International Journal of Fatigue 102 (2017) 1–8

Contents lists available at ScienceDirect

International Journal of Fatigue

journal homepage: www.elsevier.com/locate/ijfatigue

A cyclic slip irreversibility based model for fatigue crack initiation of

nickel base alloys
H.S. Ho a,⇑, M. Risbet b, X. Feaugas c
a
School of Mechanical Engineering and Henan Key Engineering Laboratory for Anti-Fatigue Manufacturing Technology, Zhengzhou University, Science Road 100, 450001
Zhengzhou, China
b
Laboratoire Roberval UMR CNRS 7337, Université de Technologie de Compiègne, Centre de Recherches de Royallieu, BP 20529, 60205 Compiègne Cedex, France
c
Laboratoire des Sciences de l’Ingénieur pour l’Environnement, UMR CNRS 7356, Université de La Rochelle, Avenue Michel Crepeau, 17000 La Rochelle Cedex, La Rochelle, France

a r t i c l e i n f o a b s t r a c t

Article history: The opportunity to define a microscopic law of fatigue crack initiation using Manson-Coffin law formu-
Received 28 December 2016 lated in terms of cyclic slip irreversibility deduced from AFM measurements is discussed for a polycrys-
Received in revised form 9 March 2017 talline superalloy with different grain sizes and precipitate sizes. The results show that the modified
Accepted 14 April 2017
Manson-Coffin law, relating cyclic slip irreversibility parameter to fatigue crack initiation life, is sustained
Available online 18 April 2017
through a two-parameter power law: e0f and c. The analysis suggests that the exponent c-value can be
related to the degree of plastic strain incompatibility between grains, and the cumulative irreversible cyc-
Keywords:
lic plastic strain to crack initiation is a relevant damage indicator for crack initiation. Consequently, our
Cyclic slip plastic irreversibility
Low-cycle fatigue
approach allows giving a physical base of engineering law.
Manson-Coffin fatigue crack initiation law
2017 Elsevier Ltd. All rights reserved.
Ni-base superalloy
Atomic force microscopy (AFM)

1. Introduction
Fatigue damage of materials that exhibits flow localization is
very often related to a surface phenomenon [1]. In these materials,
the cyclic slip irreversibility, which is a phenomenon resulted from
the slip displacements caused by the movement of dislocations in
the forward part of the fatigue loading cycle that are not fully
recovered in the reverse part of the fatigue loading cycle, is a pri-
mary source of fatigue crack initiation [2,3]. Cyclic slip irreversibil-
ity may arise from quite different physical mechanisms such as
irreversible dislocation displacements resulted from cutting of pre-
cipitates and surface roughening caused by to-and-fro glide of dis-
locations, as recapitulated in Refs. [4,5]. Some other important
factors known to promote cyclic slip irreversibility are the local
microstructure (e.g. grain size, precipitate size, deformation twin-
ning, extended stacking fault and grain boundary) and environ-
ment (e.g. oxygen and/or hydrogen interstitials) [6–10].
Cyclic slip irreversibility is difficult to be analyzed quantita-
tively [11]. Thereby, there exist many different forms of character-
izing the cyclic slip irreversibility in which they are given
numerical values, as summarized in Ref. [4]. Though, the cyclic slip
irreversibility defined as the fraction of microstructurally
⇑ Corresponding author.
E-mail address: hsinshen.ho@zzu.edu.cn (H.S. Ho).
irreversible cyclic plastic strain with respect to the total cyclic plas-
tic strain is found to be one of the most relevant microstructural
and life-determining fatigue damage parameter [11,12]. Some
works have recently demonstrated the opportunity to use atomic
force microscopy to measure slip band emergence and conse-
quently the degree of reversibility during cyclic loading [4,8,12–
14]. In low-cycle fatigue tests, a global and generic cumulative
damage rule known as the Manson-Coffin (MC) law is generally
used to express the relationship between total fatigue life and plas-
tic strain amplitude [15,16]. Thereby, three questions arise: (i)
whether the MC law can be formulated in terms of cyclic slip irre-
versibility, (ii) is the MC law changing if the early stage of damage
is considered as fatigue crack initiation life, and (iii) what are the
factors that influence the correlation between the two empirical
parameters (ductility coefficient and power exponential value)
dominating the microscopic law of fatigue crack initiation?
Since cyclic slip irreversibility arises from quite different phys-
ical processes, it becomes rather improbable to establish a univer-
sal microscopic relationship of fatigue crack initiation formulated
in terms of cyclic slip irreversibility for a large variety of materials
[5,12]. Meanwhile it is still possible to construct such relationship
for a particular material with specific cyclic slip irreversibility and
damage mode [5]. Thus, the present paper aims to test the validity
of MC rule formulated in terms of cyclic slip irreversibility for fati-
gue crack initiation of a Ni-base superalloy with different grain
sizes and precipitate sizes, where cyclic slip irreversibility is a

http://dx.doi.org/10.1016/j.ijfatigue.2017.04.007
0142-1123/
2017 Elsevier Ltd. All rights reserved.
2 H.S. Ho et al. / International Journal of Fatigue 102 (2017) 1–8
result of repeated irreversible cutting of the coherent c0 precipitate
particle during cycling, and to understand the nature and/or
physical content of the two empirical parameters of the modified
MC rule.
2. Experimental procedure
2.1. Material
The target material is a Ni-base superalloy (Waspaloy) strength-
ened by spherical precipitates (c0 , Ni3Al), with the following chem-
ical composition (wt.%): 1.80 Al; 3.33 Ti; 12.64 Co; 19.45 Cr; 7.31
Mo; balance Ni. The average grain size (D) and the average precip-
itate size (d) of the as-received state of Waspaloy are 64 lm and
42 nm, respectively. A two-step heat treatment (solutionizing
and aging treatments) is carried out on the as-received state to
modify D and d so as to obtain specimens with well-controlled
metallurgical states, i.e. different grain sizes of the matrix c with
a constant c0 precipitate size and inversely (see Ref. [8] for details
on heat treatment steps). The average precipitate size is kept
<50 nm in order to remain in the dislocation-particle shearing
deformation mode (under-age state). As the metallurgical parame-
ters such as grain size and precipitate size are the important
parameters in the present study, it is worth mentioning that the
determination of the average grain size and precipitate size are
based on the evaluation of more than 1500 grains and 300 precip-
itates using both the line intercept method and area analysis
method. The estimated grain sizes and precipitate sizes are then
analyzed using a normalization procedure elaborated in Refs.
[8,12]. The gamma function is found to describe the normalized
grain size and precipitate size distributions with a good accuracy.
As the distributions for the grain size and precipitate size seem
to be quite similar for all the specimens investigated in the present
study, the variability of the grain size and the precipitate size are
distinguished only by their average values. In the present
study, eight classes of specimens are selected with the
following average grain size (D) and average precipitate size
(d): (30 lm, 7 nm); (50 lm, 10 nm); (50 lm, 20 nm); (50 lm,
30 nm); (50 lm, 40 nm); (100 lm, 10 nm); (100 lm, 20 nm) and
(200 lm, 20 nm).
2.2. Experimental procedure
The specimens are mechanically polished, then chemically-
etched using standard metallographic techniques as detailed in
Ref. [8]. A unified surface state is obtained for the whole set of
specimens and consequently, the tooling residual stress and the
initial surface roughness are assumed not to influence the mecha-
nism of fatigue crack nucleation.
Interrupted low-cycle fatigue tests are conducted at ambient
temperature using an Instron servohydraulic machine on
hourglass-shaped cylindrical specimens. The fatigue tests are per-
formed on a controlled plastic strain amplitude (ep) of 0.05%, 0.3%
and 0.9% with a fully reversed triangular wave form at a frequency
of 0.05 Hz until the nucleation of first cracks. Cracks are considered
as nucleated when the length of a crack is equivalent to the dimen-
sion of a grain.
During fatigue testing, the fatigue hysteresis loops are recorded
in order to track the cyclic behavior and to calculate the internal
stresses. According to Dickson et al. [17], the cyclic stress ampli-
tude (ra) is composed of effective stress (reff) and backstress (X).
The backstress (X), which is to be discussed further, can thus be
expressed as:
1 1
X ¼ ra  reff with reff ¼ ðra  rr Þ þ r
2 2
where r⁄ is the thermal part of the flow stress and rr denotes the
reverse yield stress, which are measured from the experimental
hysteresis loops with a plastic strain offset of (4 ± 2)  105.
The surface states are characterized by scanning electron micro-
scope (SEM) and atomic force microscope (AFM), using respec-
tively a (FEG) Zeiss Sigma SEM and a Veeco Dimension 3100 AFM
at room temperature. For each specimen, about 8 damaged grains
and 8 undamaged grains are investigated. And for each grain, a few
regions of 10 lm  10 lm are examined. Each region extracted
using the topography-imaging mode contains around 20 slip
bands. The surface roughness, representing the arisen surface
marking from material surface, is then analyzed using the experi-
mental procedure developed in Refs. [8,12]. In addition to the sur-
face analysis, the bulk material is examined at room temperature
using JEOL JEM2011 transmission electron microscope (TEM) oper-
ating at 200 kV.
3. Results on fatigue crack initiation life, mean extrusion
height, band thickness and mean inter-band spacing
The stereological elements such as grain size (D) and precipitate
size (d) of investigated specimens are presented in Table 1. It is
shown that three classes of grain size and five precipitation states
are obtained with selected heat treatment processes. The values of
the other parameters, such as fatigue crack initiation life (Ni), mean
extrusion height (‹h›), band thickness (w) and mean inter-band
spacing (IB), of the specimens tested at three different levels of
plastic strain amplitude (ep) are also summarized in Table 1. The
values of these characteristic parameters are obtained by analyzing
SEM-EBSD maps, AFM images and TEM micrographs taken prior to
fatiguing and at the crack initiation phase.
The evolution of the fatigue crack initiation life (Ni) as a function
of the average precipitate size (d) for a series of plastic strain
amplitude (ep) is illustrated in Fig. 1. Ni is reduced by increasing
the values of d and ep, but it is less sensitive to the variation of d.
Indeed, increasing ep from 0.0005 to 0.009 induces a reduction of
Ni by 104 cycles, but the increase of d from 10 nm to 40 nm
reduces Ni by only 450 cycles. The impact of grain size on Ni
has been improved for the two precipitate sizes of 10 nm and
20 nm (c.f. Table 1). Besides, we observe a decrease of Ni (around
100 cycles) when the grain size is increased from 50 lm to 100 lm.
In this study, our observations show that the material suffers
from intragranular crack nucleation, a damage process that is asso-
ciated with the occurrence of microcracks localized at the interface
between slip bands and the matrix near the surface, as is illus-
trated in Fig. 2(a). This cracking mode is found to be independent
of the plastic strain amplitude and metallurgical state of the alloy,
which is in accordance with extensive previous works [12,18].
Under continued fatigue loading, slip bands emerge in the form
of only extrusions are formed on the surface of materials as a con-
sequence of the localization of plastic deformation caused by
shearing processes of precipitates by dislocations. These irre-
versible surface markings, presented in Fig. 2(b), arise from the
surface steps that are partially reversible during different fatigue
cycles [19,20]. This surface roughness, acting as local stress con-
centration due to strain localization, is essential for fatigue crack
initiation. Thereby, the surface roughness is studied as an indirect
measure of the principal local cyclic slip irreversibility responsible
for causing early stage of fatigue damage and for controlling fati-
gue crack initiation life in the following.
The crystallographic orientation of the surface grains is well
recognized to affect the slip reversibility and the occurrence of fati-
gue damage [11,21,22]. Meanwhile the detailed study of the height
ð1Þ of extrusions and the orientation of grains in polycrystalline 316L
has shown that though a direct proportionality is found for the
 H.S. Ho et al. / International Journal of Fatigue 102 (2017) 1–8 3

Table 1
Metallurgical states of investigated specimens.

Plastic strain Grain size (D, Precipitate size Fatigue crack initiation life Mean extrusion height Band thickness Mean inter-band spacing
amplitude (ep) lm) (d, nm) (Ni, cycle) (‹h›, nm) (w, nm) (IB, lm)
0.0005 30 7 33,500 50 80 8
0.003 30 7 1500 50 80 2.5
50 10 1100 98 121 1.75
50 20 800 107 140 2.17
50 30 700 109 146 1.93
50 40 650 116 171 2.42
100 10 950 97 130 1.7
100 20 700 98 150 2.11
0.009 30 7 320 50 80 1.0
50 40 65 40 1.5
100 10 196 69 0.8

1E+06 crystallographic orientation does not necessarily affect the damage

N i (cycle) ε p=0.0005 initiation along slip bands and thus, the full description of the
effect of crystallographic orientation of surface grains on fatigue
ε p =0.003
1E+05 damage initiation is not presented in the present paper.
ε p=0.009 The interrupted fatigue testing enables us to investigate the for-
mation of slip steps and growth of extrusions within slip bands in
1E+04
the surface grains of all the cyclically deformed specimens. The
evolution of mean extrusion height (h) vs. number of cycles (N) is
1E+03 found to be the same irrespective of the average precipitate size
and average grain size: h is found to increase slightly until
N = 20, then very significantly until N = 100, and thereafter remain
1E+02 constant until the end of fatigue crack initiation life (Ni) [8,12].
Subsequently, the mean extrusion height computed in the satura-
d (nm) tion phase (‹h›) at the crack initiation state is examined with
1E+01 respect to the average grain size (D) and average precipitate size
0 10 20 30 40 50 60 (d). The results show that the value of ‹h› exhibits an increasing
function of d (c.f. Fig. 3(b)) but remains insensitive to D, implying
Fig. 1. Fatigue crack initiation life (Ni) vs. average precipitate size (d) for a series of
plastic strain amplitudes (ep). that the irreversible motion and the free glide of mobile disloca-
tions during cyclic loading until Ni in grains of different sizes can
be considered to be identical.
In the following, the functional relationship between mean
height of extrusions and the sinus of the angle between the active extrusion height (‹h›) and the band thickness (w) computed at
slip direction and the surface, but no systematic dependence is the crack initiation phase is examined. The measurements of band
observed for the extrusion height in the active slip direction and thickness (w) are performed using TEM micrographs as shown in
the grain orientation [23]. Besides, our previous works dealing Fig. 3(a). The variations of the mean extrusion height (‹h›) and
with the correlational analysis between extrusion height, grain ori- the band thickness (w) estimated at the crack initiation phase are
entation, Schmid factor and damage initiation in surface grains of presented as a function of the average precipitate size (d) in
polycrystalline Waspaloy (the target material) has revealed that Fig. 3(b) and (c). The values of both the parameters ‹h› and w are
no direct linkage is found between grain orientation and Schmid found to increase with the increase of the average precipitate size
factor with crack occurrence [24]. It is substantiated that the (d), suggesting that ‹h› is positively correlated with w and thus, we

(a) (b)

Intragranular crack

Slip band

Grain boundary

2μm

Fig. 2. Surface microstructure obtained by SEM (a) and surface topography obtained by AFM (b) taken at the crack initiation phase for the specimen having D of 100 mm and d
of 20 nm.
4 H.S. Ho et al. / International Journal of Fatigue 102 (2017) 1–8

111

g=(1 1 1)
( 1 1 1)

001 011
Axis
(1,1,-1)
[1,13,-4]

(0,1,-1) (-1,3,-1)
(1,0,-1) (1,1,-3)

(0,0,-1)(-1,1,-3) (-1,1,-1)
(1,-1,-3) (-1,1,0)

(1,-1,-1) (-1,-1,-3)(-1,0,-1) (-3,1,-1)

(0,-1,-1)
(-1,-1,-1)(-3,-1,-1)(-1,0,0) (-3,1,1)
(1,-3,-1)
(-1,-3,-1)
(-1,-1,0) (-3,-1,1)
(0,-1,0)

500 nm (a) (-1,-3,1)

200
140 ‹h› (nm) (b) w (nm) (c)
180
120
160
100 εp
εp
80 0.0005 140
0.0005
0.003 120
60 0.003
0.009
40 100 0.009

20 80
d (nm) d (nm)
0 60
0 10 20 30 40 50 60 10 20 30 40 50 60

Fig. 3. TEM micrograph illustrating band thickness obtained at the crack initiation phase for specimen having D = 100 lm and d = 20 nm (a). Mean extrusion height (‹h›) (b)
and band thickness (w) (c) estimated in the crack initiation phase vs. average precipitate size (d) for a series of plastic strain amplitudes (ep).

are interested in plotting the evolution of ‹h› as a function of w, as
presented in Fig. 4.
It is shown in Fig. 4 that ‹h› exhibits a monotonic increase with
the increasing of w. According to Mughrabi and Risbet et al.
[21,24], the degree of plastic strain localization can be connected
to the band thickness (w), where a lower value of w corresponds
to a higher level of plastic strain localization. Thus, ‹h› is inversely
proportional to the degree of plastic strain localization in the crack
initiation phase, i.e. specimens having smaller precipitate particles
with narrower slip bands do not need to attain higher extrusion
height for the occurrence of fatigue damage. Indeed, when the pre-
cipitate particles ought to be small, the shearing processes of c0
precipitates by dislocations are easier since they do not impede
Fig. 4. Mean extrusion height (‹h›) vs. band thickness (w) for a series of plastic strain
amplitudes (ep).
the free glide of mobile dislocations during cyclic loading. As the
to-and-fro motion of these dislocations are easier, the production
of metallurgical defects such as cross slips are thus not favored,
results then in the formation of narrow bands. These narrow bands,
which hinder the relocation of the plastic deformation within the
bands, can thus accommodate only a small amount of plastic defor-
mation, explaining the low irreversibility of the plastic deforma-
tion in the bands. In other terms, the easy and reversible motion
of mobile dislocations in the specimen with small precipitate par-
ticles under continued cyclic loading promotes the creation of a
lower surface step displacement at the specimen surface for the
initiation of fatigue cracks.
Fig. 3(b) shows that crack nucleation is favored when ‹h› lies in
the neighborhood of 100 nm, suggesting that fatigue damage is ini-
tiated at the surface of the alloy when a critical value is attained. In
other words, when a critical value of the local irreversible plastic
strain accumulated in the slip band is reached, the dislocation
structures in the band are unable to accommodate any further
plastic deformation, and thus promoting the initiation of cracks
along slip bands that are particularly intense [1,24].
In the present study, the degree of plastic strain localization is
characterized by the measurements of inter-band spacing (IB)
using TEM micrographs taken at different stages in interrupted
fatigue tests. The earlier results show that there is a very negligible
‘‘temporal” variation found for IB measured at predetermined num-
ber of cycles [12], hence, it is considered hereby that IB(N) = IB(Ni)
"N. In the following, the inter-band spacing (IB) measured at the
crack initiation phase is examined as a function of the plastic strain
amplitude (ep). In Fig. 5, IB is observed to decrease with the increase
of ep. This observation is in accordance with results obtained for
other metallurgical states of Waspaloy [12,25], but also for other
material types such as austenite Fe-36 at.% Ni-12 at.% Al alloy
[26] and cooper [27].
 H.S. Ho et al. / International Journal of Fatigue 102 (2017) 1–8 5

hf1
irr ¼
pcum pcum ðN; ep Þ ð6Þ
NIB ðNi ; ep Þep 4M irr

5. Experimental validation

5.1. Empirical relationship between plastic strain irreversibility and

fatigue crack initiation life

Eq. (6) shows that pcum irr is dependent on N and our results reveal
that pcum irr first increases and then decreases with increasing N, as
exhibited in Fig. 6. pcum irr that decreases at the end of fatigue crack
initiation life is recognized to be related to the saturation of
ccum
irr;pl;loc and the continuous increase of cpl;loc as a function of the
cum

cyclic number. Fig. 6 shows that the functional dependency

Fig. 5. Inter-band spacing (IB) vs. plastic strain amplitude (ep).
4. Microstructure-based reformulation of the Manson-Coffin
fatigue crack initiation life
If the Manson-Coffin-type law is assumed to be valid for dam-
age initiation in polycrystals Waspaloy, this law relates the plastic
strain amplitude (ep) to the cycle number of crack initiation (Ni)
through a two-parameter power law [15,16]:
between pcumirr and N remains valid for all the specimens investi-
gated in the present study, in coherence with earlier results
obtained with the same alloy of only one metallurgical state tested
under three different levels of plastic strain ep [12].
At the fatigue crack initiation phase (N = Ni), pcum
irr (Ni) is found to
be dependent on ep (and hence also on Ni), as is exhibited in Fig. 7.
irr (Ni) increases when ep increases
It is shown in this figure that pcum
(or when Ni decreases). The variation of pcum irr (Ni) with ep (and Ni)
fits well with the results reported in the literature [12,13].
The Manson-Coffin relation in Eq. (2) formulated in terms of
cyclic slip irreversibility for N = Ni expressed in Eq. (6) can be
rewritten in modified form as:

ep ¼ e0f ð2Ni Þc ð2Þ hf1 D ðcþ1Þ ðcþ1Þ

irr ðN i ÞIB ðN i ÞD ¼
pcum Ni ¼ constant Ni ð7Þ
2cþ1 2Me0f
where e’f and c (<0) are the fatigue ductility coefficient and expo- Eq. (7) implies that the product pcum irr ðN i ÞI B ðN i ÞD is linked with the
nent, respectively. fatigue crack initiation life (Ni) in a well-defined manner. Thereby,
Inspired by earlier works of Mughrabi and Risbet et al., which the variations of the product pcum irr ðN i ÞI B ðN i ÞD with Ni determined
propose to formulate a life law in terms of microstructural experimentally from AFM observations are assessed for three differ-
parameters instead of the function of macroscopic plastic strain ent grain sizes. In Fig. 8, the following expressions are found:
amplitude endured by the specimen during fatigue test, the plastic


strain irreversibility pcum
irr is introduced in the present work as a irr ðN i ÞIB ðN i ÞD ¼ 3  10
pcum 15 0:38
Ni ; for D  30 lm ð8Þ
microstructural parameter to explicitly take into account the
fatigue damage resulted from irreversible plastic strain irr ðN i ÞIB ðN i ÞD ¼ 2  10
pcum 14 0:57
Ni ; for D  50 lm ð9Þ
[11,12,21,24]. The parameter pcum irr is defined as the fraction of the
microstructurally irreversible cyclic plastic strain with respect to irr ðN i ÞIB ðN i ÞD ¼ 2  10
pcum 15 0:25
Ni ; for D  100 lm ð10Þ
the total cyclic plastic strain.
ccum
irr;pl;loc
irr ¼
pcum ð3Þ
ccum
pl;loc

with ccum
irr;pl;loc is the cumulative irreversible cyclic plastic strain,
expressed as:
h
ccum
irr;pl;loc ¼ ð4Þ
D
where h is the mean extrusion height and D is the average grain
size, and ccum
pl;loc is the local plastic strain, representing the total plas-
tic strain accumulated in a slip band during N fatigue cycles. ccum
pl;loc is
expressed as a function of the volume fraction of slip bands (fv),
which itself depends on inter-band spacing (IB) and volume fraction
of one slip band (f1):
!
X
N XN
4M ep
c cum
pl;loc ¼ cpl;loc ¼ IB ðN; ep Þ pl;loc ðN; ep Þ
¼ ccum ð5Þ
k¼1 k¼1
Df1

where M is the Taylor factor and N is the number of cycles. In

shearable-precipitated alloy, the plastic strain is supported by mul-
tiplication of slip bands, and thus IB is dependent on ep, as illustrated
Fig. 6. Variation of the fraction of the microstructurally irreversible cyclic plastic
in Fig. 5. By combining the expressions for ccum irr;pl;loc in Eq. (4) and


strain pcum

as a function of the number of cycles (Ni) for different metallurgical
irr
c cum
pl;loc in Eq. (5), pcum
irr can be expressed as a function of N: states tested under three different levels of plastic strain amplitude (ep).
6 H.S. Ho et al. / International Journal of Fatigue 102 (2017) 1–8

limit, the exponents b in Eqs. (8)–(10), found to be equivalent to

0.38 for D  30 lm, 0.57 for D  50 lm and 0.25 for D  100 lm,
are in fairly good agreement with expectation.

5.2. Correlation between fatigue ductility coefficient and fatigue

ductility exponent

Subsequently, the two coefficients of the Manson-Coffin fatigue

Fig. 7. Variation of plastic strain irreversibility ðpcum
irr (Ni)) as a function of plastic
strain amplitude (ep) for different grain sizes (D).
irr ðN i ÞI B ðN i ÞD with cycle number to crack
Fig. 8. Evolution of the product pcum
initiation (Ni).
The data plotted in Fig. 8 and the relationships shown in Eqs.
(8)–(10) suggest that the product pcumirr ðN i ÞI B ðN i ÞD is empirically
connected to the fatigue crack initiation life (Ni) by a power law
in the form of:
p  að2Ni Þb
crack initiation law denoted as e’f and c are deduced from Eqs. (13)
and (14) by considering the following experimental data:
‹h› = 100  109 m, f1 = 3.3  103 (D  30 lm) or f1 = 2.0  103
(D  50 lm) or f1 = 1.0  103 (D  100 lm), and M = 3. The rela-
tionship between the two empirical parameters of MC law are fur-
ther assessed.
In the literature, it is shown by correlation that e0 f and c can be
connected through a linear log (e0 f)  c relationship [29,32]. Thus,
an log (e’f) = log (e’p)  clog (2N0 i) type correlation between e0 f
and c is assessed (c.f. Fig. 9); the coefficients e0 p and N0 i, reported
to be the coordinates of a fixed point at varying c, would have
potential to be related to the fatigue damage [33–35]. Fig. 9 exhi-
bits that the log (e0 f) value is decreasing with increasing c-value,
implying that the supposed correlation between the logarithm of
e0 f and c is rather strong. Finally, with the correlation coefficient
value (R) of 0.948, the correlation can be regarded to be rather
viable.
In an attempt to examine the factors influencing the correlation
between e0 f and c plotted in Fig. 9, the behavior of the fatigue duc-
tility exponent (c) is assessed as a function of the grain size (D) for
fatigue damage initiation in polycrystal Waspaloy. Our results
show that c exhibits a decreasing tendency with increasing D. Thus,
to understand the behavior of c vs. D, an approach based on an
overall analysis of the impact of internal stresses on the occurrence
of early fatigue damage is used in the following. It is worth noting
that the initiation of cracks along slip bands is very often argued to
be a consequence of the local stress concentration due to strain
localization [24]. Thereby, to explore the possible contribution of
the stress-strain relationship on the crack initiation process, the
average value of the long-range internal stress (backstress X) esti-
mated at the crack initiation phase is examined.
According to Feaugas et al., the backstress (X) stemming from
the long-range interaction can be decomposed into two distinct
components called intragranular backstress (Xintra) and intergranu-
lar backstress (Xinter):
X ¼ X intra þ X inter ð15Þ
ð11Þ

Eq. (11) seems to be rather similar to Eq. (7). Thus, from a compar-
ison of Eqs. (7) and (11), it follows that: 0.2
log (ε'f)
p ¼ pcum
irr ðN i ÞI B ðN i ÞD ð12Þ
-0.2
and

hf1 D -0.6
a¼ ¼ constant ð13Þ
2cþ1 2Me’f
-1
and

b¼cþ1 ð14Þ -1.4 log (εf) = -3.514 c -2.7

Morrow has suggested that for total fatigue life, c is ranging R = 0.948
from 0.5 to 0.6 [28]. Yet, lower c-values are found for fatigue c
-1.8
crack initiation life [29–31]. Thereby, the exponents b in Eqs.
-1 -0.8 -0.6 -0.4 -0.2
(8)–(10) are examined to verify if they are reasonably connected
to c + 1. With the data obtained from the literature, one expects Fig. 9. Correlation between log fatigue ductility coefficient (e0 f) vs. fatigue ductility
that b = c + 1 < 0.5 for the early-stage fatigue damage. Within the exponent (c).
 H.S. Ho et al. / International Journal of Fatigue 102 (2017) 1–8
Fig. 10. Evolution of intergranular long-range internal stresses (Xinter) as a function
of grain size (D).
where Xintra is related to strain incompatibility between grains,
which involves the crystal or grain boundary structure, while Xinter
is a consequence of the heterogeneity of the dislocation spatial dis-
tribution [36]. The results show that the backstress (X) is inter-
dependent upon metallurgical parameters such as grain size and
precipitate size, where X is found to be inversely proportional to
the grain size (D) while exhibiting a tendency of monotonically
increasing with the precipitate size (d). Our analyses then show that
the dependency of the backstress (X) on the grain size (D) is mainly
related to the intergranular long-range internal stresses (Xinter) cre-
ated by the plastic strain incompatibility between grains, in agree-
ment with previous results reported in the literature [37]. In Fig. 10,
Xinter is found to be a decreasing function of the grain size D and
thus, the functional relationship between Xinter and D can be formal-
ized using the following phenomenological law with D in
micrometers:
X inter ¼ 920  D1=2
This finding suggests that the elastic-energy stored inside the
slip bands can easily be relaxed when the grain size increases
due to the decrease in the plastic strain incompatibility resulting
from the slip transmission across the grain boundary, which
implies that the grain of large-grain-sized polycrystals can be con-
sidered as pseudo-monocrystal.
5.3. The cumulative irreversible cyclic plastic strain to fatigue crack
initiation
Ho et al. showed that there exists a critical value of cumulative
 
irreversible cyclic plastic strain to crack initiation ccum
irr;pl;loc for
fatigued polycrystal Waspaloy and they found that fatigue damage
is initiated when ccum
irr;pl;loc is equal to 1.3 ± 0.3  10
3
, evaluated by
considering the mean extrusion height computed in the saturation
phase (‹h›) at the crack initiation state in Eq. (4); it is worth noting
that on account of the grain size inhomogeneity, the value of
ccum
irr;pl;loc is computed by averaging the cirr;pl;loc ðGk Þ values estimated
cum
for all damaged grains (G) investigated in a cyclically strained spec-
imen with k is the number of examined damaged grains, as has
been proposed in Ref. [8]. It is therefore important to evaluate this
quantity based on the modified Manson-Coffin fatigue crack initi-
ation law in Eq. (7). Rewriting Eq. (4) by considering Eq. (13), we
obtain:
a2Me0f 2cþ1
ccum
irr;pl;loc ¼
D2 f 1
7
Evaluation of a critical cumulative irreversible cyclic plastic
 
strain to crack initiation ccum
irr;pl;loc via Eq. (17) for the grain sizes
of 30 lm, 50 lm and 100 lm yields very similar values of
3.3  103, 2.0  103 and 1.0  103, respectively. These values
are of the same order of magnitude as the critical value of
1.3 ± 0.3  103 computed directly from Eq. (4). However, the crit-
ical value of cumulative irreversible cyclic plastic strain to crack
initiation for fatigued polycrystals Waspaloy is found to be much
smaller than the critical value of cumulative irreversible cyclic
plastic strain to failure, estimated to lie in the range of 10 to
30, for fatigued a-iron polycrystals and Cu–31% Zn (a-brass) sin-
gle crystal [5]. In this sense, it can be expected that the character-
istic critical value of cumulative irreversible cyclic plastic strain for
a specific damage type in a particular material is rather unique, as
assumed by the modified Manson-Coffin fatigue crack initiation
law [5]. Thereby, the computed critical value of cumulative irre-
versible cyclic plastic strain equal to 1.3 ± 0.3  103 can thus be
considered as a constant for the early stages of fatigue damage in
polycrystal Waspaloy.
6. Summary
In summary, a microstructure-based reformulation of the mod-
ified Manson-Coffin law has been discussed for fatigued Ni-based
superalloy, Waspaloy. The experimental results suggest the exis-
tence of a power-law relationship between cyclic slip irreversibil-
ity parameter and fatigue crack initiation life, and show that the
two-parameter of the Manson-Coffin law can be related through
a linear log (e0 f)  c relationship. The analyses suggest that higher
exponent c-value signifies higher sensitivity to the plastic strain
incompatibility between grains and that fatigue crack initiation life
can be controlled by limiting the value of cumulative irreversible
cyclic plastic strain to crack initiation.
References
ð16Þ
[1] Suresh S. Fatigue of materials. 2nd ed. Cambridge: Cambridge University Press;
1998.
[2] Essmann U, Gosele U, Mughrabi H. A model of extrusions and intrusions in
fatigued metals. 1. Point-defect production and the growth of extrusions.
Philos Mag A 1981;44:405–26.
[3] Basinski ZS, Pascual R, Basinski SJ. Low amplitude fatigue of copper single
crystals-I. The role of the surface in fatigue failure. Acta Metall
1983;31:591–602.
[4] Shyam A, Milligan WW. A model for slip irreversibility, and its effect on the
fatigue crack propagation threshold in a nickel-base superalloy. Acta Mater
2005;53:835–44.
[5] Mughrabi H. Cyclic slip irreversibility and fatigue life: a microstructure-based
analysis. Acta Mater 2013;61:1197–203.
[6] Soboyejo WO, Mercer C. On the implications of deformation-induced crack-tip
deformation in a gamma-based titanium aluminide intermetallic. Scr Metall
Mater 1994;30:1515–20.
[7] Sangid MD, Maier HJ, Sehitoglu H. A physically based fatigue model for
prediction of crack initiation from persistent slip bands in polycrystals. Acta
Mater 2011;59:328–41.
[8] Ho HS, Risbet M, Feaugas X. The effect of grain size on the localization of plastic
deformation in shear bands. Scr Mater 2011;65:998–1001.
[9] Sriram TS, Fine ME, Chung YW. TM and surface analytical study of the effect of
environment on fatigue crack initiation in silver single crystals II: effects of
oxygen partial pressure. Scr Metall Mater 1990;24:279–84.
[10] Zhang P, Zhang ZJ, Li LL, Zhang ZF. Twin boundary: stronger or weaker
interface to resist fatigue cracking? Scr Mater 2012;66:854–9.
[11] Mughrabi H. Cyclic slip irreversibilities and the evolution of fatigue damage.
Metall Mater Trans A 2009;40:1257–79.
[12] Risbet M, Feaugas X. Use of atomic force microscopy to quantify slip
irreversibility in a nickel-base superalloy. Scr Mater 2003;49:533–8.
[13] Gerberich WW, Harvey SE, Kramer DE, Hoehn JW. Low and high cycle fatigue -
a continum supported by AFM observations. Acta Mater 1998;46:5007–21.
[14] Man J, Valtr M, Petrenec M, Dluhos J, Kubena I, Obrtlik K, et al. AFM and SEM-
FEG study on fundamental mechanisms leading to fatigue crack initiation. Int J
Fatigue 2015;76:11–8.
ð17Þ [15] Coffin LF. A study of the effect of cyclic thermal stresses on a ductile metal.
Trans ASME 1954;76:931–50.
8 H.S. Ho et al. / International Journal of Fatigue 102 (2017) 1–8
[16] Manson SS. Behavior of materials under conditions of thermal stresses. NACA
Rep 1954;1170:317–50.
[17] Dickson JL, Boutin J, Handeld L. A comparison of two simple methods for
measuring cyclic internal and effective stresses. Mater Sci Eng 1984;64:
L7–L11.
[18] Mughrabi H. On the life-controlling microstructural fatigue mechanisms in
ductile metals and alloys in the gigacycle regime. Fatigue Fract Eng Mater
Struct 1999;22:633–41.
[19] Finney JM, Laird C. Strain localization in cyclic deformation of copper single
crsytals. Philos Mag 1975;31:339–66.
[20] Depres C, Robertson CF, Fivel MC. Low-strain fatigue in AISI 316L steel surface
grains: a three-dimensional discrete dislocation dynamics modelling of the
early cycles II. Persistent slip markings and micro-crack nucleation. Philos Mag
2006;86:79–97.
[21] Mughrabi H. Cyclic deformation of multi-phase materials. In: 4th Riso inter
symp on metall and mater; 1983. p. 65–82.
[22] Ortner S, Laird C, Farrington GC. The mechanical response of copper single
crystals to corrosion – fatigue in an aqueous oxide-forming environment. Acta
Metall 1987;35:453–62.
[23] Man J, Obrtlik K, Blochwitz C, Polak J. Atomic force microscopy of surface relief
in individual grains of fatigued 316L austenitic stainless steel. Acta Mater
2002;50:3767–80.
[24] Risbet M, Feaugas X. Some comments about fatigue crack initiation in relation
to cyclic slip irreversibility. Eng Fract Mech 2008;75:3511–9.
[25] Stoltz RE, Pineau AG. Dislocation-precipitate interaction and cyclic stress-
strain behavior of a c0 strengthened superalloy. Mater Sci Eng 1978;34:
275–84.
[26] Hornbogen E, Gahr K-HZ. Microstructure and fatigue crack growth in a c-Fe-
Ni-Al alloy. Acta Metall 1976;24:581–92.
[27] Cretegny L, Saxena A. AFM characterization of the evolution of surface
deformation during fatigue in polycrystalline copper. Acta Mater
2001;49:3755–65.
[28] Morrow JD. Internal friction, dampling and cyclic plasticity. ASTM STP 378.
Philadelphia, PA: ASTM; 1965. p. 45–87.
[29] Alush H, Katz Y, Maros MB, Toth L. Some insights into the remote strain versus
fatigue life relationship. J Mater Process Technol 2004;157–158:16–22.
[30] Cooper CV, Fine ME. Coffin-Manson relation for fatigue crack initiation. Scr
Metall 1984;18:593–6.
[31] Harvey SE, Marsh PG, Gerberich WW. Atomic force microscopy and modeling
of fatigue crack initiation in metals. Acta Metall Mater 1994;42:3493–502.
[32] Krazowski AJ, Toth L. A thermodynamic analysis of the empirical power
relationship for creep rate and rupture time. Metall Mater Trans A
1997;28:1831–42.
[33] Maros M. Contribution to the physical understanding of the Coffin-Manson
law. In: Low cycle fatigue and elasto-plastic behavior of materials. Elsevier
Science; 1998. p. 99–104.
[34] Yarema SY. Correlation of the parameters of the Paris equation and the cyclic
crack resistance characteristics of materials. Strength Mater 1981;13:1090–8.
[35] Tribonod B. A fixed point in the Coffin-Manson law. Int J Fatigue
2015;81:143–7.
[36] Feaugas X, Haddou H. Grain-size effects on tensile behavior of nickel and AISI
316L stainless steel. Metall Mater Trans A 2003;34:2329–40.
[37] Feaugas X. On the origin of the tensile flow stress in the stainless steel AISI
316L at 300 K: back stress and effective stress. Acta Mater 1999;47:3617–32.