RESEARCH ARTICLE

[Link]

Mechanisms of Strengthening and Fracture in 17%Cr

Martensite/Ferrite Dual-Phase Stainless Steels
Ming-Tao Wang,* Yong-Bo Liu, Xu Zhai, Nian-Wang Li, Jian-Feng Jin,
Yong-Chao Yang, and Yan-Zhong Tian

range of mechanical properties, has been

To clarify the strengthening mechanisms of medium-chromium stainless steels recently proposed.[1] This approach has
(SSs) with carbides, ferrite, and martensite, 17%Cr SSs with varying martensite been implemented in the production of var-
contents have been prepared, and the influence of martensite on microstructure, ious steel grades using a single chemistry
for automotive body-in-white manufactur-
mechanical properties, and fracture has been investigated. According to
ing, leading to simplification of production
THERMOCALC calculations, 17%Cr SS undergo a reversible phase transition processes. This approach holds immense
between austenite and ferrite þ M23C6 through the diffusion of carbon in potential in enhancing steel-making
austenite and M23C6 in the temperature range of 850–1220 °C. As the martensite control and the sorting and recycling of
content increases, M23C6 decreases, the martensite grain size increases, and the steel scraps.[2] The 17%Cr stainless steels
ferrite grain size initially increases and then decreases. Meanwhile, the yield (SSs), which primarily exhibit a ferrite
microstructure, is among the most widely
strength and ultimate tensile strength increase, while both the uniform and
used SSs.[3] With appropriate matching
post-uniform elongation decrease. It is also found a decrease in work-hardening of elements (such as Cr and C), carbides,
index and an increase in work-hardening rate with increasing martensite content. ferrite, and austenite at elevated tempera-
Under tensile loading, three types of voids are present in 17%Cr SS: type A only in tures and corresponding martensite after
the grain boundary (GB) area with M23C6 in martensite-free material and type B quenching could form through a diverse
and C in ferrite grains close to martensite grains and at ferrite/martensite GBs, range of heat treatments. Consequently,
the delicate combinations of these phases
respectively. An increase in martensite content leads to more voids, indicating a
provide numerous possibilities to tailor
reduction in material plasticity caused by martensite. the microstructure to achieve various
mechanical properties, which means the
17%Cr SSs can serve as a potential candi-
1. Introduction date for the UniSteel concept. The effective control of microstruc-
ture in dual-phase (DP) steels results in their superior combined
A novel concept called UniSteel, which entails utilizing a single- mechanical properties, which are not present in single-phase
steel chemistry to produce diverse microstructures and a broad steel materials.[4]
Ferritic-martensitic DP steels have undergone extensive
research, but the majority of studies mainly focused on carbon
M.-T. Wang, Y.-B. Liu, X. Zhai, N.-W. Li, J.-F. Jin, Y.-Z. Tian steels such as DP600 and DP800.[5] Recent work by Scoot et al.
Key Laboratory for Anisotropy and Texture of Materials, Ministry of explored the structure–property relationship of ultrafine-grained
Education, School of Materials Science and Engineering
DP steels through V-microalloying, with a size-sensitive mean
Northeastern University
Shenyang 110819, China field structure property model employed to capture the effects
E-mail: wangmingtao@[Link] of microalloying.[6] Sun et al. confirmed that lamellar-structured
M.-T. Wang ferrite/martensite (F/M) steel exhibits an additional strengthen-
State Key Laboratory of Rolling and Automation ing effect and enhanced plasticity.[7] Regarding the deformation
Northeastern University behavior of different phases in materials composed of both soft
Shenyang 110819, China
and hard phases, Paul et al. proposed a modified empirical model
Y.-C. Yang
School of Materials Science and Engineering
for the understanding of the behavior in cold rolled DP steels at
Taiyuan University of Science and Technology the microstructural level and determine the stress and strain
Taiyuan 030024, China partitioning between its constituent phases.[8] They also pre-
Y.-C. Yang dicted the ductile failure of DP590 in the form of plastic strain
State Key Laboratory of Advanced Stainless Steel Materials localization resulting from the incompatible deformation
Taiyuan Iron & Steel (Group) Co., Ltd. between the harder martensite phase and the softer ferrite
Taiyuan 030003, China
matrix during the deformation process based on representative
The ORCID identification number(s) for the author(s) of this article volume element finite element method. The failure modes
can be found under [Link] under plane stress state with free lateral boundary and strictly
DOI: 10.1002/srin.202400217 constrained lateral boundary were discussed, and they found

steel research int. 2024, 95, 2400217 2400217 (1 of 13) © 2024 Wiley-VCH GmbH
 1869344x, 2024, 10, Downloaded from [Link] by University Of British Columbia, Wiley Online Library on [07/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
l

[Link] [Link]

that the local failure mode is closely related to the stress state in understanding of strengthening mechanisms in 17%Cr SSs,
DP590 and DP780.[9,10] Numerous observations have been which holds substantial significance in terms of controlling
made regarding the failure mechanism of DP steels, and dif- material properties and expanding its applications in various
ferent models describing the failure mechanism of DP steels fields.
were proposed. Kadkhodapour et al. found that void nucleation
occurs due to ferrite GB decohesion in the vicinity of martens-
ite grains, and void initiation is observed between closely situ-
ated martensite grains in DP800.[11] In super 13Cr, δ-ferrite is a
2. Experimental Section
soft phase compared to the surrounding matrix at high temper- The alloy was cast into a 50 kg ingot in a vacuum induction fur-
atures. It has been observed that the δ-ferrite originates crack nace, with its composition presented in Table 1. The ingot was
initiations and growth, even leading to surface cracking of
then hot forged into a 30 mm-thick square slab, followed by
seamless tubes during hot piercing.[12,13] In the study of
homogenization at 1160 °C for one hour, then subsequently
DP590 and DP780, it is believed that void initiation, growth,
hot rolled (HR) to three mm thickness. The HR sheet was then
and coalition or decohesion between ferrite and martensite
heated at 850 °C for 3 h to eliminate the martensite formed
occur in regions with higher levels of strain.[9,10] In contrast,
during HR. The hot-rolled and annealed plate was rolled by
it has been confirmed that martensite is plastically deformable,
a four-roller reversible mill, with 8 passes with an average
and the failure of the martensite islands results from crack
reduction of 8% per pass at room temperature, ultimately
propagation toward the island center based on DIC results.[5]
the cold plate with a thickness of 1.0 mm was obtained. The
Furthermore, it has been found that the GB between ferrite
initial microstructure of the material is composed of ferrite,
matrix and martensite is strong and cohesive in DP600,[5] mak-
which transforms into austenite partly after high-temperature
ing it challenging to formulate a general failure mechanism
exposure. Subsequent quenching causes the austenite to trans-
for DP steels as a whole. For 17%Cr SSs, the existence of
ferrite-austenite DP region at high temperatures has been form into martensite. The sheet was annealed at 850 °C for
observed,[14,15] and some studies explored the technical feasi- three minutes, resulting in a microstructure consisting of
bility of applying the quenching and partitioning (Q&P) treat- M23C6 and ferrite. To obtain samples with different martensite
ment to the material.[16,17] In these studies, the corresponding contents, the sheets were reheated to 940 °C for 1, 3, 5, and
DP counterparts, which have industrial feasibility and showed 10 min, respectively, followed by water quenching to induce
promising mechanical properties, were only used for compar- the transformation of γ to α’. Additionally, a CR-annealed sam-
ison with a fixed martensite content. Therefore, a detailed ple without reheating treatment was chosen as a counterpart
study on the influence of strengthening and fracture mecha- for comparison. Samples containing martensite were treated
nisms by martensite contents in the microstructure is neces- as follows: Prepared a solution by mixing 30 g of potassium
sary in 17%Cr SSs. hydroxide and 30 g of potassium ferrocyanide with 80 mL of
To reveal the strengthening and fracture mechanisms within water. Heated the solution to ≈200 °C and put the samples
the multiphase microstructure composed of carbides, martens- to the solution and maintained the temperature for four to five
ite, and ferrite, for one thing, the influence of coordinated defor- minutes. Samples containing only ferrite were treated as fol-
mation hardening between the soft and hard phases on the lows: Prepared a solution by mixing 5 g of ferric chloride and
material’s strengthening mechanisms can be clarified. For 50 mL of hydrochloric acid with 100 mL of water. Then put the
another, the interactions between dispersed precipitates and samples to the solution for 20–30 s. The sample preparation
bulk phases during the deformation and fracture processes methods for scanning electron microscopy (SEM) and OM
can be elucidated. It can be anticipated that the 17%Cr SSs observations were the same.
with a microstructure consisting of both martensite and ferrite Dog-bone-shaped tensile samples with gauge dimensions of
exhibit significantly higher strength than that with only ferrite. 60  15  1.0 mm were used to measure the mechanical proper-
However, the review of the available literatures shows that the ties. Uniaxial tensile tests were conducted at room temperature
strengthening and fracture mechanism of similar materials is using an AG-XPLUS250KN universal tensile testing machine
not universally applicable. To reveal the impact of different under displacement control at a constant crosshead speed
martensite contents on the material’s strength and fracture of 3.6 mm min1 which corresponded to a strain rate of
mechanism, a systematic study is required. 17%Cr SS plates 0.001 s1, and the loading direction was parallel to the original
with different martensite contents were prepared by heat treat- rolling direction (RD). The 0.2% yield stress, σ0.2%, true ultimate
ments at elevated temperatures for partial austenitization and tensile strength, σu, and true strain at necking, εu, were extracted,
followed by quenching. The underlying strengthening mech- with the onset of necking taken at maximum load. Three samples
anisms of the examined material were uncovered through were tested under each condition for the accuracy and stability of
microstructural observations during tensile deformation and the results.
fracture processes. Special attention was given to investigate
the evolutionary characteristics of varied phases. By analyzing
the results obtained from tensile tests, the hardening behavior Table 1. Chemical composition (wt%) of the samples.
was precisely determined. Moreover, a model based on well-
established strengthening theories was employed to quantita- C Cr Si N S Mn P Fe
tively analyze the deformation characteristics of martensite 0.04 16.39 0.28 0.05 0.005 0.53 0.035 Bal.
and ferrite. Ultimately, this research provided a definitive

steel research int. 2024, 95, 2400217 2400217 (2 of 13) © 2024 Wiley-VCH GmbH
 1869344x, 2024, 10, Downloaded from [Link] by University Of British Columbia, Wiley Online Library on [07/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
l

[Link] [Link]

Microstructure observation was carried out using SEM 3. Results and Discussion
(JSM-6510 A) and optical microscopy (OM, GX71, OLMMPUS).
The deformation behaviors of martensite and ferrite were exam- 3.1. Equilibrium Phase Diagram
ined on the RD-TD cross section. The content (area) of different
phases, average grain size, and its distribution were estimated Figure 1a illustrates the pseudo-binary phase diagram of 17%Cr
by the microstructure recognition software (Particle Complex SSs with a Cr content of 15–20% and C content of 0.02–0.07% in
Microstructure Multiparticle Image Recognition Software V2.0) the temperature range of 500–1600 °C. In the 17%Cr SS, C and N
developed by our research group with machine learning algo- are important microalloying elements. These elements are aus-
rithms.[18] Based on five fields of view, the proportion of each tenite formers in SS, capable of expanding the austenite region
phase in the microstructure and the grain size under each con- on the phase diagram and ultimately affecting the martensite
dition were determined. content.[19] Additionally, C and N exist as interstitial atoms in

Figure 1. Data from THERMOCALC calculation: a) pseudo-binary phase diagram (15–20%Cr), b) pseudo-binary phase diagram (0.02–0.07%C),
c) phases content with various austenite contents, d) C content in different phases with various austenite, e) volume fractions of phases at different
temperatures.

steel research int. 2024, 95, 2400217 2400217 (3 of 13) © 2024 Wiley-VCH GmbH
 1869344x, 2024, 10, Downloaded from [Link] by University Of British Columbia, Wiley Online Library on [07/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
l

[Link] [Link]

the crystal lattice, providing a certain solid solution strengthen- at different temperatures reveal that the volume fraction of austen-
ing effect.[20] An excess of C can precipitate in the form of M23C6, ite varies with respect to the equilibrium temperature, reaching a
which plays a role in pinning GB movement. Si primarily maximum of ≈45.41% at 940 °C and then decreasing with subse-
increases the strength and hardness of the material and influen- quent increases or decreases at the equilibrium temperatures.
ces the ratio of austenite to ferrite at high temperatures in Figure 1c shows that as the austenite content increases, the con-
steels.[21] Mn combines with S in the steel to form MnS, which tents of M23C6 and ferrite decrease due to the phase transformation
is an effective metallurgical strategy to reduce the segregation of between them, which can also be observed from Figure 2.
S. As the temperature decreases, ferrite starts to crystallize Given that the C content influences the formation and content
out of the liquid and is retained in the microstructure above of austenite and is retained in martensite after quenching, which
1220 °C. Austenite then appears in the temperature range of subsequently affects the properties, it is crucial to determine the
850–1220°C due to the retention of austenite formers, such as distribution of C in different phases with varying austenite con-
C and N, in the matrix. When the temperature drops below tents. As depicted in Figure 1d, the C content in austenite
850 °C, M23C6-type carbides are formed because of the phase trans- decreases to some extent, while the C in ferrite remains almost
formation of γ(austenite) ! α(ferrite) þ M23C6, which is reversible stable. Hence, it can be inferred that the C in M23C6, rather than
during the heat treatment process. As can be seen from Figure 1a, ferrite, diffuses into austenite during the heat treatment process,
b, both Cr and C have a significant impact on the equilibrium resulting in the formation of austenite.
phases of the alloy. With the increase of Cr, the ferrite phase region It should be noted that, considering the significant variation
expands, while with the increase of C content, the austenite phase of austenite phase content with temperature in the range of
region expands, which is consistent with the existing research 850–940 °C, the specimens with different martensite contents
results.[19] Although the calculated results indicate the presence were prepared by changing the time while keeping the tempera-
of Cr2N in Figure 1a,b, almost no Cr2N is found in the microstruc- ture constant. Although different phase proportions were
ture,[16,17,22] and it is speculated that N is only dissolved in the obtained through this dynamically controlled method, the phase
matrix according to Mola et al.[17] In Figure 1e, the phase contents transition between austenite and ferrite essentially involved the

Figure 2. Microstructures of samples under varied holding time at 940 °C and martensite content: a) 0 min, MF, b) 1 min, DP13, c) 3 min, DP25, d) 5 min,
DP35, e) 10 min, DP41, f ) martensite content.

steel research int. 2024, 95, 2400217 2400217 (4 of 13) © 2024 Wiley-VCH GmbH
 1869344x, 2024, 10, Downloaded from [Link] by University Of British Columbia, Wiley Online Library on [07/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
l

[Link] [Link]

dissolution of carbides into the ferrite matrix to form austenite, balanced GB structure,[24] typically observed during the full
ensuring the mass balance. Therefore, the qualitative analysis of coarsening stage of materials. Under the same interfacial energy,
carbon diffusion characteristics in different phases based on interfaces with higher curvature become unstable and tend to
equilibrium calculation results was acceptable, and it can provide evolve toward a shape with lower curvature. However, the pres-
some guidance for the trend analysis. ence of many sharp angular shapes of martensite in the micro-
structure indicates that this type of martensite has a high
3.2. Microstructure interfacial curvature. This implies that there is an additional driv-
ing force, such as a chemical driving force, that stabilizes this
The original CR-annealed sheets, annealed at 850 °C, have a type of interfacial structure. That is to say, at high temperatures,
microstructure consisting of a ferrite matrix and M23C6 carbides there is a significant difference in chemical energy between aus-
distributed randomly, as shown in Figure 2a and b–e show the tenite and ferrite, which allows for the formation of high-energy
microstructures with various martensite contents on the cross- interfaces with large curvature between ferrite and austenite
section (ND-TD) observed by OM. Ms and Mf of the alloy approx- which forms martensite after cold quenching. The morphology
imately fall between 200 and 60 °C, indicating that direct water and composition of the precipitate are characterized based on
quenching results in a complete transformation of austenite to SEM and energy-dispersive spectroscopy as seen in Figure 3.
martensite.[10,14] The martensite content increases with holding It was found that they are compounds containing C, Cr, and
time, while the M23C6 content decreases. The size of martensite Fe. According to the calculation results of THERMOCALC,
refers to the diameter of blocky martensite calculated based on it can be determined that the precipitate is (Fe, Cr)23C6.
the equivalent circle. Each data point came from the statistical The absence of martensite islands inside the ferrite grains indi-
analysis of 10 visual fields. The contents of the martensite phase cates that the nucleation and growth of martensite occur at the
are found to be 0% (martensite-free, MF), 13% (DP13), 25% original ferrite GBs and GB junctions.
(DP25), 35% (DP35), and 41% (DP41), as shown in Figure 2f. Figure 4 illustrates the analysis results of average grain size
By comparing Figure 2a and b–e, it is clear that the GB cur- and distribution using the software.[18] It can be observed that
vature of ferrite in the 940 °C annealed sample is lower, and the the grain size and its distribution vary with the martensite
triple junction with 120° indicates that ferrite grains become sta- content. The peak of the martensite grain size distribution curve
ble easily at 940 °C.[23] GB curvature refers to the degree of cur- shifts toward the larger value, and the average grain size increases,
vature of GBs, which can be quantitatively described through the indicating the coarsening of martensite grains. However, the size
measurement of curvature. The higher the curvature, the greater of ferrite grains first increases and then decreases. This can
the mobility of the GB. By comparing Figure 2a with Figure 2b–e, be attributed to the coarsening effect of ferrite grains at high
it is evident that the GB curvature of ferrite grains in the sample temperatures and the formation of more austenite at the expense
annealed at 940 °C is lower. The triple junction, typically associ- of consuming ferrite grains.
ated with a 120°, indicates that ferrite grains are more likely to
stabilize at this temperature. In this process, carbides dissolve in 3.3. Mechanical Properties and Strengthening Mechanism
the ferrite matrix and austenite is formed, while the ferrite grains
undergo further coarsening. The driving force for the coarsening Cold-rolled and annealed stainless steel sheets typically exhibit a
of the ferrite grains is the overall reduction in the total GB area certain degree of texture,[3] resulting in variations in mechanical
per volume. Under the driving force, the GB gradually becomes properties in different directions during tensile testing, which
flat and its curvature decreases. As a result, the GB energy in the means the material is anisotropic. To eliminate this influencing
system decreases as its area per volume reduces. The dissolution factor, all tensile tests in this experiment were conducted with the
of carbides and the high temperature further promote this pro- stretching direction parallel to the RD. The stress–strain curves
cess, as the pinning effect of carbides on GB migration weakens for samples with different martensite contents are presented in
and high temperature also facilitates its migration. The final state Figure 5a,b, with the square marks indicating the beginning of
of this process is an GB curvature of zero and ferrite GBs form a necking according to Considère criterion. Typically, carbide pre-
120° triple junction on a 2D plane, which is a relatively stable and cipitation results in precipitation hardening. In 17%Cr SS, there

Figure 3. Microstructural characterization of the precipitation: a) morphology of the precipitation and b) EDS.

steel research int. 2024, 95, 2400217 2400217 (5 of 13) © 2024 Wiley-VCH GmbH
 1869344x, 2024, 10, Downloaded from [Link] by University Of British Columbia, Wiley Online Library on [07/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
l

[Link] [Link]

Figure 4. Size distribution and average grain size in samples: a) MF, b) DP13, c) DP25, d) DP35, e) DP41, f ) average grain size.

is a certain content of M23C6 with a size of ≈1 μm as shown elongation, uniform elongation, and post-uniform elongation fol-
in Figure 3a. The occurrence of phase transformation low an opposite trend to the strength. When the martensite con-
(α(ferrite) þ M23C6 ! γ(austenite)) implies that with the increase tent increases from 0% to 41%, the yield strength and tensile
of martensite content, the content of carbides decreases sharply. strength of the material correspondingly increase from 303
According to the estimation by THERMOCALC, the carbide frac- and 472 MPa to 409 and 718 MPa. The increments are 106 and
tions of MF, DP13, DP25, DP35, and DP41 are 7%, 4%, 1.6%, 246 MPa. On average, each 1% increase in martensite content
0%, and 0%, respectively. Based on the Ashby Orowan relation- leads to a 2.59 and 6.00 MPa improvement in yield strength
0.538Gbf 0.5 and tensile strength, demonstrating the significant strengthen-
ship σ p ¼ X
p X
ln 2b , where G is the shear modulus (80 GPa
for α-Fe), b is the Burgers vector (0.25 nm), fp is the volume frac- ing effect of martensite in the alloy. Both uniform elongation
tion of precipitates, and X is the average size of precipitates,[21,25] and post-uniform elongation decrease from 21.85% and 6.45%
precipitation strengthening caused by carbides can be deter- to 11.60% and 1.95%, respectively. It is worth noting that the
mined as 21.6, 16.4, 10.3, 0, and 0 MPa, and its highest contri- yield point disappears in samples containing martensite during
bution only accounts for 7.1% of the yield strength. Therefore, tensile testing, which is consistent with previous findings.[16,17]
martensite is identified as the main factor affecting the strength The evolution of strain-hardening exponent n and work-
of the material. Increasing the martensite content results in hardening rate of samples with various martensite contents
higher yield strength and ultimate tensile strength, while total are shown in Figure 6. It can be observed that the instantaneous

steel research int. 2024, 95, 2400217 2400217 (6 of 13) © 2024 Wiley-VCH GmbH
 1869344x, 2024, 10, Downloaded from [Link] by University Of British Columbia, Wiley Online Library on [07/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
l

[Link] [Link]

Figure 5. Mechanical properties of samples with different martensite contents: a) engineering stress–strain curves, b) true stress–strain curve, c) yield
strength and ultimate tensile strength, d) total elongation, uniform elongation, and post-uniform elongation.

Figure 6. Work-hardening curves: a) strain hardening exponent n, b) work-hardening rate.

strain-hardening exponent n decreases with an increasing Figure 6b displays an increase of the work-hardening rate with
martensite content beyond 2% in Figure 6a. The uniform elon- the increase of the martensite volume fraction. This can be qual-
gation Figure 5c follows the same trend, which indicates that the itatively explained as follows: under tensile loading, 1) both
Considère criterion (nave = εuniform elongation) is applicable to 17% ferrite and martensite grains experience elastic deformation ini-
Cr SSs. tially; 2) subsequently, the softer ferrite grains undergo plastic

steel research int. 2024, 95, 2400217 2400217 (7 of 13) © 2024 Wiley-VCH GmbH
 1869344x, 2024, 10, Downloaded from [Link] by University Of British Columbia, Wiley Online Library on [07/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
l

[Link] [Link]

deformation, while martensite grains remain in the elastic γ g ¼ ½1  expðωεÞ=ω, ω is the coefficient characterizing
regime;[24,26] and 3) finally, as the surrounding hard martensite deformability of phases. λg is the geometric deformation region
grains constrain the ferrite ones, dislocations accumulate within scale, for DP steels, it can be expressed as λg ¼ kg dm =V M , kg is
the ferrite, leading to a substantial increase in their strength and the deformation zone shape factor and taken as 0.125.[28] Then
making them almost as strong as the martensite grains.[24,27] At the stress–strain relationship can be expressed as follow,[32]
this stage, the martensite grains begin to undergo plastic defor-
mation. The asynchronous deformation mechanism of ferrite σ DP ¼ σ 0 þ αMμb1=2 ½FerðεÞ=dF þ MarðεÞ⋅V M =dM 1=2 (4)
and martensite may lead to the difference in work-hardening
rate. The disappearance of the yield point, as observed in where FerðεÞ ¼ 1=pd  exp½Mpd ðε þ ε0 Þ=pd , MarðεÞ ¼
Figure 5a,b, cannot be attributed to the partitioning of C in ferrite kg =ω  kg exp ½Mωðε þ ε0 Þ=ω, pd = 0.95 and ω = 18, ε0 is the
and martensite[17] as C in ferrite remains stable and C in austen- strain when DP steel yields, ε þ ε0 is the total strain.
ite mainly comes from M23C6 dissolution according to the After the onset of plastic deformation in MF materials, the
THERMOCALC data in Figure 1d. Specifically, for the DP45 strength of the ferrite can be calculated using Equation (5).
samples under an equilibrium condition, no significant yield pla- Notably, Equation (5) indicates that the strength of the material
teau was observed. Therefore, it is speculated that the disappear- depends on its grain size, which is consistent with the well-
ance of the yield plateau is not caused by a decrease in C content established Hall–Petch relation.
of ferrite, but rather due to the fact that during the initial stages
1=2
of deformation, the ferrite grains undergo a greater amount of σ DP ¼ σ 0 þ KdF σ F ¼ σ F0 þ αMμb1=2 ½FerðεÞ=dF 1=2 (5)
plastic deformation, leading to a reduction of mobile dislocations
in ferrite. As to the martensite, it is essential to take into account both elas-
To quantitatively study the deformation process in 17%Cr SSs, tic and plastic deformation, as the stress–strain relationship can
a physical model was applied, which considered the effect of vol- be described by Equation (6).[33] Therefore, a comprehensive
ume fraction of martensite on the change in strain and stress. analysis of martensite behavior requires consideration of both
Additionally, the micro-stress–strain partitioning between mar- types of deformation.
tensite (hard phase) and ferrite (soft phase) was analyzed quan- σ M ¼ EεM (6a)
titatively during the very early stage of deformation and even
before necking. σ M ¼ K M ðεM ÞnM (6b)
The work hardening of the material primarily depends on the
statistically stored dislocations (ρs), which are related to the plas- where E is the elastic modulus, nM is the work-hardening index
tic strain during plastic deformation, and geometrically neces- of Hollomon’s equation, which is normally taken as 0.05–0.07
sary dislocations (ρG), which are related to the deformation for the low-carbon martensite, K M ¼ σ Mb =εnMu
M
, εMu , and σ Mb
compatibility of phases in materials.[28] The Taylor relationship are maximum uniform strain and ultimate tensile strength
indicates that the average flow stress (σDP) is dependent on the of martensite related with the C content, respectively.
 ela
total dislocation density (ρT):[28] ε ,ε < ε pla
εM ¼ M M pla M0 εela and εM represent the elastic
1=2
εM0 þ εM , εM ≥ εM0 M
σ DP ¼ σ 0 þ αMμbρT (1) strain and plastic strain of the martensite, respectively.
εM0 is the maximum elastic strain of martensite. εMu ¼ nM
where α, M, μ, and b are constant, Taylor factor, shear modulus,
according to Considère relationship. σ Mb can be calculated
and Burgers vector, respectively, and their values were taken as 1=2
0.04, 3, 80 GPa, and 0.25 nm, respectively.[25,27,28] The total dis- according to σ Mb ¼ 38.61 þ 3297.75CM ,[32] C M was taken as
location density is assumed to be the sum of the statistically the C content in the austenite as shown in Figure 1d. In DP steel,
stored dislocations and the geometrically necessary dislocations, the stress and strain during the tensile process are jointly influ-
ρT ¼ ρs þ ρG .[29] During tensile deformation, the relationship enced by both phases. Equation (5) describes the relationship
between statistical dislocation density and strain can be between total stress and strain. To clarify the respective contri-
expressed as,[30,31] butions of martensite and ferrite, it is assumed that the stress
(σ F , σ M ) and strainðεF , εM ) relationships of each phase are line-
dρs ¼ Mð1=bΛ − pd ρs Þdε (2) arly additive, adhering to the hybrid law, as expressed in the fol-
lowing equation,
where pd is the probability of disappearance or remotion of the
immobile dislocations, which is related to the recovery of σ DP ¼ ð1  V M Þσ F þ V M σ M (7)
the dynamic dislocations during tensile deformation, and Λ is
the mean free path of the immobile dislocations, which is approx- εDP ¼ ð1  V M ÞεF þ V M εM (8)
imated as the grain size for the relatively soft ferrite matrix.[30]
By combining Equation (5) and (6)–(8), it is possible to calculate
The geometrically necessary dislocations can be expressed as
the corresponding (εF, σF) and (εM, σM) under any given
follow,[25,28]
(εDP, σDP), thereby deriving the stress and strain ratios of the soft
ρG ¼ γ g =ðbλg Þ (3) and hard phases in the DP steel under that state. This method-
ology has been successfully applied to predict the performance of
where γg is the average strain gradient between the soft and hard ferrite-martensite DP materials,[32] demonstrating the effective-
phases, which can be expressed in the exponential form, ness of this model. To elucidate the stress and strain states of

steel research int. 2024, 95, 2400217 2400217 (8 of 13) © 2024 Wiley-VCH GmbH
 1869344x, 2024, 10, Downloaded from [Link] by University Of British Columbia, Wiley Online Library on [07/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
l

[Link] [Link]

each phase in DP steels during a tensile test, stress–strain curves and martensite under applied macro-strains in 17%Cr DPSSs
were calculated and compared with the test results. Furthermore, with different martensite contents. In Figure 7a–d, the macro-
the model was utilized to deduce the partition of microscopic scopic strains corresponding to the onsets of plastic deformation
stress and strain as a function of macroscopic strain, as illus- in martensitic grains are identified, which are consistent with the
trated in Figure 7a–d. As the strain increases, the slopes of inflection point on the curves in Figure 7e, indicating that the
the stress–strain distribution lines vary, indicating that the stress martensite content has a crucial influence on the deformation
and strain states of the two phases are not synchronized. In behavior in this type of steel.
DP13, martensite remains in the elastic stage throughout the ten- In Figure 7e, it is evident that the strain ratio of ferrite to mar-
sile process. In DP25 and DP35, the plastic deformation of mar- tensite undergoes two stages: elastic and plastic deformation of
tensite occurs in the final stage. While a considerable plastic martensite. The elastic deformation dominates the whole pro-
deformation can be observed in the mid-stage of the strain in cess. During the elastic deformation stage of martensite, the rel-
DP41. Figure 7e illustrates changes in the strain ratio of ferrite atively soft ferrite grains borne a relatively large proportion of the

Figure 7. Calculated stress–strain partitioning and tested stress–stain curves of a) DP13, b) DP25, c) DP35, d) DP41, e) strain ratios of the ferrite to
martensite.

steel research int. 2024, 95, 2400217 2400217 (9 of 13) © 2024 Wiley-VCH GmbH
 1869344x, 2024, 10, Downloaded from [Link] by University Of British Columbia, Wiley Online Library on [07/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
l

[Link] [Link]

applied strain. Once the martensite undergoes plastic deforma- crystal plasticity finite element method simulation. It was
tion, the ratio of strain borne by ferrite gradually decreases. observed that the local stress is heterogeneous at the microscopic
Under the same strain conditions, as the martensite content level. Specifically, the stress within the reversed austenite is
increases, the ferrite content decreases accordingly. This results found to be lower than that in the surrounding martensitic
in more strain being allocated to the ferrite and the ferrite work- matrix.[13,34] Despite the differences in specific phases within
hardening rate decreases. The martensite mainly undergoes the alloy, the characteristic stress distribution between soft
elastic deformation, and the hardening rate is the elasticity mod- and hard phases is consistent with our findings.
ulus, which is much higher than the work-hardening rate of the
ferrite. The final net effect, caused by the inconsistency in defor- 3.4. Fracture Characteristics and Behavior
mation, results in an increase in the work-hardening rate with an
increasing martensite content in Figure 6b. It should be noted Based on the perspective presented in ref. [15], it is difficult to
that the data in Figure 7 are based on the results of average grain identify a viable formulation that is applicable to a broad range of
size calculations. Actually, the plastic deformation of the failure mechanisms for DP steels. Therefore, it is imperative
martensite grains also start gradually because of a continuous to investigate the fracture process in 17%Cr SSs. Two sets of
distribution of grain size in the real alloys. Thus, a continuous specimens were subjected to tensile testing. To analyze the void
variation of the work-hardening curve occurs in Figure 6a,b. The characteristics, the first group of specimens was interrupted at
calculated results agree well with the trend of the tested curves, the strain of 0.09, which corresponds to the uniform elongation
indicating the validity of this model. According to the partition- of the specimen with the highest martensite content. Another
ing of stress and strain in ferrite and martensite, it is evident that group was stretched until failure occurred during the tensile test.
the martensite bears a higher load while more strain acts on Figure 8 gives details of the voids pattern in specimens of the first
the ferrite, which follows the load transfer strengthening mecha- group, and all the images are from the central area in the RD-TD
nism. The local von Mises equivalent stress distribution after cross section. Figure 8a illustrates that only a small void is visible
minor straining deformation in 13Cr SS was analyzed using in the MF specimen, located at the GBs region containing M23C6

Figure 8. Void patterns in the samples with different martensite contents: a) MF, b) DP13, c) DP25, d) DP35, e) DP41, f ) description of sampling
locations. Voids of type-A, B, and C are indicated by arrows with yellow, blue, and cyan, respectively.

steel research int. 2024, 95, 2400217 2400217 (10 of 13) © 2024 Wiley-VCH GmbH
 1869344x, 2024, 10, Downloaded from [Link] by University Of British Columbia, Wiley Online Library on [07/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
l

[Link] [Link]

(labeled as type-A). Conversely, in the DP specimens, voids are deformation. This strain mismatch between ferrite and martens-
observed around the martensite, as shown in Figure 8b–e. Two ite generates local stress perpendicular to the loading direction.
types of voids, which do not distribute randomly, can be identi- The sharp ends of martensite grains are particularly vulnerable to
fied in this case. One type (type-B) is typically found in close prox- stress/shear strain localization, which triggers delamination
imity to the martensite grains, extending in the tensile direction, between neighboring ferrite grains. This mechanism depends
which is consistent with findings in ref. [15]. The other type on local stress concentrations arising from strain incompatibili-
(type-C) is the delamination of ferrite/martensite interfaces and ties between martensite and ferrite, as confirmed by finite ele-
appears in a semi-spherical shape, indicating a weak bonding ment simulation.[15] Unlike the case in carbon steels such as
of ferrite/martensite interfaces. Moreover, it is evident that speci- tempering-treated DP800, type-C voids in 17%Cr DP SSs do
mens containing more martensite exhibit more voids in not result from enhanced ferrite/martensite boundaries due to
Figure 8b–e, which corresponds with the plasticity of the martensite decomposition and C diffusion. In contrast, most
specimens. C forms Cr23C6 due to the high Cr content in 17%Cr DP SSs,
Void nucleation at second-phase particles in the GB region can even after martensite decomposition as shown in Figure 1d.
occur through particle fracture or decohesion. The weak bonding This suggests that the ferrite/martensite interface is difficult
between ferrite and M23C6, along with stress concentrations at to strengthen by tempering treatment, in contrast, it may even
GBs during deformation, causes the formation of type-A voids be weakened by Cr23C6 precipitation. As a result, ferrite/
in the MF specimen.[35,36] Regarding the type-B voids, they martensite interfaces in 17%Cr DP SSs are susceptible to
are caused by the delamination of ferrite/ferrite interfaces. breaking under stress concentration or deformation mismatch.
Clearly, it can be assumed that carbides also play a role in this Comparing all types of voids in Figure 8, type-A voids are rela-
process.[15] In Figure 7e, ferrite is seen to undergo extensive plas- tively few and small, while type-C voids are larger and more com-
tic deformation, while martensite experiences minimal mon than type-B voids. It is likely that the ferrite/M23C6 interface

Figure 9. Fracture morphology in the central regions of samples with different martensite contents: a) MF, b) DP13, c) DP25, d) DP35, e) DP41,
f ) longitudinal section of DP25.

steel research int. 2024, 95, 2400217 2400217 (11 of 13) © 2024 Wiley-VCH GmbH
 1869344x, 2024, 10, Downloaded from [Link] by University Of British Columbia, Wiley Online Library on [07/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
l

[Link] [Link]

is stronger than the ferrite/martensite interface in 17%Cr DP during deformation, which triggers void formation due to stress
SSs, which is not the same as DP800.[15] concentration.
The tensile fracture morphology in the central region of the
interrupted samples is illustrated in Figure 9, revealing the for-
mation of dimples due to shear localization of void growth and 4. Conclusions
impingement.[37] The size and depth of the dimples vary in dif-
ferent samples, and it is known that failure occurs when void In this study, 17%Cr SSs with different martensite contents have
impingement reaches a certain stage.[10] The fracture surfaces been prepared and the influence of martensite on the microstruc-
of tensile samples consist of traces of intensive deformation ture, properties, and fracture are studied. THERMOCALC calcu-
before failures, such as micro voids, dimples, and sharp crests. lation results suggest a reversible phase transition between
The size of dimples increases with increasing ferrite content due austenite and ferrite þ M23C6 in 17%Cr SSs in the temperature
to a more intensive plastic deformation and the coalescence of range of 850–1220 °C. Furthermore, the phase transition occurs
micro-voids during a longer deformation process.[38] The MF by the diffusion of C between M23C6 and austenite, while C in the
sample displays a typical ductile dimple fracture feature, with ferrite remains constant during the process. After tempering
some big and deep dimples in Figure 9a, indicating a more inten- treatment at 940 °C followed by quenching. After heat treatment
sive plastic deformation before facture. As shown in Figure 9b–e, at 940 °C for partial austenitization and followed by quenching,
samples with more martensite exhibit fewer and shallower dim- microstructures with varying martensite contents can be
ples, implying a local shearing mechanism,[39] less homogeneous obtained. As the annealing time increases, the martensite con-
void sizes, and decreased plasticity. Therefore, it is speculated tent increases while M23C6 decreases. Martensite grain size
that samples with more martensite are more susceptible to void increases while ferrite grain size first increases and then
nucleation and display smaller void size. The fracture surface of decreases. Additionally, the increased GB mobility and the reduc-
sample with only ferrite shows a complex morphology predomi- tion of pinning particles (M23C6) cause the formation of straight
nantly characterized by dimples, indicating a ductile fracture. ferrite GBs.
Some stretching areas in samples with martensite show a small With an increase in martensite content (from 0% to 41%), the
amount of cleavage. Cleavage morphology usually occurs and can yield strength and ultimate tensile strength increase from 303
be detected in over-dense atomic planes.[40,41] It is reported that and 472 MPa to 409 and 718 MPa, respectively, while both uni-
the presence of BCC structure in the steel increases the forma- form elongation and post-uniform elongation decrease from
tion of cleavage in the fracture surface.[40] The formation of cleav- 21.85% and 6.45% to 11.60% and 1.95%, respectively. There
age planes at the two-phase interfaces is more likely to occur, is a decreased work-hardening index with increasing martensite
leading to a decrease in ductility. These elongated cracks are content, which is also consistent with the trend of uniform elon-
indicative of the intergranular fracture. Due to the weak bonding gation. While the work-hardening rate increases with higher
of ferrite and martensite, this interface is the main potential martensite content.
point for crack initiation.[40] With further progress of the tensile The void types during tensile deformation are significantly
test, the generated voids start to grow and unify. More martensite influenced by the composition of the phases in 17%Cr SSs.
increases their interfaces with ferrite and increases the possibil- In alloys without martensite, only Type-A voids, which are
ity of crack initiation. located around GBs containing M23C6, exist. Conversely, within
Referring to the characteristic microstructures of different the DP microstructure, Type-B and Type-C voids can be found in
martensite contents (Figure 1), it is found that DP25 contains ferrite grains near martensite and at ferrite/martensite GBs,
a significant amount of carbides and a certain proportion of mar- respectively. Additionally, a higher number of voids can be
tensite, making it suitable for comparing the roles of different induced by more martensite grains under the same deformation.
phases during the fracture process. Therefore, further observa-
tion and analysis of the microstructure during the fracture pro-
cess were conducted in DP25. To further investigate its fracture
behavior, the DP25 specimen was subsequently sectioned along Acknowledgements
its longitudinal axis at the approximate center of its thickness, The authors would like to acknowledge the financial support of the 111
where the initial cut had been made. The resulting fracture sur- Project, China (grant no. B20029), the Fundamental Research Funds
face, shown in Figure 9f, exhibits both martensite and small hol- for the Central Universities, China (grant nos. N2002017 and N2007011),
lows, which is caused by the detachment of M23C6 during and 2021 Annual Scientific Research Funding Project of Liaoning
Provincial Department of Education (LJKZ1335).
electrolytic polishing. The fracture surface shows cup-like sec-
tions, which are believed to be dimples resulting from the typical
ductile fracture process that occurs when voids grow and
impinge under the applied load. Martensite grains are observed Conflict of Interest
near the dimples, with one type showing sharp corners beneath The authors declare no conflict of interest.
the dimple bottom (type-B voids in Figure 8) and the other type
displaying martensite GB at the bottom of the dimple (type-C
voids). Large longitudinal voids of type-B and C are also observed
beneath the fracture surface in the area of severe plastic Data Availability Statement
deformation. Therefore, it can be concluded that the presence The data that support the findings of this study are available from the
of martensite leads to an inhomogeneous stress–strain partition corresponding author upon reasonable request.

steel research int. 2024, 95, 2400217 2400217 (12 of 13) © 2024 Wiley-VCH GmbH
 1869344x, 2024, 10, Downloaded from [Link] by University Of British Columbia, Wiley Online Library on [07/01/2025]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
l

[Link] [Link]

Keywords [19] Y. Han, Z. H. Liu, C. B. Wu, Y. Zhao, G. Q. Zu, W. W. Zhu, X. Ran,

Tungsten 2022, 5, 419.
17%Cr dual-phase stainless steel, fracture feature, mechanical properties, [20] J. Sun, H. Tang, C. L. Wang, Z. Han, S. S. Li, Steel Res. Int. 2022, 93,
microstructural morphology, strengthening mechanism 2100450.
Received: March 16, 2024 [21] T. Yang, L. J. Chai, H. Wang, G. Li, K. L. Murty, Mater. Sci. Eng., A
Revised: June 19, 2024 2023, 886, 145712.
Published online: July 29, 2024 [22] H. H. Lu, W. Q. Li, L. Y. Du, H. K. Guo, W. G. Zhang, Z. G. Liu, Mater.
Sci. Eng., A 2019, 754, 502.
[23] D. Raabe, Phys. Metall. 2014, 3, 2291.
[1] Q. Lu, Q. Q. Lai, Z. S. Chai, X. L. Wei, X. C. Xiong, H. L. Yi, [24] M. Calcagnotto, Y. Adachi, D. Ponge, D. Raabe, Acta Mater. 2011,
M. X. Huang, W. Xu, Sci. Adv. 2021, 7, 1. 59, 658.
[2] X. Y. Li, K. Lu, Nat. Mater. 2017, 16, 700. [25] N. A. Fleck, G. M. Muller, M. F. Ashby, Acta Metall. Mater. 1994,
[3] M. Y. Huh, J. H. Lee, S. H. Park, O. Engler, D. Raabe, Steel Res. Int. 42, 475.
2005, 76, 11. [26] H. G. Armaki, R. Maab, S. P. Bhat, S. Sriram, J. R. Greer, K. S. Kumar,
[4] X. Cai, X. Q. Hu, L. G. Zheng, D. Z. Li, Acta Metall. Sin. (Engl. Lett.) Acta Mater. 2014, 62, 197.
2020, 33, 693. [27] M. Delince, Y. Brechet, J. D. Embury, M. G. D. Geers, P. J. Jacques,
[5] H. Ghadbeigi, C. Pinna, S. Celotto, Mater. Sci. Eng., A 2013, 588, 420. T. Pardoen, Acta Mater. 2007, 55, 2337.
[6] C. P. Scott, F. Fazeli, B. S. Amirkhiz, I. Pushkareva, S. Y. P. Allain, [28] R. W. Cahn, Int. Mater. Rev. 1972, 17, 147.
Mater. Sci. Eng., A 2017, 703, 293. [29] M. F. Ashby, Philos. Mag. 1970, 21, 164.
[7] J. J. Sun, T. Jiang, Y. J. Wang, S. W. Guo, Y. H. Liu, Mater. Sci. Eng., A [30] Y. Estrin, H. Mecking, Acta Metall. Mater. 1984, 32, 57.
2018, 734, 311. [31] F. J. Zerilli, Metall. Mater. Trans. A 2004, 35, 2547.
[8] S. K. Paul, M. Mukherjee, Comput. Mater. Sci. 2014, 84, 1. [32] X. Wei, L. M. Fu, S. C. Liu, W. Wang, A. D. Shan, Chin. J. Mater. Res.
[9] S. K. Paul, Mater. Des. 2013, 44, 397. 2013, 6, 665.
[10] S. K. Paul, A. Kumar, Comput. Mater. Sci. 2012, 63, 66. [33] H. W. Swift, J. Mech. Phys. Solids 1952, 1, 1.
[11] J. Kadkhodapour, A. Butz, S. Z. Rad, Acta Mater. 2011, 59, 2575. [34] M. G. Lee, S. J. Kim, H. N. Han, Int. J. Plast. 2010, 26, 925.
[12] H. A. Derazkola, E. G. Gil, A. M. Marrodan, Res. Eng. 2024, [35] G. L. Roy, J. D. Embury, G. Edwards, M. F. Ashby, Acta Metall. Mater.
21, 101682. 1981, 29, 1509.
[13] H. A. Derazkola, E. Garcia, A. M. Marrodan, A. C. Fernandez, J. Mater. [36] D. Kwon, Scr. Metall. Mater. 1988, 22, 1161.
Res. Technol. 2022, 18, 2993. [37] G. A. Cingara, C. A. R. Saleh, M. K. Jain, D. S. Wilkinson, Metall.
[14] M. J. Li, Z. L. Wang, X. Chen, J. Mater. Eng. Perform. 2021, 30, 379. Mater. Trans. A 2009, 40, 3117.
[15] F. Badkoobeh, H. Mostaan, M. Rafiei, A. Bakhtiari, J. Mater. Eng. [38] H. A. Derazkola, E. G. Gil, A. M. Marrodán, D. Méresse, Metals 2021,
Perform. 2023, 32, 5028. 11, 527.
[16] H. H. Lu, H. K. Guo, W. G. Zhang, W. Liang, Mater. Lett. 2019, 240, 275. [39] I. Barsoum, J. Faleskog, Int. J. Solids Struct. 2007, 44, 1768.
[17] J. Mola, B. C. De Cooman, Scr. Mater. 2011, 65, 834. [40] H. A. Derazkola, E. Garcia, A. M. Marrodan, J. Hardell, J. Mater. Res.
[18] J. F. Jin, Particle Complex Microstructure Multiparticle Image Technol. 2023, 24, 3464.
Recognition SoftwareV2.0, Northeastern University, China 2018. [41] A. Pineau, A. A. Benzerga, T. Pardoe, Acta Mater. 2016, 107, 424.

steel research int. 2024, 95, 2400217 2400217 (13 of 13) © 2024 Wiley-VCH GmbH