0% found this document useful (0 votes)

45 views389 pages

ASTM - STP 645 - Fractography in Failure Analysis

The document details a symposium on fractography in failure analysis held by the American Society for Testing and Materials in May 1977. It discusses the application of fractographic techniques to evaluate material failures, organized into four main areas: techniques, environmental effects, fatigue, and stress and nonmetals. The volume serves as a reference for investigators and highlights the evolution of fractographic methods, particularly the use of scanning electron microscopy.

Uploaded by

mohsen.ir64

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

45 views389 pages

ASTM - STP 645 - Fractography in Failure Analysis

Uploaded by

mohsen.ir64

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

FRACTOGRAPHY IN m

FAILURE ANALYSIS

A symposium
presented at
May Committee Week
AMERICAN SOCIETY FOR
TESTING AND MATERIALS
1-6 May 1977, Toronto, Canada

ASTM SPECIAL TECHNICAL PUBLICATION 645

B. M. Strauss, Gulf Research and
Development Company, and
W. H. Cullen, Jr., U.S. Naval
Research Laboratory, editors

List price $36.50

04-645000-30

AMERICAN SOCIETY FOR TESTING AND MATERIALS

1916 Race Street, Philadelphia, Pa. 19103

Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
Copyright 9 by American Society for Testing and Materials 1978
Library of Congress Catalog Card Number: 77-91648

NOTE
The Society is not responsible, as a body,
for the statements and opinions
advanced in this publication.

Printed In Baltimore, Md.

May 1978

The symposium on "A Fractographic Approach to Failure Analysis" was

held at the American Society for Testing and Materials' Committee Week,
3-4 May 1977, in Toronto, Canada. ASTM Committee E-24 on Fracture
Testing of Materials sponsored the symposium. B. M. Strauss, Gulf Science
and Technology Company, presided as symposium chairman and W. H.
Cullen, Naval Research Laboratory, D. E. Passoja, Union Carbide Cor-
poration, and W. R. Warke, Illinois Institute of Technology, served as co-
chairmen.

Fracture Toughness Testing and Its Applications, STP 381 (1965), $19.50,
04-381000-30
Mechanics of Crack Growth, STP 590 (1976), $45.25, 04-590000-30
Fractography--Microscopic Cracking Process, STP 600 (1976), $27.50,
04-600000-30
Properties Related to Fracture Toughness, STP 605 (1976), $15.00
04-605000-30

This publication is made possible by the authors and, also, the unheralded
efforts of the reviewers. This body of technical experts whose dedication,
sacrifice of time and effort, and collective wisdom in reviewing the papers
must be acknowledged. The quality level of ASTM publications is a direct
function of their respected opinions. On behalf of ASTM we acknowledge
their contribution with appreciation.

A S T M C o m m i t t e e on Publications

Jane B. Wheeler, Managing Editor

Helen M. Hoersch, Associate Editor
Ellen J. MeGlinchey, Senior Assistant Editor
Sheila G. Pulver, Assistant Editor
Susan Ciccantelli, Assistant Editor

Introduction 1

TECHNIQUES

Application of Fractographic-Microstructural Correlations In

Evaluating Failure Mechanisms in Two Types of Steels--
J. H. STEELE, JR. AND D. F. LENTZ 5
Use of Laboratory Failure Simulation Exemplars to Study Intergranular
Fracture Modes in 9NI-4Co-0.20C Steel--J. D. YOUNGAND
ARUN KUMAR 32
Case Histories Illustrating Fractographic Analysis Techniques--
V. A. MEYN 49
Fractographic Method of Evaluation of the Cyclic Stress Amplitude
In Fatigue Failure Analysls--'A. MADEYSKIAND L. ALBERTIN 73
Discussion 82

ENVIRONMENTAL EFFECTS

Fraetographic Analysis of Gaseous Hydrogen Induced Cracking in

18Ni Maraging Steel--z. v. GANGLOFFAND S. V. WEI 87
Analysis of Fracture Morphology of Hydrogen-Assisted Cracking in
Steel and Its Welds--YONEO KXXUTA,TAKAOARAKI,AND
TOSHIO KURODA 107
Fracture of TI-8Ai-IMo-1V Alloy Fan Blade by Stress Corrosion
Cracking and Fatlgue--E. u. LEE, R. G. MAHORTER, AND
J. D. WACASER 128
Effect of the Cyclic Rate on Corrosion Fatigue and Fractography of
Type 304 Stainless Steel in Boiling 42 Percent Magnesium-
Chloride Solutlon--susu1vru HIOKIAND YOSHIHrKOMUKAI 144
Fractograplflc Observation of [Link] Cracking of AISI 304
Stainless Steel in Boiling 42 Percent Magnesium-Chloride
S O I u / I o n - - Y O S H I H I K O MUKAI, MASAKI WATANABE, AND
MASATG MURATA 164
Metallurgical Characterization of the Fracture of Aluminum Alloys--
M. D. BHANDARKAR AND W. S. LISAGOR 176

FATIGUE

Use of Microfractography in the Study of Fatigue Crack Propagation

under Spectrum Loadlng--F. R. ABELKIS 213

Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
Fractographie and Metallographlc Morphology of Fatigue Initiation
SItes--DANIEL EYLON AND W. R. KERR 235
Fractographie Analysis of Low Cycle Fatigue Specimens from a Failed
Steam Turbine Rotor--L. V. KRAMER 249
Role of Interface Chemistry in Failure of MaterlaiS--A. ~OSHI 275

STRESS AND NONMETALS

Examination of Fracture in a Pressure Vessel under Creep Conditions--
M. C. COLEMAN 297
Strength, Toughness, and Flaw Tolerance of 25.4-mm (1-in.) AHoy
Steel Lifting Chain--R. F. MCCARTNEYAND I. V. PELLEGRINO 312
Effect of the Amount and Shape of Inclusions on the Directionality
of Ductility in [Link] Steels--H. TAKADA,
K. KANEKO, T. INOUE, AND S. KINOSHITA 335
Comparison of the Threshold Stress Intensifies and Fracture Charac-
teristics for Temper Embrlttled and Deembrittled 2 1/~Cr-lMo
Steel in a Hydrogen Charging Environment--G. E. HICHO AND
C. M. GILMORE 351
Fractographic Analysis of Ceramics--~. i. MECHOLSKY, S. W. FREIMAN,
AND R, W. RICE 363

SUMMARY
Summary 383
Index 387

Introduction

The utilization of fractography as a means of determining crack origins

and mechanisms is a field of continuing development in research and for
diagnosing material failures. This symposium volume deals primarily with
the application of state-of-the-art fractographic techniques and interpre-
tations to material failures.
This symposium was organized into four broad areas: techniques, en-
vironmental effects, fatigue, and stress and nonmetals. The papers discussed
in the techniques portion offer general fractographic procedures used in
the pursuit of everyday material failures. The papers in the other three
sections discuss specific cases where the results can be applied to a broader
class of materials and failure mechanisms.
This volume will serve as a ready background reference for investigators
in its discussions of failures encountered in service and also as a fracto-
graphic atlas of typical fracture morphologies. It is interesting to note the
preponderance of scanning electron fractographs used in these studies
where, in the past, symposium volumes contained mainly transmission elec-
tron fractographs. It appears that the relative ease of specimen preparation
for scanning microscopy as well as its three-dimensional images is making
this the primary electron-optical tool in failure investigations while the trans-
mission electron microscope with its improved resolution is used for corrob-
orating fractographic interpretations.
We feel that this symposium volume is a benchmark in the development
of fractography in failure analysis and provides a sound basis for continued
progress in this area.
The collection of papers demonstrates the diversity with which fracto-
graphy is being used as an analytical diagnostic tool and the increasing
sophistication in the level of interpretation of these fractographs. We believe
that the techniques and methods presented here will continue to be refined
and serve as powerful tools in performing material failure analyses.
B. M. Strauss
Gulf Research and Development Company,
Pittsburgh, Pa. 15230, editor
W. H. Cullen, Jr.
U.S. Naval Research Laboratory, Washing-
ton, D.C. 20375, editor

1
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 1978by ASTMInternational [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
Techniques

Application of Fractographic-
Microstructural Correlations in Evaluating
Failure Mechanisms in Two Types of
Steels

REFERENCE: Steele, J. H., Jr. and Lentz, D. F., "The Application of Fractographlc-
Microstructural Correlations in Evaluating Failure Mechanisms in Two Types of Steels,"
Fractography in Failure Analysis, A S T M STP 645, B. M. Strauss and W. H. CuUen,
Jr., Eds., American Society for Testing and Materials, 1978, pp. 5-31.

ABSTRACT: Steel alloys exhibit a variety of fracture mechanisms depending upon their
composition, heat treatment, degree of cold working, and environmental and stress con-
ditions. Determination of the operative mechanisms from fractographic features re-
quires that comparisons be made between observed features and those which are
characteristic of cleavage, ductile rupture, or intergranular or hydrogen embrittlement.
In most cases this involves generating controlled laboratory fractures under conditions
which wilt produce each type individually. In addition, a correlation of microstructural
and fractographic features must be obtained to characterize these different mechanisms.
Illustrations of these two important aspects of fractographic analyses will be presented
for fracture processes occurring in drawn cup walls of low carbon sheet steel and in
bending tests of a quenched and tempered plate steel.

KEY WORDS: fractography, steels, microstructure, scanning electron microscopy,

fracture mechanisms, cleavage, intergranular, ductile rupture, hydrogen embrittlement,
drawing steel, quenched and tempered steel, plate steel, quasicleavage

O n e o f t h e m o s t i m p o r t a n t aspects o f f r a c t o g r a p h i c a n a l y s i s is a n u n d e r -
standing of the relationship between the microtopography of the fracture
s u r f a c e a n d t h e u n d e r l y i n g m i c r o s t r u c t u r e . T h i s type o f c o r r e l a t i o n b e t w e e n
t h e size, s h a p e , a n d s p a t i a l d i s t r i b u t i o n o f f r a c t o g r a p h i c a n d m i c r o s t r u c t u r a l
f e a t u r e s c a n b e o b t a i n e d b y a variety o f t e c h n i q u e s [1,2]. 3 O n e o f t h e m o s t
u s e f u l o f t h e s e is to e l e c t r o p l a t e a h a r d m e t a l l i c c o a t i n g over t h e f r a c t u r e

1 Research metallurgist, Research and Technology, Armco Steel Corporation, Research

Center, Middletown, Ohio 45043.
2 Formerly senior research metallurgist, Armco Steel Corporation, Research Center, Middle-
town, Ohio. Now deceased.
3 The italic numbers in brackets refer to the list of references appended to this paper.

5
Copyright by ASTM
9 Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright 1978by ASTM International [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
6 FRACTOGRAPHY IN FAILURE ANALYSIS

surface and then observe metallographic sections perpendicular to the

fracture plane [3]. An excellent discussion of the utility of this and other
sectioning techniques has been presented by Van Stone and Cox [4].
The importance of the concept of microstructure-fractographic correla-
tions is emphasized in this paper using two practical failure analysis prob-
lems in steels. Briefly, these may be described as:
1. Splitting in cold-drawn steel cups
2. Failure during 90-deg bend tests in a quenched and tempered plate
steel (0.3 percent carbon)
The problem of identifying the specific fracture mechanisms for these
steels (and many other ferritic alloys) is complicated by the interaction of
composition, heat treatment, cold working, and environmental and applied
stress conditions. For example, given a selected ferritic alloy composition,
microstructure (determined by the heat treatment, and cold worked state)
and environment, the fracture mechanism may change from ductile, to
intergranular, and to cleavage (including mixtures) depending upon stress
conditions. Thus, it is essential to have or obtain background information
on the fractographic features which are characteristic of these three mecha-
nisms for the particular alloy composition and microstructure when frac-
tured under controlled laboratory conditions. This information on the micro-
fractographic features which are characteristic of cleavage, intergranular
separation, or ductile rupture then can form a comparative basis for identi-
fying the fracture mode or mixture of modes occurring under field or test
conditions, which may be either uncontrolled or unknown.
Several alloy compositions, heat treatments, test conditions, sectioning
techniques, and observations are involved in characterizing the fracture
morphology in relation to the microstructure for the two examples which
are described. Although complete solutions to each of these problems have
not been developed at this point, the general observations and the relation-
ships between fractographic and microstructural features are sufficiently
definitive to illustrate the importance of the concept.

Fracture Morphology of Splits in Drawn Low Carbon Sheet Steel

In drawing quality low carbon steels, there is sometimes a tendency for
delayed cracking or splitting to occur in severely cold drawn cups. An
example of this type of fracture is illustrated in Fig. 1. The fracture initiates
at the top and crack propagation occurs in an apparently brittle 4 manner
down the cup wall terminating or blunting in some cases with a ductile
appearance. Examples of the type of fractographic features which are
characteristic of these splits are shown in Fig. 2.
4Apparently brittle is used to imply macroscopicabsence of plastic flow or necking. Also,
ductile here impliesthat some indicationof deformationoccurs macroscopically.

FIG. 1--Photomacrograph showing splitting type fracture in a drawn steel cup. The scale
shows the size in centimetres.

These scanning electron microscope (SEM) fractographs are quite diffi-

cult to interpret in terms of a ductile rupture, cleavage, or a brittle inter-
granular mechanism (or mixture thereof) without additional information
relating the observed features to the microstructure. In order to develop a
consistent interpretation of the fractographic features illustrated in Fig. 2,
which could then be used to describe the fracture mode, several sectioning
and controlled laboratory fracturing experiments were conducted and will
be described later.
Typical fracture surfaces were nickel plated to provide edge retention
and to minimize seepage during metallographic preparation. Three ortho-
gonal sections were cut from cup samples as indicated in Fig. 3 to provide
perpendicular views relative to the fracture surface. Conventional metal-
lographic techniques were used to prepare the sections which were examined
optically as well as with the SEM. Typical SEM micrographs from these
sections are presented in Fig. 4.

r-
c

FIG. 2--Scanning electron micrographs illustrating perpendicular views of the morphology o f the split fi'actures.
Micrographs (a) and (b) show areas near the imTer attd outer wall, respectively, and (c) and (d) are higher magnification m
micrographs o f typical features. m
C~
..r

z
N
N

FIG. 3--Diagram indicating orientation of metallographic sections obtained from drawn

cup walls relative to the split fracture surfaces.

The tangential and transverse sections show that the fracture path is pri-
marily along grain boundaries. They also reveal that features which have
the appearance of curled-up knife-edges or lips are present on the fracture
surface. These features which are marked with arrows in Fig. 4(a) and (b)
are very difficult to identify in the fractographs presented in Fig. 2. The
reason for this difficulty lies in the fact that the high secondary electron
intensity produced by the knife-edges also is produced by vertical steps and
the curled-up knife-edges, in most cases, are associated with vertical steps.
This type of SEM image contrast is not unusual since ductile rupture also
tends to produce a knife-edge topography between microvoids.
Careful examination of the metallographic sections also indicated that
there is a significant tendency for grain boundary rupture (as opposed to
brittle separation) at both the inner and outer skins of drawn cup walls.
This was verified by electropolishing the outer surface of a S-era-diameter
10-era-deep drawn cup which had exhibited splitting failure. Micrographs
indicating the occurrence of this ductile-type intergranular separation are
presented in Fig. 5. The occurrence of such a ductile intergranular fracture
mechanism has been reported recently by Meuris and Hornbogen [5] in

aluminum alloys exhibiting precipitate free zones along the grain boundaries
and by Hecker [6] in high strength low alloy (HSLA) steels. This grain
boundary rupture, which is currently under investigation, is thought to
have a ductile origin and to initiate splitting when external die constraints
are removed. Tension tests on cup walls indicate that the drawn material
will exhibit ductile behavior5 even with a significant amount of grain bound-
ary rupture present in the outer and inner skins. This suggests that residual
hoop stresses coupled with the grain boundary voids may cause crack propa-
gation from the outer surface into, as well as down, the wall when the cup
is removed from the die.
Sheet specimens were also cut from drawn cup walls and charged with
hydrogen in an attempt to produce intergranular fracture surfaces using
notched slow bend tests. Fractographs from various positions in a cup with
a flat top flanged region are presented in Fig. 6(a) through (d). The inter-
granular nature of these fracture surfaces is apparent, as are the elongated
grain shapes which are present in the drawn cup wall when compared to
the bottom (Fig. 6(d)) and top flange (Fig. 6(a)).
These micrographs also show some similarities to the morphology ex-
hibited by the split fractures as can be observed by comparison of Fig. 6(b)
and (c) with Fig. 2. However, there are two notable differences; first, the
knife-edge lips do not occur as frequently, and, second, extensive strain
markings are present on the intergranular facets of the hydrogen-charged
specimens. Although these strain markings could occur during the brittle
intergranular fracture of the hydrogen-charged specimens, they may in fact
reflect the grain boundary topography which existed within the cold drawn
mierostructure.
Figure 5(b) provides evidence on how this type of serrated grain boundary
topography could be formed during drawing. It shows heavy deformation
banding within a single grain and cavities formed at the tip (that is, triple
junction) and also where the bands intersect the grain boundary.
The micrographs presented in Figs. 4 and 5 also suggest how the knife-
edge lips are formed. If before final failure a majority of the triple junctions
are ruptured in a manner such as shown in Fig. S(b), then overlap between
grains which are being carried with the separate fracture surfaces will cause
their grain edges to be fretted. This will cause the grain edges or triple
junctions to be curled up without being blunted or rounded by rubbing.
Sheet Charpy specimens were prepared from drawn cup wall material so
that the crack path would have the same longitudinal orientation as the
split fractures. Specimens broken at -196 ~ had fractographic features
which were very similar to the splits as illustrated in Fig. 7(a), (b), and (c).
These micrographs were taken to illustrate the appearance of cleavage facets
side by side with the serrated grain boundary facets. The only difference

s Ductile failure in this case occurs by necking at approximately 45 deg to the tensile axis.

z
o

m
z

0
Z
m

r-
c
---I

r'-
c
m

FIG. 4--Scanning electron micrographs o f metallographic sections illustrating fracture surface m

C~
profiles (as retained by nickel plating) and the underlying microstructure. Nital etch: (a) and (b)
transverse, (c) tangential, and (d) longitudinal. Z

"n
r'-
c
FIG. S--Scanning electron micrographs illustrating grain boundary rupture which occurs in the inner and outer skins o f drawn cup m
walls. These were obtained by electropolishing the outer surface and etching with 3 percent nital. Note cavities at grain tips and deformation
bands. m
0
I
z

"11
F
c
3:1
FIG. 6--Fracture surfaces from hydrogen charged bend fractures from various positions in drawn cup : m
(a) top flange, (b) and (c) wall areas, and (d) bottom. These illustrate the elongated grain morphology in m
the cup wall and the strain markings on the facets.
-r
Z
N

--.k

r-
e
--I

"11
r"
C
m
FIG. 7 - - S c a n n i n g electron micrographs o f sheet Charpy f r a c t u r e surfaces; (a), (b), a n d (c) are f r o m l i q u i d nitrogen speci-
mens a n d (d) is a room temperature specimen. m
I
Z

between the fracture morphology in these specimens and those formed in

the cup walls is a slight increase in the amount of cleavage. Figure 7(b)
and (c) also illustrate the fractographic appearance of the serrated grain
boundaries such as the one indicated in the cross section in Fig. 5(b). Room
temperature sheet Charpy specimens had typically full oblique fractures
(ASTM Test for Plane-Strain Fracture Toughness of Metallic Materials
E 399-74). There were some very small flat regions where manganese-sulfide
stringers were apparently elongated along triple junctions as indicated in
Fig. 7(d).
If one excludes brittle intergranular fracture modes such as might be
caused by hydrogen [7] or solute segregation [8], then there are at least
three mechanisms which could account for the intergranular fracture
appearance.
I. Localized plastic flow in a thin layer on either side of a grain boundary
as described by Meuris and Hornbogen [5] where precipitate free zones
exist. These zones are not observed in the steels which exhibit the splitting
failures even though the material contains some extremely fine precipitates.
2. Grain boundary rupture at triple junctions and faces where sufficient
plastic flow occurs. This type of ductile fracture mode, in fact, may be
quite general in polycrystals where nonhomogeneity of plastic strain caused
by orientation differences must be accommodated at boundaries.
3. Microvoid formation at noncontinuous grain boundary precipitates.
This fracture mode, which cannot be completely excluded, may contribute
locally in initiating cavities at boundaries. It can be concluded, however,
from the absence of observable precipitates on the split and on the hydrogen-
charged fracture surfaces that it is extremely unlikely that this mechanism
can explain the failures.
Even though a complete understanding of the cause and the mechanism
of splitting failures has not been obtained as yet, the fractographic-micro-
structural observations which are presented provide consistent evidence
that some type of ductile grain boundary fracture is occurring during the
drawing process particularly in the skin regions where die interaction occurs.
They further demonstrate quite conclusively that the split fractures occur
by an almost completely intergranular mode. However, additional work,
which is presently underway, is needed to establish the cause of the inter-
granular fracture mode.

Fracture Morphology in Quenched and Tempered Armor Plate Steel (0.3

percent Carbon)
A problem of occasional insufficient 90-deg bend ductility in an armor
plate steel was investigated in an attempt to determine if it could be attrib-
uted to hydrogen embrittlement. The steel, which was rare earth treated
for inclusion shape control, had the following ladle chemistry.

C Mn P S Si Cr Mo

0.30 0.61 0.010 0.010 0.33 1.10 0.19

The plate used in this study was 5-ram-thick material which had been
quenched and tempered at 205 ~ to a hardness of 49 HRC. This steel had
a fine-grained tempered martensitic microstructure with no retained aus-
tenite.
Figure 8 illustrates the fracture morphology observed on the low ductility
90-deg bend specimens consisting of large dimples formed primarily by
rare earth sulfides 6 and titanium carbonitrides together with extremely
fine, shallow dimples in which second-phase particles are either absent or
below the resolution limit of the SEM. A few cleavage facets also could be
found as shown in the enlargement (Fig. 8(d)).
The problem of identifying features in the fracture morphology that
could be linked with hydrogen embrittlement in this complicated micro-
structure, which contains inclusions, prior ~ austenite grain boundaries,
lath and possibly plate martensite boundaries, as well as fine carbide parti-
cles and some proeutectoid ferrite near the plat surface, was investigated
by generating controlled laboratory fractures. These were produced by three-
point notched slow bend and Charpy impact tests at room and liquid
nitrogen ( - 196 ~ temperatures and also by a slow bend testing after
cathodically charging with hydrogen. These specimens were notched to
produce transverse fracture surfaces relative to the rolling direction.
The liquid nitrogen ( - 196 ~ fracture surfaces exhibited morphologies
as illustrated in Fig. 9. The SEM micrographs indicate that the surface
consists of a multiply faceted blocky texture with some extremely fine river
markings which are variable in orientation relative to the macroscopic
failure direction. The morphology of these flat blocky regions is controlled
by the martensitic lath structure as demonstrated by the cross-section pro-
files presented in Fig. 10. The fracture and microstructural-feature corre-
lation suggest that crack propagation occurs by the formation and linking
of very small clevage cracks ahead of the main front. This is referred to as
satellite nucleation by Beacham [9]. Some ductile shear steps also can be
observed linking the cleavage cracks in Fig. 9(a).
The features characteristic of this type of fracture have led to it being
called "quasicleavage" [1, 9] because it resembles cleavage with the apparent
exception that the facet size extends over much larger regions than the
tempered martensite features. In this microstructure, however, the facets
tend to coincide with the martensitic packets as indicated by the nickel-
plated cross sections presented in Fig. 10.
In addition there are blocky features which appear to be caused by sub-
6Possiblyoxysulfides.

z
o
p-
ro
z
-H
N
0
Z
m
<
I--
c
-H

"11
I"
c
:73
m
FIG. 8--Scanning electron micrographs o f bend test fracture specimen o f quenched and tempered plate: (a) and (b) rn
illustrate large microvoids f o r m e d by rare earth sulfides and titanium carbonitrides, and (c) and (d) provide enlarge- 0
"r
ments from areas shown in (a). Area (d) illustrates cleavage facets.
z

r-
m
z
.-I
N
0
Z
m

c
--I

71
r"
C
"n
m
m

FIG. 9--Fractographs indicating the morphology o f Charpy V-notch and three-point bend specimens
fractured at liquid nitrogen temperatures. A titanium carbonitride particle can be observed in (a) and a
rare earth sulfide in (b) as marked by arrows.

FIG. lO--MetaUographic cross sections of liquid nitrogen fracture surfaces showing pro-
files and their relationship to the microstructure.

structure within the tempered martensitic packets. These regions have a

tendency to exhibit one or more planar sides combined with fine knife-
edges. Examples are marked with the letter B in Fig. 10. These do not
appear to be "tongues" such as are formed by mechanical twinning [10].
Tear ridges between martensite packets also can be seen in Fig. 9(b) along
with the fine dimples and shear surfaces.
Two examples of inclusions also are indicated in Figs. 9(a) and (b). A
rare earth sulfide is shown in Fig. 9(b) in a rounded crater which is not
much larger than the particle itself, whereas the titanium carbonitride
shown in Fig. 9(a) (also see Fig. 11(c)) contains distinct internal crystal-
lographic cracking. This is probably cleavage parallel to the faces of the
cubical-shaped particles which have a sodium chloride (NaC1) type crystal
structure.
Figure 11 provides illustrations of the ductile fracture appearance formed
in room temperature three-point bend and Charpy V-notch specimens.
Large dimples are formed at rare earth sulfides and titanium carbonitrides.
These have a characteristic roughened appearance which is probably a re-
sult of extensive plastic deformation during void growth [2]. The ductile
Charpy and three-point bend specimens exhibited only a small central flat
fracture area with average width of about 1.4 mm and thus a 70 percent
oblique fracture (ASTM Test E 399-74). The fractographs shown were taken
from the central flat part of the ductile fracture surface.
The large dimples are surrounded by fine microvoids which in most cases
do not appear to contain particles. These regions may be examples of micro-
void sheet formation [2] although cross sections (Fig. 11(e)) do not seem
to provide confirmation of this observation. An example of a microvoid
formed several millimetres away from the fracture surface is shown in
Fig. 1l(f).
Figure 12 shows the slow bend fracture morphology generated by cathod-
ically charging S-ram-thick specimens for 48 h with hydrogen. The hetero-
geneous nature of crack propagation is evident from the low magnification
micrograph (Fig. 12(a)). The smooth areas with rounded edges which
occur along the specimen edges appear to have been formed by cracks
propagating along the tempered martensite lath boundaries as indicated
by the unusual feathery appearance illustrated in Fig. 12(b)). Nickel-plated
cross sections also suggest that the crack propagation mechanism is associ-
ated with lath boundaries as indicated in Fig. 12(c).
The central region of the specimens exhibited the same type of inter-
lath boundary morphology in small circular regions such as observed in
Fig. 12(d). These appear to have been formed by crack nucleation at ti-
tanium carbonitride particles. A metallographic section showing this type
of crack initiation is presented in Fig. 12(e). In addition, the central part
of the fracture contains microvoid regions (dimple rupture) which appear
to link the inter-lath cracks. Their morphology is illustrated in Fig. 12(f).

FIG. 1 l--Scanning micrographs illustrating fracture surface morphology and profiles from
room temperature notched three-point bend specimens, Circles in (a) indicate rare earth
sulfides, arrows titanium carbonitride, and squares iron rich particles. (c) and (d) are higher
magnification micrographs of microvoids containing a titanium carbonitride particle and rare
earth sulfide particles, respectively. (e) and (f) show the fracture surface profile, and a micro-
void formed away from the main fracture path.

FIG. 12--Scanning mierographs illustrating hydrogen charged three-point notched bend

fracture morphology. (a) Low magnification showing macroscopic appearance. (b) lath boun-
dary topography on flat regions, (c) nickel plated cross-section of lath boundary fracture,
(d) lath boundary cracks formed by cleavage of titanium carbonitride particles, (e) cross
section illustrating same effect as shown in (d), and (f) dimple rupture observed in central
region.

These fractographic-microstructure observations, although not conclusive,

suggest that hydrogen embrittlement may not be the cause of the low bend
ductility in this plate of steel. Additional studies are currently being con-
ducted to determine whether lower hydrogen concentrations can affect
microvoid formation and growth during ductile rupture in this steel. These
observations also indicate the following conclusions: (a) quasicleavage in
this plate steel exhibits a distinct relationship to microstructure in the size
of the facets and in the size and spatial arrangements of the smaller blocky
regions relative to substructure occurring within the tempered martensite
packets. (b) Hydrogen charging can embrittle martensite lath boundaries
producing a distinctive fractographic appearance. (c) Cleavage of titanium
carbonitride particles can nucleate lath boundary crack propogation in
hydrogen charged specimens and large microvoids during ductile rupture.

Discussion
It is important after discussing these two examples to emphasize several
factors which are important in making fractographic analyses with the
SEM.
1. The SEM fractographs alone can be misleading without knowledge or
background of the microstructure involved.
2. Controlled laboratory fractures, such as liquid nitrogen, Charpy im-
pact, or cathodicaUy charged hydrogen-induced embrittlement can provide
meaningful comparisons or guide lines in interpreting scanning microscope
fractographs.
3. Sectioning of fracture surfaces after plating to provide edge retention
is essential in interpreting and understanding the origin of fractographic
features.
4. The identification of hydrogen embrittlement as a cause of fracture
can be extremely difficult based upon fractography alone since it may not
produce a unique fracture mode.
In addition, the observations and interpretations discussed indicate that
SEM fractographs which in general are relatively easy to obtain may be
quite difficult to interpret unambiguously. This is emphasized by the need
for sectioning and for comparison with controlled laboratory fracture sur-
faces as pointed out for these two examples.

Summary and Conclusions

The importance of obtaining fractographic-microstructure correlations
in interpreting observed features and describing fracture mechanisms has
been illustrated by the two examples described. Also the observations, dis-
cussions, and results presented indicate that controlled laboratory fracture

experiments and the application of sectioning techniques can provide im-

portant supportive information in characterizing failure mechanisms.
The specific fractographic conclusions for the steels described have been
presented at the end of each section. The major conclusions can be sum-
marized as follows.
1. In deep drawing of low carbon steel there is a skin effect which can
cause a ductile grain boundary rupture.
2. The identification of an intergranular fracture morphology in heavily
deformed materials does not necessarily imply a brittle fracture mode.
3. Quasicleavage. in a tempered martensitic microstructure can be corre:
lated with packet size and substructure.
4. Heavy hydrogen concentrations such as can be obtained by cathodic
charging can cause martensitic lath boundary embrittlement.

Acknowledgments
The authors would like to express their appreciation to P. J. Erfort and
A. G. Golembiewski for their assistance with the laboratory work, and to
D. A. Sarno, W. G. Granzow, and A. J. Heckler for their helpful and
stimulating discussions during the course of the investigations reported here.

References
[1] Beachem, C. D. and Pelloux, R. M. N. in Fracture Toughness Testing and Its Appli-
cations, ASTM STP 381, American Society for Testing and Materials, 1964, pp. 210-244.
[2] Cox, T. B. and Low, J. R. Jr., Metallurgical Transactions. Vol. 5, 1974, pp. 1457-1470.
[3] Turkalo, A. M., Transactions, American Institute of Mining, Metallurgical, and Petro-
leum Engineers, Vol. 218, 1960, pp. 24-30.
[4] Van Stone, R. H. and Cox, T. B. in Fractography-Microscopic Cracking Process, ASTM
STP 600, American Society for Testing and Materials, 1976, pp. 5-28.
[5] Meuris, M. and Hornbogen, E., Praktische Metallographic. Vol. 13, 1976, pp. 160-171.
[6] Hecker, S. S., Metallurgical Transactions. Vol. 5, 1974, pp. 2107-2110.
[7] Bernstein, I. M., Metallurgical Transactions, Vol. 1, 1970, pp. 3143-3150.
[8] Rellick, J. R. and McMahon, C. J., Jr., Metallurgical Transactions, Vol. 5, 1974, pp.
2439-2450.
[9] Beacham, C. D. in Fracture, Vol. 1, H. Liebowitz, Ed, Academic Press, New York,
1969, Chapter 4, pp. 243-349.
[10] Brock, D., International Journal of Fracture Mechanics, Vol. 7, 1971, pp. 483-486.

Use of Laboratory Failure Simulation

Exemplars to Study Intergranular
Fracture Modes in 9Ni-4Co-0.20C Steel

REFERENCE: Young, J. D. and Kumar, Arun, "Use of Laboratory Failure Simula-

tion Exemplars to Study Intergranular Fracture Modes in 9Ni-4Co-0.20C Steel," Frac.
tography in Failure Analysis, A S T M STP 645, B. M. Strauss and W. H. CuUen, Jr.,
Eds., American Society for Testing and Materials, 1978, pp. 32-48.

ABSTRACT: During structural spectrum load testing of the aft-fuselage structure of

the B-1 aircraft, a 124-mm-long crack was detected in the 9Ni-4Co-0.20C steel pri-
mary structure. Review of the fracture location revealed that the crack path was ad-
junct to a production weld used in the fabrication of the 727-kg (1600-1b) forged and
welded horizontal-vertical stabilizer support assembly. Upon removal and examination
of the fracture on a scanning electron microscope, two distinct fracture modes were
observed: (a) at the origin, a 0.S to 1-mm intergranular fracture mode was observed,
followed by (b) 123 mm of transgranular high-stress low-cycle fatigue crack propagation.
Fatigue crack growth rate analysis on the scanning electron microscope showed that
the fatigue crack started propagation at the onset of testing, strongly suggesting a pre-
existing crack which was corroborated by the observation of an intergranular fracture
mode at the origin, which exhibited no evidence of cyclic growth.
A laboratory failure exemplar study program was conducted to establish standards
for intergranular fracture mechanisms possible in 9Ni-4Co-0.20C steel, fabricated
using the same environmental conditions as present during manufacturing. The salient
features of the fractographs of known failure mechanisms were used to establish the
cause of service failure intergranular cracking to be hydrogen embrittlement.

KEY WORDS: fractography, intergranular fracture, hydrogen embrittlement, weld-

ments

O n e o f t h e s i g n i f i c a n t i n n o v a t i o n s o n t h e U . S . A i r F o r c e B-1 a d v a n c e d
s t r a t e g i c a i r c r a f t is t h e use o f f r a c t u r e c o n t r o l in t h e d e s i g n o f all s t r u c t u r e s .
F r a c t u r e c o n t r o l is a p p l i e d to all single l o a d p a t h p r i m a r y s t r u c t u r e s . T h i s
involves five p h a s e s o f e v o l u t i o n to t h e f i n a l p r o d u c t i o n design, n a m e l y : (a)
a n a l y t i c a l d e s i g n a n d s e l e c t i o n o f m a t e r i a l s , (b) m a t e r i a l c o n t r o l ( f r a c t u r e
m e c h a n i c s ) , (c) d e s i g n d e v e l o p m e n t t e s t ( s t r u c t u r a l e l e m e n t s ) , (d) d e s i g n
v e r i f i c a t i o n t e s t (full scale s t r u c t u r e ) a n d (e) p r o o f a n d f l i g h t t e s t o f p r o t o -
type aircraft.

1 Supervisor and member of technical staff, respectively, Metallurgy Unit of Materials and
Producibility, Los Angeles Division, Rockwell International, Los Angeles, Calif. 90009.

32
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 by ASTM International [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
YOUNG AND KUMAR ON FAILURE SIMULATION EXEMPLARS 33

It was during the spectrum fatigue design verification test of the aft-
fuselage (empennage) after 991 flights of testing, that a 124-mm (4.875-in.)
long crack was detected in the 727-kg (1600-1b) cruciform shaped 9Ni-4Co-
0.20C welded steel structure member. This member supports the vertical
and both horizontal stabilizers and ties the empennage to the fuselage. De--
tailed examination of the test structure revealed that the crack path was
adjunct to the main production weld which joins the two steel forgings
which comprise the primary structure. Since 991 flights represent only 75
percent of an aircraft life (defined), the fracture surfaces were removed,
and a failure analysis was performed.
Subsequent to completion of the failure analysis and corrective action
implementation, subject test structure was repair welded on site, put back
into test, and completed 2.3 lifetimes of testing without recurrence of crack-
ing at this location.
The chemistry and mechanical properties of the 9Ni-4Co-0.20C steel
used in the structure are listed in the following table.

Chemical Composition

C Mn P S Si Cr Ni Co Mo V
0.16/0.230.20/0.40 0.010 0.010 0.2 0.65/0.85 8.5/9.5 4.25/4.75 0.9/1.1 0.06/0.12
max max max
Mechanical Properties

Ultimate Yield % % Reduc- Fracture Toughness,

Strength Strength Elongation tion in Area Klc
924 to 1020 MPa 875 to 972 MPa 14 to 19 55 to 65 121 to 143 MPax/~
(190 to 210 ksi) (180to 200 ksi) (110 to 130 ksiix/~.n. )

Procedures

Visual and Macroscopic

The location of the horizontal-vertical stabilizer spindle support fitting in
the aft-fuselage section of the aircraft and the crack path in the R / H side
plate are shown in Fig. 1. The thickness of the side plate is 27.2 m m (1.07
in.) and the crack traces on both the internal and external surface were
approximately 124 m m (4.875 in.) long. As will be seen later in the text,
the crack initiated at the toe of the weld and broke through the outer sur-
face at a repair weld on the side plate.
Specimens containing the fracture surface were removed from the struc-
ture. Figures 2 and 3 are mierographs of the specimen near the origin. The
load spectrum proved to be very helpful to determine the crack propaga-

FIG. 1--Location of horizontal stabilizer spindle support fitting in the aft-fuselage sec-
tion, and a schematic drawing showing the crack location and the origin.

tion rate, direction, etc., since the fracture surface was marked accordingly.
A total of 991 flights could be accounted for from the fractographs in the
fatigue fracture area, verifying immediate crack propagation upon start of
the test.

Scanning Electron Microscopy

The origin was examined by scanning electron microscope and the frac-
tographs are shown in Figs. 4 and 5. An intergranular fracture mode was
observed at the origin. The intergranular area at the origin was roughly
semicircular, about 0.5 mm (0.02 in.) deep and 1.0 mm (0.04 in.) wide. A
fatigue crack had initiated at the intergranular crack, which subsequently
propagated by conventional transgranular fatigue.
The grains in the intergranular area at the origin were equiaxed. No fine
structure or tear was observed on the grain facets. Electron microprobe
analysis did not reveal the presence of any corrosion products or other
foreign elements at the origin.

Metallography

Following fractographic analysis, the fracture specimen was carefully

sectioned adjacent to the origin and perpendicular to the side plate. The
specimen was mounted and polished to examine metallographicaUy the
section (see Fig. 6). The orientation of the specimen and the different struc-
tures also are identified. It can be seen that the intergranular portion of the

FIG. 2--Scanning electron micrograph showing the intergranular fracture mode at the
origin and fatigue striations originating from the intergranular area (origin).

crack was at the toe of the weld, The parent metal reveals a normal forged
structure. Another cast weld metal structure and heat-affected zone can be
seen away from the origin. Both heat-affected zones show the so-called weld
"eye-brows," which are created by the heat produced by individual weld
passes during welding.

Laboratory Failure Simulation

After establishing that the crack initiated from an origin with inter-
granular cracking, the direction of this failure analysis was diverted to
determine the cause of the intergranular crack in the 9Ni-4Co-0.20C steel
structure.
From fractography and other studies, the cause of intergranular crack-

FIG. 3--Scanning electron fractograph showing the origin.

ing at the origin was not evident. Therefore, it was decided to study the
various possible intergranular cracking mechanisms in 9Ni-4Co-0.20C
steel. Since the intergranular crack was formed sometime prior to the
start of test, and most probably during the manufacturing or processing of
the structure, or both, the complete manufacturing and processing history
for the part was reviewed. The following possibilities existed which could
have produced intergranular cracking: (a) crater cracking, (b) stress corro-
sion cracking, (c) hydrogen embrittlement, (d) elevated temperature frac-
ture, and (e) weld defect.
Laboratory failure exemplars were prepared for these mechanisms using
different variables. Attempt was made to duplicate the actual processing
conditions that the part underwent during manufacture. The fracture
surfaces produced by these known failure mechanisms were examined on

FIG. 4--[ntergranular fracture at the origin showing equiaxed grains.

FIG. 5--Detail scanning electron fractograph of the intergranular fracture at the origin.

FIG. 6--Metallographic mierosection through the origin showing the orientation of the
specimen and different areas.

the scanning electron microscope. The fracture mode, grain size and
orientation, and grain facet morphology were the features under considera-
tion to match the exemplars to the service failure.

Results and Discussion

Crater Cracking

Since the origin was determined to be at the toe of the weld, it was
considered that crater cracking might have caused the intergranular crack.

Crater cracking is caused by sudden termination of the welding process

such as pulling the weld torch away or suddenly cutting off the power
supply. Due to excessively rapid solidification and high residual stresses
generated above the equicohesive temperature, the cracks developed are
intergranular in nature. Crater cracking is a type of hot tear.
Attempts to produce crater cracks in 9Ni-4Co-0.20C steel specimens
during welding by cutting off the power supply or by pulling away the
welding torch, practically, were quite difficult. This would be expected
because of the high toughness and alloy composition of 9Ni-4Co.-0.20C
steel.
The weld crater fracture surface might be either unoxidized or oxidized
depending on whether the solidification occurred in an argon atmosphere
or in air. Since tungsten inert gas (TIG) welding was used, the power
supply may be cut off during welding with the torch still in place and the
crater crack would solidify under argon atmosphere. The other case could
be when the welding torch is suddenly pulled away and the crater crack
solidifies under a nonprotective atmosphere producing oxidation of the
fracture surface.
Crater cracks produced during welding (in argon) were broken open for
fractographic examination. The scanning electron fractographs are shown
in Figs. 7 and 8. The fracture mode is intergranular, exhibiting both
equiaxed and columnar grains. The columnar grains are near the surface
of the crater, revealing rapid solidification with a high rate of heat transfer.
Althouth the crack produced by crater cracking was intergranular in
nature, fractography revealed the following morphological differences from
the fracture surface of the service failure: (a) the grain separation was
very sharp and well defined, and (b) the grain facets exhibited a fine tear
structure not observed on the service failure.
The crater crack produced by suddenly pulling away the welding torch
loses the beneficial effect of the argon protective atmosphere, resulting in
oxidation of the surfaces exposed to air. Such a crater crack is shown in
Fig. 9. The solidified free surface is shown revealing oxidized dendrites.
Since no elevated temperature oxidation was observed on the fracture
surface in the service failure, this mechanism can be discounted totally as
a cause of failure.

Stress Corrosion Cracking

Stress corrosion cracking is a mechanical-environmental failure which
occurs in the combined presence of a sustained load and a corrosive en-
vironment. Since the crack was at the toe of the weld, where higher local-
ized residual stresses are produced by the solidifying weld metal, coupled
with restraint imposed by the adjacent metal and the part weight during
welding, one can account for the residual stresses produced during welding.

m
:lJ

-r
.<
-7
"11
I"
c
-In
m

z
r-
-<

G')

FIG. 7--Scanning electron fractograph showing columnar inter- FIG. 8--Scanning electron fractograph showing equiaxed inter-
granular cracking in weld crater, granular cracking in weld crater.

The simultaneous presence of unrelieved welding stresses along with a

humid noncontrolled atmosphere can produce stress corrosion cracking.
Three laboratory specimens of 9Ni-4Co.-0.20C steel in the as-welded
condition were subjected to stress corrosion cracking environments by
intermittent immersion in 5 percent sodium chloride (NaC1) solution in
water, with the specimens loaded in tension. All three specimens were
standard 6-ram (0.25-in.) diameter round-bar tension specimens and were
subjected to the same corrosive media, but were tested under sustained
loads of 75, 80, and 85 percent of the ultimate tensile strength, respectively.
Scanning electron fractographs of typical stress corrosion cracking fractures
are shown in Figs. 10, 11, and 12. Figure 10 reveals intergranular cracking,
however, the fracture surface is covered with corrosion products. Another
typical fractograph (Fig. 11) exhibits "mud-cracking" and recrystallized
salt crystallites on the fracture surface. A similar area was observed on
the specimen shown in Fig. 11 near the origin and is shown in Fig. 12.

Hydrogen Embrittlement
Hydrogen embrittlement is a mechanical-environmental failure which
occurs in the presence of hydrogen in metals and alloys in dissolved or
absorbed form, along with residual or static stresses. High strength steels
are very susceptible to hydrogen embrittlement and 9Ni-4Co-0.20C steel
falls in this category. Weld metal is generally more prone to hydrogen
cracking than the parent metal. Also, the heat-affected zone with higher
hardness than the parent metal is generally more susceptible to hydrogen
embrittlement than the parent metal, as has been evidenced for some
steels.
The presence of residual stresses in the part developed during the weld-
ing operation already has been described. Moisture can condense from the
atmosphere on the part prior to welding. Also, a very small leak of the
cooling-water line in the welding torch could act as a source of water and
consequently hydrogen. If the weld wire is stored unprotected in a humid
atmosphere, some moisture can be picked up by the wire. The use of non-
hydrogen controlled weld wire also can contribute to hydrogen embrittle-
ment. Hydrogen also may be produced by a corrosion reaction at the
surface which can diffuse into the steel. Thus, under these conditions,
hydrogen embrittlement could very well be the cause of intergranular crack
in the present service failure.
Standard round bar tension specimens of 9Ni-4Co-0.20C steel 6 mm
(0.25 in.) in diameter, in the as-welded condition, were cathodically charged
with hydrogen in an acid solution. Within 1 h after hydrogen charging, the
specimens were loaded at a constant load of 70 and 75 percent of the
ultimate tensile strength of 9Ni-4Co-0.20C steel, respectively. Both the
specimens failed in a totally brittle manner and no ductility was observed.

0
0

-o
ar
.<

"11
f-
t-
33
m
>
z
}>
I-
.<
cI)

FIG. 9--Scanning electron fractograph o f a crater crack f o r m e d in air. FIG. lO--Scanning electron fractograph o f 9Ni-4Co-0.20C steel speci-
Oxidized dendritic structure and oxide scale spalling are clearly visible, men failed by stress corrosion cracking.

z
o

-11

0
z
"11

r-
e

c
r"
..-I

z
m
x
m
"u

"11
FIG. l 1--Scanning electron fractograph o f another area f r o m stress FIG. 12--Scanning electron fractograph o f area at the origin o f the stress
corrosion cracking specimen revealing mud-cracking and salt crystallites corrosion cracking specimen.
on the fracture surface.
60

The scanning electron fractographs are shown in Figs. 13 and 14. The
intergranular topography and grain facet morphology very closely resemble
the intergranular area of the service failure, that is, no perceivable dif-
ferences existed between these fractographs (Figs. 13 and 14) and the
service failure (Fig. 4) when viewed at a wide range of magnifications.

Elevated Temperature Fracture

Grain boundaries are weaker than the grains at elevated temperatures
(T > Tm/2) and fractures are generally intergranular. At low temperatures,
however, grain boundary areas have higher strength than the grains, and
the fractures are generally transgranular (excluding environmental failures).
The temperature at which the transition occurs from transgranular to
intergranular fracture is called the "equicohesive temperature." Since the
origin area in the present service failure had been heated several times
above the equicohesive temperature during the welding passes, a possibility
for elevated temperature fracture exists.
Gleeble (hot tension) testing of 9Ni-4Co-0.20C steel weldments as 6-ram
(0.25-in.) diameter round specimens (with water cooled ends for gripping)
was performed at various temperatures under argon and air atmospheres.
Localized temperature adjacent to the fracture was measured by a thermo-
couple which was welded to the specimen. The temperature control was
within _+ 10~ (• 50~ Very high strain rates were applied and a con-
tinuous temperature recording was made during the test.
A weld specimen of 9Ni-4Co-0.20C steel was fractured in an argon
atmosphere at 1260~ (2300~ The intergranular fracture area is shown
in Fig. 15. The grain facets exhibit deformation bands and twins (probably
revealed by thermal etching).
Gleeble testing of a 9Ni-4Co-0.20C steel weldment at 1260~ (2300~
in air produced an oxidized intergranular fracture as shown in Fig. 16.
Wavelength dispersive microprobe analysis revealed that the oxidation
product on the fracture surface was iron oxide.
The 9Ni-4Co-0.20C steel weldment which fractured at 1315~ (2400~
in argon is shown in Fig. 17. The onset of grain boundary melting and
"tendril" formation is evident in this fractograph.
In summary, none of the fractographs produced by elevated temperature
fracture match the service fracture.

Weld Defect
The possibility of a weld defect at the origin causing intergranular crack-
ing was ruled out after examining the fracture surface of a 9Ni-4Co-0.20C
steel weldment that failed in a fatigue test, in which the crack initiated at a
lack of fusion. The scanning electron fractograph of the weld defect is

z
o

0
z
71
>
I

r-
e
33
rrl

cf -
>
.-t
5
z
m
M
m

r--
FIG. 13--Scanning electron fractograph of the fracture caused by hydro- FIG. 14--Scanning electron fractograph of another specimen failed due
gen embrittlement, to hydrogen embrittlement.
4~
01

"11
~0
0
0
fi3
"11
"u
"1"
-<

-11
>_
r-
e
:D
m

Z
3~
r-
.<
o9
o9

FIG. 15--Scanning electron fractograph o f an elevated temperature frac- FIG. 16--Scanning electron fractograph of an elevated temperature frac-
ture produced at 1260~ (2300~ in argon. ture produced at 1260~ (2300~ in air. Intergranular crack and
oxidized grain facets are evident.

z
o

3o
0
z
"1"1

F
c
;I)
m
o~

c
i-
-I
6
z
m
x
m
"13
r-
>
21J
FIG. 17--Scanning electron fractograph o f an elevated temperature FIG. 18--Scanning electron fractograph o f an intergranular crack at
fracture produced at 1315~ (2400~ in argon. Grain boundary melting a lack offuMon in the weld.
and tendril formation is evident. 4~
"4

shown in Fig. 18, which reveals a dendritic, rounded intergranular frac-

ture with strong evidence of thermal exposure.

Conclusions
An intergranular crack, roughly semicircular and about 0.5 mm deep
a n d l . 0 mm wide, was produced on the interior surface of the right-hand
side plate/weld relief hole of the horizontal stabilizer spindle support
fitting prior to the test, probably during manufacturing. The intergranular
crack most probably was caused by hydrogen embrittlement during the
period between the welding operation and the stress relieving as determined
by comparative fractography. This intergranular crack was present in the
part prior to the aft-fuselage fatigue test. Upon beginning of the test, a
fatigue crack initiated at this preexisting crack and propagated through
the side plate for 991 flights of testing, until a 124-mm-long crescent shape
crack was observed on the exterior surface of the side plate. Corrective
actions have been implemented in the welding process to preclude recur-
rence of delayed environmental failure of the type observed in this study.

Case Histories Illustrating Fractographic

Analysis Techniques

REFERENCE: Meyn, D. A., "Case Histories Illustrating Fractographlc Analysis Tech-

niques," Fractography in Failure Analysis, A S T M STP 645, B. M. Strauss and W. H.
Cullen, Jr., Eds., American Society for Testing and Materials, 1978, pp. 49-72.

ABSTRACT: The Naval Research Laboratory receives cracked and fractured parts
ranging from the exotic to the mundane: for example, titanium alloy jet engine com-
ponents, ultrahigh strength steel landing gear parts, aluminum alloy airframe sections,
and galvanized mild steel radio antenna support frames. Failure analysis involving so
many materials over such a wide range of applications demands great adaptability in
the use of fractographic techniques. The use of all techniques from the unaided eye
through low-power magnifiers to high-powered light microscopes and transmission and
scanning microscopes, will be discussed, using actual case histories for illustration. The
application of surface chemical analysis to failure analysis also will be mentioned.

KEY WORDS: failure, fractures (materials), fractography

T h e E n g i n e e r i n g M a t e r i a l s Division o f the Naval Research L a b o r a t o r y

(NRL) receives m a n y failure analysis requests every year, mostly f r o m Navy
activities, b u t occasionally from o t h e r sources. Usually, we get specimens
which a l r e a d y have u n d e r g o n e e x a m i n a t i o n or p r e l i m i n a r y t r e a t m e n t by
o t h e r o r g a n i z a t i o n s , a n d t h u s are s e l d o m in virgin condition. Nevertheless,
it is usually possible to o b t a i n c o n s i d e r a b l e i n f o r m a t i o n even f r o m severely
d a m a g e d fractures. Typically, t h e s u b m i t t e r wishes to know as m u c h as
possible o f t h e history o f the f r a c t u r e d p a r t from the t i m e o f m a n u f a c t u r e
a n d hopes t h a t we can r e c o n s t r u c t the events l e a d i n g to failure; b u t m o r e
often the investigation will end in some u n c e r t a i n t y as to the f r a c t u r e
m e c h a n i s m itself. T h e honest failure investigator m u s t a d m i t t h a t it is not
always possible to distinguish between corrosion fatigue a n d stress corrosion
c r a c k i n g (SCC), for e x a m p l e , or even between fast f r a c t u r e a n d SCC in
some instances. Therefore, a knowledge o f the c i r c u m s t a n c e s of failure a n d
t h e o p e r a t i n g stresses a n d e n v i r o n m e n t can be o f g r e a t help in d i a g n o s i n g
a failure m e c h a n i s m .

1Metallurgist, Naval Research Laboratory, Washington, D.C. 20375.

49
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 1978 by ASTM International [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
50 FRACTOGRAPHY IN FAILURE ANALYSIS

NRL Failure Analysis Techniques

Specimens are examined first and photographed as completely as pos-
sible without initial cleaning or other treatment, to an extent judged desir-
able in light of the history of the failure and the type of information likely
to be required. For example, if the composition of surface contaminations
is desired, ultrasonic cleaning and replica stripping would defeat analysis.
After such preliminary analysis, the whole fracture, or cut sections thereof
as necessary, is cleaned ultrasonically and examined in the light microscope
and scanning electron microscope (SEM). Then replicas are stripped, either
to remove further surface coatings or for transmission electron microscope
(TEM) examination. At times it is found necessary to resort to rather vig-
orous cleaning methods, such as chemical cleaning and mechanical re-
moval of dirt and surface layers. Some background on such methods have
been provided elsewhere [1]. 2
The usual techniques employed by others are used at NRL, including
low-power stereomicroscopy, light and electron microscopy, and metallo-
graphy. In addition, specialized surface analysis techniques such as electron
microprobe and Auger spectrometry are available.

Illustrative Examples
Fractography was not well developed in the early 1960s and most failure
analyses involved considerable analytical work and laboratory simulation of
fracture to identify causes of fracture. An example of the pioneering na-
ture of failure analysis of the time is the cracking and failure by ejection of
some sintered tungsten rocket nozzle liners [2,3]. One of these, ejected
from the nozzle after 30 s of a planned 1-min test firing, was intact when
recovered from a pool of water some distance away, but contained
numerous circumferential cracks on the outside which penetrated about 75
percent of the thickness. At first it was thought that the cracks occurred
upon quenching from operating temperature (about 4700 K after 20 s)
when the liner hit the water. However, further study, which included
exhaustive metaUography, X-ray diffraction, X-ray fluorescence spectro-
scopy, and fractographic examination, led to the conclusion that the cracks
were caused by thermal stresses and occurred early in the test firing: (a)
many cracks were partly coated with a thin layer of fused aluminum oxide
(AI 20 3), indicating they were there during firing. The A120 3 originated from
the solid rocket propellant; (b) the tungsten matrix contained various con-
taminants in greater quantities on one side of a crack than on the other,
again ruling out the possibility of cracking after ejection; (c) the cracks ini-
tiated on the outside surface of the liner, where tensile stresses are highest
during rapid initial heating in the early stages of firing; (d) they stopped after
2The italic numbers in brackets refer to the list of referencesappended to this paper.

penetrating 75 percent of the thickness, near the expected point of

transition from tensile to compressive thermal gradient-induced stress; and
(e) they initiated and grew at a fairly high temperature above the brittle-to-
ductile transition temperature. Figure 1 shows fractographs comparing the
service crack surface, Fig. l(a), with the room temperature fracture
surface, Fig. l(b). The room temperature fracture shows well defined inter-
granular fracture mixed with cleavage, whereas the service fracture surface
consists of pointed, drawn-out hillocks. Figure 2 shows a service crack in
cross section, suggesting the intergranular nature of the cracking and
showing considerable distortion of the grains. The entire length of all
cracks had the same appearance, which is evidently a type of plastic inter-
granular decohesion occurring at high temperature. The possibility that the
cracks might have existed before firing was ruled out because of careful
inspection before assembly, the large size of the cracks, and the likelihood
that the surfaces would have been smoothed thermally rather than made
more jagged by high temperature exposure.
The following failure analysis provided a particularly classic example of
the progression of hydrogen embrittlement cracking (HEC)/SCC propaga-
tion mechanism as the stress intensity factor increases with crack growth,
and changes after the onset of fast fracture in very high strength alloy
steels. The broken part was a jet fighter-bomber catapult socket lug which
broke during test after 617 simulated launches. The lug was made of
AMS6407 (4330Si) steel, heat treated to about 180-ksi yield strength. The
fracture, Fig. 3, bisected a threaded lube fitting hole which was filled with
grease during testing. No corrosion was evident and although the lug had
been cadmium plated, the threaded hole where cracking initiated was not.
The cracking mechanism was intergranular at the origin, Fig. 4(a), became
mixed with cleavage and dimples further along, Fig. 4(b), and ultimately
changed to 100 percent microvoid coalescence at the end of the slow crack
area, Fig. 4(c). Some evidence of fatigue was noted in a narrow band near
this point, but did not comprise a significant part of the slow crack zone.
The remainder of the fracture surface contained a mixture of cleavage,
microvoid coalescence, and intergranular cracking which is typical of fast
fracture in this steel at this strength level. In the absence of obvious cor-
rosion, the crack initiation and slow growth was attributed to HEC. The
hydrogen was probably induced during cadmium plating of the exterior
parts of the lug, and diffused into the unplated regions during storage. The
initial low stress intergranular cracking may have been caused by over-
tightening of the grease plug, the remainder of crack propagation and final
fracture occurring in increments during simulated launch testing.
The progression of fracture mechanisms from pure intergranular at the
origin through quasicleavage to microvoid coalescence as the crack
lengthens and crack tip stress increases is very diagnostic of HEC and SCC
failures in low alloy steels of this strength level. The reversion from p u r e

Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further r
01

"11

0
-4
0
:D
"o
-I-
.<
-7
"11
>_
t--
:TJ
m

Z
r-
-<

FIG. 1--Tungsten rocket nozzle insert fracture surfaces (two stage replica TEMs). (a) In-Service failure near origin.
(b) Room temperature laboratory f a s t fracture.

FIG. 2--Metallographic cross section o f in-service crack in tungsten rocket nozzle insert.
The crack is filled with epoxy resin which appears somewhat darker than the tungsten and
runs from upper left to middle right.

FIG. 3--Fracture surface o f aircraft landing gear catapult lug which failed after 617 simu-
lated launch load cycles. Slow crack growth area at upper right.

FIG. 4--Catapult lug fracture surface (two stage replica, TEAls). (a) Nearly lO0! percent
intergranular at origin of slow crack region. (b) Mixed intergranular with transgranular
cleavage and dimples midway through slow crack area. (c) 100 percent dimples within slow
crack area near onset of fast fracture.

FIG. 4--(Continued. )

microvoid coalescence to a mixture of mechanisms indicates the onset of

final overload (fast) fracture. HEC and SCC cannot be differentiated
except where obvious association of a corrosion pit with the origin is seen,
since in most cases both types of crackirig result from hydrogen activity.
The main distinction occurs in that HEC results from hydrogen already in
the metal before cracking starts whereas the hydrogen is produced by the
local corrosion processes during SCC.
An early fractographic failure analysis that still has something to teach
involved a study of an aircraft skin assembly comprising two 7075-T6
aluminum sheets overlapped and spot welded [4]. A through-thickness
crack traversed the skin assembly following a line of spot welds.
Fractographs of two areas of spot-weld fracture are shown in Fig. 5.
Although effects of corrosion are evident, both dimples and striation-like
markings can be seen. The investigators were able to prove fatigue was not
the cause by breaking other spot welds in the skin assembly by fatigue and
fast fracture, respectively, showing that the spot weld fatigue striations,
Fig. 6(a), were quite different from the service fracture markings, which
were also observed in the laboratory-produced spot weld fast fractures, Fig.
6(b). Such distinctions between other parallel markings and true fatigue
striations were not then as obvious as they might seem today and evidently
still cause some confusion.
One of the most surprising recent developments in SCC was the demon-
stration that titanium alloys are subject to this phenomenon in ordinary
water if precracked. However, it was still widely believed that titanium

>
-o
I
-<

-I1
r'-
c
m

z
r-

F I G . 5--7075-T6 aircraft skin sheet spot weld fracture surface (two-stage replicas, TEMs). (a) In-service fracture.
(b) Same.

z
f--
-<

-..t
m
I
z

c
m
FIG. 6--7075-T6 aircraft skin sheet spot weld fracture (two-stage replicas, TEMs). (a) Laboratory-produced fatigue
fracture. (b) Laboratory-produced fast fracture.
"4

alloys would not be subject to SCC if not precracked, especially in media

which did not cause severe pitting. Real shock waves radiated through the
titanium metallurgical fraternity when rocket motor tankage made of Ti-
6A1-4V (160-ksi yield) filled with reagent quality methanol began
fracturing at unexpectedly low stresses. NRL scientists were asked to verify
the failure mechanism, which was announced to be SCC in methanol. The
original service fractures were found to be a mixture of quasicleavage and
microplastic cracking at their origins, typical of aqueous SCC in
precracked alpha-beta titanium alloys. Strips of tankage material, care-
fully screened for flaws, were immersed in reagent grade methanol,
stressed to 65 percent of their yield strengths, whereupon they failed within
minutes. The fracture surface origin area, Fig. 7, was identical with the
service fractures, confirming that the service failures were caused by the
same mechanism, SCC.
The following analysis may well seem of self-evident outcome to many
but it caused considerable concern to the analyst under pressure to supply
the correct answers. Some pieces of a fractured turbojet engine compressor
blade, made of Ti-8AI-1Mo-IV alloy, were submitted for our analysis after

FIG. 7--Ti-dAll-4V rocket fuel tank stress corrosion failure in methanol, initiating from
as-machined unnotched surface (arrow). Two-stage replica, TEM.

having already undergone examination by another agency. One of four cut

pieces comprising the mating fracture surfaces at the apparent failure
origin was selected for analysis. It appeared to be in the best condition, one
piece looking corroded, the other two having been heavily dented right at
the origin edge. The fracture surface of the selected piece showed what
appeared to be cleavage facets pocked with round cavities, Fig. 8(a), and
extensive areas of unusually well-shaped dimples, Fig. 8(b). It appeared
necessary to examine the mating fracture surface to see if such features
appeared there. None were found, the typical appearance being shown in
Fig. 9(a), and a brief precision matching study (not illustrated to save
space) proved that the "dimples" appeared only on one of the mating sur-
faces. The cause of the dimples was undoubtedly etch pitting from the
inspection procedure used to look for small cracks, which comprises a blue
anodizing treatment. The major fracture mode near the origin was cleavage
with no visible striations, even when examined by TEM. Some isolated,
very coarse striations were found near the terminal end of the slow crack
zone, Fig. 9(b). The cause of crack initiation and failure was thought to be
high mean stress, low amplitude (striationless) fatigue, although this is
difficult to distinguish objectively from SCC in the absence of striations
near the origin.
Occasionally it is necessary to convince someone that a particular crack
is not only (a) a fatigue crack, but can further be classified, (b) as a
normal fatigue crack or a shear fatigue crack, as the case may be. Normal
fatigue is caused by stresses applied perpendicular to the crack plane;
shear fatigue is caused by stresses parallel to the crack plane, either
parallel to the crack front (transverse shear) or as in this case parallel to
the crack propagation direction (forward shear). Two examples serve to
demonstrate how we have approached this kind of problem. The first
example is a pretty obvious normal fatigue crack, which the submitter had
circumstantial reason to believe might have been caused by shear fatigue.
The crack, whose fracture surface appears in Fig. 10, was found in the
main bulkhead of a late model Navy jet. The material is 2024-T4. Note the
fatigue striations in an area near the origin, Fig. 10(a), and the striations
with tire tracks, Fig. 10(b). Tire tracks are caused by small hard particles,
retained by one crack surface, periodically denting the opposite surface.
The dent or tire-track spacing is an indication of the relative shear dis-
placement of the opposing crack surfaces with each successive cycle. If
shear stress components are high, tire tracks will be absent or very widely
spaced. Their existence proves that the crack opened and closed with very
little permanent shear displacement per cycle, thus that stresses were
mostly normal to the crack plane.
An example of a probable shear fatigue crack was found in the flap
torque tube fitting, of Ti-4AI-4Mn, from yet another late model Navy air-
craft. The fracture, Fig. 11, shows no striations, but abundant evidence of

FIG. 8--Turbojet engine compressor blade (Ti-SAl-lMo-1 V) fi'acture (SEMs). (a) Cleavage
with pits or voids near origin of Jracture half A. (b) Apparent dimples, same area as (a).

FIG. 9--Turbojet engine compressor blade (Ti-SAI-IMo-1 V), (a) Demonstrating clean, pit-
free cleavage near origin of fracture half B (SEM). (b) Coarse striations near end of slow
crack area (two-stage replica, TEM).

FIG. lO--Jet aircraft bulkhead (2024-T3) crack fracture surface (SEMs). (a) Fatigue stria-
tions near origin. (b) Coarse fatigue striations with "'tire tracks" near end o f slow crack area.

FIG. 11--Aircraft wing flap actuator fitting ( Ti-4Mn-4Al) crack surface (two-stage replica,
TEAl).

rubbing and burnishing, and evidence of considerable mutual shearing

motion of the surfaces. How does one prove fatigue? Well, one thing is
the very tightness of the crack; it is unlikely to have resulted from overload
in shear. Furthermore, the broken-open fracture was examined metallo-
graphically in cross section. Figure 12(b) shows that the part created by
laboratory overloading (to open the crack) shows only somewhat distorted
grains at the fracture edge, whereas the service crack, Fig. 12(a), has
indications of twinning substructure formation and microcracking under-
lying the fracture surface. These findings, though not conclusive for lack of
laboratory verification of shear fatigue-induced subsurface microstructural
changes, pointed to fatigue as the possible cause of cracking.
The fractographic analysis of an aircraft arresting gear trunnion which
failed during programmed fatigue testing created a mystery which has not
yet been solved. The part, made of 300M steel of 220-ksi yield, showed
alternate smooth (light) and rough (dark) bands on the fracture surface,
Fig. 13, attributable to regular changes in loading conditions during the
test. Although fatigue was known to have caused cracking, no striations
could be found, Fig. 13(b), a distressingly common circumstance with high

FIG. 12--Aircraft wing flap actuator fitting (Ti-4Mn-4Al) crack cross-sections (polished
and etched, light optical micrographs). (a) In-service crack. (b) Laboratory-produced fast
fracture.

strength steels--no mysteries yet. The trouble began when surface chemical
analysis, using both electron microprobe and Auger spectrometry, showed
that copper and cadmium were present on the fracture surface and were
especially concentrated in the rough, dark bands. No significant difference
in microscopic cracking mechanism could be detected between the two
types of zones, except for greater roughness on the dark ones. We could
only speculate that the copper and cadmium were mechanically dragged in
by the continual relative motion of the crack faces and that they were
preferentially retained in the dark bands because they were rougher and
hence acted as better debris traps. The copper came from copper-alloy
bushings in the trunnion, while the whole steel part was cadmium-plated.
No active role in crack initiation or propagation was suspected of either
element.
Although we have specialized in aircraft component failures, other
kinds of components also come our way. The following problem [6] is
mostly straightforward, but has an unusual twist. Large cracks were noted
traversing the 5456 aluminum alloy deck plates of an experimental hydro-
plane, and repair welding proved fruitless, somewhat like chopping off
Hydra's heads one at a time. The cracks were examined and found to be
typical of SCC, being both long transverse and short transverse inter-
granular cracking with typical "mud-crack" deposits, Fig. 14. The 5456
alloy processed to H321 is supposed to be immune to SCC, and this was
proven with stock material to be true. However, a specimen of the actual

0
0
:IJ
-o
-1-

z
I--
-<
Cl~

rn
0
-1-
z

c
FIG. 13--Carrier aircraft arresting hook trunnion bearing (300M steel)fracture surface (SEMs). (a) General view o f m
slow crack zone showing concentric bands f r o m programmed fatigue load variations. (b) Typical striationless rough
cleavage in slow crack zone.
01

"11
~J

0
:D

-o
I
-<

"n
r-
e
2E
rn

FIG. 14--Small high speed ship deck plate (5456-H321) cracks. (a) Cross sections ~oolished and etched, light optical micrographs). (b)
Mud crack appearance of fracture surfaces (two-stage replica, TEMs).

deck plate, which had seen service, cracked readily. The reason became
evident on comparing the microstructure of stock 5456-H321 plate with the
"used" deck plate, Fig. 15. The deck plate had become further aged and
thus "sensitized" to SCC by tropical sun exposure. The obvious moral to
be drawn from all this is to avoid using materials which literally change
from good to bad under one's feet.
Those who work with advanced structural materials tend to forget that
more ordinary materials also cause expensive problems. Many Navy struc-
tural applications make use of plain hot rolled galvanized steel tubing,
probably the same material used for backyard fences, certainly not exotic,
but indispensable for such things as antenna frameworks and mounting
supports. The state of the art in producing welded structures of this steel is
well established, it is tolerant of error, yet look what can happen. Figure 16
shows the corner section of an antenna support framework, with a fracture
surface where a sound weld should have been. A large proportion of this
and up to 75 percent of other fractures on the assembly were perfectly
coated with zinc, which was applied after welding by hot dipping. The
zinc-free areas showed fatigue striations, Fig. 17(a), and the zinc-coated
areas, after inhibited acid stripping, showed a rather corroded looking
appearance, Fig. 17(b). It is evident that final in-service failure was caused
by fatigue, but the original cracks which weakened the structure obviously
formed after welding, either before hot zinc coating or during coating. The

FIG. 15--Small high speed ship deck plate (5456-I-1321) material (polished and etched,
light optical micrographs). (a) Piece of plate from stock. (b) Piecefrom deck of ship showing
over-aging.

FIG. 16--Radio antenna support f r a m e (galvanized hot-rolled low-carbon steel). Arrows

p o m t to cracks in box section emanating f r o m weld area.

latter seems quite possible, since liquid zinc cracking is a very well known
phenomenon and could be expected to occur if the residual stresses caused
by welding were not relieved before dipping. Further, the zinc-coated frac-
ture surface areas showed no signs of dimples to be expected of residual
stress cracks.
The final example of failure analysis is rather unusual. High power radio
transmission generates considerable induced radio frequency (RF) current
flow in nearby metal structures, including the supporting structures of the
antennas. Damage often occurs at points of intermittent or poor electrical
contact between such things as elevator cables and their pulleys, couplings,
joints, etc. In two cases antenna tower elevator cables, which are made of
cold-drawn high carbon steels, were severely damaged by what are very
similar to welding arc strikes, Fig. 18. The damage consists of brittle, un-
tempered martensite formation, pits, extensive metal removal, and cracks,
Fig. 19(a). The fractured ends of individual strands exhibited brittle
cleavage and intergranular cracking, Fig. 19(b). Very little can be done
about such damage except to keep all affected parts clean, and frequently
inspect and remove damaged components.

FIG. 17--Radio antenna support frame (galvanized hot-rolled low carbon steel) fracture
surfaces (two-stage replicas, TEMs). (a) Fatigue striations in slow crack zone not coated by
zinc. (b) Corroded surface with possible grain boundary fracture in originally zinc-coated
area stripped with inhibited acid.

-11
-n
0
0
0
2o

"1"
.<

I--
c=
-n
m

z
I'--
-<

FIG. 1 8 - - R a d i o transmission tower elevator cable R F arc damage.

C)
0
G)
]3
})
-o
-1-

z
r-
..<
co
.-t
m
0
z
z

FIG. 19--Radio transmission tower elevator cable R F arc damage. (a) Polished and etched cross-section showing f-
ro
ffl
untempered rnartensite formation and cracks (light optical micrograph). (b) Fracture surface o f broken cable wire.
Note cleavage in untempered martensite at arrow, decrease in grain size with distance f r o m surface.
",4

Acknowledgments
The material used in this paper is the product of many others, and it is
with gratitude that the author thanks the following for their contributions:
C. D. Beachem, E. P. Dahlberg, and B. F. Brown, who showed the way;
T. C. Lupton, J. E. Flint (deceased), and W. S. Kenton, who paved the
road; J. DeVault and H. C. Wade, the current bricklayers; C. T. Fujii, a
silent co-author; and our sponsors through many years, the Office of Naval
Research, Naval Air Systems Command, and Naval Facilities Engineering
Command.

References
[1] "Fractography and Atlas of Fractographs," Metals Handbook, 8th ed., American Society
for Metals, Vol. 9, 1974.
[2] Dahlberg, E. P., "An Electron Microscope Study of Crack Surfaces in A Tungsten Rocket
Nozzle Insert," U.S. Naval Research Laboratory Memorandum Report 1217, Sept. 1961.
[3] Meussner, R. A. and Goode, R. J., "An Examination of Three Tungsten Rocket Nozzle
Insert Failures," U.S. Naval Research Laboratory Memorandum Report 1204, Aug. 1961.
[4] Beachem, C. D. and Kenton, W. S., "The Fractographic Study of Fatigue and Monotonic
Fracture Surfaces in Spot Welds in 7075-T6 Aluminum Alloy," U.S. Naval Research
Laboratory Memorandum Report 1545, July 1964.
[5] Meyn, D. A., Dahlberg, E. P., and Beachem, C. D., "Analysis of Stress Corrosion Crack-
ing of Ti-6AI-4V Fuel Tank Material in Methyl Alcohol," U.S. Naval Research Laboratory
Memorandum Report 1744, Jan. 1967.
16l Fujii, C. T., Beachem, C. D., Meyn, D. A., and Brown, B. F., "Study of the Fracture
Mechanism of 5456-H321 Aluminum Alloy," U.S. Naval Research Laboratory Memoran-
dum Report 2422, April 1972.

Fractographic Method of Evaluation of

the Cyclic Stress Amplitude in Fatigue
Failure Analysis*

REFERENCE: Madeyski, A. and Albertin, L., "Fractographle Method of Evaluation

of the Cyclic Stress Amplitude in Fatigue Failure Analysis," Fractography in Failure
Analysis, A S T M STP 645, B. M. Strauss and W. H. Cullen, Jr., Eds., American
Society for Testing and Materials, 1978, pp. 73-83.

ABSTRACT: The first part of this paper describes laboratory work results substan-
tiating the correlation of fatigue striation spacings with crack growth rate and with
cyclic stress intensity amplitude, AK. The results were obtained by transmission elec-
tron microscope (TEM) replica fractography of a fracture mechanics type wedge-open-
ing-loading (WOL) steel specimen previously tested in fatigue.
In the second part of this paper, an illustration of a practical application of this frac-
tographic approach is described in detail. The laboratory test results from the first part
are applied successfully to the service failure analysis of a crankshaft. Cyclic stresses
calculated using a combined fractographic-fracture mechanics method were in good
agreement with the values obtained from a theoretical and experimental stress analy-
sis.

KEY WORDS: metal surface, fractures (materials), fractography, failure, transmission

electron microscopes, fatigue (materials), striations, crack growth rate, stress intensity,
stresses, stress analysis

As o u r knowledge a n d u n d e r s t a n d i n g o f f r a c t u r e p h e n o m e n a improves,
so does t h e a m o u n t o f i n f o r m a t i o n which we can o b t a i n f r o m a failure
analysis o f metallic c o m p o n e n t s . In the past, an investigator usually was
satisfied with identifying the type o f failure (for e x a m p l e , fatigue), locating
the f r a c t u r e origin, d e t e r m i n i n g t h e m a t e r i a l c o m p o s i t i o n , m i c r o s t r u c t u r e ,
a n d p r o p e r t i e s , a n d p r o v i d i n g a g e n e r a l e x p l a n a t i o n for t h e failure.
M o r e recently, however, it was realized t h a t t h e r e is a definite relation
between fatigue striation spacings, S, a n d t h e cyclic stress intensity a m p l i -
t u d e , A K . This r e l a t i o n s h i p , in t u r n , m a k e s possible the calculation o f the

*Original experimental data were measured in U.S. customary units.

1Senior engineers, Materials Evaluation and Application Department, Westinghouse Re-
search and Development Center, Pittsburgh, Pa. 15235.

73
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 1978 by ASTM International [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
74 FRACTOGRAPHY IN FAILURE ANALYSIS

cyclic stress amplitude Ao responsible for the crack advance. The great
practical importance of knowing the fracture stress amplitude in the com-
ponent failure is obvious.
The following is a contribution towards a better knowledge of the cor-
relation between the fatigue striation spacings, S, the crack growth rate,
da/dN, and the stress intensity amplitude, AK. The equation relating S to
AK then is applied successfully to a case of a broken crankshaft.

Correlation of Fatigue Striation Spacing S With Crack Growth Rate

d a / d N and Cyclic Stress Intensity Factor Amplitude, AK
The experimental part of this investigation involved fatigue testing of a
2T (that is, 50.8 mm (2 in.) thick) wedge-opening-loading (WOL) speci-
men, followed by fractography. The specimen configuration and dimen-
sions are shown in Fig. 1. The material was a 3.5Ni-I.5Cr-O.5Mo-O.1V
steel. The specimen was tested at 10 Hz, in 24~ (75~ air, with a con-
stant load range AP = 62.3 kN (14 000 lb), that is, with Pmin = 4.45 kN
(1000 lb) and Pm~ = 66.7 kN (15 000 lb).
The arrangement of the transmission electron microscope (TEM) replica
grids on plastic tape on the fracture surface of the specimen is shown in
Fig. 2. Replicas B to M were examined in the TEM. The location of the
replicas in terms of the crack depth, a, and stress intensity factor ampli-
tude, AK, is shown in Fig. 3. About 15 fatigue striation areas from each
replica were photographed and analyzed. Examples of striations near the
beginning of the fracture (Replica B) and close to the overload fracture
area (Replica M) are shown in Figs. 4 and 5, respectively. The striation
spacings in Fig. 4 are 0.1 #m (4.7/An.), whereas those in Fig. 5 are 1.8/~m
(71.5 #in.), although the average spacing for the Replica M was 1.5/~m
(57.4/An.). It is interesting to note that despite a fast crack propagation in
the region of the Replica M, the striations were still very distinct and uni-
form; however, between the fatigue striation islands there were large areas
of dimpled rupture. This duplex mode of the fracture propagation is well
indicated both in Fig. 2 and in Fig. 6. Figure 2 shows the roughness of the
fracture which starts beyond the Replica K, whereas Fig. 6 clearly displays
the large disparity between the fatigue striation spacing and the measured
crack growth rate per cycle d a / d N for Replicas L and M. In contrast,
Replicas B to K show an excellent agreement between the fatigue striation
spacings and da/dN. The fracture appearance, as seen in the TEM, also
indicated that striated fatigue was the only mode of the crack propagation
in Replicas B to K.
Figure 7 shows the striation spacings S versus AK plotted on double
logarithmic paper, so that the data could be compared with the Bates-
Clark relation 2
2Bates, R. C. and Clark, W. G., A S M Transactions Quarterly, Vol. 62, June 1969.

DW9, 61IDA 1/

__1 !~ w

Relative Dimensions
W : 2.55B
a = [Link] ~,[ H

D = 0.70 B
WI= 3"29 B L ~'-- Thds, T
T = 0.625 B ___ Wl _ _

Side grooves - 0.05 B deep, 45~ included angle,

0.2M mm ([Link] in.) root radius

F I G . I - - W O L type compact fracture toughness specimen (relative dimensions in terms of

thickness B).

FIG. 2--Location o f replica grids on tape on the specimen fracture.

Stress Intensity Range AK - M Pa ' , ~

30 40 50 60 70 80 90 100 llO 12o

d.O " I " I " I v I ' I I I I I
lOO
M

K L

90
3.5

.E I
I
80 ~
2~
3.0

7O ~-

~ 2.5

} x:

2.0
50

40
1.5 l I , I , I m I * I I I I I i
20 30 40 50 60 70 80 90 tO0 110
Stress Intensity Range •K - ksi 4 " ~

FIG. 3--Replica locations with respect to crack depth and stress intensity range.

S = 6 (1)

where
S = fatigue striation spacing,
AK = stress intensity factor peak-to-peak amplitude Kmax - Kin= in the
given cycle, and
E = Young's modulus of the metal.
Each vertical bar represents the total range of the averages of striations
spacings from each small group of parallel striations, that is, the range
from Sm~ to Sm,x observed for the given replica. The average value of S for
a complete replica (marked by an open point in Fig. 7) was calculated from
the average spacings of at least 15 groups, randomly distributed over the
whole area of the replica.

FIG. 4--Replica B fatigue striations.

FIG. 5--Replica M fatigue striations.

Stres~ lntensi~.y Range AK - ~ Pav"~

30 40 50 60 ?0 80 99 100 110 120
280

260
~ L0

6.5

240
- 6.0

._= 220 5.5 E

:t :x
I i
t~
200 5.0

180
o= ~4.5

da ~4.0
Tff J
o

~, 140 u
3.5

7 i
z 120 3.0 z

a~
g~ 2.5

2.0

M c)
o 1.5

40 , 1.0

I
20 H 0.5

20 30 40 50 60 70 80 90 100 110
Stress Intensity Range AK - ksi

FIG. 6--Comparison of fatigue striation spacings (points on the diagram) with crack
growth rate (continuous l i n e ) .

The statistical distribution of the group averages for each replica was
found to be close to normal. Good correlation between the experimental
data and Eq 1, particularly in the early part of the crack propagation, is
clearly visible in Fig. 7. Thus, this laboratory work provided additional
supporting evidence for the validity of the Bates-Clark equation.

Application of the Bates-Clark Equation to Failure Analysis

The following example illustrates the application of the Bates-Clark
equation to the stress analysis of a broken ductile iron crankshaft.
The average chemical composition of the crankshaft was 3.5C, 0.40Mn,
2.40Si, 1.15Ni, and 0.06Mg. The crankshaft had been heat treated by aus-

Stress Intensity Range ~K - M Pa v'm

20 30 40 60 80 100 150
100 I I ! I I'r I It
80 2. O0
60 M 1.60

~
50 1.20
40 1.00
30 K L O.80
.c: 0.60 E
I 20 O.50 I

~
u',

"~ lO G
O.40

0.30~.~
8 F 0.20 o
E "v-

5
4 0. I 0 ~

/,/-,--s=6C~)~__ o.o~

I I I I I l III
10 20 30 40 50 60 80 100 150
Stress Intensity Range AK - kst dTn

F I G . 7--Comparison of fatigue striation spacings from fractography (points on the dia-

gram) with the Bates-Clark equation (continuous line).

tenitizing at 885~ (1625~ quenching in oil, and tempering to produce

the following average mechanical properties:
brinell hardness = 250,
tensile strength = 827 MPa (120 ksi),
yield strength (0.2 percent offset) = 655 MPa (95 ksi),
endurance limit (estimated as 35 percent of the tensile strength) = 283
MPa (41 ksi), and
elongation in 50.8 mm (2 in.) = 5 percent.
The crankshaft broke through a crank, next to a pin journal, as shown in
Fig. 8. The origin of the fracture was in the transition zone between a
straight and a filleted section of the crank, where two surface machining
operations formed a junction resulting in a step. Scanning electron
microscope examination of the origin revealed no material defects which
could have been responsible for the initiation of the crack.
The general topography of the fracture was typical of low stress-high
cycle fatigue. In order to determine the fatigue striation spacings, a two-
stage plastic-carbon (platinum shadowed) replica technique was used. The
average striation spacing near the origin was 0.1 #m (4.7 #in.), with a

FIG. 8--Crankshaft fracture.

range of 0.08 to 0.18/~m (3.2 to 7.2 tzin.). Further along the fracture the
spacings reached 0.76 ~m (30 tdn.).
This information was used to calculate the stress intensity factor ampli-
tude ~ from the Bates-Clark Eq 1
/ S \ v2

Substituting the value of S = 4.7 • 0.1 ~m (10 -6 in.), and the Young's
modulus E = 165.5 x 10 3. MPa (24 x 103 ksi) for ductile iron, we
obtained AK = 23.3 MPa ~ (21 ksi x/~-.) near the origin of the fracture.
In order to apply the linear fracture mechanics methods to the calcula-
tion of the stress amplitude which caused the cracking, the following sim-
plifying assumptions were made: (a) the shape of the initial crack was
semielliptical, (b) the applied load was in bending only, and (c) the stress
was essentially alternating, that is, the mean stress was practically zero.
The stress intensity factor K for a surface crack in a plate subjected to
bending may be estimated using the results presented by Smith. 3

K = Msoa/~Q (3)
3Smith, F. W., "Stress Intensity Factors for a Semi-Elliptical Surface Flaw," Structural
Development Research Memorandum No. 17, The Boeing Company, Aug. 1966.

where
MB = crack depth factor,
a = stress, MPa,
a = crack depth, millimetres, and
Q = crack shape parameter.
The semielliptical surface crack at the fracture origin was well marked by
an arrest line at a = 2.5 m m (0.1 in.), so that the ratio of the crack depth
a to its length 2C could be easily determined. This ratio was found to be
equal to 0.2, for which Q = 1.3 (Fig. 9). The factor M a depends both on
a/2C and on a/t, where t is the total thickness of the plate (crank).
In this particular case there was MB = 0.94 (Fig. 10).
Since the stress intensity amplitude AK = Kma. - Kmm, and the
nominal stress amplitude

AO = O'max - O'min

then from Eq 3

AK = M , Aof~vr~Q

so that

AK
Aa = MBx/-~Q

Substituting the previously found numerical values of the parameters, the

stress peak-to-peak amplitude was found to be Aa = 317 MPa (46 ksi),
that is, 159 MPa ( _+23 ksi).

o.3o~ ////

0 lOI- / / / / r = CompleteEllipticalIntegral
9 l / / / / of ~heS~on~Ki~

l.O 1.5 2.0 2.5

F I G . 9--Crack shape parameter curves (after Tiffany and Masters 4).

4Tiffany, C. F. a n d Masters, I. N. in Fracture Toughness Testing and its Application,
ASTM STP 381, American Societyfor Testing and Materials, 1965, pp. 249-277.

1.6 I , i ~
--l~a a/2C=0

1.2 ~

--- Estimate "- ~ ~

o =Mr g=O
.2 21

,0 ~ i I . ~ ~ -J
.0 .l .2 .3 .4 .5
a~
FIG. lO--Approximate stress intensity factors for semielliptical surface cracks in bending
at c~ : 0 (after Smith (see footnote 3)).

Finite element stress analysis indicated an alternating stress of 193 MPa

(_+ 28 ksi). Actual stress measurements using strain gages were also per-
formed on a similar crankshaft tested under simulated service conditions.
The stress measured at a location corresponding to the crack origin gave
an alternating stress 103 MPa (15 ksi) in tension and 138 MPa (20 ksi) in
compression.
Thus, a good agreement was found between the stress amplitude calcu-
lated from fractography using the Bates-Clark equation, and the stresses
determined both experimentally with strain gates, and analytically using
the finite element technique.

DISCUSSION

S u s u m u Hioki 1 (written discussion)--I quantitatively analyzed fatigue

machine parts using fatigue striations and fracture mechanics and com-
pared the calculated value with finite element method (FEM) results or
measured stresses actually. I found approximately _+10 percent errors
between the value from fracture surface and FEM or measuring results.
[Link] is what is your opinion and your data on that problem?
1Doctor of Engineering, Merl, Hitachi, Ltd., Hitachi-Tsuchiura, Japan.

A. Madeyski and L. Albertin (authors' closure)--A discrepancy between

the measured striation spacings and the calculated crack growth rate or the
cyclic stress amplitude may result from many causes. For example, the
crack extension is often measured microscopically on the sides of the speci-
men, whereas the crack tip may be in form of an arc, so that most of the
crack front is actually further ahead than the reported length. Unless this
is taken into account, which is not always possible, most of the correspond-
ing fatigue striations, in fact, will be located further along the specimen
than considered in the calculations. When other methods of crack length
measurement are applied, utilizing, for instance, ultrasonics or electric
resistance changes, the results are usually average values over the whole
crack front. Furthermore, there are limitations to the spatial
resolutions shown by these methods. The measured values of the crack
length are subsequently used to calculate the crack growth rates, which
again usually involves some averaging and "smoothing" errors. Thus, the
reported crack growth rates are inherently of somewhat limited accuracy.
Similarly, fatigue striation spacings may also vary widely from point to
point, even within a small area of the fracture. Their individual values are
governed by the local properties of the material and by the microstress
intensities at each specific spot, rather than by the calculated macrostress
of that region.
Despite these variations, the average values of the striation spacings and
the corresponding crack growth rates or the macrostress intensities in any
region of a specimen fracture are usually in reasonably good agreement, as
shown in Figs. 5 and 6. However, in view of the variations just discussed,
the errors of _+10 percent found by Hioki should be considered quite rea-
sonable and not excessive.

Fractographic Analysis of Gaseous

Hydrogen Induced Cracking in 1 8Ni
Maraging Steel

REFERENCE: Gangloff, R. P. and Wei, R. P., "Fraetographle Analysis of Gaseous

Hydrogen Induced Craeklng in 18Ni Maraging Steel," Fractography in Failure Analy-
sis, A S T M STP 645, B. M. Strauss and W. H. Cullen, Jr., Eds., American Society
for Testing and Materials, 1978, pp. 87-106.

ABSTRACT- Fracture surfaces, produced by sustained load gaseous hydrogen assisted

cracking of 18Ni maraging steel, were examined by scanning electron microscopy.
Cracking developed along boundaries associated with the maraging steel micro-
structure. The specific crack path depended on test temperature and correlated with
the influence of this variable on crack growth rates. At low temperatures, crack
growth proceeded predominantly along prior austenite grain boundaries. Increasing
temperature produced a continuously increasing amount of transgranular quasicleav-
age associated with lath martensite boundaries. The proportions of quasicleavage
fracture correlated with temperature induced reductions in the crack rate over those
predicted from low temperature Arrhenius behavior. Both reduced hydrogen pressure
and yield strength decreased the temperature for the onset of the transition to trans-
granular fracture but had no influence on the crack path. The lower strength steel
fracture morphology contained an increased proportion of features typical of ductile
rupture. The crack path through the microstructure was independent of the applied
stress intensity factor, and hence crack growth rate, from very low values to those
beyond the Stage I and Stage II transition for all temperature conditions. Com-
parison between fractographic observations and known sites for hydrogen segrega-
tion suggested that microstructural features play a significant role in the mechanism
for gaseous hydrogen embrittlement.

KEY WORDS: fractography, crack propagation, hydrogen embrittlement, alloy steels,

fractures (materials)

H i g h - s t r e n g t h steels c a n b e severely e m b r i t t l e d by h y d r o g e n d i s s o l v e d in
t h e m i c r o s t r u c t u r e ( i n t e r n a l h y d r o g e n e m b r i t t l e m e n t ) , by e x p o s u r e to ex-
ternal hydrogen producing environments (hydrogen environment embrittle-

1Metallurgist, General Electric Corporate Research and Development Center, Schenectady,

N.Y. 12301.
2 Professor of Mechanics, Department of Mechanical Engineering and Mechanics, Lehigh
University, Bethlehem, Pa. 18015.

87
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 1978 by ASTM International [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
88 FRACTOGRAPHY IN FAILURE ANALYSIS

ment), and by exposure to gaseous hydrogen over a wide range of pres-

sures [1-11]. 3 Aside from generic or phenomenological distinctions, there
may not be any fundamental differences between these forms of embrittle-
merit. Embrittlement is typically manifested by changes in the fracture sur-
face morphology as compared to that produced in an inert environment.
As such, fractographic characterization of surfaces produced by well docu-
mented laboratory experiments can be used to aid in identification of
causes for service failures and can provide insight into the mechanisms of
embrittlement. Systematic fractographic analyses of hydrogen embrittled
high strength steels are few and do not document the dependence of crack
path on interactive effects between critical variables for gaseous hydrogen
embrittlement.
Gaseous hydrogen and hydrogen producing environments generally pro-
duce an intergranular 4 (along prior austenite grain boundaries) crack path
in both 18Ni maraging [13-15] and quenched and tempered alloy steels
[6,8,16]. A small percentage of transgranular cleavage or quasicleavage
typically accompanies intergranular separation [6,13]. An increase in the
stress intensity factor was alternately reported to: (a) increase the amount
of ductile tearing that coexisted with intergranular fracture [8,16,17], (b)
produce a transition in fracture morphology from intergranular separation
to a "rough, ductile" mode [7], or (c) produce a transition from intergran-
ular separation to transgranular cleavage (or quasicleavage) and sub-
sequently to microvoid coalescence [I8,I9]. Composition [20] and
tempering temperature [13,21] were observed to alter the fracture mor-
phology from intergranular separation to transgranular cleavage (or quasi-
cleavage) for high strength quenched and tempered alloy steels. A correlation
[21] was established between increasing yield strength and a transition
in crack path from transgranular cleavage to intergranular separation.
This correlation, however, does not appear to apply to maraging steel.
These steels were embrittled along transgranular boundaries in the
optimally aged condition, while intergranular separation was observed for a
lower strength, underaged form of the alloy [5].
A recent study [4,22] established that temperature, hydrogen pressure,
stress intensity factor, and material yield strength interact to establish the
kinetics for sustained load gaseous hydrogen assisted crack propagation in
18Ni maraging steel. Results from the complementary fractographie
analysis are reported here. The objective of this analysis was to define frac-
ture surface characteristics associated with gaseous hydrogen embrittle-
ment as influenced by a range of chemical, mechanical, and material
variables.
3The italic numbers in brackets refer to the list of references appended to this paper.
4Fracture mechanisms are characterized as cleavage, quasicleavage, intergranular separa-
tion, or microvoid coalescence [12]. Cleavage is defined as bond separation along a crys-
tallographic plane. Quasicleavage is assumed to describe separation along interfaces
associated with transgranular microstructural features.

Experimental Procedure

Material, Specimen, and Test Procedure

A 0.60-cm-thick plate of 18Ni(200) maraging steel and a 0.47-cm-thick
plate of 18Ni(250) maraging steel were selected for study. Chemical compo-
sition, heat treatment, and mechanical properties for each steel are given
in Table 1. Prior austenite grain size equaled 15 #m for the !8Ni(200)
maraging steel, and 20/~m for the higher strength 250-grade material. The
reverted austenite content of each steel was less than 5 volume percent
[231.
Fatigue precracked wedge-opening-load (WOL) specimens, machined in
the longitudinal (LT) orientation, were tested in purified hydrogen gas
under sustained load to develop data on crack growth kinetics. This test
method provided increasing stress intensity with increasing crack length. A
range of known K levels and crack speeds was obtained from a single
specimen. Experimental procedures and test data are given elsewhere
[4,22].

Fractographic Procedure
Fracture surfaces were examined with a scanning electron microscope
(SEM) in the secondary mode at 20 kV. Working distance was 11 mm,
and all specimens were tilted 20 deg about an axis parallel to the direction
of crack growth. Morphological features normal to this axis were either
foreshortened or lengthened due to specimen tilt. Features parallel to the
direction of cracking were undistorted. The complete fractured specimen
half was examined, thus eliminating artifacts associated with sample sec-
tioning. All fractographs were obtained from areas near the midthickness
region of each specimen. Distances from the fatigue crack tip to areas of
interest were measured with a micrometer stage; corresponding K and
crack speed values were determined from growth rate data.

Results and Diseussions

Crack Growth Kinetics Summary

Crack growth in gaseous hydrogen was strongly affected by test tem-
perature and stress intensity level for both 18Ni(200) and 18Ni(250) mar-
aging steels [22]. The crack speed increased rapidly with increasing K over
a low stress intensity range defined as Stage I. Above a critical K level, the
crack speed was independent of K for a wide range of values (Stage II).
The mean Stage II growth rate increased with increasing test temperature
for a low temperature regime defined as Region A. Within an intermediate

-i1

o
O

"D
"1"
-<
TABLE 1--Composition, heat treatment, and mechanical properties.
T=
Chemical Composition, weight % >_
r-
Maraging Steels C P S N Ni Co Mo Ti Fe Heat Treatment a c-m
rrl
18Ni(200) 0.002 0.003 0.007 0.004 17.9 7.5 2.8 0.18 balance 900~ IA h, AC+
815~ 1/2 h, W Q + z
ays = 1270 MN/m2; Outs = 1330 MN/m2; elongation = 12.8% of 5.08 cm 482~ 16 h, WQ t-
18Ni(250) 0.004 0.001 0.001 0.002 17.9 7.2 4.8 0.40 balance 927~ I h, AC+
ays = 1650 MN/m2; outs = 1720 MN/m2; elongation = 10.0% of 2.54 cm 482 ~ 3 h, AC

a AC = air cooled and W Q = water quenched.

temperature range, Region B, Stage II growth rates increased at slower

rates, passed through a maximum, and decreased with increasing tempera-
ture. The crack growth rate decreased with decreasing hydrogen pressure
and material yield strength; however; these variables did not alter the two
stage K and Region A/B temperature dependencies.
Stress intensity and temperature dependencies of the hydrogen
embrittled crack growth rate are presented in ensuing sections with corres-
ponding fractographic results. Two yield strength levels and several
hydrogen pressures are discussed.

Fractographic Observations
Fracture surfaces produced by gaseous hydrogen embrittlement were
macroscopically flat without shear wails for all K range and material
strength levels investigated as illustrated in Fig. 1. Fractures produced by
Stage II cracking in Region A were characterized as the baseline condition
for each steel. The effect of stress intensity was defined, and the potential
for a fracture morphology change corresponding to the Stage I to II tran-
sition was evaluated. The effect of temperature on the Stage II growth rate
suggests that there may be a corresponding effect on the crack path; this
possibility was investigated.
Fracture Morphology for Stage H Cracking in Region A--Stage II crack-
ing in both 18Ni(200) and 18Ni(250) maraging steels was predominantly
intergranular along prior austenite grain boundaries for all Region A tem-
perature and hydrogen pressure conditions. The crack morphology was de-
fined by five components, Figs. 2 and 3:
1. A predominant component of intergranular separation along prior
austenite grain boundaries
2. Grain facet markings and boundary phase cracking
3. Microcracks along prior austenite grain boundaries out of the macro-
scopic plane of fracture (that is, secondary cracks)
4. A small component of transgranular quasicleavage within prior aus-
tenite grains
5. A small amount of ductile tearing
Comparison of Components 1 through 4 with the characteristics of mar-
aging steel fracture surfaces produced in an inert environment indicates
that hydrogen embrittlement produced a fracture mode transition. The
18Ni maraging steels fail by microvoid nucleation, growth, and coalescence
when stressed in a nonaggressive environment [24]. Transgranular cleavage
was not observed for these materials impact loaded in inert environment
for the temperature range under study [25].
The 18Ni maraging steels fractured in hydrogen producing environments
exhibited a predominantly intergranular fracture morphology with a small

FIG. 1--Macrophotographs of 18Ni(250) maraging steel fracture surfaces produced by sus-

tained load cracking in gaseous hydrogen.

z
o

m_
0
Z
-n

0
-t
0
0

"13
I

Z
r'-

FIG. 2--Scanning electron fractographs o f Stage II, Region A hydrogen cracking in 18Ni(250) maraging steel. (P//2 = 133
k N / m 2, T = 228 K, K = 23 M N / m 3/2.) (The arrow indicates the macroscopic direction o f crack growth in each figure. )
tD
6O

amount of transgranular quasicleavage and microcracking [5,13-15,26].

The susceptibility of prior austenite boundaries to hydrogen embrittlement
is well documented [27-30]. The component defined as "quasicleavage"
was correlated [5] with hydrogen induced cracking along lath martensite
boundaries. This separation process, while clearly associated with the
microstructure of the iron-nickel martensite, may be complex. The marten-
sitic transformation for maraging steels produces a lath morphology with
individual units forming on { 111 }A habit planes in multiple packets within
a parent austenite grain [31]. While either the low angle lath interface
[31,32] or the incoherent packet boundary [31,32] could be embrittled by
hydrogen, the latter boundary is probably most susceptible [29]. This inter-
face provides an obstacle to dislocation motion and was identified as a site
for hydrogen segregation through a trapping mechanism [33,34]. Evidence
[32] suggests that inert environment transpacket cleavage occurs through
parallel laths of an iron-nickel martensite packet. Since this fracture mode
was not observed for maraging steel [24,25], it is unlikely that this mecha-
nism is operative during hydrogen assisted cracking.
The 18Ni(200) maraging steel fracture surface included a sixth mor-
phological feature that was not observed for the 250-grade steel. This
component, illustrated by Feature A in Fig. 3, was indicative of me-
chanical rupture over several grain diameters. Apparently, hydrogen is less

FIG. 3--Fractograph of Stage II, Region A gaseous hydrogen cracking in 18Ni(200) mar-
aging steel. (P tt2 = 133 kN/m 2 , T = 217 K, K = 44 MN/m 3/2 .)

effective in embrittling prior austenite and lath martensite boundaries in

the lower strength maraging steel.
Effect of Stress Intensity Factor for Cracking in Region A--Fractographic
results for each grade of 18Ni maraging steel, cracked in Region A at
several hydrogen pressures, indicate that increasing stress intensity within
Stages I and II had no influence on the hydrogen assisted intergranular
crack path. Fractographs supporting this conclusion are contained in Figs.
4 and 5 in conjunction with corresponding crack growth rate data [22].
The intergranular fracture surface morphology was unaffected by increas-
ing stress intensity from the fatigue precrack tip, through the Stage I to II
growth rate transition, to K levels well within the growth rate plateau for
18Ni(250) maraging steel, Fig. 4. Similar fractographs characterizing crack
growth in 18Ni(200) maraging steel are presented in Fig. 5. These results
describe a limited K range. Additional observations were made and indi-
cated that the fracture morphology was similar for stress intensity values
between 33 and 67 MN/m 3/2.
Results presented in Figs. 4 and 5 are not consistent with those studies

FIG. 4--The influence of stress intensity Jactor on the Region A fracture surJace mor-
phology produced by gaseous hydrogen embrittlement (PH2 = 133 kN/m 2, T = 228 K) of 18
Ni(250) maraging steel. Fracture surface features are numbered as described for Fig. 2.

"11
:lJ

0-I
0
0

-0
"i"
.<

"11
>_
f-
C
20
m

z
r-
GO
O~

F I G . 5 - - T h e influence o f stress intensity factor on the Region A fracture surface morphology produced in 18Ni(200) maraging
steel, (PH 2 = 133 k N / m 2, T = 217 K). Fracture surface features are numbered as described f o r Fig. 2.

that reported a fracture mode transition at the Stage I to II growth rate

transition [7,18,19]. The maximum K level investigated for Region A
cracking was equal to about 40 percent of the estimated critical stress in-
tensity value for fast fracture. Increased ductile tearing would be expected
for higher K levels [8,16,17].
Examination of the interface between the fatigue precrack and the initial
portion of the hydrogen assisted crack revealed the absence of a resolvable
"stretch" zone [20]. A typical fractograph that supports this conclusion is
contained in Fig. 4. While a micron size stretch zone may be a prere-
quisite to initiate hydrogen embrittlement in lower strength steels [20],
there is no evidence to suggest that this process was essential for cracking
of high strength steels such as the 18Ni maraging variety.
Effect of Test Temperature for Stage II CrackingkTest temperature in-
fluenced the fracture surface morphology for hydrogen assisted Stage II
cracking in 18Ni maraging steel, consistent with the influence of this vari-
able on the Stage II growth rate [22], Comparison between fracture sur-
faces of 18Ni(250) maraging steel, produced by cracking in 133 kN/m z
hydrogen (H2) at the temperature extremes of Regions A and B, indicates
that a predominantly transgranular fracture morphology with respect to
prior austenite grains was produced at Region B temperatures as illustrated
in Fig. 6. The Region B fracture morphology is, in contrast to Region A
results, defined by four components:

1. A small amount of intergranular separation, with attendant grain sur-

face markings
2. A predominant component of transgranular quasicleavage associated
with lath martensite platelet or packet interfaces
3. A component of mechanically ruptured material
4. Microcracks of both intergranular and transgranular orientation out
of the macroscopic fracture plane (that is, secondary cracks)

The transition from intergranular to transgranular cracking with increasing

temperature was confirmed by metallographic sections of fractured speci-
mens characterized in Figs. 6(a) and (b). These results are shown in Fig.
7. The cross section through the crack produced in Region A revealed ex-
tensive hydrogen assisted separation of prior austenite boundaries adjacent
to the macroscopic crack plane. Cross sections of the specimen embrittled
at Region B temperatures revealed extensive transpacket lath martensite
cracking along the main crack surface and through adjacent grains, Fig.
7. A small proportion of prior austenite boundary separation was noted.
Fractographic results indicate that the transition from intergranular to
transgranular cracking was continuous with increasing temperature above
the Region A/B boundary. A qualitative correlation was established be-

FIG. 6-- The influence of test temperature o n t h e Stage 11 fracture surface morphology
produced in 18Ni(250) maraging steel at two hydrogen pressures. Features in fractograph
(b) are numbered as described in Fig. 2.

tween the reduction factor s [22] and the amount of transgranular quasi-
cleavage. The fracture morphology transition is complete at the tempera-
ture where reduction factor (RF) = 6.4, Figs. 6(a) and (b). The fracture
path was predominantly intergranular at the Region B temperature where
RF = 1.9 at 57 kN/m 2, Figs. 6(d) and (e). A clear transgranular character
developed at the higher temperature where the reduction factor equaled 3.9,
SThe reduction factor is defined as the ratio between the Stage II growth rate that would be
realized if Region A cracking persisted to the indicated temperature and the actual Stage II,
Region B growth rate.

Fig. 6(c). Analogous results were obtained for the 18Ni(200) maraging steel
as illustrated in Fig. 8. The proportion of transgranular lath martensite
boundary failure increased with increasing temperature as the RF increased
from 1.8 to 3.2.
Temperature was the critical variable that defined the crack path for
hydrogen embrittlement of the 18Ni maraging steels. Degree of crack tip
constraint, magnitude of the crack growth rate, hydrogen pressure, and
stress intensity level had-no influence on the crack path transition illus-
trated in Figs. 6 and 8 as shown by the following considerations. While yield
strength decreased by about 10 percent with increasing temperature from
213 to 336 K, the information contained in Figs. 6 and 8 could be pre-
sented for comparable values of the parameter 1/B(K/ay,) 2 [22]. As such,
it is concluded that the Region A to B transitions in growth rate and frac-
ture surface morphology were not related to decreasing crack tip constraint.
While Region B cracking typically required higher stress intensity levels
[22], data indicate that K p e r se had no effect on the crack path for either
Region A or Region B embrittlement. Fractographs contained in Figs. 6
and 8, while obtained at specific K levels, also described the fracture
morphology over the complete K range studied and parenthetically noted.
Intergranular cracking was produced at relatively high stress intensity
levels for 18Ni(200) maraging steel, Fig. 8. Fractographs contained in Fig.
9 indicate that the Region B transgranular fracture surface was unaffected
by increasing K through the Stage I to II growth rate transition analogous
to Region A results. Comparison qf Fig. 9(a) with Figs. 6(c) through (e)
confirms the existence of the intergranular to transgranular crack path
transition at a constant K level of 41 MN/m a/2. While cracking in Region
B was associated with higher crack speeds, crack growth rate had no in-
fluence on the crack morphology. Comparison between Figs. 6(/5) and (c)
indicates that the fracture path transition was observed at near constant
crack growth rate. The Region B transgranular crack morphology shown in
Fig. 9(a) was produced at a slower crack growth rate (4 x 10-6 m/s) than
the speeds for Region A cracking at 133 kN/m 2 (Fig. 6(a)) and at 57
kN/m 2 (Fig. 6(e)). Region B transgranular cracking in 18Ni(200) maraging
steel was produced at slower growth rates than Region A intergranular
cracking in the 250-grade steel, as illustrated by comparison between Figs.
6 and 8. Mclntyre [20] reported that the crack path for hydrogen embrittle-
ment of an alloy steel was unaffected by changing crack growth rate over
four orders of magnitude. This observation is in agreement with current
findings.

Mechanistic Implications
Fundamental understanding of hydrogen embrittlement of high strength
steels remains elusive [4]. Fractographic results, correlated to quantitative

Z
0

m
0
z
"11

"l-

z
i"

FIG. 7--Metallographic cross sections through cracks produced by gaseous hydrogen embrittlement of 18Ni (250) maraging steel at tempera- N
tures in Regions A and B. Etchants: (a) and (c) 10 ml acetic acid + 10 ml nitric acid (HN03) + 15 ml hydrochloric acid (HC1) + glycerine, (b)
10 g chromic acid (Cr03) + 100 ml water (H20) at 6 V, 5s, (d) 95 ml aqueous ferric chloride (FeCI2) + 5 ml HCI.
..&

FIG. 8--The influence of test temperature on the Stage H fracture surface morphology
produced in 18Ni(200) maraging steel at two hydrogen pressures.

crack growth kinetics data, provide some qualitative insight into this prob-
lem.
Microstructural Aspects of Cracking--There is, at present, no evidence
that unequivocally identifies the location of the fracture site for external
hydrogen embrittlement. Hydrogen assisted failure is probably governed by
a critical combination of local tensile stress and hydrogen concentration
[17]. Several models [10] are based on the hypothesized dominant role
of the crack tip stress state, however, current fractographic findings indi-
cate that microstructural features play a significant role in defining the
fracture site. Microautoradiographic studies [33,34] of 18Ni(250) maraging

F I G . 9--The influence o f stress intensity factor on the Region B fracture surface morphology
produced by gaseous hydrogen embrittlement ( P / / 2 = 57 k N / m 2 T = 297 K) o f 18Ni(250)
maraging steel.

steel showed significant segregation of hydrogen at prior austenite grain

boundaries and detectable segregation at interfaces associated with the lath
martensite structure. Hydrogen was segregated to these microstructural
features far in excess of that estimated lattice hydrogen concentration
produced by hydrostatic stress effects [10,17]. Microstructural sites for
hydrogen segregation correlate with the preferred path for cracking in
Region A.
Effects of Stress Intensity--The continuity of the fracture surface mor-
phology through the Stage I to II growth rate transition is consistent with
the hypothesis [22] that a single embrittlement mechanism is operative
over the complete stress intensity range. Stage I reflects mechanical con-
trol of the crack speed, while the Stage II K-independent plateau region is
established by the rate limiting kinetics of the chemical sequence for em-
brittlement. This result impacts models for the temperature-hydrogen pres-
sure dependency of Stage II cracking [4].
Effect of Temperature--Fractographic results indicated that increasing
temperature through the Region A to B growth rate transition produced

a continuously increasing amount of transgranular cracking, presumably

associated with lath martensite interfaces. It was hypothesized [22] that
Region B cracking is produced by a reduction in the total amount of
hydrogen supplied to fracture sites resulting from a temperature induced
decrease in equilibrium surface adsorption of hydrogen. It is further spec-
ulated that, as temperature increases, the balance between hydrogen segre-
gated to prior austenite and lath martensite boundaries changes to favor
the latter. Qualitatively, hydrogen transport to supply the transgranular
cracking mode could be expected to increase compared to diffusion along
prior austenite grain boundaries, the mode presumably operative at lower
temperatures.
Effect of Yield Strength--Invariance of the hydrogen assisted crack mor-
phology with increasing yield strength indicates that the same embritfle-
ment mechanism is operative for each grade of 18Ni maraging steel. The
increased proportion of ductile rupture observed for the lower strength
steel could be related to a reduction in the overall amount of embrittled
material and an increase in the proportion of mechanical rupture between
hydrogen embrittled microcraeks. Moon and Landes [21] reported an inter-
granular to transgranular crack path transition as yield strength decreased
and Kth for cracking in hydrogen sulfide (HzS) increased for an alloy steel.
Results presented in Figs. 6 and 8 indicate that this effect can be explained
on the basis of temperature. Specifically, the temperature for the onset of
Region B cracking decreased with reduced yield strength. Isothermal tests
can intersect different regions of cracking to produce an intergranular, Re-
gion A type crack path in the high strength steel, and a transgranular, Re-
gion B morphology in the lower yield strength material.

Conclusions
Fractographic analysis supplemented an extensive study of the kinetics
of gaseous hydrogen assisted cracking in 18Ni maraging steel. The fol-
lowing conclusions were drawn.
1. Temperature determined the crack path morphology in each steel
which, in turn, was directly related to the temperature dependence of the
crack growth rate. (a) Crack growth in the low temperature (Region A)
regime proceeded along prior austenite grain boundaries. (b) Increased
temperature above the Region A/B transition value produced a continuously
increasing proportion of transgranular quasicleavage associated with lath
martensite boundaries. The amount of transgranular cracking was qualita-
tively correlated to the degree of temperature induced deviation from Arrhe-
nius behavior.
2. Fracture morphologies were independent of stress intensity factor and
crack growth rate through the Stage I to II transition for all temperature
and hydrogen pressure conditions investigated.

3. E a c h g r a d e o f 18Ni m a r a g i n g steel e x h i b i t e d identical f r a c t u r e mor-

phologies for c o m p a r a b l e regions a n d stages o f c r a c k i n g .
F r a c t o g r a p h i c observations were i n t e r p r e t e d in t e r m s o f h y p o t h e s i z e d
m e c h a n i s m s for gaseous h y d r o g e n e m b r i t t l e m e n t . It was c o n c l u d e d t h a t
h y d r o g e n segregation to p r i o r austenite a n d lath m a r t e n s i t e b o u n d a r i e s
m u s t be c o n s i d e r e d as a significant factor in developing m e c h a n i s m s for
gaseous h y d r o g e n e m b r i t t l e m e n t o f high strength steels.

Acknowledgments

T h e a u t h o r s gratefully a c k n o w l e d g e informative discussions with G. W .

S i m m o n s , K. Klier, a n d P. S. Pao. S u p p o r t o f this research b y t h e N a t i o n a l
A e r o n a u t i c s a n d Space A d m i n i s t r a t i o n (Lewis Research Center) u n d e r N A S A
G r a n t N G R 39-007-067, the A m e r i c a n I r o n a n d Steel I n s t i t u t e u n d e r Proj-
ect No. 62-259, a n d t h e N a t i o n a l Science F o u n d a t i o n u n d e r G r a n t D M R
74-10489 is a c k n o w l e d g e d .

References
[I] Jewett, R. P., Walter, R. J., Chandler, W. T., and Frohmberg, R. P., "Hydrogen
Environment Embrittlement of Metals," NASA Report CR-2163, National Aeronautics
and Space Administration, Washington, D.C., 1973.
[2] Walter, R. J. and Chandler, W. T., "Influence of Gaseous Hydrogen on Metals," NASA
Report CR-124410, National Aeronautics and Space Administration, Washington, D.C.,
1973.
[3] Hudak, S. J., Jr., Masters thesis, Lehigh University, 1972.
[4] Gangloff, R. P., Ph.D. thesis, Lehigh University, 1974.
[5] Dautovich, D. P. and FIoreen, S. in Proceedings, International Conference on Stress
Corrosion Cracking and Hydrogen Embrittlement of Iron Based Alloys, J. Hochmann, J.
Slater, and R. W. Staehle, Eds., National Association of Corrosion Engineers, Houston,
Tex. 1974, in press.
[6] Sawicki, V. R., Jr., Ph.D. thesis, Cornell University, 1971.
[7] Kerns, G. E., Ph.D. thesis, Ohio State University, 1973.
[8] Nelson, H. G. and Williams, D. P. in Proceedings, International Conference on Stress
Corrosion Cracking and Hydrogen Embrittlement of Iron Based Alloys, J. Hochmann,
J. Slater, and R. W. Staehle, Eds., National Association of Corrosion Engineers, Houston,
Tex., 1974, in press.
[9] Troiano, A. R. and Fidelle, J. P. in Proceedings, Conference L 'Hydrogene Dans Les
M'etaux, Paris, 1972, p. 31.
[10] Gerberich, W. W., Chen, Y. T., and St. John, C., Metallurgical Transactions, Vol. 6A,
1975, p. 1485.
[11] Johnson, H. H. in Proceedings, International Conference on Stress Corrosion Cracking
and Hydrogen Embrittlement of Iron Based Alloys, J. Hochmann, J. Slater, and R. W.
Staehle, Eds., National Association of Corrosion Engineers, Houston, Tex., 1974, in press.
[12] Beaehem, C. D. and Pelloux, R. M. N. in Fracture Toughness Testing and Its Appli-
cations, ASTM STP 381, American Society for Testing and Materials, 1965, p. 210.
[13] Fidelle, J. P., Legrand, J., and Couderc, C., "A Fractographic Study of Hydrogen Gas
Embrittlement in Steels," Paper No. F71-8, The Metallurgical Society of the American
Institute of Mining, Metallurgical, and Petroleum Engineers, 1971.
[14] Wei, R. P. and Landes, J. D., Materials Research and Standards, Vol. 9, 1969, p. 25.
[15] Stavros, A. J. and Paxton, H. W., Metallurgical Transactions, Vol. 1, 1970, p. 3049.

[16] Nelson, H. G., Williams, D. P., and Tetelman, A. S., Metallurgical Transactions, Vol.
2, 1971, p. 953.
[17] Oriani, R. A. and Josephic, P. H., Acta Metallurgica, Vol. 22, 1974, p. 1065.
[18] Dautovich, D. P. and Floreen, S., Metallurgical Transactions, Vol. 4, 1973, p. 2627.
[19] Beachem, C. D., Metallurgical Transactions, Vol. 3, 1972, p. 437.
[20] Mclntyre, P. in Proceedings, International Conference on Stress Corrosion Cracking and
Hydrogen Embrittlement of Iron Based Alloys, J. Hochmann, J. Slater, and R. W.
Staehle, Eds., National Association of Corrosion Engineers, Houston, Tex., 1974, in press.
[21] Moon, D. M. and Landes, J. D., Scripta Metallurgia, Vol. 10, 1976, p. 121.
[22] Gangloff, R. P . a n d Wei, R. P., Metallurgical Transactions, Vol. 8A, 1977, p. 1043.
[23] Pampillo, C. A. and Paxton, H. W., Metallurgical Transactions, Vol. 3, 1972, p. 2895.
[24] Cox, T. B. and Low, J. R., Metallurgical Transactions, Vol. 5, 1974, p. 1457.
[25] Spaeder, G. J., Metallurgical Transactions, Vol. 1, 1970, p. 2011.
[26] Carter, C. S., Metallurgical Transactions, Vol. 1, 1970, p. 1551.
[27] Wayman, M. L. and Smith, G. C., Metallurgical Transactions, Vol. 1, 1970, p. 1189.
[28] Bernstein, I. M., Metallurgical Transactions, Vol. 1, 1970, p. 3143.
[29] Rath, B. B. and Bernstein, I. M., Metallurgical Transactions, Vol. 2, 1971, p. 2845.
[30] Bernstein, I. M., Materials Science and Engineering, Vol. 6, 1970, p. 1.
[31] Krauss, G. and Marder, A. R., Metallurgical Transactions, Vol. 2, 1971, p. 2357.
[32] Roberts, M. J., Metallurgical Transactions, Vol. 1, 1970, p. 3287.
[33] Lapasset, G., Laurent, J. P., Avcouturer, M., and Laeombe, P. in Proceedings, Con-
ference L 'Hydrogene Dans Les M'etaux, Paris, 1972, p. 108.
[34] Laurent, J. P., Lapasset, G., Avcouturer, M., and Lacombe, P. in Hydrogen in Metals,
I. M. Bernstein and A. W. Thompson, Eds., American Society for Metals, Metals
Park, Ohio, 1974, p. 559.

Analysis of Fracture Morphology of

Hydrogen-Assisted Cracking in Steel and
Its Welds

REFERENCE: Kikuta, Yoneo, Araki, Takao, and Kuroda, Toshio, "Analysis of Frac-
lure Morphology of Hydrogen-Assisted Cracking in Steel and Its Welds," Fractography
in Failure Analysis, A S T M STP 645, B. M. Strauss and W. H. Cullen Eds., American
Society for Testing and Materials, 1978, pp. 107-127.

ABSTRACT: The relationships between fracture morphology of hydrogen-assisted

cracking, mierostructure, and crystallographic orientation were investigated using
the scanning electron microscope.
In the fracture morphology of hydrogen-assisted cracking, quasicleavage fracture,
intergranular fracture, and dimple rupture were observed. In the quasicleavage frac-
ture caused by hydrogen, subcracks often were observed along the boundary of mar-
tensite lath or ferrite lath, or both, and at the interface between the matrix and car-
bides.
The unit fracture facet for hydrogen-assisted cracking, therefore, was defined as
the region between subcracks. The unit fracture facet for hydrogen-assisted cracking
was smaller than that for cleavage fracture of martensite, upper bainite, and bainitic
ferrite. Using the etch pit method the crystallographic orientation of the fracture mor-
phology of hydrogen-assisted cracking was found to be the { 110} plane.
The fracture morphologies of hydrogen-assisted cracking can be categorized as one
of four types, each of which can be explained by the microscopic diffusion behavior
of hydrogen.

KEY WORDS: hydrogen embrittlement, fractography, crystallography, orientation,

three-point bendingtest, implant weld cold cracking test, cleavage, fractures (materials),
subcrack, unit fracture facet, hydrogen cold-work-peak height

It is g e n e r a l l y said t h a t steel c a n a b s o r b h y d r o g e n f r o m p r o d u c t i o n , pro-

cessing, a n d use u n d e r n a t u r a l e n v i r o n m e n t , a n d t h u s have its p r o p e r t i e s
i m p a i r e d d u e to h y d r o g e n - a s s i s t e d c r a c k i n g . H y d r o g e n - a s s i s t e d c r a c k i n g is
r e c o g n i z e d as h y d r o g e n e m b r i t t l e m e n t , d e l a y e d c r a c k i n g , weld cold c r a c k -
ing, l a m e l l a r t e a r i n g a d j a c e n t to i n c l u s i o n s e l o n g a t e d in t h e r o l l i n g direc-
1Professor and associate professor, respectively, and Drs. of Engineering, Department of
Welding Engineering, Osaka University, Osaka, Japan.
2 Research instructor and Dr. of Engineering, Welding Research Institute of Osaka Univer-
sity, Osaka, Japan.

107
Copyright by ASTM
9 Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright 1978by ASTM International [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
108 FRACTOGRAPHY IN FAILURE ANALYSIS

tion, and environmental embrittlement occurring in use under natural

environments or a hydrogen sulfide (H2S) atmosphere.
Recently, electron fractography has been used to investigate the causes
of various fracture morphologies. Although complex, the characteristic frac-
ture morphology has been related to stress level and microstructure. There-
fore, causes and mechanisms of fracture can be suggested approximately
by observations of fracture morphology. The morphology of hydrogen-
assisted cracking includes dimple rupture (DR), quasicleavage fracture
(QC), and intergranular fracture (IG). However, these fracture morphol-
ogies do not necessarily exhibit characteristics of fracture induced by only
hydrogen but also are observed in the case of other cracking processes. It
is difficult to isolate uniquely the characteristics of fracture morphology in-
duced by hydrogen. However, it is important to establish the characteris-
tics of the fracture morphology which are relevant to investigations of causes
of premature failure.

Experimental Procedure
Present work was carried out for commercial HT-80A and B steels
(80 kg/mm 2 class high strength steel). In the implant weld cold-cracking
test, the steels were compared on the basis of variations in sulfur content.
These chemical compositions are given in Table 1.
Ordinary V-notch specimens were used for the three-point bending test
and delayed cracking test as shown in Fig. l(a). The specimen used for
the implant weld cold-cracking test is shown in Fig. l(b). This specimen
was prepared from sections in the longitudinal ( X - ) and short transverse
(Z - ) direction of plate.
Specimens were hydrogenated by cathodically charging for the three-
point bending test by a high-temperature hydrogenation treatment for the
delayed cracking test and by manual bead-on-plate welding for the implant
test. These hydrogenating conditions are given in Table 2.
The three-point bending test was carried out with a bending rate of 1
ram/rain by an Instron-type tension machine. For the delayed cracking test
and the implant weld cold-cracking test, the applied gross stress was main-
tained at a constant until complete failure occurred. Details of the appa-
ratus used are given in Fig. 2.
Fracture surfaces obtained by the tests were observed by scanning elec-
tron microscopy and two-stage replica method. Inclusions on the fracture
surfaces were analyzed by the electron probe. Crystallographic orientation
on the fracture surfaces was measured using the etch pit method [1].3 In
body-centered-cubic iron and steel, pits were observed on { 110} and { 100}
planes and exhibited shapes as shown in Fig. 3.
3The italicnumbersin brackets referto the list of referencesappendedto this paper.

m
.H

i--
0
z

Z
TABLE 1--Chemical c o m p o s i t i o n s o f materials used, weight percent.
I-
.<
Materials C Si Mn P S AI Mo Ni Cr Cu V B Ti
if)
HT-80A 0.17 0.35 1.31 0.021 0.023 0.03 0.42 0.32 0.49 0.26 0.04 0.003 0.015 0
HT-80B 0.10 0.24 0.79 0.009 0.004 0.06 0.50 1.05 0.41 0.23 0.03 0.002 ... "11

"11
0
c
an
[11

0
:IJ
"1"
0
r
o
Q
.<

(a) ~:

T ,
- ~00

(b)
M10

16 2

FIG. 1--Specimengeometries. (a) 45-deg V-notchedspecimenfor three-point bending test.

(b) Specimen prepared from longitudinal rolling (X) and short transverse (Z) direction of
plate for implant weld coM-cracking test.

Results and D i s c u s s i o n

Fracture Morphology o f Hydrogen Embrittlement and Others

The fracture strength of material ordinarily depends on prior austenite
grain size or lath width and length [2,3]. The fracture morphology was
divided between intercrystalline cracking along austenitic grain boundaries
and transgranular cracking across the austenite grain. Moreover, the crack-
ing could be classified into such groups as ductile fracture with consider-
able plastic deformation, brittle fracture with little plastic deformation,
and fatigue cracking with gradual development of fracture by the cyclic
stress.
First, typical ductile and brittle fractures are shown in Fig. 4. Figure
4(a) was produced, for example, by the three-point bending test
performed with a crosshead speed of 1 mm/min at 25~ This fracture
appeared at failure after plastic deformation. The fracture surface was
caused by microvoid nucleation and coalescence at inclusions resulting in
dimple rupture pattern.
Next, typical brittle fracture is shown in Fig. 4(b). This fracture was
found, for example, in impact tests at low temperatures of large heat in-
put welded weld-bonds in high strength steel [4]. The characteristic of this
fracture is a cleavage step or a river pattern (CF).
The fracture resulting from an impact test of the synthetic weld heat af-
fected zone (HAZ) with a bainitic structure is shown in Fig. 4(c) in which
cleavage facets enclosed by tear ridges are recognized. It is difficult in this
region to identify whether any characteristics of cleavage fracture exist or
not. For this reason, this fracture is named a quasicleavage fracture (QC).

f--

0
TABLE 2--Methods of hydrogenation and hydrogen contents. z

Z
Methods of Hydrogen Con- ),
Hydrogenation Conditions of Hydrogenation tents, ppm Remark r-
-<
o~
i
Cathodic charging current density: 80 mA/cm 2 for 3 h at 20 ~ 8 three-point bending test
electrolyte: 5% H2SO4 in Poison P (20 mg/litre) 0
"11
High temperature hydrogen atmosphere (1 atm) at 950~ for 2 h and water 4 three-point bending delayed crack-
method quenching. "11
ing test
Shielded metal arc used electrode: El1016 (4 mm) as received 12 implant weld cold-cracking test
welding welding conditions: 25 V, 180 A and 150 ram/rain speed 0
c
m

O
-0
I
0
0
G)
-<

..g
..g
..A

(a) (b)
1
E

S A speclmen
C Load cell
A
B
C
D
E
Implant speclmcn
Backing plate
Load cell
Load
Water jacket

FIG. 2--Schematic diagrams of experimental apparatus. (a) Three-point bending test for
delayed cracking test. (b) Implant weld cold-cracking test.

{110} plane {100} plane

FIG. 3--Schematic illustration of faceted etch pits.

Fracture of hydrogen-charged specimens of HT-80A steel with various

microstructures tested by the three-point bending test with a crosshead
speed of 1 mm/min is shown in Fig. 5. In Fig. 5(a), (b), and (c), equidis-
tant subcracks are oriented approximately vertical to the main crack sur-
faces. These fracture morphologies clearly differed from those of ductile
fracture or brittle fracture shown in Fig. 4. Moreover, in Fig. 5(d) and (e),
a striation-like or ripple pattern was identified between subcracks. This
fine surface topography generally is observed by using a replica method as
shown in Fig. 5(e). Phillips [5] and Meyn [6] also observed the same
pattern in the fractures of a high strength steel with a high hydrogen
content and a titanium alloy, respectively. The formation of the pattern

t'-
0
z

z
I"
-<

0
-I1
"11

-is
Ill

0
FIG. 4--Typical fracture modes of hydrogen free materials. (a) Ductile fracture (dimple rupture, DR). (b) Brittle fracture (CF) tested at O~ (c)
"0
Quasicleavage fracture (QC) tested at - 75 ~ "1"
0
P
0
-<

-r
co

has not been explained clearly, but it is considered to be due to the

collapse of slip planes after considerable plastic deformation without the
formation of voids, such as dimples. The pattern is formed because mate-
rials deform plastically not only on the slip plane but also on the cross-slip
plane within the individual laths. However, for these materials it is not dif-
ficult to explain the role of hydrogen in the formation of the pattern.

Characteristics of Fracture Morphology for Hydrogen Embrittlement

It was found that hydrogen embrittlement fracture differed from dimple
and cleavage fractures. However, based only on the information from the
scanning electron microscope, it was conjecture to determine the charac-
teristics of fracture. Since this was the case, crystallographic orientation of
the fracture surface was investigated by the etch pit method. The crystallo-
graphic orientation of hydrogen embrittlement fracture, for example, as
shown in Fig. 5(f), was a { 110} plane on which the etch pits are hexagonal.
It is well known that the crystallographic orientation of cleavage is a { 100}
plane on which etch pits are square.
The relation between fracture facet size and prior austenite grain size for
martensitic structure is shown in Fig. 6. It was clear that the distance be-
tween subcracks corresponded to martensite lath or ferrite lath width.
Facet size of the hydrogen embrittlement fracture was a third or a half of
that of brittle fracture [7]. The schematic representation is shown in Fig. 7.
In this figure, the fracture path was classified into one of three types:
translath fracture, interlath fracture, or intergranular fracture. Actually,
these three types came out as a mixture. From the aforementioned results,

(a)
100 i i i i (b)
O UNIT FACET(H-FREE) i00
O UNIT FACET(H-CHARGED) ON--UIIT FACET(H-FREE)
' i ,
&MARTENSITE LATH OUNIT FACET(H-CHARGED)
80 ~MARTENSITE COLONY A MARTENSITE LATH
80

c, 40
60
/ -~ 60

~ 4o

20 20

0 0 I I I .I
0 40 BO 120 160 200 0 40 80 120 160 200
PRIOR AUSTENITE GRAIN SIZE PRIOR AUSTENITE GRAIN SIZE
(~) (u)

F I G . 6--Variation o f unit facet size, martensite lath width, and colony spacing with prior
austenite grain size for martensite structure (HT-80A steel).

translath fracture inter granular

CTL ,~,\\~/ fracture

CIL racture

I. i

FIG. 7--Schematic representation offracture profile of hydrogen embrittlement.

it was found that fracture morphology of hydrogen embrittlement was dif-

ferent from ductile fracture, brittle fracture, and quasicleavage fracture.
Therefore, the authors propose that this fracture be named quasicleavage
fracture of hydrogen embrittlement (QCHE). 4 Since this fracture occurs on
the { 110} planes and the striation-like markings are recognized in it, hydro-
gen embrittlement is considered to be glide-plane decohesion.

Fracture Morphology of Delayed Cracking Due to Hydrogen

There are various types of delayed failure of material, for example, fa-
tigue cracking, stress-corrosion cracking, creep rupture, and so on. In
these types, crack initiation occurs after a time lag without variation in ex-
ternal conditions such as stress. This phenomenon is called delayed hydro-
gen cracking or static fatigue fracture. This time delay is necessary for
the accumulation of hydrogen by diffusion at the crack tip. This type of
fracture includes season cracking, environment cracking, and weld-cold
cracking. It may be said safely that delayed cracking induced by hydrogen
is hydrogen embrittlement. However, in this paper the delayed cracking is
defined as fracture which arises under constant stress with the passage of
time.
4The definition of quasicleavage (QCHE) used in this paper is not the same as that (QC)
proposed by Beachem [8], which may be considered to be the crystallographic orientation
of the fracture as {100} plane, similar to cleavagefracture. QC is also observed in the case of
other cracking processes without hydrogen. QCrIEis considered to be cracking by glide-plane
decohesion {110} plane in conjunction with hydrogen gathering around a dislocation. It is
considered that its cracking process differs from QC, somewhat like brittle fracture.

The delayed cracking fracture of high-temperature hydrogenated

specimens and the weld cold-cracking fracture of each HAZ of specimens
prepared from longitudinal rolling (X) and short transverse (Z) direction of
HT-80A steel plate are shown in Fig. 8. Figure 8(a), (b), and (c) exhibit
fractures from the three-point bend, delayed cracking test of specimens
hydrogenated at 950~ for 2 h. The fracture morphology of the delayed
cracking was a mixture of several modes such as DR, QCnE, IG, and CF,
but the dominant fracture morphology differs according to applied stress
level. Dimple pattern was dominant with an applied stress (three-point
bending nominal stress) of 1400 MPa. The hydrogen-free specimen had a
fracture stress of 2500 MPa, while failure of the specimen, including
hydrogen, occurred under a constant load which was lower than that of the
hydrogen-free specimen. The fracture of the hydrogen-free specimen also
had the dimple pattern. It was difficult to understand whether the frac-
ture was induced by hydrogen or not. However, dimple size of the delayed
cracking appeared as small as that of the hydrogen-free specimen. QCxE
fracture was dominant with an applied stress of 1000 MPa, and QCHE and
IG were recognized with an applied stress of 750 MPa which was nearly a
critical low stress level.

Fracture Morphology of Weld Cold Cracking

Figure 8(d), (e), and (f) exhibit weld cold-cracking fractures for the longi-
tudinal rolling direction for HT-80A steel. As in the previous case, frac-
ture morphologies varied with applied stress level.
Figure 8(g), (h), and (i) exhibit weld cold-cracking fracture in the short
transverse direction for the same steel. These fractures also varied with
applied stress level and had the same general characteristics as the afore-
mentioned morphology. But, in high applied stress, fracture differed
slightly from the result. With an applied stress of 530 MPa or on leaving it
unloaded as it was for two weeks after welding, terrace and wall-like frac-
tures, namely, lameUar tearing was observed. The terrace consisted of
elongated manganese sulfide (MnS) clusters and fine dimples with inclu-
sions. The wall consisted of extended dimples. With an applied stress of
400 MPa, the terrace was the same as before and intergranular failure
appeared in the wall. With an applied stress of 300 MPa, which was nearly
a critical low stress level, the distinction between terrace and wall was not
clear, and QCnE and IG were observed. Inclusion content had no effect on
the fracture morphology. The fracture path in this case is shown in Fig. 9
which has the same characteristics as Fig. 7. From Fig. 9 it is deduced that
inclusions were not of consequence in the fracture. The steel with low
sulfur content had little tendency toward lamellar tearing. The fracture for
HT-80B steel, which was left for two weeks after welding, was observed to
have the dimple fracture with fine MnS inclusions in the dimple bottoms.

While under constant load after welding, delayed cracking occurred. Its
fracture morphology exhibited QCaE as shown in Fig. 10, and the fracture
path consisted of translath and interlath cracking. In the short transverse
direction of the low sulfur level steel, the terrace and wall-like fracture was
not observed.

Fracture Morphology of Lamellar Tearing

In steel with a high sulfur content and large elongated MnS clusters, the
terrace and wall structure was observed in fracture along the short
transverse direction obtained in the weld cold-cracking test. In the tension
test of a hydrogen-free specimen, Elliott proposed that the terrace and
wall-like fracture morphology of the hydrogen-free specimen was caused by
lamellar tearing [9]. In Japan, this phenomenon has been investigated [10]
and it has been stated that lamellar tearing was observed in high sulfur
content steel, and close attention should be paid to these inclusions. The
authors propose that the terrace and wall-like fracture of hydrogen
includes lamellar tearing in the specimen, too. It was difficult to observe
lamellar tearing in the fracture morphologies in which the terrace and wall
were not clear, and inclusions were not observed in the short transverse di-
rection specimen with high sulfur content and low stress level.
Therefore, this implies that in the weld cold-cracking test lamellar
tearing is the fracture morphology of a specimen with a high sulfur level
under high applied stress.

Effect of Hydrogen Diffusion on Fracture Morphology of Hydrogen

Assisted Cracking
It has been proposed that hydrogen embrittlement is diffusion
dependent. There is also much discussion in the literature concerning the
exact location of hydrogen in the material. Several workers have postulated
that hydrogen interacts with a "trap" of some sort. Various defects such as
voids [11], second phases [12], vacancies [13], and dislocations [14] have
been suggested as possible traps. There is, however, no direct evidence that
interaction is a necessary part of the embrittlement process.
Recently, the authors carried out a tension test and an internal friction
experiment as shown in Fig. 11 [15]. When specimens of HT-80A steel
were hydrogen charged by the high temperature method, their ductility de-
creased at an early stage of aging, steadied at 0~ and then went through
a minimum. At a later stage of aging, it recovered the value of the hydro-
gen-free specimen. Notch tensile strength (NTS) could be regarded as a
measure of ductility, and it was used as the most convenient parameter for
embrittlement. Hydrogen cold-work-peak height increased with increasing
aging time, went through its maximum, and decreased to a vanishing small

t-
O
z
})
z
I-"

0
"11
"11

E)
E
m

FIG. 8--Typical fracture modes of hydrogen delayed cracking (HT-8OA). Three-point bending test (morten~ite at 950~ charged hydrogen) at E
0
25~ tested: (a)fracture time 6 rain at applied stress 1400 MPa, (b) 30 rain, 1000 MPa, and (c) 90 rain, 750 MPa. Implant test (HT-8OA-HAZ-X): (d)
"D
3 rain, 850 MPa (e) 8 rain, 750 MPa, and (f) 50 vain, 700 MPa. Implant test (HT-8OA-HAZ-Z)z (g) 700 MPa (TS) (h) 15 rain, 400 MPa and 100 rain, -1-
300 MPa. 0
t-
O
0-<

FIG. 9--Fracture profile of hydrogen delayed fracture for ([Link].Z) applied stress
300MPa (fracture time: 100 rain).

FIG. lO--Froeture profile of hydrogen delayed fracture for (ItT-8OB-HAZ-Z) applied stress
700MPa (fracture time: 100 rain).

value. Parallelism was found between the drop in ductility and the hydrogen
cold-work-peak height.
Activation energy for hydrogen diffusion, Q, is determined from temper-
ature dependency of diffusion in the tension test [16]. In the early stage of
aging, the activation energy for hydrogen diffusion was about 3200
cal/mol. This value may express only the lattice diffusion of hydrogen. In
the latter stage of aging, it was about 8000 cal/mol. This value may result
from diffusion through microscopic defects such as dislocations and vacan-
cies.
From these experimental results, it may he concluded that hydrogen
cold-work-peak height can be proportional to hydrogen, Co, in the neigh-
borhood of the dislocation, as shown in Fig. 11 and that hydrogen first dif-
fuses toward dislocations and then away from dislocations.
It seems that the microscopic diffusion behavior of hydrogen affects frac-
ture morphology of hydrogen embrittlement as observed in this study. The
fracture morphology of hydrogen-assisted cracking can be summarized into
four types as shown in Table 3, and these can be explained by microscopic
diffusion behavior of hydrogen.
QCH~ was observed on the fracture surface obtained under medium and
low constant applied stress within the region in which delayed fracture oc-
curred. Since occurrence of this fracture takes an average or long time, the
concentration of hydrogen at dislocations may increase and ductility
decrease as shown in Fig. 11. Therefore, it seems that QCnE starts from a
region of the highest hydrogen concentration which is generally located in
the region of the highest dislocation density such as lath and grain
boundaries. It seems that IG is caused by accumulated hydrogen at the
grain boundary.
In the case of the continuous slow bending conditions, QCHE also is ob-
served. The reason [Link] may be that as hydrogen is charged cathodically,

,,.~ ~ . ~2oo /'0~ \~ 8o'ooI~_

_..i/ I\ \ d=
y -'"

0
I0 lO0 lO00
AGING TIME AT O~ (min)

FIG. 11--Relationship between ductility and hydrogen cold-work-peak height for hydro-
genated steel (HT-80).

Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
TABLE 3--Schematic representation offactors affecting the fracture morphologies of hydrogen-assisted cracking.

Concentration
of Hydrogen on Dominant Characteristic Hydrogen
Conditions of Fracture Dislocation, CD, Fracture Fracture Assisted
Applied Loading Time and Ductility Modes Morphologies Cracking
_x

--4
I 3>
ver and oct avera high, 9 subcracks at lath delayed fr
applied stress and hydrogen boundary cracking -I
I
within the range long accumulation at 9 striation-like
in which delayed lath and grain pattern in lath weld i-
fracture occurred boundary QC t~ 9 fracture path cold O
(low) through trance and cracking z
interlath
9 main fracture surface, hydrogen z
{110} plane embrittlement r-
..<

g~
9 fracture path through
Continuous slower grain boundary 0
"TI
bending and IG 9 subcrack at grain
"11
tension boundary 3g
}>
0
transitional E
applied stress state from Ill
within the range short low, DR 9 dimple pattern ductile
io
in which delayed
t hydrogen fracture to E
0
fracture occurred random HAC fracture
distribution "0
"1"
(high) 0
Continuous rapid 9 dimple pattern and r-
bending and elongated inclusion lamellas 0
tension on terrace tearing -<
Inclusion 9 terrace and wall-like
(high sulfur content) ..k
ol

Co pyr igh t b y A S TM In t'l ( al l r i g h t s r es er v ed ); Mo n D ec 2 1 1 1 : 2 5 : 3 5 E S T 2 0 1 5

D ow nlo ade d/pr inted b y
U niv ers ity of Wa sh ingto n (U n i v er si t y o f Wa sh i ng t on ) pu r su an t t o L i cen se A g r eem en t . N o f u r t h er r ep r o d u ct i o n s au t h o r i zed.
126 FRACTOGRAPHYIN FAILURE ANALYSIS

hydrogen accumulates easily at the dislocations as compared with the high

temperature charging method. Therefore, QCaE will occur.
While QCHE did not occur in the fracture surface obtained under higher
constant applied stress within the range of delayed fracture, D R and D R +
inclusions on the terraces were observed. Since occurrence of fracture takes
a short time, Fig. 11 corresponds to an early stage at which the concentra-
tion of hydrogen on dislocations may be lower and thus the ductility
higher, but still lower than the case of the hydrogen-free specimen. It
seems that this fracture, which is similar to ductile fracture, occurs at the
transitional state from ductile fracture to hydrogen embrittlement fracture;
but it is difficult to distinguish whether this fracture is concerned with
hydrogen or not.
In the case of continuous rapid bending or tensile fracture, D R also is
observed because dislocations break away from the hydrogen atmosphere,
resulting in low hydrogen on dislocations.

Conclusions
The characteristic morphology of hydrogen embrittlement was quasi-
cleavage fracture (QCttE) which involved subcracks along martensite lath
boundaries, ferrite lath boundaries, and the interfaces between the matrix
and carbides. The unit fracture facet for hydrogen embrittlement was
defined as the region between subcracks or tear ridges. The unit fracture
facet for hydrogen embrittlement was smaller than that of cleavage frac-
ture.
The crystallographic orientation of the fracture morphology of hydrogen
embrittlement was {110} planes. So, it seemed that hydrogen embrittle-
ment was caused by slide plane decohesion.
The fracture morphology of hydrogen-assisted cracking could be sum-
marized into four types and they could be explained by microscopic dif-
fusion behavior of hydrogen.

References
[1] Taoka, T., Furubayashi, F., and Takenouchi, S., Japan Journal of Applied Physics,
Vol. 4, 1965, p. 120.
[2] Petch, N. J. in Fracture, Wiley, New York, 1959, p. 54.
[3] Embury, J. D., Keh, A. S., and Fisher, R. M., Transactions,. Metallurgical Society of
the American Institute of Mining, Metallurgical, and Petroleum Engineers, Vol. 239,
1966, p. 1252.
[4] Kikuta, Y. and Araki, T., Technology Reports of the Osaka University, Vol. 26, 1976,
p. 53.
[5] Phillips, A. and Bennett, G. V., Metals Progress, Vol. 79, No. 5, May 1961, p. 97.
[6] Meyn, D. A., Metallurgical Transactions, Vol. 3, 1972, p. 2302.
[7] Terasaki, F. and Otani, H., Journal of the Iron and Steel Institute of Japan, Vol. 58,
1972, p. 1067.
[8] Beachem, C. D., Metallurgical Transactions, Vol. 3, No. 2, 1972, p. 437.

[9] Elliot, D. N., Metal Construction and British Welding Journal, Vol. 2, 1969, p. 50.
[10] Kihara, H., Suzuki, H. and Ogura, N., Journal of the Japan Welding Society, Vol. 25,
No. 2, 1951, p. 94.
[11] Podgurski, H. H., Transactions, Metallurgical Society of the American Institute of
Mining, Metallurgical, and Petroleum Engineers, Vol. 221, 1961, p. 389.
[12] Coe, F. R. and Moreton, J., Journal of the Iron and Steel Institute, Vol. 207, p. 366.
[13] Heller, W. R. in Stress Corrosion Crack Embrittlement, Wiley, New York, 1956, p. 163.
[14] Bastien, P. and Azou, P., Proceedings, First World Metal Congress, American Society
for Metals, Cleveland, Ohio, 1951, p. 535.
[15] Kikuta, Y., Sugimoto, K., Oehiai, S., and Iwata, K., Proceedings, First International
Congress, Hydrogen in Metals, Paris, 1972, p. 144.
[/6] Kikuta, Y., Araki, T., and Ochiai, S., Journal of the Japan Welding Society, Vol. 45,
No. 12, 1976, p. 1016.

Fracture of Ti-8AI-1Mo-IV Alloy Fan

Blade by Stress Corrosion Cracking and
Fatigue

REFERENCE: Lee, E. U., Mahorter, R. G., and Wacaser, J. D., "Pt-acture of TI.8AI.
IMo-1V Alloy Fan Blade by Stress Corrosion Cracking and Fatigue," Fractography
in Failure Analysis, ASTM STP 645, B. M. Strauss and W. H. Cullen, Jr., Eds.,
American Society for Testing and Materials, 1978, pp. 128-143.

ABSTRACT: Similar fractures occurred in two gas turbine engine fan blades made of
Ti-SAI-IMo-IV alloy. Their analyses and a supplementary test were performed with
the aid of scanning electron microscope (SEM) and transmission electron microscope
(TEM) fractography to identify the common mechanism(s).
The initial fracture surface of each fan blade exhibited an area of subcritieal
crack growth. The area in one fan blade contained three blue discolored zones along
the crack initiation side, whereas that in the other fan blade was discolored to white-
gray. Intergranular separation and transgranular cleavage were evident at the three
discolored zones in one fan blade and near the crack initiation site in the other. Cleav-
age-like facets were predominant in the remaining area of subcritical crack growth.
On some of those facets, striations were seen.
A hot salt stress corrosion cracking test of Ti-8AI-1Mo-IV alloy plates resulted in
discoloration, intergranular separation, and transgranular cleavage, similar to those
observed in one fan blade.
The mechanism of initial crack growth is stress corrosion cracking at an elevated
temperature in one fan blade and at ambient temperature in the other. The mechanism
of subsequent subcritical crack growth is fatigue at ambient temperature in both fan
blades.

KEY WORDS: fractography, fractures (materials), crack initiation, subcritical crack

growth, intergranular separation, transgranular cleavage, stress corrosion, fatigue
(materials), striations, stress intensity factor, microvoid coalescence

The Ti-8AI-1Mo-IV alloy has been used in aircraft gas turbine engines
for parts requiring low density and high modulus for optimum performance.
1 Metallurgist, Materials Engineering Laboratory, Naval Air Rework Facility, Norfolk, Va.
23511.
2Mctallurgist, Naval Air Development Center, Warminster, Pa. 18974.
3 Manufacturing research and development technologist, Pratt and Whitney Aircraft, East
Hartford, Conn. 06108.

128
Copyright by ASTM
9 Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright 1978 by ASTM International [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
LEE ET AL ON FRACTURE OF Ti-8AI-1Mo-IV ALLOY 129

The most critical of these parts are fan blades. Due to the rotational speed
(approximately 167 rps or 10 000 rpm), blades are subjected to high steady
(centrifugal) stresses and high frequency vibratory stresses along with air
pressure loading on the concave surface. In addition, they must operate in
salt-laden air and are subject to damage from objects ingested into the
engine. Fan blades are manufactured by several forging operations, chem-
ical descaling, heat treatment, 4 chemical milling, straightening operations,
machining of the root and midspan shroud, belt blending and sanding of
the airfoil surface, and finally a stress relief treatment. A typical fan blade
is shown in Fig. 1.
Fan blade fractures often have led to catastrophic failures of engines
which have resulted in the loss of the aircraft. On the other hand, they
have created interesting fractographic features, characterizing a combina-
tion of complex stresses or detrimental environments during manufacture
and service. Many times the fractographic features have indicated stress
corrosion cracking or fatigue as the fracture mechanism.
The stress corrosion cracking [I-10] s and fatigue [11-19] of Ti-8AI-1Mo-
1V and other titanium alloys have been studied extensively. Stress corrosion
cracking proceeds by intergranular separation at low levels of stress inten-
sity factor, K, transgranular cleavage at intermediate levels, and microvoid
coalescence at levels close to the critical stress intensity factor, K~ [8].
Fracture is accompanied by crack branching which increases with K [8].
The fatigue results in cleavage-like facets at low levels of growth rate, da/dn,
striations at intermediate levels, and coalesced microvoids at high levels
[13,161.

FIG. 1--Fan blade.

4Heat treating: (a) heat at a temperature within the range 1172 to 1200 K (1650 to 1700~
for 1 h and cool at a rate equivalent to air cool or faster; (b) heat at a temperature within
the range 839 to 867 K (1050 to 1100~ for not less than 8 h and cool in air.
SThe italic numbers in brackets refer to the list of referencesappended to this paper.

This paper presents the analyses, with fractographic evidences, of two

separate Ti-8AI-1Mo-IV alloy fan blade fractures, which occurred during
engine operation. An analysis of a hot salt stress corrosion test was made
to confirm the failure mode of one of the blades and to relate the failure
to the manufacturing process.

Procedure of Fracture Analysis

The two fractured fan blades were examined visually and with a low mag-
nification ( x 15 to • 90) optical mic~scope to determine the extent of the
fracture and the macromorphology of the fracture surface. The micromor-
phology of the fracture surface, especially that of the subcritical crack
growth area, was studied with scanning and transmission electron micro-
scope, operated at 20 and 60 kV, respectively. For the study with a trans-
mission electron microscope (TEM), chromium-shadowed carbon replicas
of the fracture surface were prepared by employing the standard two-stage
replication technique. For a further examination of fracture path, optical
micrographs of sections through the fracture were made. A blue etch ano-
dize inspection [20] was performed to detect possible alpha case 6 on a fan
blade airfoil surface.
A Ti-8AI-IMo-IV alloy plate, 20.32 by 2.54 by 0.24 cm (8 by 1 by 3/32
in.), was subjected to a hot salt stress corrosion cracking test. The plate
was loaded in a four-point bend fixture in a constant deflection mode.
The deflection used gave a nominal outer fiber stress of 75 percent of the
room temperature yield stress. Salt slurry was applied to a 0.S-era-diameter
spot and the specimen was exposed for 2 h at 811 K (1000~ During this
exposure, the stress relaxed to approximately the yield stress at 811 K
(1000~
The time and temperature of the exposure duplicated the stress relief
heat treatment used during blade manufacture. The residual stresses in
blades prior to heat treatment were not known. However, since the manu-
facturing process includes straightening operations and a significant
amount of machining, it is felt that the induced residual stresses equal to
the 811 K (1000~ yield stress are not unlikely.
After exposure, the plate was notched at the salt spot and broken by a
single rapid overload (hammer blow). The fracture surfaces then were ex-
amined visually and with the aid of a scanning electron microscope (SEM).

Results
In order to distinguish the two fractured fan blades, they are referred
to as A and B, respectively, in this paper. The results of their fracture
analyses and the hot salt stress corrosion cracking test are as follows.
6 Alpha case: brittle layer of alpha phase, stabilized by segregated interstitials, such as oxygen.

Fan Blade A
The airfoil of this fan blade was fractured at 16.51 cm (61A-in.) out-
board of the platform. Near the fracture, several cracks were detectable
along tool marks, parallel to the fracture, on the concave side of the in-
board portion. The outboard portion was missing.
The fracture surface consisted of an area of subcritical crack growth in
the mid portion and two separated areas of fast fracture on both sides
(Fig. 2). The area of subcritical crack growth was fiat, lenticular, about
3.81 cm (11/2 in.) long; this area extended through the airfoil cross sec-
tion in the middle. This area contained three semielliptic zones, discolored
to blue, along the concave side, with a beach mark in the rest. The dis-
colored zones were 0.18 cm (0.071 in.) long and 0.04 cm (0.015 in.) deep,
0.24 cm (0.096 in.) long and 0.05 cm (0.021 in.) deep, and 0.49 cm (0.192
in.) long and 0.05 cm (0.021 in.) deep. Both areas of fast fracture were
slanted and had a relatively fine texture.
The SEM fractographs of the discolored zones exhibited predominantly
intergranular and some interspersed cleavage facets with secondary cracks
(Fig. 3). There was a discernible tendency towards more intergranular facets
in the immediate vicinity of the concave side or the crack initiation site.
Deeper in these zones, secondary cracks extended farther and diverged
more. This observation of intergranular crack path and crack branching
was confirmed by the optical micrographs of sections through the dis-
colored zones (Fig. 4). Outside of these zones but within the subcritical
crack growth area, the SEM and TEM fractographs showed mostly cleav-
age-like facets, some of which contained striations normal to the river
lines (Fig. 5). However, near the transition to fast fracture area, a mix-
ture of cleavage-like facets and dimples were observed (Fig. 6). The slanted
area of fast fracture was completely covered with dimples, some of which
were elongated (Fig. 7).
The largest of the cracks, detectable near the fracture on the concave
side of the airfoil, was about 0.13 cm (0.050 in.) long. Its crack surface
was discolored to blue, and the SEM fractograph displayed mostly inter-
granular and some interspersed cleavage facets with secondary cracks. The
optical micrograph of a section through the crack also indicated intergran-
ular and transgranular crack path and crack branching.

Fan Blade B
This fan blade airfoil had four fractures, and its tip was missing. One of
the fractures was located at 3.81 cm (11/2 in.) outboard of the platform,
one branched out of the preceding one and off to the leading edge, and the
other two at and just outboard of the midspan shroud. Only the fracture
at 3.81 cm (1 89 in.) outboard of the platform had an indication of sub-
critical crack growth and was believed to be the initial one. The others

FIG. 2--Fracture surface o f Fan Blade A.

FIG. 3 - - S E M fractograph o f a discolored zone on the fracture surface o f Fan Blade A.

FIG. 4--Optical micrograph o f a section through a discolored zone on the fracture surface
o f Fan Blade A.

appeared to have occurred as the outboard portion of the blade burst

through the engine case.
The initial fracture surface consisted of a flat and lenticular mid portion
and two slanted side portions (Fig. 8). In the flat portion, fracture lines con-
verged to a point, the crack initiation site, 3.56 cm (1.4 in.) from the trail-
ing edge on the concave side. Around the crack initiation site, a thumbnail-

FIG. 5 - - S E M and T E M fractographs outside o f discolored zones in subcritical crack growth

area of Fan Blade A: (a) S E M fractograph, and (b) T E M fractograph.

FIG. 6--SEM fractograph near transition from subcritical crack growth to fast fracture in
Fan Blade A.

shaped area of subcritical crack growth, 1.78 cm (0.7 in.) long on the con-
cave side and 0.38 cm (0.15 in.) deep, was discolored to white-gray. The rest
of the flat portion displayed a rough texture, and the slanted side por-
tions were typical shear lips with a relatively fine texture.
In the immediate vicinity of the crack initiation site, a mixture of inter-
granular and cleavage facets with some secondary cracks was found (Fig.
9). The intergranular and transgranular crack path and crack branching
also were evidenced in the optical micrograph of a section through the frac-
ture. In the remaining area of subcritical crack growth, the SEM fraeto-
graph exhibited mostly cleavage-like facets, some of which contained
faint striations (Fig. 10). The presence of such striations was verified by
the TEM fractograph (Fig. 11).
A blue etch anodizing of this blade indicated presence of alpha case by
discoloring it to dark blue in an area on the concave side surrounding the
main fracture. The alpha case was about 0.008 cm (0.003 in.) thick and
had a Rockwell hardness C (HRC) 43, whereas that of the base metal was
HRC 35. Within the area of alpha case, tight cracks, mostly parallel to the
main fracture, were detectable. They were about 0.005 cm (0.002 in.) deep,
and followed an intergranular and transgranular path, as shown in the
optical micrograph (Fig. 12).

FIG. 7--SEM fractograph of slanted area in fracture surface of Fan Blade A.

FIG. 8--Fracture surface of Fan Blade B.

H o t Salt Stress Corrosion Cracking o f T i - S A l - I M o - 1 V Alloy S p e c i m e n

Before it was broken, several cracks, 0.06 to 0.33 cm (0.022 to 0.129 in.)
long, were visible and detectable with fluorescent penetrant in the area of

FIG. 9 - - S E M fractograph near crack initiation site o f Fan Blade B.

hot salt application on the surface of the specimen. The fracture surface,
which was contained in the area of the cracks, had two discolored and inter-
connected areas along the side exposed to salt (Fig. 13). The discoloration
was dark blue in the initially cracked portion and brownish yellow in the
rest. One of the areas was crescent-shaped, 0.55 cm (0.216 in.) long, and
0.14 cm (0.054 in.) deep. The other was elliptic, 0.23 cm (0.090 in.) long
and 0.11 cm (0.045 in.) deep. The SEM fractographs of both areas show
mostly intergranular and some interspersed cleavage facets with secondary
cracks (Fig. 14). Those facets, in the dark blue portion, were covered with
corrosion and oxidation products.
Discussion
The area of subcritical crack growth was discolored partly to blue in Fan
Blade A and totally to white-gray in Fan Blade B. In both blades, the
discoloration was greater near the crack initiation site on the airfoil surface.
Such a part or total discoloration of subcritical crack growth area indicates
a prior exposure of the initial crack surface to elevated temperatures or a
corrosive environment.
The common fractographic features of the subcritical crack growth areas
were intergranular and cleavage facets with branched cracks at and near

FIG. lO--SEM fractograph away from crack initiation site in subcritical crack growth area
of Fan Blade B.

the crack initiation sRe, and cleavage-like facets with occasional striations
away from it.
The observed crack initiation from external surface, greater discoloration
near the crack initiation site, intergranular separation mixed with trans-
granular cleavage, and crack branching are typical of stress corrosion
cracking in titanium alloys [8,2I]. Considering the initial crack surface
discoloration, the stress corrosion cracking evidently occurred at an elevated
temperature in Fan Blade A and at ambient temperature in Fan Blade B.
The small cracks, found near the fracture on the airfoil surface of each fan
blade, had fractographic features similar to those at and near the fracture
initiation site. The small cracks in Fan Blade A had discolored surfaces,
intergranular and cleavage facets, and secondary cracks. Those in Fan Blade
B followed intergranular and transgranular paths. These evidences suggest
that the mechanism of those crackings was identical to the initial stage
mechanism of the fracture, stress corrosion cracking at an elevated or
ambient temperature.
The temperature of the fan blade does not exceed 389 K (240 ~ during
the engine operation. Therefore, the elevated temperature stress corrosion
cracking of Fan Blade A must have occurred during the heat treatment. Its

FIG. 11--TEM fractograph away from crack initiation site in subcritical crack growth area
o f Fan Blade B.

initiation site, a spot on the airfoil surface, was presumably contaminated

by corrosive media such as moisture and salt from a fingerprint [8] and
had residual stress induced during the manufacturing process. Similar
crackings, discoloration, and fractographic features were reproduced by a
hot salt stress corrosion cracking test of Ti-8AI-1Mo-IV alloy specimens.
In Fan Blade B, the origin of the initial fracture and adjacent small
cracks were located within a patch of alpha case. The alpha case was con-
fined to a surface layer of limited thickness, 0.008 cm (0.003 in.), and
believed to have been formed by oxidation under the heat of excessive air-
foil surface blending or sanding. The preferential initiation of stress cor-
rosion cracks in such a patch of alpha case demonstrates its great suscep-
tibility to stress corrosion cracking, arising from high oxygen concentration
[22].
The mixture of intergranular separation and transgranular cleavage,
found in both fan blades, is an evidence for an initial stress corrosion
cracking at not very low crack growth velocity, V, and stress intensity fac-
tor, K. The mode of stress corrosion cracking is known to change from
intergranular separation to transgranular cleavage with increasing V and
K, creating their mixture in the transition [8].

FIG. 12--Crack in alpha case on Fan Blade B airfoil.

FIG. 13--Fracture surface o f hot salt stress corrosion cracking specimen.

The observed cleavage-like facets with occasional striations are charac-

teristic of fatigue at low da/dn and small stress intensity factor range AK
[13,16]. In Fan Blade A, those fractographic features were detectable in
the undiscolored portion of the subcritical crack growth area, having a
beach mark. This suggests a fatigue crack growth at ambient temperature
during the engine operation. In Fan Blade B, those fractographic features
were seen within the white-gray discolored area of subcritical crack growth.
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized
LEE ET AL ON FRACTURE OF Ti-8AI-1Mo-IV ALLOY 141

FIG. 1 4 - - S E M fractograph o f hot salt stress corrosion cracking specimen: (a) near crack
initiation site, and (b) away f r o m crack initiation site.

This suggests possible acceleration of t h e fatigue c r a c k growth b y corrosion

at a m b i e n t t e m p e r a t u r e d u r i n g t h e engine o p e r a t i o n .
T h e p a r t i a l microvoid coalescence n e a r the t r a n s i t i o n to fast f r a c t u r e ,
seen in F a n B l a d e A, is similar to H e r t z b e r g ' s o b s e r v a t i o n o f Ti-6A1-4V
a n d o t h e r alloys, f a t i g u e d at high d a / d n a n d A K [16].

Conclusion
On the basis o f t h e f r a c t u r e analysis results a n d discussion, t h e following
conclusions m a y be d r a w n .
1. T h e initial subcritical c r a c k growth was c h a r a c t e r i z e d b y i n t e r g r a n u -
l a r s e p a r a t i o n , t r a n s g r a n u l a r cleavage, a n d c r a c k b r a n c h i n g . Its mech-
a n i s m was stress corrosion c r a c k i n g at an elevated t e m p e r a t u r e in one f a n
b l a d e a n d at a m b i e n t t e m p e r a t u r e in t h e other.
2. T h e s u b s e q u e n t subcritical c r a c k growth was c h a r a c t e r i z e d b y cleav-
age-like faceting. Its m e c h a n i s m was a m b i e n t t e m p e r a t u r e fatigue, which
was possibly a c c e l e r a t e d b y corrosion in one fan b l a d e .

References
[I] Meyn, D. A. and Sandoz, G., Transactions, The Metallurgical Society of the American
Institute of Mining, Metallurgical, and Petroleum Engineers, Vol. 245, 1965, p. 1253.
[2] Beck, T. R., "Stress-Corrosion Cracking of Titanium Alloys, Preliminary Report on
Ti-SAI-IMo-IVAlloy and Proposed Electrochemical Mechanism," Boeing Document No.
DI-82-0554, The Boeing Company, Seattle, Wash., luly 1965.
[3] MacKay, T. L., Gilpin, C. B., and Tiner, N. A., "Stress-Corrosion Cracking of Titan-
ium Alloys at Ambient Temperatures in Aqueous Solutions," Contract NAS 7-488,
Report SM-49105-Fl, McDonnell Douglas Corporation, Santa Monica, Calif., July 1967.
[4] Piper, D. E., Smith, S. H., and Carter, R. V., Metals Engineering Quarterly, Vol. 8,
1968, p. 50.
[5] Fager, D. N. and Spurr, W. F., Transactions, American Society for Metals, Vol. 61,
1968, p. 283.
[6] Sandoz, G. in Proceedings of Conference on Fundamental Aspects of Stress Corrosion
Cracking, R. W. Staehle et al, Eds., National Association of Corrosion Engineering,
Houston, Tex., 1969, p. 684.
[7] Beck, T. R., Blackburn, M. J., Smyrl, W. H., and Speidel, M. O., "Stress Corrosion
Cracking of Titanium Alloys: Electrochemical Kinetics, SCC Studies with Ti: 8-1-1,
SCC and Polarization Curves in Molten Salts, Liquid Metal Embrittlement, and SCC
Studies with Other Titanium Alloys," Contract NAS 7-489, Quarterly Progress Report
14, The Boeing Company, Seattle, Wash., Dec. 1969.
[8] Blackburn, M. L, Smyrl, W. H., and Peeney, ]. A. in Stress-Corrosion Cracking in
High Strength Steels and in Titanium and Aluminum Alloys, B. F. Brown, Ed., Naval
Research Laboratory, Washington, D.C. 1972, p. 245.
[9] Boyd, J. D., Metallurgical Transactions, Vol. 4, 1973, p. 1029.
[10] Boyd, J. D., Metallurgical Transactions, Vol. 4, 1973, p. 1037.
[11] Meyn, D. A., Metallurgical Transactions, Vol. 2, 1971, p. 853.
[12] Irving, P. E. and Beevers, C. J., Metallurgical Transactions, Vol. 5, 1974, p. 391.
[13] Yuen, A., Hopkins, S. W., Leverant, G. R., and Rau, C. A., Metallurgical Transactions,
Vol. 5, 1974, p. 1833.
[14] Bowen, A. W., Acta Metallurgica, Vol. 23, 1975, p. 1401.
[15] Neal, D. F. and Blenkinsop, P. A., Acta Metallurgica, Vol. 24, 1976, p. 59.

[16] Hertzberg, R. W. and Mills, W. J. in Fractography--Microscopic Cracking Processes,

ASTM STP 600, American Society for Testing and Materials, 1976, p. 220.
[17] Cullen, W. H. and Stonesifer, F. R., "Fatigue-Crack-Growth Analysis of Titanium Gas-
Turbine Fan Blades," NRL Memorandum Report 3378, Naval Research Laboratory,
Washington, D.C., Oct. 1976.
[18] Yoder, G. R., Cooley, L. A., and Crooker, T. W., "A Micromechanistic Interpretation
of Cyclic Crack-Growth Behavior in a Beta-Annealed Ti-6AI-4V Alloy," NRL Report
8048, Naval Research Laboratory, Washington, D.C., Nov. 1976.
[19] Yoder, G. R., Cooley, L. A., and Crooker, T. W., "Enhancement of Fatigue Crack
Growth and Fracture Resistance in Ti-6AI-4V and Ti-6AI-6V-2Sn through Microstruc-
tural Modification," NRL Report 8049, Naval Research Laboratory, Washington, D.C.,
Nov. 1976.
[20] Vicki, F. J. in Titanium Science and Technology, R, I. Jaffee and H. M. Burte, Eds.,
Plenum Press, New York, Vol. 1, 1973, p. 733.
[21] Metals Handbook, 8th ed., Vol. 9, American Society for Metals, Metals Park, Ohio,
1974, p. 90.
[22] Seagle, S. R., Seeley, R. R., and Hall, G. S. in Applications Related Phenomena in
Titanium Alloys, ASTM STP 432, American Society for Testing and Materials, 1968,
p. 170.

Effect of the Cyclic Rate on Corrosion

Fatigue and Fractography of Type 304
Stainless Steel in Boiling 42 Percent
Magnesium-Chloride Solution

REFERENCE: Hioki, Susumu and Mukai, Yoshihiko, "Effect of the Cyclic Rate on
Corrosion Fatigue and Fractography of Type 304 Stainless Steel in Boiling 42 Percent
[Link] Solution," Fractography in Failure Analysis, A S T M STP 645,
B. M. Strauss and W. H. Cullen, Jr., Eds., American Society for Testing and Mate-
rials, 1978, pp. 144-163.

ABSTRACT: Corrosion fatigue tests of Type 304 stainless steel were conducted at
various cyclic rates between 0 and 104 rpm in boiling 42 percent magnesium-chloride
(MgCl2) solution in rotating bending. As a result, the following conclusions were ob-
tained: (a) the cyclic rate affects the failure life at the cyclic rates less than 103 rpm,
(b) the rate of crack propagation, da/dn, is proportional to the maximum stress in-
tensity factor, Kmax, that is, da/dn = C Kmaxm, where C and m depend on the cyclic
9

rate, (c) the fracture surfaces at higher cyclic rates and static loading were mostly of
the transgranular type, that is, step pattern and striation, but the intergranular type
was observed partly on the fracture surfaces at lower cyclic rates. Intergranular type
fracture surface was observed most frequently at 1 rpm.

KEY WORDS: corrosion fatigue, stress corrosion, cyclic rate, Type 304 stainless steel,
rotating bending, fractography

The resistance to the e n v i r o n m e n t a l l y assisted fracture has b e e n dis-

cussed as two big i n d e p e n d e n t problems, that is, stress corrosion cracking
(SCC) a n d corrosion fatigue [1-3]. 3 However, considering the actual en-
v i r o n m e n t , stress cycling, a n d materials used in products, m a n y strength
m e m b e r s are simultaneously subjected to SCC a n d corrosion fatigue. A few
works have been devoted to elucidate the general properties a n d the mech-
a n i s m of simultaneously occurring SCC a n d corrosion fatigue [3]. The ef-
fect of cyclic rate of applied stress seems to be essential for both d a m a g e s
1Doctor of Engineering, Mechanical Engineering Research Laboratory, Hitachi, Ltd.,
Ibaraki 300, Japan.
2Doctor and professor, Faculty of Engineering, Osaka University, Osaka 565, Japan.
3The italic numbers in brackets refer to the list of references appended to this paper.

144
Copyright by ASTM
9 Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright 1978by ASTMInternational [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
HIOKI AND MUKAI ON EFFECT OF CYCLIC RATE 145

for the combination of a certain material and environment. In this paper,

the effect of cyclic rate on corrosion fatigue of Type 304 stainless steel in
boiling 42 percent magnesium chloride (MgC12) solution was reported.

Experimental Procedure

Testing Apparatus and Method

The testing apparatus used in this study consisted of a rotary-bending
fatigue testing machine and a bath containing corrosive media as shown in
Fig. 1. The cyclic rate could be varied up to 104 rpm.
The corrosion bath was able to contain a 70-ram-long notched specimen
dipped in boiling 42 percent MgC12 solution. The leakage of the solution
through the gap between bath and rotary specimen was prevented by oilless
seals. The temperature and concentration of solution were kept constant at
143 ~ and 42 percent, respectively, by a heater and a condenser.
Corrosion-fatigue testing and fatigue testing in air at 143~ were done
using the apparatus just mentioned. In the case of the latter, the specimen
was covered by a glass tube in order to prevent environmental attack.

Material and Specimen

The specimens shown in Fig. 2 were machined from Type 304 stainless
steel sheet and solution heat treated for 2 h at ll00~ followed by water

.~f~CONDENSER
WATER ~cF~I
o
" tJ .__((

U HRATgR
SOLUTION BATH

WEIGHT
FIG. 1--Illustration of the testing apparatus.

NOTCHo It=_9r-,

I--- 125 =t: 120 '

FIG. 2--Specimen dimensions in miUimetres (stress concentration factor of the notch,
kt -- 2.3).

quenching. The chemical compositions and mechanical properties are given

in Table 1.

M e a s u r e m e n t o f Crack L e n g t h by the Electric Resistance

In order to measure the crack length at an arbitrary number of cycles

during testing, the electric resistance was measured with Type 304 stainless
steel rods between both ends of the notch of the specimen at room temper-
ature, removing load and corrosive media, which were given by a current
of 5 _ 0.2 A through both ends of the specimen, as shown in Fig. 3. There-
fore, no effects of corrosive media were assumed to exist on the measure-
ment of crack length.
The calibration curve for crack length and electric resistance is shown in
Fig. 4. It was obtained by measuring the crack length and the resistance of
several cracked specimens by rotary-bending fatigue in air and corrosive

TABLE1--Mechanical properties and chemical compositions of Type 304 stainless steel tested.
Chemical Composition, weight %

Material C Si Mn P S Cr Ni
Type 304
stainless steel 0.076 0.85 1.15 0.020 0.012 18.35 9.2
Mechanical Properties (at 143~
0.2% Proof Stress Tensile Strength Elongation,
(gage length = 25 mm)%
137.3 N/ram 2 423.6 N/ram 2 71.0

VABIABLE 1~818TEIt
BATTBItY
'H'I

ITEST SPBCIMP( 81K)N'T 4B~ S.2A

Fig. (b)/ , ~ l~ICll~- t ~ D.C.

81~CI~ (a)
(b)
FIG. 3--Circuit for measurement of crack length electric resistance.

:.100 /

1.0 /
/

" Lo
a (ram)
FIG.4--Calibration curve f o r crack length, a. and electric resistance, R.

media, 42 percent MgC12 solution. The crack length of the specimen was
obtained by measuring the temper colored part in 30-min beating at 600~
during which time the specimen stopped testing at an arbitrary number of
cycles, that is, crack length. Considering the measuring method and the re-
sults, the error of calibration curve seemed to be just reaching one, only
0.02 mm in this experiment. The contact resistance and the probable exis-
tence of MgCl2 solution at the crack tip, etc., had no effect on the data.

Stress Intensity Factor K of a Circumferentially Cracked Rotating

R o u n d Bar
Some analyses of stress intensity factor, K, of a rotating round bar are
given [4] in the following equation

K = x/Irra/([Link] + 7.1a) x a~t (1)

where
r = the distance between the crack tip and the center of the specimen,
a = crack length, and
o,~ = net stress for actual cross section.
However, the effect of crack closure in the compression side is not con-
sidered in this analysis. So, the authors derived the following equation
from Eq 1 on the assumption that the cross section of a circle of (2r + a)
diameter would resist to bending moment

/ (> + a)(t + a) (r + 3
K = 8x 0.8(2r + a) + 14.2(t + a) x \2r + aJ owo~ (2)

where
t = depth of notch, and
o~o,, = gross stress for the original cross section.
The bending stress, oN, at the notch root has been used to describe the
experimental results in this paper.

Experimental Results

Testing Results of Fatigue in Air and of Stress Corrosion Cracking in

Static Bending
Fatigue tests in air were done at room temperature and 143~ for the
cyclic rates of 1.5, 10, and 103 rpm. The results are shown in Fig. 5 and
fatigue life appears to be shorter at 143 ~ than at room temperature.
Stress corrosion cracking testing was conducted in static bending using
the corrosion fatigue testing apparatus, at a cyclic rate of 0 rpm. The re-
sults are shown in Fig. 6 and the threshold stress of SCC was 147 N/ram 2.
The testing stresses, 147, 196, and 245 N / m m 2, were selected for the cor-
rosion-fatigue testing based on the results just mentioned.

50O
4ool
"E 550'
.E 500
z 250
2OO
9 O 1,5rpm
150 . . olO J [
10~
R. T : ROOM TEMPERAT(~RE
100
10 3 10 ~ 10 ~ 10 ~ 10 7 10 8

NUMBER OF CYCLES TO FAILUR,E Nf ( c y c l e s )

F I G . 5--Fatigue strength of Type 304 stainless steel in air.

500.

200

w ~

IO0 I

10 = 10 = 10' 10 =

TIME TO FAILUI~, tf , (rain)

F I G . 6--Results of SCC o f Type 304 stainless steel in 42 percent MgCI 2 under static bending.

Results of Corrosion-Fatigue Tests

The results of corrosion fatigue at cyclic rates between 10 -4 and 10 4
rpm are shown as the relation between the number of cycles to fatigue,
N/, and the applied stress, aN, in Fig. 7.
The effect of cyclic rate can be demonstrated by plotting the number of
cycles against the cyclic rate as shown in Fig. 8. It can be seen that at the
cyclic rates lower than 10 3 rpm, the number of cycles to failure decreased
as the cyclic rate decreased. This suggests that the cyclic rate is effective on
the life of corrosion fatigue at these cyclic rates, which would provide
enough time for the material to be subjected to environmental attack. But,

500
25O
i
-~200
7

,t 150

lO0
* 5.5x 10 -4 o 1.7x 10 -2 9 1.7x 10 -I
@ 1.5 9 I x 10 9 1.5x 102
9 10 ~ eZ64x10~
i , i i

10 o 10' 10= 10 a 10' 10 s 107

NUMBER OF CYCLES TO FAILURE , Nf , (cycles)
FIG. 7 - - R e s u l t s o f c o r r o s i o n f a t i g u e o f T y p e 304 stainless steelin42percentMgCl2 in ro-
tating bending ( N f - aN curve).

107
196N/ram z IN AIR I
I0 s
Ax 145N/mm ~' I I I --L -~- - - ~

245N/mm ~ IN A I R / IO ~ "
I0 s C) 196N/ram ~ / ' ~ I~ I:~(~-"
' I /'~ "~o . . . .
[ ] 245N/mm2 I~ / I~""~. --
10 ~ . _ _

103

10=

I0',

10 i i i i i , , , ,

I0 -s I0 -4 10 -3 I0 -2 I0-' 100 I0' 10 ~ 103 104 10'

CYCLIC RATE f (tpm)
FIG. 8--Relation between number of cycles to failure, Nf, and cyclic rate, f.

the number of cycles to failure become constant at the cyclic rates higher
than 10 3 rpm, which is slightly less than the life in air, indicating that the
reduction of life due to environmental attack becomes constant.

Effect of Cyclic Rate on the Crack Initiation Life and Propagation

Properties
The behavior of crack initiation and propagation was studied by the elec-
tric-resistance method in order to clarify the effect of cyclic rate. Some of
the test results are shown as the relation between number of cycles, n,
crack length, a, and the resistance, R, in Fig. 9.
The effect of cyclic rate is shown as the relation between cyclic rate, f,
and the number of cycles to crack initiation, N i , in Fig. 10 on the assump-
tion that AT,.is at the number of cycles when a crack of 0.2 mm long is found,
as there is no significant difference between the onset of the first crack and
a 0.2-mm-long crack.
As shown in these figures, N, increased as the cyclic rate increased.
However, Ni would saturate to the value of SCC at a very slow cyclic rate.
At cyclic rates higher than 1.6 • 10-3 rpm, N i can be expressed as

Ni = A 9fo.73 (3)

O'N = 196N/ram 2
1.5 f= 1.5rpm 1 / 55
9 ~ / ! Q-
1.9 o ~ / 50

0.5 45

0.00 '
50 100
'
fl
(a)

ON= 245N/ram
21 /
1.5 ::l.5rpm/ / 55
E '~ / 50 ~-
.e
45

oo ~ 1'o ~'5 2'0 25

(b)
FIG. 9--Examples of the relation between crack length, electrical resistance, and number
of cycles.

,,~ ~w 10 s
o O'N : 1 9 6 N / w = V
~ lo'
~e ~ 10s
~1o 2 _ ./•/~
~ 10 ~ n~~O'N : 245N/ram2
~ ~ 10o
10-I . . . . . . . . .

10 -`= 10-2 10o 10 = 10'

Z
CYCLIC KATE f (tom)

FIG. lO--Relation between number o f cycles to crack initiation, Ni, and cyclic rate, f.

where
N i = number of cycles to initiation of crack,
f = cyclic rate, and
A = constant related to the applied stress, which is 730 for a stress of
196 N/ram ~ and 52.7 for 245 N / m m 2.
As the time to initiation of crack t; is equal to Ni/f, t l can be expressed as
follows

ti = A 9 f-0.27 (4)

where ti = time to initiation of crack.

The behavior of crack propagation may be discussed in terms of the rela-
tion between the cyclic rate, f , the rate of crack propagation, d a / d n , and
stress intensity factor, K . . . . as shown in Fig. 11. In this figure, the rate of
crack propagation is determined by the relation

(a~ - a , - l ) / ( n , - n,-1)

where aq is crack length at number of cycles, n q , of which data were shown

in Fig. 10, for examples. The rate of crack propagation increased abruptly
at a small Km~ value, indicating a rapid crack propagation just after the
initiation. This phenomena would be affected by the stress concentration of
the notch. After that, the rate of crack propagation would be described as
a function of Kmax in the following equation

da/dn = C. Kmax m (5)

where
da/dn = the rate of crack propagation,
Kmx = stress intensity factor, and

C, m = constants affected by the cyclic rate,f, where

C = 2.6 x 10 -8 x f-0.92

and

m = 4.8 x 10 -2 X f7.o

Effect of Cyclic Rate on Fractography

The effect of cyclic rate on the fractography was studied. No distinctive
features of the fracture surface of corrosion-fatigued specimens were ob-
served by naked eyes. However, various distinctive features of the fracture
surface were clarified by the continuous observation from the notch root
to the center of fracture surface by scanning electron microscopy. For ex-

10 t

10 o

10 ~ L ii II
:."

10 -2

10 -3
/
;i
CYCL
~,I'E f
[C

\
, +9 ~A o 8.9 x I 0-3

10 -4 vv 9 1.7x 10 -l

9 1(]

o~ 9 1.5x 102
10 s
AA
A** 9 103`
A A
A. 9 Z 6 5 x 103
10 -6 9 9 t , =* I

20 50 100
Km a x ( N/mm-~/2 )

FIG. ll--Relation between rate o f crack propagation rate, da/dn, stress intensity factor,
Kmax, and cyclic rate, f.

ample, typical fractographs of a corrosion-fatigued specimen at the stress

level, a~ = 245 N / m m 2, will be shown for the various cyclic rate, as fol-
lows.
At the cyclic r a t e f = 0 rpm, that is, at static loading, t y p i c a l f features
of fracture surface of SCC are shown as a continuous photograph from the
notch root to the crack length a = 0.8 mm in Fig. 12. No differences were
observed in the features from notch root to the center of the specimen. In
the figure, the fan-like pattern and the step pattern along the fracture di-
rection are shown and the transgranular fracture was subjected to the frac-
ture surface. These facts were the same results as in previous works [5].
At the cyclic r a t e f = 0.165 rpm, the fractographic features of a corrosion-
fatigued specimen were as follows: similar features to the static SCC were
observed from the notch root to the crack length a = 2.3 mm as shown in
Fig. 13(a), for example. On the region of the crack length longer than 2.3
mm, intergranular fracture surface was observed as shown in Fig. 13(b).
In these regions, fan-like patterns were partly observed on one grain site.
At the cyclic r a t e f = 1.5 rpm, the fractographic features had three types
of features. Fan-like patterns were observed from the notch root to the
crack length a = 1.2 mm, as shown in Fig. 14(a). On the region of crack
length, a = 1.2 to 2.0 mm, striations were observed, as shown in Fig. 14(b).
At the cyclic rate f = 10 rpm, the fractographic features were almost
similar to that at f = 1.5 rpm. The transition points were a = 1.6 mm
from fan-like pattern to striations, and a = 2.8 mm from striations to
rock candies. Typical striations, partly observed rock candy and dimple
pattern, are shown in Fig. 15(a), (b), and (c), respectively.
At the cyclic rate f = 150 rpm, the fractographic features were as fol-
lows: on the near region of notch root, transgranular fracture surface ac-
companied with fan-like pattern and ill-defined striations were observed,
as shown in Fig. 16(a). On the region of the crack length a = 0.9 to
2.4 mm, well defined striations were observed, as shown in Fig. 16(b).
The dimple patterns were observed on the crack longer than 2.4 ram.
At the cyclic rate f = 103 rpm, the fractographic features were almost
similar to the features at f = 150 rpm. On the near region of notch root,
transgranular fracture surface accompanied with fan-like pattern and ill-
defined striations were observed as shown in Fig. 17(a). On the region of
the crack length a = 0.7 to 2.1 mm, well defined striations were observed
in Fig. 17(b) and at final stage, a dimple pattern was observed, as shown
in Fig. 17(c).
The facts just mentioned are summarized as follows: (a) at static SCC,
transgranular fracture with a fan-like pattern was subjective, (b) at lower
cyclic rate, a simple transition phenomenon from transgranular fracture
to intergranular fracture was observed, (c) at medium cyclic rate, two
transition phenomena, that is, from fan-like pattern to striations and from
striations to rock candies were observed, and (d) at higher cyclic rate, tran-

o
0
0
"11
"o
"I"
.<

-7
.11
r-
t-
-11
m

z
I--
.<

if)

FIG. 1 2 - - E l e c t r o n m i c r o g r a p h o f t h e f r a c t u r e surface o f S C C at oN = 245 N / m m 2 (scale m a r k indicates 0.05 m m ) .

z
c3
E
E
)

0
z
m
"11
"11
rn
0
-t
0
"11
0
-<
0
r-

F I G . 13--Electron micrograph o f the fracture surface o f corrosion fatigue at oN = 245 N / m m 2 and f = 0.156 rpm (scale mark indicates 0.05 m m ) .
(a) a = 0. 6 m m a n d (b) a = 2 . 9 m m .
...4
m

-i1
:D

0
-4
0
0

-o
"1"
.<

-2
.11
),
B
r-
c
m

z
i-"
-<

F I G . 14--Electron micrograph o f the fracture surface o f corrosion fatigue at aN = 245 N / m m 2 and f = 1.5 rpm (scale mark indicates 0.05 m m ) . (a) a
= 0 . 9 m m , (b) a --- l . S m m , a n d ( c ) a -- 2 . g m m .

z
o

0
z
m
"11
m
0
-I
0
"11
0
.<

f--

F I G . 15--Electron micrograph o f the fracture surface o f corrosion fatigue at aN = 245 N / m m 2 and 10 rpm (scale mark indicates 0.05 ram). (a) 1.8 -.t
m m < a < 2.5ram, (b) a --- 3.1ram, and (c) a -- 3.7ram. m

Z
o

0
z
m
-I1
m
0
-4
0
"n
0
FIG. 16--Electron micrograph o f the fracture surface o f corrosion fatigue at N = 245 N / m m 2 and f = 150 rpm (scale mark indicates 0.05 mm). (a)
Omm < a < 0.56mm, and(b) a = 2.1mm. r

-.t
m

t.o

z
o

c
>_
0
z
m
"11
m
0
-4
0
"11
0
.<
0r-

F I G . 17--Electron micrograph of the fracture surface of corrosion fatigue at aN = 245 N / m m 2 and f = 103 rpm (scale mark indicates
O.05mm). ( a ) O m m < a < 0 . 6 m m , (b) a = 2 . 0 m m , and(c) a = 2 . 7 m m .
m

..L

1500

O"N = 1 4 7 N/rnm 2 ~ MPLE ON: 196N/ram2

I NT E P ~ R 2 L N ~ R / ~
1000 ~NTE RGRANU]z~ IMPLE ~I000
~AT1ON
RACKING k ~

500 500

E TRANSGRANULAR
TBANSGRANULAI~ CRACKING
CRACKING
J i i i i i = , 0 t i i t t i t t
10-3 10-2 10-' 100 101 102 103 104 10-3 10-2 10-' 100 10' 10 2 10 3 91 0 4

CYCLIC RATE f (rpm) CYCLIC RATE f (rpm)

I000 I NTEP~RANU- O'N= 245 N/ram 2

2
E

"~
z 500 TRANSGRANULAR " ~ T ~ I ATI ON
CRACK[ NG ""
E
0 i i i i i , i
10-~ 10.2 I0-' I0 ~ 10' 102 103 104
CYCLIC RATE f (rpm)

FIG. 18--Effect of cyclic rate and Kmax on the fractographic feature.

sition phenomena from transgranular fracture accompanied with fan-like

pattern and ill-defined striations to well defined striations were observed
and finally dimple pattern was observed. The summary just mentioned is
shown in Fig. 18 for the stress level, aN = 145, 196, and 245 N/ram 2.

Conclusion
Corrosion-fatigue tests of Type 304 stainless steel were conducted at var-
ious cyclic rates between 0 and 104 rpm in boiling 42 percent MgC12 solu-
tion in rotating bending. As a result, the following conclusions were ob-
tained: (a) the cyclic rate affects the failure life at cyclic rates less than
103 rpm; (b) the rate of crack propagation, d a / d n , is proportional to the
maximum stress intensity factor Km~, that is, d a / d n = C Kmax m , where C
and m depend on the cyclic rate; and (c) the fracture surfaces at higher
cyclic rate and static loading were mostly the transgranular type, but the
intergranular type was observed partly on the fracture surfaces at lower
cyclic rate. Intergranular-type fracture surface was observed more frequently
a t l rpm.

Acknowledgments

Authors express their thanks to H. Terasaki for his experimental assis-

tance throughout this work.

References
[1] Logan, H. L., The Stress Corrosion of Metals, WHey, New York, 1966.
[2] Beachem, C. D. and Brown, B. F. in Stress Corrosion Testing, A S T M STP 425, American
Society for Testing and Materials, 1966, p. 31.
[3] Wei, R. P. and Landes, ]. D., Materials Research and Standards, Vol. 9, No. 7, 1969, p. 25.
[4] Sih, G., Handbook of Stress Intensity Factors, Lehigh University, Bethlehem, Pa., 1973.
[5] Nielsen, N. A., Journal of Materials, Vol. S, 1970, p. 794.

Fractographic Observation of Stress-

Corrosion Cracking of AISI 304 Stainless
Steel in Boiling 42 Percent Magnesium-
Chloride Solution

REFERENCE: Mukai, Yoshihiko, Watanabe, Masaki, and Murata, Masato, "Fracto-

graphic Observation of Stress-Corrosion Cracking of AISI 304 Stainless Steel In Boil-
lug 42 Percent Magn~inm-Clfloride Solution," Fractography in Failure Analysis,
ASTM STP 645, B. M. Strauss and W. H. Cullen, Jr., Eds., American Society for
Testing and Materials, 1978, pp. 164-175.

ABSTRACT: The fracture surface of AISI 304 stainless steel in stress corrosion crack-
ing (SCC) in boiling 42 percent magnesium-chloride (MgC12) solution was investigated
fractographieally by scanning electron microscopy. To determine the orientation of
fracture surface, etch pits were formed on fracture surface by the electroetching method
in 1 N sulfuric acid (H2SO4) + 100 mg/litre ammonium thiocyanate (NH4SCN) solu-
tion. It was made clear from morphology of etch pit that fracture surface in SCC was
almost everywhere on the { 100} plane, namely, the path of the crack was determined
by crystal orientation and direction of stress. In addition, a striation-like pattern was
often found by careful observation on the flat surface.

KEY WORDS: fractography, orientation, fan-shaped pattern, step, tear ridge, stria-
tions, etch pit, facet, corrosion tunnel, stress intensity factor

Several fractographic studies have been m a d e on stress corrosion crack-

ing (SCC) of austenitic stainless steels by s c a n n i n g electron microscopy
[/].2 These observations have revealed some characteristic fracture surface,
n a m e l y , in t r a n s g r a n u l a r SCC the fracture surface was characterized by
fine parallel pleat patterns. Scully et al m e n t i o n e d these p a t t e r n s as traces
of corrosion t u n n e l s [1,2] a n d called t h e m f a n - s h a p e d patterns. However,
the m e c h a n i s m of f o r m i n g f a n - s h a p e d p a t t e r n s was not clarified in detail.
T h e n , in this research, the m e c h a n i s m of f o r m i n g these fracture surface
features was investigated crystallographically by m e a n s of fractographic ob-
servations with a s c a n n i n g electron microscope.
1Doctor and professor, Faculty of Engineering; doctor and professor, Welding Research
Institute; and Research associate, Faculty of Engineering; respectively, Osaka University,
Osaka 565, Japan.
2The italic numbers in brackets refer to the list of references appended to this paper.

164
Copyright by ASTM
9 Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright 1978by ASTMInternational [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
MUKAI ET AL ON AISI 304 STAINLESS STEEL 165

Materials and Procedures

The studies were carried out on AISI 304 stainless steel, with a composi-
tion as listed in Table 1. The steel was solution treated at l l 0 0 ~ (2012~
for 60 min. A bending-type testing machine was used, and the SCC tests
were made in magnesium-chloride (MgC12) aqueous solution, boiling at
143 ~ (289 OF).
Observations of fracture surfaces were made using a scanning electron
microscope. Fracture pieces were rinsed in water after removal from the test
environment, dried with methanol, and stored in desiccators.
To obtain a correlation between crystal orientation and fracture mode,
etch pits were formed on the fracture surfaces by an electroetching method
as follows.
The specimens were electroetched at - 0 . 1 5 V in 1 N sulfuric acid
(H2SO4) + 100 mg/litre ammonium thiocyanate (NH4SCN) solution at
40~ (104~ When the potential was established to - 0 . 1 5 V against a
saturated calomel electrode, pits were formed with { 111 } planes [3] as
shown in Fig. 1. The resulting morphology of pits formed on { 100}, { 110},
and { 111 } planes are shown in Fig. 2.

TABLE 1--Chemical composition of used material.

Material Chemical Composition, weight %

C Si Mn P S Cr Ni

AISI 304 0.076 0.85 1.15 0.020 0.012 18.35 9.2

I I
I
I
I

IT. I I~I/
I
I t
t I
J r

FIG. 1--Octahedron consisting of { l l I } planes.

FIG. 2--Typical etch pits.

Results and Discussion

Fan-Shaped Patterns
The fracture mode of AISI 304 stainless steel in SCC was mainly trans-
granular fracture. The fracture surface consisted of many facets and was
characterized by fan-shaped patterns as shown in Fig. 3.
Fan-shaped patterns were observed by tilting the specimen about 40 deg
in the scanning electron microscope (example shown in Fig. 4). It was
made clear that the fan-shaped pattern was formed with tear ridges and
SCC steps.

FIG. 3--Fracture surface of SCC of AISI 304 stainless steel in boiling 42 percent MgCI2
solution.

FIG. 4--Fan-shaped pattern.

The orientation relation between the direction of fan features and crack
growth direction was studied. To show the situation clearly at crack tips in
SCC, striation marks were put on the fracture surface by intermittently
applying fatigue loads to the specimen in the middle of the SCC test, caus-
ing the fan-shaped patterns to be interrupted intermittently by fatigue
striations (Fig. S). It became clear that the direction of crack growth coin-
cided with the direction of the fan, as sketched in Fig. 6.
Figure 6 also shows the frequency distribution of angles between fan and
macroscopic crack growth direction. It shows that macroscopic crack
growth direction corresponds statistically with the direction of the fans,
although microscopic crack growth directions vary. The crack initiation
point and crack growth direction may be determined by studying the direc-
tions of fans.
In addition, diameters of facets in fracture surfaces were in the range of
120 to 150 gm (Fig. 7), corresponding with the crystal grain size.

Orientation o f Crystals in Fracture Surfaces

Several theories have been proposed concerning crystallographic orienta-
tions of fracture paths in SCC. The { 100} theory [3,4], {110} theory [5],

FIG. 5--Relation betweenstriations offatigue and SCC steps.

r ~ ' T l ' ' ' ' l ' ' ' ' l ' ' i,

Fan-shaped-pattern
100

Total numbers : 6 0 2
~.50
LL
L 1__

7-
V
F
0 I [ I
F z_q I J
100 50 0 5O 100
e (deg)
FIG. 6=--Frequency distribution angle between f a n and macroscopic crack growth direction.

250 . . . . i ' ' ' '

20O
A A
A
D o o~ A
o~ 150
N ..... _~~ _ ~ ~ ~,~-o-~-~% -~ ~-.
o oz~ o
// s
~ 100
o /
o / Grain size av. = 131.86 a ~ l l m ) Kin( k ( ] - m m -3/2
u_
o 3 40
5C -~ 6 19.6
-o 9 40
I I ,,,
O0 , I , I
1 2 3
K ( !07" N'm -3/2)
FIG. 7--Relation between facet size and K value. B = thickness o f specimens and K in =
initial stress intensity factor.

{ 111} theory [6, 7], and others about crystallographic orientations of frac-
ture surfaces scarcely have been proven until now.
The crystallographic orientations of fracture surfaces in SCC were in-
vestigated by forming etch pits on fracture surfaces. Figure 8 shows the
morphology of pits on a fracture surface. It became clear that the fracture
planes were { 100} planes almost everywhere, and the crack growth direc-
tion was < 110> by studying the relation between morphology of pits and
directions of fans. Fracture planes were parallel to the { 100} planes even
across twin and grain boundaries.
The mechanism of forming fan-shaped patterns is illustrated in Fig. 9(a)
and (b).
When the maximum principal stress acts vertically to a { 100} plane, the
crack surface shows a flat plane. When the crack crosses a twin boundary,
the {100} plane is tilted and the resulting crack surface contains con-
siderable steps. Occasionally, secondary cracks are formed at steps. The
step planes are perpendicular to a { 100} plane, because they are formed
along the { 100} plane, too.
Consequently, the step plane becomes a part of the crack growth plane.
Occasionally, tear ridges were found on step surfaces formed by shear
stresses acting on their planes.

FIG. 8--Morphology of etch pits on fracture surface of SCC.

FIG. 9--(a) Fractograph o f fracture surface at twin boundary, and (b) schematic view o f
crack growth on twin boundary.

Fan-shaped patterns are not traces of corrosion tunnels, but consist of

fiat facets containing fine steps and a few tear ridges. Fracture surfaces
were parallel to {100} planes. From these experimental results, the
mechanism of SCC can not be explained solely by a corrosion process. It is
necessary to consider mutual effects of both corrosion processes and
mechanical fracture processes to understand the mechanisms of SCC.

Striation-Like Pattern
Striation-like patterns were found almost everywhere to be perpendicular
to the directions of the fans. Figure 10 shows matching striation-like pat-
terns and their stereo pairs. Figure 11 illustrates the schematic view of a
cross section of the striation-like pattern constructed from observations of
Fig. 10 and shows how the striation-like pattern found in SCC at static
loading matched convex and concave surfaces.
The effects of stress intensity factors, K, on forming striation-like pat-
terns were investigated, as shown in Fig. 12. It is clear that intervals of
striation-like patterns did not depend on stress intensity.
Figure 13 shows the relation between the morphology of pits and stria-
tion-like patterns, and Fig. 14 illustrates the schematic view of Fig. 13.

FIG. l O--Matching stereo pairs of a striation-like pattern.

Crack growth
FIG. 11--Schematic view o f section of a striation-like pattern.

I 1 i I l I l I I

5'
llJ O O
9 O
4-

~ (D i
"03-
>
% Oo0
~V

.c_ 2 -
c
.(2_
~t
A~
go A
3
6
30
~ 0

32
1
' l 1

'O 9 30
O0 I I I 1 I I
1 2 3
K([Link]-~2)
FIG. 12--Relation between the stress intensity factor and intervals of striation-like pat-
terns. B = thickness of specimens and K in = initial stress intensity factor.

From these observations, the characteristics of striation-like pattern in

SCC are as follows.
1. Intervals of striation-like patterns did not depend on stress intensity.
2. The direction of stripes in striation-like patterns was < 110> which
was perpendicular to crack growth directions.
3. Striation-like patterns were found locally.
Consequently, striation-like patterns were found when the maximum
principal stress activated slip along the most favorable slip planes and slip
directions in the crystals.

Conclusion
A fractographic study was made on AISI 304 stainless steel fractured by
SCC. The results are summarized as follows.
1. The direction of fans in fan-shaped patterns coincided with crack
growth directions.
2. Average diameters of facets were almost the same as average diame-
ters of the crystals.

FIG. 13--Relation between morphology of pits and a striation-like pattern.

<110>
I
Crack growth direction
I
/_&
., c'o~ -r:--
T .or ridge -x( - - -
-f

FIG. 14--Relation between slip Hnes and orientation of crystal grain.

3. Fracture surfaces were almost everywhere oriented on the {100}

plane, and crack growth directions were < 110 >.
4. Striation-like patterns that were found locally on fracture surfaces
were fine slip lines formed perpendicular to directions of crack growth.

Acknowledgments

W e a c k n o w l e d g e t h e c o o p e r a t i o n o f G . T s u j i i , a s t u d e n t at O s a k a U n i -
versity, who was h e l p e d to s t i m u l a t e progress.

References
[l] Harston, J. D. and Scully, J. C., Corrosion, Vol. 25, 1969, p. 493.
[2] Harston, J. D. and Seully, J. C., Corrosion, Vol. 26, 1970, p. 387.
[3] Otani, N., Aihara, K., and Takamoto, S., Journal of the Japan Institute of Metals, Vol.
33, 1969, p. 432.
[4] Denhard, E. E., Jr., Masters thesis, Johns Hopkins University, 1957.
[5] Bakish, R. and Robertson, W. R., Acta Metallurgica, Vol. 4, 1956.
[6] Pickering, H. W. and Swann, P. R., Corrosion, Vol. 19, 1963, p. 373t.
[7] Nielsen, N. A., Corrosion, Vol. 20, 1964, p. 104t.

Metallurgical Characterization of the

Fracture of Aluminum Alloys

REFERENCE: Bhandarkar, M. D. and Lisagor, W. B., "Metallurgical Charaeteriza.

tton of the Fracture of Aluminum Alloys," Fractography in Failure Analysis, ASTM
STP 645, B. M. Strauss and W. H. Cullen, Jr., Eds., American Societyfor Testing and
Materials, 1978, pp. 176-209.

ABSTRACT: A systematic investigation was conducted to examine the fracture be-

havior of the structural aluminum alloys 2024, 6061, 7075, and 7178 (in selected heat
treatments) tested under several controlled conditions. The investigation included both
time independent (tensile, shear, and precracked notch-bend) fractures and time de-
pendent (fatigue and stress corrosion) fractures. Specimens were obtained from both
sheet and plate material and tested in longitudinal and transverse orientations. Strain
rate effects on fracture morphology were examined in tension and shear tests. Fatigue
fracture studies included an examination of the influence of minimum-to-maximum-
load ratio on fracture morphology. Second-phase particles observed on fracture sur-
faces and metallographieally prepared sections and corrosion products associated
with stress corrosion fractures were analyzed chemically using scanning electron
microscopy and energy-dispersiveX-ray analysis. Fracture morphology was related to
the microstructural features, the testing conditions, and the form of commercial
product. The characteristic and distinguishing fracture features were reported for the
different alloys. The information obtained from the systematic and controlled studies
described here should prove useful in the analysis of service failures that often occur
under complex service conditions.

KEY WORDS: fractures (materials), tensile properties, shear stress, bending, fatigue
(materials), stress corrosion, fractography, microstructure, aluminum alloys

F r a c t o g r a p h i c analysis of microscopic fracture processes has a s s u m e d

increasing i m p o r t a n c e in studies of material failures occurring in service.
T h e several different types of load-related failures t h a t occur i n structural
a l u m i n u m alloys can be g r o u p e d u n d e r two b r o a d categories, namely,
those t h a t occur u n d e r m o n o t o n i c (static tensile, shear, a n d flexural)
l o a d i n g a n d those that occur in a time d e p e n d e n t fashion (resulting from
cyclic stresses or sustained loads in the presence of a corrosive environ-
m e n t , or both). Several investigations have b e e n c o n d u c t e d in the past to
1Senior metallurgist, Anamct Laboratories, Inc., Berkeley, Calif. 94710.
2Materials research engineer, NASALangley Research Center, Hampton, Va. 23665.

176
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 by ASTMInternational [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
BHANDARKAR AND LISAGOR ON ALUMINUM ALLOYS 177

study the fracture behavior of aluminum alloys under both monotonic and
discretely changing load conditions. Attempts have been made to relate
fracture behavior to microstructure and testing conditions [1-28]. 3 How-
ever, most investigations of this type have been performed as a supplement
to a broader program with fracture conditions selected or obtained for
reasons other than fractographic analysis. As a result, the information
available generally is not developed systematically with respect to the
variables contributing to the fracture phenomenon. In addition, documen-
tation is needed of results that can be obtained from the now widely used
techniques of scanning electron microscopy and energy-dispersive X-ray
analysis.
The problems just cited pose difficulties in analyzing service failures
because component failures often occur under complex loading and
environmental conditions. Fracture analysis of such failures requires a
detailed understanding of fracture appearance resulting from each of these
conditions and how their interaction influences the fracture surface ap-
pearance of the failed component.
A recent effort which should aid in the solution of these problems
involved the documentation of tensile, fatigue, stress rupture, thermal
fatigue, and stress-corrosion fracture morphologies of some aluminum and
titanium alloys, steels, and superalloys [29]. The present investigation was
conducted to examine in detail the microstructure and fracture morphology
of selected aluminum alloys tested under several controlled conditions with
extensive use of the scanning electron microscope and associated X-ray
chemical analysis. The influence of test variables, microstructure, and alloy
product form on fracture morphology are presented, and the applicability
of the results to failure analysis is discussed.

Experimental P r o c e d u r e

Materials
The investigation included aluminum alloys 2024, 6061, 7075, and 7178
in sheet and plate forms. The alloys were selected because of their wide-
spread use in aerospace structures. Sheets were 1 mm thick and plates
were 25.4 mm thick. Chemical compositions of the alloys are listed in
Table 1.

Specimens and Test Procedures

Tension tests were conducted using 12.7-mm-diameter round specimens
machined from plate material, and sheet specimens 12.7 mm wide in
accordance with ASTM Tension Testing of Metallic Materials (E 8-69).
3The italic numbers in brackets refer to the list of referencesappendedto this paper.

-n

"0
"1"
-<

TABLE 1--Chemical compositions of alloys.

-11
Compositions, Weight Percent
r-
C
Alloy Cu Mg Zn Si Fe Cr Mn Ti Others AI "11
m
2024 3.8 to 4.9 1.2 to 1.8 0.25 a 0.50 0.50 0.10 0.30 to 0.90 0.15 remainder
z
6061 0.15 to 0.40 0.8 to 1.2 0.25 0.4 to 0.8 0.70 0.15 to 0.35 0.15 0.15 0.15 remainder
7075 1.2 to 2.0 2.1 to 2.9 5.1 to 6.1 0.50 0.7 0.18 to 0.40 0.30 0.20 0.15 remainder i-
.<
7178 1.6 to 2.4 2.4 to 3.1 6.3 to 7.3 0.50 0.7 0.18 to 0.40 0.30 0.20 0.15 remainder
cD

a Single compositions represent the m a x i m u m a m o u n t s allowed.

Strain rate conditions included: (a) 0.00008/s to failure, (b) 0.025/s to

failure, and (c) 0.00008/s to yielding followed by 0.0008/s until failure.
Shear tests were performed at crosshead displacement rates of 0.04 and
1.27 mm/s using thin sheet specimens [30] as shown in Fig. l(a). All ten-
sion and shear tests were performed using both longitudinal and long
transverse specimens. Three-point bend tests were performed on specimens
machined from plates of each alloy and were tested with the plane of the
crack normal to both longitudinal and transverse directions with propaga-
tion in the short transverse direction (Fig. l(b)). Fatigue tests were con-
ducted using longitudinal and transverse sheet specimens containing a cen-
trally machined (electrical discharge machining) slot as shown in Fig. l(c).
Tests were conducted in room temperature air at minimum-to-maximum-
load ratios (R values) of 0.05 (at nominal maximum stresses of 51, 64, and
255 MPa), 0.8 (at nominal maximum stress of 255 MPa), and - 1 (at
nominal maximum stresses of 102 and 126 MPa). Stress-corrosion tests
were performed using C-ring specimens stressed to 75 to 95 percent of yield
strength and tested in accordance with ASTM Recommended Practice for
Alternate Immersion Stress Corrosion Testing in 3.5% Sodium Chloride
Solution (G 44-75) except that specimens were exposed until fracture
occurred. Details of specimen dimensions and orientations with respect to
the rolling direction are shown in Fig. l(d).

Metallurgical and Fractographic Analysis

Microstructure and fracture surface morphology were characterized
using the optical microscope and the scanning electron microscope.
Second-phase particles that were observed on fracture surfaces and
metallographically polished sections of the alloys were analyzed using the
scanning electron microscope and energy-dispersive X-ray chemical
analysis (EDAX) [31]. Conventional metallographic procedures were used
for microstructural analysis and specimens were etched with a reagent (0.5
percent hydrofluoric, 2.5 percent nitric, 1.5 percent hydrochloric acids,
balance water).

Results

Mechanical Property Determinations

Tensile and shear properties of sheet material and tensile and fracture
toughness properties of plate material are shown in Table 2. Stress intensity
factors (KI) at failure were calculated using the K expression in ASTM
Test for Plane-Strain Fracture Testing of Metallic Materials (E 399-74).
The tensile values are indicative of typical values obtained from conventional
commercially produced product form [32]. Effects of specimen orientation
were also typical and an order of magnitude difference in strain rate had

~ ---~1 ROOTRADIUS=
-- 0.127

6.36TI" '!- - I--~-

-25.44
L LZ5.4-4

(o) SHEAR SPECIMEN (SHEET) (b) NOTCH-BEND SPECIMEN(PLATE)

t /SEE, DETA,L RADIUSNOT ~SPECIMEN

CRITICAL ~ . ~ TRANSVERSE
C.L.
~=~]~i~ SPECIMEN
c,~ TJ~-- LCNGITUDIN~_
~--,~o3- ~
-320,6 .l' i'~_o.,~.L.
-~ ~--0.254
DETAIL A

(C) SHEET FATIGUE SPECIMEN

/- . . . . . . . . -a

I" THICK
PLATE- ~
//,~ { ~ l

, ,,@y/

/z //ii

~o~ ~,~,~ I
i I //
/

I I/

(d) STRESS CORROSION SPECIMEN (PLATE)

DIMENSIONS IN MM
FIG. 1--Specimen dimensions and orientations.

little effect on tensile behavior. However, variation in crosshead displace-

ment rate (and hence, load rate) had a substantial effect on shear strength
of all alloys with higher load rates causing 25 to 55 percent lower shear
strengths. These differences and their effect on fractographic appearance
will be discussed under fracture morphology of sheet shear specimens.

Microstructures of Alloys
The characterization of microstructure (including grain structure, second-
phase particle distribution, and X-ray chemical analysis) was used to relate
the effects of these microstructural variables on fracture morphology where
possible.
Figure 2 shows optical and scanning electron microscope photomicro-
graphs of 2024-T351 plate and visual displays of the X-ray displays of three
distinct second-phase particles identified on the polished sections. These
results are indicative of the type of information obtained for all alloys in
both plate and sheet form. The photomicrographs reveal the elongated grain
structure and second-phase particle distribution. The X-ray displays reveal
the relative intensities of the elements present in the specific particles.
These data must be corrected for matrix contribution caused by penetration
of the electron beam beyond the particle and into the matrix, which causes
portions of the display to result from the irradiated matrix.
After matrix corrections and comparison with identified particles in
the literature [31,33,34], Particles A and C in Figs. 2(b) and (c) were iden-
tified as copper aluminide (CuA12) and a metal aluminide of the form (MA16)
with copper, iron, and manganese in substitution (Cu, F.e, Mn) AI6. In
addition to finding particles previously identified by other investigators, all
alloy plates and sheets were found to contain complex particles not reported
previously. The stoichiometry of these particles could not be determined,
but identification was positive based on constant elemental intensity ratios
obtained on several particles. Table 3 lists particles identified in each alloy
plate or sheet, whether or not it has been reported previously, and the frac-
ture surfaces on which it was found. Particles for which stoichiometry could
not be identified are shown with all element symbols contained within the
parentheses.
The presence of second-phase particles and their condition (whether
they were cleaved or fragmented) can indicate their contribution to the final
fracture process. The absence of fractured second-phase particles, the pre-
sence of voids adjacent to particles that are not fractured, and the presence
of unbroken particles in relief on fracture surfaces suggest particle-matrix
separation only, with little plastic flow around particles. Also, the presence
or absence of fractured particles suggested magnitudes of loading in shear
and fatigue tests, discussed in later sections.
The microstructure of all plates were similar with regard to grain struc-

"11
2D

TABLE 2--Properties of alloysinvestigated. o

Yield Elonga- Fracture Ultimate

Strength, Ultimate tion, Tough- Shear "1"
Alloy a n d Federal Product Orienta- 0 . 2 % Off- Strength, % in ness, Strength,
Specification Form tion set, M P a MPa 5 cm M P a 9 m 1/2 MPa
"11
2024-T351 plate L 423 501 20 40 ...
QQ-A-250/4d T 370 484 18 36 . ,. r-
c
2024-T3 sheet L 362 500 24 . .. 356 a
199 b m
QQ-A-250/4d T 314 476 27 ... 357 a z
207 b
6061-T4 plate L 205 256 26 . . . . . .
QQ-A-2S0/lld T 201 277 22 33
6061-T4 sheet L 180 281 27 ... 2"1"7a
167 b
QQ-A-250/11d T 170 276 30 . .. 216 a
170 b
6061-T651 plate L 305 318 22 44 ...
QQ-A-250/1 ld T 303 328 16 44 ...
6061-T6 sheet L 301 331 22 ... 235a
174 b

Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
QQ-A-250/lld T 285 325 16 ... 239" "s-
185 b
7075-T651 plate L 557 594 12 41 9 Z
0
QQ-A-250/12d T 539 595 13 36 c 999
ms
7075-T6 sheet L 519 567 23 ... 401a
178 b
QQ-A-250/12d T 588 663 17 ... 405 a :33
186 b
Z
7178-T651 plate L 589 636 11 29 c . .. 0
QQ-A-250/14d T 564 619 11 23 r ...
7178-T6 sheet L 568 617 19 ... 424 a O~
210 b
QQ-A-250/14d T 550 636 19 ... 443" 0
203 b
0
NOTE--Aluminum Alloy, Plate and Sheet (QQ-A-250/4d) Z
Aluminum Alloy, Plate and Sheet ( Q Q - A - 2 5 0 / l l d )
r
A lu m in um Alloy, Plate and Sheet, Alclad (QQ-A-250/12d) C
A lu m in um Alloy, Sheet and Plate (QQ-A-250/14d)
a Tested at crosshead speed of 2.54 m m / m i n . Z
b Tested at crosshead speed of 76.2 m m / m i n .
c Meets ASTM Test E 399-74 validity criteria for valid KIc.
r-
t-
O
-<

0
z

FIG. 2--Optical micrograph (a) and scanning electron micrographs (b) and (c) of a plate specimen of 2024-T351. The E D A X pattern from second f-
c
phase particles A, B, and C are shown in (d), (e), and (f), respectively.

r-
r-
0

Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
TABLE 3--Second-phase particle identification in aluminum alloy fracture surfaces.
O0
Particles Identified by Energy Previously Identi- Fracture Surfaces on
Alloy Product Form Dispersive X-ray Analysis fled in Alloy Which Identified
"n
2D
2024-T351 plate CuA12 yes tensile, notch bend
0
(Cu, Fe, Mn) AI6 yes tensile, notch bend --4
(Cu, Fe, Mn, Si, AI) no tensile, notch bend 0

2024-T3 sheet CuAI2 yes tensile, shear, fatigue

(Cu, Fe, Mn) AI6 yes tensile, fatigue -0
(Cu, Fe, Mn, Si, AI) no tensile, shear, fatigue I
-<
6061-T4, plate Mg2Si yes tensile, notch bend
6061-T651 Si yes tensile, notch bend
(Fe, Si, Cr, Mn, AI) no tensile, notch bend
6061-T4, sheet (Fe, Cr, Mn, AI) no tensile, shear, fatigue
6061-T6 c
m
7075-T651 plate (Fe, Cu, Cr, Zn, Mn, A1) no notch bend, tensile
(Cu, Fe, Zn, Cr, Mn, AI) no tensile, notch bend z
(Fe, Cr, Si, Mn, AI) no notch bend
Mg2Si yes tensile, notch bend .<
CuMgAI2 yes notch bend
(Mg,Si,Zn,Cu,A1) no notch bend
7075-T6 sheet FeAI3 yes tensile, shear, fatigue
Cu 3ZnA13 yes tensile
(Fe,Cr,Mn,Cu,Si,AI) no shear, fatigue
Mg 2Si yes fatigue
CuMgAI 2 yes tensile, fatigue
7178-T651 plate CuAI2 yes tensile, notch bend
CuMgAI 2 yes tensile, notch bend
(Fe,Cu,Zn,Cr,Mn,AI) no tensile, notch bend
7178-I"6 sheet FeA13 yes tensile, fatigue
Mg 2Si yes fatigue
(Cu,Zn,Mg,AI) no tensile
(Cu,Fe,Zn,Cr,Mn,AI) no shear, fatigue

ture and second-phase distribution; however, more banding of small second-

phase particles was observed in 7075 and 7178 plates. In general, second-
phase distribution in sheet materials was more random with little or no
banding or segregation. Some of the particles identified by X-ray analysis
are of the type which contribute to the precipitation strengthening process
during heat treatment. However, when present in large sizes (Fig. 2(b) and
(c)), they are probably the result of incomplete dissolution during solution
treating. The materials, however, were still within mechanical property
tolerances required by the specifications listed in Table 2.

Fracture Morphology of Alloys

Plate Tension Specimens--The fracture morphology in plate tension
specimens is illustrated by the scanning electron fractographs of Fig. 3.
Strain rate variations within the range examined and specimen orientation
did not affect fracture morphology.
'The dimpled rupture shown in Fig. 3 was the predominant failure mode
in alloys 2024-T351, 6061-T4, and 6061-T651. The large dimples contained
cleaved or shattered second-phase particles. Dimples on fracture surfaces
of 6061-T4 and 6061-T651 tension specimens were more equiaxed and ex-
hibited an increasing degree of size regularity. Nonuniform dimple sizes
and irregular dimple shapes were common in alloy 2024-T351 as illustrated
in Fig. 4. Areas with fine dimples, nucleated possibly at aging precipitates
or dispersoids, or both, are marked A in the figure.
Plate tensile fracture of 7075-T651 and 7178-T651 exhibited a predomi-
nance of relatively smooth and flat facets (Figs. 3(d) and (e), and Fig. 5(a)),
which, on high magnification examination, revealed shallow, submicron-size
dimples (Fig. 5(b)). The areas between~smooth facets exhibited ductile tear-
ing and pockets containing cracked second-phase particles. The smooth
appearance of the facets and their approximate 45-deg orientation to the
applied load suggested that the facets were stretched regions formed by ex-
tensive tearing along glide planes. Optical microscopy of sections inter-
secting the fracture surfaces indicated that the facets were formed by trans-
granular fracture. The fine shear dimples observed on the smooth surfaces
were of the proper approximate size range to have been nucleated by strength-
ening precipitates or dispersoids, or both.
Sheet Tension Specimens--Dimpled rupture was the prime failure mode
in sheet tension specimens of all the alloys of the present investigation, as
is evident from the scanning electron fractographs in Fig. 6. Strain rates
within the range examined and specimen orientations did not affect frac-
ture morphology.
Generally, all specimens contained equiaxed dimples in local fracture
areas normal to the applied load and elongated dimples resulting from com-
bined shear and normal stresses in fracture areas oblique with respect to

7~
3~
:70

Q
0

0
z

P
c
FIG. 3--Scanning electron fractographs of tensile plate specimens of aluminum alloys: (a) 2024-T351, (b) 6061-T4, (c) 6061-T651, (d) 7075-T651, z
and (e) 7178-T651.

==&
oo
(O

FIG. 4--Scanning electron fractographs of tensile plate specimens of 2024-7351.

FIG. 5--Scanning electron fractographs of a tensile plate specimen of 7178-"1"651.

the applied load. Regions that failed under a predominantly shear stress
exhibited fracture features similar to those in shear test specimens as would
be expected. Sheet tensile fracture morphology was essentially similar in
the different alloys. In general, a finer, more uniform dimple size was ob-

f.-

G)
0

0
z
>
FIG. 6--Scanning electron fractographs of sheet tension specimens of aluminum alloys, (a) 2024-T3, (b) 6061-T4, (c) 6061-T6, (d) 7075-T6, and (e) P
c
7178-T6.
Z
G:

>
r=.
f-
0..<
(I)

served compared to plate material, and less second-phase particles were

present on the fracture surfaces of sheet specimens.
Sheet Shear Specimens--Sheet shear fractures were essentially similar
for all the alloys tested at the crosshead displacement rate of 0.04 mm/s in
the present investigation. MacroscopicaUy, the fractures were smooth.
Microscopic appearances were characterized by large shallow elongated
dimples and smooth areas, as illustrated by the scanning electron fracto-
graphs in Fig. 7. Examination of the smooth areas at higher magnification
revealed fine, shallow dimple formation, possibly nucleated by strengthen-
ing precipitates or dispersoids, or both (Fig. 8). When the crosshead dis-
placement rate was changed to 1.27 mm/s, the shear ultimate strengths
were lowered by 42 to 56 percent in alloys 2024-T3, 7075-T6 and 7178-T6,
and by 21 to 26 percent in alloys 6061-T4 and 6061-T6 (see Table 2). The
shear fracture morphology was not affected in the 6061 material, but the
faster strain rate produced fractures in alloys 2024-T3, 7075-T6, and 7178-T6
which were rougher and contained more cracked particles than slow strain
rate shear fractures (Fig. 9). Specimen orientation with respect to the rolling
direction did not influence either shear fracture morphology or shear ulti-
mate strength in the sheet alloys investigated.
Notch-Bend Specimens--The microscopic fracture morphology of notch-
bend specimens of alloys was essentially similar to the plate tensile frac-
tures. The major difference between appearances of tensile and notch-bend
fractures occurred in alloys 707S-T6S1 and 7178-T651 where secondary
cracks and deep grooves were observed only in notch-bend specimens (Fig.
10). Smooth facet formation was more regular (presumably due to
controlled crack propagation) in notch-bend specimens than in plate ten-
sion specimens.
Sheet Fatigue Specimens--The results described herein characterizing
fatigue fracture morphology are limited to that area of fracture surface
which was produced by fatigue crack growth. The areas associated with the
fast fracture portion of failure were similar to sheet tension specimens with
more tendency to be planar.
At R = 0.0S and nominal maximum stresses of S1 and 64 MPa, fatigue
fractures exhibited relatively smooth areas separated by ductile tear ridges
and voids formed by particle matrix separation as illustrated in Fig. 11
(voids identified by A). Void formation and tearing were more dominant in
2024-T3, 6061-T4, and 6061-T6 (Fig. ll(a), (b), and (c)) than in 7075-T6
and 7178-T6 (Fig. ll(d) and (e)). The latter two alloys exhibited a pre-
dominance of relatively smooth areas containing poorly defined striations
and slip band markings that were not, in most instances, clearly distin-
guishable from each other (see areas near B in Fig. 11). The rare occur-
rence of cleaved particles in voids, the presence of unbroken particles adja-
cent to some voids and in relief on the fracture surfaces suggested that
dimples were nucleated mostly by particle-matrix interface failure at

maximum stress values in this range (unbroken particles are illustrated at

C in Fig. 11). Specimens tested at R = 0.05 and a nominal maximum
stress of 255 MPa exhibited greater proportions of void formation and
ductile tearing than in fatigue fractures of specimens tested at the lower
maximum stresses. Particle cracking also was observed at the higher max-
imum stress value. These results are consistent with the findings of other
investigations [6,12,18,20].
Fatigue fractures of specimens tested at R = 0.8 and nominal maximum
stress of 255 MPa were similar to high maximum stress, low R value frac-
tures and they also exhibited particle cracking. Striations were observed
even less under these stress conditions. Specimens tested under fully
reversed loading were similar in fracture morphology to constant amplitude
tension-tension specimens, except for some contact damage resulting in
local areas of surface smearing.
Scanning electron microscope fracture morphology normally associated
with fatigue failure is most often characterized by identification of stria-
tions. In sheet materials, ductile tear ridges are a far more prominent char-
acteristic on the fracture surface. Striations are often difficult to distinguish
from slip band formation and are often poorly defined. The reporting of
striations as the prominent fracture feature in fatigue failure of thin gage
materials is often misleading.
Stress-Corrosion Specimens--The prominent fracture surface charac-
teristic associated with stress-corrosion specimens is corrosion leaf forma-
tion which is illustrated in Fig. 12 by the scanning electron microscope
fractographs of a 2024-T351 specimen tested in the short transverse direc-
tion (orientation 1, Fig. 1). Corrosion leaves result from intergranular
cracking along grain boundaries parallel to and nearby the primary crack
front. The intergranular cracking is revealed more clearly in Fig. 13 by the
optical micrograph of a section approximately normal to the fracture
surface. Intergranular stress corrosion cracking (SCC) of this type is com-
monly encountered in aluminum alloys [3,13,16,22]. The degree of leaf
formation is affected by alloy susceptibility and orientation. It may also be
related to magnitude of stress.
On a microscopic scale, relatively smooth intergranular facets were pre-
dominant on stress-corrosion fractures. The fracture surfaces were coated
with corrosion products which could be classified in three general cate-
gories including nonadherent particle debris, adherent dried mud-flat
crack patterns, and adherent continuous coating. Loosely adherent
corrosion debris visible in Fig. 12(a) was removed upon extended ultrasonic
cleaning of the specimen, as shown in Fig. 12(b).
Typical higher magnification appearance of intergranular facets and cor-
rosion products are illustrated in Fig. 14 by the scanning electron fracto-
graphs of a 7178-T651 alloy specimen tested in the short transverse direc-
tion (orientation 1, Fig. 1). The fractograph in Fig. 14(b), obtained by

z
o
-11

.11
)>
z

Q
0

0
z
3>
c
FIG. 7--Scanning electron fractographs of sheet shear specimens, tested at the crosshead separation rate o f 2.54 mm/min, o f aluminum alloys: z
(a) 2024-T3, (b) 6061-T4, (c) 6061-T6, (d) 7075-T6, and (e) 7178-T6. c

t-
O
.<

.=&
cD
-4

FIG. 8--Scanning electron fractograph of a 2024-73 sheet shear specimen, tested at the
crosshead separation rate of 2.54 mm/min.

magnifying the area near A in Fig. 14(a), illustrates the dried mud-flat pat-
terns in the corrosion product. The fractograph in Fig. 14(c), obtained by
magnifying the area B in Fig. 14(a), illustrates the smooth corrosion
product coating with faint dried mud-flat cracks in some areas. X-ray
chemical analysis indicated the loosely adherent particulate debris con-
tains chlorides, probably aluminum and sodium chloride salt deposits from
the alternate wet-dry cycles. Neither cracked nor smooth deposits
contained chlorides suggesting they were hydrated or oxide products of
aluminum. X-ray patterns obtained on both smooth and cracked corrosion
products were similar. Cracked products could be the result of excessive
thickening of smooth products and subsequent cracking during the drying
process.

Diseusslon

Interrelationships of Fracture Variables

Alloy Composition and Microstructure--Alloy composition had a limited
effect on fracture morphology as compared with other test variables. The
fracture appearances seemed to group into two categories with the 2000
and 6000 series alloys exhibiting similarity and the two 7000 series alloys
appearing similar. The observed appearance of the two groups in plate
product form (dimpled structure versus ductile tear ridge formation)

FIG. 9--Scanning electron fractographs of 7075-T6 sheet shear specimens tested at cross-
head separation rates of(a)2.54 mm/min and (b) 76.2 ram~rain.

FIG. l O--Scanning electron fractographs of a 7075-]'651 notch-bend specimen.

appears to be the result of both generic alloy composition differences and

the effect of microstructure (banding, segregation). This grouping also fol-
lowed for the sheet fatigue specimens, but it generally did not follow for
shear or stress-corrosion specimens.

Large second-phase particle distribution also contributed to fracture ap-

pearance in plate tension and notch-bend specimens and sheet fatigue
specimens. Dimple size and uniformity in 6061 alloy was attributed to the
dense, homogeneous distribution of intermediate size (0.2 to 1 #m) parti-
cles. The extremely heterogeneous distribution of particles in 7000 series
alloys contributed to the ductile tear ridge fracture appearance in plate
specimens and smoother appearance (lower local ductility) of sheet fatigue
specimen fracture surfaces.
Fracture morphology was not affected significantly by alloy composition
and microstructure in sheet tension or stress-corrosion specimens, but in-
creased surface roughness and particle cracking was observed on sheet
shear specimens at higher strain rates.
Product Form and Load Conditions--Finer, more uniform dimple forma-
tion was observed in sheet tension specimens which was attributed to less
elastic constraint imposed near microvoid nucleation sites and the general
absence of large second phase particles (larger than 8 /zm) in all sheet
specimens. Larger particles in plate product contributed to a greater
degree in the final fracture process.
Specimen orientation had little or no observable effect on fracture
appearance, but strain rate changes affected the shear strengths of all the
alloys examined and the fracture surface appearance of the 2024 and 7000
series alloys tested in shear. In sheet fatigue specimens fracture feature
variations were more associated with the maximum tensile load
experienced than with the load excursion (varying R value). This dif-
ference was manifested by presence or absence of voids caused by particle-
matrix interface failure and particle cracking, both of which occurred
mostly at high maximum tensile stress levels.
Variation in stress level and crack growth direction may also be deduced
from observations of fractures produced by SCC. Fracture surfaces which
have been formed by a rapidly growing crack may contain less corrosion
leaf formation. This would indicate that higher stress levels associated with
crack growth direction could be deduced from the manner in which the
corrosion leaves peel from the surface.

Application to Failure Analysis

An attempt was made in the present investigation to characterize, in
several high strength aluminum alloys, variables influencing fracture
surface morphology. Several of the aspects that are summarized here are
discussed in greater detail elsewhere [31]. The principal objective of the
present paper was to identify the main features of fracture morphology of
the alloys as influenced by the variables related to alloy and product form,
microstructure, and testing conditions, and to highlight some points that
are of importance in failure analysis. The ultimate goal of the study was to

z
o

"11

Q
0

0
z

p-
C
FIG. ll--Scanning electron fractographs of sheet fatigue specimens at R = 0.05: (a)2024-T3, (b) 6061-T4, (c) 6061-T6, (d) 7075-T6, and (e) 7178-T6.
Nominal maximum stress was 64 MPa for 2024-]'3 and 51 MPa for others.
c

r"
f-
0

0
f.o

"11
an
0
.-I
0
0
"v
I.<

c
-11
m

FIG. 12--Scanning electron fractographs o f a 2024-7"351 stress-corrosion specimen tested in the short transverse direction: (a) after Hmited ultra-
sonic cleaning and (b) after extensive ultrasonic cleaning.

:1]
7~

I-

Q
0
-is

0
z

FIG. 13--Optical micrograph of a section approximately normal to the fracture surface o f a 2024-T351 stress corrosion specimen tested in orientation z
c
3 (Fig. l(d)).

I-
r-
0.<
tD

FIG. 14--Stress corrosion fracture o f a 7178-T651 specimen tested in the short transverse
direction: (a) scanning electron fractograph, (b) higher magnification o f area near A in (a),
(c) higher magnification o f area near B in (a).

provide quantitative information which could be used to describe more

accurately service failure conditions by analyzing the appearance of the
failed component. Typical questions which arise when a service failure is
experienced are as follows.
1. Can material quality be identified or excluded as cause of failure?
2. Can fatigue, stress-corrosion, embrittlement, or other service-related
environments be identified as cause of failure?
Failure analysis methodology using microscopic, fractographic analysis
has not progressed much beyond attempts to answer these basic questions.
Most handbook information deals with specific, isolated examples which
do not provide general guidelines for use in failure analysis work. Although
the present investigation fell short of providing quantitative information re-

iating fracture variables, it did identify some important additional

considerations which may have some general utility in advancing failure
analysis methodology by fractography in aluminum alloys and providing
direction for future work.
Fracture surface appearance of failures which occurred under macro-
seopically ductile conditions showed limited microscopic ductility
associated with tear ridge formation. The random orientation or specific
directionality of tear ridges could be used to indicate whether random
crack nucleation and growth or specific crack propagation direction had
occurred (for example, 7075 and 7178 plate tensile and notch-bend be-
havior).
In thinner gage materials, well defined striations were not the principle
indicator of fatigue damage. Distribution of voids and cracked second-
phase constitutents provided indications of load magnitude in fatigue or
shear fracture related failures. Strain rate affected both shear ultimate
strength and fracture surface morphology.
Observations of fracture surface appearance of failures which occurred
by stress-corrosion cracking provided indications of direction of crack
growth (for example, degree and orientation of corrosion leaf formation).
Fracture surfaces which experienced chemical reaction with environment or
exposure to contaminants were examined microscopically and by energy
dispersive X-ray analysis both before and after cleaning. This was
necessary to separate deposits which resulted from the fracture process
from those which occurred as a result of handling after failure.
These observations were identified from considerations of general trends
in the present program. Additional work, perhaps using automated
analysis techniques, will be necessary to determine if such observations can
be further quantified. Such information would be invaluable in furthering
effective failure analysis by fractography.

Acknowledgments
The research leading to the publication of this report was performed
during the tenure of M. D. Bhandarkar as a National Research Council
Resident Research Associate at NASA Langley Research Center, Hampton,
Va. The authors acknowledge Alberta Saunders, Materials Division, NASA
Langley Research Center for considerable assistance in the experimental
phase of the program. Appreciation is extended to Anamet Laboratories,
Berkeley, Calif., where M. D. Bhandarkar is currently employed, for
encouragement in the completion and presentation of the report.

References
[1] Forsyth, P. J. E. and Ryder, D. A., Metallurgica, Vol. 63, 1961, pp. 117-124.
[2] Jacoby, G., Experimental Mechanics, Vol. 5, No. 3, 1965, pp. 1-18.

[3] Sprowls, D. O. and Brown, R. H., Proceedings, Conference on Fundamental Aspects of

Stress Corrosion Cracking, The Ohio State University, 11-15 Sept. 1967, pp. 466-512.
[4] Jacobs, A. J., Proceedings, Conference on Fundamental Aspects of Stress Corrosion
Cracking, The Ohio State University, 11-15 Sept. 1967, pp. 530-560.
[5] Hunter, M. S. and McMillan, J. C. in Electron Fractography, ASTM STP 436, American
Society for Testing and Materials, 1968, pp. 196-211.
[6] Broek, D. inFracture 1969, Chapman and Hall, London, 1969, pp. 754-764.
[7] Unwin, P. N. T. and Smith, G. C., Journal of the Institute of Metals, Vol. 97, 1969,
pp. 299-310.
[8] Singh, S. N. and Flemings, M. C., Transactions, Metallurgical Society of American
Institute of Mining, Metallurgical and Petroleum Engineers, Vol. 245, 1969, pp. 1811-1819.
[9] Tanaka, J. P., Pampillo, C. A., and Low, J. R., Jr. in Review of Developments in
Plane Strain Fracture Toughness Testing, ASTM STP 463. W. F. Brown, Jr., Ed.,
American Society for Testing and Materials, 1970, pp. 191-215.
[10] Wei, R. P., Engineering Fracture Mechanics, Vol. 1, 1970, pp. 633-651.
[11] Kirmcin, I., Metallurgical Transactions, Vol. 2, 1971, pp. 1761-1770.
[12] Kershaw, J. and Liu, H. W., International Journal of Fracture Mechanics, Vol. 7, 1971,
pp. 269-276.
[13] Hartman, A., Lievers, J. W., and van der Vet, W. J., "Study of the Growth of Stress
Corrosion Cracks in the Aluminum Alloy 7075, Part I: Investigation on the Corrosive
Medium," Report No. NLR TR 71090U, National Aerospace Laboratory, The Nether-
lands, 1971.
[14] Low, J. R., Jr., Van Stone, R. H., and Merchant, R. H., "An Investigation of Plastic
Fracture in Aluminum Alloys," NASA Report CR-131100, National Aeronautics and
Space Administration, 1972.
[15] Peel, C. J., Wilson, R. N., and Forsyth, P. L E., Metal Science Journal, Vol. 6, 1972,
pp. 102-106.
[16] Speidel, M. O. and Hyatt, M. V. in Advances in Corrosion Science and Technology,
Vol. 2, Plenum Press, New York, 1972, pp. 115-335.
[17] Broek, D., Engineering Fracture Mechanics, Vol. 5, 1973, pp. 55-66.
[18] Brock, D., "Some Contributions of Electron Fractography to the Theory of Fracture,"
Report No. NLR TR 72029U, National Aerospace Laboratory, The Netherlands, 1973.
[19] Jones, D. L. and Liebowitz, H., Engineering Fracture Mechanics, Vol. 5, 1973, pp. 397-
402.
[20] EI-Sondani, S. M. and Pelloux, R. M., Metallurgical Transactions. Vol. 4, 1973, pp.
519-531.
[21] Hahn, C. T. and Simon, R., Engineering Fracture Mechanics, Vol. 5, 1973, pp. 523-540.
[22] Geschwind, G. and Adler, P. N., Corrosion/73, The International Corrosion Forum
Devoted Exclusively to the Protection and Performance of Materials, Anaheim, Calif.,
19-23 March 1973.
[23] Metals Handbook, Vol. 9, American Society for Metals, 1974.
[24] Rosenfieid, A. R., Price, C. W., Martin, C. L, Thompson, D. S,, and Zinkham, R. E.,
"Research on Synthesis of High Strength Aluminum Alloys, Part I, The Relation Be-
tween Precipitate Microstructure and Mechanical Properties in Aluminum Alloys," Tech-
nical Report AFML-TR-74-129, Part I, Air Force Materials Laboratory, Dayton, Ohio,
1974.
[25] Van Stone, R. H., Merchant, R. H., and Low, I. R., Jr. in Fatigue and Fracture Tough-
ness--Cryogenic Behavior, ASTM STP 556, American Society for Testing and Materials,
pp. 93-124, 1974.
[26] Wood, W. E. and Gerberich, W. W., Metallurgical Transactions, Vol. 5, 1974, pp.
1285-1294.
[27] Hahn, G. T. and Rosenfield, A. R., Metallurgical Transactions A, Vol. 6A, 1975, pp.
653-670.
[28] Nageswararao, M., Gerold, V., and Kralik, G., Journal of Materials Science, Vol. 10,
1975, pp. 515-524.
[29] Plttinato, G. F., Kerlin, V., Phillips, A., and Russo, M. A., SF,M/TEM Fractography
Handbook, Technical Report AFML-TR-75-159, Air Force Materials Laboratory, Day-
ton, Ohio, 1975.
[30] Aerospace Industries Association, Standard Test Procedure ARTC 13-5-1.

[31] Bhandarkar, M. D. and Lisagor, W. B., "Metallurgical Characterization of the Fracture

of Several High Strength Aluminum Alloys," NASA TP-1086, National Aeronautics and
Space Administration, Langley Research Center, Hampton, Va., 1977.
[32] Metals Handbook, Vol. 1, American Society for Metals, 1961.
[33] Aluminum, Vol. L Properties, Physical Metallurgy and Phase Diagrams, K. R. Van
Horn, Ed., American Society for Metals, 1967.
[34] Metals Handbook, Vol. 7, American Society for Metals, 1972.

Use of Microfractography in the Study of

Fatigue Crack Propagation under
Spectrum Loading

REFERENCE: Abelkis, P. R., "Use of Microfraetography in the Study of Fatigue

Crack Propagation under Spectrum Loading," Fractography in FailureAnalysis, ASTM
STP 645, B. M. Strauss and W. H. Cullen, Jr., Eds., American Society for Testing
and Materials, 1978, pp. 213-234.

ABSTRACT: Use of electron microscope fractography in the study of fatigue crack

propagation under spectrum loadings is based on the evaluation of 7075-T651 alumi-
num alloy specimens from an experimental program. Microscopically established crack
propagation data by striation spacing measurements are shown to provide cycle-to-cycle
evidence of crack propagation behavior under spectrum loading that normally cannot
be established under macroscopic observation. Crack propagation rate retardation and
acceleration is illustrated through fractographie striation measurements. Contribution
and interaction of various loads of a transport wing loads spectrum are illustrated
fractographically from the test program.

KEY WORDS: fractography, electron microscopes, spectrum loading, aluminum alloys,

crack propagation, fatigue (materials)

Nomenclature

a H a l f c r a c k length (hole not included)

a: T o t a l (final) c r a c k length
Aa/AN Rate o f fatigue c r a c k growth p e r cycle (also da/dN)
Aa/AF Rate o f fatigue c r a c k growth p e r flight
FbF Flight-by-flight
GAG Ground-air-ground
K Stress intensity factor
n N u m b e r o f cycles
No: Life c o r r e s p o n d i n g to as as d e t e r m i n e d by striation c o u n t i n g from
fractographs
N/ Total (final) test life
1Senior engineer/scientist, Douglas Aircraft Company, McDonnell Douglas Corporation,
Long Beach, Calif. 90846.

213
Copyright by ASTM
9 Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright 1978by ASTM International [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
214 FRACTOGRAPHY IN FAILURE ANALYSIS

R Stress ratio, Smin/Smax o r Kmi,/Kmax

S Stress
S,.f Flight mean stress
S,~g Ground mean stress
SEM Scanning electron microscope
TEM Transmission electron microscope
t Sheet thickness
#m micrometre, 10-6 m or 3.937 x 10-s in.
This paper deals with the study of fatigue failure process through the use
of electron microscope fractography. The emphasis is on the crack propa-
gation stage under fatigue spectrum loading.
The fatigue failure process can be divided into two stages: (a) crack nu-
cleation stage during which the load-carrying capacity of the structure ex-
periences no appreciable reduction, and (b) crack propagation stage during
which the load-carrying capacity is gradually reduced as a function of the
crack length and fracture toughness of the material.
Complete failure will occur in either stage when the applied load exceeds
the load-carrying capacity at that time. The length of Stage 1 depends on
the initial fatigue quality of the material and structure. The length can be
diminished substantially if the material in its basic form, or through manu-
facturing, contains flaws which act as strength reducers relative to standard
material allowables or which initiate cracking in a short time. The length
of Stage 2 depends on the material and structure basic crack propagation
properties and the fatigue loadings and environment in which it operates.
To design safe and economical structures, both stages of the failure process
must be understood. Safety in the second stage is enhanced by thorough-
ness and frequency of inspections.
Crack propagation is a measurable quantity. This quantity can be deter-
mined on the macroscopic level by observing the crack propagation on the
side of a specimen, or it can be established on the microscopic level by
studying the fracture surface through the use of the electron microscope.
The advantage of the latter method is that it provides cycle-by-cycle evi-
dence, in the form of striations, of crack propagation behavior that normally
cannot be established from macroscopic observations. This paper illustrates
the use of the electron microscope in establishing the crack propagation
rates of specimens subjected to a spectrum loading.

Fracture Surface and Striations

The fracture surface is a fingerprint, or a record, of the loading exper-
ienced by the specimen or test structure in service. Figure 1 shows sample
test and service fatigue fracture surfaces. Certain features of the loading
are visible on the macroscopic scale. This is particularly true at longer crack

FIG. 1--Sample test and service fatigue fracture surfaces--aluminum alloys.

lengths. Larger loads in the spectrum produce a rougher surface which ap-
pears darker, while lower loads produce a smoother and brighter surface.
Various such features are visible in Fig. 1. A block of higher load cycles is
represented by the dark band "h" in Specimen (c) while the following block
of lower load cycles is the brighter band "e". In Specimens (d) and (e) at

FIG. 2--Examples of striation markings under various loading conditions--aluminum

alloys.

longer crack lengths, the darker bands are due to single high loads. Thus,
limited knowledge can be obtained from such macroscopic observations
about the loading spectrum and, if the load's spectrum and sequence are
known, about the crack propagation rate. However, detail information
about cycle-by-cycle crack propagation rates and load interaction effects
can be obtained on the microscopic level through striation counting and

measurement of spacings between striations. In this context, striations are

considered to represent successive positions of the crack front. Examples of
striation markings, under various loading sequences, are illustrated in
Fig. 2.
To obtain crack propagation data from fractographs, one must be able
to recognize and relate striations to known loading cycles. The problem is
rather straightforward for constant amplitude loading. However, it be-
comes a complex and difficult problem for spectrum loadings. Interpreta-
tion of the striations in this study was based on the following observations
and findings [1,2]. 2
1. Crack advances only due to increasing tension loading. However, stria-
tion profiles and appearance depend on loading and unloading portions
of the cycle.
2. Crack propagation can occur only during that portion of the loading
cycle in which the crack is fully open at the crack tip.
3. Crack closure may occur before the minimum load is reached.
4. A cycle may not produce a striation and crack advance, because (a)
previous loading has completely retarded (stopped) cracking, or (b) loading
is below the threshold value for propagating cracks.
5. A striation may not be visible in a fractograph, because (a) it has
been obliterated because of fracture surface damage, mechanical or chem-
ical, or (b) the angle at which electron microscopy was performed did not
reveal the striation.
6. Crack propagation rates from microfractographs are established by
measuring the distance between striations and knowing the crack length at
that point. The spacing between striations of the same loading in one frac-
tograph or adjacent fractographs may vary much more than the called for
change in crack length. This could be due to a number of factors. On the
microscopic level, cracks propagate on many planes and in many directions.
What is seen on the side of the specimen is the net propagation of the
total crack front which exhibits less scatter. Also, locally the crack may be
retarded by inclusions, or the fracture surface, as viewed by the electron
microscope, may not be perpendicular to the electron beam, and the re-
sulting fractograph will not represent the true dimension. To establish the
true dimension, the replicate or the specimen should be rotated in the
electron microscope to produce the largest spacing between striations. An
example of such striation spacing variation for constant amplitude (CA)
loading is given in Fig. 3. Because of such variations, the measurements are
(a) made in the direction of the general crack front propagation, (b) aver-
aged over a number of cycles, and (c) made along a somewhat straight
line, usually in the middle of the specimen. From this it follows that to
establish valid crack propagation rates by striation measurements, one

2The italic numbersin bracketsreferto the list of referencesappendedto this paper.

FIG. 3--Striation spacing variation within one fractograph--constant amplitude loading.

must obtain data over a reasonably long crack length and cannot depend
on the rate established from one fractograph.

Experimental Program
Unless otherwise noted, the fractography work reported herein was done
on a center-hole notched 7075-T651 bare sheet specimen, 6.35 mm (0.25
in.) thick, 38.1 mm (1.5 in.) wide, with a 6.36 mm (0.250 in.) diameter
hole. Fatigue cracks were initiated from the hole and the crack length a
is taken from the edge of the hole. The program consisted of testing 81
specimens under constant amplitude and spectrum loadings. Fractography
work was done on 25 of these specimens. A summary of the loadings on
these specimens is presented in Table 1. Total test lives and crack lengths
are given in Table 2. Specimens identified as second group specimens were
made and tested a year later than the first group. Because of a difference
in hole preparation (drilled versus drilled and reamed for the second group),
the test lives of the second group specimens were approximately twice as
long as those of the first group for the same type of loadings. However,
crack propagation rates were about the same, indicating that the difference
in hole preparation influenced only the length of crack initiation stage.
The loadings on the specimens were derived from a loads spectrum typi-
cal of a short-medium range jet aircraft. The stresses represent the wing

Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized
Spectrum Loadings: TABLE l--Test loadings.
FbF Spectrum Flight Loads Spectrum
S
S max3~i
S max^ II
max l ~ . ~:L
Sm I l

Sm Smax~ ~,~/ n
Smin3~
g ~ ' V V " L
m
Specimen I11,112,113 122,123 133 141 163" 211,212 221,222 242* 251" X211,X212 _x
Sm -34.5 (-6.0) -34.5 (-5.0) &O
Smgax -21.4 (-3.1) I -13.8 (~2.0) -21.4 (-3.1) 0
Smin -47.6 (-6.9) -55.2 (-8.0 -47.6 (-6.9) Z
fY fy C
is 86.2 (12.6) 72.4 (I0.5) 86.2 (12.5) 72.4 (10.5) 86.2 (12.5) GO
mf m
Smaxl lll.O (16.1) 93.1 (13.5) 111.0 (16.1) 9 93.1 (13.5) 111.0 (16.1)
Sminl 61.4 (8.9) 51.7 (7.5) 61.4 (8.9) 51.7 (7.5) 61.4 (8.9) 0
"II
Smax2 120.0 (17.4) !100.7 (14.6) 120.0 (17.4) ]00.7 (14.6) 12o.o (17.4)
Smin2 52.4 (7.6) 44.1 (6.4) 52.4 (7.6) 44.1 (6.4) 52.4 (7.6)
Smax3 166.2 (24.1) 139.3 (20.2) 124.1 (18.0) 139.3 (20.2) 139.3 (20.2) 166.2 (24.1) 139.3 (20.2)139.3 (20.2) 166.2 (24.1) -n
Smin3 6.2 (0.9) 33.1 (4.8) 48.3 (7.0) 33.1 (4.8) 4.8 (0.7) 6.2 (0.9) 4.8 (0.7) 4.8 (0.7) 6.2 (0.9)
0
-n
Constant Amplitude Loadings:
0
Specimen Sm Smax .-.I
Smin R 0
Flight Loads X311,X313 86.2 (12.5) 166.2 (24.1) 6,2 (0.9) 0.036 -rl
C y cl e s 311-2,312-2" 86.2 (12.5) 166.2 (24.1) 6.2 (0.9) 0.036 Stresses shown are test section
333 86.2 (12.5) 124.1 (18.0) 48.3 (7.0) 0.39 gross area stresses, MN/m2 (KSI).
GAG X411 59.3 (8.6) 166.2 (24.1) (Net/Gross) Area Stress = 1.20 -r
-47.6 (-6.9) -0.29
Cycles 423-I* 46.2 (6.7) 139.3 (20.2) -47.6 (-6.9) -0.34
433 38.6 (5.6) 124.1 (18.o) -47.6 (-6.9) -0.38 *Second group specimens.
452 26.2 (3.8) 86.2 (12.6) -34.5 (-5.0) -0.4
(D

-n
[ i ]6_:35mm(.25in.) -n
>
I --4 I--6.35 I~ 0
--4
0

>
"U
Specimen NF a f mm (in.) Na f Na f / N f N ,~r b I
-<
111 11 030 8.64 (0.340) 6 436 0.58 3 140
112 8 568 5.72 (0.225) 4 568 0.53 2 835
113 13 533 9.53 (0,375) 12 514 0.93 5 697
122 17,067 11.10 (0.437) 7 404 0.43 4 129 j....
123 14 750 8.81 (0.347) 8 821 0.60 5 919 C
133 22 340 12.70 (0.50) 9 067 0.41 4 626
141 15 049 10.74 (0.423) 6 883 0.46 4 487 m
163 66 546 10.13 (0.399) 6 936 0.10 3 635
211 21 178 10.67 (0.42) 10 867 0.51 4 726 Z
212 14 921 7.47 (0.294) 6 787 0.46 3 919 j-.-
X212 16 566 9.70 (0.382) 8 216 0.50 4 475 -<
221 16 194 12.78 (0.503) 8 737 0.54 4 136
222 36 101 13.44 (0.529) 9 785 0.27 5 777
242 35 083 c . . . . . . . . . . . .
251 41 612 e . . . . . .
X311 25 680 7.11 (0.280) 11 261 0.44 8 ~102
X313 31 893 7.62 (0.30) 15 533 0.49 9 498
311-2 55 018 9.86 (0.388) 18 755 0.34 14 352
312-2 75 087 6.99 (0.275) 11 947 0.16 10 429
333 214 824 14.66 (0.577) 61 062 0.28 31 711
X411 24 854 5.08 (0.20) 11 020 0.44 6 258
423-1 78 925 7.39 (0.291) 20 488 0.26 14 361
433 45 390 6.86 (0.27) 14 348 0.32 9 741
452 474 514 13.72 (0.54) 54 321 0.12 39 953

, Flights for spectrum loadings, cycles for constant amplitude Ioadings.

bAa = 0.30 to 5.08 mm (0.012 to 0.20 in.).
e Grip end failures.

lower surface outboard of the landing gear. The objective of the exper-
imental program was to subject the specimens to a flight-by-flight (FbF)
spectrum loading and then to, individual components of this spectrum, in
order to evaluate their independent contributions to fatigue damage and
crack propagation. In this manner, all loadings were divided into four types,
using the first digit of the specimen number for identification:
1. FbF spectra
2. Flight loads spectra
3. Flight loads constant amplitude (CA) loadings
4. Ground-air-ground (GAG) CA loadings, which represent the transi-
tion from the compression ground loads to the tension flight loads

Crack Propagation Rates under Constant Amplitude Loading

The simplest fatigue cyclic loading is constant amplitude. It is a loading
with constant mean and amplitude (Fig. 2). The crack propagation rates
under CA loading usually are presented as

da/dN versus AK for a given R

where
~kK = Kmax -- Kmin,
= stress intensity factor range,
K = (a~). k = stress intensity factor,
a = applied gross area stress,
a = crack length, and
= correction factor for the effect of hole, width, stiffener, etc.
These data usually are obtained by measuring the crack propagation on the
side of a specimen during a test. Use of electron microscope fractography
to obtain these data by striation spacing measurements is too tedious, time-
consuming, and less accurate due to large scatter. Only for special studies,
such as checking the rates at short crack lengths, at multiple crack nu-
cleation sites, or in the interior of the specimen where the crack is not vis-
ible on the surface, is use of fractography justified. An example of da/dN
data obtained in this manner is shown in Fig. 4. The scatter in such data is
quite obvious here although the data does follow generally expected da/dN
data trends as indicated by the Forman [3] equation. Also, it should be
noted that negative R rates are not the same as R = 0 rates.

Crack Propagation Rates under Spectrum Loadlngs and Load Interaction

Effects
The simplest spectrum loading is a two-load-level block spectrum, Fig.
5. Fractographic analysis shown here was done on a 7075-T651 bare sheet

a film in.

i
zXK (MN/m3/2) x lO3 psi I,/T-6:.
Kc 45,32Q 41,200
n 2.48 2.48
C 2.19 x lO "lO 9.94 x lO-12

FIG. 4~7075-T651 crack propagation rates, constant amplitude loading, AK = Kmax - Kmin.

center crack, t = 6.35 mm (0.25 in.), 304.8 mm (12 in.) wide specimen.
This is an example of studying load interaction effects in the simplest
form: transition from high to low and from low to high load and what
effect it has on crack propagation rate. The number of cycles chosen for
each block, on the basis of plastic zone size due to the high load, were such
as to produce, at the end of each block, a crack propagation rate
approaching that of CA loading. Microfractographs were obtained at load
transition points and at midpoint of the low loading blocks. The data show

Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized
"' da/dI~, i n , / c y c l e
X
~I 0 0 i
da/dN, mm/cycle
x
% ~
o
/
o
/ o
~....---- o I ~ ~
P~
\ 0 / - -
\ 0
\ O' v - -
!
II ii
\ '. b
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized
224 FRACTOGRAPHYIN FAILURE ANALYSIS

the well known [1,2] crack retardation effect of high load on the following
lower load crack propagation rate, although this sometimes is considered
not to occur at such a low overload ratio as 1.2 in this case. The data also
show that the crack propagation rate at higher load after the transition
from the lower load is accelerated when compared to the da/dN data at the
end of the block. This example shows that retardation and acceleration
phenomena in simple spectrum loading crack propagation can be defined
through microfractography, although care must be exercised in
distinguishing between scatter in crack propagation rates as established by
striation spacing measurements and real load interaction effects.
Next, more complicated spectrum loadings of the type shown in Table 1
are considered. Fractographs shown in Fig. 6 illustrate the striation
spacings and patterns at various crack lengths for a typical FbF spectrum
loading. Figures 7 and 8 illustrate the correlation between striations and
individual loading cycles in the spectra. The following observations can be
made about the interpretation of these fractographs.
1. Striation groups representing individual flights were identified for all
loadings and specimens.
2. Striations of individual cycles within a flight can be identified only at
longer crack lengths. At shorter crack lengths, identification is possible
only of the major loadings, such as the GAG cycle and the peak flight
load. Close to crack initiation, a complete flight appears as one striation.
Striations representing one flight can be identified at crack lengths as short
as 0.076 to 0.152 mm (0.003 to 0.006 in.).
3. Best striation identification is possible for spectra with the lower
flight peak loads. Identification of all individual striations becomes more
difficult for the spectra with the higher peak loads, in particular, the stria-
tions due to the highest load cycles. Difficulties also are encountered in
trying to identify the separation of the GAG cycle striation from the follow-
ing flight low load cycle striations, as well as the beginning of the low load
cycle striations after the flight peak loads. In general, the flight low load
striations are clearer after the GAG cycle than after the peak loads.
4. Significant markings (deep valleys or high peaks) are associated with
the GAG and the peak flight load cycles. Since these markings seem to
precede these cycles, they appear to be caused by the large unloadings pre-
ceding these cycles. These markings remain dominant in the fractographs
at shorter cracks while the striations of the lesser cycles become indis-
tinguishable.
5. Crack propagation, as measured by striations, is primarily due to the
increasing load between reversals, although the striation profile and to a
certain extent the striation width are influenced by previous loadings and
the following unloading.
Crack propagation rates for these types of spectra loadings, as established
by striation spacing measurements, are given in Figs. 9 and 10. The rates

FIG. 6--FbF spectrum loading, specimen l l 3 fractographs.

FIG. 7--Striation--loading correlation, FbF Spectra.

are per flight, the measurements being made for a group of cycles which
represent a flight. Identification of striation groupings as a flight is rather
straightforward here because the loading consisted of the repetition of
exactly the same flight (Figs. 7 and 8). The identification is first accom-
plished at longer crack lengths and then chronologically the groupings can
be traced to shorter crack lengths (Fig. 6). Several observations and inter-
pretations can be made about these rates.

FIG. 8--Striation--loading correlation, flight loads spectra.

1. Crack propagation rates were approximately the same for spectra

run in opposite directions with respect to loading sequence. Compare rates
of Specimens 211 and 212 against Specimen X212 (Fig. 10 and Table 2).
Similar results have been observed by other investigators [4].
2. There appear to be no significant trends in the crack propagation
rates with respect to the flight peak load when observing the data in Figs. 9
and 10. All rates fall within the same general scatter band. However, a
more detailed analysis of the data, such as in Table 2, shows a tendency

,~- 111 A- 122 o- 133 9 - 141 o - 163

o- 112 e - 123
o - 113
. . . . Flight Loads Spectra ( n o G A G cycle).
Scatter Band, Figure 10.
FIG. 9--Crack propagation rates, FbF spectra.

for the rate to increase as the flight peak load increases. In general, the
small differences in the rates are probably due to the fact that as the crack
propagation rate due to peak load itself increases, the rate of the lower
loads decreases due to the increasing retardation, with the two rate
changes in opposite directions tending to offset each other.
3. Within the general scatter, the crack propagation rates per flight with
or without the GAG cycle appear to be the same when comparing the rates

~- 211 o- X2ll A- 221

o- 212 o- X212 e- 222

FIG. lO--Crack propagation rates, flight loads spectra.

between Figs. 9 and 10. However, a more detailed analysis of the data,
such as in Table 2, indicates some tendency for the loading with the GAG
cycle (FbF) to have a higher rate. This is to be expected when considering
the fact that the GAG cycle contributes to the rate increase directly by
being an additional cycle and indirectly by perhaps increasing the crack
propagation rate of subsequent low load cycles.
More detailed analysis of individual striation spacings within a flight
provides information on contribution of the different types of loading cycles
to the crack propagation, as well as on possible loads interaction effects.
The contribution of individual cycles or groups of cycles to the total crack

p r o p a g a t i o n , as percentage of total, is given in T a b l e 3. Clear identification

of individual striations was possible only at longer crack lengths, a n d these
percentages are typical only of crack lengths beyond a = 2.54 m m (0.10
in.). T h e following observations can be m a d e a b o u t these percentages.
1. Crack p r o p a g a t i o n due to the G A G cycle in the F b F spectra is
small, anywhere from 7 to 24 percent, d e p e n d i n g on crack length a n d the
spectrum. Note t h a t the percentage is lower for the s p e c t r u m (Loading 16)
with the lower flight m e a n . This is a m u c h lower percentage t h a n has been
a t t r i b u t e d to this cycle in fatigue d a m a g e calculations in the past, where
the c o n t r i b u t i o n has been claimed to be as high as 80 or 90 percent.
2. The flight low load cycles, which represent the majority of cycles in
the spectra, p r o d u c e between 30 a n d 58 percent of the crack p r o p a g a t i o n
in the F b F spectra a n d less t h a n 52 percent in the flight loads spectra.
There is a t e n d e n c y for the percentage to increase as the peak load de-
creases because of the c o m b i n e d effect of the smaller crack p r o p a g a t i o n
due to the peak loads a n d the smaller r e t a r d a t i o n on the lower load rates.
T h e lower percentages for Loadings 16 a n d 25 are due to lower loads a n d
fewer cycles, respectively. Also, it is interesting to note that whereas these
lowest loads are below the fatigue limit for crack initiation, they do

TABLE 3--Contribution o f individual cycles in spectrum loading to crack propagation.

FbF Spectra:

GAG Cycle Low Loads Intermediate and Low Loads

A to Ba B to C High Loads C to F F to A

Loading n % n % n % n %

11 1 10 to 23 6 16 to 26 3 40 to 50 6 16 to 21
12 1 13 to 24 6 20 to 29 3 31 to 46 6 17 to 24
13 I 14 to 23 6 19 to 28 3 31 to 39 6 21 to 30
14 1 13 to 15 6 22 to 26 3 38 to 40 6 22 to 24
16 1 7 to 10 5 15 to 18 4 57 to 63 6 15 to 16

Flight Loads Spectra:

Low Loads Intermediate and High Loads

Fto Ato Bb B to F

Loading n % n %

21 11 29 to 46 4 54 to 71
X21 12 36 to 48 3 52 to 64
22 11 37 to 52 4 48 to 63
24 11 32 to 35 4 65 to 68
25 5 18 to 27 4 73 to 82
See Fig. 7 for cycle identification.
bSee Fig. 8 for cycle identification.

produce crack propagation even with retardation present. This is an

example of the differences that exist between crack initiation and crack
propagation failure models.
3. The three or four intermediate and high load flight cycles produce the
major portion of the crack propagation. The percentage increases with the
increase of the peak load and its relative magnitude to the lower loads.
Finally, load interaction effects were studied by comparing individual
cycle rates from spectrum loading fractographs to the CA loading rates,
shown in Fig. 11. The AK is based on the increasing load between suc-
cessive loading reversals and for negative R loadings AK = Kmax - Kmm
includes compression. Data in Fig. 11 show the following.
1. The crack propagation rate due to flight low loads (cycles defined by
Smin~ and Smax~ in Table 1) is retarded by high loads. The retardation is
greater in the flight loads spectra than in the FbF spectrum. Possible
explanation is that the GAG cycle, which includes compression,
accelerates the rate, or, in effect, reduces the retardation due to the high
loads. Similar rate acceleration due to compression overloads has been ob-
served by others [5]. Also, in the FbF spectra, the rate of the low loads
tends to be higher after the GAG cycles than after the peak load, further
indicating the acceleration effect of the GAG cycle.
2. The rate of the GAG cycle (defined by Stain and Smf or Smin and Sm~x~
in Table 1) shows some acceleration tendencies. This could be partly due to
the difficulty in clearly defining the striations for this cycle.
The effect of load interaction on the rates of intermediate and peak load
cycles (defined by Smin2 and Smax2, Sminl and Smax2 or Smin2 and S~axl, and
Smi.3 and Sma~3 or Smin2 and Sm~3 in Table 1) is more difficult to ascertain
because of difficulties in separating striation spacing for these cycles. Peak
load retardation effects on the rate of intermediate cycles were found to be
similar to the flight low loads just discussed. An example of the difference
in the rate of the intermediate loading cycle before and after the peak load
is illustrated clearly in Fig. 7, Specimen 133, CD and EF loadings. Rates
of the peak load cycles were the most difficult to establish but appeared to
be unaffected by load interactions. However, a more detailed analysis
should be performed before a firm conclusion is reached on this point.

Crack Initiation and Propagation Lives

Because of possible differences in crack initiation and crack propagation
failure mechanisms and analysis methods, it becomes important to
separate the crack initiation and crack propagation lives. For this reason,
crack propagation test lives, as determined by striation counting from
fractographs, are presented in Table 2. With the exception of one
specimen (113), crack propagation lives of most specimens represent
between 40 and 60 percent of the total test lives. This percentage decreases

FIG. ll--Comparison of crack propagation rates between constant amplitude and spectrum
individual loadings.

as the total life increases. Most of the crack propagation life is consumed
in the initial crack propagation stages where it is most difficult to establish
the rate by striation counting. Up to 50 percent of the crack propagation
life may be consumed in propagating to 0.30-mm (O.012-in.) crack length,
although a more typical value was about 30 percent, with a low of 5 per-

cent for the long life, low loading Specimen 452. The length of the crack
propagation life of a specimen depends on the type of initial notch,
loading, and specimen geometry, and the ability to find and measure stria-
tion spacings at very short cracks.
Another aspect of crack initiation that must be mentioned here is the lo-
cation of crack nucleation sites. Usually it is assumed that cracks initiate
at the "corners" of a hole. In this program, most of the cracks started in-
side the hole, as illustrated in Fig. 12, with a high degree of correlation be-
tween crack nucleation sites and the predominant hole machining tool
marks inside the hole.

Service Cracking Interpretation

Service cracking fracture surface fractographic analysis is performed
mainly to identify cracking origin, mode of cracking, and to establish
crack propagation rates. In the preceding discussion it was seen that crack
propagation rates under spectrum loadings can be established from fracto-
graphs if the loading and sequence are known. However, attempts in estab-
lishing crack propagation rates from service cracked fracture surfaces have
not been very successful unless the loading happened to be constant ampli-
tude or a particular repetitive feature of the spectrum loading was known
and could be identified in the fractograph.

Conclusions
Electron microscope fractography is a useful tool in the study of fatigue
crack propagation failure process under a known loading sequence. Crack

FIG. 12--Crack nucleation sites in the notch hole, Specimen 313.

propagation rates and loads interaction effects under spectrum loadings

can be determined by measurement of striation spacings. Use of this
method has produced the following findings and confirmations of the fa-
tigue crack propagation failure process.
1. Under spectrum loading of small notched fatigue specimens, up to 50
percent of the total test life is consumed in crack propagation that can be
defined by striations in a microfractograph.
2, Retardation and acceleration of cracking rates under spectrum
loadings due to loads interaction were identified.
3. Striations are associated with the increasing load between successive
load reversals. Also, striations and crack propagation rates appear to be
influenced by the unloading portion of the cycle.
4. Crack propagation rates established by striation measurements
exhibit large scatter.
5. Crack propagation rates under constant amplitude negative R load-
ing, as established by striation measurements, are not represented by R =
0 loading rates.
6. In the context of a transport wing lower surface loads spectrum eval-
uated in this paper, the GAG cycle was found to be not the most sig-
nificant contribution to crack propagation, representing only from 7 to 24
percent of the total.

Acknowledgment

This paper summarizes work performed at the Douglas Aircraft Com-

pany in the Structures Engineering subdivision under the sponsorship of
the Independent Research and Development (IRAD) Program of the Mc-
Donnell Douglas Corporation.
The author acknowledges and expresses his appreciation to D. L.
Brown, Materials and Process Engineering, for the preparation of the
fractographs.

References

[1] Fatigue Crack Propagation, ASTM STP 415, American Society for Testing and Ma-
terials, 1967.
[2] Damage Tolerance in Aircraft Structures, ASTM STP 486, American Societyfor Testing
and Materials, 1971.
[3] Forman, R. G. et al, Journal of Basic Engineering, Vol. 89, Sept. 1967.
[4] Schijve, J. et al, "Crack Propagation in Aluminum Alloy Sheet Materials under Flight
Simulation Loading," NRL-TR67117U, National AerospaceLaboratoryNRL, Dec. 1968.
[5] Hsu, T. M. and Lassiter, L. W., "Effects of Compressive Overloads on Fatigue Crack
Growth," AIAA Paper 74-365, American Institute of Aeronautics and Astronautics,
April 1974.

Fractographic and Metallographic

Morphology of Fatigue Initiation Sites

REFERENCE: Eylon, Daniel and Kerr, W. R., "Fractographic and Metailographle

Morphology of Fatigue Initiation Sit~," Fractography in Failure Analysis, A S T M STP
645, B. M. Strauss and W. H. Cullen, Jr., Eds., American Society for Testing and
Materials, 1978, pp. 23S-248.

ABSTRACT: An approach for analyzing fatigue failures by investigating the under-

lying microstructure of the fatigue crack origin sites is presented. A precision section-
ing technique, developed by the authors, makes it possible to slice fatigue specimens
through the crack initiation sites thus permitting concurrent fractographic and metallo-
graphic examination. This failure analysis was found useful in determining the reason
for low fatigue lives of titanium alloy powder compacts; nonmetallic inclusions were
found to be the prime source for early fatigue initiation. In superalloy powder com-
pacts, residual porosity was related to the fatigue crack origins. In fl-proeessed or cast
titanium alloy products, the shear of large colonies of transformed a-platelets and frac-
ture along grain boundary a were identified as the main initiation mechanisms. The
identification and the elimination of the metallurgical crack initiating features has
the potential for improving the fatigue life and reducing the scatter of the fatigue re-
suits, since the largest proportion of fatigue life is spent in crack initiation.

KEY WORDS- fractography, fractures (materials), fatigue (materials), fatigue life,

crack initiation, crack nucleation, titanium, titanium powder, titanium castings, titan-
ium microstructure, heat resistant alloys, superalloy powder, nonmetallic inclusions,
precision sectioning

In recent years a great deal of research effort has been directed towards
investigation of fatigue crack growth behavior of different metals and alloys.
Yet, when the whole scope of the fatigue behavior of metallic systems is ex-
amined, one cannot escape the conclusion that in most cases the majority
of the fatigue life of a component or a specimen is spent in initiating a
crack rather than propagating it. In many commercial alloys and processed
products, this portion of fatigue life will be limited by the existence of
macro- and microimperfections in the structure. The identification and
1Department of Materials Science and Metallurgical Engineering, University of Cincinnati,
Cincinnati, Ohio 45221.
2Metals and Ceramics Division, Air Force Materials Laboratory, Wright-Patterson Air
Force Base, Ohio 45433.

235
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 1978 by ASTM International [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
236 FRACTOGRAPHYIN FAILURE ANALYSIS

investigation of fatigue fracture origins and the determination of their

metallurgical nature and source can lead to improved processing of mate-
rials, hence to components with higher fatigue life. A procedure for exami-
nation of fatigue initiation sites that has the potential of providing accurate
identification of the fracture origin will be presented here.
In certain fatigue life tests of flat, thin, notched specimens, it is possible
to detect the appearance of the initial crack in the course of the test run.
Figure l(a) is a replot of high cycle, axially loaded, notch fatigue test re-
sults for Ti-6A1-4V [1] 3 tested at different maximum stress levels. The per-
cent of the fatigue life spent in initiating a 0.5-mm crack (Ni/Nt x 100) is
plotted against the total number of cycles to failure, Nt where Ni is the
number of cycles to 0.5-mm crack initiation. It can be seen that in the 104
cycle region about 85 percent of the fatigue life is spent initiating the
crack, and in the 106 cycle region about 99 percent is spent in initiation.
This plot demonstrates the role of fatigue crack initiation in determining
the majority of the total fatigue life. Thus, any microstructural imperfec-
tion that may cause an early crack initiation will severely reduce the
fatigue life. Gell and Leverant [2] showed that in directionally solidified
Mar-M200 fatigue tested at room temperature, the fatigue life can be
decreased sharply with an increase of the size of the fatigue initiating micro-
structural defects, namely pores and MC carbides. Figure l(b) is a Ni/Nt
x 100 versus Nt replot of high temperature, low cycle, fatigue test results
of Ti-11 alloy [3] (6A1-2Sn-l.5Zr-1Mo-0.35Bi-0.1Si). In this work, the high
temperature test procedure limited the determination of the initial crack
length to 0.75 mm. All data points were obtained at the same stress level,
and the results indicate a wide scatter band between 104 to 10 s cycles. It
can be seen that while the longest life specimens spent almost 90 percent of
the fatigue life in initiating an 0.75-mm crack, the shortest life specimen
spent only 25 percent in initiation. This plot demonstrates the major role of
fatigue crack initiation in determining the scatter of fatigue results.

Current Methods of Crack Initiation Location

The present fractographic methods, utilizing optical and scanning
electron microscope (SEM) equipment, permit, in most cases, the identifi-
cation of the origin of a fatigue failure on the fracture surface. However, in
order to identify the metallurgical reason for this initiation, it is necessary
to examine the underlying microstructure. The authors were able to
perform this type of examination using a precision sectioning technique
discussed in detail elsewhere [4]. The precision sectioning provides separate
or combined metallographic and fractographic images of the initiation site.
Thus it is possible to have a microstructural, chemical, and fractographic
evaluation of the site as will be explained later in detail. Another section-
3The italic numbers in brackets refer to the list of referencesappendedto this paper.

100
A

9O J CENTERNOTCHED I
j RoOl I
9 ~ 3o Hz I
/ ~o AIR I
O J ROOMTEMPERATURE I
*" 9 HEATTREATMENT: 955"c/2 hr. 815~ hrAC.[
I
CRACKLENGTHFOR INITIATION= 0.5 mm I

80104
| , , I

105
. | , , [ , , , , , , , , I
106
NUMBEROF CYCLESTO FAILURE-Nt

A
100

80
ll (6AI-2Sn-1.5Zr-IMo-.35Bil
i.- TER NOTCHED
= = 0.1
60 clan
8% (IO00~
0 MN/mm2(45 ksi), maximumstn
i.u
65~ hr AC+
40 0~ hr, AC+
o
600~ hr, AC.

CRACKLENGTHFOR INITIATION= 0.75 mm

i i = , , ,,,| 9 , , i o , , ,

2~04 105 106

NUMBEROF CYCLESTO FAILURE- Nt

FIG. 1--Percent of total fatigue life spent in initiating (a) 0.5-mm crack in Ti-6AI-4V at
room temperature and different stress levels, and (b) O. 75-mm crack in Ti-ll at 538~ and at
a constant stress level.

ing method [5] provides a combined fractographic-metallographic image at

an angle of 120 deg to the fracture surface but lacks the precision required
to cut through a small fatigue initiation site and is limited in use to a
rather random sectioning through a fracture surface. Other methods that
provide fractographic-metallographic images [6, 7] use a stop-off lacquer to
protect a specific area on the fracture surface while its surroundings are re-
moved by electropolishing. These methods also do not have the accuracy of
the precision sectioning technique, and some inclusions in the initiation
site may be polished out preferentially and therefore would not be available
for microchemical analysis. Furthermore, the angle between the fracture
plane and the electropolished face cannot be determined accurately,

making it difficult to measure the angle between the tensile axis and the
observed shear planes in fatigue crack initiation locations.

Precision Sectioning Procedure and Results

Microstructural imperfections which may initiate fatigue cracks are
located best by testing smooth cylindrical specimens with rather long gage
lengths to allow relatively large gage volumes to he tested. Axial tension-
tension loading which preserves the fracture surface is preferred. Obtain-
ing subsurface, microstructural related origins requires that surface initia-
tion be minimized; this involves ensuring a good surface finish, free of
scratches and machining marks, and good alignment of test equipment to
prevent bending stresses.
Often it is possible to identify the defect causing the crack initiation
merely by examining the fracture surface. Figure 2(a) shows the fracture
surface of a /3-III titanium alloy ([Link]-6Zr-4.5Sn) powder compact,
smooth, fatigue specimen. The processing and heat treatment conditions of
this and other alloys discussed are listed in Table 1. The average room
temperature, high cycle fatigue life of the powder compacts (at 827
MN/m 2 peak load and stress ratio R = 0.1) were found to be more than
two orders of magnitude shorter than that for the wrought product, while
other properties like tensile strength and fracture toughness remained the
same [8]. The SEM image of the fracture surface indicates subsurface
initiation in Location A. A higher magnification image (Fig. 2(b)) shows an
inclusion lying close to the fracture surface; X-ray scan analysis identified
it as being silicon dioxide (SiO2).
In other cases, the source of the crack initiation is located well under
the fracture surface. Such an instance is shown in Fig. 3(a) (same material
and test conditions as in Fig. 2(a)). The precision sectioning method [4] in-
volves identifying the fracture origin in the SEM, cutting close to the origin
on a plane perpendicular to the fracture surface, measuring the distance
from the cut face to the origin, and grinding off this distance in a standard,
plastic mold, metallographic technique. After the specimen is polished to
the exact location (Line B in Fig. 3(a)) it is possible to obtain a micrograph
of the underlying microstructure, showing details that will help in the inter-
pretation of the fatigue failure. The SEM image in Fig. 3(b) is a modifica-
tion of this method in which the mount material has been dissolved care-
fully to make it possible to obtain the fracture surface at the origin and the
underlying microstructural features in the same SEM picture. The SiO2
particle that was not visible on the fracture surface can be seen at this
stage. It is also seen that the fracture has originated as a shear type frac-
ture on Plane D which lies close to 45 deg to the fatigue tensile axis which
is the vertical direction of the photograph. Unlike Fig. 2(b), the sectioned
plane image (Fig. 3(b)) can provide more information about the shape and

FIG. 2--(a) S E M image o f a subsurface fatigue failure origin, at A, of a ~-III powder com-
pact specimen, and (b) a S i 0 2 inclusion at the origin site.

TABLE 1--Processing and heat treatment conditions for the alloys discussed.

/~-III Powder Compacts Ti-17 Ti-6A1-4V Cast

extrusion (10:1) at 788~ 845~ h, AC + hot isostatically pressed

760~ h, WQ + 800~ h, WQ + (HIP) at 900~ MN/m2/2 h
482~ h, AC 635~ h, AC

IMI-68S AF-115

forged at 1050~ HIP 1190~ MN/m 2

1050~ h, O Q + forged at 1150~
850~ h, AC 1190~ h AC +
760~ h AC

NOTE--AC = air cooled,

WQ = water quenched, and
OQ = oil quenched.

size of the inclusion as well as how it interacts with the microstructure and
how the first crack was initiated. Previous precision sectioning work on
Ti6A1-6V-2Sn [4] powder showed that, in certain cases, diffusion of a or 3
stabilizing elements from the inclusion will change the neighboring micro-
structure by stabilizing either the a-phase or the B-phase. In those cases
where the a-phase was stabilized by the existence of oxygen in the inclusion,
cracks initiated within the embrittled a-phase rather than in the inclusion.
The same fatigue failure analysis was tried and found successful in super-
alloy powder compacts of an advanced nickel base superalloy, AF-115.
These compacts showed subsurface initiation origins when hourglass speci-
mens were fatigue tested at 760~ (1400~ R = - 1 and maximum stress
= 620 M N / m 2 (90 ksi). Figure 4 is a section through a fatigue specimen
at the initiation site. It can be seen that the fatigue crack was initiated
from an area containing clusters of unhealed pores (marked A) and large
unsolutioned 3'' particles (marked B). The two curved fgacture segments
(marked A ') may be the lower part of the pores which initiated the crack.
In many cases, the subsurface fatigue origins will result from certain
microstructural features rather than from inclusions, pores, or other de-
fects. Figure 5(a) shows the microstructure of a Ti-17 (SA1-2Sn-2Zr-4Mo-4Cr)
titanium alloy forged at 1065~ (1950~ This is 200~ above the 3-transus
temperature and resulted in an acicular transformed 3 microstructure.
This material was subsequently heat treated as indicated in Table 1. The
micrograph shows a colony of a platelets, all oriented in the same direc-
tion as at A, as well as longer and wider continuous grain boundary a film
(GBa) which outlines prior 3 grain boundaries as at B and B '. The flat
cleavage-like facet, C, in the center of Fig. 5(b) is the subsurface fatigue
initiation site on an hourglass specimen tested at 1034 M N / m 2 (150 ksi),
R = 0.1 at room temperature. This specimen had lower fatigue life than

FIG. 3--(a) SEM image of a subsurface fatigue fracture origin of a H-Ill powder compact
specimen. Using the precision sectioning technique, the fracture surface was sectioned along
line B. (b) SEM combined image of the sectioned plane and the fracture surface showing
an S i 0 2 inclusion (C) located under A. The fatigue crack initiated on shear plane D.

FIG. 4--SEM image of fatigue crack initiation site in a powder compact of nickel base
superalloy A F 115 showing cluster of pores at A and unsolutioned T' at B. The crack was
initiated at A '.

specimens forged within the o~ + /3 temperature range. Figure 5(c) is a low

magnification SEM picture of the section along line D in Fig. 5(b); E marks
the location of the crack initiation. Figure 5(d) is the light microscope
image of the metallographically polished and etched area under the initia-
tion site E. This clearly shows that the fatigue initiated at planar GBc~
similar to that at B' in Fig. 5(a). This GB~ is oriented close to 45 deg to
the major tensile axis (the vertical direction of the picture) and thus pro-
moted a slip related fracture on the long ~/~ interface which was oriented
in a direction close to the maximum shear stress. Figure 6(a) is the SEM
image of the same location as in Fig. 5(d), after the mount material was
dissolved carefully. This picture gives the combined view of the fracture
surface and underlying microstructure of the initiation site at higher mag-
nification. Since the GBa forms during slow cooling from above the 13 tran-
sus temperature, it may be possible to eliminate it with fast cooling from
the forging press and thus improve the room temperature high cycle fatigue
life. The same type of GBc~ fatigue crack initiation was found associated
with large colony structure in Ti-6AI-4V castings (Table 1) fatigue tested
at room temperature at 413 MN/m 2 (60 ksi) maximum stress, R = 0.1.

FIG. 5--(a) Ti-17 fatigue specimen with acicular microstructure showing colonies of trans-
formed c~platelets at A and grain boundary ~ at B and B ". (b) Flat fracture initiation site
at C was sectioned on line D. (c) Low magnification SEM image shows the initiation site at
E. (d) Underlying microstructure showing fatigue crack initiation along the GBr phase.

Figure 6(b) clearly shows the shear type fracture origin (A) at the or//3 inter-
face (B) of a planar GBa (C) inclined about 45 deg to the major tensile
axis.
The large colonies of ct platelets (D in Fig. 6(b)) are also typical of as-
cast microstructure of Ti-6AI-4V alloy and relate to another type of micro-
structure-related fatigue crack initiation. This is shown in Fig. 7(a) for the
test conditions just described. This type of fatigue initiation is associated
with the formation of shear bands across the colonies of similarly oriented

FIG. 6--(a) High magnification SEM combined image of the fracture initiation site at A
along the GBcr at C and the underlying microstructure of the specimen shown in Fig, 5,
(b) GB~ fatigue initiation, at A, alone the e~/~ interphase boundary (B) of the GB~ (C) in the
large colony (D) microstructure of east Ti-6A1-4V.

FIG. 7--(a) Trans-colony fatigue shear initiation in cast Ti-6AI-4V originated in pore A.
(b) Trans-colony shear initiation in I M I 685 titanium alloy showing slip traces in B.

c~ platelets. In the case of Ti-ll [3] and IMI-685 [9] (6A1-5Zr-0.5Mo-0.25Si)

alloys, the trans-colony shear fracture was found to dominate both initia-
tion and the propagation stages of the fatigue failure. In Fig. 7(a), the trans-
colony fracture may have originated from the small unhealed casting pore
(A). Also in this case, the shear type fracture propagated in a direction close
to 45 deg to the tensile axis. Figure 7(b) shows a similar trans-colony shear
fatigue initiation for IMI-685 titanium alloy (Table 1). Traces of another
slip system can be seen across the colony at B. Fatigue crack initiations
of this nature were associated with a fatigue life debit [9] only when a 5-
rain hold time was applied at the maximum stress position of the fatigue
cycle.

Discussion
To demonstrate the range of usefulness and the generality of the preci-
sion sectioning fatigue crack initiation analyses, a variety of fatigue failures
in wrought and cast titanium alloy material and powder compacts of titan-
ium and nickel base superalloys were presented in the previous sections.
Since surface fatigue crack initiation may be connected to surface eondi-
tions like machining marks, corrosion, and contamination, subsurface fa-
tigue origins were examined. It was the authors' experience that most fa-
figue specimens with subsurface fracture origins yielded unambiguous
answers on the nature of these origins. On the other hand, it was very
difficult to identify the fatigue initiation mechanism in surface initiated
failures.
The subsurface fracture origins can be divided into two groups: (a) ori-
gins related to defects such as pores and inclusions, and (b) origins related
to microstructural features. Using this analysis technique, it is possible to
distinguish between these two types.
The first group was frequently found in castings and powder compacts.
The most common fatigue initiating defect in titanium castings is porosity.
In powder compacts, fatigue cracks may initiate from a larger variety of
defects. They may be metallic or nonmetallic inclusions, high interstitial
brittle phases, or powder compaction porosity. In most cases, the identifi-
cation of the fatigue-failure causing defect can lead to improved processing
that will eliminate these defects. The porosity can be reduced or eliminated
by better casting or powder handling and compaction techniques. Identify-
ing the chemical composition of the metallic and nonmetallic inclusions may
often permit their elimination at the source of contamination. In the case
of titanium alloy powder compacts, heat treated to maximum tensile strength
[10], a very low content of nonmetallic inclusions (volume fraction less than
1-ppm SiO2) was sufficient to initiate an early fatigue failure. Since the
smooth fatigue life is so sensitive to these inclusions and since it is prac-
tically impossible to trace them by chemical analysis, smooth fatigue testing

and precision sectioning of the initiation site was found to be the only
effective method for detection and identification.
Fatigue origins relating to microstructural features were found in pow-
der, wrought, and cast materials. If low fatigue life is attributed to this
type of fatigue initiation, thermomechanical or heat treatment of the alloy
may be modified sometimes to eliminate the initiation related microstruc-
tural features. The elimination of the grain boundary a in the Ti-17 alloy
(Fig. 5(a)) by fast cooling from the/~-phase is a good example.
All of the discussed types of fatigue initiations can lead to a low fatigue
life or to a wide scatter of the fatigue test results. Since the initiation of
the first small crack in fatigue often constitutes the majority of the fatigue
life of a component or specimen, the improvement of the material and pro-
cessing to make it more fatigue-initiation resistant has the potential of in-
creasing the fatigue strength and reducing the fatigue life scatter.

Conduslons
1. The majority of the fatigue life often is spent in initiating the first crack.
2. Defects in the material and some microstructural features can promote
early crack initiation and thus early fatigue failure.
3. The precision sectioning analysis permits a complete characterization
of the fatigue failure origin.
4. Using this technique, the fatigue initiation sources in wrought, cast,
and powder compact titanium and nickel base alloys were successfully ana-
lyzed and identified.
5. The identification of the fatigue initiation sources may enable one to
eliminate them from the final product and thus has the potential of increas-
ing the fatigue strength and reducing scatter in fatigue life.

Acknowledgments

The authors gratefully acknowledge the assistance of W. A. Houston,

University of Cincinnati, and A. Kiefer and D. Kirk, Monsanto Research
Corporation, for the precision sectioning and the metallography, and M. B.
Strope and A. G. Jackson, Systems Research Laboratory, for the fracto-
graphy. Also, we would like to acknowledge N. Birla and J. M. Hyzak for
providing some of the test material.

References
[1] Eylon,D. and Pierce, C. M., Metallurgical Transactions A, Vol. 7A, 1976, pp. 111-120.
[2] Gell, M. and Leverant, G. R. in Fatigue at Elevated Temperatures, ASTM STP 520,
American Societyfor Testing and Materials, 1973, pp. 37-67.
[3] Eylon, D., Hall, J. A., Pierce, C. M., and Ruckle, D. L., Metallurgical Transactions
A, VoL 7A, 1976, pp. 1817-1876.

[4] Kerr, W. R., Eylon, D., and Hail, ]. A., Metallurgical Transactions A, Vol. 7A,
1976, pp. 1477-1480.
[5] Sasaki, G. and Yokota, M. T., Metallography, Vol. 8, 1975, pp. 265-268.
[6] Shechtman, D., Metallurgical Transactions A, Vol. 7A, 1976, pp. 151-152.
[7] Chesnutt, J. C. and Spurling, T. A., Metallurgical Transactions A, Vol. 8A, 1977,
pp. 216-218.
[8] Birla, N., DePierre, V., and Adair, A. M., Technical Report AFML-TR-76-8a, Air
Force Materials Laboratory, June 1976.
[9] Eylon, D. and Hall, I. A., Metallurgical Transactions A, Voi. 8A, 1977, pp. 981-990.
[10] Eylon, D. and Bifla, N., Metallurgical Transactions A, Vol. 8A, 1977, pp. 367-369.

Fractographic Analysis of Low Cycle

Fatigue Specimens from a Failed Steam
Turbine Rotor

REFERENCE: Kramer, L. D., "Fractographic Analysis of Low Cycle Fatigue Speci-

men= from a Falled Steam Turbine Rotor," Fractography in Failure Analysis, ASTM
STP 645, B. M. Strauss and W. H. Culleu, Jr., Eds., American Society for Testing
and Materials, 1978, pp. 249-274.

ABSTRACT: Because the TVA Gallatin Number 2 steam turbine rotor burst resulted
from creep-low cycle fatigue interaction, the present study was undertaken to establish
the effect of strain range, increasing hold times, and degree of segregation in the inter-
mediate temperature range corresponding to Gallatin origin conditions. Manganese-
sulfide (MnS) inclusions were found to be preferential initiation sites independent of
hold-time duration, surface oxide spiking, and most strain ranges, but no significant
decrease in specimen cyclic lives was observed to occur with variable MnS concen-
trations corresponding to different locations in the rotor. Fractographic simulation
of the Gallatin origin was obtained by imposing a 23-h hold time on bore segregated
material at 427~ and a 2.98 percent strain range. Periphery material or decreasing
hold times to a 4-h cycle caused a change from intergranular to transgranular fracture
at 427~ When selected hold-time tests were interrupted at less than 20 percent of ex-
pected cyclic life, early cracking was observed with the largest MnS linkup distance
occurring in bore segregated material.

KEY WORDS: fractography, chromium-molybdenum-vanadium, creep-low cycle fa-

tigue, rotors, turbines, inclusions, life (durability), separation, hold time, linkup, topo-
graphy, oxide spikes

As a result of the 19 June 1974 steam t u r b i n e rotor b u r s t in the T V A

G a l l a t i n N u m b e r Two unit, a n investigation was initiated at the Westing-
house Steam T u r b i n e Division to explain the failure. T h e s u b s e q u e n t
analyses [1-3] 2 developed the following conclusions.
1. T h e i n i t i a t i n g flaws were not gross m a n u f a c t u r i n g defects (that is, hy-
drogen bursts, u n c o n s o l i d a t e d porosity, etc.) although the forging c o n t a i n e d
positive alloy a n d m a n g a n e s e sulfide (MnS) inclusion segregation through-
out the n e a r bore origin metal volume. Because the origin was i n t e r n a l to
1Fellow materials engineer, Westinghouse Steam Turbine Division, Lester, Pa. 19113.
2The italic numbers in brackets refer to the list of references appended to this paper.

249
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 1978by ASTMInternational [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
250 FRACTOGRAPHY IN FAILURE ANALYSIS

the forging, external failure mechanisms such as stress corrosion cracking

were eliminated.
2. The origin location contained significant microstructural intergranular
creep damage associated with this near bore positive alloy and MnS inclu-
sion segregation. Elsewhere, in the rotor, this damage or microstructure
was not observed.
3. All mechanical properties testing, including creep methods, did not
show significant decreases. With the exception of the 23-h hold time 427~
strain controlled low cycle fatigue test of bore material, no other mechani-
cal specimen fractographically simulated the failed rotor. This result was
particularly surprising in that creep specimens of near origin material had
been tested over a reasonable range of Larson-Miller parameters with tem-
peratures as high as 593 ~ and times in excess of 10 000 h. These speci-
mens contained a similar microstructure to the initiating flaw, yet all creep
failures were by microvoid coalescence (that is, no intergranular cracking)
[4].
4. Using both the life fraction creep analysis as well as universal slopes
low cycle fatigue analysis, failure of the rotor could not be calculated on a
mechanical stress analysis basis. Only when a frequency modified Coffin-
Manson approach was utilized could the initiation of cracks (as defined by
Manson to be empirically 2 to 6 • 10 -4 mm in length) be calculated using
the TVA Gallatin cyclic conditions and the direct measurement of origin
fracture plane inclusion sizes and distances [3,5].
5. The initiation step leading to the critical flaw size for brittle bursting
of the rotor was concluded to result from plastic fatigue cycling of the near
bore MnS inclusions at service temperature with the steady state loading of
the turbine imposing the long hold times needed for creep damage. Since
the bore material is constrained by the surrounding rotor, this superimposed
hysteresis loop seen by the MnS inclusions is always positive (that is, the
only way to approach a symmetric hysteresis is to cool the periphery rela-
tive to the bore in such a manner that the compressive thermal stress ex-
ceeds the applied tangential stresses--an impossible condition in practice).
Because a large number of high temperature steam turbine rotors of this
1950's generic class of chromium-molybdenum-vanadium (Cr-Mo-V) air-
melted material are in service throughout the world, the Electric Power
Research Institute (EPRI) funded a research program (RP-502) to develop
calculation methods for predicting remaining rotor life given ultrasonic
data verification by dissection of actual rotors, unit operating history, and
rotor mechanical properties. This latter effort resulted in additional low cycle
fatigue testing of the Gallatin rotor to understand further the effect of ingot
segregation, cycling rate, hold time, and temperature on the initial mech-
anism to reach the critical flaw size for bursting [4,6]. Accordingly, the
EPRI test results, along with the earlier fatigue tests, are significant be-
cause of the lack of literature information on strain range and hold times

at temperatures less than 483 ~ where creep interaction effects are assumed
to be negligible. Conversely, higher temperature data exist for the current
ASTM Specification for Vacuum-Treated Carbon and Alloy Steel Forgings
for Turbine Rotors and Shafts (A 470-74), Class 8 grade [7-10] or similar
foreign grades [11-13] due to the wide applicability of this alloy to steam
turbine rotors, disks, bolts, and cylinders. Unfortunately, all such tests
correspond to steam inlet design parameters which are presumed to be
"worst case" creep-fatigue interaction conditions.
Since large turbine rotors are known to contain significant heterogeneities,
[14-16] turbine manufacturers often sample axial core bars and periphery
prolongation rings to check these variations as a condition of purchasing
production forgings. Conventional tension, stress rupture, Charpy V-notch
impact, and bend specimens often are required for these acceptance tests;
however, with one exception, low cycle fatigue testing [I 7] has not been used
to understand the role of segregation. The following paper is a summation
of both fatigue data from the Westinghouse failure investigation and the
later EPRI program. Together, the combined data are presented to com-
pare the effect of strain range, increasing hold times, and degree of segre-
gation in the intermediate temperature range corresponding to the Gallatin
origin conditions. Hopefully, a fractographic simulation of the origin topo-
graphy would result since the frequency modified low cycle fatigue analytical
model predicted a creep-fatigue interaction. The other conventional mech-
anical test methods (namely, tension, Charpy V-notch, Kk, fatigue crack
growth, and creep testing) obtained either cleavage, microvoid, or striated
fracture morphologies.

Experimental Procedure
All low cycle fatigue specimens were removed from the rotor in a tangen-
tial direction. The axial position of the specimens correspond as close as
possible to the origin location denoted in Fig. 7 of Ref 2. All specimens were
removed within 50 mm of the actual bore or periphery surfaces depending
on the degree of segregation to be studied. Due to the random nature of
the segregate banding as determined by macroetching, no reproducible
method was found to assure a full segregate band would exist completely
through the specimen gage diameter. For this reason, specimens probably
did not contain on a random basis the degree of through-section segrega-
tion that was available in the rotor itself. A typical bulk chemistry for the
origin bore material is presented in Table 1. While this area was observed
previously [2] to be segregated macroscopically, the analysis is within the
applicable specification of the 1950s material.
The specimen geometry was a threaded plain bar axial push-pull design
with a 7.6-mm reduced section diameter and a 16.5-mm gage length. All
testing was done by closed-loop str~.in control. In each case, the extensom-

TABLE 1--Chemical analysis of TVA Gallatin No. 2 IP-LP rotor material

Bore Material
Chemistry of
ASTM Method A 470-74, Gallatin 2
Element Class 8 Adjacent Origin

C 0.25 to 0.35 0.34

Mn 1.00 max 0.97
P 0.015 m a x 0.016 a
S 0.018 m a x 0.019 a
Si 0.15 to 0.35 0.27
Ni 0.75 m a x 0.07
Cr 0.90 to 1.50 1.10
Mo 1.00 to 1.50 1.43
V 0.20 to 0.30 0.25

a Met 1950s grade requirements. ASTM Method A 470-74, Class 8 values

represent improvements in subsequent steel making requirements.

eter was a linear variable differential transformer (LVDT) located external

to the surrounding electric resistant furnace. Temperature control of +__2 ~
along the specimen gage length was maintained over the test duration. Most
continuous cycle testing utilized an asymmetric triangular waveform from
zero to maximum strain and back to zero. When required, hold times were
imposed at the maximum strain point. The reasons for this particular wave-
form have been outlined previously [3], with Fig. 1 indicating the specifics
of both fatigue cycles. For comparison, a few symmetric loop tests also
have been conducted. A complete summary of test conditions are detailed
in Tables 2 and 3. All testing was done in air with no attempt to separate
environment as a test parameter. Subsequent data will show that the ex-
ternal environment did not affect specimen initiation sites.
Upon the termination of cycling, each specimen was inspected visually
under a x 60 optical binocular microscope to determine further areas of
interest such as initiation sites, secondary cracking in the gage length, etc.
Significant areas were documented optically and viewed by scanning elec-
tron microscopy (SEM). If oxidation of the specimen during the test dura-
tion was particularly bad, the specimen was cleaned electrolyticaUy by a
current reversing technique and was reinspected by SEM. For tests which
were terminated prior to the advent of gross cracking (that is, early life
specimens), the surface condition of the gage length was cleaned directly to
reveal exterior cracking and oxide spiking both optically and by SEM. All
significant specimens then were sectioned by conventional metallographic
techniques and etched using sodium tridecyl benzene sulfonate (STBS).
Grip ends of several fatigue bars including the 23-h hold time specimens
were sectioned after testing to assure that no preexisting flaws from the
forging manufacturing process were present in the relatively unstressed
ends. No flaws were found.

A STRAIN-TIME PLOT
~ MAX NO DWELL 0WELL

~" MINL

B. STRESS- TIME PLOT

0" M A X I ~ [~

TIME

C HYSTERSIS ON FIRST DWELL CYCLE

FIG. 1--Schematic diagrams for both continuous cycling and hold-time tests showing the
strain pattern imposed on the specimen, the resultant stress profile, and a typical initial hold-
time hysteresis loop.

Results and Dlscusslon

Low Cycle Fatigue Test Life Data

All low cycle fatigue life plots are presented in Figs. 2 and 3. Figure 2 is
a summation of all continuous cycle results utilizing temperature, cycling
rate, and location of specimen material as the appropriate test variables.
The significant observations in these data are the tightness of the scatter
and the intensitivity of the material to test temperature up to 427~ This
observation is particularly surprising since the 110~ data fall within the
higher temperature scatterband. Ordinarily, a material which follows "uni-
versal slopes" behavior would have a longer life presumably due to higher
tensile ductility at lower temperatures. Furthermore, the ll0~ points and
possibly the 371 ~ points in Fig. 2 show a preference for shorter periphery
fatigue lives; however, the few data points are grouped so closely that they
may not represent a sufficient statistical difference. On the other hand, both
the universal slopes deviation and this latter observation may be explainable
on the basis of differential ductilities of the banded and unbanded areas
within the specimen gage lengths since gross microstructural differences
that effect ductility (for example, grain size, amount of ferrite, etc.) were
observed elsewhere [2].
Figure 3 is the same format for all hold-time points with three tempera-
tures (371,427, and 482~ duration of hold time, and specimen location
as the test variables. For comparison, the Fig. 2 continuous cycling scatter-
band is superimposed over the hoId-time data. Three additional early life
specimens which were terminated at 20 percent or less of the respective test

TABLE 2--Summary of continuous cycle testing conducted on Gallatin No. 2 intermediate

pressure end material.

Test Life
Tempera- Cycle Total to
Specimen ture, Rate, Strain Failure,
Location ~ epm Range, % Nf Comments

Bore 110 1 2.25 206

Bore 110 1 1.12 1080
Bore 110 1 0.63 5827
Periphery 110 1 2.29 145
Periphery 110 1 1.13 692
Periphery 110 1 0.64 3964
Bore 371 1 4.80 23
Bore 371 1 2.10 319 "~
Bore 371 1 1.00 1015
Bore 371 1 0.72 2473
Periphery 371 1 4.80 3
Periphery 371 1 2.10 162
Periphery 371 1 1.00 669 iila
Periphery 371 1 0.72 1617
Bore 427 1 4.30 41 Figs. 6, 7, and 8
Bore 427 1 1.90 335
Bore 427 1 0.95 1475 a

Bore 427 1 0.60 3285 4 S

Periphery 427 1 3.05 73
Periphery 427 1 1.43 340
Periphery 427 1 0.75 1239 222"
Periphery 427 1 0.56 2359
Bore 427 100 1.83 205
Bore 427 100 0.81 1545
Bore 427 100 0.48 6080 iii~
Bore 427 100 0.33 40876

Symmetric loop tests.

c o n d i t i o n failure lifes a r e shown as t r i a n g u l a r points. Two significant p o i n t s

are evident f r o m these observations of Fig. 3. First, with the exception of a
single 4-h h o l d - t i m e point, the h o l d - t i m e d a t a d o not show a significant
decrease in life over t h e c o n t i n u o u s cycling d a t a . Thus, any creep d a m a g e
i n t r o d u c e d in the h o l d t i m e in t h e t e m p e r a t u r e r a n g e of 371 to 482~ d i d
not a p p r e c i a b l y shorten the fatigue life. Secondly, within t h e h o l d - t i m e
s c a t t e r b a n d , distinct t r e n d s in the d a t a are not obvious, p r o b a b l y b e c a u s e
s p e c i m e n - t o - s p e c i m e n m i c r o s t r u c t u r a l heterogeneities have a competitive
effect on fatigue initiation a n d growth with test t e m p e r a t u r e a n d hold-
t i m e d u r a t i o n . D e t a i l s of selected m i c r o s t r u c t u r e s will be discussed later.

E f f e c t o f Strain R a n g e (Continuous Cycling) on F r a c t u r e M o d e

F o r brevity, since the c o n t i n u o u s cycling tests showed the s a m e b a s i c
f r a c t o g r a p h i c f e a t u r e s irrespective of t e m p e r a t u r e , location of m a t e r i a l ,

Test Total Up and Life to

"11
Specimen Temperature, Total Cycle Hold Down Ramp Total Strain Failure, 3>
Location ~ Time, h Time, h Time, h Range, % Nf Comments
I'n
-In
Periphery 371 1 59/60 1/60 1.82 192
O
Periphery 371 1 59/60 1/60 1.00 2337 z
Bore 427 1 59/60 1/60 3.23 106 Figs. 9 and 10 t-
Bore 427 1 59/60 1/60 1.91 304 O
Periphery 427 1 59/60 1/60 1.74 (60) 20% Nf, Fig. 16
Bore 427 I 59/60 1/60 0.78 1800 O
-<
Bore 427 4 3 59/50 1/60 2.85 69 Figs. 11 and 12 O
Bore 427 4 3 59/60 1/60 1.59 145 r--
i11
Bore 427 24 23 I 2.98 76 Figs. 13 and 14
"-n
Bore 427 24 23 1 2.98 (12) 16% Nf, Fig. 15
Periphery 427 24 23 1 2.98 (7) 9~ Nf, Fig. 17 -t
Periphery 482 1 59/60 1/60 1.30 712 c
Periphery 482 1 59/60 1/60 2.90 97 ITI
co
a Parentheses denote initiation specimen terminated for destructive cut up. "U
Ill
O
Ill
Z
(D

bO
O1
O1

ZO
"11
73
I0 - - 0
Too' .to BORE
(3"11"C) [ ~ PERIPHERY 0
73
5
ROO'F ~A BORE
"0
I
~ 230"F - BORE .<

2 "11
,Ic
,-,,-
I--
Zm C
1 --
m

.5 z
r-
.<

if)
.2

.1 t I I lllill I , , '~lii[ , , , ,,,,,I z I , ,,,,,

101 102 103 104 105

CYCLES TO FAILURE:pNf
FIG. 2--A summary of all continuous cycling axial fatigue results.

I0
-. ~Continuous
Cycling Scatterband
5
P~
11

2 m
-~~~'-~~-. ~. _. 11
0
I z
i-"
0
:E
.5 C~
.<
H0,,
,me I Sca,,erDan, C~
r
{o,h, m

,2 iore 4hrs Perip

e,~ [] ~ 9000(!82"c) "11
--I

.I I I I I I I I I IIIIIli I I I I II111 I I I J lill c

m
i0 ~ 101 102 IO3 i0 4
"o
CYCLES TO FAILUREp Nf m
0
FIG. 3--A summary of all hold-time axial fatigue results with superposition of the continuous cycling
scatterband. m
z
or)

bO
0"1

and cycling rate, only the bore specimens which bracket the minimum and
maximum strain ranges at 427~ will be discussed. The fractography of
the lowest strain range (E, = 0.5 percent, N I = 7500 cycles) 427~ bore
specimen is shown in Fig. 4. Surface initiation occurred around 90 percent
of the gage circumference (as visually observed at x 60); however, subsur-
face initiation just adjacent to the surface (Area P) as well as internal initia-
tion (Area Q) both occurred at MnS inclusions. All initiation sites were
rubbed mechanically during the reversed ramp loading to zero strain, but
propagation topography was highly striated. Surface oxide spiking was not
observed. A metallographic section (Fig. 5) along the subsurface inclusion
in Area P revealed transgranular initiation with a slight amount of branch-
ing at the crack tip. Similar behavior was observed at secondary cracks
along the gage length below the main fracture.
By comparison, the maximum strain range (et = 4.3 percent, N I = 41
cycles) 427~ bore specimen solely initiated internally at MnS stringers.

FIG. 4--Topography of the minimum strain range (0.6 percent) bore 427~ continuous
cycling specimen fracture showing internal MnS inclusions (Area Q) near surface MnS in-
clusions (Area P), and zero deject surface locations which were all initiation sites.

FIG. S--Selected metallography through Area P in Fig. 4. Sodium tridecyl benzene sulfonate
(STBS) for etched microsections.

Figure 6 denotes the multiple level fracture surface with each level corres-
ponding to a MnS concentration. Since these inclusions were known to be
concentrated in inverse " V " segregation [2], it is not surprising that parallel
macroplanes are observed at each cluster concentration. Area Y in Fig. 6
is a typical example of this behavior. SEM fractography indicates that each
major inclusion or cluster of minor inclusions has a small patch of trans-
granular fatigue initiate from each discrete inclusion on a plane exactly
normal to the applied stress axis. Once growth proceeds sufficiently to link

FIG. 6--Topography of the maximum strain range (4.3 percent) bore 427~ continuous
cycling specimen fracture showing planar transgranular linkup of MnS inclusions internal to
the gage diameter (for example, at Area Y). Macroscopic observation of the specimen frac-
ture readily shows parallel MnS concentrations corresponding to Area Y.

up adjacent stringers, shearing occurs between these planar fatigue cracks.

As expected, a relatively small fraction of final topography is fatigue with
the bulk being shear dimples. The surface of the gage length was perfor-
ated with oxide spikes (Fig. 7); however, this phenomenon was not observed
to affect either initiation location nor propagation direction. Metallography
revealed transgranular main fracture propagation; however, secondary
cracking was significantly more branched than at lower strain ranges (Fig.
8).

E f f e c t o f H o l d Time (Constant Strain R a n g e )

In order to compare the effect of increasing hold time at approximately

constant strain range, three near bore specimens were run at 427~ to
failure. The initial hold-time test was a 1-h cycle to a strain range of 3.2

FIG. 7--Oxide spiking (arrows) on the gage section surface of Fig. 6.

percent (Nf = 106 cycles). Longer hold-time cycles of 4 h (Nf = 71 cycles)

and 23 h (Ni = 76 cycles) were tested at a total strain range of 2.85 and
2.98 percent, respectively. Details of the cycle ramp rates and hold times
are given in Table 3.
The 1-h cycle hold-time specimen cross section is presented in Fig. 9
and appears to have similar planar shearing as observed in the previous
example (Area K). Unfortunatley, the fracture topography was grossly
obliterated by oxidation due to fracture surface exposure during the test.
The gage section was penetrated uniformly with oxide spikes to 1.2 mm
depth (Area B). Metallography revealed transgranular cracking with various
degrees of branching (Fig. 10). Where MnS inclusions were present, trans-
granular cracks originated from these defects (Area S).
Detailed fractography of the 4-h cycle hold-time specimen is presented in
Figs. 11 and 12. This specimen has definite internal initiation at discrete
MnS inclusions (Areas U and J), transgranular striated fatigue linkup at
these inclusions, shearing between fatigue planes, and a preponderance of
surface spiking (Area Z) not associated with any subsurface initiation sites.
Metallography revealed all cracking to be branched but transgranular.
The 23-h hold-time specimen as-received condition is detailed in Fig. 13.
The primary fracture surface must have been open to the air for at least 20
cycles during testing, and the topography was oxidized by the furnace
atmosphere. This estimate is based on the observations of 5 percent load
drop in the stable hysteresis loop at cycle 47 as well as noticeable stable
loop shape deformation at cycle 56 (failure occurred at cycle 76). Careful
observation after electrolytic cleaning did reveal an intergranular remnant

0a)

"o
I
-<

"n
r-

z
I"
..<

FIG. 8--Branched transgranular secondary cracking off the primary fracture in Fig. 7. STBS etchant.

FIG. 9---Internal initiation (Areas K and S) at M n S inclusions and oxide spik~ag (Area B)
in the bore 1-h cycle hold-time specimen tested to failure at 427~ and a total strain range o f
3.23 percent.

0
==I
0
0
>
"o
=r
=~

"11

c
m

z
r-

FIG. lO--Selected metallography o f Areas K and S in Fig. 9 showing transgranular cracking in each case. Area K had been observed previously to
have a high local M n S concentration. However, due to the degree o f fracture surface oxidation, the subsequent electrolytic cleaning method also re-
moved most o f the M n S inclusions entrained in the oxide. Thus, no M n S inclusions remain in the microsection. S T l l S etchant.

FIG. l 1--Internal initiation (Areas J and U) at MnS inclusions in the bore 4-h cycle
hold-time specimen tested to failure at 427~ and a total strain range of 2.85 percent. Trans-
granular linkup of inclusions is obvious despite numerous oxide spikes (Area Z).

when viewed as a stereo pair. An axial metallographic section (Fig. 14)

through this specimen gage section revealed numerous intergranular secon-
dary cracks (Area F, Cracks 1, 2, and 3) below the primary plane of frac-
ture. Each crack corresponded exactly to a dark etching band of positive
segregation in a similar manner to cracking at the Gallatin No. 2 origin

FIG. 12--Selected metallograph of fatigue propagation (Area U) and oxide spiking (Area
Z). STBS etchant is used where etched.

FIG. 13--As-received condition o f the bore 24-h cycle hold-time specimen tested to failure
at 427~ and a total strain range o f 2. 98 percent. Even though the fracture had been oxidized
severely during testing, a stereo pair shows a distinct remnant o f intergranularity.

FIG. 14--Selected metallography of Fig. 13 showing greater inclusion linkup distances

associated with the darker etching, highly segregated Area F (Location 1 compared to Lo-
cations 2 and 3) Crack morphology was intergranular. STBS etchant is used where etched.

[2]. The more intense etching banded areas had closer spaced MnS incJu-
sions and thereby longer linkup distances. This statement is corroborated
by comparing Locations 1, 2, and 3 in Fig. 14. Detailed metallography of
Location 1 (Fig. 14) shows the same inclusion decohesion and branched
intergranular cracking observed in Gallatin. This behavior was also repro-
ducible in the 12-cycle early-life specimen which exhibited up to 1.2 •

10-4 mm branched intergranular linkup between inclusions internal to the

gage section (Fig. 15).

Effect of Material Location (Hold-Time Tests)

Since continuous cycling tests showed no difference in either initiation or

propagation topography with radial location (that is, variable segregation),
early-life hold-time specimens were sectioned to verify cracking morphology
differences with specimen location. Figures 16 and 17 are periphery mate-

FIG. 15--Similar intergranular initiation between MnS inclusions in another bore 24-h
cycle hold-time specimen tested at 427~ and a total strain range of 2. 98 percent. Testing was
terminated upon reaching 12 strain cycles (16 percent of life). STBS etchant.

FIG. 16--Transgranular initiation at MnS inclusions in a periphery 1-h cycle hold-time

specimen tested to 60 strain cycles (20 percent o f life) at a range o f 1.75 percent and 427~
Little segregation is observed at this position in the forging. STBS etchant.

rial microsections of 1-h (et = 1.74 percent) and 23-h (et = 2.98 percent)
hold-time specimens after 60 cycles (20 percent of Ns) and 7 cycles (9 per-
cent of Ns), respectively. In both cases, the structure was much less segre-
gated than the corresponding bore specimens; hence, the MnS mean free
distance was larger than previously noted. Slight inclusion-matrix
deeohesion occurred as well as transgranular linkup between a few
adjacent inclusions. No simulation of the GaUatin crack morphology was
observed.

General Comments
Based on the preceding experiments as well as earlier work, the initiation
mechanism leading to the development of elliptical flaws in the near bore
material of the Gallatin is most likely a creep-low cycle fatigue interaction
selectively occurring in bore segregate banded microstructure. Items sup-
porting this hypothesis are as follows.
1. Fractographic analysis of tension, impact, fatigue crack growth, frac-
ture toughness, and creep specimens could not simulate the Gallatin
failure [6,18].
2. Since the failure initiated internal to the forging, environmentally in-
duced effects were not a factor [2].
3. Based on a review of known forging quality defects, additional metal-
lography of selected macroetched rotor sections, nondestructive inspection
of the original failed rotor fragments, metallography of the specimens (in-

FIG. 17--Transgranular initiation at MnS inclusions in a periphery 24-h cycle hold-time

specimen tested to 7 cycles (9 percent of life) at a range of 2.98 percent and 427~C. Again,
little segregation is observed at this position in the forging. STBS etchant.

cluding the fatigue bar grip ends), and the appearance of the initiating
flaws themselves, no original manufacturing flaw existed in the rotor prior
to entering service with the exception of the alloy segregation and MnS
concentrations.
4. The imposition of a 23-h hold-time strain controlled fatigue cycle was
able to simulate the intergranular linkup between MnS inclusions in the
same bore microstructure as observed in the Gallatin rotor. No other fa-
tigue parameters resulted in a fractographic simulation. Furthermore,
when the bore material test was stopped at 16 percent of cyclic life, similar
linkup already had occurred when adjacent inclusions were of the same
interparticle distances as noted at the Gallatin origin (2 to 6 x 10-4 ram).
The degree of segregation and concentration of inclusions were probably
less than the actual origin location due to the random amounts of segrega-
tion in each specimen. The presence of significant intergranular linkup at
16 percent of cyclic life in the fatigue specimen thereby implies that areas
at least as heavily segregated in the actual rotor could undergo similar
early initiation. In other words, the presence of a fairly continuous planar
concentration of MnS inclusions with this 2 to 6 x 10-4 mm interparticle
spacing when linked up by a creep-fatigue interaction would be expected to
result in the critical flaw size for bursting (that is, the actual 5.5 by 0.25-
in. elliptical flaw). The 16 percent early life test was not intended to define
the exact cycle at which initiation occurred but rather to demonstrate that
crack linkup is extremely early in specimen life. Furthermore, the degree

of creep relaxation (that is, hold-time duration) to produce intergranular

linkup is not known; however, 23 h is a convenient value probably repre-
senting a realistic minimum value seen by many fossil steam turbines in
actual service.
5. Given these long hold times in fatigue, the location of material
(namely, the microstructure) has a significant effect on MnS inclusion link-
up frequency and fracture path. Since the periphery material (Fig. 16) was
less segregated, the MnS interparticle distance was greater thereby decreas-
ing the possibility of linkup. Furthermore, the unbanded periphery
material was found to be unlike the bore microstructure which was highly
duplexed in the bands [2, 6].
Although the preceding points support the creep-fatigue mechanism of
linkup between MnS inclusions of 2 to 6 • 10 -4 mm interparticle spacings,
two items are not resolved by these experiments. First, on a fatigue life
basis (NI), the hold-time loops are essentially no more damaging (that is,
shorter life) than the continuous cycle over the same strain ranges. Secondly,
when reviewing the bore hold-time data alone, the 4-h hold times appear
more damaging than either the 23 or 1-h holds. For both cases, com-
parisons with the same higher temperature data of Leven [8] for the same
class of alloy but with a homogeneous microstructure definitely imply in-
creasing hold times to be consequently more damaging than continuous
cycling. Accordingly, testing at lower temperatures such as 427~ one
would expect the same trend but with a lesser magnitude in damage with
increasing hold times. The difference between the expected and actual be-
haviors is obviously the heterogeneity of the Gallatin bore microstructure;
hence, the variable volume fraction of segregates present on a specimen-to-
specimen basis probably masks any real effect at hold time in the 371 to
483 ~ range. Early in the Gallatin failure analysis, the banded-nonbanded
volume fraction was investigated using television quantitative metallographic
techniques; however, due to the diffuse boundary of interdendritic segrega-
tion, no consistent workable method of defining segregate volume fraction
was obtained [18].

Conclusions
1. MnS inclusions preferentially act as internal crack initial sites for
both bore and periphery material at most test conditions; however, these
respective differences in inclusion concentrations had no significant effect
on actual specimen cyclic life. Only at very low strain ranges during the
continuous cycling tests were both surface and MnS initiations sites ob-
served.
2. At high strain ranges and long exposures, surface oxide spiking does
not seem to affect preferential MnS initiation.
3. Fractographic simulation of the Gallatin No. 2 rotor burst origin can

be obtained provided t h a t long hold times (that is, at least 23 h) are im-
posed on an asymmetric strain controlled loop (et = 2.98 percent) at 427~
in near bore segregated material. Removal of near bore segregation or
long hold times forced the topography to become transgranular (that is,
not typical of the actual failure topography).
4. Hold-time tests run to failure at 427~ did show essentially the same
fatigue life as observed in the continuous cycling tests; however, when three
selected specimens were terminated within 20 percent of the expected failure
life, early subsurface initiation at inclusions had occurred. The greatest
extent of planar MnS linkup occurred in the 23-h hold-time specimen from
bore material which had a 1.2 x 10 -4 m m intergranular crack after 16 per-
cent of expected life.

Acknowledgments

The author would like to t h a n k the Westinghouse Steam Turbine Division

and the Electric Power Research Institute for permission to publish these
data. In particular, the advice and comments o f B. B. Seth are gratefully
appreciated. Metallurgical analyses were conducted by D. Hamilton, T.
Calabrese, and J. M. Shinefield. T. Fabis of the Westinghouse Research
and Development Laboratories conducted the low cycle fatigue testing.
This work was sponsored under E P R I Contract R P 502, Dr. F. Gelhaus,
p r o g r a m manager.

References

[1] Hammond, J. C. and Schmerling, J. M., "Investigation of the Tennessee Valley Authority,
Gallatin Unit No. 2 Turbine Rotor Burst," Presentation before the 38th Annual Meet-
ing of the American Power Conference, Chicago, Ill., 21 April 1976.
[2] Kramer, L. D. and Randolph, D. D. in the 1976 ASME-MPC Symposium on Creep-
Fatigue Interaction, R. M. Curran, Ed., American Society of Mechanical Engineers,
New York, 1976, pp. 1-24.
[3] Weisz, D. A. in 1976 ASME-MPC Symposium on Creep-Fatigue Interaction, R. M.
Curran, Ed., American Society of Mechanical Engineers, New York, 1976, pp. 25-40.
[4] Kramer, L. D. et al, "Reliability of Steam Turbine Rotors-2nd Semi Annual Report,"
EPRI Contract No. RPS02-4, Electric Power Research Institute, 19 April 1977.
[5] Manson, S. S., Experimental Mechanics, Vol. 7, No. 5, May 1965, pp. 193-226.
[6] Kramer, L. D. et al, "Reliability of Steam Turbine Rotors--lst Semi Annual Report,"
EPRI Contract No. RPS02-4, Electric Power Research Institute, July 1976.
[7] Krempl, E. and Walker, C. D. in Fatigue at High Temperature, A S T M STP 459, Amer-
ican Societyfor Testing and Materials, 1969, pp. 75-99.
[8] Leven, M. M., Experimental Mechanics, Vol. 13, No. 9, Sept. 1973, pp. 353-372.
[9] Curran, R. M. and Wundt, B. in Reports of Current Work on Behavior of Materials at
Elevated Temperatures, A. O. Schaefer, Ed., American Society of Mechanical Engineers,
New York, 1974.
[10] Curran, R. M. and Wundt, B. in 1976 ASME-MPC Symposium on Creep-Fatigue Inter-
action, R. M. Curran, Ed., American Society of Mechanical Engineers New York, 1976,
pp. 203-283.
[11] Hill, G. L, in Thermal and High Strain Fatigue, Institute of Metals, London, 1967,
pp. 312-327.

[12] Coles, A. et al in Thermal and High Strain Fatigue, Institute of Metals, London, 1967,
pp. 270-294.
[13] Coles, A. and Chitty, A. in Thermal and High Strain Fatigue, Institute of Metals,
London, pp. 328-345.
[14] Smith, H. C. et al, "The Nature and Source of Nonmetallic Inclusions in Large Forgings,"
Presentation to the 1977 Annual Meeting of the International Forgemasters Conference,
Kyoto, 23-28 Oct. 1977.
[15] Ferdinandez, S., "Influence of 'A' Segregations on the Mechanical Properties of Forg-
ings Obtained from Vacuum Poured Ingots," Sixth International Forgemasters Meeting,
Cherry Hill, N.J., 1972.
]16] Snow, R. B., "Source of Inclusions in Forging Ingots," Sixth International Forgemasters
Meeting, Cherry Hill, N.J., 1972.
[17] Watanabe, J. et al, "A Study of Crack Initiation in Rotating Parts," Presentation to the
2nd International Conference on Mechanical Behavior of Materials, Boston, 16-20 Aug.
1976.
[18] Kramer, L. D., Unpublished data.

Role of Interface Chemistry in Failure of

Materials

REFERENCE: Joshi, A., "Role of Interface Chemistry in Failure of Materials,"

Fractography in Failure Analysis, A S T M STP 645, B. M. Strauss and W. H. Cullen, Jr.,
Eds., American Societyfor Testing and Materials, 1978, pp. 275-293.

ABSTRACT: Properties of many materials are influenced strongly by small compo-

sitional changes that occur at interfaces such as phase and grain boundaries. Knowledge
of the elements involved and whether they are segregated or present in a second phase
provides a better understanding of the failure mechanism. In many instances of grain
boundary failures, the chemistry provides the most comprehensive and useful informa-
tion. Examples of such studies discussed here include temper embrittlement and other
forms of impurity-induced low temperature embrittlement, weld-metal embrittlement,
grain boundary corrosion, and intergranular stress corrosion cracking. Failures at
phase boundaries and other more macroscopic interfaces such as encountered in fiber/
metal matrix composites, interparticle interfaces in powder metallurgy materials and
bonded (thermal compression, solder, or diffusion bonded) [Link] can be under-
stood by close examination of chemistry of these interfaces.
In many of these examples, the zone of segregation at boundaries is very small, only
a few angstroms, the same as the boundary width. The interface chemistry has been
studied successfullyin a scanning Auger microprobe by Auger electron spectroscopic
analysis of grain boundary fracture surfaces obtained inside the vacuum system.

KEY WORDS: fractography, Auger electrons, spectroscopy, embrittlement, fractures,

(materials) grain boundaries, corrosion, stress corrosion, weld metal, composite ma-
terial, hydrogen embrittlement

Properties of m a n y polycrystalline materials are i n f l u e n c e d strongly by

the presence of solid-solid interfaces such as the phase a n d g r a i n b o u n d a r i e s .
These interfaces, like surfaces (solid-vapor interfaces), are high energy
regions having a n atomic structure different from t h a t of the b u l k . F u r t h e r ,
thby often act as a source or sink for atomic defects a n d are preferred sites
for solid state reactions, such as phase t r a n s f o r m a t i o n s a n d precipitation.
The kinetics of the processes involving g r a i n b o u n d a r i e s also are different,
resulting in accelerated diffusion, growth of phases, etc. The structural
difference between the g r a i n b o u n d a r i e s a n d b u l k is the basic reason for

1Assistant director, AnalyticalLaboratory, PhysicalElectronicsIndustries, Inc., Eden Prairie,

Minn. 55343.

275
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 by ASTMInternational [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
276 FRACTOGRAPHY IN FAILURE ANALYSIS

these differences and also leads to localized variations in composition of

alloys due to equilibrium as well as nonequilibrium processes of segregation.
The differences in interface chemistry and bulk chemistry can be signif-
icant. In extreme eases, the interfaces may be occupied completely by a
layer of foreign atoms whose concentration in the bulk is in the parts per
million range. Such differences can arise due to segregation during heat
treatment or thermomechanical treatments, diffusion from surfaces, or due
to vacancy drag in certain quenched materials. The compositional changes,
even if they are small, can influence significantly many material properties,
such as mechanical strength, corrosion, and stress corrosion behaviors.
Knowledge of the elements involved and whether they are segregated or
present in a second phase provides better understanding of the failure
mechanism.
The unambiguous detection of segregated species has been difficult for
quite sometime. The chemical etching, neutron activation, and electron
probe techniques only rarely have been successful in detection. In most
techniques, the difficulty arises primarily due to inadequate resolution or
low detection sensitivity. The advances in surface analysis techniques, in
particular Auger electron spectroscopy (AES), now make it possible to
detect segregation levels as low as 0.1 atomic percent of most elements
present to within a monolayer at interfaces. At present, AES is the most
widely used technique for interface chemical studies and the purpose of
this article is to review some of the understanding provided by this tech-
nique in relating interface chemistry to material properties.

Auger Electron Spectroscopy to Study Interface Chemistry

The high surface sensitivity of the AES techniques is used in obtaining
interface chemistry. The examination of interface chemistry is facilitated
by fracturing the material in such a manner that the interfaces are made
into surfaces. Thus, in many instances of grain boundary embrittlement,
the fracture surfaces are the prior grain boundaries and analysis of such
surfaces represents grain boundary chemistry. The analysis of boundaries
in materials that do not fail by grain boundary fracture is difficult, and
techniques are being developed to facilitate studies in such materials as
well.
The principles and techniques of AES have been described in detail else-
where [I] 2 and will not be repeated here. An Auger spectrum represents
the composition of the top 5 to 20 ,~ of the surface and is presented most
commonly as the derivative of the electron energy distribution, dN(E)/dE,
as a function of kinetic energy, E. Elements present are simply identified
from the energy positions at which peaks occur in the Auger spectrum.
Since the peak-to-peak heights are generally proportional to the atomic
2The italicnumbersin bracketsreferto the listofreferencesappendedto this paper.

concentration [2], quantitative calculations simply can be made by utilizing

the elemental sensitivity factors [3,4]. In order to minimize the contami-
nation of the fracture surface from the atmosphere, the fractures are often
performed within the vacuum system. While in some cases, externally frac-
tured and field fractured surfaces have been examined with fair success,
fracture within the high vacuum system is required in most cases. Ultra-
high vacuum systems providing pressure in the 10-10 torr range having oil-
free pumping devices are found to be extremely satisfactory.
The Auger spectra also contain a wealth of chemical binding information,
which is usually derived from the peak shifts and peak shape changes.
Many metals, such as aluminum, silicon, magnesium, vanadium, tantalum,
niobium, iron, titanium as well as non-metals such as boron, phosphorus,
sulfur, and chlorine can be distinguished in their oxide forms from their
elemental forms. Carbon exhibits peak shape changes that are distinctly
different for carbide from carbonate [5], graphite, diamond, and carbon
monoxide [6]. Figure 1, from the work of Haas et al [6], depicts some of
these differences. The micrograph shown in Fig. 2 is a secondary electron
image obtained from a polished nodular cast iron using a scanning Auger
microprobe. The carbon peak shape obtained from the nodule (Fig. 2) is
typical of that of graphite while the spectrum obtained from the pearlite
region (Fig. 3) suggests the presence of both carbide and graphite. In

COonW(112) ~ ~

GRAPHITE ~ .1-
/
~, /
/

/',/
I I I I
",J
I I
220 250 290
ELECTRONENERGY,eV

FIG. 1--Spectra o f the carbon KLL Auger transitions for carbon monoxide (CO) on
tungsten (112), tungsten carbide (W2C), graphite, and diamond (after Haas et al [6]).

FIG. 2--Auger spectrum from the graphite nodule. Spectrum was obtained after sputter
cleaning the polished surface with argon ions.

addition to obtaining the point spectra from selected regions of the surface,
modern instruments also permit selected element Auger imaging. Figure 4
shows the Auger elemental images of carbon and iron from the polished
cast iron surface discussed in Fig. 3. Auger imaging of this nature at the
fracture surfaces provides information on uniformity of grain boundary
segregation and on occurrence of second phases.
Depth distribution of elements is accomplished by Auger analysis in
conjunction with ion sputtering. Presently available multiplexing units
permit simultanoous display of approximate atomic concentrations as a
function of depth. The example in Fig. 5 shows that iron oxides rather
than chromium oxides form on the surface upon oxidizing a Type 304 stain-
less steel at 500~ in air, an observation contrary to the common expecta-
tion of chromium oxide formation. Combining the ion sputtering techniques
with the capability of Auger imaging amounts to what may be closest to a
three-dimensional analysis in practical materials.

Material Failure Studies

Changes in interface chemistry often can explain material failure prob-
lems associated with interfaces. The properties that are quite often in-

FIG. 3--Auger spectrumfrom the pearlite region.

fluenced are corrosion, stress corrosion cracking, hydrogen embrittlement

behavior, and mechanical properties that encompass low and high temper-
ature embrittlement and fatigue. AES studies have made it possible to
correlate chemistry to material failure and properties, and some are dis-
cussed in the following paragraphs.

Low Temperature Embrittlement

Low temperature embrittlement is a widely investigated phenomenon.
One particular case of extensive investigation has been the failure in tem-
pered steels. The problem commonly known as temper embrittlement oc-
curs in alloy steels when tempered in the range of 350 to 525 ~ or slowly
cooled through this range of temperature and is characterized by a drastic
increase in the ductile to brittle transition temperature. The embrittlement
is reversible and completely eliminated when the steel is heated to about
600~ for a few minutes. Auger examination [7-11] of freshly fractured
surfaces of temper embrittled steels has shown the presence of elements such
as phosphorus, tin, and antimony which have long been suspected. Figure
6 shows the Auger spectrum obtained from an antimony doped (bulk con-
centration 280 ppm) experimental 3340 steel, showing grain boundary en-

FIG. 4--(a) Secondary electron, (b) carbon Auger, and (c) iron Auger images of the polished
cast iron surface.

""-. O x l -50~, .............

"-............... -..:~ : : ..'""Fe x 2
,o ..,
,.,.-.,,
.'"'Fe X 2"". """ "':':r Cr x 5 .
ffi .'" "'" ......... " ""..Oxl
9" Nix 5..'" .......... ":~"?." ............
...'Cr x 5 .'"HEATED N A R ........
~ " ' . . . . . . . . ."" 500~ FOR 75 MINUTES
ETCHING TIME

F I G . 5--Depth-composition profile for Type 304 stainless steel showing the ( A ) iron oxide,
( B ) chromium rich oxide, and ( C ) oxide/steel interface region~.

Fe
o 2oo 4oo soo aoo lOOO
ELECTRONENERGY,eV
FIG. 6--Auger spectrum obtainedfrom the intergranularfracture surface of a 3340 steel.

richment of not only antimony ( - 4 atomic percent) but also chromium,

nickel, and carbon. The carbon peak shape clearly suggests its occurrence in
the form of a carbide. Detailed studies [7] of this alloy system (Fig. 7) also
showed that the shift in the transition temperature could be directly corre-
lated to the antimony segregation. The results (Fig. 8) also demonstrate
clearly that grain boundary segregation of antimony, nickel, and chromium
is increased with increasing bulk antimony concentration suggesting that
metalloid-metal (for example, antimony-nickel) interactions predominate
leading to segregation. The extent of segregation at the boundary of the
metalloid impurities has been studied in conjunction with ion sputtering.
Figure 9 shows the summary of such studies [11] in many ferrous systems.
In all these cases, metalloid segregation occurred in a very narrow region of
less than S A at the grain boundary. While these observations are consis-
tent with the equilibrium models of segregation, presence of a nickel en-

E ~ NICKEL
o CHROMIUM 8

~ ~ 0~
4

'1
oJ .. ,
SHIFTIN TRANSITIONTEMPERATUREI~

FIG.7--Shift in transition temperature related to segregation in alloy steels (after Joshi and
[7]).
Stein

.+ :I o ANTIMONY
,, NICKEL
o CHROMIUM
lO

ANTIMONYIN BULK(PPM)
70O
0

FIG. 8--Segregation of nickel, chromium, and antimony to grain boundaries in nickel-

chromium-carbon steels versus antimony bulk concentration (after Joshi and Stein [7]).

riched region to greater depths near the boundary suggests predominance

of nonequilibrium forces in segregation of this element. Other studies have
also shown that elements such as sulfur [12], tellurium [I3], manganese
[14], and silicon [14] segregate to grain boundaries and thereby influence
the low temperature mechanical behavior. Precipitation of titanium carbides
and segregation of boron in maraging steels [15] also is shown to enhance
intergranular embrittlement.
Low temperature embrittlement and failures have been investigated in
many nonferrous systems. AES studies dearly correlated deterioration in
mechanical properties due to segregation of bismuth [16] and sulfur in
copper, phosphorus [17] in tungsten and sulfur [18] in nickel alloys.

10l x
o
S on AIS15140
S on Fe- 0.6 Sb
o p on AIS15140
0.8 o Sb on AIS13340
Sb on Fe. 2.2 Sb
J o\o v Sb on Fe. 0.6 Sb
9 Pon Fe-Ni-Cr-C-P
9 Sn on Fe- Ni- Cr- C- Sn
0.6 | v
z

~ o
0.4
.o

| 0.2 9 o o
\|
~ \x~
,~ Ill 0

5 10 15
DEPTH,A

F I G . 9--Normalized concentration of segregated solute on fracture surface versus average

amount removed by ion sputtering (after Marcus et'al [11] ).

Many of these studies were performed on specimens that failed pri-

marily in the intergranular mode upon fracture. The development of the
scanning Auger instruments makes it possible now to examine specimens
that only partially fail in the intergranular mode. Since the scanning Auger
microprobe (SAM) systems [19] combine scanning electron fractography
with Auger analysis, the areas of interest can be selected and complete
analysis performed from the selected intergranular portions. Auger ele-
mental imaging is also extremely helpful in doing the analysis. Figure 10
shows the results of a study performed on a 2.25Cr-1Mo commercial plate
steel that failed primarily by cleavage. Prior to the fracture within the
vacuum system, the specimen received a 955~ (1750~ austenitization
followed by a quench, a 6 h 695~ (1280~ plus 10 h 665~ (1225~
temper, and a 10 000 h 480~ (900~ embrittlement. The elemental
imaging clearly shows enrichment of phosphorus, molybdenum, carbon,
and chromium in the intergranular portions of the fracture surface. Spectra
obtained from the cleavage and in the high phosphorus regions (Fig. 11)
clearly show the differences between the bulk and grain boundary chemistry.
The data also strongly support the hypothesis that phosphorus segregation
occurring at the molybdenum and chromium carbide rich phases present
at the prior austenite grain boundaries causes intergranular weakness.
Embrittlement of weld metals and the heat affected zones (HAZ) of the
weld have been investigated in few alloy systems. The grain boundary
fractures that occur at the welds and HAZs appear to be some form of im-
purity induced embrittlement. Figure 12 shows an example of a 2.25Cr-lMo

FIG. lO--Auger elemental images of chromium, molybdenum, carbon, and phosphorus

obtained from the fracture surface of an embrittled 2.25Cr-1Mo steel

submerged arc weld metal failed along grain boundaries and characterized
by phosphorus and tin enrichment to the interfaces which are also rich in
chromium and molybdenum carbides. Small amounts of tin and nitrogen
also were detected.
Impurity elements segregated at the grain boundaries can further enhance
embrittlement in presence of hydrogen. The metalloid impurities are be-
lieved to be acting as hydrogen recombination poisons thereby keeping
hydrogen in an atomic state and accelerating its diffusion along the grain
boundaries, where the metalloid elements are segregated. The experiments
performed on nickel [20] suggest that tin and antimony act as such re-
combination poisons. These and other studies suggest that impurity elements

CLEAVAGE
SURFACE

Fe I
Fe Fe

t xl

Fe Fe

200 400 600 800 1000

ELECTRONENERGY,eV

FIG. 11--Auger spectra from cleavage and intergranular fracture surfaces of 2.25Cr-lMo
steel.

may play an important role in several other incidences where hydrogen

alone was believed to be the cause for intergranular failures.
The information presented here suggests that impurity elements were
detected in all events of intergranular embrittlement. The only exceptions
noted were in quenched steels [7,21] where there might only be small or
undetected segregation or that the fracture may be influenced strongly by
the internal stresses resulting from the quench.

High Temperature Embrittlement

High temperature embrittlement is complicated by the fact that diffusion
of alloying elements and impurities could occur under high temperature
and stress conditions. Studies performed on Type 304 stainless steels [22]
clearly indicated that high temperature and stress accentuate segregation
of impurity elements such as sulfur and phosphorus to grain boundaries.
Direct evidence for the grain boundary segregation of sulfur, bismuth,
tellurium, and lead has been shown by Auger spectroscopy in super alloys
that were stressed at elevated temperatures [23].

Interphase Embrittlement
Many materials, whether conventional alloy systems or composite ma-

dN
dE

0 200 400 600 800

[Link]

FIG. 12--Fracture surface and Auger spectrum from it depicting grain boundary compo-
sition of 2.25Cr-lMo submerged arc weld metal.

terials, can fail along phase boundaries. Analysis of such interfaces by

Auger spectroscopy is no different from examining the grain boundaries.
Figure 13 shows the fracture obtained within the vacuum system of an
aluminum/carbon composite material. The decohesion of the matrix and

FIG. 13--Auger spectrum showing the interface chemistry of the aluminum~carbon fiber
composite material.

carbon fibers was studied by AES analysis of the freshly opened interface.
The Auger spectrum in Fig. 13 clearly indicates the enrichment of mag-
nesium and oxygen at this interface. The magnesium Auger fine structure
(plasmon loss peaks), characteristic of its chemical environment, is sug-
gestive of existence of magnesium oxide at this interface. The magnesium-
oxide layer was believed to be the reason for poor adhesion between the
carbon fiber and the aluminum matrix. An examination of the process
variables suggested that segregation of magnesium occurs from the
aluminum during processing.
Embrittlement in powder metallurgical materials can be even more com-
plex as it can be influenced strongly by the surface chemistry of starting
materials and powder preparation techniques as well as the processing
variables. The example of a careful study [24] performed on powder metal-
lurgy compacts showed that several elements such as sulfur, oxygen, chlo-
rine, and carbon predominate at the original particle interfaces. The study
also indicated that oxygen and not sulfur or chlorine or carbon has the
damaging effects on the strength properties or density of the compacts
examined. Figure 14 shows such a correlation and clearly indicates that

0 - - I I
25 30
TENSILE STRENGTH, PSi x lO~O

FIG. 14--Variation o f sulfur, chlorine, and carbon at the fracture surface of powder metal-
lurgy iron versus tensile strength (after Joshi et al [ 2 4 ] ) .

correlations of this sort are important before arriving at conclusions on

what elements are responsible for the property deterioration.
Similar to the phase boundaries just discussed are other macroscopic
interfaces whose mechanical properties are influenced by interface chem-
istry. Examples are bonded (thermal compression, solder, or diffusion
bonded) interfaces, platings, and other thick and thin coatings. In several
of these cases, AES studies have shown that interface chemistry plays an
important role on adhesion at surfaces. A brief review on adhesion studies
may be found in Ref 1.

IntergranuIar Corrosion and Stress Corrosion Cracking

Second phase precipitation and solute segregation at grain boundaries
can significantly change the intergranular corrosion, stress corrosion crack-
ing (SCC), and liquid metal penetration properties of the material. Some
of the early work done on understanding the intergranular corrosion proper-
ties of stainless steels and recent studies on intergranular SCC properties
will be discussed in the following paragraphs.
Intergranular corrosion of Type 304 (austenitic) stainless steels sensitized
in the 450 to 900~ (842 to 1652~ range has been attributed widely to
the chromium depletion resulting from formation of grain boundary car-
bides. Several corrosive environments, generally highly oxidizing in nature
such as nitric a~id (HNO3) solutions, are found in which intergranular
corrosion of stainless steels occurs in solution treated condition with no
detectable carbides at grain boundaries. Solute segregation at grain bound-
aries that was undetected by the transmission electron microscopy and other

techniques was believed to be responsible for corrosion in these solutions.

In a careful study [25] the corrosion susceptibilities of Type 304 stainless
steel in Strauss and Huey solutions as a function of heat treatment were
conducted and related to the grain boundary compositions obtained by
Auger spectroscopy. The results shown in Table 1 clearly suggest that sulfur
segregating in nonsensitized stainless steels permits intergranular attack in
nitric-dichromate (FINOa-K2Cr207) solutions while chromium depletion
and not sulfur segregation is responsible for corrosion in sulfuric acid-
copper sulfate (H2SO4-CuSO4) solutions. The evidence for chromium de-
pletion also was obtained in this study and is shown in Fig. 15. The results
obtained from other examined steels also suggest that nitrogen, phosphorus,
and silicon may play a role in deteriorating the corrosion properties in nitric-
dichromate solutions.
Stress corrosion behaviors in which the cracking occurs along grain
boundaries have been correlated with grain boundary chemistry obtained
by AES analysis of in-situ fracture surfaces. Sulfide stress cracking (SCC)
of commercial high strength steels [26] and 7075 aluminum alloys [27] to
be discussed here are particularly interesting since hydrogen is believed to
be the predominant factor in determining the cracking behavior in these
materials. The SSC study correlated the boundary chemistry of selected
high strength steels with the minimum stress at failure in hydrogen-sulfide
solutions at room temperature. The studies indicate that either phosphorus
present at grain boundaries or sulfur and manganese, precipitated in a
thin film of (iron, manganese) sulfur, degrade the SSC properties of these
materials. Correlation of average grain boundary sulfur concentration to
the minimum stress at failure is shown in Fig. 16 and demonstrates the
role of sulfur in SSC.
The stress corrosion experiments [27] on 7075 aluminum alloy were con-
ducted in 3.5 percent neutral sodium-chloride solutions. The specimens
for the tests were prepared by solution treating at various high temperatures
for 1 h, water quenched, and aged for 24 h at 120~ (250~ Auger data
from the fracture surfaces were obtained within a vacuum system on speci-
mens that were similarly treated. The effect of heat treatment on the relative
amounts of magnesium, zinc, copper, silicon, and iron in the grain bound-
aries was of most significance. In otherwise similarly treated alloys, solute
enrichment was found to increase with the solution treatment temperature
and could be correlated extremely well with the stress corrosion crack plateau
velocities. This result, as in the SSC studies just discussed, clearly suggests
that there may be other important variables besides hydrogen to account
for the observed properties. Solute or impurity segregation is one such
variable shown to play an important role in hydrogen-assisted cracking
behavior. Auger spectroscopy, by providing direct analysis of grain boundary
composition, allows this variable to be related to the corrosion and SCC
properties of materials.

-I]
2O
3~
o
TABLE 1--Auger spectroscopic analysis of Type 304 stainless steelfracture surfaces. O
fi3
:D

Corrosion Rate, "o

1"
mg/cm 2 h -<

Boiling Nitric- Modified

Treatment C/Fe Si/Fe S/Fe Cr/Fe Ni/Fe Dichromate a Solution Strauss Test b -ff
>_
r"
2 h, 1050~ WQ c 0.0128 ND d 1.230 0.664 0.102 2.25 negligible
2 h, 1050~ WQ m
+ 2 h, 850~ WQ 0.177 0.066 0.588 0.515 0.111 0.84 0.016
2 h, 1050~ WQ z
+ 2 h, 650~ WQ 0.011 0.111 0.920 0.525 0.109 2.93 negligihle "
r.<
2 h, 1050~ WQ
+ 2 h, 600~ WQ 0.0645 ND 0.881 0.548 0.103 2.64 negligible
2 h, 1050~ WQ
+ 3 days, 650~ WQ 0.283 ND 0.850 0.459 0.119 0.87 0.447

a Weight loss rate determined at the end of 14-h corrosion test in nitric-dichromate solutions.
b Weight loss rate determined from a 60-h corrosion test.
c WQ = water quenched.
dND = none detected.

- Cr
0.7 9 -Ni
Cr BULKLEVEL ~ s
O 0.6 o
I--
<rr" 0.5 n-.
~- 0.4 [Link] ''=

0,3' O. 15

0.2 : 9 N;BULKLEVEL-e 0. I 0

0.1 0.05

0 I i I I I0 0
200 600 IOOO 1,4OO 20 O
APPROXIMATE ATOMIC M O N O L A Y E R S R E M O V E D

FIG. 15--Concentration profiles of chromium and nickel as a function of distance from

fracture surface of a Type 304 stainless steel sensitized by heating at 600~ for 2 h (after Joshi
and Stein [25]).

0.08

o X-125 [8)
0.07 o X-12518)

0.06
$ v-150(tOl
x-140i121 1.5
0.05 !
= 0~4

1.O ~_

0.03
o P-110(C123)
o P-110[11)
0.02
o.5 ~

P-11Q(11)o ~ o P-11O(1111
0.01

o P-110 115) o X-90 181

/ 1121
3O
MINIMUM STRESS AT FAILURE (% of UBSI

FIG. 16--Sulfur content of the predominantly intergranular fracture surface as a function

of minimum stress at failure in hydrogen sulfide solution~ [26].

While these studies are attempts to relate fracture surface composition

to corrosion and stress corrosion properties, there also have been attempts
involving direct examination of SCC failed specimens as well as in-situ
SCC experiments inside the Auger system, with some success. The analyses
in most practical cases are hampered by the massive corrosion products or
films that form on the surfaces that need to be analyzed.

Summary
Recent advances in the technique of Auger electron spectroscopy have
made it possible to evaluate the role of interface chemistry in failure of
materials. In many of the systems studied, impurity or solute segregation
occurring in a narrow region at grain boundaries has been found to deteri-
orate the mechanical properties at room and elevated temperatures as well
as the corrosion and stress corrosion properties. The segregants either de-
crease the fracture surface energy of the material or lower the plastic strain
energy associated with fracture or both, thereby weakening the interfaces.
The studies also have shown that solute-hydrogen interactions predominate
in some alloy systems thereby possibly accelerating the penetration of hydro-
gen along grain boundaries and fmally resulting in hydrogen assisted
cracking.

Acknowledgments

The author expresses his sincere appreciation to C. T. Hovland for ob-

taining several spectra and electron micrographs presented in the article
and to G. E. Riach and L. E. Davis for critically reviewing the manuscript.

References
[1] Joshi, A., Davis, L. E., and Paimberg, P. W. in Methods of Surface Analysis, Elsevier,
New York, 1975, p. 159.
[2] Weber, R. E. and Johnson, A. L., Journal of Applied Physics, Vol. 10, 1969, p. 314.
[3] Davis, L. E., MacDonald, N. C., Palmberg, P. W., Riach, G. E., and Weber, R. E.,
Handbook of Auger Electron Spectroscopy, 2nd ed., Physical Electronics Industries,
Eden Prairie, 1976.
[4] Davis, L. E. and Joshi, A. in Surface Analysis Techniquesfor Metallurgical Applications,
ASTM STP 596, American Societyfor Testing and Materials, 1976, p. 52.
[5] Joshi, A. and Davis, L. E., unpublished research.
[6] Haas, T. W., Grant, J. T., and Dooley, G. L in Adsorption-Desorption Phenomena,
Academic Press, New York, 1972, p. 359.
[7] Joshi, A. and Stein, D. F. in Temper Embrittlement of Alloy Steels, ASTM STP 499,
American Societyfor Testing and Materials, 1972, p, 59.
[8] Stein, D. F., Joshi, A., and LaForce, R. P., Metallurgical Transactions Quarterly, Vol.
62, 1969, p. 776.
[9] Marcus, H. L. and Palmberg, P. W., Transactions, American Institute of Mining,
Metallurgical and Petroleum Engineers, Vol. 245, 1969, p. 1164.

[10] Viswanathan, R., Metallurgical Transactions, Vol. 2, 1971, p. 809.

[11] Marcus, H. L., Hackett, L. H., and Palmberg, P. W. in Temper Embrittlement of Alloy
Steels, ASTM STP 499, American Society for Testing and Materials, 1972, p. 90.
[12] Ramasubramanian, P. V. and Stein, D. F., Metallurgical Transactions, Vol. 3, 1972,
p. 2939.
[13] Rellick, J. R., McMahon, C. J., Jr., Marcus, H. L., and Palmberg, P. W., Metallurgical
Transactions, Vol. 2, 1971, p. 1492.
[14] Joshi, A., Stein, D. F., and Palmberg, P. W., Metallurgical Transactions, Vol. 6A,
1975, p. 2160.
[15] Johnson, W. C. and Stein, D. F., Metallurgical Transactions, Vol. 5, 1974, p. 549.
[16] Joshi, A. and Stein, D. F., Journal of the Institute for Metals, Vol. 99, 1971, p. 178.
[17] Joshi, A. and Stein, D. F., Metallurgical Transactions, Vol. 1, 1970, p. 2543.
[18] Johnson, W. C., Doherty, J. E., Kear, B. H., and Gramei, A. F., Scripta Metallurgia,
Vol. 8, 1974, p. 971.
[19] MacDonald, N. C., Riach, G. E., and Gerlach, R. L., Research and Development, Vol.
27, 1976, p. 42.
[20] Latanision, R. M. and Opperhauser, H., Metallurgical Transactions, Vol. 5, 1974, p. 483.
[21] Schulz, B. J. and McMahon, C. J., Jr. in Temper Embrittlement of Alloy Steels, ASTM
STP 499, American Society for Testing and Materials, 1972, p. 104.
[22] Joshi, A. and Gurinsky, D. H., presented at the 6th Annual Spring Meeting of the
American Institute of Mining, Metallurgical, and Petroleum Engineers, Pittsburgh,
22 May 1974.
[23] Walsh, J. M. and Anderson, N. P. in Surface Analysis Techniques for Metallurgical
Applications, ASTMSTP 596, American Society for Testing and Materials, 1976, p. 58;
Superalloys: Metallurgy and Manufacture, Proceedings of the Third International
Symposium, 12-15 Sept. 1976, Seven Springs, Pa., Claitors, Baton Rouge, 1976.
[24] Joshi, A., Wiidermuth, J., and Stein, D. F., International Journal of Powder Metallurgy,
Vol. 11, 1975, p. 137.
[25] Joshi, A. and Stein, D. F., Corrosion, Vol. 28, 1972, p. 321.
[26] Joshi, A., presented at Corrosion/77 Conference, 14-18 March 1977, San Francisco,
Calif.
[27] Joshi, A., Shastry, C. R., and Levy, M., unpublished research.

Examination of Fracture in a Pressure

Vessel under Creep Conditions

REFERENCE: Coleman, M.C,, "Examination of Fracture in a Pressure Vessr under

Creep Conditions," Fractography in Failure Analysis, ASTM STP 645, B. M. Strauss
and W. H. Cullen, Jr., Eds., American Society for Testing and Materials, 1978,
pp. 297-311.

ABSTRACT: The relevance of creep crack growth data obtained from uniaxial labora-
tory experiments to the assessment of defects in plants is being investigated in a pres-
sure vessel research program involving full size half-percent chromium-molybdenum-
vanadium components containing machined defects. The first vessel in this program con-
tained external circumferential notches in a large pipe and was tested at 565 ~ and 62.5
MN/m 2 steam pressure, during which time crack growth and deformation in the vessel
were monitored. Final failure occurred in an explosive manner after 1583 h. This paper
concentrates on the fractographic and metaUographic aspects of the failure analysis.
Examination of the vessel revealed three fracture modes across the failed ligament.
Intergranular creep fracture occurred immediately ahead of the machined notch,
followed by 4S-deg ductile shear involving extensive deformation and finally low ductility
shear fracture causing fast failure. These observations are interpreted primarily in
terms of the mechanisms involved, with some consideration given to the mechanics of
the fracture. The intergranular creep fracture mode is considered typical of that associ-
ated with plant failures and the significance of this in defect assessment is mentioned.
The transfer to ductile shear is discussed in terms of the formation of void sheets, and
net section yielding is related to uniaxial tensile data. The change in mode to fast
shear fracture, attributed to gross overloading, produced an explosive failure.
The main implication of the failure analysis is that defects in this material can be
assessed using mechanics describing intergranular creep fracture. Even so, it is demon-
strated that in a nominally ductile situation final failure can be rapid and catastrophic.

KEY WORDS: fractography, metallography, fractures (materials), pressure vessels,

steels, elevated temperature, creep properties, voids, crack propagation, plastic deforma-
tion, shear properties

In recent years the increased application of nondestructive testing (NDT)

techniques to steam generating plants in the Central Electricity Generating
Board has revealed a high incidence of crack-like defects, particularly in
w e l d e d c o m p o n e n t s [1]. 2 T h e s e d e f e c t s a r e u s u a l l y i n t h e w e l d m e t a l a n d
1 Research officer, Central Electricity Generating Board, Research Department, Marchwood
Eu~gineering Laboratories, Marehwood, Southampton, Hampshire, SO4 4ZB, England.
ZThe italic numbers in brackets refer to the list of references appended to this paper.

297
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 1978 by ASTM International [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
298 FRACTOGRAPHY IN FAILURE ANALYSIS

heat affected zone (HAZ) and may arise from the welding process or, more
usually, during heat treatment or early in the service life of the vessel. In
plants operating at high temperatures, growth of these defects can occur by
creep, and it is important to be able to assess the significance of any defects
in terms of the integrity of the plant.
The assessment route generally used for chromium-molybdenum-vanadium
(CrMoV) materials is based on correlations between creep crack growth rate
and the linear elastic stress intensity factor [2-4]. However, since creep is a
time dependent process, the linear elastic approach should not be strictly
applicable and, indeed, a number of alternative models based on net section
stress [4], crack opening displacement [5], and reference stress [6] have
been suggested as being more appropriate to describe creep crack growth.
In all cases, the mechanics of the models and the mechanisms of cracking
to which they apply are based on observations made in relatively small
uniaxial laboratory tests. It is important therefore to examine their validity
in large components so they may be used with confidence to assess defects
in full size plants.
Accordingly, a program has been initiated to study crack growth in
pressure vessels under conditions directly relevant to steam generating
plants, an essential part of the work being the metallurgical aspects of the
mechanisms of failure. This paper considers the first experiment on a cir-
cumferentially notched pressure vessel fabricated in 1/2CrMoV material
and deals specifically with the fractographic and metallographie analyses of
crack growth and failure9 The features observed are described in detail,
and the mechanisms of fracture are related to certain aspects of the me-
chanics of crack growth, details of which are reported elsewhere [7].

Experimental Procedure and Results

Pressure Vessel Design and Testing

A relatively simple pressure vessel based on a main steam pipe containing
external circumferentially machined notches was selected for the experiment.
The pipe dimensions were 350-ram outside diameter, 60-mm wall; the defect
orientation corresponded to the circumferential mode often observed in
plant as weld HAZ cracking. It was necessary to give careful consideration
to the shell and notch design [7] since the hoop stress would be greater
than the axial stress and could lead to failure by an axial split rather than
by crack propagation. The final vessel and notch dimensions are shown in
9 r

Fig. 1; only the complete circumferential notches are considered here.

The vessel was constructed in 1/2CrMoV steel normalized at 970~ and
tempered for 4 h at 690~ The microstructure consisted of ferrite, with a
grain size of 70 #m, containing alloy carbides, about 5 percent bainite and
elongated silicate-type inclusions distributed in an axial direction along the

113O1
1325
A B C D E
F

225 150 l 1 7 5 ~ 5 ~ 7 5 L 250

Dimemions in men

Notch geometry
Notch detai]s
Circumferentlal 3.0
Identity Length Depth
A Fully 24,ram
circumferential
B " 28
C " 30
D " 27
E 250 mm + run out 30 ~ . ~ root radius
F 250 mm + run out 28

FIG. 1--Details of the circumferentially notched pressure vessel.

pipe, parallel to the original hot working direction. Testing took place in a
purpose-built pressure vessel testing facility [8]. The vessel was positioned
vertically in a bell furnace, in which the temperature was maintained at
565 +_ 5~ and was internally pressurized using steam at 62.5 MN/m 2.
This produced a mean diameter hoop stress of 150 MN/m 2 and a maximum
axial stress of 106 MN/m 2 on the ligament ahead of the 30 mm-deep notch.
At this notch, crack growth was monitored continuously using the potential
drop technique [9]. Activity in this region was also followed using an
acoustic emission coincidence counting technique. In addition, capacitance
strain gages were used to monitor displacement across the 28-mm-deep
notch. The NDT inspections were carried out at ambient temperature to
measure crack growth at the notches and, on these occasions, measure-
ments also were made on the shell and notches across specific locations to
determine the creep strains and displacements that occurred during each
test period.

Observations During Pressure Testing

At the first inspection, after 290 h, no crack growth was found at any of
the notches. The first indication of growth came from the 30-mm notch
after about 700 h when the potential drop and acoustic emission monitoring
techniques recorded increased voltages and stress wave emissions, respec-
tively. An inspection at 1192 h confirmed these indications and showed that

growth had started at all the notches. Crack growth data continued to be
generated during the test and was substantiated by further NDT inspections
at 1361 and 1577 h. The ultrasonic inspection also showed that crack growth
was asymmetrical at the 30-mm notch and, although for much of the test
there was less than 2-mm difference, at the final inspection the average
minimum and maximum crack depths were 35 and 38 ram, respectively.
This was also accompanied by an increase in the notch edge opening dis-
placement on the side of maximum crack depth, indicating that bulging
was taking place in the vessel as cracking extended from the root of the
machined notch, as shown in Fig. 2. Failure finally occurred in an explosive
manner after 1583 h with the vessel fracturing across the ligament ahead
of the 30-mm notch.

Failure Examination
After failure, an extensive metallurgical examination was made on the
fracture faces. A macroscopic examination was carried out to categorize
the different modes of fracture, and the areas of each were determined
using a quantitative television microscope (QTM). Several fractographic
and metallographic specimens were taken from the fracture faces and were
prepared for examination using conventional techniques. The microstruc-

FIG. 2--Bulging displacement and crack growth apparent at 30-mm-deep notch after
1361 h at 5 6 5 ~ and 62.5 M N / m 2 internal pressure.

tural and topographical features of each mode were studied then in detail
using optical and scanning electron microscopes (SEM).
In addition, specimens were taken from a number of positions around
each of the other three fully circumferential notches for comparison with
those from the fractured ligament. Other sections from the failed vessel
were used to determine the values of strain in the bore regions ahead of
each notch by using a surface roughness measuring instrument to determine
the spacing of the markings on the machined surfaces. The variation in
spacing of the machine markings even before testing was such that this
technique was only capable of detecting strains in excess of 3 percent.
As can be seen in Fig. 1, the vessel also contained two-part circum-
ferential notches. However, this portion was salvaged complete from the
failure and was fabricated into a separate vessel which is currently under
test.

Examination Results
The general appearance of the fracture surface is shown in Fig. 3 and
schematically, in Fig. 4. The numbers 1 to 8, which relate to the positions
where strain measurements were made at inspections during the test, were
used throughout the failure examination as reference positions.
Three distinct regions were apparent between the root of the 30-mm-deep
notch and the bore of the vessel.
Region/--This was a region of oxidized crack growth, coplanar with
the notch, extending in from the root, all around the circumference, and
varying in depth from about 7 mm at Position 2 to 12 mm at Position 6 as
shown in Fig. 4.
Region//--The fracture between Position 4 increasing to 8 inclined at a
45-deg angle and had a rough oxidized surface to within a few millimetres
of the bore in most places. Between Positions 5 and 6 where bulging was a
maximum, as indicated in Fig. 4, this type of fracture was continuous to
the bore.
Region IIl--Over the remainder of the failed ligament, the fracture also
followed a 45-deg angle but was smooth and less darkened by oxidation.
This type of fracture was particularly noticeable between Positions 8 through
1 to 4, where the surface also contained striation type markings as can be
seen between Positions 8 and 1 in Fig. 3.
From the distribution of the three fracture regions in Fig. 4, it can be
seen that there is a rough line of symmetry about the diameter through
Positions 2 to 6. From this representation, the proportion of each region
was determined using a QTM. Region I, oxidized crack growth extending
from the notch, represented 39 percent of the failed ligament, Region II,
the rough 45-deg fracture, 17 percent, with Region III, the smooth 45-deg
fracture, accounting for the remaining 44 percent. The symmetry of fracture

FIG. 3--General appearance of the failed ligament surface, x 1/4.

also was reflected in the symmetry of deformation observed at the bore of

the vessel. Between Positions 4 and 8, the bore had a necked-down ap-
pearance, this effect being particularly marked between Positions 5 and 6,
the region of maximum bulging. In contrast, immediately opposite this
area and adjacent to Region III, bore deformation was not apparent.

Fractography a n d Metallography
Fractographic examination of Region I confirmed that where bulging
occurred, between Positions 4 and 8, the surface was covered completely
with oxide up to where Region II started and showed that on the opposite
side of the vessel the last miUimetre or so before Region III was clear of
oxide. In this area, grain boundary cavitation and intergranular cracking
typical of creep fracture were observed, Fig. 5. Metallographic sections re-
vealed cavities on grain boundaries linking to give short cracks and side
branching of the main crack, as shown in Fig. 6. Between Positions 4 and 8,
the creep crack reached a maximum depth of 12.5 mm and creep damage
was confined to a strip 5 mm to either side of the main crack. Between
Positions 8 through 1 to 4, the creep fracture extended generally to 7 mm,
reaching 10 mm in one region, and the damaged material was confined to
within 3 mm of the main crack.

mum

Axis of
syrnmetr
/

F I G . 4--Schematic representation of the three regions o f fracture across the failed ligament
ahead of the 30-mm-deep notch. • 1/4.

Regions II and III, both 4S-deg fractures, exhibited considerable contrast

in their fractographic and metallographic features. Region II, on the side
of the vessel where bulging occurred, was a ductile shear fracture showing
clear evidence of deformation with deep elongated dimples throughout
the fracture surface. This is shown in Fig. 7 and is very similar to the ap-
pearance of a highly ductile high temperature tension test fracture surface.
The microstructure in this area was heavily deformed, parallel to the
longitudinal axis of the vessel, and it was apparent that crack propagation
occurred in a ductile manner by necking down and linking between the
many elongated voids, as shown in Fig. 8. At about Position 5, the ductile
shear fracture extended fully to the bore and deformation was still apparent
in the bore of the vessel some 30 mm below the fracture tip.
Region III, on the side of the vessel between Positions 8 through 1 to 4
where bulging was not apparent, was a remarkably featureless low ductility
shear fracture, as shown in Fig. 9. The fracture surface in general was very
flat and contained smooth areas free of detail as well as many shallow
dimples that in most cases were less than 10 #m in diameter. The striations
on the fracture surface appeared as simple steps, and no fractographic
features were observed to account for this phenomenon. The metallographic

FIG. 5--SEM showing the intergranular creep fracture region immediately ahead of the
30-mm-deep notch, x300.

FIG. 6--Optical micrograph showing the grain boundary cavitation and cracking associated
with the intergranular creep fracture, x200.

FIG. 7--SEM showing the 45-deg ductile shear fracture surface, x300.

FIG. 8--Optical micrograph showing the heavily deformed microstructure and voids associ-
ated with the 45-deg ductile shear fracture. >(200.

FIG. 9--SEM showing the smooth 45-deg low ductility shear fracture su(face, x300.

examination revealed a limited amount of deformation immediately adjacent

to the fracture surface and to a much lesser extent in a layer about 300 #m
thick, where deformation appeared as slip bands across the ferrite grains,
as shown in Fig. 10. Apart from this, the remainder of the material in the
ligament appeared as in the as-received condition.
Metallographic examination of sections through the three circumferential
notches that did not fracture completely revealed intergranular creep
fracture similar to that shown in Fig. 6. The extent of the cracking in-
creased from 2 mm for the 24-mm-deep notch to about 3.5 mm for the
28-mm-deep notch, and in all cases grain boundary cavitation and crack-
ing was confined to within 3 mm of the crack path and tip. There was no
evidence of any other mode of fracture and, outside the creep cracking
regions, the parent material appeared to be unchanged from the as-received
condition.
Finally, surface roughness measurements detected no strain in the bore
of the vessel ahead of the three unfractured circumferential notches. A
similar result also was obtained from measurements made up to the edge
of the low ductility shear fracture, at about Position 2. In contrast however,
on the side of the ductile shear fracture, at Position 5, the strain increased
from 0 to 4 to 8 percent at distances of 75, 50, and 25 mm from the fracture
tip, reaching 28 percent strain at the fracture edge.

FIG. lO--Opt&al micrograph showing the limited deformation closely associated with
the 45-deg low ductility shear fracture, x200.

Discussion
The fractographic and metallographic analyses of the failure revealed
that Region I consisted of grain boundary cavitation and intergranular
cracking. This was also the only mode observed ahead of the three notches
that did not fail. This is typical of intergranular creep fracture and indicates
that the pressure vessel experiment has produced the features generally
associated with high temperature plant failures, albeit in a normalized
and tempered material.
The extent of the damage associated with the creep fracture can be used
to assess the stress conditions controlling the failure process. In an elastic
situation the plastic zone radius at the root of a notch can be determined
from the models of Irwin [10] and Dugdale [11]. For the 30-mm notch in
the pressure vessel, these models predict a radius of approximately 2 mm,
the exact value depending on the yield stress. Hence there will be a stress
distribution ahead of the notch decaying from the yield stress, about 230
MN/m 2, at the edge of the plastic zone to the net section stress in the liga-
ment, 106 MN/m 2. Consequently, creep damage would be expected to
occur over at least a radius of 2 mm even if the controlling parameter was
the crack tip stress alone. In the present work, the radius of creep damage

ranges from 5 mm at the 30-mm notch to about 3 mm at the other notches.

This indicates that creep damage has occurred even outside the notch root
plastic zone, as would be expected from the level of the stress distribution,
but nevertheless is closely confined to this notch root region. Furthermore,
it confirms that even though the pressure vessel was constructed in a ductile
material crack growth and failure has not occurred simply by plastic
collapse of the ligament.
The creep fracture extended to a depth between 37 and 42.5 mm from
measurements at failure, while the NDT assessment at the inspection prior
to failure indicated depths from 35 to 38 ram. Bearing in mind the accuracy
of ultrasonic measurements, +_ 2 ram, and the fact that some creep crack
growth most probably occurred in the short test period immediately prior
to failure these results are in good agreement. Consequently, the data gen-
erated throughout the life of the vessel, which have been reported separately
to show a relationship between crack growth rate and reference stress [7],
relate specifically to crack propagation occurring by intergranular creep.
The mode of fracture in Region II was ductile shear. In the reviews of
the mechanisms of ductile or plastic fracture, based mainly on room tem-
perature laboratory tests, Rosenfield [12] and Sullivan [13] show that the
dominant factors are the formation of microvoids and the concentration
of shear in sharply defined bands. In general, a uniaxial test gives a region
of microvoid coalescence in the center of the specimen while a void-sheet
mechanism [14,15] results in shear bands of deformation forming at the
crack tip. Growth occurs initially in a direction macroscopically normal to
the tensile axis, transferring to 45-deg shear as a free surface is approached.
In a similar manner, it appears that in the present failure the latter stages
of creep crack growth, although remaining normal to the axial stress in the
vessel, occurred while shear deformation was taking place in the ligament
ahead of the crack tip. This would account for the gross bulging observed
on the vessel between Positions 4 and 8 towards the end of testing when
intergranular creep cracking was still occurring. In addition, the heavy
oxide layer observed at the limit of the creep fracture between Positions
4 and 8 may be attributed to easy ingress of the atmosphere resulting from
shear deformation in the ligament displacing the cracked surfaces. Finally,
the creep mode of fracture transferred to fully ductile shear. It is known
that shear separation can be rapid [16], which is consistent with the earlier
deduction that shear occurred late in the vessel life.
In terms of the mechanics involved, a model proposed by McClintock [17],
based on uniaxial data, indicated that the transfer from homogenous flow,
analogous to the creep fracture mode, to localized shear occurs simply
when the loads for each mechanism are equal. The model however related
largely to idealized situations where the size and distribution of microvoids
are readily determined. In the present work, the size and distribution of
microvoids is not regular but, nevertheless, all combinations have been

analyzed. The McClintock model predicts that even where creep fracture
is observed, the load for shear fracture is always less than that for homo-
genous flow, and in many cases is negative. Clearly, therefore, it has not
been possible to account for the transfer from creep to ductile shear fracture
on the basis of this model.
The mechanics of the final stage of ductile shear, however, can be
discussed quantitatively from the results of the present work. The QTM
measurements show that at the end of ductile shear only 44 percent of the
original ligament was remaining. At an internal pressure of 62.5 MN/m 2
and assuming uniform loading, the stress on the remaining ligament would
be about 245 MN/m 2 which is some 5 percent greater than the mean 0.2 per-
cent proof stress for the material [18]. In addition, the mean strain for
failure in these materials is 30 percent at 565~ and the maximum bore
strain measurement of 28 percent clearly indicates that this region has
necked down fully in a tensile fashion to fracture in a ductile shear mode.
As the experiment progressed, it was anticipated that a leak situation
would finally arise due to the asymmetry of crack growth, but in fact a
rapid explosive failure occurred by shear involving little deformation. How-
ever, it is apparent that the remaining ligament was stressed above the
proof stress and the material would be in a condition for plastic collapse.
In addition, if a leak occurred, and little evidence was found for this, the
leak rate did not depressurize the vessel sufficiently to reduce the ligament
stress below that for plastic collapse before the low ductility shear fracture
had fully propagated. From previous work [16] it is clear that coalescence of
microvoids in the void sheet can be catastrophic. Thus, the difference in
the extent of deformation associated with the ductile shear in Region II
and the low ductility shear in Region III is almost certainly a reflection of
the high plastic strain rate in the latter which led to high velocity crack
propagation around the remaining ligament and explosive failure.
In summary, failure of the pressure vessel occurred across the ligament
ahead of the 30-mm-deep notch and involved three fracture mechanisms in
sequence. Initially creep fracture occurred ahead of the notch tip with
damage closely confined to the crack plane. This extended asymmetrically
and, on the side where bulging was most apparent, reached a depth of
42.5 mm before transferring to ductile shear fracture. This mechanism
produced gross deformation and ductile crack extension through to the
bore on one side of the vessel. At this instant, the remaining ligament was
subjected to a stress of yield magnitude, and the final failure occurred by
rapid crack growth involving a low ductility shear fracture running around
the ligament. It is clear from the striation markings on Region III and the
axis of symmetry across the failure surface that the crack ran out in both
directions from the position where ductile shear reached the bore and that
the two accelerating crack fronts met at the opposite side of the vessel
when total separation resulted.

Concluding Remarks
The main implication of the failure analysis is that defects in normalized
and tempered 1/2CrMoV material now can be assessed using mechanics
known to describe the intergranular creep fracture mechanism. In addition,
it is clear that even in ductile materials operating under nominally ductile
conditions, rapid and catastrophic failure can occur. However, in operating
plants, while intergranular creep fracture is the expected mode of crack
propagation, neither the shear modes nor catastrophic failure are likely
to occur. They have only arisen in this work because an extremely large
defect of axisymmetric geometry was allowed to remain in a vessel sub-
jected to nominally tensile loading.
Future work in this area is intended to examine the more usual plant
problems of defect growth in weld metals and HAZs. In this respect, the
present work is considered to provide a sound basis for future pressure
vessel failure analyses.

Acknowledgments
The author would like to thank his colleagues for their interest and
valuable discussion of this work. The work was carded out at Marchwood
Engineering Laboratories of the Central Electricity Generating Board and
is published by permission of the Director.

References
[1] Toft, L. H. and Yeldham, D. E., "Welding Research Related to Power Plant," Pro-
ceedings of an International Conference, Central Electricity Generating Board, London
1972, pp. 5-19.
[2] Siverns, M. J. and Price, A. T., Nature, Vol. 228, No. 5273, 1970, pp. 760-761.
[3] Siverns, M. J. and Price, A. T., International Journal of Fracture Mechanics, Vol. 9,
No. 2, 1973, pp. 199-207.
[4] Neate, G. J. and Siverns, M. J., "Creep and Fatigue in Elevated Temperature Appli-
cations," Proceedings of an Inteimational Conference, Paper No. C234, Institution of
Mechanical Engineers, London, 1973.
[5] I-Iaigh, J. R., PhD thesis, Council for National Academic Awards, London, 1973.
[6] Williams, ]. A. and Price, A.T., Journal of Engineering Materials and Technology,
1975, pp. 214-222.
[7] Coleman, M. C., Price, A. T., and Williams, J. A., Fracture, Vol. 2, 1977, pp. 649-662.
[8] Eaton, N. F. and Rowley, T., "Experimental Evaluation of Creep Behavior of Welded
Vessels," International Institute of Welding Colloquium, Toronto, 1972.
[9] Coleman, M. C., Fidler, R., and Williams, J. A. in Detection and Measurement of
Cracks, The Welding Institute, Cambridge, 1976, pp. 40-44.
[10] Irwin, G. R., Metals Engineering Quarterly. Vol. 3, 1963, p. 24.
[11] Dugdale, D. S., Journal of the Mechanics and Physics of Solids, Vol. 8, 1960, p. 100.
[12] Rosenfield, A. R., MetallurgicalReviews, Vol. 13, No. 121, 1968, pp. 29-48.
[13] Sullivan, C. P., "A Review of Some Microstructural Aspects of Fracture in Crystalline
Materials," Bulletin No. 122, Welding Research Council, New York, 1967.
[14] Rogers, H. C., Transactions, Metallurgical Society of the American Institute of Mining,
Metallurgical, and Petroleum Engineers, 1960, Vol. 218, pp. 498-506.

[15] Cox, T. B. and Low, J. R., Metallurgical Transactions, Vol. 5, 1974, pp. 1457-1470.
[16] Bluhm, J. I. and Morrissey, R. J., Fracture, Proceedings of the First International
Conference, The Japanese Society for Strength and Fracture of Materials, Vol. 2, 1966,
pp. 1739-1780.
[17] McClintoek, F. A. in Ductility, American Society for Metals, Metals Park, Ohio,
pp. 255-277.
[18] Johnson, R. F., May, M. J., Truman, R. J., and Mickleraith, J., "High-Temperature
Properties of Steels," Proceedings of a Conference, ISI Publication No. 97, The Iron
and Steel Institute, 1967, pp. 229-263.

Strength, Toughness, and Flaw

Tolerance of 25.4-mm (1-in.) Alloy Steel
Lifting Chain*

REFERENCE: McCartney, R. F. and Pellegrino, J. V., "Strength, Toughness, and

Flaw Tolerance of 25.4-mm (1-in.) Alloy Ster Lifting CImln," Fractography in Failure
Analysis, ASTM STP 645, B. M. Strauss and W. H. Cullen, Jr., Eds., American
Society for Testing and Materials, 1978, pp. 312-334.

ABSTRACT: The interrelationship of strength, toughness, and flaw tolerance of

2.5.4-mm (1-in.) alloy steel lifting chain procured from five chain manufacturers has
been investigated. Both International Standards Organization (ISO) Grade 63 and
Grade 80 chains were tested. The susceptibility of these chains to sudden fracture in
service in the presence of flaws is of interest to producers and users, and it was antici-
pated that the results would be helpful in arriving at a recommended upper limit of
strength consistent with safety, inspection, and economic considerations.
Small round-bar tension-test and Charpy impact-test specimens were machined
from each chain to characterize the materials. Fracture-mechanics calculations of
critical flaw sizes then were conducted by using a Charpy-Kk correlation. The results
showed that the Grade 80 chain, having a Rockwell C hardness (HRC) of 40, is
susceptible to brittle fracture in the presence of small flaws. Tension tests on specimens
from chain samples containing machined flaws 2.54 mm (0.1 in.) deep confirmed the
calculations. Grade 80 chain containing the machined flaw fractured at less than half
the load required to break unnotched chain; however, Grade 63 chain, having a
HRC of 33, was virtually unaffected by the presence of a 2.54-mm (0.1-in.) deep flaw.

KEY WORDS: steel chain, mechanical properties, fractures (materials), toughness,

notch sensitivity, fractography

T h e p r e s e n t i n v e s t i g a t i o n was c o n d u c t e d to i n d i c a t e t h e i n t e r r e l a t i o n s h i p s
o f s t r e n g t h , t o u g h n e s s , a n d flaw t o l e r a n c e o f 2 5 . 4 - m m (1-in.) alloy-steel
l i f t i n g c h a i n s p r o c u r e d f r o m five c h a i n m a n u f a c t u r e r s . T h e s u s c e p t i b i l i t y o f
these c h a i n s to s u d d e n f r a c t u r e in service i n t h e p r e s e n c e o f flaws is o f
i n t e r e s t to p r o d u c e r s a n d users, a n d it was a n t i c i p a t e d t h a t t h e results
w o u l d b e g e n e r a l l y h e l p f u l in a r r i v i n g at a r e c o m m e n d e d u p p e r l i m i t o f
s t r e n g t h c o n s i s t e n t with safety, i n s p e c t i o n , a n d e c o n o m i c c o n s i d e r a t i o n s .
* original experimental data were measured in U.S. customary units.
I Associate research consultant and project analyst, respectively, United States Steel
Corporation, Research Laboratory, Monroeville, Pa. 15146.

312
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 by ASTM International [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
MCCARTNEY AND PELLEGRINO ON STEEL LIFTING CHAIN 313

The chain manufacturing process consists of forming the link, electrical-

resistance welding, normalizing, quenching (in water), and tempering.
Usually only one weld is present on a side of the link; however, at least one
manufacturer welds both sides of the link in their process. Tempering at
about 480~ (900~ results in the strength level required for International
Standards Organization (ISO) Grade 63 chain, and tempering at about
345~ (650~ results in the higher strength of ISO Grade 80 chain. A
material tensile strength of 1035 MPa (150 ksi) and hardness level of 33
according to Rockwell hardness, C scale (HRC), are characteristic of
Grade 63, and tensile strength in the range 1240 to 1380 MPa and (180 to
200 ksi) and hardness of 40 HRC are typical for Grade 80 chain. The
present paper includes a characterization of chain materials, a description
of full-size chain tests, the application of fracture mechanics to calculate
flaw tolerance at various nominal stress levels, and a description of notched-
chain tests on Grade 63 and Grade 80 chain.

Materials and Experimental Work

The chemical compositions of the various chain samples included in this
investigation are shown in Table 1. Tension and standard-size Charpy V-
notch impact tests were conducted on specimens machined from chain
links as shown in Fig. 1. On chain links with two welds, base metal proper-
ties were determined on specimens where the gage section was offset from
the weld. Hardness profiles across the diameter of the links were determined
at 1.6-ram (1/16-in.) intervals, and full-size cross sections of the links were
examined metallographically. In addition, full-size chain tension tests
(5 links) were conducted on Grade 80 and Grade 63 chains. The chain
coded 401 was not included in the full-size tests because of insufficient
material, and the specimen coded 420 could not be tested because the
19-ram (3/4-in.) chain furnished did not fit the testing fixtures.
The susceptibility of these chains to sudden fracture in service in the
presence of flaws was determined by employing an empirically developed
correlation of Charpy test results and K k values 2 and then using the Ktc
values to calculate critical crack sizes for given stress levels through fracture-
mechanics equations. Full-size tests were conducted on additional Grade 63
and Grade 80 chains to determine the effects of notches machined into
chain links on the breaking loads.

Results and Discussion

Characterization of Chain Materials

It is apparent from the mechanical compositions described in Table 1
2Barsom, J. M. and Rolfe, S. T, in Impact Testing of Metals, ASTM STP 466, American
Society for Testing and Materials, 1970, p. 281,

"n
>
0
0
TABLE 1--Chemical composition o f chains investigated, percent.
>
-o
Probable AISI Specimen I
-<
Steel Grade Code C Mn P S Si Cu Ni Cr Mo V Ti A1 N B

8620 400 0.22 0.86 0.013 0,024 0.35 0.21 0.46 0.60 0.21 0.010 0.002 0.028 0.006 0.0001 "11
8630 401 0.31 0.82 0.005 0,031 0.25 0.18 0.50 0.57 0.16 0,008 0,003 0.016 0.006 0.0001 _>
8630 402 0.30 0.77 0.010 0.030 0.27 0.27 0.48 0.57 0.16 0.008 0,003 0.048 0.008 0.0001 r-
E
86B20 403 0.18 0.81 0,006 0.020 0.14 0,13 0.48 0.54 0.15 0.009 0.026 0.036 0.008 0.0008 a
420 0.18 0.90 0,039 0.034 0.25 0,03 0.39 0.46 0.10 0.009 0.025 0.072 0,006 0.0008 a m
94817
8620 444 0,23 1.07 0,018 0.035 0.28 0.21 0.43 0.58 0.20 0,007 0.001 0,035 0,006 0.0001
z
8620 450 0.22 1.07 0.018 0.029 0.24 0.25 0.42 0.65 0.20 0,008 0.002 0.030 0.006 0.0001 >
f-
8620 166 0.22 1.06 0.020 0,033 0.27 0.22 0.41 0.58 0.20 0.009 0.002 0,048 0.006 0.0001
8620 167 0.23 1.08 0.018 0.035 0.27 0.23 0.42 0.58 0.20 0,008 0.002 0,037 0.006 0.0001 oo
8620 0.18 0.70 0.040 0.040 0.20 0.40 0.40 0.15
Specification 0.23 0.90 max max 0.35 0.70 0.60 0.20
8630 0.28 0.70 0.040 0.040 0.20 0.40 0.40 0.15
Specification 0.33 0.90 max max 0.35 0.70 0.60 0.20
94817 0.15 0.75 0.040 0.040 0.20 0.30 0.30 0.08
Specification 0.20 0.100 max max 0.35 0.60 0.50 0.15

aBoron steels, minimum O.O00S percent boron content.

CENTER-
LINE

TENSION-TESTSPECIMENS CHARPYV-NOTCHIMPACT-TEST
SPECIMENS
FIG. 1--I~yout showing location of specimens.

that various grades of steel are used by the different manufacturers. AISI
8620, 86B20, 8630, and 94B17 were among those used.
The results of round-bar tension tests on specimens machined from the
chain links are shown in Table 2. Specimens across the welds exhibited
essentially the same yield strength and tensile strength as the base metal,
but the elongation and reduction of area across the welds were lower, as
expected. Similarly, the Charpy V-notch impact-test energies were sub-
stantially lower in the weld than in the remainder of the link, Table 3 and
Fig. 2 . In general, the notch toughness of the welds and of the base metal
was substantially higher for the Grade 63 chain than for the Grade 80 chain.
The behavior is consistent with the expected effects of the tempering treat-
ments.
The hardness surveys across the links are summarized in Table 4. Maxi-
m u m , minimum, and average values are shown, but hardness measure-
ments were made every 1.6 m m (1/16 in.) across the link diameter. The
hardness survey and an accompanying metallographic examination of the
full cross section indicate that the chains had been properly heat treated
before the other mechanical tests were conducted on the chains.

Full-Size Chain Tests

An elastic stress analysis 3 of an open chain link indicates that the maxi-
m u m tensile fiber stress occurs on the outer surface at the ends of the link
3Goodenough, G. A. and Moore, L. E., "Strength of Chain Links," Bulletin 18, Engi-
neering Experimental Station, Universityof Illinois, 1907.

-i1

0
--.I
0
TABLE 2--Results of tension tests a on specimens machined f r o m chain links.

Chain Size Yield Strength Tensile Elongation Reduction "1-

-<
Specimen (Diameter, in.) Specimen (0.2% Offset), Strength, in 1 in., of Area,
Code and Grade Location ksi % % %
-11
400 1, 63 base metal 147.0 157.2 18.0 65.1 >_
weld 145.9 154.9 10.0 39.7 f-
c
401 1, 80 base metal 191.4 210.2 13.5 56.2 .1-1
weld 188.6 210.8 10.5 44.9 m

402 1, 80 base metal 189.8 205.3 14.0 53.7

z
weld 184.2 200.3 3.5 13.0
403 I, 80 base metal 164.0 178.6 14.0 63.4 r-

weld 164.0 175.8 4.0 17.6

420 3A b base metal 141.6 162.9 16.0 63.7
weld 139.6 I58.6 14.0 22.6
450 1, 80 base metal 161.2 180.6 14.0 61.0
weld 149.7 164.2 8.0 43.0

Conversion Factors--
1 in. = 25.4 ram, and
1 ksi = 6.895 MPa.
a 0.252-in.-diameter specimen.
b Not tension tested as chain, therefore grade unknown.

Chain Size 0 ~ Energy 0~ 72 ~ Energy --I

72 ~ Z
Specimen (Diameter, in.) Specimen Absorbed, Shear, Absorbed, Shear, m
-<
Code and Grade Location f t . lb % ft. lb %
z
400 1, 63 base metal 78 to 78 100 to 100 80 to 88 100 to 100 o
weld 19 to 54 20 to 100 57 to 80 100 to 100 -v
401 1, 80 base metal 16 to 17 15 to 15 18 to 21 35 to 40 m
t-
weld 7 to 7 10 to 10 9 to 14 [Link] 35 r-
,m
402 1, 80 base metal 1S to 15 15 to lS 17 to 17 30 to 30
weld 5 to 5 10 to 10 5 to 7 5 to S :~
403 1, 80 base metal 12 to 19 15 to 15 43 to 45 60 to 60 Z
0
weld 11 to 12 15 to 15 15 to 19 35 to 40
420 1Aa base metal 20 to 21 1S to 15 26 to 37 35 to 50 0
Z
weld 15 to 21 15 to 20 19 to 26 60 to 70
450 1, 80 base metal 15 to 15 S to S 20 to 22 20 to 20 o)
--i
weld 8 to 12 S to S 11 to 1S 20 to 20 m
m
i---
Conversion Factors-- r
w
1 in. = 25.4 mm, "11

1 f t . l b = 1.36 I, and
~ = S / 9 ( ~ - 32). Q
a Not tension tested as chain; therefore grade unknown. 0
-r
>
Z

GO
-dw
",4

I I i I I I I
=~~oI-. o..O ~ - - ~ ~
6oI ,,o i O BASE METAL

~:1~176
oDO r , J, , ", I / ," , , ' 1 ~
.~.icI 9 WELD
I ft ib = 1 36J
~C ~ 5/9t~ " 321
<~ 4 0 F . ~ . / o BASEMETAL ~ 4 t inch = 254 mm
~^~'~ ~~ o 9 WELD >"

'I
~. 01~60 I l I 1 I ! I I I I | I i I
LU --80 0 80 160 m 0 80 160 240
TEMPERATURE, ~ TEMPERATURE, ~
A, CODE400 B, CODE450
! I I I I I
[Link] o BASE METAL -J 8 O BASE METAL

Ji
9 WELD

== j O _ _ O ~ O "~'o
o 9
20]- = O--.-'-'~ 11-1(3 21 oo_-;/ .:..-~. 1-,""
-~BO l .~. . _ . -I - . ~ .I
I 9I I I Sl ,,,
z I I I I I I I i
0 80 160 240 m 80 0 80 160 240
TEMPERATURE, ~ TEMPERATURE, ~
C, CODE401 D, CODE403
~1001. l ! l l l I I I /|~/ "-" 1 00 ] I I I l I I | 1 7
OBA E AL o A E E'AL

4o~ J ~ ,0L
~ .___o_o___~j
[

o-o -1 o ZUl-o.O~=l-.~..---~ 9~ ' ~ --i

ol i ; ,
',--~--T--:'T'J~, oI .if,', , , , ~ i'l
m -80 0 80 160 240 m -80 0 80 160 240
TEMPERATURE, ~ TEMPERATURE, ~
E, CODE402 F, CODE420

FIG. 2--Energy absorbed in Charpy tests of chains investigated.

TABLE 4--Hardness determinations on cross sections of chain links.

HRC
Chain Size
Specimen (Diameter, in.) Weld
Code and Grade Maximum Minimum Average Centerline =

400 1, 63 34.3 32.5 33.5 28.1

401 1, 80 43.5 40.5 42.5 45.3
402 1, 80 42.5 40.2 41.9 43.9
403 1, 80 40.3 38.5 39.4 33.7
420 ~Ab 41.5 40.0 40.7 31.8
450 1, 80 41.0 34.8 37.9 32.4
166A 1, 80 42.3 35.4 39.4 ...
166B 1, 80 42.5 37.0 39.9 ...
167A 1, 63 34.5 31.5 34.0 ...
167B 1, 63 34.8 30.2 34.0 ...

Conversion Factor--
1 in. = 25.4 ram.
a Hardness converted from diamond pyramid hardness (DPH).
b Not tested as chain, therefore grade unknown.

at Position A in Fig. 1, where the stress equals 2 P/A (where P is the load,
and A is the cross-sectional area of the link). The locations of next highest
tensile stress are Positions B, where the stress equals 1.8 P/A. The weld is
located at Position B also, and sudden fracture of chain in service is re-
portedly usually observed at this location.
In the full-size chain tension tests, the load-displacement curves departed
from linearity when the stress at Position A reached the yield strength of
the material. These loads are listed in Tables 5 and 6 under the column
heading, Yield Load. For example, if the yield load of 396 kN (89 kips) for
the specimens coded 400 is substituted into the expression for the stress at
Position A, 2 P/A, the tensile stress at Position A is calculated to be 1015
MPa (147.4 ksi), in good agreement with the yield strength 1013 MPa
(147.0 ksi) Table 2, measured on the round-bar tension-test specimen.
The minimum breaking load specified for 25.4-mm (1-in.) chain of
Grade 63 is 639 kN (143.6 kips), and the minimum proof test load is
365 kN (82 kips). Specified breaking and proof-test loads for 25.4-mm
(1-in.) chain of Grade 80 are 812 and 464 kN (182.4 and 104.2 kips), re-

TABLE S--Results of full-size chain tension tests.

Chain Size Yield Maximum

Specimen (Diameter, in.) Load, Load, Elongation
Code and Grade kips kips %

400 1, 63 89 175.0 18.3

401 1, 80
402 1, 80 137 2~t:0 li.2
403 1, 80 105 186.0 18.0
420 90 '*
166A 1, 80 167 16919 "8.()
444 1, 80 107 213.5 16.1
167A 1, 63 86 160.5 15.8

a Not tested as chain, therefore grade unknown.

b Broke in weld.

TABLE 6--Results of full-size notched-chain tension tests.

Chain Size Yield Maximum

Specimen (Diameter, in.) Load, Load, Elongation
Code and Grade kips kips %

166B a 1, 80 ... 91.0 . ..

167B a 1, 63 77 157.0 ...

Conversion Factors--
1 in. = 25.4 mm, and
1 kip = 4.45 kN.
aNotch 0.006 in. wide by 0.100 in. deep machined in center link of test length.

spectively. The working-load limit for both chains is 50 percent of the proof
load, resulting in a safety factor of 3.5.
Two unnotched specimens of Grade 63 chain, coded 400 and 167A,
broke at loads of 779 and 714 kN (175 and 160.5 kips), respectively (Table 5),
well above the specified minimum of 639 kN (143.6 kips). Both fractures
were ductile and occurred when the links necked and then sheared, Figs. 3
and 4.
Three unnotched specimens of Grade 80 chain, coded 402, 403, and 444,
broke at loads of 908, 828, and 948 kN (204, 186, and 213 kips), respectively,
all in excess of the specified minimum of 812 kN (182.4 kips). All three
fractures were ductile shear, Figs. 5, 6, and 7. A fourth unnotched Grade 80
specimen, coded 166A, broke in the weld at a load of 752 kN (169 kips),
below the specified minimum, Fig. 8. This weld fracture originated at a
subsurface weld imperfection, Area B, Fig. 9(a), and then propagated in a
brittle manner, Area C, Fig. 9(a). The inclusions present in the fracture
origin in Fig. 9(b) were identified using energy dispersive X-ray analysis as
manganese silicate and manganese sulfide. The fracture mode in Area B
was primarily dimpled rupture. Figure 9(c) shows that the fracture propa-
gated primarily by a cleavage fracture mode. During the electrical-resistance
welding of the chain link, the two surfaces being welded together are upset-
forged. The resulting plastic flow tends to orient the fibrous inclusions
more parallel to the weld surface. As a result, the mechanical properties
across the weld would be similar to through thickness rather than longi-
tudinal properties of the bar from which the chain links were made.

Notch Sensitivity o f Chains 4

In order to appraise the susceptibility of chains to sudden fracture in the

presence of flaws, fracture-mechanics calculations were made to relate stress,
toughness, and flaw size, Table 7. The flaw size required to cause cata-
strophic fracture at a given stress level is called the critical flaw size. Nomi-
nal stress levels of 690, 1035, and 1240 MPa (100, 150, and 180 ksi) were
assumed to approximate, respectively, (a) a working stress frequently
reached, (b) the yield strength of Grade 63, and (c) the yield strength of
Grade 80 chain. Two levels of toughness were assumed: (a) a Charpy V-
notch impact test energy of 5 J (4 [Link]) which corresponds to about the
lowest toughness measured in any of the chain tested; this toughness corre-
lates empirically with a critical stress intensity (Kk) of 27.5 MPa/~J-m (25
ksii~/]~-.), and (b) a Charpy energy level of 22 J (16 ft 9lb) which corresponds
to the average toughness measured in the Grade 80 chain tested at - 17.8~
(0~ this toughness correlates with a Kic of 55 MPa/x/m (50 ksi i~]~-.). A
4For the reader wishingmorebackgroundin fracture mechanics,refer to Rolfe, S. T. and
Barsom, J. M., Fracture and Fatigue Control in Structures: Application of Fracture Me-
chanics, Prentice-HallInc., 1976.

Z
m

>
z

m
r-
r"
m

z
0
0
z
m
m
m
f-
r-
"11

0
"1"

FIG. 3--Photographs of specimen Code 400 25.4-mm (I-in,) Grade 63 chain. Fracture occurred through base metal with evidence
of prefracture deformation. (a) Five-link full-size chain test, • 0.3. (b) Fractured link in full-size chain test, x O.75.
r
..A

"11
:lJ

0
0
:D
>
"12
ar
-<

r-
c
m

z
I'-
.<

FIG. 4--Photographs o f specimen Code 167.4 25.4-mm (I-in.) Grade 63 chain. Fracture occurred through base metal with evi-
dence o f prefracture deformation. (a) Five-link full-size chain test, • 0.3. (b) Fractured link in full-size chain test, x 0. 75.

z
m
.<

m
F-
r-
m
0

z
0
0
z

m
m
r"
F-

FIG. $--Photographs of specimen Code 402 25. 4-ram (1-in.) Grade 80 chain. Fracture occurred through base metal with evidence
of prefracture deformation. (a) Five-link full-size chain test, x 0.3. (b) Fractured link in full-size chain test, x O.75.

"o
I.<

-11
r"
e--
m

r-
.<

FIG. 6--Photographs o f specimen Code 403 25.4-mm (1-in,) Grade 80 chain. Fracture occurred through base metal with evidence o f
prefracture deformation. (a) Five-link full-size chain test, x 0.3. (b) Fractured link in full-size chain test, x O. 75.

z
m
.<

z
o
"ID
m
I-
F-
ro

3]
z
0
0
z
O~
m
m
r-
r-
-11

z
FIG. 7--Photographs of specimen Code 444 25.4-mm (1-in.) Grade 80 chain. Fracture occurred through base metal with evidence of pre-
C)
fracture deformation. (a) Five-link fuU-size chain test, • 0.3. (b) Fractured link in full-$ize chain test, • 0. 75. -1-

GO
01

"11
r-
c
m

z
r-
.<

FIG. 8--Photographs o f specimen Code 166A 25. 4-mm (1-in.) Grade 80 chain. Fracture occurred through the weld with little evidence
o f prefracture deformation. (a) Five-link full-size chain test, x 0.3. (b) Fractured link in full-size chain test, • 0. 75.

crack geometry in which the crack length on the surface was four times the
crack depth was assumed, and the fracture-mechanics surface flaw equation,
Kic = 1.1ox/~-/Q, was used to calculate critical flaw depths. (Q is a crack
geometry factor with a value of 1.2 when o/oys = 1 and the crack length
-- 4 times the crack depth.) For example, for an assumed nominal stress
level (o) of 1035 MPa (150 ksi) and Kxc 27 MPa/x/-m (25 ksi ~ , "fhe
critical flaw depth (ac) is 0.23 mm (0.009 in.). Susceptibility to sudden
fracture in service under yield-stress loading in the presence of flaws of a
few mils in depth is obviously undesirable. If the Kz~ is 55 MPa/x/m (50
ksii,~.), the critical flaw size is only 0.89 mm (0.035 in.) for the same
nominal stress level of 1035 MPa (150 ksi). The opposite extreme--that is,
the high level of toughness exhibited by the Code 400 Grade 63 Chain (82 J
(60 ft-lb) in the weld)--indicates totally ductile behavior because a flaw
size approaching the chain diameter could be tolerated and failure could
occur only by overloading. For example, the Charpy V-notch impact test
energy of 82 J (60 ft- lb) correlates empirically with a critical stress intensity
(KI~) of 218 MPa/x/m (198 k s i x / ~ . At the yield stress level of 1035 MPa
(1S0 ksi) exhibited by the Code 400 Grade 63 chain, the calculated critical
flaw depth (a~) is 14.2 mm (0.56 in.), more than half the diameter of the
chain. Clearly, ductile behavior is assured in this chain.
In order to demonstrate the effect of flaws on the breaking strength of
chains, specimens of both Grades 63 and Grade 80 chains were notched at
Position A in Fig. 1 by using a 0.15-mm-thick (0.006-in.) milling cutter.
Position A was selected for the notch because (a) the elastic stress analysis
at that point was known, and (b) it was not possible to place the milling
cutter at Position B. A slot 0.15 by 2.54 by 10.9 mm (0.006 by 0.10 by
0.43 in.) in Grade 63 chain had a negligible effect on breaking strength; the
unnotched chain (Code 167A) exhibited a breaking strength of 714 kN (160.5
kips), Table 5, and the notched chain (Code 167B) broke at 699 kN (157.0
kips) Table 6 and Fig. 10. In contrast, the unnotched Grade 80 chain (Code
444) broke at a load of 950 kN (213.5 kips), Table 5, and the notched
chain (Code 166B), Fig. 11, broke at 405 kN (91.0 kips), Table 6. This
degradation of breaking strength of the Grade 80 chain in the presence of
an artifical notch 2.54 mm (0.1 in.) in depth was predictable from the
fracture-mechanics calculations. In fact, a back-calculation using (a) a
405-kN (91.0-kips) load, (b) a level of toughness associated with energy
absorption at 27 J (20 ft. lb) in the Charpy test (the same toughness mea-
sured at room temperature in the Grade 80 chain), and (c) an estimated
decrease in tensile fiber stress of 20 percent due to the bending stress
gradient with depth, resulted in a calculated critical flaw depth of 1.78 mm
(0.07 in.) The machined flaw, which is not as sharp or as effective in pro-
ducing failure as a natural crack of the same depth, therefore caused the
fracture to occur as predicted.
It is recognized that the experiment just described, in which the machined

FIG. 9--Specimen Code 166,4 25.4-mm (1-in.) Grade 80 chain. Fracture origin (B) exhibits
large inclusions and dimple rupture, and the fracture propagated (C) in a cleavage mode. (a)
Fracture surface of weld failure o f broken link in full-size chain test, x 2. (b) Scanning elec-
tron fractograph of fracture origin on broken link in full-size chain test, x 1500. (c) Scanning
electron fractograph o f propagating fracture on broken link in full-size chain test, x1500.

FIG. 9--Continued.

notches are located at Position A, is only a demonstration of the reduction

in load-carrying capacity of chain in the presence of a notch of arbitrary
size in the body of the most probable origin of a fracture. Furthermore, the
fracture toughness of the weld is the lowest of any location in the chain
link, so a weld flaw becomes critical and propagates at a lower stress level
than if a flaw of the same size were located elsewhere in the link. Therefore,

TABLE 7--Fracture-mechanics calculations o f critical flaw sizes.

Critical Flaw Depth, a c, in.

Nominal For Charpy For Charpy

Stress Level, V-Notch = 4 ft. lb or V-Notch = 16 [Link] o r
o, ksi KI~ = 25 ksi ~/in~-. K i c = 50 ksi~,/iff.

100 0.020 0.080

150 0.009 0.035
180 0.006 0.025

Conversion Factors--
1 in. = 25.4 mm,
1 ksi = 6.895 MPa,
1 ksi x/~. = 1.1 MPa/x/m, and
1 f t . l b = 1.36J.

z
m

z
o
"13
m
i-
r-
rn
0
-n
Z
0
0
z
oo
.-4
m
m
r-
FIG. lO--Photographs of specimen Code 167B 25.4-mm (I-in.) Grade 63 chain. Fracture occurred through the machined notch with evidence E
of prefracture deformation. (a) Five-link full-size chain test, x 0.3. (b) Fractured link in full-size chain test, x O.75. (c) Fractured surface showing "11
--4
machined flaw and ductile shear fracture, x2.0. -2
0
0"r

GO
..L

~o
z
m
.<
3>
z
c~
-o
m
r-
f.-
m
:o
z
0
0
z
cl)
FIG. l 1--Photographs of specimen Code 166B 25,4-mm (1-in,) Grade 80 chain. Fracture occurred through the machined notch with little evidence m
m
of prefracture deformation. (a) Five-link full-size chain test, x 0.3. (b) Fractured link in full-size chain test, x O.75. (c) Fractured surface showing r-
machined flaw and brittle (cleavage)fracture, • 2. O. l--
-4

0
I

the toughness of the weld material appears to be the parameter that governs
the susceptibility of chain to brittle fracture in service.

Summary
The present investigation was conducted to examine the interrelation-
ships of strength, toughness, and flaw tolerance of 25.4 mm (1-in.) alloy
steel lifting chain procured from five chain manufacturers. Small round-bar
tension-test and Charpy V-notch impact-test specimens were machined
from each chain to characterize the materials. Then an empirically de-
veloped Charpy-Kk correlation was employed, and fracture-mechanics
calculations of critical flaw size were carried out. The results showed that
Grade 80 chain, exhibiting a hardness of 40 HRC and relatively low tough-
ness, is susceptible to brittle fracture in the presence of small flaws. In
contrast, the high level of toughness for Grade 63 chain, exhibiting a hard-
ness of 33 HRC, can tolerate significantly larger flaw sizes resulting in a
chain failure when the tensile stresses in the remaining ligament exceed the
yield stress.
Tension tests on specimens from chain samples containing machined
notches 2.54 mm (0.1 in.) deep confirmed the calculations. Grade 80
chain containing the machined notch fractured at less than half the load
required to break unnotched chain; however, Grade 63 chain was virtually
unaffected by the presence of an 2.54 mm (0.1 in.) deep notch. The amount
of degradation of the breaking strength of the Grade 80 chain containing
the flaw was in agreement with fracture-mechanics calculations.

Authors' Note
It is understood that the material in this paper is intended for general
information only and should not be used in relation to any specific appli-
cation without independent examination and verification of its applicability
and suitability by professionally qualified personnel. Those making use
thereof or relying thereon assume all risk and liability arising from such
use or reliance.

Effect of the Amount and Shape of

Inclusions on the Directionality of
Ductility in Carbon-Manganese Steels

REFERENCE: Takada, H., Kaneko, K., Inoue, T., and Kinoshita, S., "Effect of the
Amount and Shape of Inclusions on the Directionality of Ductility in Carbon-Manganese
Steels," Fractography in Failure Analysis, ASTM STP 645, B. M. Strauss and W. H.
Cullen, Jr., Eds., American Society for Testing and Materials, 1978, pp. 335-350.

ABSTRACT: The present work has been made to elucidate the quantitative effects of
the amount and shape of inclusions in carbon-manganese steel plates on the tensile
and impact ductilities in connection with test directions.
The influence of rolling conditions on ductility can be expressed as a function of the
shape parameter of the inclusions in terms of the aspect ratio of elliptical manganese
sulfide on the polished surface.
Tensile and impact ductilities in transverse and through-thickness direcltions were
improved by rare earth metal (REM) additions to such an extent that the ductilities of
REM-treated steel specimens were nearly the same as those of the REM-free steel
specimens, although the inclusion content of the former was as much as two or three
times more than that of the latter.
Efforts were made to express ductility in one parameter regardless of the inclusion
shape and test directions, and it was found that the ductility was closely related to an
inclusion area fraction on the ductile fracture surface.

KEY WORDS. fractography, steels, ductility, inclusions, directionality

It is well known that impact and tensile ductilities of steel plates are de-
creased to a greater extent by the presence of nonmetallic inclusions acting
as nucleation sites of void in ductile fracture [1,2], 2 so that efforts are made
in production of clean steel by steel makers. Moreover, during the hot rol-
ling of aluminum-killed steel, the manganese-sulfide (MnS) inclusion de-
forms and the elongated inclusion reduces the ductility not only in the
through-thickness direction but also in the transverse direction [3, 4].
In recent years, since the importance of the through-thickness ductility
1Chief researcher, research metallurgist, research metallurgist, and senior researcher, re-
spectively, Department of Physical Metallurgy, Central Research Laboratory, Kobe Steel,
Ltd., Kobe, Japan.
2The italic numbers in brackets refer to the list of references appended to this paper.

335
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 1978 by ASTM International [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
336 FRACTOGRAPHY IN FAILURE ANALYSIS

has been emphasized in fabricating large and complex constructions, the

additions of rare earth metal (REM) or zirconium have been studied inten-
sively to prevent the harmful effects of elongated inclusions [5,6]. For in-
stance, these alloying elements have been utilized in lameUar-tearing-resis-
tant steel.
The MnS inclusion is known to deform to a larger extent at a lower rather
than higher rolling temperature [7], and the influence of the volume frac-
tion and shape of the elongated inclusions on ductility have been studied
extensively [1,4, 8, 9]. However, no attempt has been made to introduce a
parameter expressing the ductility, regardless of test directions and inclusion
variables.
The present study-is intended first to obtain the quantitative relationship
between the ductility of rolled plates and the shape of inclusions which
would be affected by rolling temperatures and rolling reduction ratios.
Secondly, we have attempted to express the ductility of rolled plates in
all directions as a function of one parameter irrespective of the test direc-
tions, rolling conditions, or volume fraction of inclusions and, moreover,
to know how the ductilities are affected by the addition of REM as a sulfide-
shape control agent.

The specimens used in this study were from heats with base compositions
of [Link]. The sulfur and REM contents and the general chemical
composition are given in Table 1.
Specimens A and B were obtained as slabs of commercial grade, and
Specimens C and D were 90-kg ingots from laboratory heats. Specimen
blanks from the slabs and the laboratory ingots were hot-rolled at high
( - 1250~ and low ( - 900~ austenite temperatures at various rolling re-
duction ratios. In order to keep the rolling temperatures as constant as
possible, specimens were reheated to the rolling temperature every two or
three rolling passes. Rolling temperatures were estimated to range from
1100 to 1250~ for high temperature rolling and from 780 to 900~ in the
case of low temperature rolling.

TABLE l--Chemical composition of steel (weight percent).

Specimen C Si Mn P S A1 Ce La

A 0.13 0.24 1.34 0.009 0.017 0.027 . . o

B 0.1S 0.37 1.35 0.018 0.008 0.032 0.007 0.005

C 0.15 0.41 1.52 0.012 0.009 0.036 0.006 0.006
D 0.14 0.45 1.64 0.012 0.024 0.044 0.061 0.008

Tensile and Charpy impact ductilities of the rolled plates were measured
in the longitudinal and transverse directions after normalization at 920~
Since the rolled plates were too thin to be tested directly in the through-
thickness direction, dummy materials were friction-welded to each side of
the rolled surface, and then Charpy and tension specimens were machined
after normalization at 920 ~
In Specimen A, which had no inclusion shape-control elements, some
elongated MnS and aluminum oxide (A1203) inclusions were observed on
the polished surface, while nondeformed sulfide and oxisulfide of REM
and some A1203 predominated in REM-treated steel (Specimens B, C,
and D).
Major axis (a) and minor axis (b) of each elliptical inclusion in rolled
plates were measured on the microphotographs taken at x 400 or x 1000
of the polished surface parallel to the rolling direction. The minimum
detectable inclusion size for the measurements was about 0.5 gin. For the
inclined inclusions, the projection lengths were measured as shown in Fig.
1. The measurements were carded out on more than 50 inclusions out of
approximately 10 fields. The value of log (a/b) was calculated for each in-
clusion and the mean value was defined as the inclusion shape parameter
of the rolled plate.

Rolling Direction

inclined inclusion

Inclusion Shape Parameter 9 Log o / b

FIG. l--Measurement of inclusion shapeparameter.

After Charpy and tension tests, the ductile fracture surfaces of each rolled
plate of REM-free and REM-treated steels were examined with a scanning
electron microscope (SEM).
Since an inclusion acts as an origin of ductile fracture, a certain correla-
tion was expected between the amount of inclusion existing on the ductile
fracture and the ductility of the steel. Therefore, area fractions of sulfide
and alumina inclusions on the fracture surface were measured using a quan-
titative television microscope (Quantimet 720) on the photographs of S-Kct
and AI-Ka specific X-ray images at x 400 on one side of the fracture sur-
face of the broken Charpy and tension test specimens. The measured areas
were a total of 0.18 mm 2 of the fracture surfaces.

ExperimentalResults

Influence of Rolling Conditions

Tensile ductility (reduction of area) of each direction in Specimen A
(REM-free steel) is illustrated against the rolling reduction ratio in Fig. 2.
The reduction of area in the longitudinal direction remained almost un-
changed with variations in rolling conditions, but in the transverse and the
through-thickness directions, the values decreased not only with increasing
rolling reduction ratio but also with decreasing rolling temperature. The
Charpy shelf energy changed in a similar fashion as did the reduction of
area with the rolling reduction ratio and rolling temperature, as shown in
Fig. 3. A significant feature was that the tensile and impact ductilities de-
creased systematically, corresponding to the variation of inclusion shape af-
fected by the temperature and rolling reduction ratio.

Heating
temperature
1250"c 900~

o 9 Longitudinal
9 Transverse
9 Through-
thickness
80 C twL
~ v~ A

"~ .~ &

"6
_~ 40

r~ 2 0

- ] I J I I I
4 6 8 I0 20 40
Rolling Reduction Ratio

FIG. 2--Effects o f rolling reduction ratio and temperature on reduction o f area.

On the other hand, the reduction of area of REM-treated Specimen B

is shown against the rolling reduction ratio in Fig. 4. These results suggest
that the influence of rolling conditions on the ductility of the REM-treated
steel is much smaller than is the case with the REM-free steel. This is in
good agreement with previous investigations [6, 9].

Heating ]
temperature (
1250~ gO0*c J
o 9 Longitudinal L
/
A 9 Transverse J
!
[] 9 Through- |
thickness ~250
25

"-.._
~_ I00 q-
- .._-~.~ --.~ | ==

I I I I I I /
2 4 6 810 20 40
Rolling Reduction Ratio
F I G . 3--Effects of rolling reduction ratio and temperature on Charpy shelf energy in Spec-
imen A.

8O - ~ ~ ~ _ ~ : _

o 60 - Heating
<~ temperature
1250~ 900~
"6 4O - o 9 Longitudinal
_g z~ 9 Transverse
-o 20 _ n 9 Through-
I1: thickness

I I I I I I I
4 6 810 20 40 60
Rolling Reduction Ratio
F I G . 4--Effects of rolling reduction ratio and temperature on reduction o f area in REM-
treated Specimen B.

Effect of Inclusion Shape Parameter

The results just mentioned in Specimen A suggest that the ductility in
each direction depended primarily upon the shape of MnS which deformed
under various rolling conditions.

The reduction of area was plotted against the inclusion shape parameter,
log (a/b), in Specimen A rolled at high or low temperature and is shown in
Fig. 5. In the longitudinal direction, the effect of shape parameter on the
tensile ductility was small, but the ductility of the transverse and through-
thickness directions decreased in proportion to the increasing log (a/b).
When the value of log (a/b) exceeded approximately 1.8, the relation in the
through-thickness direction deviated from the linearity.
The Charpy shelf energies of each direction also were affected by inclusion
shape and, as shown in Fig. 6, these decreased with increasing log (a/b).
It was observed that the ductility decreased as the rolling reduction ratio
increased or as rolling temperature dropped, and this effect is ascribed to
the inclusion shape change as shown in Figs. S and 6.

Influence of Volume Fraction of Inclusion

Although the influence of the inclusion shape on the ductility of rolled
steels was important, total volume fraction of inclusions had a similar effect.
Volume fraction of inclusion was regarded as the same value as the inclusion
area fraction on the polished surface.
Figure 7 shows the relationship between Charpy shelf energy and inclu-
sion volume fraction of REM-free specimens on plate rolled to a reduction
ratio of 12 at low temperature and on some commercially rolled plates which
had similar chemical compositions and inclusion shape parameters of log
(a/b) of about 1.8.
Figure 8 shows the similar relation in REM-treated Specimen B rolled
to reduction ratio of 16, Specimens C and D rolled to reduction ratio 9 at

Heating
temperature
o ~ [] 1250Oc
9 9 9 900Oc

Longitudinal
o
80

60 Transverse
L ~A~
"
~r~
.-~ 40
1O
Through- - . c ~ .
2o thickness "~11_..__ _~..m._

i l I i I [ I
1.0 12 1.4 1.6 IB 2-0 ?..2
Inclusion Shape Parameter, Log a / b
FIG. S--Dependence of reduction o f area in each direction upon inclusion shape parameter
in Specimen A .

Heating
femperuture
o zx n 1250"c
9 9 9 9O00c
25 250

20 .~00

15 50 z

~.~. Transverse

I0 IO0 ,,~

5 - ~ . 50
Through -
thickness
I I J J I i l

1.0 1.2 1.4 1,6 I JB 2.0 22

Inclusion Shape Parameter , Log o / b

FIG. 6--Dependence of Charpy shelf energy upon inclusion shape parameter in Speci-
men A.

Transverse
~ o~
inal
250

2O0

150v
E
Z

6
e- .=
ta I0 I00 w

(/)
5O
Through - o "-'---r~"
thickness
I I I i I _

Q04 006 0.1 02

Inclusion Volume Fraction ( % )
FIG. 7--Influence of inclusion volume fraction on Charpy shelf energy in REM-free steels.

low temperature, and some commercially rolled plates having similar chem-
ical compositions.
These figures indicated that the impact ductility of each direction in
REM-treated steel was higher than that in REM-free specimens when com-

25
\ ~ongitudinal
"~ 2O 200
E
.lg I o-,%
Z
_- \ Transverse ~'~ 15O
15

ID
W I0 "~ \\ A C
UJ
t-
h\a, " \-, '&-

ff) 9 \AA IU

Through- ~. [] \
5 thickness ~ \\ 50
u\
I I I I I
004 (~06 al 02
Inclusion Volume Fraction (%)
FIG. 8--1nfluenceof inclusion volumefraction on Charpyself energyin REM-treatedsteels.

pared having the same amount of inclusions. Figure 9 shows the influence
of the volume fraction of inclusions on the reduction of area of through-
thickness direction in both kinds of steel9 It also can be seen that the REM
addition resulted in the formation of nondeformed inclusions and rendered
a larger tensile ductility than REM-free specimens. The Charpy shelf energy
and reduction of area of REM-treated steels fell almost on the same level
as those of REM-free steels, even though the former contained two or three
times as many inclusions as that of the latter.

Influence o f Inclusion Area FraCtion on the Ductile Fracture Surface

Figure 10 shows the fracture surface of the Charpy specimen in the lon-
gitudinal direction in Specimen A rolled to the reduction ratio of 23 at
1100 to 1250~ The MnS inclusions were found at the bottom of large
dimples. The fracture surface of the transverse direction of the same speci-
men is shown in Fig. 11, and MnS also can be seen lying at the bottom of
large elongated dimples. Both fracture surfaces were composed mainly of
large dimples with small dimples around them. Figure 12 shows the frac-
ture surface of the through-thickness direction in Specimen A rolled to the
reduction ratio of 12 at 780 to 900~ It can be seen that MnS inclusions
are plate-like and also lie parallel to the fracture surface. The photographic
features of the fracture surface and appearance of MnS inclusions on it
in Specimen A varied significantly with test directions, while in a REM-

[] : E M - treated

~e~
4060 D

[]\
20 \x\\at
REM-free \

Incl0.04 nQ06 Froction

0,1 (%)02.
I I 1 l I

usioVolume
FIG. 9--Influence of inclusion volume fraction on reduction of area of through-thickness
direction in REM-treated and REM-free steels.

[Link]--Fracture surface o f longitudinal Charpy specimen in Specimen A rolled to reduc-

tion ratio of 23 to 1250~

[Link]--Fracture surface of transverse Charpy specimen in Specimen A rolled to reduc-

tion ratio of 23 at 1250~

treated specimen, globular or cluster-type inclusions predominated as

shown in Fig. 13. This figure shows the fracture surface of the through-
thickness direction of commercial steel (0.11C-1.4Mn-0.005S-0.02Ce), and
no significant difference was found in the features of inclusions with regard
to the test directions.
As mentioned previously, tensile and impact ductilities were found to be
expressible as a function of the inclusion shape parameter log (a/b) but
they still could not be expressed except with reference to test directions.
Several previous investigations had pointed out that the anisotropy of
ductility was due primarily to the presence of elongated inclusions [I0,11].
Since the inclusions, acting as nuclei of fracture, appear on the fracture
surface [1,4], it is reasonable to expect that the inclusion area fraction on
the ductile fracture surface is related closely to ductility in each direction.
As indicated in Fig. 14, the inclusion area fraction on the fracture surface
increased not only with log (a/b) but also with the test direction altering
from longitudinal to transverse and transverse to through-thickness. It
should be noticed that this trend is very closely related to ductility.

FIG.12--Fracture surface of through-thickness Charpy specimen in Specimen A rolled to

reduction ratio o f 12 at 900 ~

Charpy shelf energy and tensile fracture strain el are plotted against the
log of the inclusion area fraction on the fracture surface and are shown in
Figs. 15 and 16, respectively. Here tensile fracture strain el is related to
the reduction of area p by the equation ci = In 1/(1 - p). The results of
REM-treated steel and other commercial steel plates with different sulfur
contents are also plotted in both figures.
As can be seen in Figs. 15 and 16, it is noteworthy that the ductilities of
all steel specimens show the same relationship to inclusion area fraction
on the fracture surface. This relationship is independent of any factors such
as rolling conditions, sulfur contents, REM additions, or test directions,
which control the shape and volume fraction of inclusions. This relationship
will be discussed later. It also is seen in these figures that the ductilities
increase with the decrease of inclusion area fraction on the fracture surface.
However, the ductilities deviated from the linear relation and leveled off
as the inclusion area fraction on the fracture surface decreased to zero.
This can probably be attributed to the fact that ductility came to be gov-
erned primarily by other second-phase particles as the amount of inclusions

FIG.13--Fracture surface o f through-thickness Charpy specimen in commercial REM-

treated steel plate.

~ Through-

,.g
o

< ~I0
--- ~ .

Longitudinal

I I~ I I
1.0 1.4 1.8 22
Inclusion Shape Porometer, Log o/b
FiG.14--Inclusion area fraction existing on the Charpy fracture surface as a function o f
inclusion shape parameter.

- o~ REM- REM- Test

2 5 [treoted [free /direction 250
t .~ I " ~ Or~itt'd~~

~ o/ I IThrough-

~o ioo

- $
50 u)

I i I I i i i t

0.2 0.4 I 2 4 IO 20 40
Inclusion Area Fraction on the Fracture Surface (%)
[Link]--Relationship between Charpy shelf energy and the logarithm of inclusion area
fraction on the fracture surface.

became extremely small. The fracture strain approached eI = 1.64, which

corresponds to the fracture strain of a ferrite-pearlite structure of 0.15C
steel [12].

Discussion

The results described in this paper indicate that there exists a definite
relation between ductilities and inclusion area fraction on the ductile frac-
ture surface regardless of test directions, roiling conditions, or the amount
and shape of inclusions of the specimens.
It is generally known that there is an approximate equation between true
stress (a) and true strain (e)
o = k e" (1)
where k and n are material constants of the specimen.
Assuming Eq 1 is satisfied until the fracture occurs, then the equation
is rewritten as

~i -- k eF (2)

where o / i s the true fracture stress corrected by the Bridgman method and
e/is the true fracture strain. Suppose the effect of the inclusion is sufficiently
small, o/would be equal to the fracture stress of the matrix, Ore, and e f t o
the fracture strain of the matrix, era, then Eq 1 is rewritten as

~., = k e ,." (3)

As the inclusion area fraction on the fracture surface increases, the fracture
stress of the matrix increases according to the equation

Om = of (1-2f~) (4)

1.6 ~ o

,~1.2 z~ ~
~ ~ o []
c

(/~O.B Ire~ free ~direction I

;
~a4
i/ ~.'. I"
I - [-r,.==,~-.~ I
oX~',~
h

L_~I " Ithick~ I "

I I [ ] I [ I I I t I I
Q2 0.4 I 2 4 I0 20 40
Inclusion Areo Froctionon the Frocture Surfoce(%)
[Link]--Relationship between tensile fracture strain and the logarithm of inclusion area
fraction on the fracture surface.

1.6! o Longitudinol
9', Transverse
L~ o Through-
12 thickness

._= oo\,
~. 0.8 \~o /~'f
n= n
~m (I-2fo)
(/1 ~\ / [~m "1.64
\ n -0.2
X
8 0.4 o \\
~_ Q',\ o

o \ []

I [ 1 J I I

0.1 02. 0.3 0.4 Q5 0.6

Inclusion Areo Fraction on the
Fracture Surface, 2 fa
FIG. 17--Calculated and experimental relationships between strain to fracture and inclusion
area fraction existing on the fracture surface.

where fa is the inclusion area fraction on one of the fracture surfaces out
of a pair of broken specimens. The relation between r and ~z is obtained
from Eqs 2, 3, and 4 and is given by

~f = em'(1 - 2f.) (5)

Since e m is the fracture strain of the matrix, it will be independent of the

variables relating to inclusions and is expected to be a fracture strain at an
extremely lowf~, being regarded as em = 1.64 from Fig. 16. Equation 5
predicts the ductility as a function offa. Figure 17 shows the calculated
result, letting the strain hardening exponent n be equal to 0.2. This is in
good agreement with experimental values.

Conclusion
Impact and tensile ductilities were measured in longitudinal, transverse,
and through-thickness directions of rolled steel specimens of two groups.
In the first group, the shape of inclusion varied with rolling conditions,
and in the second group the inclusion shape was controlled by the addition
of REM.
The ductilities varied to a larger extent with the variables governing the
shape and amount of inclusions such as rolling conditions, sulfur contents,
and the addition of REM. However, it was indicated that the ductilities
were expressed by one unique parameter, the inclusion area fraction on
the fracture surface, regardless not only of inclusion variables but also of
the test directions. Moreover, this parameter would be more useful for
estimating the ductilities of materials containing second-phase particles
if we could predict it in a nondestructive way. For this reason, further in-
vestigations are under way to achieve the expression of the parameter as
a function of shape parameter and volume fraction of inclusions.

References
[1] Gurland, J. and Plateau, J., Transactions, American Society for Metals, Vol. 56, 1963,
pp. 442-454.
[2] Ashby, M. F., Philosophical Magazine, Vol. 14, 1966, pp. 1157-1178.
[3] Vogels, H. A., Dahl, W., Hengstenberg, H., and Briining, F., Archivff4r das Eisen-
hf~ttenwesen, Vol. 33, 1962, pp. 649-659.
[4] Baker, T. J. and Charles, J. A., "Effect of Second-Phase Particles on the Mechanical
Properties of Steel," The Iron and Steel Institute, 1971, pp. 79-87.
[5] Luyckx, L., Bell, J, R., McLean, A., and Korchynsky, M., Metallurgical Transactions,
Vol. 1, 1970, pp. 3341-3350.
[6] Croll, L E. and Macdonald, L K., The Journal of Australian Institute of Metals, Vol.
19, 1974, pp. 161-167.
[7] Maunder, P. J. H. and Charles, J. A., Journal of The Iron and Steel Institute, Vol. 206,
1968, pp. 705-715.
[8] Thomason, P. F., Journal of the Institute of Metals, Vol. 96, 1968, pp. 360-365.

[9] Bernard, G., Grumbach, M., and Moliexe, F., Metals Technology, Voi. 2, 1975, pp.
512-521.
[10] Morrison, W. B., Metals Technology, Vol. 2, 1975, pp. 33-41.
[11] Grange, R. A., Metallurgical Transactions, Vol. 2, 1971, pp. 417-426.
[12] Inoue, T. and Kinoshita, S., Transactions of the Iron and Steel Institute of Japan, Vol.
17, 1977, pp. 245-251.

Comparison of the Threshold Stress

Intensities and Fracture Characteristics
for Temper Embrittled and Deembrittled
21/4Cr-1Mo Steel in a Hydrogen
Charging Environment

REFERENCE: Hicho, G. E. and Gilmore, C. M., "Comparison of the Threshold

Stress Intensities and Fracture Characteristics for Temper Embrittled and Deembrlt-
tied 2V, Cr-IMo Steel in a Hydrogen Charging Environment, Fractography in Failure
Analysis, A S T M STP 645, B. M. Strauss and W. H. CuUen, Jr., Eds., American
Society for Testing and Materials, 1978, pp. 351-362.

ABSTRACT: Fracture toughness tests and fractographic examinations were conducted

on temper embrittled and deembrittled 21ACr-lMo steel using double cantilever-beam
specimens. The tests were conducted in an aqueous/acetic acid solution containing
hydrogen sulfide (H2S). The threshold stress intensity of the temper embrittled specimen
tested in the H2S environment was lower than that of the deembrittled specimens tested
in a similar environment. For the purpose of comparing fracture appearances, temper
embrittled and deembrittled specimens were fractured in air.
The fracture appearances of the temper embrittled and the deembrittled specimens
tested in the H2S environment and the temper embrittled specimen fractured in air
were predominantly intergranular. The fracture appearance of the deembrittied speci-
men fractured in air exhibited a transgranular ductile mode of failure. These results
indicate that the embrittling effects due to temper embrittlement and to the H2S en-
vironment act cooperatively in reducing the threshold stress" intensity of 21ACr-lMo
steel.

KEY WORDS: steels, temper embrittlement, hydrogen embrittlement, fracture prop-

erties, fractography

One of the more important metallurgical problems occurring in the

petrochemical industry concerns the effects of a hydrogen sulfide (H2 S) en-
vironment on the fracture toughness of 21ACr-lMo steel. This steel is
often used in gas-well tubings and petrochemical units because of its re-
sistance to hydrogen attack, especially at high temperatures. However, ex-
1 Metallurgist, National Bureau of Standards, Washington, D.C. 20234.
2 Associate professor, George Washington University, Washington, D.C. 20006.

351
Copyright by ASTM
9 Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright 1978 by ASTM International [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
352 FRACTOGRAPHYIN FAILURE ANALYSIS

posure of this steel to H2S from "sour" gas may promote hydrogen-
assisted stress corrosion cracking (HASCC) which leads to a reduction in
the ability of the steel to resist crack growth.
The HASCC effects have been studied on numerous alloys. An examina-
tion of some of the fundamental theories of stress corrosion cracking was
presented elsewhere [I ] 3
In addition to the problem of HASCC, temper embrittlement sometimes
results from prolonged exposure to high temperatures. Temper embrit-
tlement has been studied extensively and has been discussed in an in-
terpretive review by McMahon [2]. Temper embrittlement occurs when
certain alloy steels are heated or cooled within a particular temperature
range. This results in the segregation of the elements phosphorus, arsenic,
antimony, and tin to the prior austenitic grain boundaries [3-7] allowing
grain boundary failure to occur more readily. These elements have also
been known to promote HASCC in steel [8].
Investigations of the effects of a combination of temper embrittlement
and hydrogen embrittlement on various materials have been published.
Cabral et al [9] have presented results describing the effects of the segre-
gation of impurities on hydrogen-induced cracking in high strength steel.
Their results showed that the threshold stress for tempered nickel-chromium
steel tested in a sulfuric acid (HzSO4) solution was reduced when the steel
was aged at 773 K (932 ~
Yoshino and McMahon [10] performed experiments on 5 percent nickel-
chrominm-molybdenum-vanadinm (NiCrMoV) steel in the embrittled and
nonembrittled condition in a H2 SO4 environment and observed a reduction
in the K threshold of the embrittled material. They attributed their re-
sults to the presence of grain boundary impurities. These impurities were
said to reduce the grain boundary cohesion, thus allowing grain boundary
fracture to occur more readily.
Viswanathan et al [11], in their work on 4340 steel, determined that
temper embrittlement caused a reduction in the K~c as well as the K ~ in
H,S. Their results showed that phosphorus had the most significant effect
on Km~ and KIc. In addition, they stated that for low strength steels, im-
purity segregation caused weakening of the grain boundaries and sub-
sequently reduced the K ~ .
McMahon et al [12] have shown recently that as the grain boundary con-
centration of embrittling impurities increases, the intergranular cohesive
strength and the threshold stress intensity for hydrogen-induced cracking
decreases. These results indicate that the threshold stress intensity is re-
duced by a combination of these phenomena.
This investigation uses fracture mechanics to show that the threshold
stress intensity, Kth, for 2%Cr-lMo steel double cantilever beam (DCB)

3The italicnumbersin brackets referto the list of referencesappendedto this paper.

specimens temper embrittled and tested in an aqueous H2S environment,

was lower than the Kth for the same steel deembrittled and tested in a
similar H2 S environment. It was believed that these are the first results of
this type of 2188 steel.
Fractographic results are presented which show the fracture appearance
for the test and specimen conditions described. In addition, fracture tough-
ness and fractographic results are presented for temper embrittled and
deembrittled specimens tested in air.

Experimental Procedure
The test methods and procedures followed in this investigation were
those recommended by Heady [13].
The steel used in this investigation was temper embrittled by subjecting
it to a temperature range of 616 to 727 K (650 to 850~ for 10 000 h and
subsequently characterized by the determination of the chemical compo-
sition, tensile properties, and impact properties.
For comparison purposes, it was necessary to deembrittle a portion of
the as-received material. This was accomplished by subjecting the steel to
866 K (1100~ for 75 rain and quenching it in room-temperature water.
Fracture toughness tests were conducted on the wedge-loaded DCB
specimens. A drawing of the DCB specimen is shown in Fig. 1. To avoid
the problem of side cracking, the specimens were machined from the plate
so that their orientation corresponded to the S-T orientation as described in
ASTM Test for Plane-Strain Fracture Toughness of Metallic Materials
(E 399-74).
The specimen thickness used in this investigation was 9.5 rnm (0.375
in.), and this thickness did meet the size criteria established for plane
strain conditions as stated in ASTM Test E 399-74. The size criteria and
its relation to HASCC is now discussed.
According to ASTM Test E 399-74, the primary criteria for plane strain
conditions is that B, the specimen thickness, determined by the following
equation be satisfied

B> 2.5(K~ 2
_ - - (1)
k Cry~/

This equation was not developed to apply to environmentally induced

fracture, and its applicability is open to question [14]. For example, in a
discussion of stress corrosion test methods, Parkins [15] states that when
very ductile materials are tested and plane strain conditions are to be fol-
lowed, a very thick specimen is required. Since most corrosion failures

A S. + l/io

I
q~e
~.- .040" "~.003

~ ,

II -~\\\\\\\\~
I 1
.OION~.OO2RAOiUS
GROOVE ROOT
SECTION AA 9 3/~ R. (ARMATURE OR
~g/ TYP BOTH SIDES SECTION BB CHARPYCUTTER)
(ARItaITURs CUTTER)
I1~1"TO 11/4"R. A~PTABLE

FIG. 1 - - T h e D C B specimen.

occur in very ductile material in thin sections, it is implied that the frac-
ture mechanics criteria as to specimen size is not applicable. On the other
hand, Brown [16] suggests that fracture mechanics could he used in stress
corrosion work, "as a means of referencing stress in a body containing a
crack in a manner applicable to various geometries." It appears that using
fracture mechanics to determine the threshold stress intensity due to a cor-
rosive environment is acceptable.
Precracking of the DCB specimens was accomplished by inserting a
wedge into a specimen that was cooled previously in liquid nitrogen. The
wedge was tapped and the specimen was observed for pop in. Care was
taken so that the displacement created during precracking did not exceed
the displacement necessary to obtain the initial K value.
Once a crack was observed in the specimen, the crack length was
measured to the nearest 0.25 mm (0.01 in.) using a microscope accurate to
0.0025 mm (0.0001 in.). Each crack length was measured three times on
each side of the specimen. The average of these six measurements was taken
as the initial crack length.
Preliminary test results of Interrante and Hicho [17] served as a basis for
determining the initial K value to which specimens were to be loaded in
this investigation. Their work revealed that for 2aACr-lMo steel (HRC 20)
loaded to an initial K of about 110 MPa d-m (100 ksi ~-n.), the Kth due to
the H2S environment was 33 MPa v ~ (30 ksi i,,/in~-.). As a result of that
work, the initial stress intensities were set at the values listed in Table 1.
Additional preliminary work showed that the incubation period for crack
initiation was about 3 to 5 h and that crack growth had terminated after
five days.
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions aut
"1"
0
-i-
0

z
o
0
TABLE 1--Fracture and fractograpkic test results. Specimens 2, 7, and 8 were wedge-loaded DCB specimens. Specimens 10 and 11 were loaded as
single-edge notch specimens.
0
-n
m
Test Initial
Condition K Kth 0
z

Deem- Temper Fracture -r

Specimen brittled Embrittled H2S Air MPadm ksiix/~. MPa ksix/'m. Appearance -n
m
-r
10 X X 129 117 . . . . . . intergranular 0
11 X X 204 186 ... ductile
2 X X 66 60 53" 48 intergranular
7 X X 96 87 77 70 intergranular
8 X X 112 102 82 75 intergranular
m
co
E-
-4
m
z
Co

roa
Oa

After loading to the initial K value, the deembrittled specimens were

placed in a test environment consisting of distilled water containing 1A
weight percent glacial acetic acid saturated with H~S at one atmosphere.
The environment was charged each day with H2S for 5 min. After five
days, the specimens were removed from the environment. Crack growth
was measured on each specimen and the K,h calculated. From the K,h re-
sults for the deembrittled specimens, the temper embrittled specimen was
loaded to 66 MPa ~ (60 ksi ix/~-.) and placed in the test environment for
five days.
For comparison of fracture appearance, deembrittled and temper em-
brittled specimens were tested in air. The specimens were precracked ac-
cording to the previously described method, then placed in a tensile
machine for loading as one would load a single-edge notch specimen. The
load necessary to obtain crack instability was obtained and the stress in-
tensity factor was calculated using the equation for single edged notch
specimens [18].
Fractographic examinations were conducted on all of the specimens
using a scanning electron microscope (SEM). The surfaces were cleaned
in flowing hydrogen gas prior to examination [17] to remove a corrosion
product which formed on the specimens subjected to the H2S environment.

Results and Discussion

Table 1 shows the results of the fracture toughness tests and subsequent
fracture appearances. The results show that the temper embrittled speci-
mens tested in the H2S environment and in air had lower threshold stress
intensities than the deembdttled specimens tested in the H2 S environment.
Figure 2 shows a macrograph representative of the fracture surface of

FIG. 2--Macrograph of Specimen 2 showing three distinct areas o f crack growth: Area A,
the precrack region; Area B, crack growth due to the test environment; and Area C, crack
growth due to the specimen being pulled apart at liquid N2 temperature (X6).

a specimen. Three areas of crack growth are evident: Area A, the pre-
crack region; B, crack growth due to the H2S environment; and C, crack
growth due to being pulled apart at liquid N2 temperature.
Figure 3(a) and (b) are SEM fractographs (Area B) of the temper
embrittled and deembrittled specimens tested in the 1-12S environment. The
fracture appearance of both specimens was intergranular.
Figure 4(a) and (b) are SEM fractographs of the temper embrittled and
deembrittled specimens tested in air. The fracture appearance of the
temper embrittled specimen was intergranular whereas the fracture ap-
pearance of the deembrittled specimen was transgranular ductile.
Based primarily on the fracture appearances of the specimens, in par-
ticular the absence of shear lips, it was concluded that plane strain con-
ditions existed. Final crack measurements revealed that growth exceeded
the plastic zone size [19] established for plane strain conditions. These
results indicated that the gth values obtained in this investigation are
valid.
The results of the chemical analyses, Table 2, shows the weight per-
centages of the temper embrittling elements arsenic, antimony, tin, and
phosphorus. The weight percentages of these elements usually are reported
in parts per million for normal steels, but they were found to be abnormally
high in this steel. It was concluded that the segregation of these elements
to the prior austenitic grain boundaries and the exposure to 616 to 727 K
(650 to 850~ for 10 000 h produced the temper embrittlement observed
in the steel. It was also concluded that temper embrittlement was the
primary cause of the intergranular fracture observed in the fractographs.
A measure of the extent of temper embrittlement is revealed in the
Charpy impact test results. Impact test results for the temper embrittled
and deembrittled material are shown in Fig. 5. The results clearly showed
that the temper embrittled material (intergranular fracture) absorbed less
energy at a given temperature than the deembrittled material.
Table 3 shows the tensile results for the temper embrittled material. Vis-
wanathan et al [11] have shown that for a 4340 steel at the yield strength of
approximately 1150 MPa (167 000 psi), the segregation impurities produce
a weakening of the grain boundaries with a reduction in the K~s~. The
yield strength of the material used in this investigation was 421.5 MPa
(62 000 psi). It was concluded from a comparison of these yield strengths
that grain boundary weakening should also be expected to occur in the
material used in this investigation.
In addition to the temper embrittlement, the H2S environment is also
considered to be a promoter of intergranular fracture.
The H2S environment, in conjunction with temper embrittlement, pro-
duced conditions that favored intergranular failure. This strong tendency
for intergranular failure subsequently led to a reduction in the Kth o f
21ACr- 1Mo steel.

FIG. 3--SEM fractographs of the temper embrittled Specimen 2, (a), and the deembrit-
tled Specimen 8, (b) taken in the crack growth Region B shown in Fig. 2. Both specimens
were tested in the H2S environment. Fractures were primarily intergranular.

FIG. 4--SEM fractographs of the temper embrittled Specimen 10, (a) and the deembrittled
Specimen 11, (b). Both specimens were tested in air. Fracture of (a) was primarily inter-
granular whereas (b) was transgranular ductile.

TABLE2--Chemical composition of the as-received, temper

embrittled 2 88Cr- l Mo plate (weight percent).

C 0.14
Mn 0.65
P 0.018
S 0.006
Si 0.27
Cr 2.32
Ni 0.20
Mo 1.05
Cu 0.20
As 0.016
Sn 0.026
Sb 0.0044

TABLE3--Room temperature tension tests results for the temper embrittled 2 88 plate.

Specimen Orientation
Longitudinal Transverse

Ultimate 599.8 MPa 586.1 MPa

tensile strength (87 000 psi) (85 000 psi)
Yield strength, 421.5 MPa 420.6 MPa
0.2% offset (62 000 psi) (61 000 psi)
Elongation, 30% 29%
5.08 cm (2 in.)
Reduction of area 71% 68%

Coneinsiom

Experimental results have shown that subjecting temper embrittled

21ACr-lMo steel to a H2S environment reduces the threshold stress in-
tensity. These results indicate that temper embrittlement and a H2S en-
vironmerrt act cooperatively in promoting intergranular fracture of 21ACr -
1Mo steel.

Acknowledgments

The authors wish to thank D. Truax of the Standard Oil Company of

California for the test material used in this investigation. The chemical,
tensile, and impact data on the test material also were supplied by Truax.
The authors wish to acknowledge the following members of the Mechanical
Properties Section of the National Bureau of Standards for their efforts:
C. G. Interrante, J. G. Early, D. E. Harne, L. C. Smith, C. H. Brady, and

200 - - 271.2

"T MBRITTLED

150 - - 2 0 3 . 4
>..

,..., I 0 0 -- 155.6

n-
O
o3
,,=[
.
50 - - 67.8
;jx x
I I I I I I I
-50 0 50 I00 150 200 250
227 255.2 268.6 310.8 339 366 :594
TEST TEMPERATURE, ( F / K )
FIG. 5--Charpy V-notch impact test results for the temper emhrittled and deembrittled
test material.

L. E. Ketron, who aided in the preparation of the photographs and graphs

used in this paper.

References
[1] The Theory of Stress Corrosion Cracking In Alloys, L C. Scully, Fxl., North Atlantic
Treaty Organization, Scientific Affairs Division, Brussels, 1971.
[2] McMahon, C. J., Jr. in Temper Embrittlement in Steel, A S T M STP 407, American
Society for Testing and Materials, 1968, pp. 127-167.
[3] Steven, W. and Balajiva, K., Journal of the Iron and Steel Institute, Vol. 193, 1959,
pp. 141-147.
[4] Marcus, H. L. and Palmberg, P. W., Transactions, Metallurgical Society of the
American Institute of Mining, Metallurgical, and Petroleum Engineers, Vol. 245, !960,
pp. 1664-1666.
[5] Palmberg, P. W. and Marcus, H. L., Transactions, American Society for Metals,
Vol. 62, 1969, pp. 1016-1018.
[6] Stein, D. F., Joshi, A., and Laforce, R. P., Transactions, American Society for Metals,
Vol. 62, 1969, pp. 776-783.
[7] Low, J. R., Jr., Stein, D. F., Turkalo, A. M., and Laforce, R. P., Transactions, Metal-
lurgical Society of the American Institute of Mining, Metallurgical, and Petroleum
Engineers, Vol. 242, 1968, pp. 14-24.
[8] McCright, R. D. and Staehle, R. W., International Conference on Stress Corrosion
Cracking and Hydrogen Embrittlement of Iron Base Alloys, Firming, France, 1973.
[9] Cabral, U. Q., Hache, A., and Constant, A., Computed Rendus Acad ofSei, Vol. 260T,
1965, p. 6887.
[10] Yoshino, K. and McMahon, C. J., Jr., Metallurgical Transactions, Vol. 5, 1974, p. 363.
[11] Viswanathan, R. and Hudak, S. J. in Effect of Hydrogen on the Behavior of Materials,
American Institute of Mining, Metallurgical, and Petroleum Engineers, 1975, pp. 262-
272.

[12] McMahon, C. J., Jr., Briant, C. L., and Banerji, S. K., Fracture 1977, Vol. 1, IC4F,
Waterloo, Canada, 1977, pp. 363-373.
[13] Heady, R. B., "Sulfide Corrosive Cracking in Gas and Oil Wells. XIV," Technical
Progress Report BRC-Corp 5-74-B, Shell Oil Company, Houston, Texo
[14] Novak, S., Engineering Fracture Mechanics, Vol. 5, 1973, pp. 727-763.
[15] Parkins, R. N. in The Theory of Stress Corrosion Cracking In Alloys, 1971, pp. 449-468.
[16] Brown, B. F. in Stress Corrosion Cracking of Metals--A State of the Art, ASTM
STP 518, American Society for Testing and Materials, 1971, pp. 3-15.
[17] Interrante, C. G. and Hicho, G. E. in Stress Corrosion--New Approaches, ASTM STP
610, American Society for Testing and Materials, 1976, pp. 349-365.
[18] Tada, H., Paris, P. C., and Irwin, G. R. in The Stress of Analysis of Cracks Handbook,
Del Research, 1973, p. 216.
[19] Irwin, G. R., Engineering Fracture Mechanics, Vol. 1, No. 2, 1968, pp. 241-255.

Fractographic Analysis of Ceramics

REFERENCE: Mecholsky, J. J., Freiman, S. W., and Rice, R. W., "Fraetographie

Analysis of Ceramics," Fractography in Failure Analysis, ASTM STP 645, B. M.
Strauss and W. H. Cullen, Jr., Eds., American Society for Testing and Materials,
1978, pp. 363-379.

ABSTRACT: The development of fracture surface observations into a quantitative

tool for analysis of brittle fracture is reviewed. The effect of temperature, strain
rate, and residual stresses on the "mirror constant," A, is discussed through recent
experimental results, as well as new analysis of established literature. The current
theories of mirror formation in glasses, glass ceramics, single crystals, and poly-
crystalline ceramics are compared to the results of a number of investigators. The
relationship of the mirror constant to fracture toughness, that is, Kic, is discussed.
Methods to incorporate the effects of grain size and internal stress on fracture are
demonstrated. Finally, application of fracture surface analysis to scientific and applied
problems are enumerated.

KEY WORDS: fractography, fracture surface analysis, ceramics, fractures (materials),

brittle materials, mechanical properties

Since at least the 1920s, c h a r a c t e r i s t i c f r a c t u r e features have b e e n ob-

served on glass [1,2] 2, single crystals [3], n a t u r a l polycrystalline m a t e r i a l s
like b a s a l t [4,5], a n d i n t e r m e t a l l i c fibers [6]. I n a d d i t i o n , W a l l n e r [7]
n o t e d c h a r a c t e r i s t i c lines f o r m e d on the " m i r r o r " surface o f glass f r o m t h e
intersection o f the p r i m a r y stress front with t h e stress front g e n e r a t e d f r o m
a s e c o n d a r y flaw. F r o m these early observations o f c h a r a c t e r i s t i c features,
relationships between these features a n d the f r a c t u r e stress in brittle m a t e -
rials was noted, in p a r t i c u l a r by S h a n d [8], T e r a o [9], a n d S m e k a l [10]. It
was also shown [8] t h a t t h e s e " W a l l n e r lines" can be used to calculate t h e
velocity o f the c r a c k front. These c h a r a c t e r i s t i c f e a t u r e s have been shown to
be r e l a t e d to t h e f r a c t u r e origin [11,12], velocity o f c r a c k p r o p a g a t i o n [13],
a n d critical stress intensity [14,15] a n d strain intensity [16] at f r a c t u r e .
These relationships are used to analyze t h e f r a c t u r e o f brittle m a t e r i a l .
This p a p e r will review t h e c u r r e n t state o f u n d e r s t a n d i n g o f f r a c t u r e sur-

1Research ceramic engineer, supervisory research engineer, and supervisory research

ceramic engineer, respectively, Naval Research Laboratory, Washington, D.C. 20375.
2The italic numbers in brackets refer to the list of references appended to this paper.

363
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 by ASTM International [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
364 FRACTOGRAPHY IN FAILURE ANALYSIS

face analysis and will discuss the important applications of this technique
to brittle failure.

Fracture Surface Analysis

Fracture in brittle materials occurs from defects like machining or
mechanically produced cracks, pores, and inclusions. Although actual
defects are not necessarily regular in shape, they usually can be idealized
by semielliptical flaws in the manner prescribed by Randall [17]. For
surface flaws that are perpendicular to the applied stress, one can relate
the size of the original defect~ c, to the failure stress, a, through fracture
mechanics equations a similar to those used for metals

Kic
~fc - ~ (1)

where Kic is the critical stress intensity factor, and 9 is an elliptical in-
tegral of the second kind which accounts for the geometry of the crack.
Specifically, flaws due to machining have been characterized and shown
to agree well with Eq 1 to predict failure of brittle materials [18]. Pores,
either singly or in groups, act as the source of failure and can be analyzed
according to Eq 1 as long as the relationship of the pores to the local
microstructure is taken into consideration [19]. Likewise, inclusions [20]
and grain boundary grooves [21] can be treated as fracture-initiating flaws
in such a way that Eq 1 can be utilized.
While observation of the fracture-initiating flaw is important to analysis,
in many cases flaws are nonplanar, too small, do not have clear boundaries,
or a chip has come out that includes the flaw, so that one cannot estimate
the flaw size. The real power of fracture surface analysis is in utilizing the
fracture boundaries that form outside the flaw to describe fracture be-
havior, regardless of the flaw shape and any difficulties in identifying the
fracture origin. Four definitive regions of fracture, surrounding fracture-
initiating flaws, have been observed in brittle materials [22,23] (Fig. 1).
The first region, generally smooth and commonly known as a mirror,
is bounded by another region of small radial ridges, known as mist, which
in turn is bounded by an even rougher area, known as hackle, which in
turn is bounded by macroscopic crack branching. The distances from
the fracture initiating flaw to these boundaries, that is, ri, Io, rcb, have
been shown experimentally to be related to the fracture stress, a

o r 1/2 = constant = A (2)

3This expressionneglects the correctionsfor a back-facefree surface and plastic flow at the
crack tip as given in Ref 17 because, for most cases, these are not applicable for ceramics.
Also, failure generallyoccurs at or near the surface.

FIG. 1--Schematic of the shape and general appearance of fracture mirror and related
features on typical brittle fracture surface.

These radii are commonly referred to as mirrors in brittle materials, so

that the distinction is made between inner (mirror-mist) and outer (mist-
hackle) mirror boundaries with their corresponding "mirror constant," A.
It also should be indicated that in many instances the specimen size is too
small to contain all of these boundaries so only some of the features are
seen.
In order to understand this relationship, it is necessary to examine the
fracture process, at least qualitatively. When a flaw begins to propagate as
a sharp crack, its vdocity increases with increasing crack length until it
nears the terminal velocity for that material, approximately 0.6 of the
shear wave velocity. Since the strain energy associated with the lengthening
crack plus its kinetic energy can no longer be used to increase its velocity,
other processes commence. These processes lead to the four definitive re-
gions surrounding the fracture origin (Fig. 1). These regions are observed
as long as the specimen size is large enough to accommodate the features,
since the crack basically propagates as though it were in an infinite body
regardless of specimen size. The first feature formed is a bright, shiny
region; in glass, it if known as the fracture mirror. As the terminal velocity
is approached, nucleation in the vicinity of the crack tip of microcracks
(mist) occurs which are energetically unable to propagate over large dis-
tances. As the crack propagates further, enough energy becomes available
so that the secondary cracks now can propagate over longer distances,
leading to hackle. In all materials, when the crack becomes large enough,
macroscopic crack branching accurs, although this phenomena is not
always observed because the specimen is too small or there is propagation
into a compressive region. The attempts to explain crack branching quanti-
tatively will be discussed later. At this point we will show the results of the
experimentally observed fracture mirror.
Micrographs of representative fracture surfaces of glass-ceramic, fine-
grain polycrystalline and single-crystal ceramics, and glass, showing the
areas of fracture initiation, are presented in Figs. 2 and 3. In comparing
Figs. 1, 2, and 3, one can see the similarity of most brittle materials. In

many ceramics, there is an identifiable region surrounding the failure

origin that reflects light better than regions further from the origin and
usually is bounded by small flakelike particles (mist). The beginning of
hackle is clearly evident on these micrographs, as it is on most other
ceramics examined, and closely resembles that observed on glasses.
While there is some evidence of mist formation, this feature is not always
discernible in polycrystalline ceramics, especially in large grain or highly
porous materials. In large grain ceramics, many of the fracture steps left
from transgranular failure on the easy cleavage plane or on grains of
limited misorientation focus onto the fracture origin thus making identifi-
cation of the source of failure easier. Cleavage or fracture steps as well as
hackle markings also appear on the fracture surfaces of single crystals

FIG. 2--Optical photographs of fracture surfaces of (a)arsenic trisulfide (As 2S 3), and (b)
glass carbon showing source of failure (F), mist (M), and hackle (H) regions. The fracture
surface of As 2S 3 is similar to those of silicate glasses, whereas in glassy carbon the separa-
tion of mist and hackle is greater.

depending on the orientation of crack propagation. However, in single

crystals, the occurrence of mist and hackle depends on the fracture plane
and the direction of propagation on this plane [22].
In glasses, the energy needed to initiate or propagate a secondary crack
is isotropic so the formation of mist is expected to occur at an energy near
that required to form hackle, as is observed experimentally. In polycrystal-
line ceramics, however, secondary cracks can be nucleated on cleavage
planes within grains or along grain boundaries at an energy near that of
single crystals, and hence much less than that required to propagate the
primary crack through the polycrystalline microstructure, leading to much
earlier formation of mist. For these secondary cracks to extend to form
hackle means that more energy must be supplied to the fracture process.
In polycrystalline ceramics, this energy comes from the greater extension
of cracks away from the inner (mirror-mist) boundary leading to larger
ratios of hackle to mist radii [15]. However, even with individual differences,
the features on the fracture surfaces of glasses, polycrystalline, and single
crystal ceramics generally obey Eq 2.
Logarithmic plots of flexural strength versus the outer mirror radii are
presented in Figs. 4 and 5 to demonstrate the difference in mirror constant
for several representative materials. The curves are the best fit straight
lines of slope - 0.5. The use of this slope is based on the assumption that
Eq 1 is valid for these materials. In fact, in many cases, the best linear
least squares fit had a slope of -0.5. Extended data for glasses and ce-
ramics were given in previous papers [12,15]. Strength mirror size curves
for ADP [24], hot pressed alumina [25], sapphire [15,26], silicon nitride
[23], and AISiMag 614 [23] are also given in the literature. There are, how-
ever, external factors than can also influence the mirror constant. These
are discussed in the following sections.

Effect of Temperature
Both Shinkai [27] and Mecholsky [28] found that the mirror constant in
glass was higher at - 150~ than at 20~ Mecholsky [28] also found that not
only was the mirror constant higher at - 150~ but also that the inner mir-
ror to flaw size ratio was much smaller (6:1 compared to 13:1), a value
commonly observed in polycrystalline ceramics. This decrease indicates
that this temperature decrease produces more than just an increase in
fracture toughness. The limited research in this area and at elevated
temperature [29] indicates that much more research is needed in the area
of fracture surface analysis at low and high temperatures.

Effect of Residual Stresses

In a study of the effect of tempering on strength, Kerper and Scuderi [30]
found the mirror size-modulus of rupture relationship to be independent
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductio
368 FRACTOGRAPHY IN FAILURE ANALYSIS

FIG. 3mOptical fractographs of (a) baria-silica (3BaO.5Si02) glass ceramic, (b) hot-
pressured silicone nitride (Si3N4 ), and (c) single crystal ammonium diphosphate (ADP). Broad
arrows indicate outer mirror demarcations; thin arrows in (a) and (b) indicate inner mirror
boundaries; thin arrows in (e) may be outer mirror boundary with no inner mirror (mirror-
mist) boundary present. Black arrows indicate machined flaws which acted as the source of
failure. The failure origin in (b) is an inclusion (black dot approximately in the center of the
broad arrows), The dashed line is just inside the outline of the outer mirror.

FIG. 3--Continued.

i r i I I I Ill l I } ] I
20o x AszS3 -- 3 O O
o GLASSYCARBON
~ l 9 BOROSILICATE
v 9 SODA LIME 200
~
SODA
i i v v
~ ASIOz GLASS --

E 80
Z zx O
-LEACHED ~ ~5,..~ ~ ~,~"':~,,..~m~...,~ a -- I O 0 ~
~ 6o - 8o~

~= 4 0 6 0 u~

20 --

/ I I I I i I i I L I L I I i
0,2 03 0 4 0 5 O6 0 8 1.0 2.O '%0 6.0
OUTER MIRROR RADIUS (ram)

FIG. 4--Fracture stress versus outer (mist-hackle) mirror radius of silicate and nonoxide
glasses. Solid lines represent slope o f -O. 5 [12].

iO00 I I I .... [ I I I [ .... I

0 ZIRCONIA
Q BORON CARBIDE
9 pYROCERAM 9606 - -
DO
x GRAPHITE
9O
o _- 8O
500
70
400 oo _-6O
50
~" 3OO

i
D=

I--

~ loo

0.2 0.3 0.4 0.6 1.0 2.0 3s 4.0 0.6 1QO

OUTER MIRROR RADIUS (ram)

FIG. S--Fracture stress as a function o f outer mirror (mist-hackle) radius f o r representative

polycrystalline ceramics. Solid lines are least squares fit of slope - 0.5 (10 3 psi = 1 ksi) (1 ksi
= 6. 9 MPa) [1S].

of location of fracture origin for new glass. For semitempered and tempered
glass, they found that the slope of the logarithmic plot of 6 versus r '~ did
not equal 0.5 as predicted by Eq 2. However, there are two approaches to
plotting these data: (a) plot the data on a log-log plot and use regression
analysis to determine the slope or (b) assume Eq 2 holds and attribute
deviations in the intercept to extrinsic variations such as residual stress
[23,31]. For example, if unlike Kerper and Scuderi [30] who followed the
first approach, one assumes that Eq 2 is valid and plot ~ versus r '~ for
these data (Fig. 6), a residual stress of approximately 70 MPa is predicted
in their tempered glass as might he expected.

Effect of Strain Rate

Although the values of fracture stress are changed with rate of loading,
the mirror constant should be relatively insentitive to this parameter. 4
Kirchner [31] used fracture mirrors to study impact of various glasses and

4Although the average fracture stress increases with an increase in the rate of loading, the
distance to the mirror boundaries would correspondingly decrease, thereby not changing the
value of the mirror constant.

2 4 6 8 i0 12
A l u m l n o s l l l c a t e Glass
Data of Kerper and Scuderl
30
200 / /
Tempered
15C
~.o

lOC

o ~
i0 20 30 40 50 60 70

F I G . 6--The data o f R e f 30 replotted assuming Eq 2 is valid. This gives an estimate o f the resi-
dual stress present for each heat treatment.

polycrystalline ceramics both at room and elevated temperatures. He made

the assumption that the relationship in Eq 2 is valid for different rates of
loading, which recently has been substantiated for glasses by Bradt et al
[32]. More recently, Kirchner [33] has shown that even impact loading
producing Hertzian cone fractures lead to fracture mirrors which can be
used in the analysis of brittle fracture for dynamic loading. Pohanka et
al [24] have shown that mirror formation from dynamic stresses in piezo-
electric ceramics produced by voltage application give rise to mirror con-
stants that agree with those observed for the normal fracture (Fig. 7).

Criteria for Formation of Fracture Features

Many attempts have been made to try to explain the formation of mist,
hackle, and crack branching. These phenomena can basically be repre-
sented by three approaches; crack branching occurs: (a) at a characteristic
or critical velocity [34]; (b) at particular strain intensity [16] or stress
intensity [15,35,35] values; or (c) when the strain energy release rate ex-
ceeds that required for propagating planar cracks [37-42].
Most of the velocity flaw size relations are of the form

vo(m
where v is the velocity of the crack of length, r, and c is the critical flaw
size. Because c/r at branching is small compared to 1, however, large
changes in c/r make only small changes in velocity, so that estimates of
ratios of (mist-hackle) radius to flaw size from velocity data are subject
to large inaccuracies. It is unclear at this time whether the three criteria
are different or whether they really result in the same predictions of frac-
ture mirror formation and boundaries. That is, Bansal [42] and Abdel-
Latif et al [40] have related the mirror constant to the critical stress in-
tensity factor, including a kinetic energy term similar to the approach of
Johnson and Holloway [14] through the energy balance approach suggested
by Roberts and Wells [38].
Regardless of the criteria which govern the formation of mist and
hackle, Eq 2 is still valid and can be used in fracture surface analysis.
Fracture mechanics expressions (Eq 1) can be combined with fracture
surface analysis (Eq 2) to explain quantitatively the mirror boundaries [15].
The mirror constant has been shown to be related to the critical stress in-
tensity factor, the fracture stress, and the mirror to flaw size ratio

OKlc
(4)

This relationship states that the mirror constant is proportional to the

critical stress intensity factor at failure, s These results are compared to
actual measurements of mirror constants in Table 1. In fact, qualitatively
comparing all these results implies that the velocity is constant for a given
c/r ratio. In the velocity formulations, however, numerical predictions of
mirror constants from these equations are not in general agreement with
actual observations (Table 1), but there are some parameters in the equa-
tions that are difficult to determine, so one can understand if accurate pre-
dictions are not possible. The formulation which includes the mirror to
flaw size ratio [15] agrees well with this observation. This implies that the
distance to the fracture mirror boundaries most likely includes any kinetic
energy effects on crack propagation. In addition, these equations suggest
that the velocity, strain energy release rate, and strain intensity criteria are
really different expressions of the same phenomenon. It is clear, however,
from the fact that boundaries do not occur in certain directions in single

s This is the same form of the equation with slight modifications as given by the strain in-
tensity proponents. Since the velocity is related to the (c/r) ratio, this form is also the same as
the energy balance proponents as well as the constant velocity proponents for tbrmation of
crack branching, Table l.

"~.X ,,t ADP 42m 0 DYNAMIC

~ ; ' t ~ X STATIC

~ ~o~ o

o
I I I I I I 1 I I
0,1 0.2 0.4 0.6 0.810 2 4 6 8 tO
MIRROR SIZE (ram)
FIG. 7--Static and dynamic fracture stress versus mirror sizefor ADP single crystal. The static
strength was measured on several sizes of 45-deg z-cut ADP bars in three-point flexure. The dy-
namic strength was measured by driving the bars piezoelectrically in a free-free longitudinal
mode until failure occurred [24].

crystals [22] that all of these are necessary but not sufficient conditions
for boundary formation.
If we plot available data as A versus K~ (Fig. 8) as suggested by Eq 4
with certain exceptions discussed later, the data are seen to fall within a
band centered on a mirror to flaw size ratio of 13 to 1 (for semicircular
flaws). There is no evidence to suggest that the points falling either above
or below this trend represent failure of a particular type, for example,
machine flaws, pores, large grains, etc. This trend implies that the ratio
of outer mirror size to flaw size is a constant in glasses and ceramics inde-
pendent of crystal structure (but not direction in crystals), composition, or
mierostructure. The data in Fig. 8 would tend to suggest that Eq 4 is valid
for a wide range of ceramics and that if the mirror size can be measured on
the fracture surface a n d Kit is also known, an estimate of the flaw size
can be obtained for that specimen. The validity of this approach has been
shown for many glasses, polycrystalline ceramics, and singlecrystal ceramics
as well as glass ceramics [15]
The lithia-silica glass ceramics are notable exceptions to the relationship
in Fig. 8. The deviation of these data would predict that their mirror to
flaw size ratio would be about 8 to l, when in reality, measurement of the
mirror to flaw size ratio gives a number around 13 to 1 [43]. The difference
has been explained by the presence of microcracking at the top of the pri-
mary crack front which would be an energy-absorbing process in this mate-
rial, analogous to the plastic zone in metals. The existence of microcrack-
ing in lithia-silica glass ceramics is well established [44,45]. In fact, the
amount of deviation to the right of the curve provides a measure of the
amount of microcracking that is occurring in the material, that is, at the
given mirror constant the intersection with the line in Fig. 8 would give
the critical stress intensity factor for a lithia-silica glass ceramic without
microcraeking.

TABLE 1--Comparison of criteriafor crack branching,a "rt

Author(s) (A/Kk) Formulation (A/Kk) (boundary) (A/KIc) Measured
Johnson and Holloway [14] (mirror-mist) 2.5 G)
2~ - k---~2 (mist-hackle) 2.7 ~.
1.2 (branching) "D
ar
.<
Congelton and Petch [13] ~ 0.8 (branching) 1.4

Abdel-Latif, e rb
a n d T r e s[4s0]lBradt, I ( 3 1 -'2 k P ~ ) ] 1/2 1.8 (mirror-mist) 2.3 t~--Fro

>
Bansal [42] 2~ 1/2 1.6 (mirror-mist) 2.3
3kov2 .~

Mecholsky, Freiman, and Rice ~ 1.25 ] 1/2 2.8 (mirror-mist) 2.7

[15] [2 ( c ) ] 3.2 (mist-hackle) 3.1

NOTE--v = 1500 m/s,

p = 2.5 g/era 2, and
k = 44.
a Soda-lime glass.
bTensile ease; since k = 44 or v = 1500 m/s cannot be used because the term in parenthesis becomes negative;
k = 22 and v = 1000 m/s were used as suggested in the reference.

20.0 I I I I I I

x SINGLE CRYSTALS
o POLYCRYSTALLINE CERAMICS Si3 N4 (HS 130)
18.0 GLASSESro
I/9 ~
AO=Y(-~-) KIC
16.0
r

14.C At SiMag614

12.(D
; /H RAI2~O5
I0.0 -- ///'*SIC
~o MgO, / __
/ .B 4 c

o 8.0 SPINEL(NRL)
/
< 9 ZYTTRITE
BaT 03 .......
- BAlz O~~ ,/~9- PYROCERAM
')bQU':)•IUz 9606
(LiF- MgO) SrZrO 9
6.0-- ~ ~ o 5i/~?MULLITE o Li20.2SiO2
" ~ E RV/6T/ SAPPHIRE
BaTi03o XMgO
9Li20.2SiO2
4.0 PZT/
/ 9 GRAPHITE
x,~/~'" Mg F2
r~ -"'SPINEL
2.0 ~k~GLASSES
/ \ "znse
/ "GLASSY CARBON
CADP
Or I I I I I I
0 1.0 2.0 3.0 4.0 5.0 6.0 70
KI(; (MN/m3/2 )

FIG. 8--Outer mirror (mist-hackle) contant, Ao, as a function o f the critical stress in-
tensity, K lc, for ceramic materials [15].

Other materials also have been shown to exhibit microcracking [46] (for
example, Poco graphite [47], zirconia [48], alumina-zirconia [49], silicon
nitride, some aluminas, and many other noncubic ceramics). Modifications
of Eq 2 must be made to account for the effect of microcracking in any
analysis.
Another factor that must be considered is that when the fracture in-
itiating flaw is of the order of the size of the local microstructure, single
crystal rather than polycrystalline fracture mechanics applies. That is, the
calculations predicting either strength or flaw size demonstrate that a
value of "re approaching that of a single crystal is appropriate for mate-
rials in which flaws are contained in one or two grains [50,51]. Because of
the large size of the crack compared to the microstructure, the stress in-
tensity of the mirror boundary as represented by the mirror constant, A,

is a measure of the average fracture toughness of the material. It should not

be surprising then that this value cannot be used to predict failure con-
ditions for the local properties in the materials where the flaw size is
smaller than the large grains; Eqs 1 and 4 must be modified by including
the single crystal rather than the polycrystalline K~c [51].
Also, if internal stress is present due to either thermal expansion anisot-
ropy in noncubic ceramics such as beryllium oxide (BeO) and alumina or
as a result of phase transformations, Eq 1 must be modified to account for
this stress [52]

~Klc
oa + (oi) = 1.12,~- (5)

where (o'i) is the effective internal stress acting on the flaw. Since even a
small flaw may encompass several grains, there will be some averaging of
the stresses around its perimeter. Hence, the value of <o;> will depend on
the ratio of the flaw size to the grain size. (oi) would be expected to ap-
proach zero as the flaw size increases, namely, as the perimeter of the flaw
averages more and more of the tensile and compressive components of the
internal stress in the body. Studies [53] have corroborated the hypothesis
that (ol) decreases from a value approaching the theoretical limit in the
material for very small flaws, to zero for large flaws.

Application of Fracture Surface Analysis

Observation of characteristic markings from fracture surfaces can be
used not only in the study of basic phenomena of crack propagation but
also as an aid in determining the effects of various mechanical phenomena:
polishing, grinding, impact, and processing. Several examples are given
here to demonstrate the usefulness of fracture surface analysis.
Mecholsky et al [18] used the relationship between fracture mirror and
flaw sizes to study the effect of machining on the shape of the fracture in-
itiating flaws in the fracture of soda lime glass. From this study it was
shown that mirror size measurements followed Eq 2 even when the frac-
ture-initiating flaw was out of the plane of fracture or a very comi31icated
type flaw, thus demonstrating the power of fracture mirror measurements
over local measurements at the crack tip. In addition, they showed that
fracture energy, and thus Kic, was independent of flaw geometry. They
further observed that the area of the flaw divided by the square of the
outer mirror radius was a constant in agreement with the subsequent pre-
diction of Bansal [54]. In addition, the use of fracture surface analysis ex-
plained differences of strength produced by grinding in two different
directions. These studies have been extended to polycrystaUine ceramics

and single crystals and to polishing of ceramics. Since polishing is com-

monly a random process of fine grinding, a failure occurs from the worst
case, generally an elongated flaw, a/b ~ 0.5 in both glasses and poly-
crystalline ceramics. The case for single crystals is complicated by elastic
anisotropy, but the study of fracture surface analysis has provided a tool to
investigate the effect of anisotropy on flaws produced by grinding, and,
hence, on fracture.
Another example is in the field of high strength optical fibers. In drum-
to-drum proof testing of optical fibers, there are numerous breaks. It
cannot always be determined whether the breaks occur when full tension is
on the fiber or before it. By use of fracture surface analysis, one can deter-
mine not only the strength at failure but also the source of failure. For
example, by observing the fracture surface Mecholsky et al [55] determined
that two fibers which were thought to be proof tested at 1379 MPa
(200 000 psi) failed at much lower strength, around 345 MPa (50 000 psi),
because of a crack and a dust particle which were fused to the fiber
surface, respectively. The latter, most likely, occurred during the drawing
process, causing a sharp flaw upon cooling. These fibers obviously failed
before they reached the full stress on the drum. Without fracture surface
analysis, one would have suspected that these two fibers reached the 1379-
MPa (200 000-psi) level. An important result that came from another study
of optical fibers by Maurer et al [56] was that the mirror constant for fiber
rods is the same for that as silica bars tested in three-point bending, thus
extending the strength-mirror size relationship for glasses.
Although they will not be discussed in detail here, there are a number of
other applications of fracture surface analysis to ceramics. These include
the analysis of stresses during impact [31], failure stress predictions due to
thermal shock [29], and analysis of failures of piezoelectric components
which occur during voltage application [25,57].

Sllltnl~gkl~
Fracture in ceramic materials results in fracture surface features known
as mirror, mist, and hackle. The formation of mist and hackle, which
represent different stages of secondary crack formation, occurs due to the
excess energy of the moving crack over what is needed to propagate it. The
boundaries of the mist and hackle regions have been shown to be related
quantitatively to the stress at fracture as well as to the critical flaw size in
the material. It was demonstrated that the so called mirror constant, A,
can be thought of as a measure of [Link] intensity factor at the mist and
hackle boundaries. The ratio of A to the critical stress intensity factor for
fracture, Kk, is a constant for a wide range of ceramic materials. In any
quantitative analysis of fracture of polycrystalline ceramics, however, one

m u s t t a k e i n t o a c c o u n t a n y i n t e r n a l o r r e s i d u a l stresses in t h e b o d y as well
as c o n s i d e r t h e r a t i o o f t h e c r i t i c a l f l a w size to t h e size o f m i c r o s t r u c t u r a l
f e a t u r e s . F i n a l l y , a n u m b e r o f a p p l i c a t i o n s o f f r a c t u r e s u r f a c e analysis w e r e
discussed.

References

[1] Preston, F. W., Journal of the American Ceramic Society, Vol. 14, 1931.
[2] Preston, F. W., Journal of the Society of Glass Technology, Vol. 10, 1926, pp. 234-269.
[3] Zaffee, C. A. and Wardon, C. O., Acta Crystallographica, Vol. 2, Part 6, 1949, pp.
377-382.
[4] Iddings, J. P., American Journal of Science, Vol. 31, 1886, p. 321.
[5] Preston, F. W., Proceedings of the Royal Society, B, 1930.
[6] Davies, G. J. and Broom, N. D., The Philosophical Magazine, 1972.
[7] Wallner, H., Zeitschrifl fur Physik, Vol. 114, 1939, pp. 368-378; Ceramic Abstracts,
Vol. 19, No. 6, 1940, p. 137.
[8] Shand, E. B., Journal of the American Ceramic Society, Vol. 37, No. 12, 1954, pp. 559-
572.
[9] Terao, N., Journal of Physics, Proceedings of the Physical Society, Japan, Vol. 8, 1953,
pp. 545-549.
[10] Smekal, A., Journal of the Society of Glass Technology, Vol. 20, 1936.
[11] Krohn, D. A. and Hasselman, D. P. H., Journal of the American Ceramic Society, Vol.
54, No. 8, 1971, p. 411.
112] Mecholsky, J. J., Rice, R. W., and Freiman, S. W., Journal of the American Ceramic
Society, Vol. 57, 1974, p. 440.
[13] Congelton, J. and Petch, N. J., The Philosophical Magazine, Vol. 16, 1967, p. 749.
[14] Johnson, J. W. and Holloway, D. C., The Philosophical Magazine, Vol. 14, 1966, p. 731.
[15] Mecholsky, J. J., Freiman, S. W., and Rice, R. W., Journal of Material Science, Vol.
11, 1976, pp. 1310-1319.
[16] Kirchner, H. P., "Criteria for Fracture Mirror Boundary Formation in Ceramics,"
Proceedings of ICM-II, Boston, Mass., 1976.
[17] Randall, P. N. in Plain Strain Crack Toughness Testing of High-Strength Metallic
Materials, ASTM STP 410, W. F. Brown, Jr. and J. E. Srawley, Eds., 1966, pp. 88-126.
[18] Mecholsky, J. J., Freiman, S. W., and Rice, R. W., Journal of the American Ceramic
Society, Vol. 60, No. 3-4, 1977, pp. 114-117.
[19] Evans, A. G. and Tappin, G., Proceedings of the British Ceramic Society, Vol. 20,
1972, p. 275.
[20] Baratta, F. I., Driscoll, G. W., and Katz, R. N. in Ceramics for High Performance
Applications, Proceedings of the 2nd Army Materials Technology Conference, Hyannis,
Mass., Nov. 1973.
[21] Coble, R. L., Journal of the American Ceramic Society, Vol. 54, 1971, p. 59.
[22] Rice, R. W., Surface and Interfaces of Glass and Ceramics, Freehette, Lacourse, and
Burdick, Eds., Plenum, New York, 1972.
[23] Kirchner, H. P. and Gruver, R. M. in Proceedings of Symposium on Fracture Me-
chanics of Ceramics, Bradt, Hasselman, and Lange, Eds., Plenum, New York, Vol. 1,
1973, pp. 309-321.
[24] Pohanka, R. C., Smith, P. L., and Pasternak, J., "Report of NRL Progress," U.S.
Naval Research Laboratory, Jan. 1975, p. 21.
[25] Kirchner, H. P. and Gruver, R. M., The PhilosophicalMagazine, Vol. 27, 1973, p. 1433.
[26] Becher, P. F., U.S. Naval Research Laboratory, private communication.
[27] Shinkai, N., Japanese Journal of Applied~ Vol. 14, No. 1, 1975, pp. 147-148.
[28] Mecholsky, J. J., U.S. Naval Research Laboratory, unpublished data.
[29] Kirchner, H. P., Gruver, R. M., and Sotter, W. A., Ceramic Finishing Company
Report No. 2, 1974.
[30] Kerper, M. J. and Scuderi, T. G., Journal of the American Ceramic Society, Vol. 44,
No. 12, 1965, pp. 953-955.

[31] Kirchner, H. P. and Sorter, W. A., Ceramic Finishing Company Report No. 1, 1974.
[32] Bradt, R. C., Pennsylvania State University, private communications.
[33] Kirchner, H. P. and Gruver, R. M., Ceramic Finishing Company Report No. 5, Jan.
1977.
[34] Yoffee, E. H., The Philosophical Magazine, Vol. 42, 1951, p. 739.
[35] Clarke, A. B. J. and Irwin, G. R., ExperimentalMechanics, Vol. 23, 1966, pp. 321-330.
[36] Petch, N. J. in Fracture, H. Liebowitz, Ed., Academic Press, Vol. 1, 1968, pp. 351-393.
[37] Mort, N. F.,Engineering, Vol. 165, 1948.
[38] Roberts, D. K. and Wells, A. A.,Engineering, Vol. 178, 1954, p. 1820.
[39] Berry, J. P., Journal of Mechanics and Physics of Solids, Vol. 8, 1969, pp. 194-216.
[40] Abdel-Latif, A. I. A., Tressler, R. E., and Bradt, R. C., International Journal of Frac-
ture Mechanics, to be published, 1977.
[41] Dynamic Crack Propagation, G. C. Sih, Ed., Nordoff, 1973.
[42] Bansal, G. K., The Philosophical Magazine, to be published.
[43] Mecholsky, J. J. and Freiman, S. W. "Fracture Surface Analysis of Glass Ceramics,"
Proceedings of Eleventh International Congress on Glass, Prague, Czechoslovakia, July
1977.
[44] Sahoo, M., Rao, A. S., and Nadeau, J. S., Technical Report, Center for Materials
Research, University of British Columbia, 1972.
[45] Freiman, S. W. and Hench, L. L., Journal of the American Ceramic Society. Vol. 55,
1972, p. 86.
[46] Wu, C. Cm., Rice, R. W., Freiman, S. W., and Mecholsky, J. J., submitted to the
Journal of Material Science, 1977.
[47] Meyer, R. W., Zimmer, J., and Almon, M. C., Report No. ATR74 (7408) 2, Aerospace
Corp. March 1974.
[48] Green, D. J., Nieholson, P. S., and Embory, J. D., Journal of the American Ceramic
Society, Vol. 56, 1973, p. 619.
[49] Claussen, N., Journal of the American Ceramic Society, Vol. 59, No. 1-2, 1976, pp.
49-51.
[50] Rice, R. W. in Proceedings of Symposium on Fracture Mechanics of Ceramics, Pennsyl-
vania State University, July 1973.
[51] Freiman, S. W., Mecholsky, J. J., Rice, R. W., and Wurst, J. C., Journal of the Ameri-
can Ceramic Society, Vol. 58, 1975, p. 406.
[52] Pohanka, R. C., Rice, R. W., and Walker, B. E., Journal of the American Ceramic
Society, Vol. 59, 1976, p. 71.
[53] Pohanka, R. C., Freiman, S. W., and Bender, B. A., Journal of the American
Ceramic Society, Vol. 61, 1978, pp. 1-2.
[54] Bansal, G. K., Journal of the American Ceramic Society, Vol. 59, No. 1-2, 1976, pp.
87-88.
[55] Mecholsky, J. J., Freiman, S. W., and Morey, S. M., Bulletin of the American Ceramic
Society, accepted for publication, Dec. 1977.
[56] Maurer, R. C., Miller, R. A., Smith, D. D. and Trondeen, J. C., ONR Contract
N00014-73-C-0293, Coming Glass Works Technical Report, March 1974.
[57] Pohanka, R. C., Rice, R. W., Pasternak, J., Smith, P. L., and Walker, B. E., Proceed-
ings of Workshop on Sonar Transducer Materials. Smith and Pohanka, Eds., U.S.
Naval Research Laboratory, Nov. 1976.

CopyrightbyASTMInt'l(allrightsreserved);MonDec[Link]ST2015
Downloaded/printedby
UniversityofWashington(UniversityofWashington)[Link].
STP645-EB/May 1978

Summary

This symposium volume is divided into four areas: techniques, environ-

mental effects, fatigue, and stress and nonmetals.
In the techniques section, the'common approach was to develop a set of
reference fractographs of cracks with known histories for comparison to the
fractographs from service failures. Steele and Lentz utilized both microscopy
and fractography to characterize cleavage, ductile rupture, intergranular
separation, and hydrogen embrittlement in low carbon steel and then com-
pared the resulting fractographs to those characterizing failures in drawn
cup walls, and quenched and tempered plate. Young and Kumar simulated
the manufacturing environment which caused hydrogen embrittlement of
9Ni-4Co-0.20C steel to produce the same intergranular fracture found in
a service failure of the aft-fuselage structure of the B-1 aircraft.
Meyn presents examples of a wide range of materials and components
which had failed and the techniques employed to identify the fracture mode.
The techniques used consist of the entire range of resolution from the
unaided eye to the transmission electron microscope.
Madeyski and Albertin are more general in their suggested techniques.
They discuss the unconventional technique of looking at electrically con-
ductive replicas in the scanning electron microscope and compensating for
mirror image effects by electronically reversing the image. They also have
gotten good correlation in relating striation spacing in fatigue failures to
macroscopic da(AK)/dN by means of the Bates-Clark relation and present
a practical example.
Papers on environmental effects deal primarily with hydrogen embrittle-
ment phenomena, corrosion fatigue, and stress-corrosion cracking. Gang-
loft and Wei examined the mechanism of hydrogen-induced cracking in
18Ni maraging steels and conclude that the crack path depends mainly on
temperature. At low temperatures, intergranular cracking proceeded
along prior austenite grain boundaries, and as the temperature increased
the cracks became transgranular and propagated along the lath martensite
boundaries. This was in complete agreement with the results presented by
Kikuta et al, who made the same observations in other steels. Both papers
agreed that the hydrogen-assisted cracking mechanism can be characterized
by the role of microstructural sites participating in hydrogen diffusion re-
sulting in the embrittlement process.

383
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 byASTMInternational [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
384 FRACTOGRAPHY IN FAILURE ANALYSIS

Lee et al discuss two similar failures in Ti-8AI-IMo-IV gas turbine fan

blades. Both failures had initiated by stress corrosion cracking: one at high
temperature, the other at ambient. The remainder of the subcritical crack
growth in both failures was found to be fatigue at ambient temperature.
Laboratory controlled hot salt stress corrosion tests provided the basis for
comparison with the service failures.
Corrosion fatigue was also examined by Hioki and Mukai. Their work
was on Type 304 stainless steel in boiling 42 percent magnesium chloride.
They utilized a parametric approach in determining the service life of this
material in this environment. They conclude that the cyclic rate affects
fatigue life at rates less than 103 cycles per minute (cpm), that the rate of
crack propagation is proportional to the maximum stress intensity factor,
and that the fractography, at higher cyclic rates and static loading, is
transgranular, and at lower cyclic rates, such as 1 cpm, intergranular.
Mukai et al examined the same material in the same hostile environment
under constant load and found that the orientation of the fracture surface
was in the (100) plane. They also contend that striation-like markings ap-
peared in the flat regions. This normally would not be expected in non-
fatigue service.
There are several interesting approaches to examining fatigue failures
and their causes. Bhandarkar and Lisagor examined stress corrosion crack-
hag and fatigue as well as time independent fracture processes on four
aluminum alloys. What resulted is a systematic compilation of fractographic
features which serve as a basis for examining failures in these and similar
aluminum alloys.
Abelkis also studied fatigue in aluminum but concentrated on correlating
crack propagation under spectrum loading with fracture morphologies.
Excellent correlation between changes in loading schemes and striation
spacing are demonstrated. Eylon and Kerr analyzed fatigue failures by in-
vestigating the microstrueture of the origin sites using a precision section-
ing technique. This technique was extremely useful in determining that
nonmetallic inclusions were fatigue initiation sites in titanium alloy powder
compacts. In superalloy compacts, porosity was related to fatigue initiation.
Kramer found manganese-sulfide (MnS) inclusions to be the cause of a
fatigue failure in a steam turbine rotor. His analysis included an exhaustive
study of alternative mechanisms before concluding that it was a fatigue
failure, while Takada et al examined the effect of quantity and shape of
MnS inclusions on ductility of carbon-manganese (C-Mn) steel plates.
They determined that ductility was related to the inclusion area fraction on
a ductile fracture surface.
Joshi discussed, in general terms, the application of surface analysis
methods such as electron spectroscopy for chemical analysis (ESCA) and
Auger electron analysis in the determination of temper embrittlement,
grain boundary corrosion, and intergranular stress-corrosion cracking. The

ability to analyze several monolayers of a surface without the masking

effects of the bulk metal has already proved to be a powerful addition to
fractographic studies.
The effects of stress on material behavior were examined by various
methods. Most authors employed both microscopy and fracture mechanics.
Hicho and Gilmore found in comparing deembrittled and embrittled
21ACr-lMo steels that the fracture morphology of the deembrittled speci-
men was transgranular, while temper embrittlement resulted in an inter-
granular failure. Stress corrosion cracking tests showed that these em-
brittling effects result in a marked lowering of the threshold stress intensity.
Creep mechanisms in pressure vessels were discussed by Coleman. He
observed that intergranular separation initiated at a notch in the vessel
wall by means of classical grain boundary cavitation. This initiating
mechanism was followed by two shear mechanisms, one relatively ductile
and one with low ductility.
McCartney and Pellegrino investigated the relationship of strength,
toughness, and flaw tolerances in steel-lifting chains by means of fracture
mechanics. Their studies showed that higher strength chains were more
notch sensitive and were susceptible to brittle failure if they contained
flaws 1 in. deep.
Perhaps the newest area discussed at the symposium was the Mecholsky
et al paper on the fractographic analysis of ceramics. The introduction of
the concept of mirror constant and its relationship to fracture toughness
was discussed as well as the current theories of mirror formation. Several
examples of typical ceramic failures and their important features are dis-
cussed. We will look forward to additional work in this area in the future.

B. M. Strauss
Gulf Research and Development Company,
Pittsburgh, Pa. 15230, editor

W. H. Cullen, Jr.
U.S. Naval Research Laboratory, Washing-
ton, D.C. 20375, editor

Index
A Chemical analysis of fracture sur-
faces (see Auger electron
Acoustic emission, 299 ft.
spectroscopy, electron mi-
Alloy steels (see Steels, specific
croprobe analyzer)
types)
Chemical environments (see En-
Alpha-beta titanium alloys (see
vironments)
Titanium alloys)
Cleaning fracture surfaces, 50
Alpha titanium alloys (see Titanium
Cleavage facets, 116
alloys)
Cleavage fractures, 11, 51, 59, 110,
Aluminum alloys, specific types
128 ff., 320
2024-T4, 59, 176 ft.
Controlled fracture, 5 ft.
5456-H321, 64, 67
Corrosion-fatigue fractures, 144 ft.
6061, 176 ft.
Corrosion leaves, 195
7075, 176 ff., 218 ff.
Corrosion pits, 59
7178, 176 ft., 284 ff.
Corrosion products, 195
Auger electron spectroscopy, 64,
Corrosive environments, 288
275 ff.
Crack arrests 217, 224
Austenite, 110 ff.
Crack-growth rate, 73 ft., 99, 308
Austenitic stainless steels (see Steels,
Crack initiation, 150 ft., 168, 214,
specific types)
233, 235 ff., 250
Crack origins, 33 ft., 128 ft.,
B 235 ft., 363 ft.
Crack propagation, 151 ff., 214,
Bainite, 110 ft.
297 ft., 363 ft.
Beach marks, 213 ft.
Cracks, 364
Block loading, 213 ft.
Creep-fatigue interaction, 249 ft.,
Brittle striations, 135
297 ft.
Cryogenic-temperature fracture, 11,
C 21
Crystallographic orientation, 107 ft.
Carbide particles, 27-30
Carbon replicas, 130
D
Cathodic charging, 11 ft., 108 ft.
Ceramics, 363 ft. Dimples, 51, 153-162, 187, 198,
Chain links, 312 ft. 320
Charpy tests, 312, 337, 357 Ductile fractures, 88, 207, 320

387
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Copyright9 byASTMInternational [Link]
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
388 FRACTOGRAPHY IN FAILURE ANALYSIS

Ductility, 119, 207, 335 ft. Hydrogen embrittlement, 32 ff,

41-42, 87 ft., 107 ft., 257 ft.,
E 351 ff.,
Hydrogen-embrittlement cracking,
Electron microprobe analyzer, 50, 51, 55
64 Hydrogen-embrittlement fractures,
Elevated-temperature fractures, 44 11, 30, 87 ft., 351 ft.
Energy-dispersive spectrometers,
176 ft. I
Environments, 49 ft., 87 ft., 107
ft., 128 ft., 144 ft., 164 ft., Impurities, 352
351 ft. Inclusions, 238, 249 ft., 320, 335 ft.,
Etch pits, 108, 116-117, 164 ft. 364
Intergranular fractures, 20, 32 ft.,
F 44, 51, 88, 128 ff., 250, 283,
288, 302, 357
Fatigue crack growth rate, 73 ft., Interphase embrittlement, 285
213 ff., 365 Iron-nickel alloys (see Maraging
Fatigue fractures, 51 steels)
Fatigue striations, 55, 59, 73 ft.,
112, 128 ft., 153, 164 ft.,
L
172 ft., 194, 217 ft.
Fracture origins, 133, 363 Low alloy steels (see Steels, spe-
Fracture profiles, 133 cific types)
Fracture strain, 345, 363 Low carbon steels (see Steels, spe-
Fracture-surface matching, 59 cific types)
Fracture toughness, 353 ft. Low cycle fatigue fractures, 249 ft.
Low temperature fractures, 279 ft.
G
M
Glass, 363 ff.
Grain boundary fracture, 10 Machining marks, 364
Grain boundary cavitation, 306 ft. Macrofractography, 356
Maraging steel
It 18Ni, 87 ft.
Martensite, 21, 27, 94, 110
Hackel marks, 364 Metallography, 24-25, 50
Hardness, 315 Microcracks, 91
Heat affected zone, 110, 283, 297 Microplastic cracking, 58
ff. Microstructure, 5 ft., 102 ft.,
High cycle fatigue fractures, 236 176 ff.
High strength low alloy steels (see Microvoid coalescence fractures,
Steels, specific types) 51, 88, 91, 110 128 ft., 250,
High strength steel (see Steels, 308
specific types) Microvoids, 20, 201
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
INDEX 389

Mirror, 364 ff. Scratches, 131, 233

Mist, 364 ff. Sectioning, 236 ff.
Mud-crack pattern, 64 Secondary cracks, 112, 130, 170,
252, 260, 367
N Segregation, 200, 251 ft., 281
Shear fractures, 50, 243, 308 ff.
Nickel, superalloy Slant fractures, 297 ff.
AF-115, 240
Slip, 242
Nondestructive inspection, 214 Slip planes, 116, 173
Notched specimens, 110 ff., 236 ft.,
Spectrum fatigue loading, 213 ft.
320 Stainless steels (see Steels, specific
Notches, 233 types)
Steel plate, 5 ft.
O Steels, specific types
Oxide spikes, 249 ft. AMS 6407 (43305i), 51 ff.
Cr-Mo-V, 249 ff., 298
P HT80 (A, E), 107 ff.
Low carbon sheet, 5 ff.
Pits (see also Etch pits), 59 Mar-M-200, 236
Plastic-carbon replicas, 79 0.15C-1.5 Mn, 335 ff.
Plating, 7 2.25Cr-lMo, 283, 351 ft.
Pores, 240, 249, 364 [Link]-0.40Mn-2.40Si-l.15Ni-
Precipitate particles, 176 ft. 0.06Mg, 78 ft.
Pressure vessels, 297 [Link]-0.1V, 74 ft.
Primary cracks, 128 ff. 5Ni-Cr-Mo-V, 352
9Ni-4Co-0.20C, 32 ff.
Q 18Ni maraging, 87 ff.
300M, 63
Quantitative fractography, 300, 337
304 stainless, 144 ff., 164 ff.,
Quasicleavage fracture, 21, 88, 110
285
3340, 279
R Stress-corrosion cracking, 39-40,
Radial zone, 364 ft. 51, 129 ft., 144 ft., 275 ft.,
Replicas (see Carbon replicas) 289
Reversed-bending fractures, 144 ft. Stress-corrosion cracking fractures,
River patterns, 116, 153 352
Rock candy fractures, 153 Stresses, 73 ft., 176 ff., 285, 298
Rotor steel (see Steel, specific Stress-intensity factor, 73 ft.,
types) 128 ff., 147, 363 ft., 351 ff.
Rupture, grain boundary, 10, 20 Stress raisers, 367
Stretch zone, 97
Striations (see Fatigue striations)
S
Surface crack origins, 33, 28 ft.,
Satellite nucleation, 21 235 ff., 363 ft.
Copyright by ASTM Int'l (all rights reserved); Mon Dec 21 [Link] EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorize
390 FRACTOGRAPHY IN FAILURE ANALYSIS

T U
Tearing fracture, 118 Ultrasonic cleaning of fracture
Tear ridges, 110, 164 ft., 198 surfaces, 50
Temper embrittlement, 279 ff.,
351 ft.
Temperature, 87 ft., 285, 249 ft., V
284, 297 ff., 367
Tensile fractures, 249 ff. Voids, 297 ff., 308
Tension tests, 315, 337
Tire tracks, 59
Titanium, 235 ff. W
Titanium alloys, 55 ff.
Titanium alloys, specific types Wallner lines, 363
B-11 (powder compact), 238 ff. Welds, 32 ff., 55, 283, 297 ft.,
IMI-685, 246 312 ff.
Ti-4A1-4Mn, 59 Weld-crater cracking, 38-39
Ti-6A1-4V, 58, 236, 242-243
Ti-6A1-6V-2Sn (powder), 240
Ti-8AI-IMo-IV, 58, 128 ff. X
Ti-11, 236
Ti-17, 240 X-ray spectrometers, 44, 50, 176 ft.
Transgranular fractures, 44, 88,
128 ff., 260
Tungsten, 50 Z
Two-stage replicas (see also Carbon
replicas), 130 Zinc, liquid, embrittlement, 67

STP 600-1976
100% (1)
STP 600-1976
263 pages
Fractography of Ceramic and Metal Failure
100% (1)
Fractography of Ceramic and Metal Failure
415 pages
STP700 EB.14153 1 Fracture Mechanics... 1980
100% (2)
STP700 EB.14153 1 Fracture Mechanics... 1980
586 pages
ASTM - STP436 Electron Fractography
No ratings yet
ASTM - STP436 Electron Fractography
236 pages
SEM Techniques in Failure Analysis
No ratings yet
SEM Techniques in Failure Analysis
10 pages
Aerospace Fractography Guide
0% (1)
Aerospace Fractography Guide
17 pages
Stages of Failure Analysis Explained
No ratings yet
Stages of Failure Analysis Explained
12 pages
DS63 - (1985) References and Conference Proceedings Towards The Understanding of Fracture Mechanics
100% (1)
DS63 - (1985) References and Conference Proceedings Towards The Understanding of Fracture Mechanics
58 pages
ME4255 NUS Failure Analysis
No ratings yet
ME4255 NUS Failure Analysis
23 pages
STP 601-1976
100% (1)
STP 601-1976
593 pages
SEM Fractography of Nonmetals
100% (1)
SEM Fractography of Nonmetals
2 pages
09 Microscale Fracture Surface Morphologies
No ratings yet
09 Microscale Fracture Surface Morphologies
25 pages
Metallography Failure Analysis
100% (1)
Metallography Failure Analysis
18 pages
Fatigue and Fracture Mechanics, 34 TH Volume: Steven R. Daniewicz, James C. Newman, and Karl-Heinz Schwalbe, Editors
No ratings yet
Fatigue and Fracture Mechanics, 34 TH Volume: Steven R. Daniewicz, James C. Newman, and Karl-Heinz Schwalbe, Editors
599 pages
08 Macroscale Fracture Appearances
No ratings yet
08 Macroscale Fracture Appearances
18 pages
Chapter 10 Practical Fracture Mechanics in Design
100% (1)
Chapter 10 Practical Fracture Mechanics in Design
12 pages
ASTM STP 801 Corrosion Fatigue
No ratings yet
ASTM STP 801 Corrosion Fatigue
538 pages
Fractography - An Important Tool For Analyzing Metal Failurews
No ratings yet
Fractography - An Important Tool For Analyzing Metal Failurews
5 pages
Scanning Electron Microscope Fractography in Failure Analysis of Steels PDF
No ratings yet
Scanning Electron Microscope Fractography in Failure Analysis of Steels PDF
8 pages
Fractography of Ceramics and Glasses
No ratings yet
Fractography of Ceramics and Glasses
4 pages
ASTM STP 513 Stress Analysis and
No ratings yet
ASTM STP 513 Stress Analysis and
314 pages
PE429 - Lectures 3 & 4
No ratings yet
PE429 - Lectures 3 & 4
19 pages
07 Introduction To Types and Appearances of Fracture
No ratings yet
07 Introduction To Types and Appearances of Fracture
8 pages
N1 15-03 - 2022 Susane Sherrer ADM Guidance
No ratings yet
N1 15-03 - 2022 Susane Sherrer ADM Guidance
22 pages
STP 559-1974
No ratings yet
STP 559-1974
319 pages
Fractography of Metals and Plastics
No ratings yet
Fractography of Metals and Plastics
7 pages
Analyzing
No ratings yet
Analyzing
3 pages
Fractography of Metals and Plastics
No ratings yet
Fractography of Metals and Plastics
7 pages
Fractography of Glass 1994 PDF
100% (1)
Fractography of Glass 1994 PDF
305 pages
Failure Analysis of Engineering Materials
0% (1)
Failure Analysis of Engineering Materials
88 pages
Fracture Analysis
No ratings yet
Fracture Analysis
6 pages
02 Preface
No ratings yet
02 Preface
2 pages
Metallography in Failure Analysis PDF
83% (6)
Metallography in Failure Analysis PDF
300 pages
STP 560-1974
100% (1)
STP 560-1974
263 pages
STP 1296-1997
100% (2)
STP 1296-1997
640 pages
Lecture - Other Fracture Analysis Techniques
No ratings yet
Lecture - Other Fracture Analysis Techniques
11 pages
Failure Analysis and Fractography - Module 4
No ratings yet
Failure Analysis and Fractography - Module 4
7 pages
MTM Self Study
No ratings yet
MTM Self Study
1 page
CCT 3rd Cie - Merged
No ratings yet
CCT 3rd Cie - Merged
12 pages
Fracturesurfaceanalysis PDF
No ratings yet
Fracturesurfaceanalysis PDF
3 pages
Fractography: Observing, Measuring and Interpreting Fracture Structure Topography
No ratings yet
Fractography: Observing, Measuring and Interpreting Fracture Structure Topography
18 pages
Lecture 2 Alt
No ratings yet
Lecture 2 Alt
46 pages
Modern Analytical Techniques in Failure Analysis of Aerospace, Chemical, and Oil and Gas Industries
No ratings yet
Modern Analytical Techniques in Failure Analysis of Aerospace, Chemical, and Oil and Gas Industries
16 pages
Fractography for Engineers
No ratings yet
Fractography for Engineers
11 pages
Lecture 3
No ratings yet
Lecture 3
38 pages
Fractography - Wikipedia
No ratings yet
Fractography - Wikipedia
6 pages
STP 381-1965 (1981)
No ratings yet
STP 381-1965 (1981)
416 pages
MTS 226 Materials and Manufacturing Processes: Instructor: Dr. Hamid Jabbar Hamid - Jabbar@ceme - Nust.edu - PK
No ratings yet
MTS 226 Materials and Manufacturing Processes: Instructor: Dr. Hamid Jabbar Hamid - Jabbar@ceme - Nust.edu - PK
44 pages
Fractography: Analyzing Fracture Surfaces
No ratings yet
Fractography: Analyzing Fracture Surfaces
1 page
Frac To Graphy
No ratings yet
Frac To Graphy
639 pages
06 Fractography of Ancient Metallic Artifacts, and Restoration and Conservation Aspects
No ratings yet
06 Fractography of Ancient Metallic Artifacts, and Restoration and Conservation Aspects
16 pages
Mechanical Failure Analysis Techniques
No ratings yet
Mechanical Failure Analysis Techniques
77 pages
Reduction Gearbox Shaft Failure
No ratings yet
Reduction Gearbox Shaft Failure
6 pages
Engineering Failure Analysis
No ratings yet
Engineering Failure Analysis
7 pages
Garza Mendez 2007
No ratings yet
Garza Mendez 2007
4 pages
Failure Analysis: Al Emran Ismail, PHD
No ratings yet
Failure Analysis: Al Emran Ismail, PHD
12 pages
Job Title
No ratings yet
Job Title
6 pages
Audit System Improve Change 10
No ratings yet
Audit System Improve Change 10
53 pages
New Jobs
No ratings yet
New Jobs
27 pages
IELTS Listening Answer Sheet: Candidate Name Candidate No
No ratings yet
IELTS Listening Answer Sheet: Candidate Name Candidate No
1 page
IELTS Reading Answer Sheet: Academic General Training Day Month Year
No ratings yet
IELTS Reading Answer Sheet: Academic General Training Day Month Year
1 page
Scrum Cheat Sheet-Dr - Hosseini
No ratings yet
Scrum Cheat Sheet-Dr - Hosseini
1 page
Training
No ratings yet
Training
11 pages
A Nite Element Analysis of The Inelastic Relaxation of Thermal
No ratings yet
A Nite Element Analysis of The Inelastic Relaxation of Thermal
16 pages
KaiNexus Product Comparison Worksheet
No ratings yet
KaiNexus Product Comparison Worksheet
44 pages
Advanced Linear Scheduling Program With Varying Production Rates For Pipeline Construction Projects
No ratings yet
Advanced Linear Scheduling Program With Varying Production Rates For Pipeline Construction Projects
12 pages
Comparing Fatigue Corrosion IF Steel
No ratings yet
Comparing Fatigue Corrosion IF Steel
8 pages
1 Students Proof Schemes For Proving and Disproving of Propositions Final Paper
No ratings yet
1 Students Proof Schemes For Proving and Disproving of Propositions Final Paper
23 pages
Petrochemical Blast-Resistant Design
100% (1)
Petrochemical Blast-Resistant Design
7 pages
Understanding Motion Graphs Explained
No ratings yet
Understanding Motion Graphs Explained
11 pages
Bluehill Test Development Manual
No ratings yet
Bluehill Test Development Manual
58 pages
Motor's Bearing Details
No ratings yet
Motor's Bearing Details
9 pages
LuK TecBr WDC 210x297 LowRes EN
No ratings yet
LuK TecBr WDC 210x297 LowRes EN
28 pages
(Ebook PDF) Mechanical Behavior of Materials 2nd Edition Download
100% (6)
(Ebook PDF) Mechanical Behavior of Materials 2nd Edition Download
56 pages
CAD & CAM Exercises Guide
No ratings yet
CAD & CAM Exercises Guide
48 pages
At 171702
No ratings yet
At 171702
4 pages
GE LVMotor UltraVertical GEA30732 0
No ratings yet
GE LVMotor UltraVertical GEA30732 0
8 pages
dx4542 Partes
No ratings yet
dx4542 Partes
229 pages
Thermodynamics Mind Map
No ratings yet
Thermodynamics Mind Map
8 pages
Miniature Linear Motion Series - : L12 Specifications
No ratings yet
Miniature Linear Motion Series - : L12 Specifications
4 pages
SERVICE REPORT - 200CMD MBBR STP-GPS - THUMRAIT-signed
No ratings yet
SERVICE REPORT - 200CMD MBBR STP-GPS - THUMRAIT-signed
4 pages
Components and Installation Handbook Landirenzo Omega Gi
No ratings yet
Components and Installation Handbook Landirenzo Omega Gi
49 pages
Understanding Centripetal Acceleration Proof
No ratings yet
Understanding Centripetal Acceleration Proof
4 pages
Pneumatic Test Procedure Guide
No ratings yet
Pneumatic Test Procedure Guide
6 pages
Perbedaan Spec Hyundai STARGAZER
No ratings yet
Perbedaan Spec Hyundai STARGAZER
12 pages
4th Sem. / Chemical Engg (Pulp Paper Tech) Subject:Heat Transfer Section-D
No ratings yet
4th Sem. / Chemical Engg (Pulp Paper Tech) Subject:Heat Transfer Section-D
2 pages
L1706D Ii Driveline
No ratings yet
L1706D Ii Driveline
206 pages
4521 BR en
No ratings yet
4521 BR en
2 pages
Company Profile Pt. Dutaflow
No ratings yet
Company Profile Pt. Dutaflow
62 pages
Nptel For Beam & Shell PDF
No ratings yet
Nptel For Beam & Shell PDF
10 pages
044 - Mikropor-Mng-Bro-Us - 111021 Psa Nitrogeno
No ratings yet
044 - Mikropor-Mng-Bro-Us - 111021 Psa Nitrogeno
12 pages
Compact Style Precise Servo Technology Substantial Positioning Freedom Space Saving Construction
No ratings yet
Compact Style Precise Servo Technology Substantial Positioning Freedom Space Saving Construction
2 pages
Pressure Relief Valve Sizing Guide
No ratings yet
Pressure Relief Valve Sizing Guide
5 pages
Sewer Bypass Pumping
No ratings yet
Sewer Bypass Pumping
6 pages
Daily HSE Report for LPG Terminal Project
No ratings yet
Daily HSE Report for LPG Terminal Project
4 pages
Volvo BM L150 BML150 Wheel Loader Service Repair Manual Instant Download
No ratings yet
Volvo BM L150 BML150 Wheel Loader Service Repair Manual Instant Download
22 pages
150 KLD Structural Design 21.05 Calculations PDF
100% (1)
150 KLD Structural Design 21.05 Calculations PDF
16 pages
Seismic Design for Damping Systems
No ratings yet
Seismic Design for Damping Systems
54 pages