Ferrite Transformation from Fe-0.3N Austenite

Abstract

Ferrite transformation behavior of Fe-0.3 mass%N binary alloy was studied in the temperature range between 500°C and 750°C. By isothermal holding in the α + γ two phase region, two morphologies of Allotoriomorphic ferrite (AF) and Widmanstetten ferrite (WF) formed from the prior γ grain boundary. On the other hand, in the case of isothermal holding at 500°C, nitride-free bainitic ferrite (BF) was formed at the beginning, and changed to bainite accompanied with γ’ precipitation in further holding. The retained γ was obtained by decreasing the transformation temperature in the two phase region, and maximum volume fraction of 9% was obtained at 600°C. AF had near K-S OR with one side of the adjacent prior γ and grew into the grain without K-S OR. On the other hand, WF and BF had near K-S relationship with the matrix. Both WF and BF had significantly higher energy dissipation than AF, and the energy dissipation of AF is due to interfacial friction. On the other hand, the strain energy associated with the transformation was dominant in WF.

1. Introduction

Nitrogen is an interstitial element and thermodynamically stabilizes the austenite (γ) phase. It is known that the behavior of nitrogen in iron shows similar tendency to carbon with respect to the concentration dependence of the change in the A_e3 or Ms temperature and the lattice constant.^1,2,3,4) On the other hand, due to the difference in thermodynamic stability between iron nitride and iron carbide in γ, the eutectoid temperature and concentration, A_e1, are different between Fe–C and Fe–N binary systems; 727°C and 0.77 mass%C, for Fe–C and 590°C and 2.46 mass%N for Fe–N. The A_e1 point in Fe–N is lower in temperature and higher in concentration than Fe–C.

Since such characteristics of nitrogen steel can be applied to a new kind of TRIP steel using retained γ stabilized by N enrichment. We recently investigated the reverse transformation behavior of intercritically annealed Fe-0.3N binary alloy at the ferrite (α) + γ two-phase region, and demonstrated that about 10% of retained γ can be obtained.⁵⁾

In the cold-rolled TRIP steel, the microstructure control was performed by the two step of heat treatment of intercritical annealing in the α + γ two phase region and austempering treatment in the temperature range where bainite transformation occurs afterwards, and high strength-high ductility is achieved. Therefore, control of ferrite transformation is also indispensable because the size and volume fraction of pro-eutectoid α greatly affect the strength and ductility of the material.

Proeutectoid α forms from γ in hypoeutectoid steel is most frequently observed microstructure in steel, and its morphology changes with the degree of supercooling from A_e3 line. Allotriomorphic ferrite (AF) is formed at austenite grain boundary when the degree of undercooling from Ae3 line is small at high temperatures whereas needle-like or plate-like Widmanstetten ferrite (WF) nucleates at low temperatures and with a high degree of supercooling.^6,7,8) Regarding the crystallographic characteristics of AF, the orientation of proeutectoid α formed from the γ grain boundary during the cooling almost satisfies the K-S orientation relationship (OR) with one side of adjacent prior γ grain, but it has no specific OR with the other γ grain, and shows preferential grows into the γ grain without near K-S OR.^9,10) On the other hand, WF formed from γ grain boundaries has an orientation of near K-S OR with prior γ grain, so far.^9,11) Although there are many studies on the ferrite transformation behavior of the Fe–C alloy, that of Fe–N alloy has not been clarified yet, except in an ultra-low nitrogen alloy¹²⁾ or in a hypereutectoid alloy.¹³⁾ Therefore, in this study, to clarify the ferrite transformation behavior of Fe–N alloy, the morphology and transformation kinetics of proeutectoid α, and the partitioning behaviors of nitrogen at α/γ interface were investigated.

2. Experimental Procedure

Commercial pure iron was used as a starting material, of which chemical composition and the calculated A_e1 and A_e3 temperatures by ThermoCalc using the TCFE9 database are shown in Table 1. The sample was cut into the dimension of ^l18 × ^w8 × ^t0.7 mm and then mechanically polished to the thickness of 0.5 mm. The polished samples were nitrided at 1000°C for 1 hour in H₂/NH₃ mixture atmosphere using a gas nitriding technique.¹⁴⁾ The total pressure during nitriding was fixed to 1 atm and the partial pressure of NH₃ gas was 0.2 atm. After the nitriding, sample was dropped into a salt bath heated to various temperatures in the range from 500°C to 750°C and isothermally annealed for periods up to 3600 s, followed by quenching into iced water. After these heat treatments, the samples were mechanically polished with emery papers, diamond sprayed buff and colloidal silica. Microstructures were characterized using field emission scanning electron microscopy (FE-SEM) after etching using a 2% Nital solution. The volume fraction of ferrite was determined by a point counting method using SEM micrographs. The volume fractions of γ’-Fe₄N and retained γ were measured by X-ray diffraction (XRD) using Co target with 35 kV and 40 mA. During XRD measurement, the texture effect was removed by the in-plane rotation and the inclination of sample.

Table 1. Chemical composition, Ae1 and Ae3 temperatures of the initial material (mass%).

	C	Si	Mn	P	S	Cr	Ae1	Ae3
Pure Iron	0.011	0.02	0.02	0.001	0.003	0.01	590°C	827°C (0.3mass%N)

The N concentration at α/γ interface was measured by FE-EPMA with an acceleration voltage and current of 4 kV and 10 nA, respectively. Since nitrogen is a light element, the nitrogen concentration was determined using a calibration curve obtained from six standard specimens (Pure Fe, Fe-0.197N, Fe-0.307N, Fe-0.98N, Fe-2.0N (mass%) and γ’-Fe₄N), the procedure of which is described in detail in Ref. 15. The α orientation map was acquired using EBSD (TSL OIM Analysis 7) equipped to field-emission SEM with an acceleration voltage and step size of 15 kV and 50 nm, respectively. The orientation of prior γ was calculated based on orientations of martensite using the reconstruction method previously reported.^16,17)

3. Results

Figure 1 shows the OM images of Fe-0.3N alloy isothermally treated at 500°C–750°C for various period of times. At the isothermal holding at 750°C, AF is formed at γ grain boundary at 5 s holding. Transformation rapidly progresses at 60 s, and the γ grain boundaries are covered by AF thickened. It is thought that film-like α morphology is due to the coalescence of α grains having the same crystal orientation nucleated at the γ boundary. Besides, it is seen that secondary WF nucleates at the AF grows into the γ grain. With increasing the holding time, both of AF and WF grew together. With decreasing the transformation temperature, WF becomes dominant. At 600°C, AF and WF are formed along the γ grain boundary, and Martensite-Austenite constituent (MA) exists between adjacent WFs as shown in Figs. 1(c) and 1(d). At the 5 s holding, the prior γ grain was wholly covered by AFs and WFs, and there was no significant change in the thickness of WF with increasing the holding time. On the other hand, the AF grew to increase in thickness by coalescence with adjacent AF. It is considered that the α transformation at 600°C progressed mostly by the lengthening of WF, and no iron nitride was observed between the adjacent WFs even at 3600 s holding. As shown in Figs. 1(e) and 1(f), bainitic ferrite (BF) and degenerate pearlite (DP) were formed from the prior γ grain boundaries at 500°C. In addition, MA exists between the adjacent BFs. DP has a ferrite structure in which the cementite phase is dispersed in a hypoeutectoid Fe–C alloy,¹⁸⁾ but in the case of nitrogen steel, it is considered that γ’-Fe₄N is formed instead of cementite. With increasing the holding time, the MA became thinner and divided into finer pieces with the growth of BF. On the other hand, the thickness of DP and proeutectoid α did not change compared to the early stage of holding.

Fig. 1.

Microstructures of isothermally treated samples. (a) 5 s and (b) 3600 s at 750°C, (c) 5 s and (d) 3600 s at 600°C, (e) 5 s and (f) 3600 s at 500°C, respectively. γ(M) denotes martensite transformed from untransformed austenite.

Figure 2 shows SEM images of the sample isothermally treated at 500°C. In the OM image shown in Figs. 1(e) and 1(f), the BF grew to the inside of the grains at the holding time of 5 s, and the bainite transformation seemed completed, but it can be seen that MA exists between the adjacent BFs and transformation was not completed in Fig. 2(a). In addition, it was observed that some untransformed γ decomposed into α and film-like γ’ at 3600 s holding, as indicated by the arrow in Fig. 2(b). Such a change is similar to the case of bainite transformation where MA turns to BFs+ cementite (θ) in carbon-contained steels. As described above, it was clarified that bainite transformation accompanied by precipitation of γ’ occurred when the Fe-0.3N alloy was isothermally treated at 500°C.

Fig. 2.

SEM images of isothermally treated sample at 500°C for (a) 5 s and (b) 3600 s, respectively.

Figure 3 shows the XRD patterns of the isothermally treated samples at various temperatures for 3600 s. In the 750°C holding (Fig. 3(a)), only diffraction peaks derived from α were confirmed, and the diffraction peaks derived from γ and iron nitrides such as α”-Fe₁₆N₂ and γ’ were not confirmed. In addition, there is a shoulder in the lower angle side of 110α peak indicating the existence of the martensite phase transformed by quenching of low nitrogen concentration γ. In the case of 600°C (Fig. 3(b)), the diffraction peaks derived from α and a shoulder at the low angle side of 110α peak were also observed as in the case of the 750°C holding. Also, diffraction peaks derived from retained γ were confirmed at 2θ = 59.1°, 88.2°, 109.7°, and 117.1°, respectively. In the case of the 500°C holding, the diffraction peaks of γ and γ’ were observed at the early stage, and the peak intensity of γ’ increased and the peak intensity of γ decreased with increasing holding time. At 3600 s holding, only diffraction peaks of α, γ and γ’were observed (Fig. 3(c)).

Fig. 3.

XRD patterns of isothermally treated samples for 3600 s at (a) 750°C, (b) 650°C and (c) 500°C, respectively.

Figure 4 shows the transformation kinetics curves of the isothermally treated samples at various temperatures. In the 750°C sample (Fig. 4(a)), the α volume fraction in the early stage was small, and reached to only about 10% after the holding for 10 s. Then the transformation proceeded more rapidly and an equilibrium volume fraction of 65% was obtained at 3600 s. At 600°C, as seen in Fig. 4(b), the α transformation became faster due to the increasing driving force and the α volume fraction after 10 s where the kinetics got slowdown was much higher than that of 750°C. After 10 s, the α volume fraction still gradually increases up to 3600 s. From the XRD measurement, it is shown that the amount of retained γ increased with increasing the holding time due to nitrogen enrichment between WFs, and 9% of retained γ was obtained at 600 s holding and remained the same during further holding. In the case of 500°C, about 85% of transformation progressed at 5 s, and 6% of retained γ was obtained. The transformation rate gradually increased because of the decomposition of untransformed γ to α + γ’ with increasing the holding time. The transformation was completed at 3600 s holding because the volume fraction of γ’ reached the equilibrium amount of 5%.

Fig. 4.

Transformation kinetics curve at (a) 750°C, (b) 600°C, and (c) 500°C, respectively. (Online version in color.)

Figures 5(a) and 5(b) show IPF maps at the α/γ interface in the specimens transformed at 750°C and 600°C, respectively. The angle in the figure is the deviation angle from the exact K-S OR with the prior γ grain. As shown in Fig. 5(a), the AF formed at γ grain boundary holds near K-S orientation relationship , of which deviation angle is less than 5°, with one side of the adjacent prior γ grains while the deviation angle was more than 10° with the opposite γ grain. AF tends to grow into the non K-S OR side presumably due to an incoherent nature of α/γ interface on this side. On the other hand, as shown in Fig. 5(b), both of the WF formed at the 600°C (and also the BF formed at the 500°C) held orientation relationships, of which deviation angles of less than 5° from the K-S OR, with the γ matrix, resulting both the WF/γ and BF/γ interfaces have high coherency.

Fig. 5.

Bcc IPF maps of isothermally treated samples held for (a) 10 s at 750°C and (b) 5 s at 600°C, respectively. γ(M) denotes martensite transformed from untransformed austenite. (Online version in color.)

Figure 6 shows the results of variant distribution analysis using EBSD analysis for sample isothermally treated at 500°C for 3600s. BF hold a specific orientation relationship with respect to the parent austenite, and 24 different variants can be formed within a single grain in the K-S OR. These 24 variants are classified into 4 variant groups which consists of 6 variants sharing the parallel relation of close-packed plane for same (111)_γ (CP group), and they are also grouped according to the 3 distinctive variants of the Bain correspondence (Bain group).²⁰⁾ In this figure, same color denotes the same Bain or CP group, respectively. In addition, white and black line in the figure represents the low angle (5° <θ <15°) and high angle (15° <θ) grain boundary, respectively. As shown in the Bain map in Fig. 6(a), many low angle grain boundaries are observed in the same color region, indicating that the variants of the same Bain group are greatly developed. On the other hand, in the CP map, there are few low angle grain boundaries in the same color region. This result indicates that Fe-0.3N is formed in the same Bain group rather than the same CP group. It has been reported in the isothermally treated Fe-2Mn-(0.05–0.75)C alloy at 500°C that variants with same Bain group develops collectively after bainitic ferrite having a specific orientation is preferentially formed at grain boundaries.¹⁹⁾ In addition, it has been also reported that the driving force for transformation is small in the bainite transformed at a relatively high temperature, and the variants having same Bain group is generated to minimize the energy at the interface.²⁰⁾ Furthermore, after one variant nucleates, a variant sharing lattice-invariant deformation is preferentially nucleated at the interface between the preexisting variant and the matrix.²⁰⁾ Therefore, it is considered that the development of the variant having same Bain group in this study might be due to the small misorientation at the variant boundary and the sharing of lattice-invariant deformation mode as in previous reports on Fe–C based alloys.

Fig. 6.

(a) Bain map and (b) CP map of isothermally treated sample at 500°C for 3600 s. White and black lines denote low angle (5° < θ < 15°) and high angle (15° < θ) grain boundary, respectively. (Online version in color.)

Figure 7 shows the α/γ interfacial nitrogen concentration profiles of the isothermally treated samples. The broken line in the figure denotes the equilibrium nitrogen concentration in γ calculated by Thermo-Calc. software, and the Δθ denotes the deviation angle from the exact K-S OR. At 750°C, the nitrogen concentration of the 10 s holding sample was lower at α/γ interface from the equilibrium value, and approached to a similar nitrogen concentration to bulk composition away from the interface. In addition, it was confirmed that the nitrogen concentration of γ at the interface tends to be lower on the non K-S side than the near K-S side. With increasing the holding time, the deviation from the equilibrium nitrogen concentration became smaller, and reached the equilibrium concentration of 0.75 mass%N with the decrease of the concentration gradient in γ. At 600°C, the interface nitrogen concentration was about 0.6 mass% in the early stage of the transformation, and it showed larger deviation from the equilibrium nitrogen concentration of 2.3 mass%N than that of the 750°C. Even after prolonged holding, such deviation remains although there is large scatters and some γ region reaches to the equilibrium with a nitrogen concentration.

Fig. 7.

Nitrogen concentration profile accross α/γ interface in isothermally treated samples. (a), (b) 750°C, (c), (d) 650°C, (e), (f) 600°C for 10 s and 3600 s.

Figure 8 summarizes the nitrogen concentration at α/γ interface plotted on the Fe–N binary phase diagram. The nitrogen concentration of the 500°C sample is a calculated data from the XRD measurement, because since both of MA and γ’ had film shape, it is difficult to distinguish MA and γ’ in the image. At 750°C and 650°C holding, the nitrogen concentration at the α/γ interface is lower nitrogen concentration side from the A_e3 line at lower transformation temperature or shorter holding time. It can be seen that the nitrogen concentration at the α/γ interface approaches the Ae3 line with increasing the holding time. In addition, the deviation from the Ae3 line is larger for the WF/γ interface than for the AF/γ interface from the comparison of 750°C and 600°C. At 500°C, the nitrogen concentration at BF/γ interface exceeded the T₀ line at 5 s holding, increased with increasing holding time, and stopped at the nitrogen concentration crossing the Aγ’ line (2.2 mass%N). This may be due to the precipitation of γ’ in utransformed γ at this temperature well below the A_e1 temperature.

Fig. 8.

Summary of the N concentration at α/γ interface. (Online version in color.)

4. Discussion

4.1. Energy Dissipation in Ferrite Transformation of Fe–N Alloy

The α/γ interface nitrogen concentration showed deviation from the equilibrium nitrogen content depending on the transformation temperature and the microstructure, and it is thought that such deviation of the interface nitrogen concentration results from various energy dissipations consuming the driving force of ferrite transformation.^21,22)

Figure 9 shows (a) a schematic diagram for calculating energy dissipation from the interfacial N concentration in Fe–N binary system and (b) the change in energy dissipation with respect to the holding time. Since the diffusivity of N is much faster than that of substitutional atoms at given temperature, the chemical potential of N will be practically constant over the interface region.   

$${\mu }_{\text{N}}^{\alpha }={\mu }_{\text{N}}^{\gamma }$$(1)

Fig. 9.

(a) Schematics of N concentration profile across migrating α/γ interface and corresponding energy dissipation estimated from interfacial N concentration in γ and (b) calculated energy dissipation at each condition. (Online version in color.)

Thus, the energy dissipation can be derived by the chemical potential difference of Fe in α and γ, as shown by the red arrow in Fig. 9(a)   

$$\mathrm{\Delta }{G}_{dis.}={\mu }_{Fe}^{\gamma }-{\mu }_{Fe}^{\alpha }$$(2)

At 750°C, the amount of energy dissipated in the early stage of transformation is about 70 J/mol and decreases with increasing the holding time as shown in Fig. 9(b). It can also be seen that the energy dissipation for the same holding time increases as the transformation temperature decreases for both of AF and WF. In addition, as a result of the calculation of energy dissipation amount for AF and WF, WF showed a larger amount of energy dissipation than AF. In general, the total energy dissipation (ΔG_dis) in Fe–C binary system is mainly composed of two terms:   

$$\mathrm{\Delta }{G}_{dis.}=\mathrm{\Delta }{G}_m+\mathrm{\Delta }{G}_{strain}$$(3)

where ΔG_m is the dissipation due to lattice friction that occurs when atoms move across the interface accompanying the change of crystal structure from fcc to bcc, and ΔG_strain is the dissipation due to the strain energy associated with transformation. The amount of ΔG_m can be derived by   

$$\mathrm{\Delta }{G}_m=v\cdot V_m/M$$(4)

where M is the intrinsic mobility of interface that is closely related to the interfacial coherency, V_m is the molar volume, v is the interface velocity, respectively. It is clear that the amount of ΔG_m is proportional to interface velocity and reciprocal of intrinsic mobility. Thus, the ΔG_m cannot be ignored when the v is large, e.g., in early transformation stage.

Figure 10 shows the measured half thickness of α at several temperature and holding time. At any temperature, the change in the α thickness follows the parabolic law, and the interface velocity v was calculated from the slope of this figure. Since there is no report on the intrinsic mobility M for Fe–N alloys, following equation reported by Hillert et al. for the AF in Fe–C alloys was used.²³⁾   

$$M=0.0058exp\left(-140\mathrm{\hspace{0.17em} }000/RT\right) \mathrm{\hspace{0.17em} }[\mathrm{m}\cdot \mathrm{m}\mathrm{o}\mathrm{l}/\mathrm{J}\cdot \mathrm{s}]$$(5)

Fig. 10.

Measured half thickness of ferrite at 750°C and 600°C.

The intrinsic mobility M calculated using the Eq. (3) was M = 4.12×10⁻⁹ m·mol/J·s for 750°C and 2.43×10⁻¹⁰ m·mol/J·s for 600°C, respectively. Although the calculated ΔG_m is lower than ΔG_dis, the ratio of ΔG_m to the amount of total energy dissipation is large, and especially it is relatively close to ΔG_dis at the non-K-S interface as shown in Table 2. Therefore, the factor of energy dissipation in the 750°C can be roughly explained by the energy dissipation due to interfacial friction. The strain energy due to volume change may be expected with migration of the incoherent interface. But, at this temperature, self-diffusion of iron may be also extensive so that there is small accumulation of volumetric strain energy. On the other hand, as shown in Table 3, at 600°C, the ratio of ΔG_m to the total amount of energy dissipated in 5 s holding is small, and the ratio becomes even smaller with increasing holding time. Therefore, it can be understood that the factor of energy dissipation at 600°C is mainly caused by ΔG_strain, due to contribution from volumetric and shear component in transformation strain and less plastic accommodation by diffusion and dislocation slip. In the case of 500°C, it showed the total larger dissipation due to high strain energy,^24,25,26) and the precipitation of γ’ as shown in Fig. 4(c) also causes the decrease in the nitrogen concentration at α/γ interface.

Table 2. Comparison of calculated ΔG_m and ΔG_dis at 750°C.

Time (s)	v (nm/s)	Calc. ΔGm (J/mol)	ΔGdis (J/mol) (non K-S)	ΔGdis (J/mol) (K-S)
10	213	52	63	97
60	87	21	48	80

Table 3. Comparison of calculated ΔG_m and ΔG_dis at 600°C.

Time (s)	v (nm/s)	Calc. ΔGm (J/mol)	ΔGdis (J/mol)
10	49	200	746
60	14	58	717
600	4	18	502

4.2. Bainitic Transformation of Fe–N Binary Alloy

The bainite structure is classified according to the morphology of BF.²⁷⁾ The rod-like BF extending in one direction along with the habit plane and the plate-like BF are defined as upper and lower bainite, respectively. Furthermore, the upper bainite is further classified into B-I, B-II, and B-III types, depending on the presence or absence of precipitates and their locations. The B-I type is a lath-like ferrite structure without precipitates, the B-II type is a structure in which precipitates exist between the adjacent laths, and the B-III type is a structure in which precipitates are regularly arranged in the lath, respectively. For the microstructure of the B-I type bainite in the Fe–C system, pearlite or MA exist due to the carbon enrichment into the untransformed γ accompanied with the growth of BF. As shown in Fig. 3, B-I type bainite in which MA exists between BFs was obtained at the early stage of transformation and changed to B-II type bainite with precipitation of interlath γ’ during prolonged isothermal holding. In addition, as shown in Fig. 8, the nitrogen concentration in the retained γ of the sample isothermally treated at 500°C exceeded the T₀ line and had a lower nitrogen concentration than the Aγ’ line in the early stage of the transformation of 5 s holding. Then the nitrogen concentration reached to the Aγ’ line with increasing the holding time, in good agreement with microstructure evolution at this temperature.

5. Conclusions

In this study, the morphology and transformation kinetics of ferrite and bainite as well as and partioning of nitrogen at α/γ interface were investigated in an Fe-0.3mass% N alloy. Following results were obtained.

(1) By isothermal holding in the α + γ two phase region, two types of α were nucleated. One is the AF formed from the prior γ grain boundary, and another one is WF mainly formed at the AF. On the other hand, in the case of isothermal holding at 500°C, nitride-free (B-I type) BF was formed at the beginning, and changed to B-II type bainite with γ’ precipitation in further holding.

(2) Retained γ was obtained by decreasing the transformation temperature in the α+γ two phase region, and maximum volume fraction of 9% was obtained at 600°C. On the other hand, bainite transformation results in less amount of retained γ due to high α fraction with γ’ precipitation.

(3) AF had near K-S OR with one side of the adjacent prior γ and grew into the grain without K-S OR. On the other hand, WF and BF had near K-S relationship with the matrix, implying high coherency with the matrix. As a results, WF had significantly higher energy dissipation than AF, and the energy dissipation of AF is considered to be due to interfacial friction. On the other hand, the strain energy associated with the transformation was dominant in WF.

Acknowledgement

This research was supported by Iketani Science and Technology Foundation and ISIJ Research Promotion Grant, respectively.

References

• 1)   Y.  Imai: Tetsu-to-Hagané, 51 (1965), 2336 (in Japanese).
• 2)   L.  Cheng,  A.  Böttger,  Th. H.  de Keijser and  E. J.  Mittemeijer: Scr. Metall. Mater., 24 (1990), 509.
• 3)   T.  Bell: J. Iron Steel Inst., 206 (1968), 1017.
• 4)   T.  Tsuchiyama,  K.  Inoue,  K.  Hyodo,  D.  Akama,  N.  Nakada,  S.  Takaki and  T.  Koyano: ISIJ Int., 59 (2019), 161.
• 5)   M.  Sato,  S.  Matsumoto,  G.  Miyamoto and  T.  Furuhara: Scr. Mater., 156 (2018), 85.
• 6)   C. A.  Dube,  H. I.  Aaronson and  R. F.  Mehl: Rev. Metall. Paris, 55 (1958), 201.
• 7)   H. B.  Aaron and  H. I.  Aaronson: Metall. Trans., 2 (1971), 23.
• 8)   H. I.  Aaronson: Decomposition of Austenite by Diffusional Processes, Interscience, New York, NY, (1962), 442.
• 9)   T.  Furuhara,  G.  Miyamoto,  H.  Saito and  T.  Maki: Proc. Int. Conf. on Solid-Solid Phase Transformation in Inorganic Materials, Vol. 1, TMS, Warrendale, PA, (2005), 5.
• 10)   G.  Miyamoto,  R.  Hori,  B.  Poorganji and  T.  Furuhara: Metall. Mater. Trans. A, 44 (2013), 3436.
• 11)   A. D.  King and  T.  Bell: Met. Sci., 8 (1974), 253.
• 12)   Y.-C.  Liu,  F.  Sommer and  E. J.  Mittemeijer: Metall. Mater. Trans. A, 39 (2008), 2306.
• 13)   N.  Nakada,  N.  Fukuzawa,  T.  Tsuchiyama,  S.  Takaki,  T.  Koyano,  T.  Iwamoto and  Y.  Omori: ISIJ Int., 53 (2013), 139.
• 14)   Z.-Q.  Liu,  G.  Miyamoto,  Z.-G.  Yang and  T.  Furuhara: Acta Mater., 61 (2013), 3120.
• 15)   J.  Takeuchi,  T.  Iguchi and  T.  Ishihara: Netsu Shori (J. Jpn. Soc. Heat Treat.), 21 (1981), 26.
• 16)   G.  Miyamoto,  N.  Takayama and  T.  Furuhara: Scr. Mater., 60 (2009), 1113.
• 17)   G.  Miyamoto,  N.  Iwata,  N.  Takayama and  T.  Furuhara: Acta Mater., 58 (2010), 6393.
• 18)   T.  Furuhara,  T.  Moritani,  K.  Sakamoto and  T.  Maki: Mater. Sci. Forum, 539–543 (2007), 4832.
• 19)   T.  Kaneshita,  G.  Miyamoto and  T.  Furuhara: Acta Mater., 127 (2017), 368.
• 20)   N.  Takayama,  G.  Miyamoto and  T.  Furuhara: Acta Mater., 60 (2012), 2387.
• 21)   J. R.  Bradley and  H. I.  Aaronson: Metall. Trans. A, 12 (1981), 1729.
• 22)   M.  Hillert: Acta Mater., 47 (1999), 4481.
• 23)   M.  Hillert and  L.  Höglund: Scr. Mater., 54 (2006), 1259.
• 24)   H.-D.  Wu,  G.  Miyamoto,  Z.-G.  Yang,  C.  Zhang,  H.  Chen and  T.  Furuhara: Acta Mater., 149 (2018), 68.
• 25)   G.  Miyamoto,  K.  Yokoyama and  T.  Furuhara: Acta Mater., 177 (2019), 187.
• 26)   T.  Furuhara,  T.  Chiba,  T.  Kaneshita,  H.-D.  Wu and  G.  Miyamoto: Metall. Mater. Trans. A, 48 (2017), 2739.
• 27)   Y.  Ohmori,  H.  Ohtani and  T.  Kunitake: Trans. Iron Steel Inst. Jpn., 11 (1971), 250.