Reverse Transformation Behavior in Multi-phased Medium Mn Martensitic Steel Analyzed by in-situ Neutron Diffraction

Abstract

The reverse transformation behavior during heating in Fe-10%Mn-0.1%C (mass%) martensitic alloy consisting of α’-martensite, ε-martensite and retained austenite was investigated using the in-situ neutron diffraction. When the temperature was elevated with a heating rate of 10 K/s, the ε→γ reverse transformation occurred first at the temperature range of 535–712 K, where Fe and Mn hardly diffused. In the temperature range where the ε→γ reverse transformation occurred, the full width at half maximum of the 200γ peak increased, indicating that the austenite reversed from ε-martensite contains high-density dislocations. In addition, the transformation temperature hardly depends on the heating rate and the crystal orientation of the reversed austenite was identical to that of the prior austenite (austenite memory), which suggests that the ε→γ reverse transformation would proceed through the displacive mechanism. After completion of the ε→γ transformation, the α’→γ reverse transformation occurred at the temperature range of 842–950 K. When the heating rate is low (<10 K/s), the reverse transformation start temperature significantly depends on the heating rate. It could be because the diffusional reverse transformation accompanying the repartitioning of Mn occurs. On the other hand, a higher heating rate (≥10 K/s) resulted in the disappearance of the heating rate dependence. This was probably due to the change in the transformation mechanism to the massive-type transformation, which is diffusional transformation without repartitioning of Mn.

1. Introduction

Fe-9%Ni-0.1%C alloys (mass%) with martensitic structures have been widely used as cryogenic steels because Ni-enriched austenite with high stability can be formed by intercritical annealing, and the alloy exhibits an excellent strength-toughness balance in cryogenic environments. Recently, Ni is expected to be replaced by Mn, a low-cost alternative element, from a commercial perspective. For the Fe-10%Mn-0.1%C alloy that the author focused on in reference,¹⁾ the A_c1 temperature is close to that of the Fe-9%Ni-0.1%C alloy, and thus, austenite with high stability is formed by intercritical annealing, similar to the Fe-9%Ni-0.1%C alloy. In addition, we confirmed that excellent toughness was obtained by intercritical annealing after an appropriate thermomechanical process.²⁾ However, the martensitic structures obtained by quenching are quite different for the two alloys. The as-quenched Fe-9%Ni-0.1%C alloy possesses a lath martensitic single structure, whereas the microstructure of the as-quenched Fe-10%Mn-0.1%C alloy consists of multiple phases of fine α’-martensite, ε-martensite, and retained austenite. This is because the addition of Mn results in a decrease in the stacking fault energy of austenite, and the γ→ε→α’ two-step martensitic transformation is promoted. Because of the mixed martensitic structure consisting of ε-martensite and α’-martensite phases, the medium-Mn steel has two paths of different reverse transformations to austenite from each martensite during heating, leading to a complicated nucleation and growth behavior of austenite, which is different from that of the Fe-9%Ni-0.1%C alloy. To improve the mechanical properties of medium-Mn martensitic steel, it is important to elucidate the reverse transformation mechanism for controlling the microstructure of steel. In this study, the reverse transformation behavior in the multi-phased Fe-10%Mn-0.1%C alloy was investigated using the in-situ neutron diffraction method, which makes the analysis of multiple phases (α’-martensite, ε-martensite and retained austenite) possible.

2. Experimental Procedure

A 50 kg ingot of Fe-10%Mn-0.1%C alloy was prepared by vacuum melting. The chemical compositions of the alloys are listed in Table 1. The ingot was hot-rolled and austenitized at 1473 K for 1.8 ks, followed by water quenching to obtain the initial martensitic structure (as-quenched specimen). Microstructural observations were performed by electron backscatter diffraction (EBSD) analysis using a field-emission scanning electron microscope (SIGMA500, ZEISS, Germany) operated at an accelerating voltage of 20 kV. The reverse transformation behavior during heating was examined by in-situ neutron diffraction using a time-of-flight neutron diffractometer (iMATERIA) at the Japan Proton Accelerator Research Complex (J-PARC).³⁾ The as-quenched specimen (70 mm×10 mm×2 mm) was heated at a rate of 10 K/s from ambient temperature to 1073 K during irradiation with a neutron beam (beam size: 20 mm×20 mm, beam power: 529 kW), and the diffraction line profiles were obtained from the backscatter bank (145°<2θ<175°). The diffraction line profiles during heating were divided every 1 s. In addition, the neutron diffraction line profile of the as-quenched specimen was measured at ambient temperature for 600 s, and the phase fraction was calculated via Rietveld Texture Analysis (RTA) using MAUD software.⁴⁾ Dilatometry was performed using a Transmaster II instrument (Advance Riko, Japan) for specimen sizes of φ3 mm×10 mm, and the heating rate was changed between 0.5 and 50 K/s.

Table 1. Chemical composition (mass%) of Fe-10%Mn-0.1%C alloy.

C	Si	Mn	P	S	Al	Fe
0.097	0.019	10.03	<0.001	0.0029	0.012	bal.

3. Results

3.1. Microstructure of As-quenched Specimen

Figure 1(a) shows a crystallographic orientation map of the as-quenched Fe-10%Mn-0.1%C alloy. Because the microstructure consists of multiple phases, that is, α’-martensite (bcc), ε-martensite (hcp), and retained austenite (fcc), its morphology is finer and more complex than that of the lath martensite structure in low-C steels. Figure 1(b) shows a reconstruction map of prior austenite grains (PAGs) before quenching. It was calculated from the crystal orientations of α’-martensite shown in Fig. 1(a) using austenite reconstructing method based on the Kurdjumov-Sachs orientation relationship (K-S OR) ({111}_fcc//{011}_bcc, <1-10>_fcc//<1-11>_bcc).⁵⁾ The PAGs size of the as-quenched specimen is estimated to be approximately 140 μm. Figures 1(c), 1(d), 1(e) show the high-magnified crystallographic orientation maps of the as-quenched specimen ((c) all phases, (d) hcp and (e) fcc). The α’-martensite grains were very fine, and their average grain size was approximately 0.46 μm. In addition, the α’-martensite and ε-martensite shapes are not plate-like, but granular with a small aspect ratio. It was reported that such a unique microstructure was formed via a two-step γ→ε→α’ martensitic transformation during quenching; plate-like ε-martensite was initially formed in austenite grains, and then fine α’-martensite nucleated inside the ε-martensite.⁶⁾ Figure 2 shows the neutron line profile of the as-quenched specimen and the diffraction peak position of each phase, calculated using RTA. The diffraction peaks of α’-martensite, ε-martensite, and retained austenite were confirmed (lattice constants: a_α’=0.2877 nm, a_γ=0.3588 nm, a_ε=0.2528 nm, and c_ε=0.4094 nm), and many peaks overlapped with each other. RTA revealed that the volume fractions of α’-martensite, ε-martensite, and retained austenite in the as-quenched specimen were 73, 18, and 9 vol.%, respectively.

Fig. 1. Orientation maps of Fe-10%Mn-0.1%C alloy before heating: (a) low magnification, (b) reconstructed austenite map of (a), (c)–(e) high magnification ((c) all phases (d) hcp (e) fcc). (Online version in color.)

Fig. 2. Neutron diffraction line profile of Fe-10%Mn-0.1%C alloy before heating.

3.2. Reverse Transformation Behavior during Heating

Figure 3(a) shows the change in the neutron line profiles every 100 K during heating at a heating rate of 10 K/s from ambient temperature. In addition, Fig. 3(b) shows the changes in the integrated intensity of the 200α’, 200γ and 1011ε peaks with increasing temperature. These peaks did not overlap with the other peaks. For the 1011_ε peak, its integrated intensity decreased from 535 K, and the peak completely disappeared at 712 K. Furthermore, the integrated intensity of the 200γ peak increased as that of the 1011_ε peak decreased, and then stopped increasing when the 1011_ε peak disappeared. This indicates that the ε→γ reverse transformation occurred in this temperature range. For the ε→γ transformation, the transformation start temperature (A^ε_s) and transformation finish temperature (A^ε_f) were determined as 535 and 712 K, respectively. It should be noted that the Fe and substitutional solute atoms hardly diffused in this temperature range. In addition, the integrated intensity of the 200α’ peak hardly changed during the ε→γ transformation, which means that α’-martensite is unrelated to the ε→γ transformation. In the temperature range between 842 and 950 K, the 200α’ peak decreases and the 200γ peak increases at the same time with increasing temperature, and thus the α’→γ reverse transformation can be determined to have occurred in this temperature range; here, A^α’_s and A^α’_f are 842 and 950 K, respectively. Figure 3(c) shows the changes in the full width at half maximum (FWHM) of the 200γ peak during heating. The FWHM at ambient temperature is relatively large because the austenite peak is expected to broaden owing to the residual stress and dislocations introduced by the shape change accompanying the martensitic transformation. The FWHM monotonously decreased with increasing temperature below A^ε_s (<535 K), which was due to a decrease in the residual stress. However, the FWHM increased remarkably in the temperature range of the ε→γ transformation (535–712 K). Because it is difficult to consider that the dislocation density of preexisting retained austenite increases significantly owing to the small volume expansion during the ε→γ transformation, it is reasonable to assume that the newly formed reversed austenite contains high-density dislocations. After the ε→γ transformation is complete, the FWHM monotonically decreases with increasing temperature. The α’→γ transformation starts above 842 K; however, the FWHM does not increase, unlike the case of the ε→γ transformation, but continuously decreases.

Fig. 3. Changes in (a) neutron diffraction line profile, (b) integrated intensity of 200α’, 200γ and 1011_ε and (c) full width at half maximum of 200γ during heating. (Online version in color.)

4. Discussion

The mechanisms of reverse transformation are classified into diffusional transformation with atomic diffusion and displacive transformation accompanying martensitic structural change, almost without atomic diffusion. Furthermore, there are two types of diffusional transformations: with and without the repartitioning of solute elements between the matrix and reversed phases. The latter is a massive transformation that proceeds with short-range diffusion near the phase boundary.⁷⁾ To identify these reverse transformation mechanisms, it is necessary to discuss the repartitioning behavior of alloying elements, the heating rate dependence of the transformation temperature, and the inheritance of the crystal orientation of prior austenite (austenite memory).^8,9,10) In addition, the lattice defects introduced during transformation also provide important evidence for distinguishing whether the mechanism is diffusional or displacive; thus, in-situ neutron diffraction, which can measure the dislocation density, is an effective method to identify the reverse transformation mechanisms. Based on the results of in-situ neutron diffraction analysis, the mechanisms of ε→γ and α’→γ transformation in the Fe-10%Mn-0.1%C alloy were investigated as follows.

4.1. Reverse Transformation from ε to γ

The ε→γ transformation occurred at a relatively low temperature range, where the diffusion of Fe and Mn hardly occurred, and austenite with high-density dislocations was formed by the transformation, as suggested in Fig. 3(c). This indicates that the ε→γ transformation can be classified as a displacive transformation.¹¹⁾ The high-density dislocation of austenite could be generated by the plastic deformation of the soft austenite phase to accommodate the lattice strain induced by the displacive reverse transformation. In addition, it has been reported that a displacive reverse transformation has little heating rate dependence for the transformation temperature,¹²⁾ and that the crystal orientation of reversed austenite is the same as that of prior austenite (austenite memory).¹³⁾ Figure 4 shows the dilatometry curves indicating the dependence of the transformation temperature on the heating rate. Because it is confirmed that the ε→γ and α’→γ transformations occur independently, as described above, the reverse transformation temperature can be estimated from the dilatometry curves. Dilatometry curve (I) is the result of the as-quenched specimen measured at a heating rate of 10 K/s. It is known that volume expansion and contraction occur during the ε→γ and α’→γ transformations, respectively.^14,15) Therefore, the temperatures at which the volume expansion and contraction begin correspond to the A^ε_s (≒520 K) and A^α’_s (≒830 K), respectively. The transformation start temperatures obtained by dilatometry were almost the same as those obtained by neutron diffraction (Fig. 3(b)). The changes in the transformation temperature measured by dilatometry are shown as a function of the heating rate (0.5–50 K/s) in Fig. 4(b). The T₀ temperatures calculated using Thermo-Calc. (Database: SSOL7) are also shown (dashed lines). In the ε→γ transformation, there is little heating rate dependence of A^ε_s, and the temperature is approximately 80 K higher than T₀ at any heating rate. The difference between T₀ and A^ε_s corresponded to a Gibbs free energy change of 315 J/mol. In addition to A^ε_s, there was no heating rate dependence on A^ε_f. Therefore, the ε→γ transformation mechanism itself does not depend on the heating rate.

Fig. 4. (a) Dilatometric curves of Fe-10%Mn-0.1%C alloy heated to (I) 1073 K and (II) 723 K at a heating rate of 10 K/s and (b) heating rate dependence of transformation start/finish temperature. (Online version in color.)

Figures 5(a), 5(b) show the crystallographic orientation maps of the specimen heated to 723 K (just above A^ε_f) at a heating rate of 10 K/s ((a) all phases and (b) fcc). The sample was held at 723 K for 5 s after heating and then quenched with N₂ gas. Figures 5(c), 5(d) represent high-magnified images of square region in Fig. 5(a) ((c) all phases and (d) fcc). It was confirmed that the ε→γ transformation yielded a larger amount of austenite compared to the as-quenched specimen. Austenite with an identical crystal orientation was collectively formed, and the orientation was the same as that of the prior austenite. This is a typical phenomenon that suggests an austenite memory. In addition, the dilatometry curve (II) in Fig. 4(a) shows that volume expansion due to the ε→γ transformation is observed during heating, but only monotonic volume contraction occurs during cooling, which demonstrates that the γ→ε and γ→α’ transformations hardly occur during cooling from temperatures below A^α’_s, owing to the stabilization of austenite by C repartitioning from α’-martensite to austenite during heating and cooling. From the above results, it is concluded that the ε→γ transformation occurs through the displacive transformation mechanism, as predicted from the results of neutron diffraction.

Fig. 5. Orientation maps of Fe-10%Mn-0.1%C alloy after heating to various temperatures: (a)–(d) heated to 723 K ((a)(b) low magnification ((a) all phases (b) fcc), (c)(d) high magnification ((c) all phases (d) fcc), (e)–(h) heated to 1003 K ((e)(f) low magnification ((e) all phases (f) fcc), (g) austenite reconstructed map of (e), (h) enlarged view of area enclosed by black frame in (g)). (Online version in color.)

4.2. Reverse Transformation from α’ to γ

The reversed austenite formed from α’-martensite does not contain high-density dislocations, as shown in Fig. 3(c), suggesting that the α’→γ transformation is expected to proceed through the diffusional transformation mechanism. As shown in Fig. 4(b), the α’→γ transformation temperature depends on the heating rate when it is low (< 10 K/s). Han et al. investigated the heating rate dependence of the α’→γ transformation temperature in Fe–(5–9)%Mn-0.05%C alloy with an α’-martensite single phase, and reported that the transformation temperature depended on the heating rate at a low heating rate (< 15 K/s).⁸⁾ In addition, they found that the reverse transformation proceeded with Mn diffusion and Mn-enriched austenite was formed during heating. Assuming that the same behavior occurs in this study, the heating rate dependence in the α’→γ transformation at a low heating rate should be caused by Mn repartitioning that rate-limits the α’→γ transformation. On the other hand, the α’→γ transformation temperature exhibited no heating rate dependence above 10 K/s. Han et al. also reported that Mn repartitioning did not occur, and the transformation temperature did not depend on the heating rate when it was high (≧ 15 K/s) in the Fe-9%Mn-0.05%C alloy.⁸⁾ Therefore, they concluded that a displacive transformation occurred. However, Yang et al. indicated that it was a massive-type transformation because pre-deformation promoted the transformation, and the difference in Gibbs free energy between α’-martensite and austenite was too small to cause a displacive transformation.¹⁰⁾ In addition, considering that the dislocation density of reversed austenite formed by the α’→γ transformation is not large in this study (heating rate:10 K/s) (Fig. 3(c)), the transformation mechanism is expected to be diffusional transformation without Mn repartitioning, that is, a massive-type transformation.

Figures 5(e), 5(f) show the crystallographic orientation maps of the specimen heated to 1003 K (just above A^α’_f) at a heating rate of 10 K/s ((e) all phases and (f) fcc). The specimen was quenched with N₂ gas immediately after the heating. The amount of retained austenite is less than that in Figs. 5(c) and 5(d) because C is homogenized in the fully reversed austenitic structure and the austenite stability is lowered. Figure 5(g) shows the reconstruction map of the PAGs for Fig. 5(e), corresponding to the austenitic structure at 1003 K. PAGs with the same size as that of the as-quenched specimen (Figs. 1(a), 1(b)) can be mainly observed, which means that the austenite memory appeared through the reverse transformation. In the case of the massive-type α′→γ transformation, it is not necessary to satisfy the K-S OR.^16,17) However, since there are already austenite with the original austenite orientation, that is retained austenite and displacive reversed austenite formed from ε-martensite, the α’→γ reverse transformation will proceed with the growth of these austenites, regardless of whether the reverse transformation mechanism is a massive-type transformation. As a result, an austenite memory inevitably develops. On the other hand, it was found that fine austenite grains were dispersed within the PAGs, for example, in the square region in Fig. 5(g). Figure 5(h) shows an enlarged map of this area. The size of the fine austenite grains was approximately 10–20 μm. Each fine austenite grain (γ₁, γ₂, and γ₃) has a different crystal orientation from that of the surrounding austenite (γ₀), and they seem to be formed by another mechanism. Such austenite grains might be nucleated with the massive-type mechanism above A^α’_s without the appearance of austenite memory, or formed through recrystallization of austenite with high-density dislocation introduced via a displacive ε→γ transformation.

5. Conclusions

The reverse transformation behavior in the multi-phased martensitic Fe-10%Mn-0.1%C alloy during heating is summarized below.

Fine α’-martensite, ε-martensite, and retained austenite were present in the as-quenched specimen. When the temperature is elevated, the ε→γ reverse transformation occurs first with a displacive mechanism, leading to the introduction of high-density dislocations into the reversed austenite. After completion of the ε→γ transformation, the α’→γ transformation occurs at a higher temperature. When the heating rate was less than 10 K/s, it was considered that a diffusional reverse transformation mechanism proceeded with the repartitioning of Mn based on the evidence that the reverse transformation start temperature significantly depended on the heating rate; however, a faster heating rate resulted in the disappearance of the heating rate dependence. This is probably due to the change in the transformation mechanism from diffusional reverse transformation with repartitioning Mn to massive-type reverse transformation without repartitioning.

Acknowledgements

Neutron experiments at the Materials and Life Science Experimental Facility of the J-PARC were performed under a user program (proposal No. 2021BM0008).

References

• 1)   T.  Maeda,  S.  Okuhata,  K.  Matsuda,  T.  Masumura,  T.  Tsuchiyama,  H.  Shirahata,  Y.  Kawamoto,  M.  Fujioka and  R.  Uemori: Mater. Sci. Eng. A., 812 (2021), 141058. https://doi.org/10.1016/j.msea.2021.141058
• 2)   K.  Matsuda,  T.  Maeda,  T.  Masumura,  T.  Tsuchiyama,  Y.  Kawamoto,  H.  Shirahata,  M.  Fujioka and  R.  Uemori: Extend Abstract 5th International High Manganese Steel Conference (HMnS2022), (Linz), (2022), 219, E-book.
• 3)   Y.  Onuki,  T.  Hirano,  A.  Hoshikawa,  S.  Sato and  T.  Tomida: Metall. Mater. Trans. A, 50A (2019), 4977. https://link.springer.com/article/10.1007/s11661-019-05415-6
• 4)   A.  Saville,  A.  Creuziger,  E. B.  Mitchell,  S. C.  Vogel,  J. T.  Benzing,  J.  Klemm-Toole,  K. D.  Clarke and  A. J.  Clarke: Integr. Mater. Manuf. Innov., 19 (2021), 461. https://doi.org/10.18434/mds2-2400
• 5)   G.  Miyamoto,  N.  Iwata,  N.  Takayama and  T.  Furuhara: Acta Mater., 58 (2010), 6393. https://doi.org/10.1016/j.actamat.2010.08.001
• 6)   S.  Uranaka,  S.  Tanaka,  T.  Masumura,  T.  Tsuchiyama,  R.  Uemori and  H.  Shirahata: J. Jpn Soc. Heat Treat., 59 (2019), 271 (in Japanese).
• 7)   M.  Enomoto: Materia Japan., 54 (2015), 65 (in Japanese). https://doi.org/10.2320/materia.54.65
• 8)   J.  Han and  Y. K.  Lee: Acta Mater., 67 (2014), 354. https://doi.org/10.1016/j.actamat.2013.12.038
• 9)   H.  Shirazi,  G.  Miyamoto,  S.  Hossein Nedjad,  T.  Chiba,  M.  Nili Ahmadabadi and  T.  Furuhara: Acta Mater., 144 (2018), 269. https://doi.org/10.1016/j.actamat.2017.10.068
• 10)   D. P.  Yang,  D.  Wu and  H. L.  Yi: Scr. Mater., 174 (2020), 11. https://doi.org/10.1016/j.scriptamat.2019.07.040
• 11)   N.  Nakada,  R.  Fukagawa,  T.  Tsuchiyama,  S.  Takaki,  D.  Ponge and  D.  Raabe: ISIJ Int., 53 (2013), 1286. https://doi.org/10.2355/isijinternational.53.1286
• 12)   G.-S.  Sun,  J.  Hu,  B.  Zhang and  L.-X.  Du: Mater. Sci. Eng. A, 732 (2018), 350. https://doi.org/10.1016/j.msea.2018.07.024
• 13)   S.  Matsuda and  Y.  Okamura: Trans. ISIJ., 14 (1974), 363. https://doi.org/10.2355/isijinternational1966.14.363
• 14)   H.  Schumann: Eisenhutt., 38 (1967), 647 (in German). https://doi.org/10.1002/srin.196704234
• 15)   H.  Schumann: Eisenhutt., 40 (1969), 1027 (in German). https://doi.org/10.1002/srin.196904421
• 16)   T. B.  Massalski: Metall. Mater. Trans. A, 33 (2002), 2277. https://doi.org/10.1007/s11661-002-0351-7
• 17)   A.  Borgenstam and  M.  Hillert: Acta Mater., 48 (2000), 2765. https://doi.org/10.1016/S1359-6454(00)00102-6