Inconsistent effects of austempering time within transformation stasis on monotonic and cyclic deformation behaviors of an ultrahigh silicon carbide-free nanobainite steel

Inconsistent effects of austempering time within transformation stasis on monotonic and cyclic deformation behaviors of an ultrahigh silicon carbide-free nanobainite steel

Materials Science & Engineering A 751 (2019) 80–89 Contents lists available at ScienceDirect Materials Science & Engineering A journal homepage: www...

4MB Sizes 0 Downloads 6 Views

Materials Science & Engineering A 751 (2019) 80–89

Contents lists available at ScienceDirect

Materials Science & Engineering A journal homepage: www.elsevier.com/locate/msea

Inconsistent effects of austempering time within transformation stasis on monotonic and cyclic deformation behaviors of an ultrahigh silicon carbidefree nanobainite steel

T



Jiali Zhaoa, Fucheng Zhanga, , Bo Lvb, Zhinan Yangc, Chen Chena, Xiaoyan Longa, Xiaojie Zhaoa, Chunhe Chua a b c

State Key Laboratory of Metastable Materials Science and Technology, Yanshan University, Qinhuangdao 066004, China College of Environmental and Chemical Engineering, Yanshan University, Qinhuangdao 066004, China National Engineering Research Center for Equipment and Technology of Cold Strip Rolling, Yanshan University, Qinhuangdao 066004, China

A R T I C LE I N FO

A B S T R A C T

Keywords: Bainite Austempering time Microstructure Mechanical property

Given the transformation stasis after the incomplete transformation of carbide-free bainite, this paper aimed to study the effects of austempering time within transformation stasis on the bainite microstructure, monotonic deformation behavior, and especially cyclic deformation behavior using an ultrahigh silicon (2.59 wt%) steel. With increasing austempering time, the dislocation density of bainitic ferrite decreased, carbon content of retained austenite slight increased, and carbon distribution in retained austenite blocks gradually homogenized. The best combination of strength and ductility was obtained through a longer austempering time within transformation stasis. But the longer austempering time, indeed, did not result in longer fatigue life. The opposite trend could be explained by the fact that the primary factors affecting these mechanical properties were different. The lower density of mobile dislocation pre-existed in the starting microstructure of the samples austempered for a longer time was primarily responsible for its lower cyclic hardenability. Moreover, one retained austenite block with completely homogeneous carbon distribution was only transformed to one martensite grain, which increased the cyclic softening and degrade the fatigue life. By contrast, the deformation-induced martensite transformation from the retained austenite with higher mechanical stability in the samples austempered for a longer time enhanced strain hardenability at higher monotonic tensile strains and well delayed the necking, thus improving the combination of strength and ductility.

1. Introduction

same chemical composition have identical free energy. In fact, the growth of bainitic ferrite is always displacive in which a shear mechanism is involved. There is an invariant-plane strain deformation with an elastic energy barrier (also called stored energy, ~400 J/mol) between bainitic ferrite and austenite [5–7]. Bhadeshia et al. [8] argue that the growth of bainite should require an extra driving force due to this elastic energy barrier. According to this theory, the T0′ curve takes into account this elastic energy barrier [9]. When the carbon content of untransformed austenite approaches the one given by the T0′ curve, the free energy change between bainitic ferrite and austenite is no longer greater than the elastic energy barrier [8]. Thus, bainite transformation cannot be performed thermodynamically, suggesting the transformation stasis stage is reached [10]. The transformation stops prematurely when the retained austenite reaches approximately a carbon level defined by the T0′ curve, rather that much greater equilibrium carbon level

As well known, bainite is formed by a “displacive” or “military” transfer of the iron and substitutional solute atoms from austenite to ferrite. There is no long-range redistribution of substitutional solutes while the carbon atoms, which are in interstitial solution, can migrate at rates many orders of magnitude greater than the iron or substitutional solute atoms [1–3]. For carbide-free bainite, the carbide precipitations are inhibited and therefore there is no other sinks for the carbon getting partitioned than between austenite and bainitic ferrite [4]. Accordingly, after displacive growth, the excess carbon atoms in bainitic ferrite are quickly rejected to the surrounding austenite. Growth of carbide-free bainite is initially considered to follow the T0 criteria. The T0 curve defines the locus of all points on a plot of temperature versus carbon concentration where austenite and ferrite of the



Corresponding author. E-mail address: [email protected] (F. Zhang).

https://doi.org/10.1016/j.msea.2019.01.100 Received 17 October 2018; Received in revised form 24 December 2018; Accepted 21 January 2019 Available online 30 January 2019 0921-5093/ © 2019 Elsevier B.V. All rights reserved.

Materials Science & Engineering A 751 (2019) 80–89

J. Zhao, et al.

thinned to perforation on a TenuPol-5 twin-jet unit with an electrolyte consisting of 7% perchloric acid in ethanol at room temperature and at an operating voltage of 30 V. X-ray diffraction (XRD, Rigaku D/max-2500/PC, 40 kV, 200 mA) tests were carried out by using Cu-Kα radiation [38]. XRD data were collected from a 2θ range of 40–100° in step scanning, with a step width of 0.02° and counting time of 2 s. For each test state, three XRD samples were examined to obtain average value. The volume fraction of retained austenite (Vγ) was determined on the basis of a direct comparison method [39] of the integrated intensity of the austenite (200)γ, (220)γ, and (311)γ peaks and the bainite (200)α and (211)α peaks. This helped avoid bias due to crystallographic texture [40]. The carbon content of retained austenite (Cγ) was calculated by using the lattice parameters of the retained austenite as reported in [39]. Dislocation density was calculated from the line profile of X-ray measurement where breadth of the microstrain distribution, ε, was deduced from the Williamson and Hall plot. The dislocation density thus calculated from the microstrain broadening was [41]:

predicted for the paraequilibrium condition. This phenomenon is called incomplete transformation [9,11–13]. Therefore, carbon-enriched retained austenite is not thermodynamically stable. After the transformation stasis stage, the austenite decomposition resume with increasing austempering time. The sequence, in general, is carbide precipitation first [14] followed by dislocation free ferrite or pearlite forming [15–17] and final bainite plates coalescing [18]. This is consistent with the concept of the thermal stability of retained austenite, i.e., the resistance of retained austenite to thermally decompose into more thermodynamically stable phases during maintaining at the bainitic transformation temperature or tempering temperature [19,20]. The carbide-free bainite microstructure and mechanical properties strongly depend on the austempering time [21–24]. In recent years, researchers have reported a significant amount of work into the carbide-free bainite microstructure and mechanical properties after specific time interval [25–31]. Mandal et al. [32] have reported that the best combination of strength and ductility is observed when austempering time is 10–30 min in a 0.13 wt% C and 1.2 wt% Si steel. The carbides are precipitated when austempering time increases from 30 min to 60 min, which accelerates the initiation of dimples and cleavage facets and deteriorates the tensile properties. Mondal et al. [33] have worked on the optimisation of the austempering time for high strength and elongation in a 0.3 wt% C and 1.76 wt% Si steel. They finally conclude that at austempering time of 30 min maximum yield strength (1557 MPa) with good elongation (15.5%) is obtained. The yield strength and elongation decrease with the increase in holding duration from 30 min to 60 min. This is related to the carbide formation in nano-scale with prolong austempering, which further reduces the carbon content of retained austenite and therefore decreases its thermal stability against martensite formation upon quenching after specific time interval during isothermal annealing. Hence, ultrahigh silicon steels should be developed to virtually delay the carbide precipitation even when the austempering time is extremely long. In addition, structural components are unavoidably subjected to the cyclic loadings, which results in fatigue failure. An understanding of the effects of austempering time within transformation stasis on the low cycle fatigue properties is critical to the design of structural components. Therefore, it is the aim of this work to research the effects of austempering time within transformation stasis on the bainite microstructure, monotonic deformation behavior, and especially cyclic deformation behavior using an ultrahigh silicon steel. The deformation and failure mechanisms of the monotonic tensile and the low cycle fatigue were investigated carefully.

ρ=

k ε2 ∙ F b2

(1)

where for body-centred cubic metals, k equals to 14.4 with the Burgers vector of dislocations, b, along < 111 > . The value of F is assumed to be 1. Monotonic tensile and low cycle fatigue tests were carried out using an MTS810 servohydraulic tensile testing machine. Monotonic tensile tests were conducted at a strain rate of 2 × 10−3 s−1 using standard tensile samples with a gauge diameter of 5 mm and a gauge length of 25 mm. Low cycle fatigue tests under total strain control, with total strain amplitudes lying between 0.52 × 10−2 and 1.0 × 10−2, were conducted using a triangular strain waveform (strain ratio, R = −1) at a constant strain rate of 6 × 10−3 s−1. Cylindrical samples with a gauge diameter of 5 mm and a gauge length of 10 mm were prepared for the low cycle fatigue tests. Tensile and low cycle fatigue tests were respectively conducted with three samples for each test status to check the repeatability. 3. Results 3.1. Bainite microstructure The expansion curves of the ultrahigh silicon steel austempered at 350 °C for 0.5 h and 4 h were shown in Fig. 1a. These expansion curves indicated that bainite transformation proceeded rapidly within 0.5 h and the transformation stasis occurred between 0.5 h and 4 h. After austempered, both samples were quenched by liquid nitrogen to room temperature, during which no turning point was observed in the expansion curves, as shown in Fig. 1b. This result indicated that when the bainite transformation was completed at 350 °C, the martensite start temperatures of retained austenite was lower than room temperature and therefore retained austenite did not transform to martensite during the cooling process after austempering. The microstructures of the investigated steel austempered at 350 °C for 0.5, 1, 2, and 4 h were verified by SEM and TEM. It was observed that the samples austempered at various times presented a similar microstructure. Therefore, the 0.5 h and 4 h samples were selected as the representatives and presented in Fig. 2. No carbide precipitations were observed in the TEM microstructures, indicating that ultrahigh silicon (2.59 wt%) indeed retarded carbide formation. The microstructure consisted essentially of a mixture of two phases, carbide-free bainitic ferrite and carbon-enriched retained austenite. To determine the true thicknesses of bainite plates and retained austenite films, totally 1000 measurements were made from TEM images by using the mean linear intercept method, and the final data were stereological corrected by the factor π/2 [42]. The results revealed that the true thicknesses of bainite plates and retained austenite films were approximately 80 and 40 nm, respectively. Comparing the TEM images of the samples austempered

2. Experimental An ultrahigh-silicon high-carbon nanobainite steel was studied and its chemical composition was Fe–0.70 C–2.59 Si–0.63 Mn–0.59 Cr wt%. The purpose of ultrahigh silicon was to ensure that no carbides were precipitated at the high bainite austempering temperature for a long period of time in high carbon steels [34–36]. A 40 kg cast of the investigated steel was produced using a vacuum induction furnace and then hot forged with a forging ratio of about 6. To ensure that the bainite transformation at 350 °C was terminated, the austempering time was set at 0.5 h and above, based on the expansion curve measured using a DIL805A/D dilatometer (Fig. 1a). The samples for the measurements of microstructure and mechanical properties were initially heated to 900 °C for 60 min and then placed in a salt bath furnace at 350 °C for 0.5, 1, 2, and 4 h followed by air cooling to room temperature. These samples are denoted as 0.5, 1, 2, and 4 h samples hereafter. The microstructures of undeformed samples and of ones close to the fatigue fracture surface were characterized by scanning electron microscopy (SEM, Hitachi SU-5000, 15 kV) and transmission electron microscopy (TEM, JOEL-2010, 200 kV). SEM samples were prepared through mechanical polishing and etching in 4% Nital solution [37]. TEM slices were ground to 30 µm using a SiC abrasive paper and then 81

Materials Science & Engineering A 751 (2019) 80–89

J. Zhao, et al.

Fig. 1. Relative length change as a function of (a) austempering time and (b) austempering temperature. Ms and Ms′ are the martensite start temperatures before and after isothermal holding, respectively.

shown in Fig. 3b. It can be seen from Fig. 3b that the diffusion distances of carbon in blocky retained austenite of the 0.5, 1 and 2 h samples were much less than the maximum equivalent radius of blocky retained austenite of ~2 µm. This result showed that a carbon concentration gradient occurred in some blocky retained austenite of the samples austempered for shorter time, where the carbon content of blocky retained austenite near the interfaces was higher than that of the interior. With increasing austempering time, the carbon concentration gradient in blocky retained austenite tended to promote the carbon diffusion near the interface into the interior of blocky retained austenite. Fig. 3b

for different times, the dislocation density decreased clearly with increasing austempering time, as shown in Fig. 2b, d. Seeing that the microstructural morphology and phase size did not change with increasing austempering time, we selected the 0.5 h sample to measure the equivalent radius of blocky retained austenite (Rγb) using the Image-Pro Plus software from more than 20 SEM images and its distribution was presented in Fig. 3a. Clearly, the equivalent radius of blocky retained austenite was distributed within 2 µm. The diffusion distances of carbon in blocky retained austenite of the samples austempered at 350 °C for different times was calculated by Eq. (2) [43], as

Fig. 2. SEM and TEM images of samples austempered at 350 °C for (a, b) 0.5 h and (c, d) 4 h. γb is blocky retained austenite. 82

Materials Science & Engineering A 751 (2019) 80–89

J. Zhao, et al.

Fig. 3. (a) Distribution of equivalent radius of blocky retained austenite (Rγb). (b) Average diffusion distance of carbon in blocky retained austenite as a function of the average time required for carbon homogenization in austenite at 350 °C.

decreased slightly due to diffusion into the interior, the carbon atoms in bainite plates can diffuse into blocky retained austenite to compensate for the carbon loss of blocky retained austenite near the interfaces. This caused that the carbon content of blocky retained austenite near the interfaces to maintain the one given by the T0′ curve and thus bainite transformation cannot be performed thermodynamically. Following this theory, the carbon content of retained austenite increased with increasing austempering time. Morales-Rivas et al. [44] also observed an increase in the carbon content of retained austenite during prolonged holding. They attributed this increase to a release of the carbon segregated at dislocation core resulting from the dislocation recovery in bainitic ferrite. As can be observed in Fig. 4, a decrease in the dislocation density of bainitic ferrite from 5.73 ± 0.04 × 1015 m−2 to 5.01 ± 0.05 × 1015 m−2 indeed occurred with increase austempering time. Therefore, it can be speculated that the carbon atoms captured at the dislocation in bainite plates perhaps compensated for the carbon loss of blocky retained austenite in the vicinity of the interfaces as the austempering time was extended from that needed to complete bainite transformation.

showed that the diffusion distance of carbon in blocky retained austenite of the 4 h samples reached the maximum equivalent radius of blocky retained austenite, indicating that the carbon distribution in blocky retained austenite of the 4 h samples became homogenized.

l̅ =

Q ⎞ 6tD0exp ⎛− ⎝ RT ⎠

(2)

where l¯ is the average diffusion distance of carbon in blocky retained austenite or bainite plates. t is the average time required for carbon homogenization in blocky retained austenite or bainite plates. D0 is a constant (D0γ = 0.10 × 10−4 m2/s, D0α = 0.62 × 10−6 m2/s). Q is the activation energy of carbon diffusion (Qγ = 135.7 kJ/mol, Qα = 80.4 kJ/mol). R is the gas constant (8.314 J/(mol K)). T is the austempering temperature in kelvin (623.15 K). The volume fraction (Vγ) and carbon content (Cγ) of retained austenite and the dislocation density of bainitic ferrite (ραBF) were measured by XRD, as shown in Fig. 4. The volume fractions of retained austenite in all microstructures were almost identical, approximately 22.5 vol%, indicating the retained austenite was indeed not decomposed during the long-term bainite transformation stasis. The carbon content of retained austenite increased slightly from 1.19 ± 0.03 to 1.40 ± 0.03 wt% during prolonged holding. According to the calculation by Eq. (2), the carbon diffusion from bainite plates (with a true thickness of ~80 nm) to retained austenite occurred within 0.01 s, while it taken 4 h to accomplish carbon homogenization in blocky retained austenite during austempering at 350 °C. Consequently, when the carbon content of blocky retained austenite near the interfaces

3.2. Monotonic deformation behavior Fig. 5a showed the tensile properties of the ultrahigh silicon steel austempered for different times. With increasing austempering time from 0.5 to 4 h, the yield and ultimate strengths decreased, while the uniform and total elongations increased. The 4 h sample obtained highest product of strength and elongation, i.e., the best combination of strength and ductility. Fig. 5b showed the comparison of the XRD patterns between the undeformed samples and the fractured samples near the fracture. XRD patterns of all the undeformed samples showed the presence of a predominantly ferrite with some retained austenite, the volume fractions of which were approximately 22.5 vol% as shown in Fig. 4. But no retained austenite peaks were observed in the XRD patterns of all the fractured samples near the fracture, indicating that the volume fraction of retained austenite much below the detection limit of the XRD. In other words, all of the retained austenite in the samples austempered for different times transformed to martensite during tensile deformation. Therefore, the monotonic tensile properties were primary affected by the transformation induced plasticity (TRIP) effect [45,46]. It will be discussed later in detail. 3.3. Cyclic deformation behavior

Fig. 4. Volume fraction (Vγ) and carbon content (Cγ) of retained austenite, as well as dislocation density of bainitic ferrite ( ραBF ), as a function of aus-

Fully reversed strain-controlled fatigue tests were conducted on the 0.5 h and 4 h samples. In Fig. 6, the total, elastic, and plastic strain amplitudes (εt , εe , and εp respectively) corresponding to half-life were

tempering time. Error bars represent 95% confidence intervals. 83

Materials Science & Engineering A 751 (2019) 80–89

J. Zhao, et al.

Fig. 5. (a) Yield strength (YS), tensile strength (TS), uniform elongation (UE), total elongation (TE), and product of strength and elongation (PSE) as a function of austempering time. Error bars represent 95% confidence intervals. (b) The comparison of the XRD patterns between the undeformed samples and the fractured samples near the fracture. The abbreviation “a.u.” represents arbitrary units.

amplitude, primarily attributed to the higher yield strength of the former sample (Fig. 5a). Quantitative assessment from Fig. 7c showed that, comparing with the 0.5 h sample, the cyclic hardening rate of the 4 h sample decreased and its cyclic softening rate increased under any given total strain amplitude. During cyclic deformation the metastable retained austenite in the 0.5 and 4 h samples transformed to martensite. The transformation ratios of retained austenite of both the samples under all total strain amplitudes were shown in Fig. 8. As well known, mechanical stability referred to the resistance of retained austenite to transform into martensite upon deformation [19,20,49]. Following this concept, compared with the 4 h sample, there was more retained austenite transformed to martensite in the 0.5 h sample under any given total strain amplitude, indicating the mechanical stability of retained austenite increased with increasing austempering time. The mechanical stability of retained austenite was generally controlled by its grain size, morphology, and carbon content [50,51]. In this paper, grain size and morphology of retained austenite were similar in both the samples (Fig. 2). Therefore, the higher mechanical stability of retained austenite in the 4 h sample arose from its higher carbon content. With increasing total strain amplitude from 0.52 × 10−2 to 1.0 × 10−2, the transformation ratios of retained austenite in the 0.5 h and 4 h samples increased and reached the maxima of 35 ± 1.3% and 26 ± 1.5%, respectively, at the total strain amplitude of 0.52 × 10−2. The ratio of the volume fraction of blocky retained austenite to that of total retained austenite estimated according to equations published in the literature [52] was approximately 48.3%, much larger than the maximum transformation ratios of retained austenite in both the samples during cyclic deformation. As well known, blocky retained austenite was less mechanical stable than filmy retained austenite and therefore tended to undergo deformationinduced martensite transformation preferentially under the cycle loading [53]. It was quite likely that most of the retained austenite transformed to martensite was blocky during cyclic deformation. The amount of filmy retained austenite transformed to martensite during cyclic deformation was so few that it can be ignored. Therefore, the transformation behavior of blocky retained austenite during cyclic deformation was mainly analyzed later. Fig. 9 typically showed the TEM microstructure and fatigue cracks in the 0.5 h sample cycled to failure (Nf ~ 1118) under the total strain amplitude of 1.0 × 10−2. It can be seen from Fig. 9a that one retained austenite block was transformed to multiple small martensite grains. As well known, the carbon was an important chemical element enhancing the mechanical stability of retained austenite [45,51]. It was possible that the inhomogeneous carbon distribution within the interior of retained austenite blocks (Fig. 3) significantly affected their local deformation-induced martensitic transformation. The newly formed martensite and the surrounding untransformed austenite bore a

Fig. 6. Strain–life curves of samples austempered at 350 °C for 0.5 and 4 h. εt , εe , and εp are the total, elastic, and plastic strain amplitudes corresponding to halflife, respectively.

plotted against the number of reversals to failure on a log–log scale. Clearly, higher fatigue life occurred in the 0.5 h sample, instead of in the 4 h sample with higher plasticity (Fig. 5). The relationships of Basquin and Manson–Coffin, as shown in Eq. (3) [47], were used to fit the fatigue data, indicating that the elastic strain amplitudes were considerably higher than the plastic strain amplitudes under all total strain amplitudes selected in this study. Therefore, the elastic strain amplitudes played a leading role. The fatigue life of the investigated steel strongly depended on the strength. This phenomenon was also found in other high strength materials, such as pre-hardening Hardfield steels [48] and high strength aluminium alloys [47], etc.

εt = εe + εp =

σ ′f E

(2Nf )b + ε′f (2Nf )c

(3)

Where εt , εe , and εp are total, elastic, and plastic strain amplitudes corresponding to half-life, respectively. σ′f and b are fatigue strength coefficient and exponent (Basquin), respectively. ε′f and c are fatigue ductility coefficient exponent c (Manson-Coffin), respectively. E is elastic modulus. 2Nf is number of reversals to failure. Fig. 7a, b demonstrated the similar cyclic stress responses of the 0.5 h and 4 h samples under various total strain amplitudes. Initial cyclic hardening followed by saturation and then cyclic softening till failure under low total strain amplitude, or direct cyclic softening till failure under high total strain amplitude was demonstrated in both the samples. Obviously, the cyclic stress amplitudes of the 0.5 h sample were always higher than those of the 4 h sample at any given strain 84

Materials Science & Engineering A 751 (2019) 80–89

J. Zhao, et al.

Fig. 7. Cyclic stress response under different total strain amplitudes of samples austempered at 350 °C for (a) 0.5 h and (b) 4 h, as well as their (c) corresponding cyclic hardening and softening ratios.

In comparison, the TEM microstructure and fatigue cracks in the 4 h sample cycled to failure (Nf ~ 781) under the total strain amplitude of 1.0 × 10−2 were shown in Fig. 10. Note that, one retained austenite block was only transformed to one martensite grain (confirmed by the electron diffraction of α martensite in Fig. 10b), so that the newly formed martensite was adjacent mainly to surrounding bainite plates (Fig. 10a). A high density of dislocations got entrapped at the αM/αBF interfaces, making them “shaggy” in appearance (Fig. 10a). Fig. 10c, d clearly revealed that lots of fatigue cracks appeared in the vicinity of the fatigue fracture surface and propagated along the αM/αBF interfaces. 4. Discussion In the present paper the samples austempered at 350 °C for various times within transformation stasis exhibit similar cyclic stress responses, i.e., initial cyclic hardening followed by cyclic saturation and then softening, or by direct cyclic softening till failure, depending on the total strain amplitude applied (Fig. 7a, b). However, the cyclic hardening ability decreases and the cycle softening ratio increases with increasing austempering time (Fig. 7c). As already known, phase fractions, phase sizes, and morphology of bainite microstructures are similar in all the samples austempered for various times (Figs. 2 and 3), so that the difference in cyclic hardening/softening behaviors is presumably related, directly or indirectly, to the dislocation density of bainitic ferrite and the carbon content (especially local carbon content) of retained austenite (Figs. 3 and 4). During bainite transformation, mobile dislocations are introduced by the plastic accommodation of the shape deformation which is an

Fig. 8. Transformation ratio of retained austenite under different total strain amplitudes in the 0.5 and 4 h samples fatigue tested to failure. The negative value of the ordinate indicates a reduced volume fraction of retained austenite. Error bars represent 95% confidence intervals.

Kurdjumov–Sachs (K–S) orientation relationship, i.e., (111)γb// (011)αM, [01¯1]γb//[1¯1¯1]αM, as confirmed by the electron diffraction in Fig. 9b. As observed from Fig. 9c, d, there were only a few fatigue cracks close to the fatigue fracture surface of the 0.5 h sample. Moreover, these fatigue cracks did not propagate along the deformation-induced martensite but through the bainite sheaves. 85

Materials Science & Engineering A 751 (2019) 80–89

J. Zhao, et al.

Fig. 9. In the 0.5 h sample cycled at ± 1.0 × 10−2 total strain amplitude (Nf ~ 1118) (a) the deformation-induced martensite from blocky retained austenite and (b) selected area electron diffraction pattern of the αM/γb interfaces in (a), as well as (c) low- and (d) high-magnification SEM images of fatigue cracks. γb, αM, and αBF are blocky retained austenite, deformation-induced martensite, and bainite plates, respectively. 1

invariant-plane strain with a large shear component and a volume change normal to the plane [54]. In such a case, Bhadeshia et al. [55,56] assert that most of pre-existent dislocations in bainitic ferrite are mobile dislocations. The nanoscale bainite plates with the true thickness of approximately 80 nm (Fig. 2) in the investigated steel limit the mean free path of mobile dislocations and thus promote their multiple/cross slips under the cycle loading. These multiple/cross slips tend to make mobile dislocations entangle and immobile during the initial cycles, resulting in the sharp decrease in the density of mobile dislocations. An empirical relationship between of the density of mobile dislocations and the effective stress acting on mobile dislocations discovered by Johnston and Caillard [57–59] is shown in Eq. (4). It is indicated that the effective stress acting on mobile dislocations (necessary to maintain the plastic strain rate imposed) increases significantly with the decrease in the density of mobile dislocations during plastic deformation. Therefore, reasonably, the decrease in the density of mobile dislocations can enhance the initial cycle hardening in low cycle fatigue domain. This theory is also applied by Qian et al. [60]. They, by careful calculation, report that the primary reason for the initial cyclic hardening of bainite steels with extremely high density of dislocations is neither the increase in dislocation density nor the deformation-induced martensite transformation, but the change in the dislocation mobility. Therefore, comparing with the 4 h sample, the higher cyclic hardening ability of the 0.5 h sample is mainly attributed to its higher density of mobile dislocation pre-existed in bainitic ferrite (Fig. 4).

ε¯p ⎞ n σ* = τ0 ⎛⎜ ⎟ ⎝ Φρm b ⎠

(4)

Where σ* is the effective stress acting on mobile dislocations. τ0 is the effective stress at a unit velocity of dislocations. ε̅p is the plastic strain rate. Φ is the geometric factor. ρm is the density of mobile dislocations. b is the Burgers vector. n is the dislocation velocity–stress exponent. Retained austenite tends to transform to martensite during the cyclic hardening stage of the low cycle fatigue [61–63]. Given that the martensite transformation can absorb considerably more of the energy used for fatigue crack propagation, the initiation of fatigue cracks is delayed and the first stage of fatigue crack propagation is extended [60,64]. However, this positive effect is limited by the low transformations ratio of retained austenite (Fig. 8) and only occurs in the moment of shear transformation (i.e., the cyclic hardening stage of the low cycle fatigue [61]). After the cyclic hardening stage, the newly formed martensite is likely to become a fast path for fatigue crack propagation due to its low toughness [65–67]. In this paper, most of fatigue cracks propagate along the deformation-induced martensite blocks in the 4 h sample but unexpectedly through the bainite sheaves in the 0.5 h sample. Correspondingly, one retained austenite block can transform to multiple small martensite grains in the 0.5 h sample but to only one large martensite grain in the 4 h sample (Figs. 9 and 10). There must be a certain relationship between them. Because one retained austenite block can transform to multiple small martensite grains in the 0.5 h sample, the small martensite grains in retained austenite blocks are mainly adjacent to nearby retained 86

Materials Science & Engineering A 751 (2019) 80–89

J. Zhao, et al.

Fig. 10. In the 4 h sample cycled at ± 1.0 × 10−2 total strain amplitude (Nf ~ 781) (a) the deformation-induced martensite from blocky retained austenite and (b) selected area electron diffraction pattern of αM in (a), as well as (c) low- and (d) high-magnification SEM images of fatigue cracks.

life. However, all the retained austenite is transformed to martensite during the monotonic tensile deformation (Fig. 5b), owing to the extremely high strain levels in excess of 30% (Fig. 5a). Consequently, the deformation-induced martensite transformation is the most primary factor affecting the monotonic tensile properties. Owing to the higher carbon content of retained austenite in the 4 h sample than that the 0.5 h sample (Fig. 4), the retained austenite in the 4 h sample possesses higher mechanical stability. The proportion of retained austenite transformed to martensite at larger strain regime is larger in the 4 h sample. This relaxes the local stress concentration at higher strains and well delays the necking in the 4 h sample, hence increasing the elongation, especially uniform elongation (Fig. 5a). For the 4 h sample, although the lower dislocation density strengthening and solid solution strengthening from bainite ferrite (Fig. 4) are not conducive to improving the yield strength, the progressive increase in volume fraction of hard phase martensite formed at larger strain regime well enhances the strain hardenability, which is more beneficial to the increase in the tensile strength (Fig. 5a) [49]. Thus, the 4 h sample obtains the higher elongation (especially uniform elongation) without the large reduce in the tensile strength, i.e., the best combination of strength and ductility (Fig. 5a).

austenite (Fig. 9a). The shear characteristic of martensite transformation requires coherent (or semi-coherent) interfaces between the newly formed martensite and the parent austenite, which effectively reduces the resistant force of dislocation movement from martensite to nearby retained austenite [68]. Therefore, the stress/strain between two phases can be well coordinated through αM/γb interfaces. The stress concentration is slight, and fatigue cracks cannot be initiated at the αM/γb interfaces but only through bainite sheaves (Fig. 9c, d). Filmy retained austenite possesses excellent plasticity to suppress the initiation and propagation of fatigue cracks [69], resulting in the rare presence of fatigue cracks close to the fatigue fracture surface of the 0.5 h sample. In contrast, because one retained austenite block can only transform to one large martensite grain in the 4 h sample, the newly formed martensite grains are adjacent to surrounding bainite plates (Fig. 10a). As the cycling progresses, a high density of dislocations is piled up at the αM/αBF interfaces (Fig. 10a) due to the development of large localized elastic incompatibility stresses between martensite blocks and bainite plates. This essentially accelerates the initiation and propagation of fatigue cracks from the αM/αBF interfaces (Fig. 10c, d), as reported by Hu and Johansson [65,67]. This phenomenon, unavoidably, increases the cyclic softening and degrades the fatigue life (Fig. 7). Longer austempering time in transformation stasis does not increase the low cycle fatigue life (Fig. 6) but improve the combination of strength and ductility of the monotonic tensile samples (Fig. 5a). It is because the primary factors affecting these mechanical properties are different. During cyclic deformation with maximum total strain amplitude of 1.0% (Fig. 6), the small transformation ratios of retained austenite (Fig. 8) limit the contribution of the TRIP effect to the fatigue

5. Conclusions In this work, bainite microstructure, monotonic deformation behavior, and especially cyclic deformation behavior of an ultrahigh silicon carbide-free nanobainite steel austempered at 350 °C for different time within transformation stasis were investigated. Based on the 87

Materials Science & Engineering A 751 (2019) 80–89

J. Zhao, et al.

mechanical tests and accompanying microstructural analyses, the following conclusions can be drawn:

[13] S. Li, R. Zhu, I. Karaman, R. Arróyave, Development of a kinetic model for bainitic isothermal transformation in transformation-induced plasticity steels, Acta Mater. 61 (2013) 2884–2894. [14] B. Avishan, C. Garcia-Mateo, S. Yazdani, F.G. Caballero, Retained austenite thermal stability in a nanostructured bainite steel, Mater. Charact. 81 (2013) 105–110. [15] E.P. Klier, T. Lyman, The bainite reaction in hypoeutectoid steels, Trans. ASM 158 (1944) 395–422. [16] T. Furuhara, T. Yamaguchi, G. Miyamoto, T. Maki, Incomplete transformation of upper bainite in Nb bearing low carbon steels, Mater. Sci. Technol.-Lond. 26 (2010) 392–397. [17] Y. Xia, G. Miyamoto, Z.G. Yang, C. Zhang, T. Furuhara, Direct measurement of carbon enrichment in the incomplete bainite transformation in Mo added low carbon steels, Acta Mater. 91 (2015) 10–18. [18] C. Liu, Z.B. Zhao, S.D. Bhole, Lathlike upper bainite in a silicon steel, Mater. Sci. Eng. A 434 (2006) 289–293. [19] A. Kammouni, W. Saikaly, M. Dumont, C. Marteau, X. Bano, A. Charaï, Effect of the bainitic transformation temperature on retained austenite fraction and stability in Ti microalloyed TRIP steels, Mater. Sci. Eng. A 518 (2009) 89–96. [20] J. Hidalgo, K.O. Findley, M.J. Santofimia, Thermal and mechanical stability of retained austenite surrounded by martensite with different degrees of tempering, Mater. Sci. Eng. A 690 (2017) 337–347. [21] P. Luo, G.H. Gao, H. Zhang, Z.L. Tan, R.D. Misra, B.Z. Bai, On structure-property relationship in nanostructured bainite steel subjected to the quenching and partitioning process, Mater. Sci. Eng. A 661 (2016) 1–8. [22] C. Hofer, F. Winkelhofer, H. Clemens, S. Primig, Morphology change of retained austenite during austempering of carbide-free bainite steel, Mater. Sci. Eng. A 664 (2016) 236–246. [23] Q. Zhou, L.H. Qian, J. Tan, J.Y. Meng, F.C. Zhang, Inconsistent effects of mechanical stability of retained austenite on ductility and toughness of transformation-induced plasticity steels, Mater. Sci. Eng. A 578 (2013) 370–376. [24] Y.Q. Huo, X.H. Long, Z.H. Zhou, J.G. Li, Bainite transformation and TRIP effect in 20Mn2SiVB steel, Mater. Sci. Eng. A 438–440 (2006) 158–161. [25] G. Sidhu, S.D. Bhole, E. Essadiqi, D.L. Chen, Characterization of isothermally heattreated high carbon nanobainite steels, J. Mater. Eng. Perform. 22 (2013) 3070–3076. [26] S. Sharma, S. Sangal, K. Mondal, Development of new high-strength carbide-free bainite steels, Metall. Mater. Trans. A 42 (2011) 3921–3933. [27] S. Baradari, S.M.A. Boutorabi, Effects of isothermal transformation conditions on the microstructure and hardness values of a high-carbon Al–Si alloyed steel, Mater. Des. 86 (2015) 603–609. [28] M.N. Yoozbashi, S. Yazdani, Mechanical properties of nanostructured, low temperature bainite steel designed using a thermodynamic model, Mater. Sci. Eng. A 527 (2010) 3200–3205. [29] S. Chen, G.Z. Wang, C. Liu, C.C. Wang, X.M. Zhao, W. Xu, Correlation of isothermal bainite transformation and austenite stability in quenching and partitioning steels, J. Iron Steel Res. Int. 24 (2017) 1095–1103. [30] F.G. Caballero, M.K. Miller, C. Garcia-Mateo, Carbon supersaturation of ferrite in a nanocrystalline bainite steel, Acta Mater. 58 (2010) 2338–2343. [31] F.G. Caballero, H.K.D.H. Bhadeshia, Very strong bainite, Curr. Opin. Solid State Mater. 8 (2004) 251–257. [32] D. Mandal, M. Ghosh, J. Pal, S.G. Chowdhury, G. Das, S.K. Das, S. Ghosh, Evolution of microstructure and mechanical properties under different austempering holding time of cast Fe–1.5Si–1.5Mn–V steels, Mater. Des. 54 (2014) 831–837. [33] A. Misra, S. Sharma, S. Sangal, A. Upadhyaya, K. Mondal, Critical isothermal temperature and optimum mechanical behaviour of high Si-containing bainite steels, Mater. Sci. Eng. A 558 (2012) 725–729. [34] X.Y. Long, J. Kang, B. Lv, F.C. Zhang, Carbide-free bainite in medium carbon steel, Mater. Des. 64 (2014) 237–245. [35] L.H. Qian, Q. Zhou, F.C. Zhang, J.Y. Meng, M. Zhang, Y. Tian, Microstructure and mechanical properties of a low carbon carbide-free bainite steel co-alloyed with Al and Si, Mater. Des. 39 (2012) 264–268. [36] M.N. Yoozbashi, S. Yazdani, T.S. Wang, Design of a new nanostructured, high-Si bainite steel with lower cost production, Mater. Des. 32 (2011) 3248–3253. [37] J.L. Zhao, Z.N. Yang, F.C. Zhang, Study on carbide-free bainite microstructure and mechanical properties of 70Si3Mn steel, J. Yanshan Univ. 39 (2015) 199–212. [38] J.L. Zhao, F.C. Zhang, B.D. Yu, H. Liu, Bainite microstructure and its tempering stability of 70Si3MnCrMo steel, Iron Steel 52 (2017) 65–75. [39] J. Kang, F.C. Zhang, X.W. Yang, B. Lv, K.M. Wu, Effect of tempering on the microstructure and mechanical properties of a medium carbon bainite steel, Mater. Sci. Eng. A 686 (2017) 150–159. [40] C. Garcia-Mateo, F.G. Caballero, H.K.D.H. Bhadeshia, Mechanical properties of lowtemperature bainite, Mater. Sci. Forum 500–501 (2005) 495–502. [41] G.K. Williamson, R.E. Smallman III., Dislocation densities in some annealed and cold-worked metals from measurements on the X-ray Debye-Scherrer spectrum, Philos. Mag. 1 (1956) 34–46. [42] L.C. Chang, H.K.D.H. Bhadeshia, Austenite films in bainitic microstructures, Mater. Sci. Technol. 11 (1995) 874. [43] G.H. Gao, H. Zhang, X.L. Gui, Enhanced work hardening capacity in a lean alloy steel treated by a “disturbed” bainitic austempering process, Acta Mater. 101 (2015) 31–39. [44] L. Morales-Rivas, H.W. Yen, B.M. Huang, M. Kuntz, F.G. Caballero, J.R. Yang, C. Garcia-Mateo, tensile response of two nanoscale bainite composite-like structures, Jom-US 67 (2015) 2223–2235. [45] T. Sourmail, F.G. Caballero, F. Moudian, D. De Castro, M. Benito, High hardness and retained austenite stability in Si-bearing hypereutectoid steel through new heat treatment design principles, Mater. Des. 142 (2018) 279–287.

1. With increasing austempering time the true thicknesses of bainite plates and retained austenite films (approximately 80 and 40 nm, respectively) and the size distribution of retained austenite blocks did not change. The changes in the microstructures included a decrease in the dislocation density of bainitic ferrite and a carbon enrichment and homogenization in retained austenite. The longer time austempering within transformation stasis increased the combination of strength and ductility, but did not result in longer fatigue life. 2. The decrease in the fatigue life of the samples austempered for a longer time result from two reasons. On the one hand, the lower density of mobile dislocation initially existed in the starting microstructure weakened the contribution of mobile dislocation to the cyclic hardening. On the other hand, one retained austenite block with completely homogeneous carbon distribution was only transformed to one martensite grain, so that the large localized elastic incompatibility stresses developed at the interfaces between martensite blocks and bainite plates. This accelerated the initiation and propagation of fatigue cracks, therefore increasing the cyclic softening and decreasing the fatigue life. 3. The small retained austenite transformation ratios limited the contribution of the TRIP effect to the fatigue life. But the TRIP effect significantly affected the monotonic tensile properties because all the retained austenite transformed to martensite during the monotonic tensile deformation. In the samples austempered for a longer time, retained austenite possessed higher mechanical stability and therefore can transform to martensite at larger strain regime. Thus, the necking was been well delayed and the strain hardenability at higher strains been well enhanced, which significantly increased the elongation (especially uniform elongation) without the large reduce in the tensile strength. Acknowledgement This work was supported by the National Natural Science Foundation of China [No. 51831008], the National Key R&D Program of China [No. 2017YFB0304501], and the Research Program of the College Science & Technology of Hebei Province [No. QN2018144]. References [1] H.K.D.H. Bhadeshia, A.R. Waugh, Bainite: an atom-probe study of the incomplete reaction phenomenon, Acta Metall. 30 (1982) 775–784. [2] F.G. Caballero, M.K. Miller, C. Garcia-Mateo, Tracking solute atoms during bainite reaction in a nanocrystalline steel, Mater. Sci. Technol.-Lond. 26 (2010) 889–898. [3] F.G. Caballero, M.K. Miller, S.S. Babu, C. Garcia-Mateo, Atomic scale observations of bainite transformation in a high carbon high silicon steel, Acta Mater. 55 (2007) 381–390. [4] A. Varshney, S. Sangal, S. Kundu, K. Mondal, Super strong and highly ductile low alloy multiphase steels consisting of bainite, ferrite and retained austenite, Mater. Des. 95 (2016) 75–88. [5] H.K.D.H. Bhadeshia, Rationalisation of shear transformations in steels, Acta Metall. 29 (1981) 1117–1130. [6] H. Chen, K.Y. Zhu, L. Zhao, S.V.D. Zwaag, Analysis of transformation stasis during the isothermal bainitic ferrite formation in Fe–C–Mn and Fe–C–Mn–Si alloys, Acta Mater. 61 (2013) 5458–5468. [7] H.K.D.H. Bhadeshia, Bainite in Steels: Theory and Practice, 3nd ed., Mancy Publ., Leeds, UK, 2015. [8] H.K.D.H. Bhadeshia, D.V. Edmonds, The mechanism of bainite formation in steels, Acta Metall. 29 (1980) 1265–1273. [9] A. Saha Podder, H.K.D.H. Bhadeshia, Thermal stability of austenite retained in bainitic steels, Mater. Sci. Eng. A 527 (2010) 2121–2128. [10] A.M. Ravi, J. Sietsma, M.J. Santofimia, Exploring bainite formation kinetics distinguishing grain-boundary and autocatalytic nucleation in high and low-Si steels, Acta Mater. 105 (2016) 155–164. [11] H.I. Aaronson, W.T. Reynolds Jr., G.R. Purdy, The incomplete transformation phenomenon in steel, Metall. Mater. Trans. A 37 (2006) 1731–1745. [12] H.D. Wu, G. Miyamoto, Z.G. Yang, C. Zhang, H. Chen, T. Furuhara, Incomplete bainite transformation in Fe-Si-C alloys, Acta Mater. 133 (2017) 1–9.

88

Materials Science & Engineering A 751 (2019) 80–89

J. Zhao, et al.

[58] J.P. Hirth, A brief history of dislocation theory, Metall. Trans. A 16A (1985) 2085–2090. [59] D. Caillard, J.L. Martin, Thermally Activated Mechanisms in Crystal Plasticity, Elsevier Ltd., Oxford, UK, 2003. [60] Q. Zhou, L.H. Qian, J.Y. Meng, L.J. Zhao, F.C. Zhang, Low-cycle fatigue behavior and microstructural evolution in a low-carbon carbide-free bainite steel, Mater. Des. 85 (2015) 487–496. [61] J. Kang, F.C. Zhang, X.Y. Long, B. Lv, Low cycle fatigue behavior in a mediumcarbon carbide-free bainite steel, Mater. Sci. Eng. A 666 (2016) 88–93. [62] M. Abareshi, E. Emadoddin, Effect of retained austenite characteristics on fatigue behavior and tensile properties of transformation induced plasticity steel, Mater. Des. 32 (2011) 5099–5105. [63] F.C. Zhang, X.Y. Long, J. Kang, D. Cao, B. Lv, Cyclic deformation behaviors of a high strength carbide-free bainite steel, Mater. Des. 94 (2016) 1–8. [64] T.B. Hilditch, I.B. Timokhina, L.T. Robertson, E.V. Pereloma, P.D. Hodgson, Cyclic deformation of advanced high-strength steels: mechanical behavior and microstructural analysis, Metall. Mater. Trans. A 40A (2009) 342–353. [65] Z.G. Hu, P. Zhu, J. Meng, Fatigue properties of transformation-induced plasticity and dual-phase steels for auto-body lightweight: experiment, modeling and application, Mater. Des. 31 (2010) 2884–2890. [66] S. Ackermann, D. Kulawinski, S. Henkel, H. Biermann, Biaxial in-phase and out ofphase cyclic deformation and fatigue behavior of an austenitic TRIP steel, Int. J. Fatigue 67 (2014) 123–133. [67] J. Johansson, M. Odén, Load sharing between austenite and ferrite in a duplex stainless steel during cyclic loading, Metall. Mater. Trans. A 31 (2000) 1557–1570. [68] Y. Wang, K. Zhang, Z.H. Guo, N.L. Chen, Y.H. Rong, A new effect of retained austenite on ductility enhancement in high strength bainite steel, Mater. Sci. Eng. A 552 (2012) 288–294. [69] X.Y. Long, F.C. Zhang, C.Y. Zhang, Effect of Mn content on low-cycle fatigue behaviors of low-carbon bainite steel, Mater. Sci. Eng. A 697 (2017) 111–118.

[46] J.L. Zhao, B. L, F.C. Zhang, Z.N. Yang, L.H. Qian, C. Chen, X.X. Long, Effects of austempering temperature on bainitic microstructure and mechanical properties of a high-C high-Si steel, Mater. Sci. Eng. A 742 (2019) 179–189. [47] A. Niesłony, C.E. Dsoki, H. Kaufmann, P. Krug, New method for evaluation of the Manson–Coffin–Basquin and Ramberg–Osgood equations with respect to compatibility, Int. J. Fatigue 30 (2008) 1967–1977. [48] C. Chen, B. Lv, F. Wang, F.C. Zhang, Low-cycle fatigue behaviors of pre-hardening Hadfield steel, Mater. Sci. Eng. A 695 (2017) 144–153. [49] C. Garcia-Mateo, F.G. Caballero, T. Sourmail, M. Kuntz, J. Cornide, V. Smanio, R. Elvira, Tensile behaviour of a nanocrystalline bainitic steel containing 3 wt% silicon, Mater. Sci. Eng. A 549 (2012) 185–192. [50] J.B. Seol, D. Raabe, P.P. Choi, Y.R. Im, C.G. Park, Atomic scale effects of alloying, partitioning, solute drag and austempering on the mechanical properties of highcarbon bainitic–austenitic TRIP steels, Acta Mater. 60 (2012) 6183–6199. [51] P.J. Jacques, E. Girault, A. Mertens, B. Verlinden, J.V. Humbeeck, F. Delanny, The developments of cold-rolled TRIP-assisted multiphase steels. Al-alloyed TRIP-assisted multiphase steels, ISIJ Int. 41 (2001) 1068–1074. [52] H.K.D.H. Bhadeshia, D.V. Edmonds, Bainite in silicon steels: new compositionproperty approach Part 1, Met. Sci. 17 (1983) 411–419. [53] Z.N. Yang, Y.L. Jia, F.C. Zhang, M. Zhang, B. Nawaz, C.L. Zheng, Microstructural evolution and performance change of a carburized nanostructured bainitic bearing steel during rolling contact fatigue process, Mater. Sci. Eng. A 725 (2018) 98–107. [54] J.H. Ryu, J.I. Kim, H.S. Kim, C.S. Oh, H.K.D.H. Bhadeshiaa, D.W. Suh, Austenite stability and heterogeneous deformation in fine-grained transformation-induced plasticity-assisted steel, Scr. Mater. 68 (2013) 933–936. [55] H.K.D.H. Bhadeshia, Bainite in Steel: Transformations, Microstructure and Properties, 2nd ed., IOM Communications Ltd., London, UK, 2001. [56] C. Garcia-Mateo, F.G. Caballero, Ultra-high-strength bainitic steels, ISIJ Int. 45 (2005) 1736–1740. [57] W.G. Johnston, J.J. Gilman, Dislocation velocities, dislocation densities, and plastic flow in lithium fluoride crystals, J. Appl. Phys. 30 (1959) 129–144.

89