Short Article
Enhancing the TRIP Effect by Pre-Ferrite in the Hot Rolled Fe-8Mn-4Ni-3Al-0.1C Steel
2025 年 65 巻 4 号 p. 571-575
詳細
2025 年 65 巻 4 号 p. 571-575
As is well known, reconstruction of the initial microstructure before an inter-critical annealing (IA) can optimize the final microstructure as well as improve mechanical performance. In this study, the initial microstructure with “pre-ferrite + martensite” was designed via two-stage cooling in the rolling process. The final microstructure observation showed that bimodal size distributed austenite and double morphologies of austenite. Moreover, lath-shaped austenite existed in the martensitic matrix; blocky austenite occurred at the boundaries of pre-ferrite. According to the numerical simulations of the IA process, blocky austenite tended to be more stable with smaller size and higher Mn concentrations than lath-shaped austenite. The characteristics of austenite were similar at different IA temperatures, but the lower IA temperatures, the smaller size of austenite. An excellent combination of strength (1308 MPa) and ductility (34%) was obtained at 928 K, which was ascribed to a positive TRIP effect induced by the austenite with a large stability gradient.
Medium-Mn steels (MMnS), as one of the third generation advanced high-strength steels for automobile steels, have received considerable attention from industry and academia in the last decade.1,2,3) Among MMnS, hot rolled ones may become the most hopeful candidate for car parts (B-pillar, cross member, etc.) due to their excellent comprehensive mechanical properties and relatively economical manufacturing costs.
In general, the MMnS with the initial microstructure of full martensite suffered from an inter-critical annealing process (IA), where ordinary martensite transformed into lath-shaped austenite (γ) and ferrite. During IA, Mn/C atoms can diffuse from martensite to austenite, which enhances the chemical stability of austenite as well as that austenite could be retained at ambient temperatures.3,4,5,6,7) Metastable austenite would transform into hard martensite with deformation, i.e., transformation-induced plasticity (TRIP) effect, resulting in a localized work hardening to delay necking. Accordingly, the TRIP effect can enhance strength and ductility simultaneously. Moreover, by optimizing the stability of austenite, excellent products of strength and elongation (PSE) of 30–50 GPa% were achieved.1,2,3,4,5)
As is well known, the TRIP effect was directly related to the stability of austenite and its proportion in MMnS. The characteristics of retained austenite were strongly affected by an initial microstructure before IA. According to the study of Han et al.,5) an initial microstructure with “pre-existing ferrite + austenite + martensite” was obtained by pre-annealing in hot-rolled Fe-6.93Mn-1Al-1.07Si-0.095C steel. Accordingly, lath-shaped, blocky, and granular austenite occurred in the final microstructure, which resulted in a positive TRIP effect. Shao et al.8) reported an initial microstructure with “coarse δ-ferrite + martensite” resulting from an addition of 3%-Al in hot-rolled Fe-4.99Mn-3.03Al-0.63Si-0.2C steel. After IA, lath-shaped austenite and bimodal size distributed lath-shaped ferrite were obtained in the final microstructure, which resulted in a large TRIP effect. Though reconstruction of the initial microstructure based on the conventional fully martensitic matrix was widely investigated, it was still a lack of knowledge about the initial microstructure with “pre-ferrite (αpre) + martensite (α′)”, which was constructed via two-stage cooling in hot rolling process. Moreover, the growth mechanisms of austenite surrounding pre-ferrite and martensite, as well as their correlation with Mn/C partitioning kinetics during IA and corresponding deformation behavior were worth studying.
In the present work, we prepared MMnS via a “rolling process with two-step cooling + one-step inter-critical annealing”. We mainly focus on the correlation between austenite characteristics and the initial microstructure containing pre-ferrite. Meanwhile, Mn/C partitioning behavior, austenite reversion kinetics and the final mechanical properties were investigated.
The chemical composition of the designed steel was Fe-7.88Mn-3.95Ni-3.3Al-1.15Cr-0.1C (wt.%). The addition of Mn, Ni, and Cr ensured the chemical stability of austenite with relatively short IA time, while the addition of Al increased the temperature of the “austenite + ferrite region” to ensure the efficiency of rolling and heat treatment. The steel was cast into an inner 56mm-diameter mold and then hot-forged. The obtained billet was homogenized at 1373 K for 2.16×104 s, hot rolled in four passes to a thickness of 12 mm during 1083–1113 K (recrystallization-controlled rolling), followed by two-step cooling to ambient temperature. Two-step cooling was designed to accomplish the following objectives: (a) online cooling realized the precipitation of fine ferrite at prior austenite grain boundaries (PAGB); (b) water quenching ensured the transformation of austenite into martensite. The as-rolled microstructure, i.e., the initial microstructure was 12% ferrite (pre-ferrite) and 88% martensite, as shown in Fig. 1(c). The hot-rolled steel was further reheated to 928 K and 978 K for 1.8 ks, and then air cooled to ambient temperature. Thermodynamic calculations using Thermal-Calc software and the TCFE8 database suggested that austenite phase fractions were 60% at 928K and 70% at 978K, as shown in Fig. 1(b). The steels with different isothermal heat treatment temperatures of 928 K and 978 K were designated as “928K” and “978K” ones, respectively. The details of thermo-mechanical processing and heat treatments are shown in Fig. 1(a).
To analyze the influence of pre-existed ferrite on the microstructures, the sample was cut along rolling and thickness directions and characterized using field-emission scanning electron microscopy (SEM, Hitachi SU5000), electron backscatter diffraction (EBSD, OIM Collection 7.3), transmission electron microscope (TEM, FEI Talos F200X) and X-ray diffraction (XRD, MiniFlex600). The SEM samples were mechanically polished and etched in 4% nitric acid. The EBSD samples were mechanically polished and electropolished in a solution with “10 vol.% perchloric acid + 90 vol.% ethanol”, then determined at 30 kV by a scanning step of 0.05 μm. The data obtained were analyzed using the AZtecCrystal software. The TEM samples were thinned using dual-jet electro-polisher at a working voltage of 30 V in a 7% perchloric acid/glacial acetic acid solution at room temperature and were characterized at acceleration voltage 200 kV. The XRD samples were investigated using Cu radiation at voltage 40kV and current 15mA with a range of measuring diffraction angle (2θ: 40–110°), and the volume fraction of retained austenite was calculated from the integrated intensities of (200)α, (211)α, (200)γ, (220)γ and (311)γ diffraction peaks, according to the formula in Ref.9)
2.3. Tensile TestsThe specimens were machined into dog-bone-shape along the hot rolling direction with the gauge size of 10 × 4 × 2 mm3 and then conducted on the tensile testing machine (Autograph DCS-R-5000) at an initial strain rate of 2×10−3 s−1. All the reported tensile properties were taken as the averages of two or three repeated measurements. SEM was used to observe the fracture surfaces of specimens.
Figures 2(a) and 2(d) show representative SEM micrographs of 928K and 978K samples. Annealed martensite dominated the microstructures, which was inherited from as-quenched martensite. Typically, annealed martensite was enveloped by the white squares and the martensitic lath bundles were clear. Pre-ferrite was enveloped by a dotted ellipse. When IA temperature increased from 928 K to 978 K, the width of laths significantly increased, and the size of the pre-ferrite slightly decreased. Figures 2(b) and 2(e) show the phase distribution based on EBSD data. As can be seen, annealed martensite was composed of alternating lathy ferrite (αlath) and lathy austenite (γlath), while αpre was surrounded by blocky austenite (γblock). However, some austenite grains were not detected due to their small size and a relatively large scanning step of EBSD. This deduction can be supported by the TEM results in Figs. 2(g)–2(i). The width of γlath in the 928K sample varied in the range of 45–157 nm, as shown in Fig. 2(h); while it was in the range of 53–214 nm in the 978K sample, as shown in Fig. 2(g). The size range of γblock was larger than γlath, ranging from 10 nm to 210 nm, as shown in Fig. 2(h). Through Fig. 2(i), it can be seen that the positions of some austenite nucleation were close, and they can almost be considered as “a large austenite” at a low magnification field of view. So, it can be inferred that some large sized γblock may be “aggregates” of small γblock. In addition, there were a small amount of γblock with larger sizes due to nucleation at complex grain boundaries or sub-grain boundaries, which provided higher interfacial energy.10) Therefore, there were two morphologies of austenite, and the size of blocky austenite was smaller than lath-shaped ones. The statistical data of grain size in samples was summarized in Table 1. When IA temperature increased from 928 K to 978 K, the average width of lath-shaped austenite increased from 112 nm to 136 μm; the average size of blocky austenite increased from 27 nm to 35 nm. In steels whose initial microstructure were fully martensitic, almost all austenite was lath-shaped, with few blocky austenite at PAGBs.2,3,4) This suggested that blocky austenite nucleated and grew from boundaries of pre-ferrite and pre-ferrite effectively increasing the amount of blocky austenite.
| Samples | Width of αpre | Width of αlath | Width of γlath | Diameter of γblock |
|---|---|---|---|---|
| As rolled | 680 | 420 | – | – |
| 928K | 712 | 216 | 112 | 27 |
| 978K | 745 | 198 | 136 | 42 |
Figure 2(c) displays an orientation image map of austenite near a pre-ferrite in a 928K sample. The blocky austenite maintained individual orientation, while some lath austenite presented in the same variant. The typical representation of the crystallographic orientation relationship between austenite and neighbor ferrite was shown in Fig. 2(g), which corresponds to the analytical results of the region in Fig. 2(c). The lath-shaped ferrite (a part of annealed martensite) labeled as 5 was surrounded by lath-shaped austenite labeled as 3 and 4, and both 3 and 4 satisfied the K–S orientation relationship with 5 based on the (111) and (110) pole figure of austenite and ferrite, respectively. However, blocky austenite labeled as 2 was surrounded by pre-ferrite labeled as 1, and there was no K-S orientation. For lath-shaped austenite, they were reverted from a martensitic structure based on the “austenite memory”.9) Blocky austenite nucleated individually at the ferritic boundary with a non-K-S relationship between them and adjacent ferrite.
3.2. Austenite Growth ModesTo better understand the growth of the austenite and elements partitioning behavior with the presence of pre-ferrite in the martensitic matrix, the reversion kinetics of the two-type austenite in the 928K sample were simulated using DICTRA software based on TCFE8 and MobFe3 database assuming local equilibrium (LE). Figures 3(a) and 3(b) are the simulation cells of blocky austenite and lath-shaped austenite, respectively. The initial sizes of two-type austenite were assumed to be 1 nm. The half width values of ferrite and martensite lath were set at 340 nm and 210 nm, according to the statistics of initial microstructure. The initial chemical composition of phases was determined via bulk composition and calculations of thermo-calc software. Based on the above settings, the calculated results in Figs. 3(c)–3(f) can approximately represent the C/Mn diffusion profiles.
The evolution of C and Mn concentration near the γ/αpre interface is shown in Figs. 3(c) and 3(e). As IA time increased, the γ/αpre interfaces migrated into pre-ferrite, accompanied by a decrease in C concentration and an increase in Mn concentration. When the time exceeded 1000 s, Mn concentration within austenite tended to stabilize at 13.5 wt.%. The results of C and Mn concentration near the γ/α′ interface in Figs. 3(d) and 3(f) were like the above, but Mn concentration at 12.8 wt.%. Besides, at the same time, the migration distance of γ/αpre interface was smaller than that of γ/α′ interface, which implied that the size of the blocky austenite was relatively small in the final microstructure. This was consistent with microstructural observation. Therefore, based on the “size effect”11) and martensitic transformation thermodynamics, blocky austenite with high Mn composition and small size has higher stability than lath-shaped austenite.
As shown in Fig. 3(g), the lath-shaped austenite growth kinetics near the γ/α′ interface can be divided into 3 stages:12) In stage-1 with IA time did not exceed 0.001 s, the γ/α′ interface movement was controlled by rapid C diffusion in martensite, referred as negligible-partitioning local equilibrium (NPLE). In stage-2 with IA did not exceed 1000 s, the austenite content continued to increase as well as the slope of the curve (growth rate) increased. It can be seen from Fig. 3(e), C concentration almost reached equilibrium in austenite and martensite, respectively, while the Mn concentration of austenite kept increasing. This suggested that the growth rate increased with the increase of Mn concentration in γ, and the growth was gradually governed by Mn diffusion in martensite, referred to partitioning local equilibrium (PLE-1). In stage-3, with IA exceeding 1000 s, the variation of austenite volume fraction tended to be slow. Combined Mn concentration tended to be homogeneous with IA time from 1000 s to 1800 s in Fig. 3(e). This suggested that the growth was gradually governed by Mn diffusion in austenite (partitioning local equilibrium, PLE-2). Unlike martensite, pre-ferrite was precipitated during the rolling process. As was well known, the C/Mn concentration in αpre was low, like the result in Fig. 3(c). This led to almost no NPLE stage in Fig. 3(g). While the PLE mode controlled by Mn diffusion was limited by the low Mn concentration in pre-ferrite and it posed challenges for the rapid movement of the γ/αpre interface. Thus, the size of lath-shaped austenite was larger than that of block austenite and this was the reason for the bimodal characteristic in the distribution of austenite size. Moreover, larger size induced dilution of the average Mn concentration of austenite. Consequently, the chemical and mechanical stability of lath-shaped austenite was lower than that of block austenite.
The total volume fraction of austenite calculated was roughly equivalent to the one obtained by XRD. As shown in Fig. 3(g), the error may come from some information that had not been taken into account, such as lattice defects in martensite, carbide formation, and dissolution, grain size variation of constituents, etc. Anyway, the simulation results displayed well that the pre-ferrite in the martensite matrix induced the formation of austenite with different stability. In addition, the above simulation was just an example of “average size”. In fact, there were many sizes of pre-set austenite and lath martensite. Therefore, the initial microstructure with “αpre +α′” caused a large stability gradient of austenite in the final microstructure.
3.3. Tensile BehaviorsThe engineering stress–strain curves and corresponding work-hardening rate curves are presented in Figs. 4(a) and 4(b), and the detailed mechanical properties were listed in the table in Fig. 4(a). The 928K sample showed yield strength (σ0.2) of 793 MPa, ultimate tensile strength (σb) of 1308 MPa and total elongation (δ) of 34%. As a comparison, the 978K showed a 54 MPa decrease in σ0.2 and a 4% decrease in δ. The reduction in σ0.2 in the 978K was probably due to grain coarsening of αpre with IA temperature. Compared to αlath and γlath, αpre with larger size and lower dislocation density was easily deformed at the early stage of the tensile behavior.4,10) The fracture surface subjected to the tensile test showed large and small dimples in both samples (Figs. 4(d) and 4(e)). This showed good ductility and consistent tensile results.
To study the tensile behavior, the work hardening index curves were obtained by derivation of the true stress-strain curve, as shown in Fig. 4(b). At the outset of deformation, the work-hardening index decreased rapidly before yielding. The reason was related to the dynamic dislocation recovery in ferrites.1,2) Then, the work hardening index increased from the lowest point. For conditional steels with full martensite as the initial microstructure, the curves reached a peak quickly and then decreased slowly. The curves were relatively smooth and had only one peak.2,3,4,5) However, the multi-peak work-hardening behavior was found in 928K and 978K samples with pre-ferrite. Correspondingly, remarkable serrated plastic flow was observed in Fig. 4(a), i.e. dynamic strain aging (DSA). According to classical DSA theories, DSA resulted from dynamic interactions of mobile dislocations with solute atoms, during tensile deformation. However, using in situ magnetic induction measurements combined with ex situ XRD analysis and DIC techniques, Sun et al.13) reported that mobile dislocations might be additionally pinned by the small martensite embryos, which was dominated by the stability of austenite. Moreover, Cai et al.14) proposed the serrations was related to discontinuous TRIP effect, which primarily resulted from austenite with different degree of stability. The appearance of peaks in the work-hardening index curves originated from the positive TRIP effect of a significant quantity of austenite with similar stability at a specific strain. Taking the 978K sample as an example, when the strain reached 0.08, a considerable austenite with similar stability transformed into martensite. Simultaneously, martensite pinned a large number of dislocations. Hence, the work hardening index increased dramatically to form a peak (point A) in the work hardening index curve as well as a corresponding steep occurred in the true stress–strain curve. With increasing the true strain (points A to B), dislocations break away from their carbon atmosphere or martensite. Since martensite is a hard phase compared to the surrounding austenite ferrite, martensite undergoes elastic deformation. The surrounding ferrite and austenite undergo coordinated deformation with dislocation slip. This resulted in a drop to point B. With further increasing strain (points B to C), the true strain reaches the next critical strain, and another batch of austenite with similar but higher than ones in A point stability will transform into martensite. Points C-D-E had the same mechanism as points A-B-C but required greater strain to activate. The multi-peak work-hardening behavior demonstrated that there was a large stability gradient in the retained austenite. This was consistent with previous microstructural observations and simulations. It was noteworthy that when the curve entered the F region, the specimen necked (n< nT). But when the strain increased, there was still a peak formed in the curve. Obviously, the soft ferrite in the sample was hardened by first-deformation and then cooperated with the TRIP effect. This suggested that the occurrence of the TRIP effect was accompanied by adaptive coordination of ferrite during deformation. Anyway, the pre-ferrite in the martensite matrix achieved a multi-peak work-hardening behavior by inducing the formation of austenite with a large stability gradient.
Figure 4(f) displays the volume fraction of retained austenite before deformation and after fracture in 928K and 928K samples. There was 51% retained austenite in the 928K sample, and after tensile deformation to fracture, 11% retained austenite remained. In contrast, the volume fraction of austenite in the 978K sample was reduced from 53% to 8%. Obviously, there was a larger TRIP effect in the 978K sample compared with the 928K sample. The higher peak value in the work hardening curve of the 978K sample was also illustrated that a larger TRIP effect in the 978K sample compared with the 928K sample. This was due to the lower mechanical stability resulting from the larger austenite grain size, which was evaluated in microstructure observation.
The correlation between the δ and σb can be represented via the so-called banana-curves. Figure 4(c) summarized the δ and σb of medium-Mn steels in this study and hot rolled MMnS in literatures1,2,3,4,5) with similar Mn/C levels. As can be seen, superior tensile properties would be obtained for the steels in this study, though their δ was relatively low. Anyway, it was suggested that pre-ferrite enhanced the TRIP effect in hot rolled MMnS and obtained a strength-ductility balance as high as 44.6 GPa%, which was much higher than the target of the 3rd-generation AHSS (30- 35 GPa%).
A detailed study was conducted about the effect of pre-ferrite on Mn/C partitioning behavior, austenite reversion kinetics, and deformation behaviors in MMnS. The main conclusions were as follows:
(1) There were two types of austenite induced via the initial microstructure with “pre-ferrite + martensite”. Lath-shaped austenite existed in the martensitic matrix; blocky austenite occurred at the boundaries of pre-ferrite. The size of austenite increased with the increase of the IA temperature.
(2) Based on the size effect and Mn/C concentration, the initial microstructure with pre-ferrite caused a large stability gradient of austenite in the final microstructure.
(3) An excellent combination of strength (1308 MPa) and ductility (34%) was obtained at 928 K, which was ascribed to a positive TRIP effect induced by the austenite with a large stability gradient.
I write on behalf of myself and all co-authors to confirm that the results reported in the manuscript are original and neither the entire work, nor any of its parts have been previously published. The authors confirm that the article has not been submitted to peer review, nor has been accepted for publishing in another journal. The authors confirm that the research in their work is original, and that all the data given in the article is real and authentic. If necessary, the article can be recalled, and errors corrected.
MMnS: Medium-Mn steels
IA: inter-critical annealing process
TRIP effect: transformation-induced plasticity effect
PSE: products of strength and elongation (GPa%)
αpre: pre-ferrite
α′: martensite
PAGB: prior austenite grain boundaries
γlath: lath-shaped austenite
γblock: blocky austenite
αlath: lath-shaped ferrite
σ0.2: yield strength (MPa)
σb: ultimate tensile strength (MPa)
δ: total elongation (%)
