Effects of {332}〈113〉 Deformation Twinning on Fatigue Behavior of Ti–Mn System Alloys

Abstract

The effects of {332}⟨113⟩ deformation twinning, one of the unique deformation modes for metastable β-type Ti alloys, on the fatigue behavior of Ti–Mn system alloys were investigated focusing on fatigue strength, fatigue crack initiation and propagation. Ti–7Mn and Ti–5Mn–3Mo (mass%) alloys which are primarily deformed by dislocation slips and {332}⟨113⟩ deformation twins, respectively, were subjected to fatigue tests conducted in tensile-tensile mode at room temperature, followed by fracture surface and deformation microstructure analyses. We found for the first time the Ti–5Mn–3Mo alloy has higher fatigue strength as compared to the Ti–7Mn alloy owing to the formation of the {332}⟨113⟩ deformation twins. The {332}⟨113⟩ deformation twins are to some extent responsible for the plastic strain accumulation in place of the dislocations during cyclic deformation. Thus, {332}⟨113⟩ deformation twinning prevents the accumulation of dislocations during cyclic deformation, thereby suppressing fatigue crack initiation. Moreover, formation of the {332}⟨113⟩ deformation twins around crack tip decreases stress concentration at the crack tip and changes the crack propagation direction, as a result, crack propagation speed is decreased. These results indicate that the {332}⟨113⟩ deformation twining is crucial for improving the fatigue properties of metastable β-type Ti alloys.

1. Introduction

Metastable β-type Ti alloys are widely employed for aerospace,^1–3) biomedical,^4–6) chemical⁷⁾ and energy⁸⁾ applications due to their good workability, high specific strength, excellent corrosion resistance and good biocompatibility. In addition, they are one of the promising candidates for the development of high-strength structural materials. The strength of these alloys could be increased by precipitation hardening through microstructural control including the formation of α and/or ω precipitates under various thermomechanical treatments.⁹⁾ High-strength metastable β-type Ti alloys, such as Ti–29Nb–13Ta–4.6Zr,¹⁰⁾ Ti–15V–3Cr–3Sn–3Al,¹¹⁾ Ti–10V–2Fe–3Al¹²⁾ and Ti–3V–11Cr–3Al¹³⁾ that exhibit an ultimate tensile strength (UTS) greater than 1200 MPa after specific thermomechanical treatments¹⁴⁾ have been developed. However, these high-strength Ti alloys are composed of rare and expensive elements such as Nb, V and Ta, which are difficult to obtain in the future due to their extensive industrial use and limited supply.^15,16) In previous studies, Mn was selected as an alloying element due to its β stabilizing effect and lower cost when compared to other β stabilizer elements.¹⁷⁾ The solutionized Ti–Mn alloys consist of bcc matrix and nano-sized athermal ω precipitates which results in its high strength as compared to the widely used α+β type Ti–6Al–4V alloy.^18–21) In particular, Ti–Mn alloys after 90% cold rolling have a UTS higher than 1800 MPa.²²⁾ However, the elongation to fracture (EL) of these high-strength Ti–Mn alloys is lower than 10%.^18–22)

Moreover, the mechanical properties of metastable β-type Ti alloys depend strongly on the stability of the bcc matrix and the deformation mode.²³⁾ The deformation mode in these alloys vary depending on the stability of the bcc matrix as follows: stress-induced α′′ martensite,^24–26) {332}⟨113⟩^27–32) and {112}⟨111⟩^33,34) deformation twinning and dislocation slip. For instance, the alloys deformed by {332}⟨113⟩ deformation twins exhibit better strength-ductility balance than those deformed by dislocation slips due to twin-induced plasticity (TWIP).^35,36) Recently, Gloriant et al.³⁷⁾ and Tsuchiya et al.³⁸⁾ have reported tensile deformation behavior of Ti–Nb and Ti–Mo alloys, respectively, focusing on the formation of {332}⟨113⟩ deformation twins. Similarly, Raabe et al.³⁹⁾ and Gloriant et al.⁴⁰⁾ have studied formation mechanism of the {332}⟨113⟩ twins in Ti–Nb alloys. Tobe et al.⁴¹⁾ have reported possible twinning modes in metastable β-type Ti alloys based on the theory of the crystallography of deformation twinning.

In our previous studies, we found that the addition of Mo to the Ti–Mn alloys promotes the formation of {332}⟨113⟩ deformation twins, and consequently, the solutionized Ti–Mn–Mo alloys exhibit a UTS higher than 900 MPa and an EL higher than 30%.^42,43) The effect of Mo addition on the formation of {332}⟨113⟩ deformation twins focusing on the critical resolved shear stress (CRSS) has already been discussed in a previous study.⁴⁴⁾ Furthermore, the influence of {332}⟨113⟩ deformation twinning on the athermal ω phase during tensile loading in the Ti–Mn–Mo alloys has been previously studied.⁴²⁾ These studies indicate that the Ti–Mn–Mo alloys could be used for structure materials in the future due to their excellent tensile properties. However, the fatigue properties of the Ti–Mn–Mo alloys and effects of the {332}⟨113⟩ deformation twinning on the fatigue behavior have not been investigated yet.

Thus, in the present study, the fatigue properties of Ti–Mn and Ti–Mn–Mo alloys were examined by fatigue tests conducted in tensile-tensile mode at room temperature, with a particular focus on the formation of {332}⟨113⟩ deformation twins. Moreover, in order to investigate the effects of {332}⟨113⟩ deformation twins on the fatigue behavior such as crack initiation and propagation, fracture surface and deformation microstructure of specimens after cyclic deformation were observed.

2. Experimental Procedures

In order to examine the effect of {332}⟨113⟩ deformation twinning on fatigue properties, the stability of bcc matrix must be constant. It is well known that value of valence electron to atom ratio (e/a) is one of the indexes for the stability of bcc matrix. Thus, nominal chemical compositions of Ti–Mn and Ti–Mn–Mo alloys were selected to Ti–7Mn (mass%) and Ti–5Mn–3Mo since the e/a of these alloys are the same. The Ti–7Mn and Ti–5Mn–3Mo ingots were prepared by plasma arc melting using high-purity Ti, Mn and Mo in an Ar atmosphere. The ingots were initially subjected to a homogenization treatment for 43.2 ks at 1273 K in an Ar atmosphere followed by water quenching. The homogenized ingots were then hot-rolled to 70% reduction in thickness at 1173 K in air. To retain the β phase in the alloys, the plates were solution-treated for 3.6 ks at 1173 K in an Ar atmosphere, followed by water quenching. Hereafter, the solutionized Ti–7Mn and Ti–5Mn–3Mo alloys are referred to as TM7 and TMM53, respectively. The chemical compositions of these alloys were analyzed using inductively coupled plasma optical emission spectroscopy (for metallic compositions) and infrared absorption (for oxygen and nitrogen).

The microstructure and constituent phases of the alloys were investigated by optical microscopy (OM), X-ray diffractometry (XRD) and transmission electron microscopy (TEM). The specimens used for microstructure characterizations were mechanically polished using waterproof emery papers of up to #2000 grit, and then up to a mirror-like finish using a colloidal SiO₂ suspension. XRD analyses were performed using a Cu target at an acceleration voltage of 40 kV and a current of 30 mA. TEM observations were carried out at an accelerating voltage of 300 kV.

Tensile properties of the alloys were measured using an Instron-type testing machine with a strain rate of 1.7 × 10⁻⁴ s⁻¹ at room temperature (RT). The fatigue properties of the alloys were measured using an electro-servo-hydraulic testing machine. The tests were conducted at a stress ratio (R) of 0.1 and a frequency (f) of 10 Hz in the tension-tension mode at RT. The maximum stress at which the tested specimen does not undergo failure after 1 × 10⁶ cycles (run out) is regarded as the fatigue limit. The plastic strain accumulated in the specimens during the fatigue tests was evaluated based on plastic hysteresis energy (E_p).⁴⁵⁾ E_p was defined as the product of the area of stress-strain hysteresis loop during fatigue tests and the volume of gauge part of the specimens. The actual stress and strain for a specific number of cycles were measured using a real-time measurement system and the area of stress-strain hysteresis loop was determined by calculating integral of the hysteresis loop. The gauge dimensions of tensile and fatigue test specimens were 5.0 × 1.5 × 1.0 mm and 5.0 × 2.0 × 2.0 mm, respectively. These were obtained from the solutionized alloy plates using an electro-discharge machining followed by mechanical surface polishing using waterproof emery papers and colloidal SiO₂ suspension.

The deformation structure of the tensile and fatigue tested specimens was investigated by electron backscatter diffraction (EBSD) and TEM analyses. The fracture surface of the fatigue tested specimens was observed using a scanning electron microscope (SEM).

3. Results

3.1 Microstructure and phase stability of TM7 and TMM53

Table 1 shows the actual chemical composition and e/a ratio calculated from actual chemical compositions for TM7 and TMM53. It is noted that both the alloys have almost same value of e/a. This suggests that the alloys prepared in the present study have comparable phase stability. In addition, O and N contents are less than 0.09 and 0.003 mass%, respectively. These values are substantially low and any variation in the O content will be too insignificant to have a noticeable effect over the microstructure and mechanical properties of the two alloys.

Table 1 Actual chemical composition (mass%) and e/a ratio for TM7 and TMM53.

Figure 1 shows the XRD profiles of TM7 and TMM53. Diffraction peaks for the β(110) and β(211) planes are observed in the figure. Figure 2(a) and (b) shows OM images of the alloys. This result, together with the XRD profiles, indicates that the matrix of the alloys is composed of equiaxed β grains. The average grain diameter (D) of the β grains is slightly decreased from approximately 496 µm to 335 µm by Mo addition due to the low diffusion coefficient of Mo in Ti. Figure 2(c) and (d) shows selected area electron diffraction (SAED) patterns and dark-field images obtained from diffraction spots of the ω phase for the two alloys. The nano-sized athermal ω phase is dispersed uniformly in both alloys. The formation of the athermal ω phase is affected significantly by the phase stability of the β phase. Hanada et al. reported that the phase stability of the β phase can be estimated by 0002 to 222 reciprocal lattice vector ratios (d^*₀₀₀₂/d^*₂₂₂).⁴⁶⁾ The values of d^*₀₀₀₂/d^*₂₂₂ calculated by SAED patterns are 0.662 each for TM7 and TMM53. This result also indicates that both TM7 and TMM53 have same phase stability.

Fig. 1

XRD profiles of TM7 and TMM53.

Fig. 2

Typical OM images (a), (b) and TEM images (c), (d) of TM7 (a), (c) and TMM53 (b), (d). (c), (d) SAED patterns taken with beam direction B = [110] and dark-field images obtained from ω spot.

3.2 Deformation behavior of TM7 and TMM53

Deformation behaviors of TM7 and TMM53 were investigated by tensile tests. The yield stress (YS), UTS and EL for TM7 and TMM53 are listed in Table 2. These alloys exhibit completely different tensile properties, while they have same phase stability. TMM53 shows comparable UTS as TM7 while YS of the alloy is lower than that of TM7, which is considered to be attributable to large work hardening. It is also noted that TMM53 demonstrates larger EL than TM7. Figure 3 shows EBSD inverse pole figure (IPF) map of the tensile deformed TMM53. Band like deformation structures which have a misorientation angle of approximately 50.5 degree can be seen in the deformed specimen, suggesting that the {332}⟨113⟩ deformation twinning takes place in TMM53 during the tensile deformation. In addition, slip lines also can be observed in the tensile deformed TMM53. The CRSS for the deformation twinning (377 MPa) is lower than that for the slip (446 MPa) in the alloy.⁴⁴⁾ This means that deformation mode for TMM53 changes from {332}⟨113⟩ deformation twinning to the deformation twinning + dislocation slip with increasing tensile stress.

Table 2 YS, UTS and EL for TM7 and TMM53.

Fig. 3

An IPF map of TMM53 tensile deformed to fracture at RT (a) and misorientation profile along the line between points A-A′ in the IPF map (b).

This suggests that the strong work hardening and high ductility of the alloys are caused by {332}⟨113⟩ deformation twins. On the other hand, the deformation twins cannot be observed in TM7. This is because Mo addition can lower the CRSS for {332}⟨113⟩ twinning, which is well documented in a previous paper.⁴⁴⁾

3.3 Fatigue properties of TM7 and TMM53

Figure 4 shows the maximum cyclic stress-number of cycles to failure (S-N_f) curves for studied alloys, evaluated by cyclic deformation at R = 0.1 and f = 10 Hz. It is interesting to note that the fatigue strength of TMM53 which is primarily deformed by {332}⟨113⟩ deformation twinning is higher than that of the TM7 in whole fatigue life region, though these alloys have similar UTS. Moreover, the fatigue limits for TM7 and TMM53 alloys are approximately 150 MPa and 250 MPa, respectively.

Fig. 4

S-N_f curves for TM7 and TMM53 cyclically deformed at R = 0.1 and f = 10 Hz.

Figure 5 shows SEM fractographs for TM7 and TMM53 cyclically deformed at a σ_max = 600 MPa. As shown in Fig. 5(a) and (b), the main crack was initiated from specimen surface in both alloys. The fracture surfaces can be divided into two regions; stable crack growth and final fracture regions. There is no significant difference in the area of these regions for TM7 and TMM53 alloys. In addition, striations and dimples can be found in the crack propagation and final fracture regions, respectively, for both alloys (Fig. 5(c)–(f)). These characteristic similarities between the fracture surfaces of TM7 and TMM53 alloys mean that the fracture mode for both the alloys is the same, indicating an difference in fatigue strength between TM7 and TMM53 alloys is not caused by the fracture mode. Thus, it is supposed that the variations in the fatigue strength of TM7 and TMM53 is due to the differences in their fatigue behavior such as crack initiation and propagation during cyclic deformation.

Fig. 5

SEM fractographs for TM7 and TMM53 after the fatigue tests. (a), (c), (d) TM7, σ_max = 600 MPa, N_f = 16818 cycles, (b), (e), (f) TMM53, σ_max = 600 MPa, N_f = 34086 cycles. (a), (b) Low magnification images of the fracture surface, (c), (f) striations observed in stable crack growth regions, (d), (f) dimples observed in final fracture regions.

It is well known that dislocations accumulated and pile-upped during cyclic deformation lead to fatigue crack initiation.⁴⁷⁾ Thus, in order to investigate the crack initiation behavior of the studied alloys, dislocation structure of TM7 and TMM53 were observed by TEM. Figure 6 shows TEM images of TM7 and TMM53 cyclically deformed to 1000 cycles at σ_max = 450 MPa. Numerous dislocations tangled each other can be seen in TM7 (Fig. 6(a)), while only a few dislocations are observed in TMM53 (Fig. 6(b)). Therefore, it is suggested that the crack initiation life of TMM53 is longer than that of TM7. This seems to be one of the reasons for the high fatigue strength of TMM53.

Fig. 6

TEM bright-field images of dislocation structure in TM7 (a) and TMM53 (b) cyclically deformed at 1000 cycles at σ_max = 450 MPa.

Figure 7 shows the average widths of striation (W_s) on the fracture surfaces of TM7 and TMM53 as a function of the distance from crack initiation site (d_s). The width of striation corresponds to the cracks propagation length per cycle. As shown in Fig. 7, W_s of TMM53 is smaller than that of TM7. This indicates that the crack propagation rate of TMM53 is lower than that of TM7. It is also noted that W_s of TM7 increases rapidly at d_s > 1500 µm. This means that the crack propagation in TM7 is accelerated with increasing number of cycles. On the other hand, in TMM53, the slow crack propagation in the stable crack growth region increases the fatigue strength.

Fig. 7

W_s of TM7 and TMM53 as a function of d_s; TM7, σ_max = 600 MPa, N_f = 16818 cycles. TMM53, σ_max = 600 MPa, N_f = 34086 cycles.

4. Discussions

In order to investigate effects of {332}⟨113⟩ deformation twinning on the fatigue behavior of metastable β-type Ti alloys, deformation behavior and fatigue properties of TM7 and TMM53 were examined. As a result, it was found that the fatigue strength of TMM53, which is primarily deformed by {332}⟨113⟩ deformation twinning, is higher than that of TM7. It is suggested that the high fatigue strength of TMM53 is related to the long crack initiation life and slow crack propagation compared to those of TM7. Thus, the effect of {332}⟨113⟩ deformation twinning on the fatigue behavior of TMM53 was discussed, focusing on crack initiation and propagation.

4.1 Effect of {332}⟨113⟩ deformation twinning on crack initiation

It is well known that {112}⟨111⟩ deformation twins decrease fatigue properties of bcc alloys including β-type Ti alloys, as they become a crack initiation site.^48,49) It is noted that the amounts of shear for {112}⟨111⟩ and {332}⟨113⟩ twins are 0.707 and 0.354, respectively.⁴¹⁾ A large shear generally leads to high stress concentration, resulting in fatigue failure. The shear of {332}⟨111⟩ twins is smaller than that of {112}⟨111⟩ twins, which is favorable for the fatigue properties. We also reported that the propagation of {112}⟨111⟩ deformation twins are disturbed by the other twin variant, resulting in high stress concentration around the intersection.⁴⁴⁾ In contrast, {332}⟨113⟩ deformation twins can pass through the other variant, and therefore, stress concentration around the intersection is not significant.⁴⁴⁾ This means that {332}⟨113⟩ deformation twins does not decrease crack initiation life.

As described earlier, fatigue crack initiation is caused by high density of dislocations accumulated in the course of plastic deformation during fatigue tests.⁴⁷⁾ Figure 8 shows the variation in E_p with σ_max at a number of cycles (N) of 10000 cycles in TM7 and TMM53. Note that the E_p corresponds to the work necessary for plastic deformation during each cycle.⁴⁵⁾ In other words, in the stress-controlled fatigue test, higher E_p results in higher plastic strain. As shown in Fig. 8, there is no remarkable difference in E_p between TM7 and TMM53 at all σ_max, while the dislocation density of the cyclically deformed TMM53 is lower than that of TM7 (Fig. 6). On the other hand, {332}⟨113⟩ deformation twins with a misorientation angle of approximately 50.5 degree can be seen in TMM53 cyclically deformed to 34086 cycles at σ_max = 600 MPa (Fig. 9). This means that the plastic strain accumulated in TMM53 during cyclic deformation is not caused by dislocations only. This result indicates that not only dislocations but also {332}⟨113⟩ deformation twins are responsible for the plastic strain accumulated during cyclic deformation, which leads to a decrease in dislocation density. Therefore, it is possible to conclude that {332}⟨113⟩ deformation twinning is effective on suppression of fatigue crack initiation.

Fig. 8

Variation in E_p with σ_max for TM-7 and TMM-53 at N = 10000 cycles.

Fig. 9

An IPF map of TMM53 cyclically deformed to 34086 cycles at σ_max = 600 MPa (a) and misorientation profile along the line between points A-A′ in the IPF map (b).

4.2 Effect of {332}⟨113⟩ deformation twinning on crack propagation

The crack propagation rate at stable crack growth region for TMM53 is slower than that for TM7 (Fig. 7). Figure 10 shows an OM image of crack growth in TMM53 cyclically deformed to 34086 cycles at σ_max = 600 MPa. The crack and the twins intersect each other, and consequently, the crack is deflected at {332}⟨113⟩ deformation twins. This means that {332}⟨113⟩ deformation twins are formed near the crack tip and then suppress the crack propagation. Figure 11 shows schematic drawings for crack propagation in TMM53 during cyclic deformation. First, {332}⟨113⟩ deformation twins are formed in a plastic zone around a crack tip (Stage I). The stress concentration around the crack tip can be relieved by the formation of the twins, which results in a decrease in crack propagation rate. Thereafter, the cracks pass through or stop at the twins (Stage II). The twin boundary acts as an effective barrier to the crack propagation, and therefore, the crack propagation direction is changed at the boundaries. This also leads to a retardation of crack propagation (Stage III). Thus, one can conclude that {332}⟨113⟩ deformation twinning is important to extend crack initiation life and decrease crack propagation rate during cyclic deformation, resulting in improved fatigue properties of metastable β-type Ti alloys.

Fig. 10

An OM image of a crack propagating in TMM53 cyclically deformed to 34086 cycles at σ_max = 600 MPa.

Fig. 11

Schematic drawings for crack propagation at TMM53 during cyclic deformation.

5. Conclusion

The fatigue strength and behavior of TM7 and TMM53 were examined with particular focus on the effects of {332}⟨113⟩ deformation twinning on fatigue crack initiation and propagation. The following conclusions could be drawn from the present study:

1. (1)    TMM53 which is deformed primarily by {332}⟨113⟩ deformation twinning shows higher fatigue strength than TM7 in whole fatigue life region, while these alloys have similar UTS.
2. (2)    In TMM53, {332}⟨113⟩ deformation twins are partly responsible for the plastic strain accumulation during cyclic deformation instead of dislocations. Therefore, {332}⟨113⟩ deformation twinning leads to low dislocation density, and consequently, fatigue crack initiation is suppressed, compared with TM7.
3. (3)    {332}⟨113⟩ deformation twins formed around a crack tip decreases the stress concentration at the tip, resulting in lowering of crack propagation rate. In addition, the fatigue cracks are stopped and deflected at the twin boundaries, which leads to a decrease in crack propagation rate.

Acknowledgments

This work was supported in part by Grant-in-Aid for Scientific Research (C) (Grant No. 16K06771) from the Japan Society for the Promotion of Science (JSPS), Japan and The light Metal Education Foundation, Japan.

REFERENCES

• 1)   R.R.  Boyer: Mater. Sci. Eng. A 213 (1996) 103–114.
• 2)   M.  Peters,  J.  Kumpfert,  C.H.  Ward and  C.  Leyens: Adv. Eng. Mater. 5 (2003) 419–427.
• 3)   V.A.R.  Henriques,  P.P.  Campos,  C.A.A.  Cairo and  J.C.  Bressiani: Mater. Res. 8 (2005) 443–446.
• 4)   M.  Niinomi: Mater. Sci. Eng. A 243 (1998) 231–236.
• 5)   M.  Niinomi,  M.  Nakai and  J.  Hieda: Acta Biomater. 8 (2012) 3888–3903.
• 6)   J.E.G.  González and  J.C.  Mirza-Rosca: J. Electroanal. Chem. 471 (1999) 109–115.
• 7)   I.  Gurrappa: Mater. Charact. 51 (2003) 131–139.
• 8)   R.W.  Schutz and  H.B.  Watkins: Mater. Sci. Eng. A 243 (1998) 305–315.
• 9)   I.  Weiss and  S.L.  Semiatin: Mater. Sci. Eng. A 243 (1998) 46–65.
• 10)   D.  Kuroda,  M.  Niinomi,  M.  Morinaga,  Y.  Kato and  T.  Yashiro: Mater. Sci. Eng. A 243 (1998) 244–249.
• 11)   M.  Okada: ISIJ Int. 31 (1991) 834–839.
• 12)   G.  Srinivasu,  Y.  Natraj,  A.  Bhattacharjee,  T.K.  Nandy and  G.V.S.  Nageswara Rao: Mater. Des. 47 (2013) 323–330.
• 13)   R.A.  Rowe,  J.M.  Dupouy and  M.B.  Bever: Trans. Metall. AIME 218 (1960) 821.
• 14)   B.G.  Pound: Acta Metall. Mater. 42 (1994) 1551–1559.
• 15)   L.  Erdmann and  T.E.  Graedel: Environ. Sci. Technol. 45 (2011) 7620–7630.
• 16)  U.S. Department of Energy, Critical Materials Strategy Summary, (2011).
• 17)   K.  Cho,  M.  Niinomi,  M.  Nakai,  J.  Hieda,  P.F.  Santos,  Y.  Itoh and  M.  Ikeda: Ceram. Trans. 251 (2014) 21–30.
• 18)   K.  Cho,  M.  Niinomi,  M.  Nakai,  J.  Hieda,  P.F.  Santos,  Y.  Itoh,  T.  Hattori and  M.  Ikeda: Mater. Sci. Forum 783–786 (2014) 1232–1237.
• 19)   K.  Cho,  M.  Niinomi,  M.  Nakai,  P.F.  Santos,  A.M.  Liens,  M.  Ikeda and  T.  Hattori: Ceram. Trans. 254 (2015) 1–12.
• 20)   P.F.  Santos,  M.  Niinomi,  K.  Cho,  M.  Nakai,  H.  Liu,  N.  Ohtsu,  M.  Hirano,  M.  Ikeda and  T.  Narushima: Acta Biomater. 26 (2015) 366–376.
• 21)   P.F.  Santos,  M.  Niinomi,  H.H.  Liu,  K.  Cho,  M.  Nakai,  Y.  Itoh,  T.  Narushima and  M.  Ikeda: J. Mech. Behav. Biomed. Mater. 59 (2016) 497–507.
• 22)   K.  Cho,  M.  Niinomi,  M.  Nakai,  H.H.  Liu,  P.F.  Santos,  Y.  Itoh,  M.  Ikeda,  M.  abdel-hady Gepreel and  T.  Narushima: J. Alloy Compd. 664 (2016) 272–283.
• 23)   M.  Ikeda,  M.  Ueda,  R.  Matsunaga,  M.  Ogawa and  M.  Niinomi: Mater. Trans. 50 (2009) 2737–2743.
• 24)   T.W.  Duerig,  G.T.  Terlinde and  J.C.  Williams: Metall. Trans. A 11 (1980) 1987–1998.
• 25)   S.  Miyazaki,  T.  Imai,  Y.  Igo and  K.  Otsuka: Metall. Trans. A 17 (1986) 115–120.
• 26)   T.  Grosdidier and  M.J.  Philippe: Mater. Sci. Eng. A 291 (2000) 218–223.
• 27)   G.  Carter,  H.M.  Flower,  G.M.  Pennock and  D.F.  West: J. Mater. 12 (1977) 2149–2153.
• 28)   O.A.  Yelkina and  R.M.  Lerinman: Phys. Met. Metallogr. 45 (1979) 78–83.
• 29)   S.  Hanada,  M.  Ozeki and  O.  Izumi: Metall. Trans. A 16 (1985) 789–795.
• 30)   S.  Hanada and  O.  Izumi: Metall. Trans. A 18 (1987) 265–271.
• 31)   Y.  Takemoto,  M.  Hida,  E.  Sukedai and  A.  Sakakibara: J. Japan Inst. Metal. 53 (1989) 1004–1012.
• 32)   T.  Furuhara,  K.  Kishimoto and  T.  Maki: Mater. Trans. 35 (1994) 843–850.
• 33)   H.G.  Paris,  B.G.  LeFevre and  E.A.  Starke: Metall. Trans. A 7 (1976) 273–278.
• 34)   H.  Xing and  J.  Sun: Appl. Phys. Lett. 93 (2008) 031908.
• 35)   C.  Brozek,  F.  Sun,  P.  Vermaut,  Y.  Millet,  A.  Lenain,  D.  Embury,  P.J.  Jacques and  F.  Prima: Scr. Mater. 114 (2016) 60–64.
• 36)   J.Y.  Zhang,  J.S.  Li,  Z.  Chen,  Q.K.  Meng,  F.  Sun and  B.L.  Shen: J. Alloys Compd. 699 (2017) 775–782.
• 37)   A.  Ramarolahy,  P.  Castany,  F.  Prima,  P.  Laheurte,  I.  Péron and  T.  Gloriant: J. Mech. Behav. Biomed. Mater. 9 (2012) 83–90.
• 38)   X.H.  Min,  K.  Tsuzaki,  S.  Emura and  K.  Tsuchiya: Mater. Sci. Eng. A 528 (2011) 4569–4578.
• 39)   M.J.  Lai,  C.C.  Tasan and  D.  Raabe: Acta Mater. 111 (2016) 173–186.
• 40)   P.  Castany,  Y.  Yang,  E.  Bertrand and  T.  Gloriant: Phys. Rev. Lett. 117, 245501.
• 41)   H.  Tobe,  H.Y.  Kim,  T.  Inamura,  H.  Hosoda and  S.  Miyazaki: Acta Mater. 64 (2014) 345–355.
• 42)   P.F.  Santos,  M.  Niinomi,  H.  Liu,  K.  Cho,  M.  Nakai,  A.  Trenggono,  S.  Champagne,  H.  Hermawan and  T.  Narushima: Mater. Des. 110 (2016) 414–424.
• 43)   P.F.  Santos,  M.  Niinomi,  K.  Cho,  H.  Liu,  M.  Nakai,  T.  Narushima,  K.  Ueda and  Y.  Itoh: Mater. Trans. 58 (2017) 271–279.
• 44)   R.  Morioka,  K.  Cho and  H.Y.  Yasuda: Mater. Sci. Forum 941 (2018) 1360–1365.
• 45)   T.  Nakano,  H.Y.  Yasuda,  N.  Higashitanaka and  Y.  Umakoshi: Acta Mater. 45 (1997) 4807–4821.
• 46)   S.  Hanada: Bull. Jpn. Inst. Met. 25 (1986) 755–764.
• 47)   K.  Tanaka and  T.  Mura: J. Appl. Mech. 48 (1981) 97–103.
• 48)   S.  Hashimoto,  H.  Ikehata,  A.  Kato and  H.  Kato: Interface Sci. 7 (1999) 159–171.
• 49)   Y.M.  Hu,  W.  Floer,  U.  Krupp and  H.-J.  Christ: Mater. Sci. Eng. A 278 (2000) 170–180.