22] and to the DMIC Technical Note, 1 Feb. 1966 [23],
which contains a summary of the whole subject of accelerated crack
propagation of precracked titanium alloys. (The program of the 1967
Symposium at Battelle Memorial Institute is not known to the writer
at the time of this writing.)
Fraetography in the Study of Radiation Effects

ASTM Committee E-10 on Radioisotopes and Radiation Effects
has published four special technical publication volumes which discuss
the effects of radiation on engineering materials and on flow and frac-
ture behavior [24-27]. Two ASM-AEC monographs have been pub-

lished which also consider radiation effects on cladding, structural ma-
terials, and ferritic steels [28,29].
Although there are marked effects of neutron irradiation on the ten-
sile and impact properties of steel as Trudeau [28] has detailed, there
appears to be few reported applications of electron fractography on
ferritic, irradiated steels. Wilson and Berggren (see bibliography,
Chapter 3 in Ref 28) found that an ASTM A 106 steel irradiated at
300 C broke after 6 per cent elongation with no necking down. The
fracture surface showed about 50 per cent cleavage and 50 per cent
intergranular fracture.
Bush [29] has summarized the effects on neutron irradiation on the
mechanical properties of stainless steels. Marked increases in the yield
strength occur with less pronounced increases in tensile strength. De-
creases in elongation and reduction of area are observed. These effects,
however, are decreased in magnitude when post-irradiation annealing
and testing is carried out at or above 350 C. There is a nominal effect of
irradiation on the impact strength of austenitic stainless steels; in fer-
ritic stainless it is more marked.
In considering the loss of ductility phenomena which austenitic
steels exhibit following exposure to nuclear reactor environments, one
must distinguish between the water-cooled reactors operating in the
300 C temperature range and the gas-cooled (superheat) reactors
operating at about 600 C.
In the former case fracture is intergranular and apparently the result
of preferential grain boundary corrosion by the reactor coolant with
Cr +6 ions accelerating attack (in the absence of carbide precipitation
in the grain boundaries). The role of stress is one of acceleration of
grain boundary dissolution and of the rate of intergranular fracture.
The importance of the role of radiation is not clearly established.
Figures 13, 14, and 15, all of Type 304 stainless steel cladding, illus-
trate this type of failure. It should be pointed out that the "dimple-
like" areas in Fig. 14 contrast with the smooth intergranular separa-
tion shown in Fig. 13. That the features are not dimples was shown by
stereoexamination of this area of the fracture. They are small hills and
not to be confused with the equi-axed dimples characteristic of ductile
rupture. Figure 15 is considered to be a faceted intergranular fracture
(corrosion) surface on which crystallographic etching has oc-
curred [30].
When austenitic stainless steels and high-nickel alloys are exposed to
irradiation under superheat conditions, the temperatures, - 6 0 0 C, re-
sult in intergranular carbide precipitation. Very long exposures (2000
h) cause matrix strengthening which in turn promotes intergranular
fracture in room temperature tension tests. Figures 16 and 17 compare
Z
Z
0
z
0
z
0
z
o
o
FIG. 1 3 - Grain b o u n d a r y separation in f a i l e d T y p e 304 stainless FIG. 1 4 - N o d u l a r intergranular f r a c t u r e in Type 304 stainless
steel cladding [30] (• steel cladding [30] (• "'
z
,-i-
FIG. 1 5 - F a c e t e d fracture surface in Type 304 stainless steel FIG. 16--Tensile fracture in us-received Hastelloy-X [31]
cladding' [30] (x6000). (•
F IG. 17 - Extraction replica of mixed ductile-brittle fracture in Hastelloy-X after ex-

posure ~( 2000 h at 565 C [31] (•
the ductile fracture characteristics of Hastelloy X (22 per cent

chromium, 19 per cent iron, 9 per cent molybdenum, balance nickel
plus cobalt) in the as-received condition with those developed in the
same alloy after 2000 h at 565 C. In the latter case the fracture is
partially intergranular (brittle). Separation occurs along the interface
of an austenite grain and a continuous grain boundary network of
precipitated carbide. When a grain boundary is not conveniently
oriented relative to the axis of the applied tensile load, ductile trans-
granular rupture occurs [31 ].
The effects of irradiation show up only in tension tests at 600 or
700 C. Irradiated specimens fracture in a brittle intergranular mode.
Unirradiated specimens are ductile. The research investigations at the
General Electric Vallecitos Nuclear Center indicate that irradiation
enhances matrix strengthening and matrix-carbide precipitation which
in turn would result in the increased high temperature brittleness ob-
served in the irradiated austenitic alloys. Electron fractographs of ir-

radiated Type 304 stainless steel, Incoloy 800, and Inconel 600
appear in Figs. 18 through 21. Carbides of several different morpho-
logical forms are shown in the carbon-extraction replicas of Figs. 18
and 19. These fractographs were made at the failure region of Type 304
superheat fuel cladding [32]. Inconel 600 alloy cladding exhibited a
completely intergranular fracture when tensile tested at 700 C (Fig.
20). Incoloy 800 showed a partly intergranular, partly ductile fracture
on tensile testing at 600 C (Fig. 21) [33].
Zircaloy-2 (1.2 to 1.7 per cent tin, 0.07 to 0.20 per cent iron, 0.05 to
0.15 per cent chromium, 0.03 to 0.18 per cent nickel, <100-ppm
nitrogen, balance zirconium) and other zirconium alloys are discussed
by Bush [29] as materials of construction for water-cooled nuclear
reactors. The effect of irradiation is perhaps best reflected in the notch
FIG. 18-Fracture of irradiated fuel cladding (Type 304 steel) at 600 C superheat.
Extraction replica [32] (•
.7,
z
0
z
0
z
0
z
.1.
9
O
FIG. 19-Fracture of irradiated Type 304 steel fuel cladding at FIG. 20--lntergranularfracture at 700 C in irradiated lncone1600 ~,~
600 C superheat. Extraction replica [32] (x30,O00). fuel cladding [33] (x6000). t~n
F I G . 21 - Mixed ductile-brittle fracture at 600 C in irradiated lncoloy-8OO fuel clad-

ding [34] (•
tensile properties as a function of cold work, temperature of irradia-

tion, integrated flux, and direction. Above about 40 per cent cold work
notch strength and ductility decrease very sharply at high flux densi-
ties. The effects of irradiation on the fractographic characteristics of
Zircaloy-2, however, are somewhat confusing. Figures 22 and 23 are
included to show the comparison of ductile fracture in unirradiated
Zircaloy-2 cladding for fuel rods with the partially brittle fracture of
irradiated alloy. Figure 23 shows cleavage and intergranular separa-
tion, as well as ductile rupture [34].
Electron fractographic studies have also been made at Battelle
Memorial Institute of fission-fragment effects3 in Type 304 stainless
steel and Zircaloy-2 for short time exposures at 150 C. The effect of
neutron irradiation3 was also examined on these alloys whose stress-
3 T h e n e u t r o n e x p o s u r e w a s a b o u t 9.2 • 10 TM nvt; the fission-recoil e x p o s u r e w a s
a b o u t 2.1 • l 0 '4 f r a g m e n t s c m -2.
7"
rrl
z
o
z
.<
o
z
N
,N
N
o
z
,-r
2
0
Q
F I G . 22--Fracture in unirradiated burst specimen o f Zircaloy-2 F I G . 23--Fracture in irradiated burst specimen o f Zircaloy-2
[34] (• [34] (x6000). ",4
strain behavior was determined by bend tests. The electron fractog-

raphy did not reveal any changes in the fracture mode of either the
Type 304 stainless steel or the Zircaloy-2 alloy as a result of exposure
to neutron irradiation or fission-fragments. In all specimens fracture
occurred by a dimpled rupture mechanism [35]. Other work in which
fission-fragment damage in irradiated Type 304 stainless steel resulted
in intergranular cracking and microstructural effects at grain bound-
aries has been reported by Rosenbaum et al [36].
Summary
The examples of application of electron fractography to environ-
mental effects (chemical and radiation) on fracture presented in this
paper are only a brief sampling of past and present research areas in
which this technique has demonstrated its value. Its utility is growing
in a wide spectrum of service failure analyses. Concurrent research is
developing new applications and an increased facility and confidence
in the interpretation of fractographic features. Hopefully, electron
fractography will ultimately provide reliable differentiation and ex-
planation of many of the presently puzzling fracture phenomena metals
and alloys exhibit in their stress-strain-environmental relationships.
Acknowledgment
The author wishes to acknowledge gratefully the contributions of
several others to this paper. C. D. Beachem (Naval Research Labora-
tory) provided the electron fractographs of the titanium alloys. H. S.
Rosenbaum and U. E. Wolff (General Electric Co.) contributed all of
the figures and pertinent references to the section on Fractography in
the Study of Radiation Effects. J. L. McCall contributed the report of
the Battelle study on fission-fragment effects.
References
[t] Phillips, A. et al, "'Specific Applications of Electron Fractography," Section 6,
Electron Fractography Handbook, AFML-TR-64-416, 31 Jan. 1965.
[2] Elsea, A. R. and Warke, W. R., "Electron Microscopic Fractography," DMIC
Memorandum 161, 21 Dec. 1962.
[3] Tiner, N. A. and Gilpin, C. B., "Microprocesses in Stress Corrosion of Mar-
tensitic Steels," Corrosion, Vol. 22, No. 10, Oct. 1966, pp. 271-279.
[4] Phelps, E. H. and Mears, R. B., "The Effect of Composition and Structure of
Stainless Steels Upon Resistance to Stress-Corrosion Cracking," First Interna-
tional Congress on Metallic Corrosion, London, 10-15 April 1961, Butterworths,
1962, pp. 319-327.
[5] Matsushima, I., Deegan, D., and Uhlig, H. H., "Stress Corrosion and Hydrogen
Cracking of 17-7 Stainless Steel," Corrosion, Vol. 22, No. 1, Jan. 1966, pp. 23-27.
[6] Douglass, D. L., Thomas, G., and Roser, W. R., "Ordering, Stacking Faults, and
Stress-Corrosion Cracking in Austenitic Alloys," 2rid International Congress on
Metallic Corrosion, New York, 11-15 March 1963, National Association of
Corrosion Engineers, 1966, pp. 66-79.
[7] Nielsen, N. A., "The Role of Corrosion Products in Crack Propagation in
Austenitic Stainless Steel. Electron Microscopic Studies," Physical Metallurgy of
Stress Corrosion Fracture, Metallurgical Society Conferences, Vol. 4, Inter-
science, New York, 1959, pp. 121-154.
[8] Nielsen, N. A., "Nature of Initial Corrosion of Stressed Austenitic Stainless Steel
by Chloride Ions," 2nd International Congress on Metallic Corrosion, New York,
11-15 March 1963, National Association of Corrosion Engineers, 1966, pp. 116-
123.
[9] Mahla, E. M. and Nielsen, N. A., "An Oxide Replication Procedure for Stainless
Steels and High Nickel Alloys," Journal of Applied Physics, Vol. 19, 1948, p. 378.
[10] Picketing, H. W., Beck, F. H., and Fontana, M. G., "Wedging Action of Solid Cor-
rosion Product During Stress Corrosion of Austenitic Stainless Steels," Corro-
sion, Vol. 18, No. 6, June 1962, pp. 230t-239t.
[11] Logan, H. L., McBee, M. J., and Kahan, D. J., "Evidence for an Electrochemical-
Mechanical Stress Corrosion Fracture in a Stainless Steel," Corrosion Science,
Vol. 5, 1965, pp. 729-730.
[12] McEvily, A. J. and Bond, A. P., "On Film Rupture and Stress-Corrosion Crack-
ing," Environment-Sensitive Mechanical Behavior, Gordon and Breach, New
York, 1966, pp. 421-443.
[13] Brown, B. F., "A New Stress-Corrosion Cracking Test Procedure for High
Strength Alloys," Materials Research & Standards, Vol. 6, No. 3, March 1966.
[14] Stough, D. W., Fink, F. W., and Peoples, R. S., "The Stress Corrosion and Pyro-
phoric Behavior of Titanium and Titanium Alloys," TML Report No. 84, DMIC,
Battelle Memorial Institute, Columbus, Ohio, 15 Sept. 1967.
[15] Stress-Corrosion Cracking of Titanium, ASTM STP 397, American Society for
Testing and Materials, 1966.
[16] Turley, R. V. and Avery, C. H., "Elevated-Temperature Static and Dynamic Sea-
Salt Stress Cracking of Titanium Alloys," Stress-Corrosion Cracking of Titanium,
ASTM STP 397, American Society for Testing and Materials, 1966, pp. 1-30.
[17] Martin, G., "Investigation of Long-Term Exposure Effects Under Stress of Two
Titanium Structural Alloys," Stress-Corrosion Cracking of Titanium, ASTM STP
397, American Society for Testing and Materials, 1966, pp. 95-121.
[18] Hatch, A. J., Rosenberg, H. W., and Erbin, E. F., "Effects of Environment on
Cracking in Titanium Alloys," Stress-Corrosion Cracking of Titanium, ASTM
[19] Rideout, S. P., Louthan, M. R., Jr., and Selby, C. L., "Basic Mechanisms of Stress-
Corrosion Cracking of Titanium," Stress-Corrosion Cracking of Titanium, ASTM
[20] Meyn, D. A., Dahlberg, E. P., and Beachem, C. D., "Analysis of Stress-Corrosion
Cracking of Ti-6AI-4V Fuel Tank Material in Methyl Alcohol," NRL Memoran-
dum Report 1744, Jan. 1967, Naval Research Laboratory, Washington, D.C.
[21] Meyn, D. A., "Micromechanical Metallurgy (A study of the Crystallographic
Orientation of Cleavage Facets Produced by Stress-Corrosion Cracking of
Ti-7AI-2Nb-ITa in Water)," Report of NRL Progress, Naval Research Labora-
tory, Aug. 1965, pp. 21-23.
[22] Dahlberg, E. P., "An Annotated Bibliography of Recent Papers and Reports on
the Subject of Ambient Temperature Aqueous Stress-Corrosion Cracking of
Titanium and Titanium Alloys," AD 642128 Oct. 1966, Naval Research Labora-
tory, Washington, D.C.
[23] Jackson, J. D. and Boyd, W. K., "The Stress-Corrosion and Accelerated Crack-
Propagation Behavior of Titanium and Titanium Alloys," DMIC Technical Note,
1 Feb. 1966, DMIC Battelle Memorial Institute, Columbus, Ohio.
[24] Radiation Effects on Materials, ASTM STP 233, American Society for Testing
and Materials, 1959.
[25] Reactor Structural Materials: Engineering Properties as Affected by Nuclear Re-
actor Service, ASTM STP 314, American Society for Testing and Materials, 1962.
[26] Radiation Effects on Metals and Neutron Dosimetry, ASTM STP 341, American
Society for Testing and Materials, 1963.
[27] Flow and Fracture of Metals and Alloys in Nuclear Environments, ASTM STP
380, American Society for Testing and Materials, 1965.
[28] Trudeau, L. P., Radiation Effects on Toughness of Ferritic Steels Jor Reactor
Vessels, Rowman and Littlefield, New York, 1964.
[29] Bush, S. H., Irradiation Effects in Cladding and Structural Materials, Rowman
and Littlefield, New York, 1965.
[30] Low, J. R., Jr., Wolff, U. E., and Cowden, W. C., "Electron Microscopic Examina-
tion of Failed Stainless Steel Fuel Cladding- Dresden Advanced Fuel Assembly
PF-1 and PF-4," General Electric Company Report, APED-4242, March 1966.
[31] Comprelli, F. A. and Wolff, U. E., "Stability of High-Nickel Alloys in Superheated
Steam," AEC Research and Development Report, GEAP-4745, Atomic Energy
Commission, Nov. 1964.
[32] Hazel, V. E. et al, "Fuel Irradiations in the ESADE-VBWR Nuclear Superheat
Loop," AEC Research and Development Report, GEAP-4775, Atomic Energy
Commission, March 1965.
[33] Beaudreau, B. C. et al, "Tensile Properties of Fuel Clad Materials Irradiated in
Superheat Environments," AEC Research and Development Report, GEAP-4754,
Atomic Energy Commission, April 1965.
[34] Williamson, H. E. et al, "AEC Fuel Cycle Program Examination of Zircaloy Clad
UO2 Fuel Rods Operated in the VWBR to 10,000 MWD/TU," AEC Research and
Development Report, GEAP-4597, Atomic Energy Commission, March 1965.
[35] DeMastry, J. A. et al, "Fission-Fragment Effects in Structural Materials," Report
No. BMI-1766, 22 March 1966, Battelle Memorial Institute, Columbus, Ohio.
[36] Rosenbaum, H. S., Armijo, J. S., and Wolff, U. E., "Fission-Fragment Damage to
Type 304 Stainless Steel Fuel Cladding," AEC Research and Development Re-
port, GEAP-5002, Atomic Energy Commission, 1966.
B. V. Whiteson/ A. Phillips/ V. Kerlins, t
and R. A. R a w e t
Special Fractographic Techniques

for Failure Analysis
R E F E R E N C E : Whiteson, B. V., Phillips, A., Kerlins, V., and Rawe, R. A.,

" S p e c i a l F r a c t o g r a p h i c T e c h n i q u e s for F a i l u r e Analysis," Electron
Fractography, A S T M STP 436, American Society for Testing and Materials,
1968. pp. 151-178.
ABSTRACT: in order to assist the investigator of service failures, work was

performed using electron fractographic methods to resolve three separa,'e
problems that have not been solvable using the more conventional macro- or
light microscopic techniques. Three independent problems were examined,
and solutions were achieved. These were: (1)determination of fracture direc-
tion in thin sheet metal components, (2) differentiating between hydrogen em-
brittlement and stress corrosion in high-strength steels, and (3) determination
of applied cyclic stress as a function of fatigue striation spacing.
The results of the investigation indicated that:
1. Fracture direction in thin sheet-metal components can be determined by
replicating the acute angle shear lip. The direction of open dimples with respect
to the fracture edge indicated the fracture direction in the plane of fracture.
2. There is reasonable evidence that stress corrosion and hydrogen em-
brittlement fractures in high-strength steel can be separated by the following:
(a) Hydrogen embrittlement fractures initiated subsurface, while stress corro-
sion fractures initiated at the free surface. (b) The corrosion attack in hydrogen
embrittlement fractures, if present, was random while in stress corrosion the
oxidation products were more concentrated at the nucleus. (c) Stress corrosion
fractures had stronger indications of secondary cracking than hydrogen em-
brittlement failures. (d) The fracture features on the hydrogen embrittled speci-
mens were more clearly defined than those on the stress corrosion fractures.
3. A correlation was established between cyclic stress and fatigue striation
spacing for several aluminum alloys over a wide range of alternating and mean
stresses in thicknesses of 0.050 and 0.500 in. The correlation was empirically
derived.
KEY WORDS: electron microscope, fractography, failure analysis, fracture,

replication methods, hydrogen embrittlement, stress corrosion, fatigue (ma-
terials), crack propagation rates, fracture toughness, aluminum, steel, sustained
load testing, intergranular fracture, cadmium plating, high-strength materials
1 Chief, Metallurgy Development Section; group leader, Electron Microscope Labora-
tory; research and development engineer, Electron Microscope Laboratory; and group
leader, Development Metallurgy, respectively, Metals-Ceramics Branch, Materials
Research and Production Methods Dept., Missiles and Space Systems Div., Douglas
Aircraft Co., Santa Monica, Calif. Mr. Rawe is a personal member of ASTM,
151
Although service failure analysis has many aspects, its final objective
is structural improvement. Since it is essential to examine the fracture
origin, a technique must be developed for the determination of fracture
direction in sheet materials which exhibit none of the usual macro-
scopic features associated with fracture direction. If the mode of frac-
ture at the origin is hydrogen embrittlement or stress corrosion, dis-
tinguishing between these two modes is a difficult task since the
fracture appearances are quite similar. In case of fatigue, it would be
advantageous to be able to calculate propagation stresses by observing
the variation in striation spacings. The following three sections of this
paper present the results of work directed towards solving these
specific problems:
1. Determination of fracture direction in thin sheet-metal compo-
nents.
2. Differentiation between hydrogen embrittlement and stress cor-
rosion in high strength steel.
3. Determination of applied cyclic stress as a function of fatigue
striation spacing.
P r o b l e m 1 - D e t e r m i n a t i o n o f Fracture Direction
in Thin Sheet-Metal Components
In service failure analysis, an extremely important aspect of the
investigation usually centers around locating the origin or nucleus of
the fracture. Generally, the fracture origin holds the key to the reason
for failure and also paves the way for corrective action. If the origin is
not clearly identifiable visually, the most common method of locating it
is to trace the fracture direction on the failed components back to a
common intersection point. The ability to trace fracture direction on
the failed components generally depends on macroscopic features such
as chevron marks, localized zones of plane strain, hackle marks [I],2
etc. This task becomes difficult, if not impossible, when all of the frac-
ture is oblique shear, which is usually the case with relatively thin,
ductile materials that are used in the skins of aircraft and the fuel tanks
of missiles and launch vehicles. In this case there are no macroscopic
features that are clearly identifiable with fracture direction, much to the
frustration of the investigator.
The purpose of this work was to find out if the electron microscope
could provide a method to determine fracture direction in thin sheet-
metal components.
Experimental Procedure
The materials investigated for the determination of fracture direction
were several high-strength steels, aluminum alloys, and a nickel base
superalloy. Two thicknesses, 0.050 and 0.125 in., were employed for
WHITESON ET AL ON TECHNIQUES FOR FAILURE ANALYSIS 153
each alloy. Two types of specimens were tested in order to obtain a

tensile tear (Fig. la) and a combined shear and tear fracture (Fig. lb).
All materials failed with fractures that had 100 per cent oblique shear,
Fig. 2, with no areas of plane strain that provided any macroscopic
indication of fracture direction.
O" O"
t
C e nt e r Slot
7
2 0 in.
Stiffeners to Prevent
Specimen Buckling
L a. T e n s i l e T e a r S p e c i m e n
l
o"
O"
t@ ,/" ff
2 in. l]
[~ __irecti?,
8 dl *~
b. Combined Tear & Shear Specimen
FIG. ] - Configuration of test specimens.
FIG. 2 - The appearance o f the fracture surface o f tensile tear and shear specimens.
Results
In all specimens, fracture direction was discernible using the elec-
tron microscope, and it is believed that the technique for determining
fracture direction is applicable to all materials that fail by dimple
rupture [2]. For simplicity, none of the individual materials tested are
delineated in the illustrations, since the fractographic features analyzed
are identical, except for the relative size and abundance of the observed
dimples.
The basic concept of determining fracture direction that applies to
all materials which fail by dimple rupture is illustrated in Fig. 3. For all
FIG. 3-Relationship between orientation of open dimples along fracture edge and
fracture direction.
specimens tested, the open elongated dimples (dimples with the open
end of their parabolas toward the fracture edge) along the edge of the
rupture point back to the fracture origin (Fig. 3a). The open dimples
occur exclusively on the acute angle shear lip. The closed elongated
dimples (dimples with the closed ends of their parabolas toward the
fracture edge) along the obtuse angle shear lip and those found in the
center of the fracture were usually randomly oriented and could not be
used for fracture direction determination (Fig. 3b and c).
The acute angle shear lip was replicated by the plastic-carbon
technique [2] in such a way as to overlap the edge of the fracture. In
order to orient the replica with respect to the fractured specimen, one
or both corners of a rectangular replica were cut so as to indicate the
desired orientation. The replica was then placed in the electron micro-
scope so that the replica orientation with respect to the fracture could
be maintained in the electron fractograph. The fracture direction was
then related back to the actual specimen. The complete technique is
illustrated in Fig. 4.
Fracture Surface ~ ~,~ ii A rebii~:r YRSefer~earicee
Direction
Cut to Identify Replica

STEP 3 Reference Direction
Shadow and Deposit Carbon by Normal
Two Stage Plastic--Carbon Technique,
FIG. 4 - F r a c t u r e direction replication technique.
It was very important to obtain a good replica of the fracture edge

and the open dimples immediately adjacent to it. This was essential
because the orientation of the open dimples had to be determined
locally with respect to the edge. In obtaining an accurate indication of
the edge line, low magnification (approximately • electron micros-
copy techniques were used instead of optical microscopy because of
resolution and depth of field requirements. A series of overlapping
electron fractographs were taken along the edge region in order to
obtain an accurate sense of the dimple direction. A typical low magni-
fication electron fractograph showing open dimples along a fractured
edge is shown in Fig. 5.
FIG. 5 - Open dimples along fracture edge indicate fracture direction (x400).
Discussion
The sensitivity of dimples to fracture direction can be explained by
considering the basic difference between shear and tear dimples. In
case of a "pure" shear fracture, the dimples point in opposite directions
on mating fracture surfaces [2]. In case of a "pure" tear, the dimples
point in the same direction (back to the fracture origin) on both mating
fracture surfaces. Since from a practical standpoint no fracture occurs
by pure shear alone, that is, some tearing always accompanies a run-
ning fracture in thin sheet material, the resulting elongated dimples
observed on a fracture surface are the end product of a combined shear
and tear type of fracture. If a fracture propagates predominantly by a
tear mechanism, the elongated dimples would point more strongly in
UUUUUUUU
UUUUUUUU
Front
BIccccccl
A C CCCCCJ
Pure S h e a r Pure T e a r 9
Y ~- " O p e n "
~ Dimples I" T e a r
A Z/UV(-/U(_.,'V(_./ / Component
F r %0r
=uuuuuuuul
BI I
L,UUUUUU~U .esu,ta~J~
\Shear
ned T e a r & S h e a r ~'''cDi~ Component
Dimples
FIG. 6-Relationship between open dimple orientation and fracture origin.
the direction of fracture origin than if it propagated primarily by a

shear mechanism. This concept of resultant combined dimples is
illustrated in Fig. 6.
It is not clear, however, why open dimples are consistent in their
orientation with respect to fracture direction while closed dimples have
shown inconsistency. Assuming that a fracture separation can be non-
symmetrical, it is difficult to explain why this nonsymmetry should
only affect closed dimples since their apparent "twin" is an open dimple
which is oriented in a systematic and orderly fashion. Perhaps dis-
continuous propagation at the center of the fracture (voids occurring
ahead of the crack front) could account for the lack of orderly dimple
orientation in this region.
Conclusions
1. It is possible to determine fracture direction in thin sheet-metal
that fails in an oblique shear mode, and by dimpled rupture, by observ-
ing the orientation of open dimples along the fracture edge at the acute
angle shear lip.
2. Replicas must be made so that they overlap the fracture surface
and the sheet-metal face, coupled with low magnification microscopy,
in order to obtain an accurate assessment of the open dimple orienta-
tion with respect to the fracture edge.
3. The type of material examined is not important, as long as dimples
are visible on the fracture surface.
P r o b l e m 2 - E l e c t r o n Fractographic Technique for

Distinguishing Stress Corrosion from Hydrogen
Embrittlement in High-Strength Steels
Stress corrosion and hydrogen embrittlement have been a common

and exceptionally perplexing source of service failures in the high-
strength, medium-carbon, low-alloy steels. These fractures most com-
monly occur on components that have long life spans and exist under
some level of sustained stress even when the component is not in use.
There have been attempts to distinguish between the appearances of
the fractures caused by stress corrosion and hydrogen embrittlement
without substantial success because the two failure types exhibit a
remarkably similar fracture appearance [2]. It would be of great
importance in failure analysis to have the ability to recognize differ-
ences in the fracture appearances of the two failure mechanisms so
that appropriate corrective action can be taken.
Experimental Procedure
Sustained load tests were conducted to produce failures due to
stress corrosion or hydrogen embrittlement. The alloys selected were
forged 4340, 4330M, and D6AC. The mechanical properties are
shown in Table 1. The sustained load specimen configuration is shown
in Fig. 7.
T A B L E 1 - M e c h a n i c a l properties of steels used

for hydrogen embrittlement and stress corrosion
tests.
Strength
Material Thickness, inY Ft, Ft~
4340 ................ 0.050 277.2 240.9

4330M ............. 0.050 224.4 193.8
D 6 A C .............. 0.050 289.6 243.2
" C u t from 4Vz-in. billet in the transverse grain

direction.
To evaluate hydrogen embrittlement, three distinct cadmium plating

processes were used to provide qualitative differences in the hydrogen
levels in the test specimens:
1. Acid pickle plus cadmium fluoborate plating.
2. Bright cadmium cyanide plating.
3. Cadmium fluoborate plating.
0.1")5 in. 131A" ~ 0.25 in. F 0.25 in, DIA
I ,/- t
FIG. 7-Configuration of hydrogen embrittlement and stress corrosion specimens.
All plating conformed to the requirements of QQ-P-416, except that

specimens were not embrittlement relief baked after plating. This was
done to ensure specimen failure. The stress corrosion specimens were
tested in the bare and vacuum cadmium plated condition (MIL-L-3387).
Both hydrogen embrittlement and stress corrosion tests were con-
ducted at 50, 75, and 90 per cent of yield strength. Specimens were
axially loaded in single spring jigs. The hydrogen embrittlement tests
were conducted in plastic bags containing silica gel desiccant. The
stress corrosion tests were conducted by alternate immersion in
deionized water (10 min in water, 50 min in air).
Results
The electron fractographic examination of known stress corrosion
and hydrogen embrittlement fractures revealed similar characteristics
for the steels examined, and it appeared that no exclusive single feature
precisely identifies either fracture mechanism. However, a combina-
tion of features could be used to distinguish between the two types of
failure. The features associated with stress corrosion fracture were:
1. Predominantly surface nucleation of intergranular fracture.
2. Intergranular regions show pronounced secondary cracking or
deep crevices.
3. A relatively greater amount of oxidation or corrosion attack at
the nucleus and slow growth region than in the rapid fracture area.
4. Less pronounced hairline indications on the intergranular surfaces
of stress corrosion fractures in comparison to intergranular surfaces
of hydrogen embrittlement fractures.
The fractures associated with hydrogen embrittlement were:
1. Predominantly subsurface nucleation of intergranular fracture.
2. Both the nucleus and the rapid fracture areas exhibit equal de-
grees of (postfracture) oxidation or corrosion products.
3. Evidence of partial dimples and distinct hairline indications in the
intergranular regions.
In order to determine whether a fracture nucleus was surface or
subsurface, a two-stage plastic-carbon replication technique was used.
The suspected nucleus region was marked by placing a fine scribe line
on the free surface next to the nucleus area. The plastic replica was
allowed to overlap the edge so as to include the scribe line on the free
surface. This line was then used as a guide to locate the nucleus area.
The differences between the nucleus regions of stress corrosion and
hydrogen embrittlement fractures are shown in Fig. 8. In the case of
stress corrosion, the intergranular nucleus area followed the surface
and extended straight back from the surface. In the case of hydrogen
embrittlement, the intergranular nucleus region formed predominantly
subsurface.
FIG. 8 - The nature o f nucleus region for stress corrosion and hydrogen embrittle-
ment fractures.
FIG. 9-Typical electron fractographs showing the distinguishing features of stress

corrosion and hydrogen embrittlement (•
Other features associated with each type of fracture are shown in

Fig. 9. Stress corrosion usually showed evidence of secondary crack-
ing and corrosion products in the nucleus region. The hairline indica-
tions in the intergranular regions and partial dimples which are thought
to be associated with the fracturing process are somewhat more dis-
tinct in hydrogen embrittlement than in stress corrosion fractures.
These subtle differences are illustrated in Fig. 10.
By using a microautoradiographic technique [3], it was shown that
hydrogen was associated only with the intergranular portion of the
fracture. Due to the relatively low resolution of the technique, it was
not possible to relate hydrogen accumulation at specific structural
details such as partial dimples or hairline indications.
FIG. lO-IIlustration of subtle differences between hairline indications (arrows) of

stress corrosion and hydrogen embrittlement fractures in 4340, 260/280 steel (x6500).
Discussion
It has been shown that the accumulation of hydrogen with accom-
panying crack nucleation occurs at a region of high triaxial state of
stress [3,4]. This triaxial state of stress exists subsurface at the root
of a notch; therefore, it is not unexpected to find evidence of subsurface
nucleation in hydrogen embrittlement fractures. In stress corrosion
fractures, however, surface nucleation occurs because corrosion
initiates from the surface and progresses into the material.
Intergranular fracture is associated with both stress corrosion and
hydrogen embrittlement failures, and is generally confined to the
nucleus region. Hydrogen embrittlement intergranular fracture showed
a preponderance of partial dimples and hairline indications on the
grain boundary surfaces. This suggests that a mechanical separation

has been involved, whereas in stress corrosion fractures, disregarding
corrosion products with their accompanying surface attack, show
smoother intergranular surfaces, thereby suggesting a smaller degree
of mechanical separation (dissolution).
When attempting to distinguish stress corrosion from hydrogen em-
brittlement fractures it was important to obtain a good overall impres-
sion of the nucleus area. Often, the differences between these two
failure types were subtle and examining only a few isolated areas might
have led to incorrect analysis.
In addition, because of the subtlety between these two sustained
load failure appearances, we must make use of all the facts available.
This includes knowledge of the part processing practices, service
history, and prior service failures of the same type. The electron
microscope should be used primarily to supplement knowledge rather
than be the only source of information when analyzing this type of
failure.
Conclusions
The features associated with stress corrosion fracture are:
1. Predominantly surface nucleation of intergranular fracture.
2. lntergranular regions show pronounced secondary cracking or
deep crevices.
3. A relatively greater amount of oxidation or corrosion attack at
the nucleus and slow growth region than in the rapid fracture area.
4. Less pronounced hairline indications on the intergranular sur-
faces than observed on hydrogen embrittlement fractures.
The features associated with hydrogen embrittlement are:
1. Predominantly subsurface nucleation of intergranular fracture.
2. Both the nucleus and the rapid fracture areas exhibit relatively
equal degrees of oxidation or corrosion products.
3. Evidence of partial dimples and pronounced hairline indications
in the intergranular regions.
Problem 3 - R e l a t i o n s h i p Between Cyclic Stress and

Fatigue Striation Spacing
Fatigue cracking in metal alloys is one of the most commonly ob-
served modes of structural failure in the aerospace industry. It occurs
as a result of the repeated application of loads well below the ultimate
strength of the material. When the fracture surface of a fatigue crack
is examined at high magnification by means of electron microscopy,
periodic markings, known as striations, are observed. The spacing
between striations is characteristic of the microscopic progression of

the crack front during one loading cycle.
The major objective of this program was to develop a correlation
between the striation spacings and the cyclic load history of through-
the-thickness fatigue cracks in various sheet and plate aluminum alloys.
Such a correlation would then be useful in the analysis of fatigue
failures of structures fabricated from these alloys.
Experimental P r o c e d u r e
The aluminum alloys studied included 2024-T3, 7075-T6, 7075-T73,
7079-T6, and 6061-T6 (Table 2). For the purpose of detecting an ef-
T A B L E 2 - M e c h a n i c a l properties of aluminum alloys

used for crack propagation tests.
Strength
Material Thickness, in. Ft~, ksi F~., ksi
2024-T3 .................. 0.050 51.2 69.5

51.0 69.4
52.4 69.6
2024-T3 .................. 0.500 56.0 66.0

56.6 66.6
56.4 66,6
7075-T6 .................. 0.050 75.6 83.5

76.1 84.3
76,1 84.1
7075-T6 .................. 0.500 75.6 82.4

74.7 82.0
78.2 84.3
7075-T73 ................ 0.050 66,9 77.7

67.8 78,5
65.8 76.6
7075-T73 ................ 0.500 63.8 73,7

64.1 74.2
62.0 73.0
7079-T6 .................. 0.050 65.7 71.8

64.6 72.0
64,4 72.0
6061-T6 .................. 0.050 42.1 47.4

42.0 47.4
42.3 47.6
WHITESON Er At ON TECHNIQUES FOR FAILURE ANALYSIS 165
fect of material thickness on the properties to be studied, both 0.050-in.

sheet and 0.500-in. plate of 2024-T3, 7075-T6, and 7075-T73 were
used. The remaining alloys were tested in the 0.050 in. thickness only.
Fatigue cracks were developed in 8-in.-wide by 18-in.-long speci-
mens, centrally slotted and fatigue-precracked. Growth of the fatigue
crack was monitored visually and by means of 2-in.-long crack propa-
gation gages bonded on one specimen surface in the path of the growing
crack, Fig. 11. Each gage consisted of 20 elements, equally spaced at
0.10-in. intervals. Sequential rupture of gage elements by the propagat-
ing crack was recorded by an oscillograph simultaneously with the
output from the load cell. The oscillographic trace provided a detailed
loading and cracking history of each specimen.
The cyclic stresses which were employed were selected to represent
loading environments encountered in military aircraft and to provide
evidence of the effect on striation spacing of mean stress, alternating
stress, and frequency of loading.
FIG. 11 - L o c a t i o n o f crack propagation gages on a fatigue specimen.

The cyclic stresses for fatigue crack propagation tests are presented
in Table 3. S t a n d a r d fatigue crack propagation (Table 3a) was estab-
lished in tests with constant mean and alternating stresses. In each of
these tests one of the four scheduled mean stresses and one of the five
scheduled alternating stresses characterized the cyclic stress for that
test. The loading frequency for these tests was 1000 cpm. In several
tests, a frequency o f 10 c p m w a s e m p l o y e d .
T A B L E 3 - Fatigue cracking test conditions.

(a) STANDARD FATIGUE CRACK PROPAGATION
Thick- Fre- 8.25 13.75 16.5 22.5

ness, quency, om, ksi ~,b
Alloy in. cpm o'o,--- ksiC 6.75 2.5 7.5 11.25 13.5 6.75
2024-T3 ........ 0.050 1000 X X X X X X

10 ...... X X ......
0.500 1000 ... X X X ......
7075-T6 ........ 0.050 1000 X X X X X X
10 ...... X X ......
0.500 1000 ... X X X ......
7075-T73 ...... 0.050 1000 X X X X X X
10 ...... X X ......
0.500 1000 ... X X X ......
7079-T6 ........ 0.050 1000 X X X X X X
6061-T6 ........ 0.050 1000 X X X X X X
(b) SPECTRUM-LOAD FATIGUE CRACK PROPAGATION
Thick- Fre- 13.75 8.25 13.75 22.5

ness, quency, o',n,ksi
Alloy in. cpm o-a, • ksi 2.5 7.5 11.25 6.75
2024-T3 ...... 0.050 600 X X

7075-T6 ...... 0.050 600 X X
7075-T73 ..... 0.050 600 X X
7079-T6 ...... 0.050 600 X X
6061-T6 ...... 0.050 600 X X
6061-T6 ...... 0.050 600 X X
a All stresses based on gross area.

b o-m = mean stress.
O'max -- O'min
r o-~ = alternating stress =
2
Spectrum-load fatigue crack propagation tests (Table 3b) were also

performed in order to simulate service conditions and duplicate single
mean or alternating cyclic loads. In one series of these tests the mean
stress was held constant while the alternating stress varied every ten
cycles. In the second series, the alternating stress was held constant
and the mean stress was varied every ten cycles. These tests were per-
formed at a frequency of 600 cpm.
The standard two-stage plastic-carbon technique was used in pre-
paring replicas of the fatigue crack surfaces for examination in the
electron microscope. A corner was cut off on each replica in order to
orient it with respect to the macroscopic crack propagation direction.
Striation spacings were measured from electron fractograph negatives
taken at a standard magnification of x5300. Calibration of this magnifi-
cation was accomplished using a replica of a diffraction grating con-
taining 28,800 lines/in.
The fatigue cracks examined ranged from approximately a total
length of 0.7 to 4.3 in. The selection of areas to be replicated on
individual specimens was based on the number of cycles used to propa-
gate the left or fight crack front 0.10 in. Since it was difficult to observe
distinct fatigue striations in areas where less than about 200 cycles
were used to grow the fatigue crack 0.10 in., these areas were not
replicated. For 0.10-in. increments containing more than 200 cycles of
stress, replicas were taken at significant intervals (usually 3 to 4) along
the fracture path.
The fatigue striation spacing data were obtained by examining a
specific 0.10-in. area to gain an impression of representative striation
spacings in this area. Generally, a fatigue striation spacing for a specific
area represents an average of four readings, depending on the definition
and number of striations present. Where possible, measurements were
taken from relatively flat regions that exhibited the largest uniform
striation spacings over an appreciable distance. Striations that were
influenced by second phase particles, or occurred on steep slopes, were
not considered. It was found that distinct fatigue striations could be
observed on both the normal and oblique mode of the macroscopic
fracture profile in the fatigue region.
Results
Completely detailed data from all the tests may be found in Ref 5.
Because of its tabular bulk, only representative results will be given in
this paper.
When comparing the macroscopic fatigue crack growth rate (ob-
tained by dividing the observed 0.10-in. incremental crack growth by
the number of cycles used to propagate the crack through that incre-
ment) to the microscopic growth rate (striation spacing), the behavior
observed is illustrated in Fig. 12. At relatively low crack growth rates,
the macroscopic rate was less than the microscopic rate. This differ-
ence in growth rates may be due to crack arrest periods during fatigue
cycling. At higher crack growth rates, the macroscopic rate was greater
than the microscopic rate. This can be explained by the presence of
LEGEND: ]
0 Micro Rate 0 a = + _ 2 . 5 0 kei,J
/
/k M a c r o R a t e ~ r n = 1 3 . 7 5 ksi|/
Material: 2024--T3 J
Thickness: 0.050 in. [
Frequency: 1000 cpm /
@ 314 micro--inches/cycle
200
la
180
160
140 I
o
>,
o 120
J:
o 100 - - Macroscopic C r a c k -
c Growth Rate J
o 80
o ZX
E
,~Z
60
40
/
~ C O p l
Crack
20 Growth Ra e
0
0 V 2 3 4
Total Crack Length, inches
FIG. 12-Relationship between fatigue microscopic and macroscopic crack growth

rate,
dimple rupture regions along with the striation-producing fatigue prop-

agation. In calculating macroscopic growth rate, it is assumed that the
crack propagation over a measured increment of length was due en-
tirely to fatigue. However, in reality, regions of dimple rupture have
contributed substantially to the progression of the crack front as re-
vealed by fractographic examination.
Electron fractograph analysis of spectrum loaded specimens indi-
cates that when a fatigue striation is produced, it is the result of one cy-
cle of stress. However, it has been observed that for both varying al-
ternating and varying mean stresses, when the applied mean or alter-
nating stress range changes abruptly from a high to a low level, there is
a period (at least 10 cycles) during which the fatigue crack apparently
does not propagate, Fig. 13. On the other hand, when the low level is
not immediately preceded by high levels of stress, fatigue crack propa-
gation is observed. Since this paper is only concerned with the striation
producing portion of fatigue crack propagation, the arrest phenomena
WHITESON ET AL ON TECHNIQUES FOR FAILURE ANALYS|S 169
FIG. 13-Characteristics o f spectrum loading. Mean stress = 13.75 ksi, three alter-
nating stresses (2.5, 7.5, and 11.25 ksi) applied in 10 cycle intervals. Note lack o f stria-
tions for lowest alternating stress (2.5 ksi). Arrow indicates fracture direction (x13,800).
are relevant to calculation of number of stress cycles, not to calculation

of stress intensity.
The striation spacings were examined in terms of a parameter K [6]
which describes the local intensity of stress in the vicinity of the crack
tip, as:
K=tr [ wtan ~ (+Tro.~Vjj .................... (1)
where:
o-= gross section stress,
w= panel width,
( = total crack length, and
o-u = tensile yield strength.
A representative plot of fatigue striation spacing, dd'/dN, as a func-
tion of AK, (Kmax -- Kmin), is shown in Fig. 14. The dashed lines in the
Legend:
aa o~n
+ ksi •si
0 2.50 13.75
& 7.50 13,75
[] 11.25 13,75
9 13.50 16.50
& 6.75 8,25
9 6.75 22,50
100 Material: 7075~T73

Thickness: 0 , 0 5 0 in.
~" 5O
E
E
c
@
C
-- 10
5
10 50 100 500
Microscopic Crack Growth R a t e , djL m i c r o - i n c h e s / c y c l e

dN
FIG. 14-Relationship between striation spacing and stress intensity range param-
eter, AK.
figure encompass the range of data points and suggest a relationship of

the type:
d( (AK) 4
dN- M ......................................... (2)
This relationship is somewhat in agreement with previously pub-
lished analyses [7-9]. However, examination of the figure reveals
that the proportionality constant, M, is some complex function of the
stress environment. For example, knowledge of d(/dN alone from frac-
tographic analysis will not define AK with satisfactory precision. A fur-
ther relationship between M and the stress environment must be known
to make Eq 2 useful as a correlation between striation spacing and
stress environment.
When the range of the cyclic stresses under investigation is small,
Eq 2 will appear to define a correlation between AK and striations spac-
ing (and between AK and macroscopic growth rate) and the propor-
tionality constant, M, will have a discrete value.
When the range of cyclic stresses is expanded, however, as was true
in this program as well as in Ref I0, based on macro-rate, M is revealed
as a multivalued function of the stress.
WHITESON ET AL O N TECHNIQUES FOR FAILURE ANALYSIS 171
F o r most aircraft structures the average mean stress is known (1-g

load) and is generally constant for a given aircraft gross weight. Al-
ternating stress, on the other hand, will vary and for many extreme op-
erating conditions the magnitude of the alternating stress is not known.
For this reason, striation spacings were examined with the objective of
developing a correlation between alternating stress and the distance be-
tween fatigue striations.
Logarithmic plots of striation spacing, s, versus total crack length,
( , revealed that striation spacings increase with distance from the
origin ((/2) as:
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
where:
Sw = a constant depending on material, frequency and thickness,
m = a constant depending on the stress environment, and
W = panel width.
As shown in Fig. 15 the slope, m, of the curves varies with the cyclic
stress in such a manner that all curves intersect at a common point,
1000
,=
a~
100
c
0
u
E
&
in.
~ lO
Z
1
0.1 1 10
TotaJ Crack L e n g t h , ~ , i n c h e s
FIG. 15-Striation spacings versus crack length for 7075-T73, O.O50-in.-thick panels.
172 ELECTRONFRACTOGRAPHu
(Sw, W). From this plot, however, the exact manner in which the cyclic
stress influences the slope is not immediately obvious.
Further plots of the slope as a function of alternating stress at equal
mean stresses, Fig. 16, and as a function of mean stress at equal alter-
nating stresses, Fig. 17, indicated that the slope, m, can be described
as:
m = po'a-l12O'm -113 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4)
where p is a constant nearly equal to 10, as shown in Fig. 18. Solving

for alternating stress,
p 2
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
From Eq 3,
log s / S w (6)
m-- - -
log r ........................................
Substituting in Eq 5,
[ p (log e/w)12
o'a = ~ \log s/Sw]J ............................... (7)
Equation 7 relates alternating stress to striation spacing b,y means

of two constants, p and Sw, which can be determined by test. Values for
10
Materiat: 2024--T3
Thickness: 0.050 in.
Frequency: 1000 cpm
ksi - - - - - - ~
~-,/-~ i~,.~-
2 10
Alternating Stress, 0a, 4- k s i
FIG. 1 6 - E f f e c t o f ahernating stress on slope, m.

10
Material; 2024--T3
Thickness; 0.050 in.
Frequency: 1000 cpm
m~ (~
o -a
•
6.75
I 7.50
--~"--~
,,,"~ ~ ~ - - ~ - - - - . 5
10 50
Mean Stress, o-m, ksi
FIG. ]7-Effect o f m e a n s t r e s s o n slope, m.
Material: All Alloys Tested /

Thickness: 0.050 in, 0
Frequency: 1000 cpm 8
m P
oaY= ~mYs
i
:/
E /
~---- p = 10
Q.
_o
0)
1
I
= Range for all
tests except
four points
shown. [
0
0 0.1 0,2 0.3
1
(#a)Y2 (am)V3
FIG. 18-Determination o f the value o f the constant, p, in Eq 3.
the aluminum alloys tested in this program are given in Table 4. The
value of the constant, p, varies only slightly from alloy to alloy and does
not appear to be affected by cyclic frequency. The constant, Sw, varies
more from alloy to alloy and shows a definite trend to increase with de-
creasing frequency and increasing thickness. An increase in Sw reflects
larger striation spacing (higher crack growth rate).
The ability of Eq 7 to describe the standard fatigue test data is dem-
onstrated in Fig. 19. In this figure, the alternating stress predicted by
TABLE 4 - C o n s t a n t s for use in correlating striation
spacings with cyclic stresses (Eq 7).
Thickness, Frequency,
M~efi~ in. cpm p~ $2
2024-T3 ......... 0.050 1000 10.35 5 5 0 • 10 -6

10 10.35 8 0 0 • 10 -8
0.500 1000 10.35 1 5 0 0 • 10 -~
7075-T6 ......... 0.050 1000 9.90 8 0 0 • 10 -6
10 9.90 1 0 0 0 • 10 -6
0.500 1000 . . . . . .
7075-T73 ........ 0.050 1000 9.73 8 0 0 • 10 -6
10 9.73 1 0 0 0 • 10 -6
0,500 1000 9.73 2 0 0 0 • 10 -6
7079-T6 ......... 0.050 1000 10.05 1 0 0 0 • 10 -6
6061-T6 ......... 0.050 1000 1020 8 0 0 • 10 -6
"Constant for E q 7 w h e n o-, a n d o-m are in units o f ksi.

Constant for E q 7 w h e n ~ , W, a n d s are in units of inches.
20
Legend: t
0 2024--T3
0 7075--T6
7075--T73
15 9 7079--T6
.r
9 5061--T6 J
+I
f-
D.
0 5 10 15 20
Actual Oa, _+. ksi
F I G . 1 9 - Comparison of actual and predicted a l t e r n a t i n g s t r e s s for O.050-in.-thick

panels fatigue cracked at 1000 cpm.
Eq 7 is plotted versus the actual alternating stress. Each data point in

the figure represents the average predicted alternating stress computed
for each striation spacing measurement made in a given test. For most
tests, the predicted value of alternating stress is within 1 or 2 ksi of the
actual stress.
Predicted and actual alternating stresses for spectrum load tests are
compared in Table 5 using the constants developed for a frequency of
1000 cpm.
T A B L E 5 - Comparison of actual and predicted alternating
stress from spectrum load tests.
Actual Average Predicted

Alternating Alternating
Mean Stress, Stress, Stress,
O'm, O'a, (ra, a
Material ksi • ksi • ksi
2.5 ...
2024-T3 .......... 13.75 7.5 7.86
11,25 16.30
8.25] ...
13.75~ 6.75 ...
22.5 J 6.46
2.5 ...
7075-T6 ............ 13.75 7.5 6.42
11.25 12.33
8.25] ...
13,75~ 6.75 6.36
22.5 J 6.88
2.5 ...
7075-T73 ........ 13.75 7.5 7.52
11.25 10.18
8.25] ...
13.75~ 6.75 5.31
22.5 J 8.24
2.5 ,..
7079-T6 .......... 13.75 7.5 7.60
11.25 11.78
8.25] ...
13.75~ 6.75 5.49
22.5 J 6.78
2.5 ...
606 I-T6 .......... 13.75 7.5 6.24
11.25 l 1.82
8.25] ...
13.75~ 6.75 8.24
22.5 J 7.72
a Calculated using Eq 7 and constants from Table 4.

Equation 7 can be employed in estimating the alternating stress for a

given fatigue crack when the mean stress is known. To use the equation
it is necessary to measure striation spacing, s, at a particular distance,
4[2, from the fatigue crack origin. A number of measurements at vari-
ous distances will improve confidence in the results.
As an example, assume that a fatigue fracture from a 10-in.-wide
7075-T73 structure operating at a mean stress of 18 ksi is to be an-
alyzed to determine the maximum cyclic stress experienced during op-
eration. Five striation spacing measurements are made at 4/2 distances
from the origin.
These d/2 distances are 0.5, 0.75, 1.0, 1.25, and 1.5 in. Assume that
the resulting striation spacings are 30, 56, 85, 115, and 150 tzin.,
respectively.
For the total crack length of 1.5 in. (4/2 = 0.75) and the striation
spacing of 56/An., the alternating stress is calculated as follows:
[ P (log ~/W'~q 2
-- slS JJ
= [973 (, ' " I( log 10
= 7.01 ksi
The results of calculations for the other crack lengths and striation
spacings are shown in Table 6. The average calculated alternating
stress is 7 ksi. Therefore, the maximum stress is:
O'max ~" O-m AV O-a
=18ksi+7ksi
= 25 ksi
T A B L E 6 - Calculated alternating stress for idealized example
given in text.
Distance Total Striation Alternating

from Origin, Crack Length, Spacing, Stress,
in. in. /zin. o-~, • ksi
0.50 ................ 1.00 30 6.78

0.75 ................ 1.50 56 7.01
1.00 ................ 2.00 85 7.10
1.25 ................ 2.50 115 7.04
1.50 ................ 3.00 150 7.13
Avg 7.01
WHITESON ET AL ON TECHNIQUES FOR FAILURE ANALYSIS ]77
Conclusions
1. The stress intensity range parameter AK does not satisfactorily
describe the relationship between fatigue striation spacing and cyclic
stress.
2. If one stress condition is known, such as mean stress, the empiri-
cal relationship
[ p {log
or, =___ ~ \log s/S~]J
permits a reasonably accurate calculation of the magnitude of the other
stress, that is, alternating stress, in through-crack, thin sheet.
3. The reduction of fatigue frequency from 1000 to 10 cpm results
in an increase in the striation spacing at equivalent stress conditions.
4. An increase in thickness from 0.050 to 0.500 in. results in an in-
crease in the striation spacing at equivalent cyclic stress and test fre-
quency.
5. An abrupt decrease of applied fatigue stress can result in a period
of arrested crack growth.
6. A fatigue striation is the result of one cycle of stress.
7. The appreciable difference between microscopic and macro-
scopic growth rates can be explained by the presence of dimple rupture
areas on a fatigue surface.
8. Fatigue striations can be observed on both the normal and the
oblique mode of fracture.
Acknowledgment
The authors wish to express their appreciation to the Air Force Ma-
terials Laboratory (MAAS), Research and Technology Division, Air
Force Systems Command, Wright-Patterson Air Force Base, Ohio, for
permission to publish this work. The effort was funded under Contract
AF33(615)-3014, with R. L. Henderson as the Air Force project
engineer.
References
[1] Handbook for Aircraft Accident Investigators, U.S. Naval Aviation Safety Center,
Office of the Chief of Naval Operations, 1957.
AFML-TR-64-416, Air Force Materials Laboratory, Wright-Patterson Air Force
Base, Ohio, Jan. 1965.
[3] Gilpin, C. B. et al, "Electron Microautogradiography and Its Application to the
Study of Hydrogen Distribution in Steel," Advances in Electron Metallography,
ASTM STP 396, American Society for Testing and Materials, 1966, pp. 7-20.
[4] Troiano, A. R., "The Role of Hydrogen and Other Interstitials on the Mechanical
Behavior of Metals," Transactions, American Society for Metals, Vol. 52, 1960,
p. 54.
[5] Phillips, A. et al, "Electron Fractography Handbook-Specific Applications of

Electron Fractography," AFML-TR-64-416, Supplement I, Air Force Materials
Laboratory, Wright-Patterson Air Force Base, Ohio, Dec. 1966.
[6] "Fracture Testing of High Strength Sheet Materials," ASTM Bulletin, Jan. and
Feb. 1960.
[7] Paris, P. and Erdogan, F., "A Critical Analysis of Crack Propagation Laws,"
Journal of Basic Engineering. American Society of Mechanical Engineers, Dec.
1963.
[8] Hertzberg, R. W., "Application of Electron Fractography and Fracture Me-
chanics to Fatigue Crack Propagation in High Strength AluminumAlloys," Lehigh
University, Bethlehem, Pa., 1965.
[9] Krafft, J. M., "On Prediction of Fatigue Crack Propagation Rate from Fracture
Toughness and Plastic Flow Properties," Transactions, American Society for
Metals, Vol. 58, 1965.
[10] Eitman, D. A. and Rawe, R. A., "Plane Stress Cyclic Flaw Growth of 2219-T87
Aluminum and 5A1-2.5Sn Titanium Alloys," NASA CR-54956, National Aero-
nautics and Space Administration, June 1966.
A. J. Brothers I and S. Y u k a w a a
Engineering Applications of Fractography
REFERENCE: Brothers, A. J. and Yukawa, S., "Engineering Applica-

tions of Fractography," Electron Fractography, A S TM S TP 436, American
ABSTRACT: To date the primary use of microfractography has been qualita-

tive. For this purpose the characteristic features of a specific fracture mecha-
nism are compared with those of the service failure to be identified. However,
the optimum use of fractography is that of both a quantitative and qualitative
tool. This paper illustrates its usefulness in interpreting quantitatively the
microfractographic features of various failure mechanisms, and suggests some
currently speculative but promising areas for future work.
The incidence of fatigue striations and their association with cyclic alternat-
ing loads provide the basis for quantitative interpretation of fatigue failures. It is
shown that fractographic determination of microscopic crack growth rates,
coupled with currently available analyses for macroscopic crack growth rates,
permits one to estimate relative failure stress or strain level, approximate num-
ber of failure cycles, and, in some cases, the relative stress gradient and mode of
loading. The general stress dependence of fatigue striations, if determined from
component tests, can be used directly for estimating relative operating stress
levels of service failed parts. However, the influence on crack growth rate of
geometrical factors suggests caution in comparing service failures with fracto-
graphic features of laboratory specimens. For this general case the correlations
of crack growth rate with Gritfith-lrwin fracture mechanics concepts are recom-
mended. Since these analyses consider both component and crack geometry
and mode of stressing, they are more generally applicable to a service con-
sideration.
The high magnification used in electron microfractography and its use of
replica techniques are shown to be useful in interpreting quantitatively the
metallurgical features associated with creep microvoid formation. It is shown
that fractography, in combination with standard electron diffraction techniques,
permits one to isolate and identify the distribution, size, and content of specific
metallurgical phases.
KEY WORDS: electron microscopy, fractography, fatigue (materials), cleav-

age fracture, creep rupture
Since its early use by Zapffe [1]2 for the examination of cleavage
fracture surfaces, fractography, or the study of fracture surfaces, has
been developed extensively both in its techniques and applications. Its
1 Materials and Processes Laboratory, General Electric Co., Schenectady, N.Y.
179
general uses have been described previously by Plateau et al [2],

Beachem and Pelloux [3], and others. To date, fractography has found
its greatest application as a qualitative tool with which to deduce the
features of the various fracture mechanisms, or to study in detail the
specific features of a single fracture mechanism. Illustration of its use
for this latter purpose has been demonstrated by Low [4] for cleavage
fracture, by Beachem [5] for'ductile fracture, and Forsyth [6] for
fatigue. However, to date there have been few attempts to illustrate
broadly the direct engineering applications of fractography.
It is the purpose of this paper to review some direct engineering
applications of fractography. The general thesis presented is that the
optimum practical usefulness of these techniques requires the com-
bined use and relation of quantitative fractrographic features to other
currently available engineering analyses and correlations. The specific
details of fractographic techniques are now generally well known and
have been described previously [7-9].
Although the primary emphasis is on the quantitative use of fractog-
raphy for failure analysis, the general principles illustrated are widely
applicable to all general engineering tests.
Elements o f Failure Analysis

The study of service fractures, and their comparison with com-
ponent tests, provided valuable information regarding weaknesses in
engineering design, assessment of operating conditions, and determina-
tion of the suitability of specific materials for given applications. The
information required from service failures often includes that de-
scribed below:
1. The fracture initiation mechanism.
2. The mode of loading, that is, bending, tension, torsion, thermal
stress, etc.
3. The magnitude of loads causing failure.
4. If a fatigue failure- how many cycles of alternating stresses were
involved, that is, is it high- or low-cycle fatigue?
5. The influence of environmental effects on fracture.
In some cases it is possible to assess part of the above information
solely from macroscopic examination. For example, the presence of
gross chevron markings indicate brittle cleavage, while beach marks
suggest fatigue fractures. The extent of deformation near the initiation
site and presence of shear lips on a fracture surface indicates some
measure of fracture loads and the relative temperature of fracture,
respectively. However, the fractographic features of the various modes
provide measurable quantities with which to specifically characterize a
fracture. It is, of course, to be recognized that there are limitations on
the uses of fractography, particularly for the examination of service
BROTHERS AND YUKAWA ON ENGINEERING APPLICATIONS 181
failures. In those cases where local mechanical damage, corrosion, or

oxide coatings obscure the fracture surface, useful fractographic analy-
sis is impossible. Furthermore, the successful use of fractography is
extremely dependent on the judgment of the investigator. Primarily, it
is necessary to distinguish between areas of initiation versus propaga-
tion, and to recognize the significance of various macroscopic features
of fracture surfaces.
In many cases, as will be shown later, the use of comparison speci-
mens tested under suspected operating conditions permits one to make
a direct comparison of fractographic features.
Quantitative Features of Static Fracture Modes

The various modes of static fracture are generally characterized by
the manner in which fracture occurs. These include transcrystalline
ductile fracture or, so-called dimpled rupture, transcrystalline cleav-
age fracture and intergranular fracture. The mechanism of transcrystal-
line ductile fracture has been examined extensively by Gurland and
Plateau [10], Rogers [11], and Beachem [3] and found to occur by
microvoid coalescence. Although infrequently observed as a primary
service failure mode, the incidence of microvoid coalescence or
"dimpled rupture" in low and intermediate strength steels is generally
indicative of a ductile or tough material which fractured at relatively
high stress levels. Although the engineering implications of the
fractographic features of dimpled rupture fracture have not been fully
explored, there are some applications. It has been observed, for ex-
ample, that the shape of the dimples is related to the general stress sys-
tem by which they are formed [3]. Under uniaxial tensile rupture,
microvoid growth occurs symmetrically resulting in equiaxed dimples;
under shearing or tearing action, parabolic dimples are formed, the
directionality of which provides some measure of the local direction of
crack propagation. In some cases the presence of extremely elongated
parabolas provides clues to the presence of shear loading. These
various features are illustrated in Fig. 1.
Qualitatively, the appearance of dimpled rupture in low and inter-
mediate strength steels is indicative of fracture at temperatures above
the ductile-to-brittle or cleavage-to-shear transition temperature. How-
ever, in high strength steels dimpled rupture may not be indicative of
high toughness. Although to date there have been no specific correla-
tions between dimple size and test temperature, the work of Edwards
[12], Fig. 2, illustrates the dependence of dimple depth on test tempera-
ture for a high strength steel. Since it is likely that the absolute dimple
depth is a function of material strength level, ductility, etc., this
empirical correlation will probably have to be established for each ma-
terial and condition. However, it does suggest that correlations of
.9
I
BROTHERS A N D YUKAWA O N ENGINEERING APPLICATIONS 183
--0-- Avg. Depth of Rupture DtmpJes

--X- Avg, Heiqht of Ridges
Between Q u a s i - C l e a v a g e Facets
.~..0
-0,3 .s J
I,
,/
/
/
7l /
v /
O
/
/
/
/
/
-0,2
= / x
I /
e-, g /
x /
g /
o /
/
/
/
X
-.0.1 /
J
J
X~ "
5 % Cr Die Steel
I l I i I I I i [
-200 -I00 0 I00 200 300 400 500 600 700
Test Temperature ~~
F I G . 2 - Plot of average depth of rupture dimples and average height of ridges be-
tween quasi-cleavage facets versus test temperatures-Edwards [12].
service fracture dimple depth measurement with that of controlled

comparison test specimens could provide a means of assessing frac-
ture temperature, and the focal ductility associated with fracture.
The few studies relating dimple diameter to mechanical properties
suggest that dimple size merely reflects inclusion or second-phase
particle spacing and is not directly related to mechanical properties
[3,10]. Birkle et al [13], using fractographic techniques, have shown
that the level of fracture toughness is determined primarily by the
density and distribution of the sulfide inclusions associated with sulfur
content level.
The application of fractography to the examination of cleavage frac-
ture has been performed by Low [4] and others. An example of the
engineering application of fractography to cleavage fracture is the de-
termination of the relationship of cleavage facet size or cleavage crack
length to the temperature of the ductile-to-brittle transition; in this
250
e
~ 200
150
o.
o
O
50 I I I I I I I I I I
0 0.01 0,02 0.03 0.04 0,05 0,06 0.07 0.08 0,09 0.10
CLEAVAGE CRACK LENGTH [MM)
FIG. 3 - R e l a t i o n o f transition temperature to cleavage dimensions Ni-Mo-V steels.
case, the 50 per cent fibrous fracture appearance transition tempera-

ture measured with Charpy impact specimens. This is illustrated in
Fig. 3 which shows the relationship of fracture appearance transition
temperature to cleavage crack length for Ni-Mo-V low-alloy steels of
the strength range shown. The cleavage microcrack dimensions were
determined microscopically from profile cross sections and fracto-
graphically by optical examination at magnifications up to • The
values shown represent the average of a minimum of 25 measurements
on fractured Charpy specimens tested at temperatures corresponding
to a high percentage of cleavage fracture. Since this variation in cleav-
age crack length reflects the toughness of the alloy, the measurement of
facet size, particularly in regions of complex metallurgical structures
(for example, weld regions), can provide a measure of the local tough-
ness in material through which fracture occurred.
The qualitative observations of Lubahn [14] and Crussard [15] of
microscopic regions of fibrous fracture and the appearance of curved
or deformed cleavage planes in cleavage fracture suggest further
quantitative investigation to determine the relative change in micro-
scopic quantities of fibrous fracture with temperature through the
brittle-to-ductile range.
Quantitative Features of Fatigue Fractures

The quantitative fractographic features of fatigue, or failure under
alternating stress, have been examined extensively. It has been ob-
served that there are characteristic striations associated with fatigue
failure; these general features are shown in Fig. 4, a fractograph of an
austenitic alloy component. Since each of these striations represents
the fatigue crack front on a local scale, it is possible to derive informa-
tion of direct engineering value.
By precise orientation of the fractographic shadowing direction rela-
tive to macroscopic features of the component, it is possible to esti-
mate the direction of crack growth. If it is assumed that each striation
results from a single cycle of loading, it is possible to estimate the rela-
tive number of cycles involved in crack propagation. From the spacing
of the striations one can estimate the local fatigue crack growth rate.
The relative changes in striation spacing with location, that is, their
gradient, permits one to determine whether the fatigue crack growth
rate is increasing or decreasing with increasing crack length.
Intuitively, one might expect that the fatigue crack growth rate or
striation spacing is a function of the nominal alternating stress resulting
in fatigue failure. This, in fact, has been observed as shown in Fig. 5.
FI G. 4 - Fractographic features of austenitic alloy fatigue fracture showing striated

features typical of fatigue failures (x5000).
oo
~u
I
v~
~5
This figure shows the relative striation spacing determined at the site of
initiation in two identical vanes tested in zero mean stress, completely
reversed bending at two different stress levels. At the higher alternating
stresses the crack growth rate is noticeably greater with a correspond-
ingly coarser striation spacing. This general dependence of spacing on
alternating stress suggested to Forsyth [16] that fractography could be
used to determine the alternating stress associated with component
failure. For this purpose and where possible, simulated operating com-
ponent tests are made at various nominal stresses and the dependence
of striation spacing on stress level is determined. Examples of these
relationships for various materials tested in reversed bending are
shown in Fig. 6. In these cases, the mean stress is zero. The influence
of nonzero mean stresses is discussed further below.
The correlation of striation spacing with alternating stress is suitable
for direct comparison of service failures with prototype or actual com-
ponent tests. However, caution should be exercised in attempting
direct correlation of the results of laboratory fatigue test specimens
with more complex components of different geometry. For these condi-
tions it is suggested that crack growth rates be correlated with the
stress intensity factor as proposed by Paris and Erdogan [17] and
described in detail in other publications [18,19]. With this procedure
M-252 ('~ 1250"F

-I
"
3.0
%
x
w
,.J
"~ 2.0
/
G
z
u T.R COPPEI~
8END S P E C I M ~
/ r PE I
~ 1.0
0 I I ] 1 I I I I I I I
I0 20 50 40 50 60 70 80 90 I00 I10
NOMINAL ALTERNATING STRESS, Ksl
FIG. 6-Relationship between fatigue striation spacing and alternating stress
amplitude.
the effects of component or specimen geometry and crack length on

crack growth rate can be correlated with the elastically derived stress
intensity factor, K. This term combines in a single parameter both the
stress and the effect of geometry, that is, crack length. Conceptually,
if one determines empirically the correlation between crack growth
rate and stress intensity factor for a given specimen configuration for
which K is known, it is then possible to estimate crack growth rates for
other configurations for which an analysis for stress intensity factor is
available.
Examples of these correlations for both macroscopic crack growth
rates and striation spacing measurements are shown in Fig. 7. Of
particular interest is the correlation of fractographically determined
crack growth rate as a function of stress intensity factor for part-
through surface cracks in Type 304 steel loaded in pulsating tension.
For comparison, the correlation of macroscopic crack growth with K
derived from results on 14-in.-wide through-cracked plates is also
shown. This reasonably good correlation between macroscopic and
microscopic crack growth rates including the effects of specimen
geometry and crack type suggest the usefulness of this concept in
quantitatively interpreting fatigue service failures. For example, it is
possible to predict the gradient in crack growth rate or striation spac-
ing for various modes of loading. This feature is illustrated in Fig. 8 for
an idealized simple plate geometry containing a shallow surface crack;
the stress intensity factors for these geometries were derived from Ref
20. The corresponding anticipated gradients in crack growth rate re-
sulting from these changes in stress intensity factor are also described
here. Although there is little significant difference at very short crack
lengths, there is a predicted high variation in relative crack growth
rates with increasing crack length. If one considers that edge-loading
approximates the condition of thermal loading, it is apparent that there
should be a significant difference in crack growth rates under various
modes of loading.
Although there have been to the authors' knowledge no systematic
investigations to determine the practical applicability of the above
described approach, the few isolated examples of Fig. 9 for Type 304
steel suggest that there are, in fact, apparent effects of mode of loading
on crack growth rate gradients which correlate generally with the
above analysis. In particular, the weak dependence of crack growth
rate on crack length for thermal cycling is in agreement with the effect
predicted from consideration of the stress intensity factor. Clearly, the
specific use of fractography for this application requires additional ex-
perimental verification.
The use of a macroscopic crack growth correlation to interpret the
significance of various magnitudes and gradients of striation density re-
I000
THRU-
CRACK THRU- CRACK PLATE
O-----O/ PART-THRU SURFACE CRACK ZERO TENSION CYCLING
GROWTH RATES DETERMINED Crg
9 9 a FRACTOGRAPHICALLY
CARBON Cr-Mo-V
S T E E L \ \ ALLOY
TYPE ~ ~TEEL (REF.18) K=~(wr (,oo~-REZ2o 9
-r
W
z
I Z
:1.
d r
~,oc
).
0
c.D
9i I el
K- D +oaz(i- ~o~r
I Z
g
~-(~ 0
REF ZO z
PART-THRU Z
0
SURFACE CRACK
ZERO- TENSION CYCLING
tO I I I I I I i I J i I I I I i I iJ
I0,000 I00,000 3 Z
RANGE OF STRESS INTENSITY FACTOR, K( I b - i n - 2 )
FIG. 7-Correlation o f m a c r o s c o p i c a n d m i c r o s c o p i c c r a c k g r o w t h with s t r e s s i n t e n s i t y f a c t o r .

"0
. • 4.0
3.0
z.o
t,
I--
.IX
f p--%OE LO,D,.
n. 1.0 C '1e I (tHee"aLl
3oo
<
200 BENDING
z
<
'~ I00
o
Ld TORSION
I-
EDGE
w / t ~" LOADINB
0.1 0.2 0,3 0.4

RELATIVE DEPTH OF CRACK 0 / , .
"w
FIG. 8-111ustration of analytical procedures for estimating type of loading in

fatigue service failures (schematic).
quires that all variables influencing crack growth rate be known. Al-
though linear-elastic fracture mechanics rationalizes on a semi-
empirical basis, the effect of crack length and specimen or part geom-
etry, there is to date no fully satisfactory model or analysis with which
to estimate the relative contribution to crack growth rate of mean
versus alternating stress. In those cases where service operation in-
cludes known levels of mean stress, it may be possible from previous
empirical correlations to relate the magnitude of the alternating stress
level to the striation frequency. An example of the use of fractography
for this purpose is described in Ref 9. Illustration of the effect on crack
growth rate of various ratios of mean to alternating to mean stress is
shown in Fig. 10. Although the magnitudes of the crack growth rates
I10
/
IOO A
90
o
I--
70 PART- THROUGH
SURFACE CRACK IN
I.- TENSION PLATE
ne 6O
1"
O 50
v
U
<
a~
u
I-
J
ul
X ~ x
I I I I
0 0.1 0.2 0.3 0.4
DISTANCE FROM CRACK TiP INCHES
FIG. 9 - Comparison of loading procedures on striation frequency gradient in Type

304 steel.
500 [ I II
7075-T73
U.I 0.050 IN.
I--
,1[
~. U.I
I-- --J
~ - ; 0.1| 0.3(
I
o~ u , /,L, ~
"" I00 9 nT!
I;. ~',] I
^ ~" ^ ; 1 .
F ~ i.~'
, Cf, j
~ ,=1"1=
0.55
'/},
/ /fJ
j If7
I~L I0 50 I00
STRESS INTENSITY RANGE, AK, KSI ( IN.} I/;'
FIG. 10-Relationship between striation spacing and range o f stress intensity factor
for various stress ratios (Ref 9).
191
are influenced by mean stress, the nearly similar slopes indicate that
the general power dependence of crack growth rate on the range of the
stress intensity factor still applies. This conforms generally to the
analyses proposed by Krafft [21] and recently by Forman et al [22].
Application of Fractography to Creep Rupture Fractures

Since creep rupture tests and service failures of most materials at ele-
vated temperatures result in oxidation of fracture surfaces, it is
generally not possible to examine these surfaces directly. However, in
some cases the creep-induced intergranular microvoids in regions near
the fracture surface can be exposed by fracture at low temperatures
prior to fractographic examination. This technique was first employed
by Hyam [23] to study a 1 per cent aluminum-magnesium alloy. It has
also been used on an alloy steel having low creep rupture ductility; the
specific fractographic features noted in this alloy are shown in Figs.
11-13. Figure 11 illustrates the general appearance of intergranular
microvoids in a polished cross section of a creep rupture specimen of
Cr-Mo-V alloy steel tested at 1050 F for about 8000 h. After fracturing
in liquid nitrogen to expose the microvoid surfaces, these surfaces were
replicated using standard fractographic procedures.
The first of these replicas, a photograph of which is shown in Fig. 12,
reveals the metallurgical phases which have been removed in situ from
the intergranular microcrack surface during the replication pro-
FIG. 11 -lntergranular microcracks formed during creep of a CrMoV low-alloy steel

at 1050 F stressed for 7700 h at 25,000 psi (X/0).
F I G . 12-Extraction replica of exposed inter.r microcracks showing metal
lurgical phases associated with microvoids. Large plate-like particles are M23C~; fila-
ment groups are VC (•
F I G . 13-Fractographic features of exposed microcrack surface showing distribu-

tion of particle sites (• 193
cedure. By use of electron diffraction techniques, it was possible to

isolate these individual particles and to identify their composition.
Subsequent replications removed these particles and revealed the fea-
tures shown in Fig. 13. From the specific measurement of the number
and distribution of particles present on these surfaces, it is possible to
determine quantitatively those metallurgical features contributing
directly to creep fracture on a microscopic scale.
Similar techniques for metallurgical examination can and have been
applied to those materials which can be fractured intergranularly to
exposed intergranularly disposed metallurgical phases.
Summary
Although still in its incipient development stage, the use of fracto-
graphically derived features for quantitative engineering applications
appears promising. However, the optimum practical usefulness of
these techniques will require the combined use and relation of quanti-
tative fractographic features to currently available engineering
analyses and correlations.
References
[1] Zapffe, C. A. and Clogg, M., Jr., "'Fractography-A New Tool for Metallurgical
Research," Transactions, American Society for Metals, Vol. 34, 1945, p. 71.
[2] Plateau, J., Henry, G., and Crussard, C., "Several New Applications of Micro-
fractography," Revue de Metallurgie, Vol. 55, No. 3, 1957.
Study of Micromechanisms of Fracturing Processes," Fracture Toughness Testing
and Its Applications, ASTM STP 381, American Society for Testing and Ma-
terials, 1964, p. 210.
[4] Low, J. R., Jr., "A Review of the Microstructural Aspects of Cleavage Fracture,"
International Conference on Fracture, Swampscott, Mass., t959.
[5] Beachem, C. D., "An Electron Fractographic Study of the Influences of Plastic
Strain Conditions upon Ductile Rupture Processes in Metals," Transactions,
American Society for Metals, Vol. 56, 19~3, p. 318.
[6] Forsyth, P. J. E., "Fatigue Damage and Crack Growth in Aluminum Alloys,"
Acta Metallurgia, Vol. 11, 1963, p. 703.
[7] Turkalo, A. M., "Electron Microscopy Techniques Applied tO the Problem of
Brittle Fracture," presented at ASM Symposium on Techniques to Study Frac-
ture and Fatigue, American Society for Metals' Fall Meeting, 1964 (also available
as General Electric Research Laboratory Report No. 64-RL-3855M).
[8] Dahlberg, E. P. and Beachem, C. D., "Fractography, Part XV, Some Artifacts
Possible with the Two-Stage Plastic Carbon Replication Technique," Memo Re-
port 1457, National Research Laboratory, 1963.
[9] Phillips, A. et al, "Electron Fractography Handbook," Technical Report ML-
TDR-64-416, Air Force Materials Laboratory, 1965.
[10] Gurland, J. and Plateau, J., "The Mechanism of Ductile Rupture of Metals Con-
taining Inclusions," Transactions, American Society for Metals, Vol. 56, 1963, p.
442.
[11] Rogers, H. C., "Tensile Fracture of Ductile Metals," Transactions, American
Institute of Mining, Metallurgical, and Petroleum Engineers, Vol. 218, 1960, p. 498.
[12] Edwards, A. J., "Depth Measurements on Fracture Surfaces," Report of NRL
Progress, National Research Laboratory, 1963.
[13] Birkle, A. J., Wei, R. P., and Pellissier, G. E., "Analysis of Plane Strain Fracture
in a Series of 0.45 C-Ni-Cr-Mo Steels with Different Sulfur Contents," Transac-
tions, American Society for Metals, Vol. 59, 1966, p. 981.
[14] Lubahn, J. D., "Room Temperature Crack Propagation and Size Effect in Mild
Steel," Welding Journal Research Supplement, Vol. 34, 1955, p. 518s.
[15] Crussard, C. et al, "A Study of Impact Tests and the Mechanism of Brittle Frac-
ture," Journal of the Iron and Steel Institute, Vol. 183, 1956, p. 146.
[I6] Forsyth, P. J. E. and Ryder, D. A., "Some Results Derived from the Microscopic
Examination of Crack Surfaces," Aircraft Engineering, 1960, p. 96.
[17] Paris, P. C. and Erdogan, F., "A Critical Analysis of Crack Propagation Laws,"
Transactions, American Society of Mechanical Engineers, Vol. 85, 1963.
[18] Brothers, A. J. and Yukawa, S., "Fatigue Crack Propagation in Low-Alloy Heat-
Treated Steels," Transactions, American Society of Mechanical Engineers, Vol.
89, 1966.
[19] Tiffany, C. F. and Masters, J. N., "Applied Fracture Mechanics," Fracture
Toughness Testing and Its Applications, ASTM STP 381, American Society for
Testing and Materials, 1964, p. 249.
[20] Paris, P. C. and Sih, G. C., "Stress Analysis of Cracks," Fracture Toughness
Testing and Its Applications, ASTM STP 381, American Society for Testing and
Materials, 1964, p. 30.
[21] Krafft, J. M., "On Prediction of Fatigue Crack Propagation Rate from Fracture
Toughness and Plastic Flow Properties," Transactions, American Society for
Metals, Vol. 58, No. 4, 1965, p. 691.
[22] Forman, R. G., Kearney, V. E., and Engle, R. M., "Numerical Analysis of Crack
Propagation in Cyclic-Loaded Structures," Paper No. 66-WA/Met-4, American
Society of Mechanical Engineers, 1966.
[23] Hyam, E. D., "Discussion in Structural Processes in Creep," Iron and Steel
Institute, Special Report No. 70, 1961, p. 76.
M. S. Hunter' and J. C. M c M i l l a n 2
Fractography and Microstructure of

Aluminum Alloys 7075-T651 and
7075-T7351
R E F E R E N C E : Hunter, M. S. and McMillan, J. C., " F r a e t o g r a p h y a n d

M i e r o s t r u e t u r e of A l u m i n u m Alloys 7 0 7 5 - T 6 5 1 a n d 7 0 7 5 - T 7 3 5 1 , "
Electron Fraetography, ASTM STP 436, American Society for Testing and
Materials, 1968, pp. 196-211.
ABSTRACT: Specimens from 13/s-in.-thick plate of 7075-T651 and 7075-

T7351 alloys were examined by electron transmission, light microscopy, and
electron fractography to relate microstructure, fracture topology, and mechani-
cal and fracture properties. Material in the T7351 temper had a lower yield
strength, smaller and more numerous fracture dimples, and an entirely trans-
granular fracture path. The T651 temper had a higher yield strength, fewer and
larger dimples, and a partly intergranular fracture path. It was concluded that
the beneficial effect of the lower yield strength more than offset the disad-
vantage of more numerous fracture dimple sites, and led to greater fracture
toughness in the T735I temper. The partially intergranular fracture was not
significant in relation to fracture.
KEY WORDS: electron microscopy, fractography, aluminum alloys, micro-

structure, fracture toughness
One of the primary objectives of Subcommittee II of ASTM Com-

mittee E-24 on Fracture Testing of Materials is to obtain an under-
standing of the relationships that may exist among microstructure,
topological features of fracture surfaces, and resistance to crack
propagation? -5 To achieve this objective, separate task groups were
established to examine aluminum and steel specimens. Specimens of a
1 Scientific associate, Physical Metallurgy Div., Alcoa Research Laboratories, New
Kensington, Pa.
Group engineer, Metals Research Group, Commercial Airplane Div., The Boeing
Co., Renton, Wash.
3 Beachem, C. D. and Pelloux, R. M. N., "Electron F r a c l o g r a p h y - A Tool for the
Study of Micromechanisms of Fracturing Processes," Fracture Toughness Testh~g and
Its Applications, ASTM STP 381, American Society for Testing and Materials, 1965,
pp. 210-244.
4 Burghard, H. C., Jr., and Davidson, D. L., "'Fracture Mechanisms and Fracture Sur-
face Topography," International Conferences on Fracture, Japan, Sept. 1965.
s Kerlins, V., Phillips, A., and Whiteson, B. V., "Electron Fractography Handbook,"
ML-TDR-64-416, Wright-Patterson Air Force Base, Ohio, Jan. 1965.
] 96
HUNTER AND McMILLAN ON ALUMINUM ALLOYS 197
material of known composition, fabrication history, and mechanical

and fracture characteristics were to be examined independently by task
group members using whatever techniques each would normally use in
fracture-structure diagnosis.
Material
The specimens selected for the aluminum task group were a group of
7075 alloy specimens in the T651 and T7351 tempers that had been
subjected to tension and notch toughness tests, the results of which are
given in Table 1. All were from a single lot of 1% in. plate. The speci-
T A B L E 1 - Tensile and fracture properties o f long transverse
specimens o f 13/s-in.-thick 7075-T651 " and 7075-T7351 b plate.
7075-T651 7075-T7351
Identification 183201-T 183203-T
Tensile strength, psi ............................... 86 100 70 100

Yield strength, psi .................................. 77 750 58 500
Elongation in 4D, % ............................... 10.8 11.0
Net fracture strength, psi ......................... 65 500 70 700

Kit, notched rounds, psi ~ . . ................... 28 200 31 500
K~c, center notched panels, psi ~ .......... 26 100 34 400
Kc, 1-in. panels, psi ~ ......................... 30 900 60 400
a Solution heat treated 3 h at 870 F, quenched in cold water and aged

four days at room temperature plus 24 h at 250 F.
b Solution heat treated 3 h at 870 F, quenched in cold water, aged four
days at room temperature plus a two-step artificial aging treatment (U.S.
Patent 3,198,676).
mens supplied for fractographic examination were l-in. notched-round

fracture specimens taken in the long transverse direction from the
plate. In addition to mechanical data, electron transmission micro-
graphs of the two materials were supplied. These specimens promised
to be interesting subjects for task group investigation because of the
fairly pronounced differences in tensile and fracture properties and in
microstructure between the two tempers. With the more vigorous
aging to produce the T7351 temper, tensile and yield strengths were re-
duced appreciably, but fracture properties were increased.
Methods of Examination
The examination of the fractured specimens involved several
metallographic techniques. Electron transmission examinations to re-
veal alloy structures involved machining to about 0.040 in. thick,
chemically etching in a hydrochloric acid solution to about 0.005 in.
and finally thinning electrolytically in Lenoir's solution/Specimens for
722-cm z H:~POa, 134-cm 3 HzSO4, 150-g CrO3, 140-cm z HzO.
light microstructural examinations were prepared by conventional

mechanical techniques and were examined either as-polished or after
etching in Keller's etch. Fractographic examinations were made using
two-stage, plastic-carbon replicas and anodic oxide replicas,
Results of Investigation
Microstructure
Comparing the transmission electron microstructure of the two tem-
pers (Figs. 1 and 2) discloses that the principal structural difference
lies in the size of precipitate particles within the grains. Both tempers
have a considerable number of E-phase particles (Cr2Mg3Alls) about
1000 A in size in the form of a randomly distributed dispersoid. In
addition, both have considerable grain boundary precipitate, much of
which probably developed during quenching because the thickness of
FIG. 1 - E l e c t r o n transmission. Microstructure of 7075-T651 material (• E

indicates E-phase dispersoids and Z indicates zones.
FIG. 2 - E l e c t r o n transmission. Microstructure of 7075-T7351 material (x50,O00).

Note larger and more distinct zones than T651 temper in Fig. 1. E indicates E-phase
dispersoids, Z indicates zones, and M indicates M-phase plates.
the plate prevented rapid cooling after solution heat treatment. In the
T651 temper (Fig. 1), matrix precipitation is in the form of fine zones.
In the T7351 temper (Fig. 2), however, the zones are considerably
larger and more distinct, and, in addition, numerous small plate-like
precipitates of the transition M'-phase (MgZn2) have developed. Phase
identifications are based on previous diffraction work with these alloys
and tempers, and the stated compositions for the phases are those pre-
vailing at equilibrium.
Light Microscope Examinations

Examinations of the fracture surfaces of the fracture specimens were
first made under the low power light microscope. Both tempers ex-
hibited fairly rough fracture surfaces, with relatively few smooth areas,
indicating comparatively ductile fractures. In the T651 specimens,

however, a number of fine longitudinal fissures were noted.
Cross sections of the fracture surfaces were made and were ex-
amined under the light microscope at low (xl00) and high (x500)
magnification. The specimen in the T7351 temper had a very ragged
fracture and a large number of voids near the fracture (Fig. 3), sug-
gesting more numerous fracture initiation sites. The fracture of the
specimen in the T651 temper had about the same general roughness but
comparatively few voids near the fracture. Instead, the specimen had
smooth intergranular fissures extending normal to the fracture surface
and parallel to the stressing direction (Fig. 4). These were presumably
the fissures noted during low-power examinations of the fracture sur-
faces.
Electron Fractographic Examinations

Replicas of the fracture surfaces of the fracture specimens were ex-
amined at several magnifications with the electron microscope. The
primary fracture mode throughout the fracture surfaces was the
initiation and coalescence of microvoids. Fracture of and around
second-phase particles, stretching, and some intergranular fracture
was also observed. General views of the fracture appearance at the
start of fracture, further along the fracture path, and in the final frac-
ture area are shown by Figs. 5-7. At the x3000 magnification of these
figures, differences between the two tempers were not readily apparent.
Additional details of the fractures were obtained from examinations
at higher magnification. On the basis of extensive scanning with the
electron microscope, it was apparent that the fracture of the T7351
material was almost entirely transgranular. That of the T651 material,
however, was approximately 80 per cent transgranular and 20 per cent
intergranular. In transgranular areas, both tempers had large and small
dimples associated with large and small second-phase particles, along
with localized regions of shear and tear dimples (Figs. 8 and 9). In cer-
tain areas of the T651 fracture, smooth facets at an angle to the general
fracture surface were observed as shown at A in Fig. 10. These smooth
facets are believed to represent the longitudinal intergranular fissures
observed in the cross sections of the fracture region. This is sub-
stantiated by the fact that these facets contained many comparatively
large particles, probably the grain boundary precipitates.
Examinations at still higher magnification in the dimpled areas re-
vealed a difference in the size and number of dimples between the two
tempers. In the T651 specimen (Fig. 11), the dimples were fairly large
at this magnification, and the second-phase particles associated with
the dimples were the same size as the E-phase dispersoids (Fig. 11
versus Fig. 1). With the specimen in the T7351 temper (Fig. 12), the
FIG. 3 - Fracture contour and voids in vicinity of fracture of 7075-T7351 specimen.

FIG. 4-Longitudinal intergranular fissures in vicinity of fracture of 7075-T651

specimen.
(a) 7075-T651.
(b) 7075-T7351.
Two-stage, plastic-carbon replicas
FIG. 5 - Typical fracture topographies from the fracture initiation area (x3000).
(a) 7075-T651.
(b) 7075-T7351.
FIG. 6 - T y p i c a l fracture topographies from center of fracture (x3000).
(a) 7075-T651.
(b) 7075-T7351.
FIG. 7-Typical fracture topographies from final fracture area (•
FIG. 8-Oxide replica. Typical transgranular fracture appearance in 7075-T7351

specimen (x5000).
FIG. 9-Oxide replica. Typical transgranular fracture appearance in 7075-T651

xpecimen (x5000).
FIG. 1 0 - O x i d e replica, lntergranular fracture (a), a stretched region (b), and trans-
granular dimple rupture (c) in region of longitudinal fissure in 7075-T651 specimen
(•
FIG. l 1-Oxide replica. Comparatively large dimples associated with dispersoid

particles on 7075-T651 fracture (• E indicates dimple nucleation site believed
to have been an E-phase dispersoid particle.
FIG. 1 2 - O x i d e replica. Mixture of large and small dimples on 7075-T7351 fracture

(• E and M indicate dimple sites believed to have been E-phase dispersoid and
M-phase plates, respectively.
same large dimples were observed, but, in addition, there were a num-
ber of smaller dimples associated with smaller second-phase particles,
possibly the small plate-like M'-phase (MgZn2) particles (Fig. 2) de-
veloped by the higher temperature aging treatment. The net result was
more numerous dimples of a smaller average size in the T7351 temper.
Discussion and Conclusion

In relating microstructural features to the fracture properties of the
T651 and T7351 specimens, three factors are apparent, namely, the
difference in yield strength of the two tempers, the difference in dimple
size, and the partly intergranular fracture in the T651 specimen. The
lower yield strength of the T7351 temper would promote greater frac-
ture toughness. The smaller and more numerous dimples in this temper
reflect the greater number of fracture initiation sites, which would be
expected to give this temper poorer fracture toughness. The partly
intergranular fracture of the T651 material would tend to lower fracture
toughness if the intergranular path was normal to the direction of
stress. Since the intergranular fissures were parallel to the stress, they
might be expected to improve fracture properties by blunting of the
crack front and by contributing to a biaxial rather than a triaxial
stress condition. It is believed, however, that the intergranular fissures
were so few and so shallow that they did not help or harm fracture
properties. Thus, lower yield strength contributes to improved fracture
properties of the T7351 temper, and the larger number of dimples, re-
flecting additional fracture initiation sites, tends to decrease fracture
properties. It is apparent that the yield strength advantage more than
offsets the disadvantage of number of fracture sites.
W. R . W a r k e , t N . A . N i e l s e n , 2 R . W. H e r t z b e r g , 3
M . S. H u n t e r , 4 a n d M . Hill 5
Techniques for Electron Microscopic

Fractography
REFERENCE: Warke, W. R., Nielsen, N. A., Hertzberg, R. W., Hunter,

M. S., and Hill, M., " T e c h n i q u e s for Electron Microscopic Fraetogra-
phy," Electron Fractography, A S T M STP 436, American Society for Testing
and Materials, 1968, pp. 212-230.
ABSTRACT: The basic techniques of electron microscopic fractography are

reviewed. Steps to be prior to replication for preventing damage and for
cleaning are briefly covered. The three most common replication methods are
discussed in detail. These methods are the plastic carbon, direct carbon, and
oxide techniques. Examination techniques are also discussed covering the
range for visual selection of areas to be studied through photography of repre-
sentative areas. Stereoscopic viewing is given heavy emphasis because of the
special advantages of this technique. Finally, the complementary technique of
fracture sectioning is described.
KEY WORDS: electron microscopy, replication techniques, metals, metal-

lurgy, fracture surfaces, fractography, metallography
Electron microscopic fractography is a metallurgical tool which is

finding ever-increasing application in fracture research and service
failure analyses. Skill and proficiency in the techniques employed re-
quire a good deal of practice and time. However, the ASTM Sub-
committee on Fractography felt that a brief description of the basic
techniques would be most useful to newcomers to the field and would
assist them in avoiding some of the pitfalls which those experienced in
fractography have already encountered. This report is intended to be
very narrow in scope dealing only with fractographic techniques. For
general information on other metallurgical applications of electron
microscopy, the reader is referred to previous ASTM publications
Instructor, Illinois Institute of Technology, Chicago, I11.
z Research fellow, E. I. du Pont de Nemours & Co., Inc., Wilmington, Del.
3 Assistant professor, Lehigh University, Bethlehem, Pa.
4 Scientific associate, Alcoa Research Laboratory, New Kensington. Pa.
Research metallurgist, Republic Steel Corp., Brecksville, Ohio.
212
Copyright s 1968 by ASTM International www.astm.org
WARKE ET AL ON TECHNIQUES FOR ELECTRON MICROSCOPIC FRACTOGRAPHY 213
[1,2] 8 and to standard textbooks on the subject [3-5]. Also, the inter-
pretation of fract~graphs is not considered here, but is covered in the
Air Force Fractography Handbook [6], in several review articles, and
in many articles in the open literature [7-9]. One should become
familiar.with interpretation as well as techniques before attempting to
employ this tool in either research or failure analysis.
The present paper deals with the basic techniques of replication with
an introductory section on care and cleaning of fractures and later sec-
tions on examination and on fracture sectioning. Throughout this dis-
cussion, it should be remembered that three additional important fac-
tors are required; the person employing the techniques must always be
guided by common sense, patience, and carefulness.
Care and Cleaning of Fracture Surfaces

Common sense is certainly required in preserving and cleaning frac-
ture surfaces to be examined by eledtron microscopic fractography. It
is recognized that many of the requirements set forth in this section
cannot be met in the case of many service failures. However, the more
care that is exercised to avoid damage to fracture surfaces, the more
meaningful and reliable will be the information obtained.
Unless a fracture is replicated immediately after it is produced, it
should be preserved to prevent atmospheric attack. Depending on the
nature of the specimen material, varying degrees of care must be exer-
cised in protecting the fracture surface. Any metal or alloy which may
react with the environment must be protected as soon as possible after
the specimen has been broken. It is especially necessary to protect the
fracture surfaces of steels since they rust rapidly in moist air and many
fracture surface features are thus obliterated.
One may preserve the fracture surface of the specimen by storing it
in vacuum or in a storage vessel containing a desiccant. The latter is
obviously less costly and more practical. One may also spray the frac-
ture surface with a protective coating to inhibit oxidation and corro-
sion. A commercial spray acrylic lacquer has been used for this pur-
pose, and is very suitable since it is readily soluble in acetone. Coating
with an oil or grease may suffice if the coating does not contain any
compounds which might attack the metal specimen and does not con-
tain any insoluble silicones.
If a particular specimen is to be used for electron fractography and
yield meaningful information, mechanical damage to the fracture sur-
faces must be avoided. The two fractured halves must never be placed
back together again as is often done with service failures or when a
measurement of total elongation of a tension specimen is desired. If
6 T h e italic n u m b e r s in brackets refer to the list of references appended to this paper.
such information is necessary then duplicate test pieces should be pre-

pared (that is, one specimen will yield total elongation data while the
other will furnish the fractographic information). It is also suggested
that satisfactory fractographic information be obtained first and then
elongation data measured when only one specimen is available. Also,
rubbing, fingering, picking, or erasing on the fractures will destroy vital
features which are necessary for fractographic interpretation.
When cleaning the fracture surface, it is imperative that no corrosive
chemicals be used. To remove dirt, oil, and other coatings, acetone will
usually suffice. Other organic solvents, acids, or alkaline cleaners may
also be employed but should always be tested on a trial specimen to be
sure that no etching, pitting, or corrosion will result. Commercial ultra-
sonic cleaners are available and have been used extensively by those
working in the field of fractography. Ultrasonic vibration accelerates
the action of the solvent and assists in the removal of debris.
Another cleaning technique used by many consists of stripping off
and discarding the first few plastic replicas made from the fracture sur-
face. The softened plastic flows around the loose dirt on the fracture
surface, and when the replica is stripped the dirt is removed with the
plastic. The one disadvantage to this process is that particles involved
in the fracture process, which are of great interest to the investigator,
will be removed so as to preclude their identification. Surfaces should
never be cleaned by rubbing them with an abrasive tool or brush. To be
sure, the surface will be cleaned, but the process will introduce scratch
marks and obliterate a good portion of the fracture surface.
Replication Methods
There are three techniques of replication which are commonly used
for fractographic studies. These are the plastic-carbon method, the
direct carbon technique, and the oxide replica techniques. Each of
these replication techniques has its limitations and its areas of maxi-
mum applicability. The plastic-carbon technique is quite rapid and is
nondestructive, the fracture surface being preserved for further study
if desired. However, plastic-carbon replicas are more susceptible to
artifacts than are the direct-carbon replicas. The direct carbon and the
oxide methods give the maximum obtainable fidelity of reproduction of
the surface. For example, the resolution of a direct carbon replica is
about 50 ,~ while that for a plastic-carbon replica is about 150 .~. But
single step replicas require deep etching or electropolishing of the sur-
face to free the replica. Therefore, plastic-carbon replication is used
primarily for service failure analysis and in cases where preservation of
the specimen is of prime importance, while the other techniques are
used mainly for research work on laboratory-produced fractures where
maximum fidelity is of prime importance and where production of

duplicate specimens is not a problem.
Plastic-Carbon Replication
In this two step technique, a piece of cellulose acetate tape is
softened on one side by applying a few drops of acetone. After a few
seconds, the excess acetone is allowed to run off, and the tape is
pressed against the fracture surface. Other plastics, such as nitro-
cellulose or acetobutyrate may be used with suitable solvents to soften
them. Pressure is maintained (usually with one's finger tip or an eraser)
for a few moments (10 to 15 s). The specimen is then set aside for 10 to
20 min so that the plastic may be completely hardened. The hardening
period will depend upon such things as the degree of softening before
application, tape thickness, and ambient temperature. It is important
that the tape be moistened with the proper amount of acetone. If too
little acetone is used, the tape will not be able to fill in the crevices and
other rough surface contours on the surface. If, on the other hand, the
tape is too soft it will tend to tear apart when pressed onto the fracture
surface, and, in addition, bubble artifacts may be produced. Experience
has shown that a good replica can be prepared when the tape (usually
two to ten mils thick) is soft and sticky on one side but still firm on the
other side. The correct point of stickiness can be judged when the drop
of acetone no longer flows freely on the surface of the tape and the tape
has lost some of its stiffness.
The replicating tape should be removed only when it has completely
dried. If the tape tends to peel off slowly (like removing adhesive tape
from a surface) this indicates that the tape has not completely dried.
The tape should almost "pop" off when one begins to pry it loose. Re-
moving a tape that has not been completely dried can result in stretch
lines and other artifacts. The plastic tape, now having the impression of
the fracture surface, is called the primary replica.
The replica is shadowed by vapor depositing a heavy metal or alloy,
an oxide, carbon or platinum-carbon in a vacuum evaporator at a pres-
sure of less than 0.1 ~zm. This procedure enhances the contrast of the
final replica and results in easier recognition of various surface charac-
teristics. The material employed for shadowing is not too critical at low
magnification but at higher power, granularity and lack of stability
under the beam become important considerations. Also the heavier the
atoms of the shadowing material are, the greater will be the contrast
produced by a given thickness. However, too much contrast should be
avoided since important details may be hidden by the shadows. The
amount of shadow deposited can be judged by placing a piece of white
paper with an object or a drop of oil near the replica in the bell jar and
observing the development of the shadow layer on the paper. With ex-
perience the amount of shadow can be controlled more simply by the
proper selection of the size of pellet, length of wire or amount of ma-
terial to be evaporated, and the source-to-replica distance.
Contaminating gases and vapors influence the structure of the
shadow layer. Therefore, deposition should be made at the lowest pres-
sure obtainable in a reasonable length of time. For very high resolution
studies, a cold trap on the evaporator should be considered.
A shadow angle of 45 degis usually employed. If the surface is rela-
tively fiat, the shadowing angle is often reduced to produce longer
shadows from the small elevations. Shadowing in a direction related to
some important line on the specimen such as the macroscopic fracture
propagation direction is an often used technique. By observing the
shadows on the final replica, one is always able to correlate directions
in a particular region with this propagation direction. Also, one is able
in many cases to trace the fracture to an origin when no directional
features exist in the propagation region.
Carbon is then deposited onto the shadowed primary replica. This is
usually done at an angle of 90 deg to the replica surface. In cases where
the fracture surface is very rough, the replica may be rotated on a
motor drive table while the carbon is deposited at an angle less than 90
deg. This insures a continuous uniform layer of carbon on the fracture
surface. If rotary deposition is used, the carbon film will be less likely
to break up when the plastic tape is removed during final preparation.
Also rotary carbon deposition avoids confusing double shadows. The
desired thickness of the carbon layer should depend on the particular
situation. If high resolution is important, the layer should be as thin as
possible without running the risk of breakup during dissolution of the
plastic tape. A thin carbon film has a yellowish color. A thicker film is
obviously much stronger but does not have the fidelity of the thinner
film. A thick carbon film has a gray-black color. The thickness is con-
trolled by the amount of carbon evaporated and by the source-to-
replica distance.
The next step, solution of the plastic primary replica, is the most
critical and difficult step in the preparation of a replica. A maximum of
patience and care is required. In the simplest method, the two layer
plastic-carbon composite is carefully cut up into small pieces, about
V16 in. on a side. The shape of these pieces may be used to correlate
the orientation and location of the replica with respect to the specimen.
The pieces are then placed in a spot plate or Petri dish containing a
suitable solvent, acetone in the case of cellulose acetate. With the
plastic side down, the preshadowed carbon replica will tend to float
freely much more quickly. After three washings in fresh solvent baths
the replica is ready to be fished out with a copper grid. This operation
should be carried out under a binocular microscope at • to •

magnification. The film may be more easily fished out if most of the
acetone is removed first, reducing the mobility of the film. Replica mo-
tion can also be greatly reduced by placing the clean replica in an
alcohol bath.
Difficulty is often encountered during the washing operation since
the plastic swells as much as 50 per cent while dissolving and shatters
the thin brittle carbon layer into useless small pieces. A number of pro-
cedures have been proposed to avoid or alleviate this situation:
1. Balance plastic sheet thickness with fracture surface roughness.
Rough surfaces such as ductile ruptures require thick plastic to ensure
a continuous replica. Smooth surfaces like many fatigue fractures re-
quire a thinner plastic to prevent breaking.
2. Make carbon evaporation at the lowest possible pressure.
3. Cut the acetone by as much as 3 to ! with alcohol or distilled
water.
4. Warm the acetone bath.
5. Use vapors of acetone or an extractor to wash the plastic from the
replica.
6. Embed or laminate a specimen grid in the plastic to restrict ex-
pansion.
7. Coat the carbon side of the plastic-carbon composite with
paraffin before immersing in acetone. The paraffin can be subsequently
removed without expansion by washing in benzene.
The experimenter should try the various techniques to find the one
which he is best able to employ and which works most satisfactorily on
the particular specimens he is attempting to replicate. He should also
expect to improvise on the aforementioned list to suit his needs. In
any case, the end result is a carbon replica appropriately located on a
specimen grid ready to be examined in the electron microscope.
Direct Carbon Replication

In the case of direct carbon replication, the intermediate plastic
replica is eliminated, and the carbon film for examination is prepared
directly on the fracture surface. Structure and artifacts which could
result from the plastic replica are avoided. Thus, direct carbon replicas
are employed when increased fidelity is required and when specimen
destruction is permissible.
The shadow and carbon layers for a direct carbon replica are vapor
deposited directly on the fracture surface. It is common practice to
mask off all of the specimen except the area of interest using stop-off
lacquer or tape prior to shadowing, The shape of the unmasked region
and the direction of shadow can both be used to relate directions on the
replica to the fracture surface. The same rules and precautions for
these operations outlined earlier for plastic-carbon replicas apply to

direct carbon replication as well. It is even more important to employ
rotary deposition of carbon in this case to ensure a more uniform, con-
tinuous coverage of the fracture surface.
The shadowed carbon film is freed from the specimen by etching
away the surface to which the replica is attached. It is advantageous to
cut the replica into conveniently sized pieces by scribing the shadowed
fracture surface prior to etching. Any etchant, chemical polishing
solution or electropolish which is suitable for the alloy in question may
be employed for stripping direct carbon replicas provided it does not
result in gas evolution at the specimen surface. Gas bubbles forming
and leaving the surface tear the replica rendering it useless. One etch-
ing solution which is quite generally applicable but which requires ex-
treme care in handling is a 2 to 10 per cent solution of bromine in
methanol. CAUTION severe, slow healing, third degree burns result if
- -
this bromine solution comes in contact with the skin. Extreme boiling
or bumping can occur when the bromine is added to the alcohol unless
it is added slowly with adequate stirring and cooling.
The time required to free the replica is determined by trial and error
and may range from a few seconds to several hours. Gentle agitation
may be used to assist in freeing a loosely clinging replica. Once the
replica is floating in the etchant, it is lifted out and rinsed in a series of
alcohol or distilled water baths. Dilute acid may also be used to dis-
solve debris clinging to the replica, but an acid which does not attack
the shadow metal must be used. Often, the carbon film will roll up when
stripped, but platinum shadowing will minimize the tendency to curl.
Such replicas can be salvaged by alternate immersion in alcohol and
distilled water which causes the replica to either straighten or shatter.
Once cleaned, the replicas are lifted on specimen support grids and
examined or stored in covered containers for future study.
Oxide Replication
Oxide replicas have a unique application and ability in the examina-
tion of metal surface structure by electron microscopy. Some general
advantages of oxide replicas are the following:
1. The replica consists of a thin oxide film formed to controlled
thickness on the metal or alloy surfaces. As such it represents the
actual surface layer of the specimen.
2. The oxide films can be chemically or electrochemically stripped
from the specimem Their preparation and isolation introduce a mini-
mum of artifacts to be confused with true replicated structure.
3. Excellent contrast and resolution can be obtained. Variations in
surface composition resulting from. the presence of secondary phases,
precipitates, etc., affect the thickness and structure of the replica and
thereby the electron scattering ability of the film. The films, therefore,
have improved image contrast and definition.
4. Oxide replicas are particularly characterized by their dimensional
stability. They have the least tendency of all types of replicas to deform
or flatten upon positioning and drying on a support grid. The true
"three-dimensionality" of the replicated surface topology is maintained
in the replica. The use of stereotechniques is very much facilitated.
The prime requirement of oxide replicas is that they should them-
selves be "structureless" or have minimum electron microstructure.
Oxide films which are granular, or porous, or characterized by fine
oxide crystallites cannot adequately replicate the structural detail of
the substrate metal. These requirements are best fulfilled by the oxide
films which form or can be formed by various oxidation treatments on
aluminum, titanium, nickel, the stainless steels, high nickel alloys, and
the high alloy refractory metals. As a general rule, these metals and
alloys whose normal corrosion resistance or useful passivity behavior
is associated with the presence of thin, protective surface oxide films
are the ones which can be successfully oxide-replicated.
Aluminum and Aluminum Alloys-Aluminum and its alloys are es-
pecially well suited for oxide replication since they form very tight,
structureless, adherent anodic oxide layers. Therefore, oxide replica-
tion of these materials is used extensively. A common technique 7 con-
sists of replica formation in a 3 per cent tartaric acid solution adjusted
to a pH of 5.5 with ammonium hydroxide (NH4OH). The solution is
employed at ambient temperature with the specimen as the anode, a
high-purity aluminum cathode, and a forming potential of 20 V. The
specimen is tapped frequently to dislodge the bubbles which form.
Anodization is continued until bubbles no longer form, usually 5 to
10 min.
Areas to be examined are selected and the oxides scraped from the
remaining area. The replica is next stripped from the specimen
electrolytically in 20 per cent perchloric acid in denatured alcohol.
Again the specimen is the anode, and a strip of high-purity aluminum
is the cathode. A voltage of 12 to 15 V is applied until the replicas float
off from the specimen. The replicas are then washed thoroughly in dis-
tilled water.
Stainless Steels, Nickel, and High-Nickel Alloys-Mahla and
Nielsen [10] in 1948 developed a technique to form and strip oxide
films which serve as very satisfactory replicas for the above metals and
alloys. While no extensive investigation of oxidation procedures was
made, it was found that heat-tinting the specimens in a molten mixture
of NaNO3KNO3 (equal parts by weight) produced oxide films that
when stripped are sufficiently "structureless" to function as excellent
r Hunter, M. S., private communication, Alcoa R e s e a r c h Laboratories.
replicas. In their procedure the metal specimen can be prepared by the

usual metallographic techniques or with any degree of surface roughen-
ing as by deep etching or a fracture process. The specimen is then im-
mersed in the molten nitrate bath until it is superficially oxidized to the
extent that the surface shows a light yellow interference tint. A con-
siderable degree of control of the oxidation conditions is possible, and
visual inspection of the specimen is sufficient to determine when it
should be removed from the salt bath. For instance, a specimen of
deep-etched austenitic stainless steel was oxidized for 5 rain at 800 F
(425 C) to produce a replica film of satisfactory thickness and proper-
ties. However, there is considerable latitude in the time and tempera-
ture of oxidation so that an exact specification is not critical. There is
reason to believe, however, that the oxidation procedure which pro-
duces a replica film of a given thickness, say 250 ,~, at the lowest tem-
perature of oxidation should also develop the finest crystallite structure
within the oxide. (Room-temperature, air-formed "passive" films on
stainless steel are essentially structureless and amorphous.) T. Kat-
surai [I1] has used a 50 KNO~, 50 NaNO3, 1 MnO2 (by weight) fused
salt bath to study the electron microstructure of four different stainless
steels and Armco iron. His oxidation treatment was carried out at
300 C.
Subsequent to oxidation, the specimen is cooled in air, washed with
distilled water, and dried. After 1/s in. squares have been scribed with a
razor blade or a sharp-pointed instrument on the oxidized surface, the
specimen is wholly immersed in a solution of bromine in methanol.
(The concentration of this solution is not critical up to the maximum
safe level of 10 per cent. Nickel specimens require rather dilute solu-
tions, 2 to 3 ml of bromine per 100 ml of methanol, while a stronger
concentration, 7 to 10 ml of bromine in the 100 ml of methanol is
necessary for stainless steel specimens.) As mentioned earlier, extreme
care should be exercised to avoid bromine burns. The bromine solu-
tion attacks the film-free metal exposed along the scratches and
effectively undermines the oxide film, freeing it from the metal. When
the film is completely stripped- a matter of several minutes to several
hours according to the metal composition- small squares of oxide film
float free or cling loosely to the base metal. Gentle agitation of the
solution or specimen serves to detach this film. The squares of oxide
film are collected with a piece of 200-mesh wire screening and are
washed by transfer through successive dishes of fresh methanol. The
individual squares of film are then caught on a microscope screen,
dried, and examined in the electron microscope.
If a particular area on the metal specimen is to be examined, this
area only is originally scribed in the metal surface. The one square of
film is easily isolated for investigation.
Further examples of the use of oxide replicas of this kind are given
by (a) Mahla and Nielsen [12] in studies of carbide precipitation in
stainless steel (the oxide replicas also function as extraction replicas
when inert, nonmetallic precipitates are present in the alloy microstruc-
ture), (b) by Heindenreich [13] who used thermal oxide replicas to
study Widmanst5tten precipitates in Alnico V alloy heat-treated in a
magnetic field, and (c) by Nielsen [14,15] who used such replicas in
stress-corrosion cracking research and in examining platinum-deco-
rated sites of corrosion attack in plastically strained stainless steel
Other Metals- As pointed out, all refractory metals and alloys form
protective oxide films which are potentially useful surface replicas.
Titanium (and alloys) is one of the best examples. Oxide films from the
normal room-temperature oxide to heavier oxides built up by thermal
oxidation techniques can provide useful information on the titanium
topology and microstructure. Again film-stripping can be accomplished
by the bromine-methanol technique in most of these cases. It may also
be convenient, however, to experiment with electrolytic stripping
techniques. Gulbransen et al [16] have made investigations of oxide
films electrolytically isolated from metallographically polished speci-
mens of numerous metals and alloys oxidized at high temperatures in
0.1 atm of oxygen. Their techniques may be applicable to the fracto-
graphic examination of these materials.
Special Techniques and Applications

In this section, a number of special replication techniques and appli-
cations are described. These methods in some cases entail some steps
which differ from those described earlier, but in general the main re-
quirement is a great amount of the common sense, patience, and care
which was mentioned previously.
Origins and Other Specific Areas-In both fracture research and
service failure analysis it is often desired to use fractography to deter-
mine the mode of crack initiation as well as propagation. The site of the
fracture origin can be ascertained macroscopically by observing radial
patterns or arrest marks, or microscopically by tracing back cleavage
rivers or fatigue striations. Also, it is sometimes desired to examine a
particular fracture facet or other small region on a fracture surface.
Two problems are encountered in replicating such areas. First, origins
are often located at a free surface (that is, at the edge of the fracture
surface). Secondly, the specific area of interest will, in general, be lo-
cated over a grid bar and not be accessible for examination.
Replicating the entire fracture surface, out to the edge, is not too
difficult. Trimming off of the excess plastic with a scalpel or razor blade
before stripping is often of assistance. If the excess is not removed it
can cause difficulty during shadowing and washing. If an origin is lo-
cated at the edge, shadowing should be done in a direction opposite to

the crack propagation direction. This practice avoids shadow deposi-
tion on the cut edge of the plastic replica which will often fold over
resulting in a double layer at the edge of interest. During washing,
replicas tend to crack at the edges. So, to obtain a good replica of an
origin at this location, the carbon film should be as thick and as strong
as practicable. Also, unnecessary agitation and handling of the replica
during immersion washing should be avoided.
Placing of a replica on a specimen grid so that an area of interest is
over a grid opening is largely a matter of patience. First of all, the
operation should be carried out under a stereo binocular microscope at
• to • Secondly, grids of the coarsest mesh possible should be
used to maximize the open area. Also it is possible to use grids with
one enlarged opening, either purchased or manufactured with a
scalpel blade over which the area of interest is placed. Then, either
several replicas are prepared to improve changes of success or a replica
is carefully strained from an alcohol bath under the binocular micro-
scope, checked for location at about • and either refloated, or dried
and stored depending on the result obtained.
Matching Fracture Surfaces-Although more will be said later
about the examination of replicas from matching fracture surfaces,
there are some aspects of this technique concerned with replication. If
the specimen is of an amenable shape, it is often convenient to clamp
the two halves of the fracture together so that corresponding areas on
the respective halves are opposite one another. A plastic replica is then
made spanning the interface. Some plastic is usually forced down be-
tween the two halves of the fracture, and this fin of plastic should be
trimmed off with a scalpel prior to shadowing. After carbon deposition
small areas spanning the interface are cut out and washed so that the
corresponding regions from each half of the fracture are located on the
same specimen grid facilitating location and examination. Again, the
use of a coarse grid with a greater percentage of open area is advised.
This technique is applicable to square bar or sheet specimens but not,
in general, to round or irregular pieces.
Small Specimens-On occasion wire or foil specimens too small to
be replicated by the normal plastic-carbon technique are to be studied.
In such cases some success has been obtained by taping a piece of
plastic to a glass slide, resting the fracture surface on the plastic, and
flowing a drop of acetone under the specimen. The weight of the speci-
men may be sufficient to press the fracture into the soft plastic. The
specimen can be held in self-locking tweezer to provide additional
weight. The impression is usually sunk into the plastic, and the excess
material must be trimmed away to allow adequate shadowing.
Etched Fractures- Comparison of replicas of a fracture surface be-
fore and after etching will in many cases reveal much additional infor-
mation on the role of the various microconstituents in the fracture
process. Also, light etching will facilitate extraction of precipitates.
Usually a light etch with the normal metallographic etching solution for
the alloy being studied will suffice to delineate the various phases ex-
posed on the fracture surface. Determination of the optimum etchant
and etching time will be required for each alloy studied.
Examination
The next aspect of electron microscopic fractography to be dis-
cussed is the actual examination of the fracture. In this discussion, it is
assumed that the reader has access to a transmission electron micro-
scope and is familiar with its operation.
There are two advantages of the electron microscope over the light
microscope which have resulted in electron microscopic fractography
being such a useful tool. These advantages are greater magnification
and vastly increased depth of field. The high magnifications of which
the electron microscope is capable allow the study of submicron sized
features on fracture surfaces. However, most of the information to be
obtained from fractography is visible at relatively low magnifications
for the electron microscope, and it is the depth of field of the instru-
ment which is most important. Under normal conditions, this depth of
field is on the order of several microns. The short range level dif-
ferences on a fracture surface are usually less than this amount so that
the entire field of view is in sharp focus even though the surface repre-
sented is quite rough.
Visual and Low Magnification Examination

In general, examination of a fracture surface begins at the visual level
and proceeds to succeedingly higher magnifications with electron
microscopic fractography as the final step in this process. Even prior to
replication, as much information as possible should be obtained by
optical examination of the fracture at magnifications from xl to x500.
Studies with a binocular microscope up to x50 will give information on
the location of the origin of fracture, shear lips, changes in fracture
mode as revealed by changes in texture, and the direction of fracture
propagation as indicated by chevron marks, radial markings, and arrest
marks. Higher power light microscope studies will give further infor-
mation on fracture origins and directions and may indicate the general
fracture mode (that is, cleavage may be distinguishable from inter-
granular fracture).
Photomacrography of fracture surfaces becomes difficult at magnifi-
cations above • to • Here the principal limitation is depth of
field, the lack of which often makes it impossible to clearly show high
points and crack bottoms in the fracture. Also, metallic fractures com-
monly contain many small mirror-like areas at all possible orientations,
and these disperse the light to obscure detail. Among the combinations
of illumination which can be tried to avoid this trouble are highly
directional light both from the side and from devices like metallurgical
illuminators, polarized light, and monochromatic light. Stereoscopic
photography obtained by tilting the specimen between exposures will
often be helpful. Sometimes carbon, silicon oxide (SiO), or other ma-
terials have been vapor deposited on fractures to dull the mirror effect.
The lack of depth of field is the most troublesome problem, however.
A greater depth of field may be had for ordinary photographic lenses
by stopping them down to a small opening. With modern rapid films the
hour-long exposures of the past are obviated. At the higher magnifica-
tions this is not feasible, and so Zapffe [17] and co-workers introduced
the art of examining interesting regions which were nearly planar by
turning the specimen to bring the region in view and then grinding away
interfering material so that the microscope objective could be focused
on the fracture or aspect to be studied. Successful use of the method
requires great patience, manipulative dexterity, practice, and a reason-
able concept of what the area being sought should look like. Fre-
quently, it is possible to precede the detail grinding with careful elec-
trical discharge machining so that less time is consumed than if only
grinding was used. Another means of improving the depth of field is the
use of a scanning microscope which is commercially available. At the
time of this writing there are few reports in the literature of its applica-
tion to fractography. It appears to be capable of use from below x100
to above x1500 on the specimen surface without special preparation
and appears capable of covering a usefully large field. However, the
quality, resolution, and ease of use of the instrument for fractography
remain to be fully explored.
Low magnification studies of fracture surfaces may also be carried
out on replicas. In some cases the study of the fracture's surface may
be foregone to permit use of a replica made in the same way as for
electron microscopy. A negative plastic or silicone rubber replica can
be stripped from the fractured surface and studied. The replica quality
with silicone rubber is inadequate above about xl000, but at low
magnifications it is more completely pulled out of cracks and crevices
than the usual replicating plastics and can serve as a semipermanent
record of the overall fracture appearance. These replicas can be studied
as units, or by microtomesectioning, contour maps of the fracture can
be prepared. This, however, may be too time consuming for most in-
vestigations. The quality of either plastic or silicone rubber replicas
can be greatly improved by vacuum deposition of a metal such as
aluminum to increase the reflectivity and contrast. Plastic replicas may
also be examined with transmitted light, facilitating observations of

neighboring noncoplanar facets.
Electron Microscopic Examination

Once these low magnification studies have been carried out and the
general location of important features and the fracture growth direction
have been ascertained, examination proceeds to the electron micro-
scope. Again, examination begins at the lowest magnification normally
available. At x1500 to • it is usually possible to distinguish the
features characteristic of the major fracture categories (that is,
dimpled, cleavage, intergranular, or fatigue). At this magnification, the
replica is scanned to ascertain its shape, location, and orientation and
to locate the grid openings where replica quality is highest. The relative
proportions of the various fracture modes can be evaluated, and the
location of a specific feature which was seen using light microscopy can
best be found at the lowest available magnification.
Higher magnifications are employed to study features of special
interest in detail. Very fine striations can be resolved or the interaction
of fatigue striations with precipitates can be studied. The shape and
nature of the dimple nuclei in a ductile fracture region may be evident
athigh magnification. Many other examples could be given of the use of
the higher magnifications available in the electron microscope. In any
case, as was mentioned earlier, the maximum useable magnification
for a replica is set by the resolution of the replica itself. Plastic-carbon
replicas do not yield much additional information above about x 12,000
while the corresponding magnification for direct carbon replicas is
about •
There are situations where it is desirable to employ magnifications
on the electron microscope even lower than those normally obtained.
This can be accomplished on some microscopes by simply turning one
of the lenses down or off. On other microscopes, this is not possible,
but the same end can be reached by raising the replica away from the
objective lens in a special holder. Low magnifications are employed
when studying matching fractures. The magnification is reduced to the
point where one grid opening fits on a photographic plate, and the
replicas of the two fracture surfaces are mapped out grid opening by
grid opening. Prints of these maps are then compared to find the
matching regions which can then be examined and compared at normal
magnifications. Extra-low magnification is often convenient for locat-
ing specific features such as origins since one may employ the same
magnification in the electron microscope as was used in a light micro-
scope to locate the feature. Also by going to low magnification and en-
larging, it is possible to avoid montaging with its attendant seams and
mismatches.
Photography
Facilities for photographic recording of the fracture surface ap-
pearance are available on electron microscopes. Much could be said
regarding the relative merits of various films, developers, and so forth,
but that is not the concern of this report. In this connection it should be
pointed out, however, that the same precaution is necessary in
fractography as in all metallography. Namely, care should be exercised
to photograph typical rather than unusual areas on a fracture surface
unless there is clear evidence that the unusual area has some special
significance. It can almost be said that areas having the appearance
typical of every fracture made can be found on every fracture surface.
That is, small regions of brittle fracture are often found in the most
ductile specimens, and photographic recording of such a region would
be misleading and in the case of service failure analysis could lead to
completely erroneous conclusions. So the representative-rather than
the outstanding-should be recorded as fractographs, and enough
fractographs to adequately present the proportion of various fracture
modes in fracture surface should be obtained.
Stereoscopic Microfractography
Manufacturers of present-day electron microscopes incorporate in
all but the most inexpensive models simple specimen tilting devices
such that stereopairs of micrographs can easily be taken. It is to the
user's benefit to take advantage of this feature of the electron micro-
scope (and especially in fractography) as will be summarized in the
following section.
Advantages-The obvious advantage of stereomicrography is that
the observer gets a three-dimensional view of the specimen surface.
He is able, almost instantly, to see the true toPology, to know what is
up and what is down, to distinguish specimen from supporting film,
true structure from artifact structure, etc.
Thus, stereomicrography eliminates much of the confusion that often
results in ambiguous interpretation of the surface topography in a
single micrograph. The various shadings in photographic density be-
come meaningful in terms of topology. Fractography becomes a more
complete study and fracture modes more easily identifiable when the
fracture surfaces can be visualized in their true spatial geometry.
Another value of electron stereomicrography which is perhaps un-
appreciated consists of the increased resolution afforded of surface
structural detail. Because there are two separate micrographs which
the eyes and brain fuse into a single picture, that net image is the result
of the contributions of twice the number of electrons and photographic
grains. Imperfections and "noise" in the photographic emulsion tend
to cancel out. Even when one micrograph of the pair is off focus, the
single fused image appears sharp and clear.
Procedures-The first requirement, of course, is a good replica
which faithfully reproduces the surface topography of interest and re-
mains dimensionally stable and rigid. On the latter point, oxide
replicas are far superior to carbon replicas (particularly on fracture
faces where there are re-entrant surfaces). Quantitative measurements
will be in considerable error if the replica film flattens or distorts during
preparation.
To obtain stereomicrographs, two successive pictures are taken of
the same field, tilted between exposures along a line perpendicular to
the electron beam. The "built-in" stereoangle in electron microscopes
is usually in the 6 to 10 deg range.
The stereomicrographs are viewed as prints or transparencies (posi-
tives). In both cases they are positioned or mounted along a base line
at right angles to the axis of tilt and to the electron beam. Some points
to remember are the following:
1. The stereomicrographs should be taken, if possible, from areas on
the replica that lie on the axis of tilt. Otherwise there will be magnifica-
tion differences between the two micrographs which will have to be
compensated for when making exact photographic measurements.
Positive transparencies may or may not be capable of correction.
2. The exact same field must be photographed in the two successive
pictures. The same area must be brought back to the center of the
fluorescent screen for the second micrograph (after the specimen has
been tilted through the fixed stereoangle) and usually must be re-
focused.
3. The two exposures in the stereo pair should be made to duplicate
photographic density.
Viewing-Electron stereomicrographs are most commonly viewed
as prints using a stereoscope (or stereopticon). 8 When the electron
microscope is equipped for 35-mm roll film micrography, it is con-
venient to make 35-mm transparencies. These are mounted in standard
35-mm stereo-slide mounts which may then be examined in a hand
viewer, a table viewer, or by stereo-projection onto an aluminized
screen. Table-viewing and projection-viewing both involve super-
imposed polarized images. The observer must wear polaroid spectacles
to see the three-dimensional image.
All of this equipment is commercially available, having been de-
8Three sources of supply of such devices are: (1) Air Photo Supply Corp.. 158 South
Station, Yonkers, N.Y.; (2) Abrams Instrument Corp., 602 East Shiawassee St., Lans-
ing, Mich.; and (3) Wild Heerbrugg Instruments, Inc., 366 Main St., Port Washington,
N.Y. The latter two companies market more expensive stereoscopes having parallax
bars available for quantitative measurement of differences in elevation of topographical
features present in the stereomicrographs.
veloped by several manufacturers for the once popular 3 5-mm stereo-

photography hobby. Those interested in specific equipment for view-
ing and projection should consult the annual Industrial Photographic
Catalog.
Quantitative Surface R elief Measurements- Stereophotogrammetry
is the process of measurement from stereo pairs of micrographs (but
more commonly, photographs). Measurements of elevation of topo-
graphical features in a surface replica of a fracture face, for example,
would be typical of the process. A parallax bar is employed with an
appropriate mirror stereoscope. X-direction parallactic displacement
(that is, difference in separation perpendicular to the tilt axis) is meas-
ured with a micrometer and the value inserted into the following
formula [5] to calculate an elevation value:
CSC T
AZ-- ~tX
2M
In this equation, AZ is the difference in elevation of the two points,
23, is the total stereoscopic tilt angle, M is the magnification, and
is the parallactic displacement.
To prepare a contour plot sufficient measurements must be made on
the micrograph to connect points of the same elevation for a series of
elevations bracketing the range of "roughness" present in the surface
topography. Specific instructions on the use of parallax bars are
furnished by their manufacturers (two of these were listed earlier).
Fracture Sectioning
The final technique to be described in this report is fracture section-
ing. Although this is not a fractographic technique in the strictest sense
of the term, it is closely connected and can contribute much additional
information and so is included here. By examining the fracture surface,
the appearance characteristic of various fracture modes and testing
conditions can be studied and some conclusions regarding the micro-
mechanisms of the fracture process can be reached. By combining the
study of fracture profiles with fractography at the same magnifications,
much more can be learned about the role of microstructure in the
fracture process.
Often the best expedient to obtaining good fracture profiles is to
plate the fracture with nickel or chromium so that it will not be injured
in polishing. Then it may be sectioned and examined by conventional
metallographic means. Before plating it is essential that the surface
be clean. If there are deep crevices into which replicating plastic has
been forced and then broken off or if heavy grease is on the surface,
even ultrasonic cleaning may require several changes of solvent.
Frequently the electroplate does not penetrate into recesses but
bridges them forming pockets in the sections which catch etchants and
polishing compounds. This may be largely avoided by a preliminary
plate of chemical nickel or cobalt sufficiently heavy to smooth the sur-
face outline. This plate will be laid down as deeply as the liquid can
penetrate. It may be followed by electro-nickel or hard chrome. When
sections are made through cracks it is desirable not to use a pressure-
mounting medium. Indeed, application of a room-temperature setting
mount under vacuum is desirable. Pressure mounting may bend the
parts of the cracked specimen thus opening secondary cracks which
will confuse interpretation. Since the section-polishing technique
shows the conditions only at a single plane, it is easy to misinterpret
evidence of crack paths, the influence of inhomogeneities, and the like.
Consequently it is often desirable to polish, replicate, repolish, repli-
cate, and so continue until the photographs present a composite three-
dimensional representation of the fracture. The polishing and replicat-
ing techniques employed are usually the standard ones for the alloy
being studied and so will not be described here. Often it is desirable to
polish until some feature shown in an electron fractograph is reached
in section. This is always difficult but may be simplified slightly by
making a recognizable landmark which will be reached by the polished
surface a known distance before the desired feature is reached. Micro-
hardness indentations or electrospark marks are satisfactory as land-
marks. If the process is successful, a direct comparison of surface and
profile can be made.
References
[1] Advances in Techniques in Electron Metallography, ASTM STP 339, American
Society for Testing and Materials, 1963.
[2] Advances in Electron Metallography and Electron Probe Micro-Analysis, ASTM
STP 317, American Society for Testing and Materials, 1962.
[3] Techniques for Electron Microscopy, Kay, D., ed., Blackwell's Scientific Publica-
tions, Ltd., Oxford, 1965.
[4] Thomas, G., Transmission Electron Microscopy of Metals, Wiley, New York,
1962.
[5] Heidenreich, R. D., Fundamentals of Transmission Electron Microscopy, Inter-
science, New York, 1964.
[6] Phillips, A., Kerlins, V., and Whiteson, 13. V., Electron Fractography Handbook,
ML-TDR-64-416, Air Force Materials Laboratory, Wright-Patterson Air Force
Base, Ohio, 1965.
Study of Micromechanisms of Fracturing Processes," Fracture Toughness Testing
and Its Applications, ASTM STP 381, American Society for Testing and Materials,
1965.
[8] Pelloux, R. M. N., "The Analysis of Fracture Surfaces by Electron Microscopy,"
Metals Engineering Quarterly, Vol. 5, No. 4, Nov. 1965, p. 26.
[9] Warke, W. R. and McCall, J. L., "Fractography Using the Electron Microscope,"
Technical Report No. W3-2-65, American Society for Metals.
[10] Mahla, E. M. and Nielsen, N. A., "An Oxide Replica Technique for the Electron
Microscopic Examination of Stainless Steel and High Nickel Alloys," Journal of
Applied Physics, Vol. 19, No. 4, 1948, pp. 378-382.
[11 ] Katsurai, T., "Electron Microscopic Examination of the Surface of Stainless Steel
by Means of the Oxide Replica Method," Transactions, Chalmers University of
Technology, Gothenburg, Sweden, No. 96, 1950, p. 6.
[12] Mahla, E. M. and Nielsen, N. A., "Carbide Precipitation in Type 304 Stainless
Steel-An Electron Microscope Study," Transactions, American Society for
Metals, Vol. 43, 1951, pp. 290-322.
[13] Heidenreich, R. D. and Neshitt, E. A., "Physical Structure and Magnetic Aniso-
tropy of Alnico 5. Part 1," Journal of Applied Physics, Vol. 23, 1952, pp. 352-
365.
[I4] Nielsen, N. A., Physical Metallurgy of Stress Corrosion Fracture, Rhodin, T. N.,
ed., Interscience, New York, 1959.
[15] Nielsen, N. A., "I. Nature of Initial Corrosion of Stressed Austenitic Steel by
Chloride Ions. II. Platinum Decoration of Active Sites," Corrosion, Vol. 20, No. 3,
March 1964, pp. 104t-110t.
[16] Gulbransen, E. A., Phelps, R. T., and Hickman, J. W., "Electron Diffraction and
Electron Microscopic Study of Oxide Films Formed on Metals and Alloys at
Moderate Temperatures," Industrial and Engineering Chemistry, analytical edi-
tion, Vol. 18, 1946, pp. 391-400.
[17] Zapffe, C. A. and Clogg, M., "Fractography-A New Tool for Metallurgical
Research," Transactions, American Society for Metals, Vol., 34, 1945, p. 71.

ASTM - STP436 Electron Fractography

Uploaded by

Document Informationclick to expand document informationASTM - STP436 Electron Fractography

Document Informationclick to expand document information

Copyright:

Available Formats

ASTM - STP436 Electron Fractography

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ASTM - STP436 Electron Fractography

Uploaded by

Copyright:

Available Formats

ELECTRON

ASTM SPECIAL TECHN!CAL PUBLICATION NO. 436

List price $11.00; 20 per cent discount to members

Printed in Rahway, N.J.

The Symposium on Electron Fractography was presented during

Techniques of Electron Microscopy Diffraction

Fracture Toughness Testing and Its Applications,

Advances in Electron Metallography, Vol. 6,

Fatigue Crack Propagation, STP 415 (1967), $30.00

Electron F r a c t o g r a p h y - Tools and T e c h n i q u e s - J. L. MCCALL 3

Cleavage Phenomena and Topographic F e a t u r e s - H . C. BURGHARD,JR.,

Fracture by Microscopic Plastic Deformation Processes-c. O.

Application of Electron Fractography to Fatigue S t u d i e s - J . c. MC MIL-

Environmental Effects on Fracture M o r p h o l o g y - N . A. NIELSEN 124

Special Fractographic Techniques for Failure A n a l y s i s - B . V. WroTE-

Engineering Applications of F r a c t o g r a p h y - A . j. BROTHERS AND S.

Fractography and Microstructure of Aluminum Alloys 7075-T651 and

The purpose of this symposium is to describe the present "state-of-

Copyright~ 1968 by ASTM International www.astm.org

Electron Fractography Tools and

R E F E R E N C E : McCall, J. L., " E l e c t r o n F r a c t o g r a p h y - Tools a n d Tech-

ABSTRACT: Electron fractography has recently become a valuable tool for

KEY WORDS: fractography, fracture, election microscopy

Valuable information has long been known to exist in the fracture

were able to provide some significant information regarding the be-

Use o f Optical Microscopes

Examination of Fracture Surfaces

certain types of fractures by certain characteristic markings. For ex-

FIG. 1 - D e e p field photographic microscope developed by McLachlin [13].

Examination of Metallographic Cross Sections

FIG. 2 - Fracture o f a steel specimen photographed with a deep-field photographic

difficult to relate the features seen in the microscope to those macro-

Use of Transmission Electron Microscopes

The Replication Technique

chanically abraded, or in other ways damaged. As a result, their use-

FIG. 3-Plastic-tape replication o f a service failure.

FIG. 4 - S h a d o w i n g a plastic-tape replica in a vacuum evaporator.

plastic tape replica of the fracture of a rather large service failure.

FIG. 5-Examination of stereo-pairs in a special viewer to observe the three-dimen-

slightly different angles in the electron microscope. When these pairs

The Electron Microprobe Analyzer

commercial electron microprobe instrument. The instrument makes

FIG, 6 - C o m m e r c i a l electron microprobe analyzer.

FIG. 7 - C o m m e r c i a l scanning electron microscope.

signal from a spectrometer is fed into a cathode-ray tube that scans

The Scanning Electron Microscope

FIG. 8-Scanning electron micrograph of a fracture of magnesium oxide (see foot-

Reflection Electron Microscopy

Information Obtained from Electron Fractography

failures. On many occasions, it has been shown by electron fractog-

A Fractographic Analysis of the

REFERENCE: Spitzig, W. A., Pellissier, G. E., Beachem, C. D., Brothers,

ABSTRACT: Fracture surfaces of fatigue precracked, notched tension test

KEY WORDS: fractography, fracture toughness, alloy steels, maraging steels,

The important influence of microstructure and attendant strength

NOTE--All compositions in weight per cent.

spectively) were especially prepared from electrolytically refined iron

T A B L E 2 - Mechanical properties o f steels investigated.