You are on page 1of 19

Engineering Fracture Mechanics Vol. 53, No. 6, pp.

829-847, 1996

Pergamon

0013-7944(95)00232-4

Copyright 1996 Elsevier Science Ltd


Printed in Great Britain. All rights reserved
0013-7944/96 $15.00 + 0.00

CYCLIC STRESS RESPONSE, STRAIN RESISTANCE AND


F R A C T U R E B E H A V I O R O F M O D I F I E D 1070 S T E E L
J. D. DOUGHERTY
Timken Research, The Timken Company Canton, Ohio 44706, U.S.A.
T. S. SRIVATSAN
Department of Mechanical Engineering, The University of Akron, Akron, Ohio 44325, U.S.A.
J. PADOVAN
Department of Mechanical and Polymer Engineering, The University of Akron, Akron,
Ohio 44325, U.S.A.
Abstract--A combined experimental and finite element study of fatigue crack closure in modified 1070
steel has been conducted. In this paper, the material property evaluations required for this study are
presented. The monotonic and cyclic stress-strain properties, cyclic stress response, cyclic strain resistance,
low cycle fatigue life and fracture behavior are examined. The low cycle fatigue tests were conducted using
tension-compression cycling, under total strain amplitude control, over a wide range of strain levels. The
material was found to possess medium strength and high ductility; while displaying a strain level dependent
combination of cyclic strain softening and hardening behavior. The observed softening behavior is
attributed to the rearrangement of dislocations produced by processing, formation of slip bands on the
specimen surface and the formation of microcracks. The observed hardening behavior is ascribed to
contributions from synergistic influences of dislocation multiplication, dislocation~lislocation interactions
and dislocation-microstructural feature interactions. The material followed the strain-life relationships
attributed to Basquin and Coltin-Manson. The fracture surfaces of the fatigue specimens showed distinct
regions of crack initiation, microscopic-macroscopic crack growth and sudden fracture. The low-cycle
fatigue characteristics and fracture behavior are discussed in the light of competing and mutually
interactive influences of cyclic strain amplitude, concomitant response stress, intrinsic microstructural
effects and dislocation-microstructure interactions during cyclic straining.

1. INTRODUCTION
THE SELECTIONOf a material for a structural component depends on many attributes, such as cost,
manufacturability, material properties and fatigue performance. With fatigue performance being
a combination of resistance to crack initiation, crack propagation behavior and fracture
characteristics of the material. Considerable research has shown that crack propagation is
significantly impacted by the crack closure behavior of the material. Crack closure has been
attributed to many factors, such as crack tip plasticity, oxide debris, crack plane surface roughness
and stress induced phase transformation [1]. Many experimental studies have been conducted to
quantify the effect of crack closure and several analytical or finite element modeling techniques have
been developed. However, few studies combining experimental and numerical tools have been
reported. A combined experimental and finite element study of fatigue crack closure in modified
1070 steel has recently been completed [2]. In this paper, the monotonic and cyclic stress-strain
properties, cyclic stress response characteristics, low-cycle fatigue behavior and fracture behavior
of this material are examined. Subsequent papers will present the experimental and finite element
aspects of this fatigue crack closure study.
Many engineering components demand that the competing materials offer high strength and
wear resistance properties in certain areas, while providing high ductility and fracture toughness
properties in other areas. A typical example, associated with the automotive industry, is a wheel
hearing package with integral mounting flange to facilitate bolting the package to the steering
knuckle [3]. High strength and wear resistance properties are required in the raceway areas, while
high ductility and fracture toughness properties are essential for the flange portion. This can readily
be accomplished by having the flanged component made from a forging using steel having
sufficiently high hardenability. Typically, the forging is heat treated to obtain the desired bulk
properties and then a selective hardening process, such as induction hardening, is used to obtain
829

830

J.D. DOUGHERTYet al.

the desired properties in the raceway areas. For this type of application, one potential solution is
to use forgings of 1070M steel. This modified 1070 steel has its alloy content adjusted to improve
hardenability. The heat treatment selected for this study is a quench and temper process, which
results in a fine grained microstructure consisting of tempered martensite with slight
spheroidization of the second phase carbides present in the microstructure. The aim hardness
selected was 97 HRB (16 HRC), which provides medium strength while providing good
machinability properties. This microstructure provides the key properties for the flange portion of
the bearing package.
The preponderance of experimental research effort on fatigue behavior has concentrated on
advanced materials. Relatively few studies have been conducted to comprehensively quantify and
characterize the cyclic stress response, strain resistance and fracture behavior of new steel types.
Recent pressure to improve fuel economy in the automotive industry makes improvements in the
fatigue performance, of selected steel types, essential to enabling weight minimization of high
production volume structural components.
The effect of martensite content and morphology on the fatigue behavior of low carbon steel
has been studied by Mediratta et al. [4,5]. The dual-phased (ferritic-martensitic) microsructure
exhibited cyclic strain hardening during fully reversed strain controlled cyclic deformation, with
the degree of hardening decreasing with increasing martensite content. Increasing the martensite
content was also found to improve the fatigue properties. The effect of heat treatment on the
microstructure and cyclic fatigue behavior of 8620 steel was examined by Yu et al. [6]. A quench
and tempered martensitic microstructure was found to increase strength without any decrease in
ductility compared to an as-received hot-rolled ferritic microstructure. Both of these
microstructures exhibited cyclic strain softening during fully reversed strain controlled cyclic
deformation. The microstructure resulting from the quench and temper process exhibited superior
fatigue properties.
Studies of combined ferritic-bainitic microstructures in 1Cr-Mo-V steel by Bhambri et al. [7]
and 30Cr2MoV steel by Yimin and Jinrui [8] revealed these two microstructures to exhibit strain
softening behavior during fully reversed strain controlled cyclic deformation. A comparative study
of CrMoV steel in as-received (ferritic), and quenched and tempered (martensitic) microstructures
was conducted by Singh et al. [9]. The martensitic microstructure offered a higher strength but
lower ductility material condition, which exhibited a cyclic strain softening behavior during fully
reversed strain controlled cyclic deformation. While the ferritic microstructure initially exhibited
softening followed by hardening prior to failure.
A comparison of microalloyed steels, having a dual-phase pearlitic and ferritic microstructure,
and quench and tempered steels, having a tempered martensitic microstructure, was conducted by
Farsetti and Blarasin [10]. Tests conducted under fully reversed strain controlled cyclic deformation
revealed that the microalloyed steels tended to harden while the quench and tempered steels tended
to soften. The martensitic microstructure, of the quench and tempered steel, offered higher
monotonic yield strength while cyclic yield strength values of these two microstructures were
similar. A study of the effect of mean strain (and stress) on the low-cycle fatigue behavior of
ASTM A723 quench and tempered steel, with a martensitic microstructure, was conducted by
Koh and Stephens [11]. All of these tests, conducted under strain control at strain ratios
of R = / ~ m i n / E m a x ~--" - - 2, - 1, 0, 0.5 and 0.75 showed evidence of cyclic strain softening, Furthermore,
mean stress relaxation was observed during all tests except those conducted at the lowest strain
ranges and corresponding lower response stress. A study of the effect of cold rolling on the
monotonic and cyclic stress-strain and low-cycle fatigue behavior of 1010 steel was conducted by
Yu et al. [12]. In the as-received condition, the material showed evidence of strain hardening, while
the cold-rolled counterpart exhibited strain softening during fully reversed strain controlled cyclic
deformation. The monotonic yield strength and the degree of cyclic strain softening were both
found to increase with increased amount of cold working.
Plain carbon low alloy steels (types: 1010, 1045, and 1080) were compared under fully reversed
stress controlled conditions, rather than strain control, by Eifler and Macherauch [13]. A
normalizing heat treatment sequence resulting in a dual-phased ferritic and pearlitic microstructure
was used for all steel types. The 1010 steel, at low stress amplitudes, exhibits gradual changes of
initial softening followed by hardening prior to specimen failure. However, tests at higher stress

Modified 1070steel

831

amplitudes revealed rapid initial softening followed by hardening and then gradual softening to
failure. For the 1045 steel, tests at all stress amplitude levels exhibited a more gradual softening
followed by hardening to failure. For the 1080 steel, the behavior was similar to that of the 1045
steel except the relative amount of hardening was considerably reduced. The general trend shown
by these steels was that the extent of hardening, following the initial softening behavior, decreased
with an increase in carbon content.
Microstructural influences on the fatigue behavior of SAE 1080 steel was investigated by
Muralidharan and Fuquen [14]. A normalized heat treatment process, resulting in a fully pearlitic
microstructure, and a quench and temper process, which resulted in a tempered martensitic
microstructure, were examined. Both heat treatment processes resulted in a hardness range of 21-23
HRC. Both microstructural conditions promoted softening behavior during fully reversed strain
controlled cyclic deformation. The tempered martensitic microstructure had a much higher
monotonic yield strength, but exhibited a greater degree of cyclic strain softening. This resulted
in both microstructures having nearly identical cyclic yield strengths.
A detailed analysis of these research studies performed to characterize the cyclic stress response
characteristics for the families of carbon and low alloy steels reveals the martensitic microstructures
to promote cyclic strain softening. Decreasing the carbon content tended to promote cyclic strain
hardening, while increasing the carbon content tended to promote cyclic strain softening.
Increasing the monotonic yield strength of the microstructure tended to enhance the degree of
softening during cyclic deformation. Also, the nature and extent of mechanical deformation the
material was subjected to during processing had a significant impact on cyclic stress response.
The objective of this paper is to systematically document microstructural influences on the
low-cycle fatigue properties and fracture behavior of a modified 1070 steel. Tests were conducted
under fully reversed strain controlled cyclic deformation over a wide range of total strain
amplitudes. The cyclic stress response characteristics, fatigue life and fracture behavior are
discussed in terms of the specific roles played by the concurrent and mutually interactive influences
of cyclic strain amplitude, concomitant response stress, intrinsic microstructural effects and
macroscopic aspects of fracture.

2. MATERIAL
The material selected for investigation in this study is 1070M, where the " M " denotes
modified. The chemical composition of this steel, in weight percent, is presented in Table 1. The
steel was produced and made available by The Timken Company (Canton, Ohio, U.S.A.). The
material was processed to obtain a microstructure which offered the high ductility and fracture
toughness properties essential for the bulk forging of the aforementioned flanged component. To
more precisely simulate the extent of mechanical deformation a forged component would receive,
the test specimens were machined from hot forged blanks. Forging preforms were machined from
63.5 mm diameter hot rolled solid rounds; such that, the centerline of the test specimens was
parallel to the rolling direction. Machined forging preforms were heated to 1218C, hot forged to
a 50% reduction in height and then air cooled to ambient temperature.
The forged blanks were then heat treated, using a quench and temper process, to obtain the
desired microstructure, grain size and hardness level. The quench and temper process consisted of
(a) heating to 843C for 1 hr, (b) quenching in oil and (c) tempering at 709C for three hrs. This
heat treatment sequence resulted in a fine grained tempered martensitic microstructure with slight
spheroidization of the carbides present in the microstructure. The grain size ranged from 0.011 to
0.022 mm and the resulting hardness ranged from 96 to 98 HRB (15.7 to 17.0 HRC). An optical
micrograph showing the resulting microstructure of this material is provided in Fig. l.

C
0.680

Table I. Chemicalcompositionof the 1070M steel


Mn
P
S
Si
Cr
Ni
Mo
Cu
A1
Sn
Ti
0.950 0.009 0.022 0.170 0.130 0.110 0.050 0.190 0.035 0.009 0.001

Fe
Balance

832

J . D . DOUGHERTY et al.

3. EXPERIMENTAL TECHNIQUES
3.1. Specimen description and preparation
The test specimens were machined from the hot forged and heat treated blanks. The tensile,
test specimens conformed to specifications in ASTM E8-90a and the low-cycle fatigue test
specimens conformed to specifications in ASTM E606-80. The specimen gage length used was
25.4 mm for the tensile tests and 12.7 mm for the low-cycle fatigue tests. All test specimens were
finish machined to the required surface finish and the low-cycle fatigue test specimens were given
a final grind and polish process to minimize surface irregularities and circumferential machining
marks.
3.2. Mechanical testing
Monotonic tensile tests were performed in accordance with ASTM E8-90a standards. The
tests were conducted using a fully automated closed loop servohydraulic structural test
machine (Instron). The test conditions were an ambient temperature of 23C and a laboratory air
environment with a 50% relative humidity. The test specimens were deformed to failure using
a constant cross head speed of 2.54 mm per minute. The controller had a print out facility
which enabled the load and displacement measurements, parallel to the load line, to be output
at periodic intervals and concurrently recorded using a X-Y recorder equipped with a pen
plotter.
The low-cycle fatigue tests were performed in accordance with ASTM E606-80 standards.
The load frame used was a MTS material test system with a 244.64 kN capacity. This fully
automated closed loop servo-hydraulic test machine utilizes hydraulic collets to hold the
specimens, thus eliminating thread backlash during tension-compression cycling. All tests were
conducted under total strain amplitude (AE,/2) control using a sinusoidal waveform and a
constant cyclic frequency of 0.2 Hz. All tests were initiated in tension and were conducted using
a strain ratio of R = ~min/Emax = - - 1 . The total strain amplitude control was achieved by
controlling the displacement along the specimen gage length using a 12.7mm gage length
extensometer. The stress-strain hysteresis loops were recorded on a X-Y recorder equipped with
a pen plotter. A PC-based data acquisition system was used to store test data for analysis and
interpretation.
The technique chosen for evaluating the cyclic stress-strain behavior of the 1070M steel
was the Companion Specimen Test (CST) method. In this technique, the specimens are
deformed at a constant strain amplitude until either (1) the test is suspended when the
stress-strain hysteresis loops reach stabilization or (2) failure of the specimen occurs. An intrinsic
advantage of this technique is a total elimination of strain history effects; thereby, facilitating
a comparison of test data. For all tests reported here, the CST tests were conducted to specimen
failure with the cyclic stress and strain values being taken at specimen half-life (Nf/2). The
low-cycle fatigue behavior is readily obtained from these CST tests conducted to specimen
failure. More complete details on the experimental procedures and data analysis are available
elsewhere [2].
3.3 Failure and damage analysis
Fracture surfaces of the monotonic tensile test and low-cycle fatigue test specimen were
examined in a scanning electron microscope (SEM) to (1) determine the macroscopic fracture
mode and (2) characterize the fine-scale topography and microscopic mechanisms governing crack
initiation, fatigue crack growth and fracture behavior. The distinction between macroscopic and
microscopic fatigue and fracture mechanisms is based on the magnification level at which the
observations are made. Of particular interest for the fatigue crack closure aspects of this study,
was an examination of the fracture surface for fatigue striations. This was conducted to determine
if this material is appropriate for using the existing experimental methods, which use striation
spacing to quantify crack closure behavior [15,16].

Modified 1070 steel

Fig. 1. Microstructure of the 1070M steel.

833

834

J.D. DOUGHERTY et al.

Fig. 2. SEM micrographs of the tensile fracture surface showing: (a) distribution of microvoids,
(b) microvoids of a range of sizes, (c) secondary cracking.

Modified 1070 steel

835

Table 2. Monotonic tensile test results for 1070M steel


Yield strength
MPa
ksi

Ultimate strength
MPa
ksi

589.0

721.0

85.5

104.6

Elongation
to failure
~r(%)

Reduction
in area
RA(%)

25.9

59.7

4. RESULTS AND DISCUSSION


4.1. Tensile behavior

The tensile properties, of the high carbon quenched and tempered 1070M steel, are
summarized in Table 2. The results reported are the mean values based on multiple (three) tests.
The yield strength (Sy0, defined as the stress corresponding to a plastic strain of 0.2% is 589 MPa
(85.5 ksi). The ultimate strength (S,,) is 720 MPa (104.6 ksi). The large difference between the yield
strength and ultimate strength indicates that this material exhibits a significant amount of work
hardening during monotonic deformation. The elongation to failure and reduction in area were
25.9% and 59.7%, respectively. These tensile properties reveal the quenched and tempered 1070M
steel to have medium strength and high ductility. The material exhibited a negative stress-strain
slope immediately after the onset of yielding. This phenomenon, often referred to as pop-in, results
in an abrupt discontinuity in the stress-strain curve which occurs over a very narrow strain range.
Scanning electron microscopy of the tensile test fracture surfaces revealed that the fracture
occurred by the characteristic cup and cone type of separation. The fracture behavior was found
to be consistent with failure that occurs by microvoid coalescence. The central regions of the tensile
fracture surface, where the fracture initiated, showed a rough texture while the surrounding shear
lips were relatively smooth and well defined. Fracture was predominantly ductile and the surface
topography was found to consist of a bimodal distribution of dimples and microvoids [Fig. 2(a)].
Examination of the tensile fracture surface, at higher magnifications, revealed the existence of a
population of voids of a wide range of sizes [Fig. 2(b)]. The voids were homogeneously distributed
throughout the tensile fracture surface. Although the exact nucleation site of these voids is difficult
to pin-point, their near-equiaxed shape suggests that they nucleated around the carbide particles
and inclusions in the microstructure, during the later stages of deformation, without significant
growth and the influence of shear stresses that would result in their ovalization. The shear lips were
comprised of elongated dimples. The fine microscopic voids grow and coalesce until the residual
strength of the microstructure is not sufficient to withstand the applied combination of stress and
strain, resulting in sudden fracture (overload). The central region of the fracture surface also
exhibited secondary cracking behavior [Fig. 2(c)].
4.2. Cumulative glide and cyclic stress response
The variation of stress response with fatigue cycles and cyclic strain amplitude is an important
feature of the strain amplitude controlled low-cycle fatigue process. The cyclic stress response curve,
determined by monitoring the cyclic stress amplitudes (Aa/2) during fully reversed total strain
amplitude controlled fatigue, provides useful information pertaining to the cyclic and overall
mechanical stability of the material. Mechanical stability of the intrinsic microstructural features
coupled with an intrinsic ability of the microstructure to distribute the plastic strain over the entire
microstructural volume are two key factors controlling the cyclic stress response, cyclic strain
resistance and overall stability of the material [17,18],
The cyclic stress response curve, or cumulative glide plot, of cyclic stress amplitude vs number
of cycles illustrates the path by which the material arrives at the final level of stress corresponding
to specimen failure. The stress response curves, over a range of cyclic strain amplitudes, are shown
in Fig. 3. For the purposes of clarity, the data from the low-cycle fatigue tests are presented in
two groups: (a) low strain range and (b) high strain range. The cyclic stress response is dependent
on strain level and can be characterized as follows:
(1) At the lowest total strain amplitude tested (AEt/2 = 0.21%), the material response was
initially purely elastic with an increasing cyclic stress response. This initial behavior is attributed
to increasing dislocation density within the microstructure. After about 60 cycles, plastic strains
developed and cyclic strain softening began to occur.

836

J . D . D O U G H E R T Y et al.

(2) At low strain amplitudes (AEt/2 < 0.50%), the microstructure showed evidence of
progressive softening to failure.
(3) At intermediate cyclic strain amplitudes (0.50% < AE,/2 < 1.37%), the microstructure
showed evidence of softening followed by hardening just prior to specimen failure.
(4) At higher cyclic strain amplitudes (Act > 1.37%), the microstructure revealed hardening
during the first few cycles, followed by cyclic gradual softening and cyclic stability for the majority
of fatigue life, and culminating in progressive hardening to failure.
Since low-cycle fatigue life is strongly dependent on cyclic strain amplitude, a fatigue life scaled
comparison of the cyclic stress response curves is made in Fig. 4. This figure shows the cyclic stress
response curves, for all strain amplitudes, as a function of percent life (N/Nf expressed as a

iii

175, I

!,00t

Ej

, 050,

~
2

Cycles (N)

,oo

600~

z.3v~

b) high strain range

||

10

n,=|

,|

100
1000
Cydes (N)

g,

10000

Fig. 3. Cyclic stress response curves for the 1070M steel.

Modified 1070 steel

837

ACt/2

500.

0.75~

"~ 400.

O.50t

raO

0.37%

300'
a) low strain

O. 2 1 %

200 ,' ' ' ' ' 1 ' ' ' ' 1 ' ' ' ' 1 ' ' ' ' !

....

PercentLife N/Nf(%)
800-j-

--

Act/2 = 2.40~

'I

600

Z.37~

5 0 0 ~ . ~ ~ .

1.00,

b) high strain range


400 tJ . . . . , . . . . I . . . . I . . . . i . . . . ,

PercentLife N/Nf(%)
Fig. 4. Cyclic stress response vs percent life.

percentage). For the low strain amplitudes (AE,/2 _< 0.50%) and intermediate (0.50% < AE,/2 <_
1.37%) strain amplitudes, the stress response vs percent life is characterized by:
(1) Rapid softening occurs during the first 5-10 percent of fatigue life.
(2) A nearly stabilized stress response occurs for the majority of fatigue life, starting near 10%
and culminating at up to 80% of fatigue life.
(3) The low strain amplitude tests continued to exhibit softening until specimen failure;
whereas, tests conducted at intermediate strain amplitudes exhibited hardening prior to specimen
failure.
At the higher strain amplitudes (AE,/2 > 1.37%), the rapid initial hardening was followed by
a nearly stable cyclic response for most of the fatigue life and a significant amount of hardening
was exhibited prior to specimen failure.
In Fig. 5, the cyclic stress response data is replotted as normalized stress (aN/or,) vs the number
of cycles, where stress at any cycle (aN) is normalized with respect to the stress response during

838

J . D . D O U G H E R T Y et al.

the first cycle (a,). This type of plot enables a more direct comparison of the relative magnitude
of softening and hardening which occurs at the different strain amplitudes. This plot confirms the
observations of: (a) significant amounts of softening occur at low strain amplitudes, (b) the degree
of softening decreases as the strain amplitude increases and (c) the degree of hardening prior to
specimen failure increases as strain amplitude increases.
The cyclic behavior of the 1070M steel exhibited isotropic behavior during fully reversed
tension-compression strain cycling. Plots of the tensile and compressive stress response vs the
number of cycles are plotted in Fig. 6 for total strain amplitudes of 0.50% and 1.37%. For the
lower strain amplitude, the tensile and compressive stresses were nearly identical during the initial
stages of fatigue life. The tensile stress becomes progressively less than the compressive stress during
the later stages of fatigue life. This behavior is consistent with crack growth and crack opening
reducing the specimen load only during the tensile portion of the fully reversed strain cycle. For
the higher strain amplitude, crack initiation is expected within the first few cycles and the tensile
stress remains slightly below the compressive stress over the preponderance of fatigue life.

I
0"6t a)Iowstrainrange'37~ o.!a,I
0.5

. .....,

. .....

......

....

-,

.....

-I

.....

-I

Cycles (N)
1.3

zl.2

AEt/2

2.40~

]
1.37~

!*

i.oo~

Z
0.9

b) highstrainrange

0.8

........ ,

I0

........ ,

........ ,

I00
I000
Cycles (N)

........
I0( bOO

Fig. 5. Normalized cyclic stress response curves.

Modified 1070 steel

839

700
Tension
o

600-

Compression

"" 5 0 0 .
400-

300

""I

"

'

'''"'I

'

'

''''"I

1-4

v-t

Cycles (N)
700

600.

"-" 5 0 0 .

Tension
- e . -

COml~'euion

400b)
300

A(t/2

........
1

n
10

= 1.37%
........

,
100

.......
1000

Cydes(N)
Fig. 6. Tensile and compressive stress response curves.

The progressive decrease in cyclic stress response (softening) observed at the lower strain
amplitudes and resultant enhanced fatigue life is ascribed to the concurrent and mutually
competitive influences of:
(1) Destruction of microstructural contributions to strength through dislocation-microstructural feature interactions.
(2) Growth and coalescence of the microscopic cracks to form one or more macroscopic cracks
and resultant growth of both the microscopic and macroscopic cracks.
The coalescence or linking of the microscopic cracks is a process which normally occurs at
the surface. Initially, the softening rate is large due to the microscopic cracks linking up at a large
number of sites. The sites at which the linking up occurs decrease exponentially as cycling
progresses with a simultaneous reduction in the softening rate. Initially, the microcracks which
form in the favorably oriented slip bands grow and coalesce. At the beginning of cyclic straining,
these points of favorable orientation are numerous and hence the rate of softening is high. As the
points of favorable orientation decrease, so does the softening rate. This process takes the material
to the point of saturation; wherein, no change occurs in both the size and shape of the stress-strain
hysteresis loop.
The initial hardening observed during the first few cycles, at the higher cyclic strain amplitudes,

840

J.D. DOUGHERTYet al.

results from a rapid increase in dislocation density and is aided by the conjoint influences of
dislocation~lislocation interaction, dislocation multiplication and dislocation-microstructural
interactions. Once the stress response has reached the peak of initial hardening there occurs a
dislocation configuration which tries to change upon further cyclic straining. The onset of this
dislocation rearrangement process provides an appealing rationale for the start of either stability
or softening. Another plausible explanation for the softening is based on the formation of slip
bands. At a given cyclic strain amplitude, the amount of slip band formation reaches saturation
at the peak of initial hardening. From this point onward, no new slip bands form on the surface
during continued cyclic straining. Further cyclic deformation produces softening through an
intensification of the slip bands. In the absence of local stress and strain concentrators, these
intensified slip bands are the most likely sites for crack initiation.
The theory behind the dependency of strength on grain size suggests that grain boundaries
act as obstacles to dislocation motion and hence create dislocation pile-ups at these boundaries.
For this 1070M steel having a fine-grained microstructure, the dislocation pile-ups occur at a
greater number of sites. Furthermore, the dislocations have a smaller distance to glide before they
encounter other obstacles. An increased dislocation density in the microstructure, due to continued
cyclic straining, results in dislocation concentration at locations of carbide particles or inclusions,
grain boundaries and grain boundary triple junctions. This results in high local stress concentration
in these regions. The high stress concentration facilitates the initiation of microcracks at grain
boundaries, interfaces between second phase particles and metal matrix, and cracking through the
second phase particles (carbides or inclusions). Sudden fracture (overload) of the fatigue specimens
is rationalized as being due to the concurrent and competing influences of:
(1) Matrix stress in excess of the residual tensile strength of the material.
(2) Rapid growth and coalescence of the microscopic cracks results in a reduction of the area
of uncracked matrix below that needed to sustain the applied strain and stress amplitudes.
4.3. Cyclic strain resistance: low-cycle .fatigue
The total strain amplitude (AE,/2) is the sum of the elastic strain amplitude (&o/2) and plastic
strain amplitude (AEp/2).
AEt/2 = AE~/2 + aEp/2.

(1)

Both the elastic and plastic strain amplitudes are measurable quantities which can be used to
develop empirical equations relating strain amplitude to the reversals to failure (2Nr). An empirical
relationship between the elastic strain amplitude and the reversals to failure is given by the Basquin
relationship:

&o/2 = (a~/E)*(2Nf)L

(2)

This relationship, which uses the modulus of elasticity (E = 207 GPa) and two empirical
parameters [fatigue strength coefficient, (rrf') and fatigue strength exponent (b)], dominates
strain-life behavior for the high cycle region where plastic strains are minimal.
An empirical relationship between the plastic strain amplitude and the reversals to failure is
given by the Coffin-Manson relationship:
AEp/2 = E~*(Nr)'.

(3)

This relationship, which uses two empirical parameters [fatigue ductility coefficient, (of') and
fatigue ductility exponent (c)], dominates strain-life behavior for the low cycle region where plastic
strains are significantly higher than elastic strains. Combining the above equations provides a
relationship between total strain amplitude and the reversals to failure.
The results of the low-cycle fatigue tests are summarized in Table 3. The plastic strain
amplitudes and stress amplitudes are provided for the first cycle and the specimen half-life (Nr/2).
Using least squares regression for the half-life elastic and plastic strain amplitudes, the empirical
parameters of eqs (2) and (3) were estimated. Based on these tests, the resulting low cycle fatigue
strain-life relationship for the 1070M steel studied is given by the relationship:
A,/2 = 0.0045"(2"Nf) - 0.085+ 0.18*(2*Nr) - 0.,s

(4)

Modified 1070 steel

841

Table 3. Results of low cycle fatigue tests


Plastic strain
amplitude
Stress amplitude
N= 1
N = Nr/2
N 1
Nr/2
0.00000
0.00046
443.7
273.2
0.00113
0.00179
539.9
364.3
0.00239
0.00304
540.7
404.7
0.00481
0.00499
550.5
464.9
0.00747
0.00739
540.2
509.1
0.01083
0.01100
592.8
559.5
0.01470
0.01464
580.0
591.9
0.02110
0.02097
599.5
626.9

Total strain
amplitude
0.00212
0.00374
0.00500
0.00747
0.01000
0.01370
0.01750
0.02400

Cycles to
failure
Nr

205 885
10 305
5 087
1 392
790
401
144
38

The test data and resulting empirical relationships are plotted in Fig. 7. The test data points
are represented by symbols and the best fit curves are plotted as solid lines. The values o f the
low-cycle fatigue empirical p a r a m e t e r s that satisfy the Basquin and C o f f i n - M a n s o n relationships
are summarized in Table 4. The value of c, the fatigue ductility exponent, is - 0.45 which is within
the generally observed range ( - 0 . 7 - 0.2) for a large n u m b e r o f monolithic alloys and their
composite counterparts [19,20]. The fatigue ductility coefficient (Er') is 18%, which does not accord
well with the m o n o t o n i c ductility as measured by the elongation to failure (Er) of 25.9%. The
inferior cyclic ductility is ascribed to the local stress (and strain) concentrations and deformation
characteristics o f the microstructure features (grain boundaries and second phase particles) during
fully reversed cyclic straining.
4.4. Cyclic s t r e s s - s t r a i n response
Cyclic stress-strain response is an i m p o r t a n t source o f information required for designing and
developing materials having enhanced cyclic strain resistance. This response, which describes the
relationship between cyclic flow stress and cyclic strain amplitude, is a useful aid in understanding
the strain amplitude controlled low-cycle fatigue process [21]. The m o n o t o n i c and cyclic true

0.1

0.01

Act/2

0.001

0.0001

.......

'=

.......

......

'l

......

......

,=

s,-

"-

R e v e r s a l s to F a i l u r e

'n

(2Nf)

Fig. 7. Strain amplitude-fatigue fife response.

k(MPa)
1121

n
0.15

Table 4. Monotonic and cycle parameters


k'(MPa)
n'
~//E
b
1831
0.26
0.0045
- 0.0085

Er'

0.18

- 0.45

842

J.D. DOUGHERTYet

al.

oo],
oo]
oo] / /
4oo //

~400"

~ 300,

L_LLS_L_
0

0.005

0.01 0.015 0.02 0.025


Strain Amplitude
Fig. 8. Monotonic and cyclic stress-strain curves.

stress-true strain curves are compared in Fig. 8. This curve was constructed, from the companion
specimen tests (CST) conducted to failure, by taking the true stress and true strain values at
specimen half-life. This figure shows that the cyclic stress at half-life is less than the corresponding
monotonic stress and the material is softer in the cyclic state. The amount of softening is greater
at lower strain amplitudes and progressively decreases with an increase in strain amplitude. At
strain amplitudes greater than 1.75%, the material is harder in the cyclic state than the monotonic
state.
Another means of comparing the monotonic and cyclic true stress-true strain behavior is
obtained by using the following power law equations [22].
a = k*(e0" (for monotonic tensile test)

(5)

Atr/2 = k'*(Aep/2) "~ (for low-cycle fatigue test).

(6)

For monotonic tests, eq. (5) uses two empirical parameters (k, the monotonic strength
coefficient; and n, the monotonic strain hardening exponent), which are determined by a least
squares regression from the data for one tensile test data. For fully reversed cyclic strain tests,
eq. (6) uses two empirical parameters (k', the cyclic strength coefficient; and n', the cyclic strain
hardening exponent), for which the regression is performed using data taken from half-life data
of the CST tests conducted to failure at different strain amplitudes. A summary of the monotonic
and cyclic empirical parameters, obtained from the test data of this study, is provided in Table 4.
The results for the monotonic strength coefficient (k) was 1121 MPa and the monotonic strain
hardening exponent (n) was 0.15, while the results for the cyclic strength coefficient (k') was
1831 MPa and the cyclic strain hardening exponent was 0.26. The cyclic strain hardening exponent
(n' =0.26) is substantially higher than the monotonic strain hardening exponent (n =0.15). This
indicates a much steeper cyclic stress-strain slope, which is clearly seen in Fig. 8.
In terms of plasticity induced crack closure, the difference between the monotonic and cyclic
stress-strain behavior is extremely important due to the impact on crack tip plastic zone size. Using
Irwin's [23] plane stress equation, for estimating the plastic zone size (ry) resulting from a given
stress intensity (K) and material yield strength (Sy,):
1
ry -

2re * (K/Syt)2"

(7)

A comparison reveals that the estimated plastic zone size based on the cyclic properties will
be approximately twice that obtained by using the monotonic properties.

Modified 1070 steel

Fig. 9. SEM micrographs of low-cycle fatigue fracture surface total strain amplitude of 0.75% (Nr = 1392
cycles) showing: (a) overall morphology of fracture surface (b) region of crack initiation (c) secondary
cracks in transition region (d) voids and dimples in overload region.
Fig. 10. SEM micrographs of low-cycle fatigue fracture surface total strain amplitude of 0.75% (Nr = 1392
cycles) showing: (a) transgranular striation-like markings (b) high magnification of (a).

843

844

J.D. DOUGHERTY et al.

Fig. 1I. SEM micrographs of low-cycle fatigue fracture surface total strain amplitude of 1.37% (NL,= 401
cycles) showing: (a) secondary cracks in transition region (b) transgranular striation-like markings
(c) grain boundary void coalescence (d) striation-like markings near void.

Modified 1070 steel

845

4.5. Cyclic fracture behavior


Examination of the fracture surface features of the deformed low-cycle fatigue specimens,
in a scanning electron microscope (SEM), was done at: (a) low magnifications to identify
the regions of crack initiation, crack propagation and sudden fracture (overload); and (b) higher
magnifications to identify the locations of microcrack initiation, nature of early microscopic crack
growth, microstructure influences on crack growth and fine-scaled features of the overload
region.
SEM observation of the low-cycle fatigue fracture surfaces revealed nearly similar
topographies at high cyclic strain amplitudes, with resultant short low-cycle fatigue life, and at
lower cyclic strain amplitudes, with concomitant enhanced low-cycle fatigue life. Representative
micrographs of the low-cycle fatigue fracture surface features are shown in Figs 9-11. In all of these
micrographs, the specimen and microgragh are oriented so the crack is propagating from left to
right.
At a low cyclic total strain amplitude (AE,/2 = 0.75%) and resultant fatigue life (Nf -- 1392
cycles), the crack initiated at the surface and propagated inward, leaving a distinct region of crack
propagation prior to specimen fracture which appeared semi-circular in shape [Fig. 9(a)]. The
microcracks initiated at the specimen surface, coalesced to form a macrocrack, and propagated
approximately halfway through the specimen prior to failure. Higher magnification observations
of the region of crack initiation [Fig. 9(b)] showed distinct evidence of crystallographic (stage I)
crack growth prior to propagation of the crack along the plane normal to the stress axis (stage
I1). The region of transition between slow and stable crack propagation and sudden fracture
(overload) revealed significant amounts of secondary cracking [Fig. 9(c)] in a direction parallel to
the major stress axis. High magnification observation of the overload region [Fig. 9(d)] revealed
features reminiscent of ductile fracture: (a) microscopic voids of a variety of sizes, (b) evidence of
microvoid coalescence and (c) dimpled rupture.
High magnification observations of the region of slow and stable crack growth revealed a near
featureless morphology with an absence of well defined fatigue striations (Fig. 10). Certain regions
of transgranular fracture showed evidence of extensive deformation with the presence of extrusions
and fine striation-like features. These regions were isolated and irregularly distributed throughout
the microstructure. The presence of striation-like markings was more frequent at or near the sites
of second phase carbide particles or inclusions in the microstructure.
At the higher strain amplitude (AEt/2 = 1.37%) and resultant short low-cycle fatigue life
(Nr--401 cycles) the fracture features were quite similar to those observed at the lower strain
amplitude. The region of crack propagation prior to sudden fracture was only slightly smaller than
that of the lower strain amplitude, indicating a faster crack propagation rate. The transition region
from stable crack propagation to sudden fracture revealed a higher density of secondary cracks,
which were parallel to the stress axis [Fig. l l(a)]. The regions of transgranular fracture which
showed evidence of fine striation-like markings were more prevalent but remained non-uniformly
distributed throughout the fracture surface [Fig. 1l(b)]. Formation of voids at the carbides or
inclusions lying along the grain boundaries, aided by void-void interaction and coalescence, results
in secondary cracking along the grain boundary [Fig. l l(c)]. High magnification observation of
the transgranular regions revealed the fine striation-like markings to be more prevalent near second
phase carbide particles [Fig. 1 l(d)], which may tend to shield the local fracture surface during crack
closure. Assessment of the spacing of the striation-like markings indicates that the localized crack
propagation rate is influenced by the presence of the carbide in the microstructure. High
magnification observation of the sudden fracture region revealed microscopic features indicative
of classical ductile fracture.
Interfacial strength is a dominant factor in void nucleation. Other factors, which exert an
influence on void nucleation, growth and coalescence 24,25, are namely: (1) size of the second phase
particles (carbides), (2) shape and orientation, (3) volume fraction, (4) intrinsic strength of the
particle and (5) distribution within the microstructure. Void nucleation at the second phase particles
in the microstructure is also aided by the mutually interactive influences of (i) local stress and strain
levels, and (ii) local deformation characteristics. Coalescence of the voids and their linking to an
advancing crack tip can occur due to the mutually competitive mechanisms of void sheet

846

J.D. DOUGHERTY et al.

coalescence [26] and strain localization. Consequently, cyclic ductility and strain resistance of the
material are dependent on the extent of void growth as well as their spacing.
Examination of the fracture surfaces of the low-cycle fatigue specimens, deformed to failure
over a spectrum of cyclic strain amplitudes, suggests that overload occurs prior to the crack
propagating through the vast majority of the specimen. Coalescence of the voids to microscopic
cracks and the ensuing propagation of the microscopic and macroscopic cracks, both through the
matrix and along grain boundaries and carbide/matrix interfaces, eventually causes the fracture
strength of the cyclically deformed microstructure to be exceeded during the tension portion of the
strain cycle. The resulting overload is essentially monotonic.
The lack of consistent and homogeneously distributed visible fatigue striations and
transgranular fracture striation-like features indicates that this material may not be appropriate
for methods which utilize striation spacing to determine crack closure behavior 15,16. During the
crack closure experimental portion of this study, CT specimens were tested under positive load ratio
conditions. High magnification observation of the fracture surfaces of these specimens also revealed
a lack of consistent and uniformly distributed fatigue striations, which provides further evidence
that this material may not be appropriate for striation spacing methods.
5. C O N C L U S I O N S
The present study on the cyclic stress response characteristics, strain resistance, low cycle
fatigue life and fracture behavior of modified 1070 steel reveals the following.
(1) Tensile tests revealed the material to have medium strength and high ductility. Quasi-static
fracture reveals intrinsic microstructural features to dominate the ductile failure process through
microviod formation and coalescence.
(2) The modified 1070 steel followed the Basquin and Coffin-Manson relationships, with a
single slope behavior for the variation of elastic strain amplitude and plastic strain amplitude with
reversals to failure. The cyclic ductility of the microstructure was inferior to the true monotonic
fracture ductility.
(3) Cyclic stress response, during CST tests, revealed combinations of softening, stable
behavior and hardening. The softening and hardening behavior of this steel varied as a function
of the applied cyclic strain amplitude.
(4) Cyclic stress-strain response revealed the material to be softer than the monotonic state
at lower strain amplitudes and harder than the monotonic state at strain amplitudes greater than
1.75%. Estimates of the crack tip plastic zone size using cyclic properties are twice that obtained
using the monotonic properties.
(5) Low-cycle fatigue morphology was essentially similar at the different strain amplitudes.
Fracture surfaces revealed distinct regions of crack initiation, microscopic and macroscopic crack
growth, and sudden fracture (overload). Microscopic observations of the overload region revealed
features reminiscent of ductile failure, similar to that of the monotonic tensile test specimens.
Microstructural effects on cyclic fracture resistance are exacerbated by the presence of microvoid
clusters.
Acknowledgements--The authors would like to thank The Timken Company (Canton, Ohio, U.S.A.) for supporting this
research effort.Dr J. D. Doughertywould like to thank Mike Leap for his assistancewith specimenpreparation and training
him how to conduct the tests on the load frames at Timken Research, and Chuck Mozden for training him how to use
the SEM at Timken Research.

REFERENCES
[I] A. J. MeEvily, A S T M STP 982, 35-43. American Society for Testing and Materials, Philadelphia (1988).
[2] J. D. Dougherty, Combined experimental and finite element study of fatigue crack closure in 1070M steel. Ph.D.
Dissertation, The University of Akron, U.S.A. (1994).
[3] Automotive Bearing Applications Manual. The Timken Company, Canton, Ohio, U.S,A.
[4l S. R. Mediratta, V. Ramaswamy and P. Rama Rao, Int. J. Fatigue 7, 101-106 (1985).
[5] S. R. Mediratta, V. Ramaswamy and P. Rama Rao, Int. J. Fatigue 7, 107-115 (1985).
[6] M. T. Yu, T. H. Topper and L. Wang, Int. J. Fatigue 10, 49-255 (1988).
[7] S. K. Bhambri, C. R. Prasad and R. Vasudevan, Int. J. Fatigue 9, 239-246 (1987).

Modified 1070 steel


[8]
[9]
[10]
[I 1]
[12]
[13]
[14]
[15]
[16]
[I 7]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]

847

L. Yimin and W. Jinrui, Int. J. Fatigue 14, 169-172 (1992).


V. Singh, P. V. S. S. Raju, T. K. G. Namboodhiri and P. Rama Rao, Int. J. Fatigue 12, 289-292 (1990).
P. Farsetti and A. Blarasin, Int. J. Fatigue 10, 153-161 (1988).
S. K. Koh and R. !. Stephens, Fatigue Fracture Engng Mater. Structures 14, 413-428 (1992).
M. T. Yu, D. L. DuQuesnay and T. H. Topper, Int. J. Fatigue. 12, 433-439 (1990).
D. Eifler and E. Macherauch, Int. J, Fatigue. 12, 165-174 (1990).
U. Muralidharan and R. Fuquen, Advances in Mechanical Behavior and Properties Evaluations, pp. 31-38. ASME
Winter Meeting, Chicago, U.S.A. (27 November-2 December 1988).
K. Anandan and R. Sunder, Int. J. Fatigue. 9, 217-222 (1987).
D. S. Dawicke, A. F. Grandt, Jr and J. C. Newman, Jr, Engng Fracture Mech. 36, I 11-121 (1990).
B. I. Sandor, Fundamentals of Cyclic Sti'ess and Strain. University of Wisconsin Press, Madison, Wisconsin, U.S.A.
(1972).
T. S. Srivatsan, K. Yamaguchi and E. A. Starke, Jr, Materials Science and Engineering. 83, 87-107 (1986).
B. Tomkins, Phil. Mag. 18, 1041-1049 (1968).
T. S. Srivatsan and R. Auradkar, Int. J. Fatigue. 14, 355-366 (1992).
R. W. Landgraf, J. Morrow and T. Endo, J. Mater. Sci. 4, 176-188 (1969).
J. A. Collins, Failure of Material in Mechanical Design: Analysis, Prediction, Prevention. John Wiley, New York, NY,
U.S.A. (1981).
G. R. Irwin, 1960 Sagamore Ordnance Materials Conference. Syracuse University (1961).
R. H. Van Stone and J. A. Psioda, Metall. Trans. 6A, 672~80 (1975).
R. H. Van Stone, T. B. Cox, J. R. Low and J. A. Psioda, Int. Metals Review, 30, 157-177 (1985).
T. B. Cox and J. R. Low, Metall. Trans. 5A, 1457-1470 (1974).
(Received 16 March 1995)

ElM $3/6--B

You might also like