Content uploaded by Zhen Zhang
Author content
All content in this area was uploaded by Zhen Zhang on Nov 15, 2017
Content may be subject to copyright.
Content uploaded by Zhen Zhang
Author content
All content in this area was uploaded by Zhen Zhang on Nov 15, 2017
Content may be subject to copyright.
Low-Cycle Fatigue Properties of P92 Ferritic-Martensitic
Steel at Elevated Temperature
Zhen Zhang, ZhengFei Hu, Siegfried Schmauder, Marijo Mlikota, and KangLe Fan
(Submitted November 2, 2015; in revised form January 10, 2016)
The low-cycle fatigue behavior of P92 ferritic-martensitic steel and the corresponding microstructure
evolution at 873 K has been extensively studied. The test results of fatigue lifetime are consistent with the
Coffin-Manson relationship over a range of controlled total strain amplitudes from 0.15 to 0.6%. The
influence of strain amplitude on the fatigue crack initiation and growth has been observed using optical
microscopy and scanning electron microscopy. The formation mechanism of secondary cracks is established
according to the observation of fracture after fatigue process and there is an intrinsic relationship between
striation spacing, current crack length, and strain amplitude. Transmission electron microscopy has been
employed to investigate the microstructure evolution after fatigue process. It indicates the interaction
between carbides and dislocations together with the formation of cell structure inhibits the cyclic softening.
The low-angle sub-boundary elimination in the martensite is mainly caused by the cyclic stress.
Keywords crack nucleation, cyclic softening, low-cycle fatigue,
P92 steel
1. Introduction
Several grades of 9-12%Cr ferritic-martensitic heat-resistant
steels are candidate materials for structural components for Gen
IV nuclear power plants, due to their high strength, low thermal
expansion, good corrosion resistance, and good mechanical
properties at elevated temperature (Ref 1–3). These character-
istics permit to consider them to be extensively used as
structural materials for many components in steam generators
and fusion reactors (Ref 4). P92 steel is one of these 9-12%Cr
ferritic-martensitic heat-resistant steels and its higher stress
rupture strength is obtained by an addition of 1.8% W and a
reduction of the Mo content from 1 to 0.5% (Ref 5). These
components are often subjected to repeated thermal stress
owing to temperature gradients that occur on heating and
cooling during start-ups and shutdowns. This, in turn, may give
rise to plastic deformation over a period of time, especially
under high working temperatures. Therefore, the deep under-
standing of the strain-controlled low-cycle fatigue (LCF)
behavior is an important concern for its applications. To date,
extensive works have been reported to study the LCF (Ref 6–
9), aging (Ref 10–12), and creep-fatigue interaction (Ref 13–
15) behavior of 9-12% Cr ferritic-martensitic steels and their
microstructures have been reported under different conditions.
Under cyclic loading, these steels exhibit continuous softening
as a result of microstructural instability (Ref 6–9). Previous
low-cycle fatigue studies on these steels about cyclic softening
behavior attribute this phenomenon to the annihilation of
dislocations within the lath (Ref 7), the increase of subgrain
size (Ref 8), and the elimination of grain/lath boundary (Ref 9).
However, the evidences of microstructural evolution associated
with fatigue behavior, like the cyclic stress response in the
microstructure and the formation of secondary cracks, are much
less known. The aim of this study is, therefore, to investigate
experimentally the LCF properties of P92 steels at 873 K (a
typical operating temperature of components of fossil-fired
power plants). The combined damage mechanism related to the
different total strain amplitudes was characterized first and the
mechanism of secondary crack initiation was presented in the
next part. Finally, the microstructure changes under the effect of
both cyclic strain and thermal exposure were characterized and
the relationship between the macro fatigue properties at
different softening parts and microstructural evolutions were
discussed.
2. Experimental Details
2.1 Material
The material examined in this study is P92 steel, a kind of
new 9Cr ferritic-martensitic steel. Its chemical composition
specified by standard ASTM (Ref 16) is given in Table 1.In
order to get the full tempered martensitic structure, the final
heat treatment consists of austenitizing at 1323 K for 4 h (air
cooling) and tempering at 1033 K for 6 h (air cooling). The
initial microstructures of P92 steels are shown in Fig. 1. The
examined material in as-received state is a tempered martensite
with normalizing and tempering treatment. The structure
composed of block martensites, which consist of laths deco-
rated with stringers of M
23
C
6
carbides (with particle size in the
range of 40-250 nm) that were Cr enriched at the prior austenite
grain boundaries (PAGB) and martensite lath boundaries
(MLB). Moreover, there existed some fine MX-type carbides
(with particle size of 6-20 nm) rich in V and Nb, which were
Zhen Zhang, ZhengFei Hu, and KangLe Fan, School of Materials
Science and Engineering, Tongji University, Shanghai 201804, China;
and Siegfried Schmauder and Marijo Mlikota, IMWF, Stuttgart
University, 70569 Stuttgart, Germany. Contact e-mails: 012zhangzhen
@tongji.edu.cn, huzhengf@tongji.edu.cn, siegfried.schmauder@imwf.
uni-stuttgart.de, marijo.mlikota@imwf.uni-stuttgart.de, and 12klfan@
tongji.edu.cn.
JMEPEG ASM International
DOI: 10.1007/s11665-016-1977-8 1059-9495/$19.00
Journal of Materials Engineering and Performance
homogeneously dispersed in the inter-lath regions (Ref 17). In
the process of martensitic transformation, the strong interac-
tions between austenite grain boundaries and martensitic laths
result in high dislocation densities in the range 10
14
m
2
in the
matrix (Ref 18), while typical values for dislocation densities in
pure metals and alloys are 10
12
m
2
(undeformed).
2.2 Mechanical Properties
The mechanical stress-strain curves of P92 steel at 293 and
873 K are shown in Fig. 2. After UTS was reached, the stress
declined gradually and the examined material fractured at about
422 and 264 MPa, respectively. Three tensile tests were
conducted for each temperature to acquire the average value of
the tensile properties of P92 steel and the comparative
mechanical properties at 293 and 873 K are given in Table 2.
As we can see, the mechanical parameters reduce dramatically
at 873 K.
2.3 Fatigue Tests
The P92 samples (Fig. 3) were taken from the longitudinal
direction of a commercial pipe. Fatigue tests at 873 and 293 K
were conducted in air under tensile/compression conditions
with asymmetry factor of R=1. The LCF tests were
performed using an MTS 809 servohydraulic test machine,
which was equipped with an induction heating system
MTS652.01. The temperatures of each sample were monitored
by thermocouple and controlled by an induction heating
system. All tests were started after soaking for 30 min at the
test temperatures in order to ensure the uniform temperature
throughout the samples. The tests were performed at a constant
strain rate of 0.004 s
1
at different total strain amplitudes. The
failure criterion was fracture or 30% drop in the peak tensile
stress (Ref 21).
3. Results and Discussion
3.1 Cyclic Deformation Properties
These kinds of materials generally show a continuous
softening behavior that ends in a period of stress saturation
under cyclic loading. The curves of peak cyclic stress at
different total strain amplitudes are presented in Fig. 4.Itis
observed in Fig. 4(a) and (b) that the stage of cyclic softening
can be divided into three parts, which consisted of a rapid
softening part, a saturation part, and a final failure part. The
maximum peak tensile stress for all curves was obtained during
the first cycles, and then a large amount of softening behavior
occurred from the first cycle onwards, after which there was a
Table 1 Chemical composition of P92 steel (mass percent, %)
C Mn P S Si Cr W Mo V Nb N B Al Ni
ASTM A335-2003 0.07-0.13 0.3-0.6 £0.02 £0.01 £0.5 8.5-9.5 1.5-2.0 0.3-0.6 0.15-0.25 0.04-0.09 0.03-0.07 0.001-0.006 £0.04 £0.4
Test 0.1 0.45 0.015 0.008 0.3 8.82 1.57 0.35 0.2 0.078 0.037 0.0027 0.006 0.11
Fig. 1 Tempered martensite structure of P92 steel (a) and schematic diagram of 9Cr ferritic-martensitic steel (Ref 19,20) (b)
Fig.2 Stress-strain curve of P92 steel at different temperatures
Journal of Materials Engineering and Performance
very gradual change in the stress values under all the test
conditions. Moreover, the stabilizing cycles and fatigue life
were significantly decreased with the increase of total strain
amplitude. At the first five cycles at the temperature of 293 K, a
very short hardening behavior can be observed at the total strain
amplitude above 0.55% (Fig. 4c). While at the first one cycle at
the temperature of 873 K, a similar hardening behavior is also
observed at the total strain amplitude above 0.40% (Fig. 4d),
whereas there are no hardening behaviors at the lower strain
amplitudes for both temperatures. The increment in the
dislocation density as well as the interaction between the
dislocation and precipitate contributes to the initial cyclic
hardening observed above. After this very short consolidation
phase, the peak stress decreases rapidly. Similar results are
obtained on other materials (Ref 22–24). After the very short
beginning stage, the material shows continuous rapid softening
behavior before the saturation stage and they are consistent at
all total strain amplitudes (Fig. 4e and f). At higher strain
amplitudes and resultant shorter fatigue life N
f
, the extent of
softening, or the decrease of peak stress, is more obvious than
that at lower strain amplitudes. The initial microstructure
contains a lot of dislocations, which have a trend of redis-
tributing to a low energy configuration (Ref 25) such as
dislocation network (DN), cell structure (CS) during cyclic
deformation, or disappears by annihilation process (Ref 26). It
has been reported that the most relevant microstructural
changes which take place at the beginning of test is a
significant decrease of dislocation density (Ref 27) and this
result is in agreement with a previous one obtained by Giordana
et al. (Ref 28). At 293 K (Fig. 4a), the peak of stress reaches a
saturating level after the rapid softening stage. However, at 873
K (Fig. 4b), the evolution of the stress amplitude with the
number of cycles was distinct compared with that observed at
293 K. Here the stress amplitude decreased continuously
without an obvious stabilized value at all total strain ampli-
tudes, which shows that the softening rate at this stage is
accelerated by the effect of high temperature. The cyclic stress
response under different strain amplitudes is compared at two
temperatures, as shown in Fig. 5. The amount of softening (As)
incurred during cycling was calculated according to the
following equation.
Amount of softening ð%Þ¼½ðrinitial rhalf lifeÞ=rinitial100;
ðEq 1Þ
where rinitialand rhalf life denotes the peak tensile stress at the
first and half-life cycle, respectively. As indicated, the soften-
ing amount is accelerated with increasing total strain ampli-
tudes. Moreover, it is observed that the amount of softening
at 873 K is significantly higher than that at 293 K, which is
further associated with dislocation movement and microstruc-
ture evolution at high temperature. It is interesting to note
that the amount of softening at elevated temperature changes
more distinctly, compared with that at RT, when the total
strain amplitude is lower than 0.4%. It can be concluded that
the softening behavior of the steel at elevated temperature is
influenced greatly by the strain amplitude in a particular
range.
Upon unloading and subsequent reloading, the stress-strain
hysteresis loops would develop due to the non-recoverable
plastic damage during the cyclic loading. The shape, area, and
location of fatigue hysteresis loops can reflect the fatigue
damage evolution of materials. The stress-strain plot (hysteresis
loops) obtained from the cyclic test conducted with different
strain amplitudes are shown in Fig. 6. Fig. 6a displays typical
stress-strain hysteresis loops of the second and half-life cycles
at different strain amplitudes. The hysteresis loops at both
cycles are almost asymmetric in tension and compression
stress. Irrespective of applied strain amplitudes, the maximum
tension and compression stress at half life is reduced compared
with the second cycle hysteresis loops, which is a clear
indication of cyclic softening of the material. The lower tips of
half-life hysteresis loops corresponding to different total strain
amplitudes are transferred to a common origin, in order to
check whether P92 steel exhibits Masing behavior or not, as
shown in Fig. 6b. It can be seen that the stress-strain curve for
different strain amplitudes do not overlap on each other, which
indicates non-Masing type behavior of the P92 steel at 873 K.
The non-Masing behavior is in general attributed to the phase
instability and transient dislocation substructure (Ref 29,30).
The evolution of hysteresis area and modulus obtained from the
cyclic test conducted with different strain amplitudes are shown
Fig.3 Dimension of fatigue specimen (unit: mm)
Table 2 Basic mechanical properties of P92 steel
Temperature, K 0.2% proof strength, MPa Tensile strength, MPa Elongation,% Modulus of elasticity, GPa
293 465 627 26 215
873 306 381 20 151
Journal of Materials Engineering and Performance
in Figs. 7(a) and (b). The fatigue hysteresis modulus decreases
very slowly in the whole life until the final sharp drop, as
shown in Fig. 7(a). In most of fatigue life, the hysteresis
modulus degrades slowly due to the weakening of structure
strengthening and the nucleation of microcracks, while the final
sudden drop of hysteresis modulus is ascribed to the crack
propagation in the materials. The hysteresis area associated
with fatigue hysteresis loops is the energy lost during
corresponding unloading and subsequent reloading cycles. A
major part of this energy input is dissipated into heat, and the
10 100 1000 10000
-600
-400
-200
0
200
400
600 Temperature: 293K
(a)
Peak and Valley stress / MPa
Number of cycles /
N
0.25%
0.30%
0.40%
0.55%
0.80%
5100 1000 10000
-400
-200
0
200
400
Peak and Valley stress / MPa
Number of cycles /
N
0.15%
0.20%
0.25%
0.30%
0.40%
0.60%
Temperature: 873K
(b)
2 4 6 8 10 12
450
500
550
600
Temperature: 293K
(c)
Stress amplitude, Sa=(Smax-Smin)/2, / MPa
Number of cycles /
N
0.25%
0.30%
0.40%
0.55%
0.80%
2 4 6 8 10 12
250
300
350
400
Temperature: 873K
(d)
Stress amplitude, Sa=(Smax-Smin)/2, / MPa
Number of cycles /
N
0.15%
0.20%
0.25%
0.30%
0.40%
0.60%
200 400 600
400
450
500
550
600
Temperature: 293K
(e)
Stress amplitude, Sa=(Smax-Smin)/2, / MPa
Number of cycles /
N
0.25%
0.30%
0.40%
0.55%
0.80%
100 200 300
250
300
350
400
Temperature: 873K
(f)
Stress amplitude, Sa=(Smax-Smin)/2, / MPa
Number of cycles /
N
0.15%
0.20%
0.25%
0.30%
0.40%
0.60%
Fig. 4 Cyclic stress response curves at 293 K (a, c, e) and 873 K (b, d. f), (a, b) whole life cycles; (c, d) 1-10 cycles; (e, f) 1-500 and 1-300
cycles
Journal of Materials Engineering and Performance
remaining mechanical energy causes dislocation movement
leading to permanent deformation and final crack propagation
(Ref 31,32). As shown in Fig. 7(b), the area of hysteresis loops
is increased with the increasing of total strain amplitudes.
Considering the effect of total strain amplitudes, it can be seen
that the evolution of hysteresis loop areas at the higher strain
amplitude (above 0.4%) is contrary to the smaller ones. Under
the lower strain amplitudes (less than 0.4%), the evolution of
hysteresis loop areas shows a slight increase with the progress
of cycles over most of the fatigue life followed by a sharp drop
at the onset of fracture. On the contrary, the evolution of
hysteresis loop areas decreases with the ongoing of fatigue test
under the higher strain amplitude, and similar results are
acquired by Wang et al. (Ref 33). The influence of total strain
amplitude on plastic strain amplitude and hysteresis loop area at
stable half-life cycle is shown in Fig. 7(c) and (d). It is
interesting to find that both plastic strain and hysteresis loop
area vary linearly when the strain amplitude increases, so the
value of plastic strain and hysteresis loop area could be
predicted by the fitted equation.
3.2 Coffin-Manson Relationship
The fatigue lives at stress-controlled test follows the strain-
life relationship derived by Raske and Morrow (Ref 34) and
Landgraf et al. (Ref 35) based on the relationship proposed by
Basquin (Ref 36) and by Coffin (Ref 37) and Manson (Ref 38).
The strain-life relationship is given by
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
12
16
20
24
28
Amount of softening
Total strain amplitude / %
293 K
873 K
Fig.5 Amount of softening for different test conditions
Fig.6 The fatigue hysteresis loops under different total strain amplitudes at 873 K (a) the 2nd and half lifetime; (b) the non-Masing type behav-
ior
Journal of Materials Engineering and Performance
Det
2¼r0
f
Eð2NfÞb0þe0
fð2NfÞc0;ðEq 2Þ
where Det/2 is the total strain amplitude, 2N
f
is the number
of cycles to failure, e0
fis the fatigue strength coefficient, b
0
is
the fatigue strength exponent, r0
fis the fatigue ductility coef-
ficient, c
0
is the fatigue ductility exponent, and E is the elas-
tic modulus. The cyclic stress-strain curve is plotted by
joining the vertices of the hysteresis loops at half life
(Fig. 6a). The cyclic stress-strain parameters are characterized
using power law fits by the relation
Dr
2¼KDep
2
n
;ðEq 3Þ
where Dris the total stress range, Depis the true plastic strain
range, nis the cyclic strain hardening exponent, and Kis the
cyclic strength coefficient. This equation provides a measure of
the response of a material to cyclic straining. The evaluated fati-
gue parameters in this work are given in Table 3. In general, c
0
is in the range between 0.5 and 0.7 for ductile materials
(Ref 39) and these parameters are similar to those of other con-
ventional ferritic steels. The decrease of the fatigue strength
coefficient e0
fand the fatigue strength exponent b
0
at high tem-
perature is caused by the ongoing dislocation annihilation and
the accompanied decomposition of martensite lath, which is
speeded up at high temperature. The effect of cyclic strain on
LCF of P92 steel at 873 K can also be viewed as an indication
of the resistance of the microstructure to crack initiation and
failure. The value of Kat 293 K is higher compared to that at
873 K, which correlates with the higher yield stress at lower
temperature. But the value of nincreases with an increase of
temperature, which is caused by the ongoing severe microstruc-
tural evolution at high temperature.
0.2 0.4 0.6
0.0
0.2
0.4
Plastic strain amplitude
Linear Fit of Plastic strain amplitude
Plastic strain amplitude (%)
Total strain amplitude (%)
(Δεp/2)=0.90762(Δεt/2)-0.12319
(c) 0.2 0.4 0.6
0.000
0.002
0.004
(d)
Wp=0.00876(Δεt/2)-0.00137
hysteresis loop areas
Linear Fit of hysteresis loop areas
hysteresis loop areas (mm2)
Total strain amplitude (%)
100 1000 10000 100 1000 10000
120000
140000
160000
180000
200000
(a)
Cycle Modulus (
N
/mm2)
Number of cycles (
N
)
0.15%
0.20%
0.25%
0.30%
0.40%
0.60%
Temperature: 873K
-0.001
0.000
0.001
0.002
0.003
0.004
0.005
Temperature: 873K
Area of Hysteresis Loop (mm2)
Number of cycles (
N
)
0.15%
0.20%
0.25%
0.30%
0.40%
0.60%
(b)
Fig.7 The fatigue hysteresis parameters under different total strain amplitudes at 873 K (a) the fatigue modulus vs. the cycle number; (b) the
hysteresis area vs. the cycle number; relationship of the strain amplitude with plastic amplitude (c); and hysteresis loop area (d)
Table 3 Constants in Manson-Coffin and Basquin relationships
Materials T, K r0
f=Ee0
fb
0
c
0
K (MPa) n
P92 293 0.00330 1.1962 0.04954 0.64840 705.7 0.0750
P92 873 0.00281 0.2771 0.07055 0.62612 475.4 0.0902
9Cr-1.4W-0.24V (Ref 40) 823 0.00300 0.4822 0.06000 0.64260 325.7 0.1008
Journal of Materials Engineering and Performance
3.3 Fracture Characteristics
Three representative samples that were stopped at 20% of
peak stress drop after fatigue at total strain amplitudes of 0.15,
0.30, and 0.60% are chosen to examine their fractography.
Figure 8shows the microstructure of longitudinal cross section
near the fracture, taken with an optical microscope at a lower
magnification. It can be observed that fatigue cracks are all
initiated from the surface of the samples. Meanwhile, the
number and length of cracks all increase with the increasing
total strain amplitude. In Fig. 8(d), secondary cracks are
initiated from the crack tip and propagated transgranularly into
the matrix. A comprehensive examination of fracture surfaces
deformed with different total strain amplitudes revealed
microscopic crack initiation at the surface followed by
stable microscopic crack growth. Figure 9shows SEM images
of fracture tested at different strain amplitudes at 873 K.
Particular attention is paid to observe the fatigue crack initiation
sites on individual samples as indicated by circles in Fig. 9.It
can be observed that most cracks initiate from the surface
defects (Fig. 9a) which might be caused by machine process,
and inclusions near the surfaces (Fig. 9b and c). The boundaries
between the matrix and inclusions are so weak. Under the
cyclic loading, the stresses are easily concentrated on these
sites, and eventually the fatigue cracks preferentially initiate at
these sites. The EDS results (Fig. 9d) show that the inclusions
(marked by circle) in Fig. 9c are aluminum oxides that are
produced during the casting process.
In addition, the scanning electron micrographs of the
stable crack growth region and the high magnification image
of the corresponding squared areas at different total strain
amplitudes at 873 K are shown in Fig. 10. The direction of
fatigue crack propagation is marked by the long white arrows
which are perpendicular to the loading direction. At the total
strain amplitude of 0.15% (Fig. 10a), it can be observed that
linear secondary cracks and fatigue striations are oriented
normal to the main crack propagation direction and tear ridges
are oriented parallel to the crack growth direction. It is also
important to note that fatigue cracks can bypass the inclusion
(pointed by white cross) in the matrix during growth along the
particle-matrix interface and then branch into some small
cracks. The result of EDX shows that the inclusions are
aluminum oxides which are produced during casting process.
The micro-voids can nucleate at the particle-matrix interface
and grow under the stress field of cracking during fatigue
cycling (Ref 41). In the high magnification image of
Fig. 10(A), it can be seen that secondary cracks are formed
Fig.8 Optical micrograph of longitudinal section of failed samples test under different strain amplitudes. (a, d) 0.15%; (b) 0.30%; (c) 0.60%
Journal of Materials Engineering and Performance
along fatigue striations and ended at the tear ridges and the
average width of striation at the stable crack growth area is
about 1.017 lm/cycle. At the total strain amplitude of 0.30%
(Fig. 10b), it appears that some secondary cracks are connected
together under the higher cyclic stress. In the high magnifica-
tion image of Fig. 10(B), the average width of striation at the
stable crack growth area is measured about 1.254 lm/cycle. An
interesting feature observed on the fracture of a sample tested at
the total strain amplitude of 0.60% (Fig. 10c) is the occurrence
of tire-like stripes and some parabolic-like secondary cracks.
Furthermore, the average width of striation is about 1.908
lm/cycle. The relationship between striation spacing, current
crack length, and plastic strain range is similar to that proposed
by Wareing (Ref 42), which is given by
S¼AðDepÞca;ðEq 4Þ
where Sis the striation spacing at a current crack length a,in
a test with plastic strain range Dep,A;and care constants.
The relationship between S/a ratio with plastic strain range is
graphically illustrated in Fig. 11. It shows that the striation
spacing is increasing linearly as the fatigue crack grows under
certain strain amplitude. The fitted equation is shown as fol-
lows
S¼1:4103ðDepÞ0:27aðEq 5Þ
The schematic diagrams of the mechanism of secondary
cracking during the fatigue tests are shown in Fig. 12. During
the fatigue crack propagation, a notch-like striation (Ref 43)
usually occurs by a repeated crack blunting and sharpening
process due to the slip of dislocations in the plastic zone at the
fatigue crack tip (Ref 44). The plastic flow via dislocation glide
at the tip of crack occurs to release the stress around the crack
tip, but in this case the stress relaxation might not be
sufficiently fast. As a result, a significant tensile stress acts
on the notch-like striation and the microcracks are easy to
initiate at the striation. It has been reported that PAGB and lath
boundaries are favorable sites for microcrack nucleation (Ref
41) (Fig. 12b) for this kind of ferritic-martensitic steel and the
results can be understood in terms of irreversible cyclic strain
Fig.9 Scanning electron micrographs of the fatigue crack initiation region with different total strain amplitudes. (a) 0.15%; (b) 0.30%; (c)
0.60% (d) The EDS result of inclusion (marked by cycle in c)
Journal of Materials Engineering and Performance
due to the formation of pile-up occurring preferentially near the
surface (Ref 45). During the fatigue test, it is known that most
of the plastic strains are carried by slip bands (SBs) and the SBs
may become carriers to transport the defects, such as disloca-
tions and vacancies, from the grain interiors to the grain/lath
boundaries. With further cyclic deformation, more and more
residual dislocations are piled up, as a result, some microcracks
gradually appear in different sites of these interface, as seen in
Fig. 12b. The carbides along the grain/lath boundaries can
accelerate the rate of microcrack nucleation, considering the
weakness of interface between carbide and matrix. During the
cyclic loading, when the notch-like striation occurs exactly at
the grain/lath boundaries, as seen in Fig. 12(c), the microcracks
subsequently propagate by their linking up and conjoining.
Finally, the bigger secondary crack appears. The parabolic-like
secondary cracks occur when the microcracks lying at primary
austenitic grain boundaries are connected together under higher
cyclic loading.
3.4 TEM Observation
Figure 13 shows the microstructures for as-received P92
steel and after fatigue failure at total strain amplitude of 0.4% at
873 K. The tempered martensitic structure contains martensite
Fig.10 Fracture micrographs of the stable crack growth region under different total strain amplitudes. (a, A) 0.15%; (b, B) 0.30%; (c, d) 0.60%
Journal of Materials Engineering and Performance
laths of several microns in length and approximately 0.3-0.5
lm in width. The discontinuous M
23
C
6
carbides lying along the
PAGB and lath boundaries are chromium enriched, while the
much fine particles distributed intragranularly are MX-type
carbides (Fig. 13c) enriched in niobium and vanadium (Ref
27). In the as-received state, the high density dislocation and
DN observed in the martensite lath are formed by the fierce
interactions between austenite grain boundaries and martensitic
laths during the process of martensitic transformation. After
fatigue test at 873 K, the martensite laths become wider as the
subgrain boundaries (SGB) disappear. The growth of subgrain
is a result of the elimination of low-angle boundaries which
contains subgrain and lath boundaries (Maxime Sauzay et al.
(Ref 46)). Much more free dislocation areas within the laths can
be seen (Fig. 13b and d) which indicates a significant decline of
dislocation densities in the grains. Cyclic softening is in general
attributed to the occurrence of dynamic recovery due to the
rearrangement of dislocations. The formation of dislocation
wall substructures through recovery micro-mechanisms leads to
an increase in the mean free path for dislocations (Ref 47).
Dislocations slipped out of the laths contribute to the rapid
softening at the beginning of the test, which is manifested by
the rapid decline of friction stress (Ref 28) (Fig. 14). However,
some dislocations could also be observed in laths, it is also
clear that these dislocations are pinned by precipitates inside the
subgrains (Fig. 13e). The in-line carbides within the subgrain
indicate where a previous lath boundary was completely
disappeared during cycling (Fig. 13e). As cyclic deformation
proceeded beyond the early rapid softening stage, the disloca-
tion structure is transformed into a CS with lower energy (Ref
48). Laird et al. (Ref 49) have indicated that cell formation
during fatigue cycling is triggered by the onset of multiple
glides. The fine carbides located in CS can enhance multiple
slip and induce subsequent generation of additional cell walls
through the interaction between the gliding dislocations on
different slip planes. The interactions and formation of CSs as
well as the interaction between carbides and dislocations
(Fig. 13e) can weaken the cyclic softening effect and manifest
as the decline of softening rate during the saturation stage.
Since the early work of Cottrell (Ref 50) and others (Ref 51,
52), it shows that the flow stress contains two parts: the short
distance interaction stress ‘‘friction stress’’, r
f
, and the long
distance interaction stress ‘‘back stress’’, r
b.
The friction stress
is related to the initial internal structure of the material, such as
the movement of dislocation. On the other hand, the back stress
is related to the density of long-range obstacles such as the
subgrain boundaries. As can be seen in Fig. 14, the curve of
friction and back stress shows analogs softening behavior as the
peak tensile stress curve after the first rapid softening stage. It
indicates that the softening behavior at this stage is the
comprehensive action effect between the movement of dislo-
cations and the annihilation of low-angle boundaries. Carbide
coarsening is also a common phenomenon in heat-resistant
steels during long-term aging at high temperature, and this
process would be accelerated obviously in stress field. The
relatively coarse carbides (about 250 nm in length and 100 nm
in width), surrounded by tangled dislocation, is confirmed to be
M
23
C
6
by electron diffraction and energy spectrum analysis
Fig.12 Schematic diagrams of the mechanism of secondary cracking during the cyclic loading
Fig.11 The relationship between striation spacing, current crack
length, and plastic strain range
Journal of Materials Engineering and Performance
(Fig. 13A). The carbide coarsening along grain/lath boundaries
maybe correlated with pronounced cyclic softening occurring
and be the favorable sites for microcrack nucleation. It has been
reported that the M
23
C
6
carbides along grain boundaries in
ferritic-martensitic steel will be coarsened and spheroidized
after long-time aging, whereas most of the martensite laths are
still stable and not decomposed (Ref 10,12,53), which means
the decomposition of martensite laths and the growth of
Fig. 13 TEM observations of as-received P92 steel and after fatigue failure (a, c) As-received; (b, d, e) After fatigue failure with total strain
amplitude of 0.4%
Journal of Materials Engineering and Performance
subgrains are mainly caused by the cyclic stress under high
temperature.
4. Conclusions
Based on the results of LCF test for P92 steel at 873 K and
fracture characteristics and microstructure evolution, the main
conclusions are drawn as follows:
(1) The S-N curves at different total strain amplitudes from
0.15 to 0.6% well follow the Coffin-Manson relation-
ship. The fatigue life can be expressed as a function of
the total strain amplitude:
Det
2¼0:00281 ð2NfÞ0:07055 þ0:2771ð2NfÞ0:62612
(2) Cyclic stress response of the P92 steel at 293 and 873
K reveals initial hardening at high strain amplitude dur-
ing the first number of cycles (1-5 cycles) of fully re-
versed cyclic straining followed by progressive
softening to failure, while it exhibits a continuous soft-
ening to failure at low strain amplitude. The fatigue life
is decreased while the amount of softening is enhanced
with increasing total strain amplitude and temperature.
Analysis of stabilized hysteresis loops at different strain
amplitudes show that the material exhibits a non-Masing
behavior. The relationship of strain amplitude with plas-
tic strain amplitude and hysteresis loop area can be ex-
pressed
Dep
2¼0:90762 Det
2
0:12319
Wh¼0:00876 Det
2
0:00137
(3) The fatigue cracks initiated from the surface and the
number of crack nucleation sites increased with the ris-
ing total strain amplitude. The formation of secondary
cracks is attributed to the connection of microcracks
nucleating at the grain/lath boundaries. The striation
spacing, current crack length, and plastic strain range
have a relationship:
S¼1:4103ðDepÞ0:27a
(4) The rapid softening stage at the beginning is mainly at-
tributed to the significant decrease of dislocations. The
interaction between carbides and dislocations as well as
the formation of CS inhibit the cyclic softening which is
manifested as the decline of softening rates when the
stress comes to saturation stage. The growth of sub-
grains during fatigue is mainly caused by the cyclic
stress collaborated with high temperature.
Acknowledgments
The authors gratefully acknowledge the financial support of the
National Nature Science Foundation of China (No. 50871076) and
the Key Project of Shanghai Science and Technology (No.
10521100500).
Reference
1. Y. Yi, B. Lee, J. Kim, and J. Jang, Corrosion and Corrosion Fatigue
Behaviors of 9Cr Steel in a Supercritical Water Condition, Mater. Sci.
Eng A, 2006, 429(1-2), p 161–168
2. M. Toloczko, M. Hamilton, and S. Maloy, High Temperature Tensile
Testing of Modified 9Cr-1Mo After Irradiation with High Energy
Protons, J. Nucl. Mater., 2003, 318, p 200–206
3. J. Henry, X. Averty, Y. Dai, P. Lamagnere, J.P. Pizzanelli, J.J. Espinas,
and P. Wident, Tensile Properties of 9Cr-1Mo Martensitic Steel
Irradiated with High Energy Protons and Neutrons, J. Nucl. Mater.,
2003, 318, p 215–222
4. R. Kannan, V.S. Srinivasan, M. Valsan, and K.B.S. Rao, High
Temperature Low Cycle Fatigue Behaviour of P92 Tungsten Added
9Cr Steel, Trans. Indian Inst. Met., 2010, 63(2-3), p 571–574
5. P.J. Ennis and A. Czyrska-Filemonowicz, Recent Advances in Creep-
Resistant Steels for Power Plant Applications, Sadhana, 2003, 28(3-4),
p 709–730
6. A. Nagesha, R. Kannan, G.V.S. Sastry, R. Sandhya, V. Singh, and K.
Bhanu, Shankara Rao, Isothermal and Thermomechanical Fatigue
Studies on a Modified 9Cr-1Mo Ferritic Martensitic Steel, Mater. Sci.
Eng. A, 2012, 554, p 95–104
7. M.F. Giordana, I. Alvarez-Armas, and A. Armas, On the Cyclic
Softening Mechanisms of Reduced Activity Ferritic-Martensitic Steels,
Steel Res. Int., 2012, 83(6), p 594–599
8. P. Marmy and T. Kruml, Low cycle fatigue of EUROFER 97, J. Nucl
Mater., 2008, 377, p 52–58
9. V. Shankar, V. Bauer, R. Sandhya, M.D. Mathew, and H.J. Christ, Low
Cycle Fatigue and Thermal-Mechanical Fatigue Behavior of Modified
9Cr-1Mo Ferritic Steel at Elevated Temperatures, J. Nucl. Mater., 2012,
420, p 23–30
10. I. Fedorova, A. Belyakov, P. Kozlov, V. Skorobogatykh, I. Shenkova,
and R. Kaibyshev, Laves-Phase Precipitates in a Low-Carbon 9% Cr
Martensitic Steel During Aging and Creep at 923 K, Mater. Sci. Eng A,
2014, 615, p 153–163
11. Y.T. Xu, M.J. Wang, Y. Wang, T. Gu, L. Chen, X. Zhou, Q. Ma, Y.M.
Liu, and J. Huang, Study on the Nucleation and Growth of Laves Phase
in a 10% Cr Martensite Ferritic Steel After Long-Term Aging, J. Alloys
Compd., 2015, 621, p 93–98
12. A. Kipelova, A. Belyakov, and R. Kaibyshev, The Crystallography of
M
23
C
6
Carbides in a Martensitic 9% Cr Steel After Tempering, Aging
and Creep, Philos. Mag., 2013, 93, p 2259–2268
13. A. Plumtree and H.A. Abdel-Raouf, Cyclic Stress-Strain Response and
Substructure, Int. J. Fatigue, 2001, 23, p 799–805
14. B. Fournier, M. Sauzay, C. Ca/s, M. Noblecourt, M. Mottot, L. Allais,
and A. Pineau, Creep-Fatigue Interactions in a 9Cr-1Mo Martensitic
Fig.14 The evolution of the tensile stress, the friction stress, and
back stress presented with total strain amplitude of 0.4% at 873 K
Journal of Materials Engineering and Performance
Steel: Part 1. Mechanical Test Results, Metall. Mater. Trans. A, 2009,
40, p 321–329
15. T. Fukuda, R. Sugiura, A.T. Yokobori, and T. Matsuzaki, Character-
isation of Crack Growth Behavior for W-Added 12% Ferritic Heat
Resistant Steel Under the Creep-Fatigue Interaction Conditions, Mater.
High Temp., 2015, 32, p 280–288
16. Standard Specification for Seamless Ferritic Steel-Steel Pipe for High-
Temperature Service, A335, Annual Book of ASTM Standard, 2003
17. V.K. Sikka, C.T. Ward, and K.C. Thomas, Modified 9Cr-1Mo
Steel—An Improved Steel for Steam Generator Application. Ferritic
steels for high temperature applications, Proceedings of ASM Interna-
tional Conference on Production, Fabrication, Properties and Appli-
cation of Ferritic Steels for High Temperature Applications, A.K.
Khare, Ed., Warren, PA, 6-8 October 1981. Metals Park (OH): ASM,
1983. p 65-84
18. J. Pesicka, R. Kuzel, A. Dronhofer, and G. Eggeler, The Evolution of
Dislocation Density During Heat Treatment and Creep of Tempered
Martensite Ferritic Steel, Acta Mater., 2003, 51, p 4847–4862
19. M. Hayakawa, T. Hara, S. Matsuoka, and K. Tsuzaki, Microstructural
Observation of Tempered Martensite in Medium-Carbon Low-Alloy
Steel by Atomic Force Microscopy, J. Jpn. Inst. Met., 2000, 64(6), p
460–466 (in Japanese)
20. M. Hayakawa, S. Matsuoka, and K. Tsuzaki, Observations of Prior
Austenite Grain Boundaries and Carbides in the Same Area of
Tempered Martensite in Medium-Carbon Steel by Atomic Force
Microscopy, J. Jpn. Inst. Met., 2001, 65, p 734–741 (in Japanese)
21. K.B.S. Rao, M. Valsan, R. Sandhya, S.K. Ray, S.L. Mannan, and P.
Rodriguez, On the Failure Condition in Strain Controlled Low Cycle
Fatigue, Int. J. Fatigue, 1985, 7(3), p 141–147
22. M.F. Giordana, P.F. Giroux, I. Alvarez-Armas, M. Sauzay, A. Armas,
and T. Kruml, Microstructure Evolution During Cyclic Tests on
EUROFER 97 at Room Temperature. TEM observation and Modeling,
Mater. Sci. Eng. A, 2012, 550, p 103–111
23. P.C. Chakraborti and M.K. Mitra, Room Temperature Low Cycle
Fatigue Behavior of Two High Strength Lamellar Duplex Ferrite-
Martensite (DFM) Steels, Int. J. Fatigue, 2005, 27, p 511–518
24. X. Hu, L.X. Huang, W.G. Wang, Z.G. Yang, W. Sha, W. Wang, W.
Yan, and Y.Y. Shan, Low Cycle Fatigue Properties of CLAM Steel at
Room Temperature, Fusion Eng. Des., 2013, 88, p 3050–3059
25. M.F. Giordana, I. Alvarez-Armas, M. Sauzay, and A.F. Armas,
Mechanical Behavior and Microstructure Evolution of EUROFER97
During Low-Cycle Fatigue at Room Temperature and 550C, Key Eng.
Mater., P. Sandera, Ed., Jun 28-30, 2010 (Brno), Trans Tech
Publications, 2011, p 358-361
26. D.M. Li, K.W. Kim, and C.S. Lee, Low Cycle Fatigue Data Evaluation
for a High-Strength Spring Steel, Int. J. Fatigue, 1997, 19, p 607–612
27. V. Shankar, M. Valsan, K.B.S. Rao, R. Kannan, S.L. Manan, and S.D.
Pathak, Low Cycle Fatigue Behavior and Microstructural Evolution of
Modified 9Cr-1Mo Ferritic Steel, Mater. Sci. Eng. A, 2006, 437, p 413–
422
28. M.F. Giordana, I. Alvarez-Armas, and A. Armas, Microstructural
Characterization of EUROFER 97 During Low-Cycle Fatigue, J. Nucl.
Mater., 2012, 424, p 247–251
29. S. Sivaprasad, S.K. Paul, A. Das, N. Narasaiah, and S. Tarafder, Cyclic
Plastic Behaviour of Primary Heat Transport Piping Materials:
Influence of Loading Schemes on Hysteresis Loop, Mater. Sci. Eng.
A, 2010, 527, p 6858–6869
30. L. Borrego, L. Abreu, J. Costa, and J. Ferreira, Analysis of Low Cycle
Fatigue in AlMgSi Aluminium Alloys, Eng. Fail. Anal., 2004, 11,p
715–725
31. F. Ellyin and D. Kujawski, Plastic Strain Energy in Fatigue Failure, J.
Press. Vessel Technol., 1984, 106, p 342–347
32. D. Lefebvre and F. Ellyin, Cyclic Response and Inelastic Strain Energy
in Low Cycle Fatigue, Int. J. Fatigue, 1984, 6, p 9–15
33. X.W. Wang, J.M. Gong, Y.P. Zhao, Y.F. Wang, and M.H. Yu,
Characterization of Low Cycle Fatigue Performance of New Ferritic
P92 Steel at High Temperature: Effect of Strain Amplitude, Steel Res.
Int., 2015, 86, p 1046–1055
34. D.T. Raske and J.M. Morrow, Mechanics of Materials in Low Cycle
Fatigue Testing, Manual on low cycle fatigue testing, ASTM, STP 465,
Philadelphia (PA), ASTM, 1969. p. 1
35. R.W. Landgraf, J. Morrow, and T. Endo, Determination of the Cyclic
Stress-Strain Curve, J. Mater JMLSA, 1969, 4, p 176–188
36. O.H. Basquin, The exponential law of endurance tests, Proc ASTM
ASTEA, 1910, 10, p 625-630
37. L.F. Coffin and J.F. Tavernelli, The Cyclic Straining and Fatigue of
Metals, Trans. ASME, 1954, 76, p 931–950
38. S.S. Manson, Behaviour of Materials Under Conditions of Thermal
Stress, NACA, Tech Note-2933, 1954
39. D.T. Raske and J.D. Morrow, Manual on Low Cycle Fatigue Testing,
American Society for Testing and Materials, Philadelphia, 1969, p 1
40. V. Shankar, K. Mariappan, A. Nagesha, G.V. Prasad Reddy, R.
Sandhya, M.D. Mathew, and T. Jayakumar, Effect of Tungsten and
Tantalum on the Low Cycle Fatigue Behavior of Reduced Activation
Ferritic/Martensitic Steels, Fusion Eng. Des., 2012, 87, p 318–324
41. M.N. Batista, M.C. Marinelli, S. Herenu, and I. Alvarez-Armas, The
Role of Microstructure in Fatigue Crack Initiation of 9-12% Cr
Reduced Activation Ferritic-Martensitic Steel, Int. J. Fatigue, 2015, 72,
p 75–79
42. J. Wareing and H.G. Vaughan, The Relationship Between Striation
Spacing, Macroscopic Crack Growth Rate, and the Low-Cycle Fatigue
Life of a Type 316 Stainless Steel at 625 C, Met. Sci., 1977, 11(10), p
439–446
43. X. Gong, P. Marmy, L. Qin, B.Verlinden, M.W. Wevers, and M. Seefeldt,
Effectof Liquid Metal Embrittlementon Low Cycle FatigueProperties and
Fatigue Crack Propagation Behavior of a Modified 9Cr-1Mo Ferritic-
Martensitic Steel in an Oxygen-Controlled Lead-Bismuth Eutectic
Environment at 350 C, Mater. Sci. Eng. A,2014,618, p 406–415
44. N. Farabi, D.L. Chen, J. Li, Y. Zhou, and S.J. Dong, Microstructure and
Mechanical Properties of Laser Welded DP600 Steel Joints, Mater. Sci.
Eng. A, 2010, 527, p 1215–1222
45. D.A. Lados and D. Apelian, Fatigue Crack Growth Characteristic in
Cast Al-Si-Mg Alloys: Part 1. Effect of Processing Conditions and
Microstructure, Mater. Sci. Eng. A, 2004, 385, p 200–211
46. M. Sauzay, H. Brillet, I. Monnet, M. Mottot, F. Barcelo, B. Fournier,
and A. Pineau, Cyclically Induced Softening Due to Low-Angle
Boundary Annihilation in a Martensitic Steel, Mater. Sci. Eng A, 2005,
400-401, p 241–244
47. S. Apu, B.K. Kumawat, and J.K. Chakravartty, Low Cycle Fatigue
Behavior of a Ferritic Reactor Pressure Vessel Steel, J Nucl. Mater.,
2015, 462, p 273–279
48. Y.-J. Oh, T.-B. Hwang, and B.M. Shin, A Comparison of Low-Cycle
Fatigue Properties in Stainless Steel Types 347N and 316N, Mater.
Trans., 2005, 46, p 317–322
49. C. Laird, P. Charsley, and H. Mughrabi, Low-Energy Dislocation
Structures Produced by Cyclic Deformation, Mater. Sci. Eng., 1986,
81, p 433–450
50. A.H. Cottrell, Dislocations and Plastic in Flow in Crystal, Oxford
University Press, London, 1953, p 111
51. D. Kuhlmann-Wilsdorf, Dislocation Behavior in FatigueII. Friction
Stress and Back Stress as Inferred from an Analysis of Hysteresis
Loops, Mater. Sci. Eng., 1979, 37(2), p 111–120
52. J.I. Dickson, J. Boutin, and L. Handfield, A Comparison of Two
Simple Methods for Measuring Cyclic Internal and Effective Stress,
Mater. Sci. Eng., 1984, 64, p L7–L11
53. X. Hu, L.X. Huang, W. Yan, W. Wang, W. Sha, Y.Y. Shan, and K.
Yang, Low Cycle Fatigue Properties of CLAM Steel at Room
Temperature, Mater. Sci. Eng. A, 2013, 586, p 253–258
Journal of Materials Engineering and Performance
