High Magnetic Field Effects on Cu-precipitation Behavior of Fe-1mass%Cu at 773 K

Abstract

Tramp elements in steel, such as Cu and Sn, cannot be removed by acid treatment. Since these elements condense by repeating recycling process, leading to deterioration of strength. Therefore, the methods for avoiding condensation or removing tramp elements are required. In this study, in-magnetic-field annealing process was focused on because magnetic field is effective for diffusion, phase transformation, phase diagram and precipitation. In-magnetic-field annealing of Fe-1mass%Cu at 773 K was performed in 5 and 10 T for investigating precipitation behavior of supersaturated Cu. From microstructural observation, precipitation of Cu-rich phase in Fe-matrix, and magnetic field effect on itself were not observed clearly. Increase of the hyperfine field was detected for the samples annealed at 5 T by Mössbauer spectroscopy, indicating the enhancement of the Cu-precipitation. On the contrary, hyperfine field for 10T-annealed sample was slightly smaller than that for 0 T. Therefore, in-field annealing effect on Cu-precipitation became unclear at 10 T. These magnetic field effects were discussed in the viewpoints of the change of Cu–Fe phase diagram and the atomic diffusion under magnetic field. Difference of the magnetic field effects on precipitation between 5 T and 10 T is explained by the competition between the enhancement of the driving force of the precipitation and suppression of the atomic diffusion. The obtained results indicated that there is optimized magnetic field intensity for controlling Cu-precipitation.

1. Introduction

Effects of magnetic field, including both steady magnetic field and alternative field, on materials processing have been studied.^1,2,3,4,5) In-magnetic-field annealing (IFA) realized a unique crystal growth, such a grain refinement^6,7) or coarsening,^8,9) alignment of ferromagnetic grains in non-magnetic matrix,⁸⁾ which were due to the magnetic anisotropy, dipole-dipole interactions, and the Lorentz force. Magnetic states of precipitates/reactant and matrix are also important factor for controlling phase transformation or phase equilibrium by magnetic field because of Zeeman energy E_M.^10,11) In-magnetic-field process for Fe-C-based alloys has also been reported.^12,13,14,15) For example, 10 T-class of magnetic field effectively influenced Fe–C phase diagram.^13,15) In addition, the phase fraction of γ-Fe in α + γ region was controlled by magnetic field,¹³⁾ which was explained by change of Fe–C phase diagram due to E_M of α-Fe phase.

Meanwhile, there are many reports for magnetic-field-induced suppression of atomic diffusion for both ferromagnetic and paramagnetic diffusion couples.^{16,17,18,19,20)} IFA effects have been explained by magnetostriction¹⁷⁾ and Lorentz force for the phase formations.¹⁸⁾ As described above, there are both enhancement/suppression effects on the phase transformation and phase formation under magnetic fields. IFA-induced suppression of atomic diffusion exhibited despite of magnetic states of metals, whereas difference of magnetic states of the phases plays a key role for IFA-induced enhancement of the phase transformation. Therefore, it is expected that IFA effects on precipitation in Fe-based alloys are complex.

Copper and tin in the steel were known as tramp element, which condensation by recycling induced the deterioration of strength.²¹⁾ Since tramp elements were not removal by acid treatment, alternative approaches for improving recycling process, such as avoiding condensation, removing and analyzing the distribution, have been studied.^22,23,24,25) Among them, magnetic-field-induced suppression of the segregation of Sn in Fe-0.8Sn was reported, which was discussed in the viewpoint of magnetic contribution at grain boundary.²³⁾ In addition, solidification of Fe–Cu and Fe–C–Cu alloys under magnetic field was performed by Yuan et al.²⁶⁾ Magnetic field contributed the uniform distribution of Cu in γ-Fe-C matrix of Fe–C–Cu alloy, which magnetic field effects was discussed in the viewpoint of Gibbs energy of γ-Fe-C and liquid under magnetic field. On the other hand, we focused on E_M of α-Fe and change of the phase diagram. We performed IFA for Fe-5mass%Cu alloys around eutectoid temperature, and found that the Cu-precipitation was enhanced by magnetic field when the annealing temperature was just above eutectoid temperature.²⁷⁾ It was suggested that magnetic-field-induced change of the Cu–Fe phase diagram played a key role for the IFA effects on the precipitation.²⁷⁾ However, Cu solubility in α- and γ-Fe is up to ~3 at.% around eutectoid temperature of 1123 K,^29,30) which is difficult to precipitate for steels containing minimal amount of Cu by annealing. In comparison, Cu solubility in α-Fe is rapidly decreased with decreasing temperature from eutectoid temperature. Therefore, much lower temperature than eutectoid temperature is required to precipitate Cu efficiently. If Cu condenses at grain boundary by annealing, it is expected that magnetic separation of Cu from Fe will be effective.

In this study, IFA for Fe-1mass%Cu at 773 K were performed because Cu-solubility is quite small at 773 K. In addition, large E_M is expected because α-Fe is ferromagnetic at 773 K. The precipitation behavior was evaluated by microstructural observation and Mössbauer spectroscopy.

2. Experimental

Cylindrical-shaped Fe-1mass%Cu (Fe-0.88at.%Cu) ingots with ϕ12 mm were prepared by induction melting of nominal composition of pure elements (Fe, 99.9% Cu, 99.99%) with total weight of 0.15 kg, following the casting into mild steel mold. Cu is supersaturated in α-Fe phase because of rapid cooling. From microstructural observation for as-cast sample by electron probe micro analyzer with LaB₆ source (EPMA, JXA-8230, JEOL), The morphology was almost same as annealed samples, and segregation or precipitation was not observed clearly.

In-field heat treatments at 0 (ZFA) and 5 T (5T-IFA) were carried out at 773 K by electric furnace utilizing to cryogen-free superconducting magnet (CSM) (Tamakawa Co. Ltd.). The sample was cut into quadrant. The applied magnetic field direction was parallel to the cylindrical axis. It should be noted that the annealing temperature 773 K was much lower than the Curie temperature of α-Fe ~1043 K.

The other series of Fe-1mass%Cu were annealed at 0 and 10 T (10T-IFA) using another electric furnace and 10T-CSM. In this series, the samples with semicircular shapes with 2–3 mm thick were sealed in the vacuum quartz tubes. The applied magnetic field direction was perpendicular to the cylindrical axis, which was different from 5T-IFA series. The annealing conditions (magnetic field B_a and annealing time t_a) were listed in Table 1. For both series, the sample temperature was elevated to 773 K for 10 K/min. and cooled down by furnace cooling after the annealing. Since the cooling water was circulating around the furnace, the cooling rate of sample temperature is faster than conventional furnace cooling.

Table 1. Annealing condition of Fe-1mass%Cu.

Sample	Magnetic field, Ba (T)	Annealing time, ta (h)
Batch 1	0 (ZFA)	6, 12, 24
	5 (5T-IFA)	6, 12, 24
Batch 2	0 (ZFA)	12
	10 (10T-IFA)	12

For evaluating Cu-precipitation effects, two measurements were carried out. One was the microstructural observation by EPMA. Elemental composition analysis was performed by energy dispersion X-ray spectroscopy (EDS) with ~10 nA. For evaluating the average amount of Cu-precipitation, ⁵⁷Fe Mössbauer spectroscopy using ⁵⁷Co source was performed by a conventional constant acceleration method at room temperature (RT). The powder samples were obtained by filing the bulk samples. The velocity scale was calibrated by pure α-Fe powder (hyperfine field B_hf = 33.1 T). The obtained spectra were analyzed by NORMOS program³⁰⁾ and Cu-precipitation was qualitatively evaluated by B_hf.

3. Results

Figure 1 shows the secondary electron (SE) images and backscattered electron (BSE) images of Fe-1mass%Cu annealed for 24 h at 773 K in a zero field (a)(b), and 5 T (c)(d), respectively. The obtained SE images indicated that Cu-rich phase did not precipitate and the morphology seemed homogeneous for both ZFA and IFA samples. Meanwhile, there are light and dark area in BSE images in both samples. The brightness of light and dark area seemed multi-level. The Fe and Cu compositions were evaluated by point-analysis for each two points in darker and lighter area, which were indicated in Figs. 1(b) and 1(d). Table 2 shows the compositions Fe and Cu at the points. Herein, in addition to Fe and Cu, about 1 at.% of carbon was detected due to resin. In Table 2, total compositions of Fe to Cu were set to 100% after excluding the carbon composition. The composition of Fe and Cu was not systematic with respect to brightness for both samples. In other words, the light area did not indicate the Cu-rich phase. It is suggested that the slight composition change exhibited in the thinner area than penetration depth of X-ray. Consequently, the precipitation of Cu-rich phase and its IFA effects were not detected clearly from microstructural observation.

Fig. 1.

SE and BSE images of Fe-1mass%Cu annealed for 24 h at ZFA (a) (b) and 5 T-IFA samples (c) (d), respectively. Crosses in the images indicates the points that Fe and Cu compositions were evaluated by EDS.

Table 2. Composition of Fe and Cu for Fe-1mass%Cu at the points indicated in Fig. 1.

Point	Brightness	ZFA, 24 h		5T-IFA, 24 h
		Fe (at.%)	Cu (at.%)	Fe (at.%)	Cu (at.%)
1	Light	98.64	1.36	99.00	1.00
2	Light	98.94	1.07	99.28	0.72
3	Dark	99.14	0.86	99.24	0.76
4	Dark	98.63	1.37	98.85	1.15

For evaluating average value of Cu-precipitation from bcc-lattice of α-Fe, Mössbauer spectroscopy was performed. From Mössbauer spectroscopy, it is reported that magnetic moment of α-Fe decreased by Cu substitution.³¹⁾ That is, B_hf of α-Fe indicated the average value of the Cu-precipitation. Figures 2 and 3 show the Mössbauer spectra for the ZFA and IFA samples at RT. The sextet peaks, which derived from ferromagnetism of α-Fe, are clearly observed for all samples. The calculated spectra (solid curves) well represented the experimental data (open circles). Figure 4 shows the annealing time t_a dependence of B_hf for ZFA and IFA samples. With increasing annealing time, B_hf tended to increase and approached to 33.1 T, which is B_hf of pure α-Fe. B_hf for both batches varied from 32.93 to 33.03 T with increasing t_a. According to Ref. 31, substitution of 1 at.%Cu decreased B_hf by ~1%, which is comparable to the present work. Therefore, it should be noted that precipitation behavior of supersaturated Cu from α-Fe solution is observed by Mössbauer spectroscopy. 5T-IFA samples show higher B_hf than ZFA samples. Therefore, magnetic field of 5 T enhanced the Cu-precipitation.

Fig. 2.

Mössbauer spectra for Fe-1mass%Cu for ZFA (a) and 5T-IFA samples (b) (batch 1). The solid curves were calculated by NORMOS.

Fig. 3.

Mössbauer spectra for Fe-1mass%Cu annealed at ZFA and 10T-IFA samples (batch 2). The solid curves were calculated by NORMOS.

Fig. 4.

Annealing time dependence of hyperfine field B_hf. Broken line at B_hf = 33.1 T indicates the B_hf of pure α-Fe.

Meanwhile, as seen in Fig. 4, B_hf for 10T-IFA sample shows slightly smaller than that for ZFA sample, which is different behavior to 5T-IFA. Although only one condition was performed for 10T-IFA, IFA-induced enhancement of precipitation disappeared by application of 10 T.

4. Discussion

In this study, IFA effect on precipitation was influenced by magnetic field intensity. Firstly, the effects of difference of applied magnetic field direction between 5T-IFA and 10T-IFA were discussed. The difference of the direction influences the morphology and E_M. IFA-induced anisotropic morphology, such as anisotropic shape of Cu-precipitates cannot be observed because Cu-precipitates themselves were not observed clearly. In addition, the effective E_M is decreased demagnetizing field. In this study, the effect of demagnetizing field of 10T-IFA samples were smaller than 5T-IFA because longer axis of the sample were parallel to magnetic field. However, IFA effects for 10T-IFA samples seemed to disappear. Therefore, the difference of the applied field direction was not effective in this study.

Next, we focus on two factors for controlling precipitation: one contributes for enhancement of precipitation, and another contributes for suppression. In this study, these factors are explained by change of the phase diagram and magnetic-field induced-suppression of diffusion, respectively.

Enhancement effect of Cu-precipitation is discussed based on the Cu–Fe phase diagram in magnetic field. Figure 5 shows the schematic illustration of magnetic field effect on the phase diagram. When supersaturating degree Δx is small, driving force for precipitation ΔG is linear to ΔT(H)/T_a.³²⁾ According to the Cu–Fe phase diagram, solidus line at Fe-1mass%Cu (Fe-0.88at.%Cu) is ~1020 K,²⁸⁾ therefore ΔT is evaluated to be 247 K. In Cu–Fe phase diagram, α-γ transformation at eutectoid temperature is paramagnetic-paramagnetic transition. However, since magnetic susceptibility of α-Fe is still higher than γ-Fe, α-γ transformation temperature T_α-γ rose by application of magnetic field because of difference of gain of Zeeman energy between α- and γ-phase. Difference of Zeeman energy ΔG_α-γ in paramagnetic region can be described as   

$$\begin{array}{l}\mathrm{\Delta }{G}_{\alpha -\gamma }={\mu }_0\int {}_0^H{M}_{\alpha }dH-{\mu }_0\int {}_0^H{M}_{\gamma }dH\\ \hspace{1em}\hspace{1em}\hspace{1em}={\mu }_0\int {}_0^H\mathrm{\Delta }{\chi }_{\alpha -\gamma }(T)HdH=\frac{1}{2}{\mu }_0\mathrm{\Delta }{\chi }_{\alpha -\gamma }{H}^2,\end{array}$$(1)

where M_α and M_γ are magnetization of α- and γ-Fe, respectively, and χ_α-γ is difference in magnetic susceptibility between α- and γ-Fe. As ΔG_α-γ is linear to H², T_α-γ also increases linear to H². From high-field DTA, T_α-γ for pure iron increased by ~5 K under 5 T.¹⁵⁾ Therefore, eutectoid temperature rises slightly by application of magnetic fields, resulting in increase of ΔT. When solidus curves rise 5 K at 5 T, ΔT/T_a increased by ~3%, which contribute the enhancement of precipitation. If magnetic-field-induced rise of solidus curve was simply shift of the curve without change of the curvature, Δx and solubility x_s are enhanced and reduced in magnetic field, respectively. The change of x_s can be explained by gain of E_M for α-Fe phase. Figure 6 shows the schematic illustration of Gibbs energy of α-Fe and Cu. x_s was contact point of x-G_α curve and common tangent line obtained by G_α and Gibbs energy of Cu-phase G_Cu. When applying magnetic field, G_Cu did not changed since E_M for diamagnetic G_Cu is quite small. On the other hand, E_M for G_α becomes large with increasing Fe concentration because magnetic moment and T_C of α-Fe phase decreased by adding Cu.³¹⁾ Therefore, as seen in broken curve in Fig. 6, curvature of G_α changed, resulting in the change of the common tangent line and x_s. This change of solubility enhances driving force of precipitation because the driving force is also liner to Δx/x_s as well as ΔT/T_a. However, magnetic field effects on the solubility and solidus curves are unclear at present. For investigating these effects, the phase diagram under magnetic field by CALPHAD approach and high-field DTA is required.

Fig. 5.

Schematic illustration for magnetic field effects on the Cu–Fe phase diagram.

Fig. 6.

Schematic illustration of the relationship between G_α, G_Cu, and Fe concentration. x_s(0 T) and x_s(B_a) is the solubility of Cu in α-Fe in zero field and in B_a, respectively. The broken curve indicated G_α under magnetic field. The solid and broken line indicated the common tangent line between G_α, G_Cu in zero field and in magnetic field, respectively.

Next, 10T-IFA effects was focused on. Increment of eutectoid line at 10 T will be 4 times larger than 5 T because increment of solidus line is supposed to be linear to H². When solidus line rise 20 K at 10 T, ΔT/T_a increased by ~8% by application of 10 T. However, the amount of Cu-precipitation at 10T-IFA is almost same as ZFA. This can be explained by magnetic field effect on the diffusion. 10 T-class magnetic field effects on atomic diffusion and phase formation were evaluated by reaction of diffusion couples and were discussed based on the parameters in following Arrhenius-type equation:   

$$k=k_0\text{exp}\left(\frac{-Q}{RT}\right),$$(2)

where k, k_0, Q, and R were rate constant, frequency factor, activation energy and gas constant, respectively.^17,18,19) Diffusion under magnetic field was reported for both substitutional and interstitial solutions, indicating the reduction of k₀. In substitutional solutions, such as Ni–Al¹⁸⁾ and Fe–Ga,¹⁹⁾ k₀ for forming some phases under 10 T was a fraction of k₀ in 0 T. Therefore, magnetic-field-induced suppression degree is larger than enhancement of ΔG as described above. On the contrary, enhancement effect of precipitation by ΔG was more effective than suppression effects by reduction of k₀ in 5 T, leading to the enhancement of precipitation. However, many reports for IFA effects focused on with/without magnetic field. Therefore, it is required that investigation for magnetic field intensity on the above IFA effects. Actually, IFA effects did not appear in diffusion of Ni–Ti system at 4 T,³³⁾ and Mn–Ga system at 5 T.³⁴⁾

5. Conclusions

Magnetic field effect on Cu-precipitation behavior in Fe-matrix at Fe-1mass%Cu was investigated. From microstructural observation, Cu-precipitation was not detected clearly regardless of the application of magnetic field. On the other hand, hyperfine field indicated that slight enhancement of Cu-precipitation exhibited by 5T-IFA, while magnetic field effects did not appear at 10T-IFA. The above IFA effects depending on magnetic field intensity were discussed in the viewpoint of the change of the phase diagram and atomic diffusion under magnetic fields, which contributed enhancement and suppression effects on Cu-precipitation. Therefore, there is optimum magnetic field intensity for controlling precipitation.

Acknowledgment

This work was supported by 28th ISIJ research promotion grant (Ishihara/Asada grant). High-field experiments were carried out at High Field Laboratory for Superconducting Materials, Institute for Materials Research, Tohoku University (Project No. 17H0072, 18H0032, and 19H0005). The authors are grateful to Ms. Y. Watanabe (Graduate Student, Kagoshima University) for supporting microstructural observation.

References

• 1)   T. S.  Suzuki,  T.  Uchikoshi and  Y.  Sakka: J. Eur. Ceram. Soc., 30 (2010), 2813. https://doi.org/10.1016/j.jeurceramsoc.2010.04.022
• 2)   X.  Zuo,  K.  Han,  C.  Zhao,  R.  Niu and  E.  Wang: Mater. Sci. Eng. A, 619 (2014), 319. https://doi.org/10.1016/j.msea.2014.09.070
• 3)   R.  Chen: J. Mater. Sci., 55 (2020), 4054. https://doi.org/10.1007/s10853-019-04249-2
• 4)   K.  Iwai and  H.  Ishii: ISIJ Int., 60 (2020), 707. https://doi.org/10.2355/isijinternational.ISIJINT-2019-411
• 5)   M.  Li and  T.  Tamura: Metall. Mater. Trans. A, 51 (2020), 2939. https://doi.org/10.1007/s11661-020-05725-0
• 6)   Y.  Ma,  K.  Watanabe,  S.  Awaji and  M.  Motokawa: Appl. Phys. Lett., 77 (2000), 3633. https://doi.org/10.1063/1.1330227
• 7)   J.  Wang,  S.  Yue,  Y.  Fautrelle,  P. D.  Lee,  X.  Li,  Y.  Zhong and  Z.  Ren: Sci. Rep., 6 (2016), 24585. https://doi.org/10.1063/1.1330227
• 8)   C.  Lou,  Q.  Wang,  T.  Liu,  N.  Wei,  C.  Wang and  J.  He: J. Alloy. Compd., 505 (2010), 96. https://doi.org/10.1016/j.jallcom.2010.06.050
• 9)   Y.  Mitsui,  R. Y.  Umetsu,  K.  Takahashi and  K.  Koyama: J. Magn. Magn. Mater., 453 (2018), 231. https://doi.org/10.1016/j.jmmm.2018.01.026
• 10)   Y.  Mitsui,  K.  Koyama and  K.  Watanabe: Mater. Trans., 54 (2013), 242. https://doi.org/10.2320/matertrans.M2012310
• 11)   R.  Kobayashi,  Y.  Mitsui,  R. Y.  Umetsu,  K.  Takahashi,  M.  Mizuguchi and  K.  Koyama: Mater. Trans., 58 (2017), 1511. https://doi.org/10.2320/matertrans.MAW201709
• 12)   M.  Shimotomai and  K.  Maruta: Scr. Mater., 42 (2000), 499. https://doi.org/10.1016/S1359-6462(99)00381-4
• 13)   J.-K.  Choi,  H.  Ohtsuka,  Y.  Xu and  W.-Y.  Choo: Scr. Mater., 43 (2000), 221. https://doi.org/10.1016/S1359-6462(00)00394-8
• 14)   T.  Garcin,  S.  Rivoirard,  C.  Elgoyhen and  E.  Beaugnon: Acta Mater., 58 (2010), 2026. https://doi.org/10.1016/j.actamat.2009.11.045
• 15)   Y.  Mitsui,  Y.  Ikehara,  K.  Takahashi,  S.  Kimura,  G.  Miyamoto,  T.  Furuhara,  K.  Watanabe and  K.  Koyama: J. Alloy. Compd., 632 (2015), 251.
• 16)   X. J.  Liu,  Y.  Lu,  Y. M.  Fang and  C. P.  Wang: Calphad, 35 (2011), 66. https://doi.org/10.1016/j.calphad.2010.12.004
• 17)   H.  Fujii and  S.  Tsurekawa: Phys. Rev. B, 83 (2011), 054412. https://doi.org/10.1103/PhysRevB.83.054412
• 18)   C.  Li,  Z.  Yuan,  R.  Guo,  W.  Xuan,  Z.  Ren,  Y.  Zhong,  X.  Li,  H.  Wang and  Q.  Wang: J. Alloy. Compd., 641 (2015), 7. https://doi.org/10.1016/j.jallcom.2015.04.061
• 19)   Y.  Mitsui,  T.  Hagio,  K.  Takahashi and  K.  Koyama: ISIJ Int., 60 (2020), 807. https://doi.org/10.2355/isijinternational.ISIJINT-2019-470
• 20)   Y.  Watanabe,  Y.  Mitsui,  R.  Kobayashi,  R. Y.  Umetsu and  K.  Koyama: J. Alloy. Compd., 887 (2021), 161310. https://doi.org/10.1016/j.jallcom.2021.161310
• 21)   I.  Daigo,  L.  Fujimura,  H.  Hayashi,  E.  Yamasue,  S.  Ohta,  T. D.  Huy and  Y.  Goto: ISIJ Int., 57 (2017), 388. https://doi.org/10.2355/isijinternational.ISIJINT-2016-500
• 22)   T.  Nishi,  S.  Fukagawa,  K.  Shinme and  T.  Matsuo: ISIJ Int., 39 (1999), 905. https://doi.org/10.2355/isijinternational.39.905
• 23)   S.  Tsurekawa,  K.  Okamoto,  K.  Kawahara and  T.  Watanabe: J. Mater. Sci., 40 (2005), 895. https://doi.org/10.1007/s10853-005-6507-2
• 24)   L.  Zhang and  K.  Wagatsuma: ISIJ Int., 53 (2013), 2201. https://doi.org/10.2355/isijinternational.53.2201
• 25)   S.  Kuzuhara,  K.  Kudo,  H.  Hayashi and  I.  Daigo: ISIJ Int., 59 (2019), 2120. https://doi.org/10.2355/isijinternational.ISIJINT-2019-235
• 26)   Y.  Yuan,  K.  Sassa,  K.  Iwai,  Q.  Wang,  J.  He and  S.  Asai: ISIJ Int., 48 (2008), 901.
• 27)   K.  Sato,  Y.  Mitsui,  S.  Kuzuhara and  W.  Ito: CAMP-ISIJ, 33 (2020), PS-36, CD-ROM (in Japanese).
• 28)   Q.  Chen and  Z.  Jin: Metall. Mater. Trans. A, 26 (1995), 417. https://doi.org/10.1007/BF02664678
• 29)  National Institute for Materials Science: AtomWork, http://crystdb.nims.go.jp, (accessed 2021-10-08).
• 30)   R. A.  Brand: WinNormos-for-Igor, https://www.uni-due.de/~hm236ap/keune/deutsch/Brand/r.a.brand.homepage.htm#Normos, (accessed 2021-08-03).
• 31)   C. L.  Chien,  S. H.  Liou,  D.  Kofalt,  W.  Yu,  T.  Egami and  T. R.  McGuire: Phys. Rev. B, 33 (1986), 3247. https://doi.org/10.1103/PhysRevB.33.3247
• 32)   K.  Oikawa: Bull. Iron Steel Inst. Jpn., 25 (2020), 26.
• 33)   H.  Nakajima,  S.  Maekawa,  Y.  Aoki and  M.  Koiwa: Trans. Jpn. Inst. Met., 26 (1985), 1.
• 34)   Y.  Watanabe,  T.  Hagio,  R.  Kobayashi,  Y.  Mitsui and  K.  Koyama: Mater. Trans., 60 (2019), 2195. https://doi.org/10.2320/matertrans.MT-MAW2019013