Materials Science and Engineering A 534 (2012) 288–296

Contents lists available at SciVerse ScienceDirect

Materials Science and Engineering A

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

Role of microstructure in the low cycle fatigue of multi-phase steels

Timothy Hilditch a,∗ , Hossein Beladi b , Peter Hodgson b , Nicole Stanford b
a
School of Engineering, Deakin University, Pigdons Rd, Waurn Ponds, VIC 3217, Australia
b
Centre for Material and Fibre Innovation, Deakin University, Pigdons Rd, Waurn Ponds, VIC 3217, Australia

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

Article history: The low cycle fatigue (LCF) behaviour of several commercially-produced multiphase steels was studied;
Received 19 September 2011 including dual-phase (DP) and transformation induced plasticity (TRIP). In addition, a novel TRIP980
Received in revised form hybrid microstructure was examined that consisted of coarse ferrite grains along with low temperature
11 November 2011
bainite regions interspersed with retained austenite. Fully reversed strain controlled fatigue tests were
Accepted 16 November 2011
conducted on the different steels to determine the cyclic stress response and strain to failure. The effects
Available online 1 December 2011
of the cyclic deformation on the microstructures were analysed using electron backscattered diffraction
(EBSD) and X-ray diffraction (XRD). Results showed that the initial cyclic hardening behaviour and low
Keywords:
EBSD
cyclic softening ratio observed in the TRIP steels was not necessarily due to austenite to martensite
X-ray diffraction transformation. Differences between the austenite transformation behaviour of the conventional and
Steel novel hybrid TRIP microstructures was related to the different surrounding phases and the size of the
Austenite retained austenite.
Fatigue © 2011 Elsevier B.V. All rights reserved.
Phase transformation

1. Introduction result in well-formed dislocation substructures in the ferrite that

become more refined with higher strain amplitudes [4,5]. It has
Low-carbon multiphase steels have been frequently used in been reported that an increase in martensite content increases the
applications that require a good combination of strength and fatigue life for relatively small martensite volume fractions [6,7],
ductility, such as body structures in the automotive industry. Dual- most likely due to a more uniform distribution of the dislocation
phase (DP) steels were developed with a high volume fraction substructure in the ferrite [8].
of ferrite to ensure a relatively low yield point and good work The presence of retained austenite in the microstructure can
hardening rates at low strains, as well as regions of martensite decrease fatigue life in some instances [9,10], while enhancing
to increase tensile strength. In transformation-induced plasticity fatigue life in others [10]. Austenite to martensite transformation
(TRIP) steels, metastable austenite is retained in the microstructure can decrease fatigue life for steels with high volume fractions of
that will transform to martensite upon straining to further increase austenite, such as metastable austenitic steels, provided the applied
strength and ductility. Conventional advanced high strength steel strain amplitude is above a certain threshold for the transforma-
(AHSS) grades such as DP and TRIP have been extensively utilised tion to occur [11,12]. The newly formed martensite can provide a
in the automotive industry due to their good room temperature path for fast crack propagation. The improvement in fatigue life
formability and crash performance. Currently there is a signif- for multi-phase steels with retained austenite is associated with
icant effort into developing novel multiphase microstructures, the transformation of retained austenite to martensite in front of
especially those based on retained austenite to further improve the crack tip [3,13]. This transformation is triggered by the strain
the strength/ductility combination [1,2]. While the developmental energy in front of a crack tip and has been linked with a delay of
focus is often on formability, it is vital to understand the impact of crack initiation [10], as well as retardation of stage I crack propa-
cyclic loading and/or deformation on the microstructures of these gation [10].
materials to ensure good performance in service. While it is clear that retained austenite does typically transform
It has been shown in low cycle fatigue that strain partioning can in front of a propagating crack tip, the impact of the retained austen-
occur, where strain is concentrated in the softer phase [3]. In the ite on the fatigue performance in low cycle fatigue is not necessarily
case of multi-phase steels, such as DP steels, repeated plastic strains clear, especially for steels with relatively low retained austenite
contents (below 20%). This understanding is necessary to aid in
the design of novel complex multi-phase steel microstructures
∗ Corresponding author. Tel.: +61 3 5227 2265; fax: +61 3 5227 2028. that have good fatigue-resistance, especially for the automotive
E-mail address: [email protected] (T. Hilditch). industry. The present study examines the fatigue performance and

0921-5093/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.msea.2011.11.071
 T. Hilditch et al. / Materials Science and Engineering A 534 (2012) 288–296 289

Table 1 Electron microscopy was carried out on either a LEO 1530 SEM
Chemical composition of steels (wt%).
or a Supra VP, both of which were operated in high current mode
Steel C Mn Si Mo to improve backscatter electron yield. The EBSD system used was
TRIP590 0.06 1.15 1.53 0.00 an HKL Channel 5 system, and the data acquisition was carried out
TRIP780 0.23 1.70 1.36 0.00 at 20 kV. The backscattered imaging was done using a Zeiss angular
DP780 0.12 1.97 0.92 0.00 selective backscattered detector (ASB). This is a relatively new low
TRIP980 0.26 2.00 1.96 0.31 angle collection detector with greater spatial resolution over tradi-
tional high angle collection detectors. Images were taken at a small
working distance of ∼4 mm and an accelerating voltage of 20 kV.
microstructural development of different multi-phase steels. The
steels studied include several commercial AHSS, along with a lab-
3. Results
oratory developed hybrid steel microstructure with an excellent
combination of strength and ductility containing ferrite, retained
3.1. Initial microstructures
austenite and a high volume fraction of low-temperature bainite.

The starting microstructures of the four steels examined in this

2. Experimental procedure
study are shown in Fig. 1. The quantitative measurements for the
various microstructural features in the steels were made with opti-
2.1. Materials
cal microscopy, EBSD and XRD. It can be seen that TRIP780, DP780
and TRIP590 all have quite similar morphologies; comprising of
The materials examined in the study included a commercially
at least 60% ferrite with a grain size ranging from 2 to 7 ␮m.
produced uncoated TRIP steel (TRIP590), a commercially produced
The one significant difference between these three alloys was that
uncoated dual-phase steel (DP780), and two laboratory produced
the dual phase steel does not contain retained austenite. TRIP590
TRIP steels (TRIP780 and TRIP980). Compositions of the steels are
contained approximately 5.5 wt.% retained austenite and TRIP780
shown in Table 1.
contained approximately 15 wt.% retained austenite. TRIP980, on
Engineering stress–strain properties were determined in accor-
the other, had a uniquely different microstructure consisting of
dance with AS 1391-1991 on a 100 kN screw-driven load frame at a
large polygonal ferrite grains (∼25 ␮m average grain diameter) dis-
constant cross-head velocity of 1.5 mm/min. Specimens had a nom-
persed between a bainite matrix interspersed with small retained
inal gauge length of 25 mm and width of 6 mm and were tested in
austenite regions (∼8 wt.%).
the longitudinal direction using a non-contact video extensometer
to measure strain. All specimens had a nominal thickness of 2 mm. 3.2. Static stress–strain response

2.2. Fatigue testing The engineering stress–strain response for the studied steels is
shown in Fig. 2. TRIP590 and TRIP780 both had similar high total
Fully reversed strain amplitude (R = −1) fatigue tests were elongations, with TRIP780 having significantly higher strength and
performed in accordance with ASTM E606-92 on a 25 kN servo- uniform elongation. This increase in strength and uniform elonga-
hydraulic load frame. Specimens had a 2 mm width and 7.9 mm tion can be attributed to the higher retained austenite content of
parallel length. Strain amplitudes in the range from 0.004 to 0.010 this steel. DP780 and TRIP980 both had similar yield strength and
were applied in strain control mode using a 5 mm gauge length total elongation; however TRIP980 had a higher uniform elongation
clip-on extensometer at a constant strain rate of 0.02 s−1 . To pre- and tensile strength. Both DP780 and TRIP980 displayed continu-
vent buckling under large compressive strains, anti-buckling guides ous yielding with very high initial work hardening rates, in contrast
were used. Epoxy was placed under the extensometer edges to to TRIP590 and TRIP780 that showed both discontinuous yielding
prevent slippage or crack initiation at these points. Failure was and comparably lower initial work hardening rates.
determined as a load drop of 10% from that of the stabilised hys-
teresis loop. 3.3. Strain life

2.3. X-ray diffraction The total strain amplitude versus reversals to failure (Fig. 3)
shows that DP780 and TRIP980 had the highest strain life at low
The percentage of retained austenite was measured using X-ray total strain amplitudes (0.004). At such low total strain amplitudes
diffraction (XRD). Measurements of the X-ray spectra were made the number of reversals to failure will be strongly related to the
using Cu K␣ radiation over a 2 angular range of 41–92◦ . The area yield strength of the materials, as higher yield strength means a
of each peak was measured from the spectra using X’pert High- larger elastic strain component in the applied total strain ampli-
score Plus software. Estimates in the volume fraction were made tude. Since it is the plastic strain component that contributes most
using the direct comparison technique described in [14]. Conse- to the damage leading to failure, it follows that strain life should be
quently the area available for XRD was quite small. To ensure assisted by a higher yield strength for a given total strain amplitude.
self-consistency in the data, measurements of the as-received Looking now to the higher strain amplitudes, the strain life of
materials were made on the same size and dimension samples. TRIP980 was lower than the other steels (as shown by the low slope
of the trend line in Fig. 3). At higher total strain amplitudes, the
2.4. Electron microscopy plastic strain component becomes dominant and typically materi-
als with higher ductility, such as TRIP780 and TRIP590 in this study,
Electron backscattered diffraction (EBSD) was carried out on have a higher strain life [15].
selected specimens to further examine the changes in microstruc- It is clear from the plastic strain amplitude versus reversals to
ture that occur during fatigue loading. Samples were prepared for failure (Fig. 4) that TRIP980 had the lowest life at all plastic strain
EBSD by mounting and polishing using standard metallographic amplitudes. The plastic strain amplitude has been calculated as the
techniques. This was followed by ∼2 min polishing with OPS. This plastic strain component of the total strain amplitude at the half life
was a sufficient sample preparation method to produce high quality [16]. The Coffin–Manson equation was used to calculate the fatigue

EBSD data as well as clear backscattered electron images. ductility coefficient (ε f ) (εf ) and exponent (c), shown in Table 2,
290 T. Hilditch et al. / Materials Science and Engineering A 534 (2012) 288–296

Fig. 1. Initial microstructures of the examined steels showing (a) TRIP590, (b) TRIP780, (c) DP780, and (d) TRIP980. Note lower magnification in (d).

1200 0.01
TRIP980 TRIP590

1000 TRIP780
Engineering Stress(MPa)

TRIP780
Plastic Strain Amplitude

DP780
800
DP780 TRIP980

600
0.001

400 TRIP590

200

0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0001
1000 10000 100000
Engineering Strain
Reversals to failure (2Nf)
Fig. 2. Engineering stress–strain response of the examined steels.
Fig. 4. Plastic strain amplitude versus reversals to failure for the examined steels.

TRIP590
which also includes a TRIP780 and DP590 from a previous study by
TRIP780 Hilditch et al. [4].
0.01 DP780 It is worth noting that the steels containing retained austenite
Total Strain Amplitude

TRIP980 all have a higher exponent than the two dual-phase steels, showing
that their fatigue life is comparatively better at higher plastic strain
amplitudes. The higher slope of the Coffin–Manson plot, or fatigue

Table 2
Coffin–Manson parameters for the different steels.

Steel Fatigue ductility Fatigue ductility


exponent (c) coefficient (εf )

DP590 [4] 0.52 0.22

0.001 DP780 0.48 0.23
100 1000 10000 100000 TRIP590 0.58 0.42
Reversals to Failure (2Nf) TRIP780 [4] 0.54 0.28
TRIP780 0.66 0.80
Fig. 3. Total strain amplitude versus reversal to failure for the examined steels. TRIP980 0.58 0.18
 T. Hilditch et al. / Materials Science and Engineering A 534 (2012) 288–296 291

a 0.9 b 0.9

0.8 0.8

Fatigue Ductility Coefficient

Fatigue Ductility Coefficient

0.7 0.7

0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.10 0.20 0.30 0.40 0.50
Uniform Elongation Total Elongation

Fig. 5. Fatigue ductility coefficient versus (a) uniform tensile elongation and (b) total tensile elongation for the different steels. Note that TRIP steels are represented by a
closed marker and DP steels by an open marker.

ductility exponent can be explained, in part, by the higher uniform stress has been plotted as a function of the plastic strain amplitude.
ductility of the TRIP steels. Low cycle fatigue is an accumulation of Plastic strain was used as it is considered the primary factor influ-
small tensile plastic strains, with the fatigue ductility coefficient an encing the hardening behaviour. The plot shows the general trend
extrapolation of the plastic strain–life curve to 1 reversal to failure. of an increase in cyclic hardening with increasing plastic strain
While tensile ductility is not necessarily a reliable measure of strain amplitude. While most of the steels showed a similar level of cyclic
life in low cycle fatigue performance, materials with a higher ductil- hardening for a given plastic strain amplitude, TRIP980 hardened
ity tend to have a higher strain life. The plot in Fig. 5 shows a general significantly more than the other steels. This can at least partially
trend that the fatigue ductility coefficient increases with increasing be attributed to the static tensile behaviour, as this steel had the
tensile elongation (uniform and total). The trend is more marked highest work hardening rate at strains up to 5%. Note that DP590
with total elongation and, in fact, the fatigue ductility coefficient from Hilditch et al. [4] is not included in Fig. 7 as it softened at all
(extrapolation to one reversal) is quite similar to the total elonga- applied strain amplitudes.
tion for most of the steels. The obvious exception was TRIP780 that Fig. 8 shows the cyclic softening ratio versus the plastic strain
showed a significantly higher fatigue ductility coefficient than the amplitude for all steels, with the cyclic softening ratio calculated as
other steels and higher than its own total elongation would sug- [18]:
gest. This result appears to be linked more with poor performance
max −  
 Nf
in terms of strain life at low plastic strain amplitudes, which artifi- max 2
cially inflates the slope of the data. This suggests that the data may softening ratio = (1)
not be linear at higher plastic strain amplitudes. max

where  max is the peak stress amplitude and   is the stress

 Nf
max 2
3.4. Cyclic stress response
amplitude at the half life.
The cyclic stress amplitude response at high strain amplitudes There was a significant difference between the softening
for all steels showed initial cyclic hardening, followed by cyclic soft- behaviour of the TRIP steels compared to the DP steels. The TRIP
ening (Fig. 6a–d). At lower strain amplitudes similar behaviour was steels all show an increased resistance to softening compared with
apparent for three of the steels (DP780, TRIP590 and TRIP980); the DP steels, which has also been noted in previous studies [19,20].
whereas TRIP780 cyclically softened from the first cycle. Initial At very low plastic strain amplitudes, TRIP590 softened to a simi-
cyclic hardening at low strains is more likely to occur for steels lar extent to the DP steels. It is worth noting that TRIP590 had the
with a high initial tensile work hardening rate, such as DP780 lowest amount of retained austenite of the examined TRIP steels,
and TRIP980, as cyclic hardening is caused by an increase in the suggesting that at very low strain amplitudes the austenite does
dislocation density and their subsequent interaction. Softening fol- not play a noticeable part in the cyclic stress response. The cyclic
lowing initial hardening normally results from the formation of softening ratio decreased with increasing plastic strain amplitude
lower energy dislocation substructures, such as cells or microbands for the DP steels, while the behaviour of the TRIP steels was more
within the ferrite [17]. Softening that occurs from the first cycle at scattered.
low strain amplitudes is thought to be due to the formation and
spreading of dislocation sources along the specimen gauge length 3.5. Retained austenite measurement
during the first cycle.
Both the initial cyclic hardening behaviour and cyclic softening X-ray diffraction (XRD) was used to measure the retained
ratio for all steels are shown in Figs. 7 and 8. Included in Figs. 7 and 8 austenite volume fraction of TRIP980 and TRIP780 at different
are other data for two different steels, a TRIP780 and DP590 from stages during fatigue at total strain amplitudes of 0.004 and 0.008.
a previous study by Hilditch et al. [4]. In Fig. 7 the amount of cyclic The retained austenite volume fractions in Fig. 9 were measured
hardening from the end of the first tensile reversal to the peak cyclic in the as-received condition, after the first cycle, at the peak
292 T. Hilditch et al. / Materials Science and Engineering A 534 (2012) 288–296

600 700
0.008 0.009
0.008
0.004

Cyclic Stress Amplitude (MPa)

550 650 0.006
Cyclic Stress Amplitude (MPa)
0.006 0.004

500 600

450 550

400 500

350 450

300 400
1 10 100 1000 10000 1 10 100 1000 10000
Cycles (Nf) Cycles (Nf)
(a) TRIP590 (b) TRIP780
800 1000
0.009 0.008
950 0.007
750 0.008
Cyclic Stress Amplitude (MPa)

Cyclic Stress Amplitude (MPa)

0.006
0.006 900 0.005
700 0.004
0.004
850
650
800

600 750

550 700

650
500
600
450
550
400
500
1 10 100 1000 10000
1 10 100 1000 10000
Cycles (Nf) Cycles (Nf)
(c) DP780 (d) TRIP980
Fig. 6. Cyclic stress amplitude versus number of cycles for (a) TRIP590, (b) TRIP780, (c) DP780 and (d) TRIP980 at a range of strain amplitudes between 0.004 and 0.009.

140 0.20
TRIP590 TRIP590
TRIP780 0.18 DP590 [4]
120 TRIP780 [4] TRIP780
DP780 0.16 TRIP780 [4]
Cyclic Hardening (MPa)

TRIP980 DP780
100
Cyclic Softening Ratio

0.14 TRIP980

80 0.12

0.10
60
0.08

40 0.06

0.04
20
0.02
0
0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.00
0 0.002 0.004 0.006 0.008
Plastic Strain Amplitude
Plastic Strain Amplitude
Fig. 7. Cyclic hardening (calculated as the difference between the first cycle stress
and peak cyclic stress versus the plastic strain amplitude for the four steels in this Fig. 8. Cyclic softening ratio versus plastic strain amplitude for the four steels from
study and TRIP780 from [4]). the present study as well as DP590 and TRIP780 from [4].
 T. Hilditch et al. / Materials Science and Engineering A 534 (2012) 288–296 293

16 The retained austenite volume fraction of TRIP980 showed no

TRIP780 - 0.004
variation during fatigue from that measured in the as-received
Retained Austenite Volume Fraction (%)

TRIP780 - 0.008
14
TRIP980 - 0.004 material for either of the strain amplitudes.
TRIP980 - 0.008
12
3.6. Microstructural development during cyclic loading
10

Samples of the TRIP780 and TRIP980 were examined using

8
backscattered imaging and EBSD to characterise the substructural
6
development of these steels during cyclic loading. The TRIP980
steel showed a significant increase in dislocation density in the fer-
4 rite at the peak stress cycle (Fig. 10), and this is likely to result
in the observed cyclic hardening. At the half life, a well-defined
2 dislocation substructure had been formed with a cell-wall spacing
of approximately 400–500 nm. SEM observation showed no dis-
0
cernable substructure development in the ferrite for TRIP780. The
As-Received First Cycle Peak Cycle Half Life
formation of such a well-defined substructure in TRIP980 can be
Fig. 9. Austenite volume fractions – measurement error is approximately 1% (com- explained by the larger ferrite grains that allow a larger plastic
bination of equipment uncertainty and specimen variation). Note that TRIP780 did activity within them [21]. Conversely, the very small ferrite grains
not show any cyclic hardening at the 0.004 strain amplitude, and thus for compar- in TRIP780 do not allow sufficient plastic activity to generate a
ison sake, the peak stress was taken at 31 cycles (which was the same number of
well-defined substructure.
cycles as for the peak stress at the 0.008 strain amplitude for TRIP780).
All samples were examined optically for evidence of surface
relief after fatigue loading. There were no features of note in
any sample except for the TRIP980 steel. In that case there was
cyclic stress and at the half life. At the 0.008 strain amplitude, noticeable surface relief and an example after fracture at a total
TRIP780 had a large amount of austenite transform in the first cycle strain amplitude of 0.008 is shown in Fig. 11. It is clear from
(∼3.5%), with another ∼1.5% transforming between the first cycle Fig. 11 that there was a significant amount of strain partitioning in
and peak stress. Although the retained austenite volume fraction TRIP980, with strain being concentrated in the large, blocky ferrite
was not measured after 1 cycle at 0.004, the volume fraction would grains.
have been higher than after a first cycle strain of 0.008. Thus it Optical microscopy and electron microscopy were also used to
can be assumed that a similar amount of austenite transformed examine the steels for evidence of microcracks in the sample inte-
between the first cycle and peak cycle at 0.004 as to that mea- rior as well as at the sample surface. This revealed that there was no
sured at 0.008. It is worth noting that the peak stress occurred evidence of cracking in the TRIP780 samples at either the peak or
at a significantly higher number of cycles for the 0.004 strain half cycle life. For the TRIP980 steel surface cracking was observed
amplitude than 0.008. There was no measured retained austenite in samples after half of the cycle life. In all cases observed the cracks
transformation between the peak stress and half life at either strain were restricted to the ferrite grains and showed significant plastic
amplitude. deformation in the region of the crack path (Fig. 12).

Fig. 10. Low angle backscatter SEM images of the ferrite grains in TRIP980 during fatigue at a total strain amplitude of 0.008 showing the as-received condition (a); and
subsequent substructure development after the peak cycle (21 cycles) (b), after the half-life (301 cycles) (c), and after fracture (675 cycles) (d).
294 T. Hilditch et al. / Materials Science and Engineering A 534 (2012) 288–296

TRIP980 can be attributed to the distribution and size of the

austenite within the microstructure shown in Fig. 13. The austen-
ite was interspersed throughout the predominantly bainitic phase
in the material. It is clear from the surface relief observations
shown in Fig. 11 that there was a significant amount of strain
partioning in TRIP980, with strain being concentrated in the large
blocky ferrite grains. The significantly higher yield strength of the
bainite/austenite regions means those phases only experienced
relatively small plastic strains (due to the large elastic strain com-
ponent). It has previously been discussed in the literature that
the immediately surrounding phases can affect retained austen-
ite stability as bainite does not transfer global stress and strain to
the retained austenite as effectively as either martensite or fer-
rite [23,24]. In addition, there is a dependence on the morphology
and size of the retained austenite itself, with smaller and needle,
or lath-like austenite more stable [24,25]. It can be seen in Fig. 13
Fig. 11. Optical image of the surface of TRIP980 after fatigue at a total strain ampli- that the approximate retained austenite grain size was an order
tude of 0.008 showing obvious surface relief from the large blocky ferrite grains. of magnitude smaller in the TRIP980 compared to the TRIP780.
Despite the lack of austenite transformation, there was signifi-
4. Discussion cant hardening in TRIP980 that can be attributed to the larger
ferrite grain size. Lucas and Gerberich [26] have shown that larger
4.1. Austenite transformation and cyclic hardening ferrite grains can result in a higher cyclic strain hardening expo-
nent due to the subgrain cell structure that develops during cyclic
TRIP780 underwent austenite-to-martensite transformation deformation.
during the initial hardening stage of the fatigue test at high strain
amplitudes; however, it only exhibited a relatively small amount 4.2. Cyclic softening resistance
of cyclic hardening (∼30 to 40 MPa). At the lower strain ampli-
tude, the TRIP780 actually softened slightly (∼15 MPa) from the In this study the softening ratio of the TRIP steels was noticeably
first cycle, despite the retained austenite that was transforming. lower than the dual-phase steels. Softening that occurs post-
This is not completely consistent with the reports of Sugimoto cyclic hardening typically results from the generation of additional
[22] that showed in a TRIP-aided multi-phase steel with 13% vol- mobile dislocations and the formation of a dislocation cell structure
ume fraction retained austenite that the initial cyclic hardening with lowered internal stress within the ferrite. This is followed by
increment increases significantly with increasing austenite trans- the formation of deformation bands within the grains and shear
formation. The lack of significant hardening for TRIP780Nb despite bands, or persistent slip bands that can develop across multiple
austenite transformation is potentially associated with the for- grains, eventually leading to failure. It has previously been observed
mation and spreading of dislocation sources in the ferrite. It is for both austenitic steels and TRIP steels that cyclic softening after
expected there are competing effects between the softening via the initial hardening phase is arrested through the formation of
dislocation source generation and hardening via the martensite martensite [20,22,27]. For TRIP780 in the present study, it cer-
formation. tainly appears that austenite transformation in the initial hardening
The lack of hardening in the TRIP780 contrasts with TRIP980 phase would be at least partially responsible for the low cyclic
that showed significantly higher cyclic hardening at all strain softening ratio.
amplitudes despite no retained austenite transformation (∼80 MPa TRIP980 also showed a high resistance to softening with a cyclic
at 0.004 and ∼120 MPa at 0.008). The lack of transformation in stabilised stress that was higher than the stress after the first cycle
(similar to TRIP780). This was despite no bulk retained austen-
ite transformation in the early stages of fatigue. Yokoi [20] has
suggested that TRIP steels have higher resistance to cyclic soft-
ening due to compressive residual stress in the surface region
caused by transformation retarding the propagation of microc-
racks. SEM observation in the current study of the TRIP780 fatigued
microstructures at the half life showed no evidence of microcracks.
Fig. 12 shows that crack initiation in TRIP980 occurred exclusively
in the ferrite grains, and where possible the crack path was through
the middle of the large ferrite grains. This mechanism of crack ini-
tiation has been observed in DP steels, where strain localises in the
ferrite, resulting in microcracks forming in slip bands within the
ferrite [8]. Thus neither the bulk transformation of austenite dur-
ing initial hardening nor the retardation of microcracks by retained
austenite transformation was responsible for the low cyclic soften-
ing ratio in TRIP980. The resistance of this steel to cyclic softening
is likely to be the result of one of two possibilities. It is possible that
the isolated ferrite grains were unable to form deformation bands
that crossed multiple grains, or else the highly refined ferrite sub-
structure resulting from the high plastic strain could not develop
into a lowered internal stress state.
Fig. 12. EBSD images of a crack originating at the surface of the TRIP980 sample At larger crack sizes than those shown in Fig. 12, it is expected
initiating and propagating exclusively through the ferrite. that the crack path would not be exclusively through the ferrite
 T. Hilditch et al. / Materials Science and Engineering A 534 (2012) 288–296 295

Fig. 13. EBSD image of the initial microstructure for TRIP980 (left) showing the retained austenite regions in yellow dispersed within the low temperature bainite matrix;
and TRIP780 (right) showing retained austenite closely an evenly distributed throughout the ferrite matrix. Note higher magnification in (right). (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of the article.)

in TRIP980, with the plastic zone likely to affect retained austenite 5. Conclusions
within the bainitic regions. Thus it is possible that some crack retar-
dation via local austenite transformation in front of larger cracks is The low cycle fatigue testing and subsequent microstructural
possible for this steel. It is not considered likely that this would have examination for a number of TRIP and DP steels has shown that:
a significant effect on the strain life, however, as for larger cracks
the increased plastic zone size and crack growth rate would not be • TRIP980 had a significantly lower fatigue life for a given plastic
as significantly slowed by this mechanism as surface or microcracks strain amplitude than the other steels. This has been attributed
would be. to a high level of strain partioning in the low volume fraction of
It is expected that there would be some interaction between the ferrite that led to pronounced surface relief and subsequent crack
retained austenite and surface cracks for TRIP780 due to the evenly initiation in that phase.
and closely spaced distribution of retained austenite, as shown in • Bulk retained austenite transformation during the early stages of
Fig. 13. While there were no obvious surface cracks at the half life low cycle fatigue as shown for TRIP780 appears to be detrimen-
(hence this interaction would not affect the cyclic softening ratio), tal to fatigue life at low strain amplitudes. The higher austenite
crack retardation via austenite transformation may be occurring for transformation in this steel compared to TRIP980 was due a
this steel late in the fatigue test. reduced stability resulting from the different surrounding phases
and larger austenite grain size.
4.3. Relationship between static and cyclic mechanical behaviour • Retained austenite measurements for TRIP980 and TRIP780 sug-
gest that cyclic hardening was significantly less dependent on the
While TRIP steels have been shown in numerous studies to austenite to martensite transformation than other factors such as
perform better than DP steels in fatigue for a given set of tensile the deformation mechanisms occurring in the ferrite.
properties, TRIP980 had a lower fatigue life for the examined strain • The TRIP steels all showed significantly lower cyclic softening
amplitudes than DP780 despite a greater uniform elongation and ratio than the dual-phase steels. The higher cyclic softening resis-
tensile strength. This lower fatigue life is believed to be related tance was not necessarily related to transformation of retained
to the high level of strain partioning into the relatively small vol- austenite to martensite. For TRIP780, retained austenite did
ume fraction of ferrite, in addition to the protection of the austenite transform during initial cycles that contributed to the low soft-
within the bainitic region from stress levels high enough to trans- ening ratio, however, no retained austenite transformation was
form the austenite to martensite (both during bulk cyclic straining measured for TRIP980.
as well as from strain-induced transformation resulting from sur-
face cracks in the ferrite). Acknowledgements
TRIP780 had a similar fatigue life than TRIP590 for a given
plastic strain amplitude despite higher strength levels and uni- The authors would like to acknowledge the work of Srikanth
form elongation. The XRD results showed that a significant volume Vegi in the preparation and testing stage, as well as Dr. Andrew
fraction of retained austenite (∼1.5%) transformed to marten- Sullivan for advice on electron microscopy techniques.
site during the early stages of cyclic deformation for TRIP780.
The comparatively poorer fatigue life relative to the other steels
suggests that this early austenite transformation is not helpful References
in prolonging fatigue life at low plastic strain amplitudes. The
[1] D.V. Edmonds, K. He, F.C. Rizzo, B.C. DeCooman, D.K. Matlock, J.G. Speer, Mater.
high fatigue ductility exponent and relatively improved strain life Sci. Eng. A 438–440 (2006) 25–34.
at higher strain amplitudes suggests that this retained austen- [2] E. DeMoor, P.J. Gibbs, J.G. Speer, D.K. Matlock, J.G. Schroth, Iron Steel Tech. 7–11
ite transformation was not harmful, and potentially beneficial, to (2010) 96–105.
[3] A.M. Sarosiek, W.S. Owen, Mater. Sci. Eng. 66 (1984) 13–34.
strain life for steels with a similar microstructure at higher strain [4] T.B. Hilditch, I.B. Timokhina, L.T. Robertson, E.V. Pereloma, P.D. Hodgson, Metall.
amplitudes. Mater. Trans. A 40 (2009) 342–353.
296 T. Hilditch et al. / Materials Science and Engineering A 534 (2012) 288–296

[5] Z.G. Wang, Z.M. Sun, S.H. Ai, Mater. Sci. Eng. A 113 (1989) 259–265. [17] H.J. Roven, E. Nes, Acta Metall. Mater. 39 (1991) 1719–1733.
[6] A.M. Sherman, R.G. Davies, Int. J. Fatigue (1981) 36–40 (January). [18] S.G. Hong, S.B. Lee, Int. J. Fatigue 26 (2004) 899–910.
[7] S.R. Mediratta, V. Ramaswamy, P. Rama Rao, Int. J. Fatigue (1985) 101–106 [19] Z.G. Hu, P. Zhu, J. Meng, Mater. Des. 31 (2010) 2884–2890.
(April). [20] T. Yokoi, K. Kawasaki, M. Takahashi, K. Koyama, M. Mizui, JSAE Rev. 17 (1996)
[8] Z.G. Wang, G. Wang, W. Ke, H. He, Mater. Sci. Eng. 91 (1987) 39–44. 191–212.
[9] G. Baudry, A. Pineau, Mater. Sci. Eng. 28 (1977) 229–242. [21] I. Alvarez-Armas, M.C. Marinelly, J.A. Malarria, S. Degallaix, A.F. Armas, Int. J.
[10] Z.Z. Hu, M.L. Ma, Y.Q. Liu, J.H. Liu, Int. J. Fatigue 19 (1997) 641–646. Fatigue 29 (2007) 758–764.
[11] G.R. Chanani, S.D. Antolovich, Metall. Trans. 5 (1974) 217–229. [22] K. Sugimoto, M. Kobayashi, S. Yasuki, Metall. Mater. Trans. A 28a (1997)
[12] U. Krupp, C. West, H.-J. Christ, Mater. Sci. Eng. A 481–482 (2008) 713–717. 2637–2644.
[13] X. Cheng, R. Petrov, L. Zhao, M. Janssen, Eng. Fract. Mech. 75 (2008) 739– [23] I. Timokhina, P. Hodgson, E.V. Pereloma, Metall. Mater. Trans. A 358 (2004)
749. 2331–2341.
[14] B.D. Cullity, Elements of X-ray Diffraction, third ed., Addison-Wesley, 1967. [24] J. Chiang, B. Lawrence, J.D. Boyd, A.K. Pilkey, Mater. Sci. Eng. A 528 (2011)
[15] B. Yan, D. Urban, Characterization of Fatigue and Crash Performance of New 4516–4521.
Generation High Strength Steels for Automotive Applications (Phase I and Phase [25] A. Basuki, E. Aernoudt, J. Mater. Process. Tech. 89-90 (1999) 37–53.
II), AISI/DOE Technology Roadmap Program Report, AISI, January 2003. [26] J.P Lucas, W.W. Gerberich, Int. J. Fatigue 7 (1985) 31–38.
[16] BS7270:1990, Method for Constant Amplitude Strain Controlled Fatigue Test- [27] K. Basu, M. Das, D. Bhattacharjee, P.C. Chakraborti, Mater. Sci. Tech. 23 (2007)
ing, British Standards Institution, 1990. 1278–1284.