Effects of Strain Rate on the Deformation Mechanism of Ultra-high Strength TWIP Steel

Abstract

The deformation mechanism of Fe-20Mn-0.6C twinning-induced plasticity (TWIP) steel was studied with respect to different strain rates ranging from 10⁻⁴ to 10³ s⁻¹. Moreover, the microstructure of the ultra-high strength TWIP steel at each strain rate was characterized by transmission electron microscopy (TEM). The TWIP steel exhibits three distinct strain hardening stages with increasing true strain. In stage II, dσ/dε shows a plateau at the strain rates of 10⁻³ to 10⁻¹ s⁻¹, while dσ/dε continuously decreases in the other stages with increasing strain rate. The deformation mechanism of TWIP steel under the high strain rate was a process in which the deformation twin and the dislocation slip promoted and restricted each other. When the strain rate is higher than 10² s⁻¹, the increase in the adiabatic heating temperature (approximately 143°C) suppresses the secondary twinning and enhances the softening effect.

1. Introduction

Twinning induced plasticity (TWIP) steel possesses low fault energy characteristics, excellent plasticity, and high strain hardenability during the deformation process. Moreover, the high strength and negligible low-temperature brittle transformation of TWIP steel make it a promising choice for lightweight automotive applications.^1,2,3) Furthermore, the fabrication and forming process of TWIP steel have garnered much attention due to its excellent plasticity, optimal toughness, and widespread utilization in the automotive industry.^4,5,6) One should note that the typical strain rate of TWIP steel is in the range of 10⁻¹ to 10¹ s⁻¹. Moreover, the mechanical properties under high strain rate deformation (10²–10³ s⁻¹) have also gained significant focus due to the hazard of high-speed vehicle collision.^7,8,9,10,11)

Recently, different research groups^12,13,14,15) have conducted in-depth studies on the mechanical behavior of TWIP steel under different strain rates. The results revealed that TWIP steel exhibits different deformation characteristics under static, quasi-static (e.g., strain rate of 10⁻⁶–10⁻¹ s⁻¹), and dynamic loadings (e.g., strain rate of 10⁰–10³ s⁻¹). It has been reported that deformation twins can be generated in face-centered cubic austenite grains of TWIP steel, even at a reduced strain rate of 10⁻⁴ s⁻¹, in the temperature range of −70–400°C. Moreover, Kim J K et al.¹⁶⁾ had demonstrated that dynamic strain aging (DSA) was very likely due to the interaction between dislocations with a jerky motion pattern and point-defect complexes such as C–V complexes. This mechanism allows for rapid strain aging at the room temperature, when the diffusivity of the mobile interstitial solute C atoms is very low. Other authors¹⁷⁾ concluded that this phenomenon is also related to the strain rate, temperature, and other conditions of plastic deformation. Grässel et al.¹⁸⁾ and Frommeyer et al.¹⁹⁾ had observed that the Fe–Mn–Si–Al system, a typical TWIP steel, does not exhibit any strain rate sensitivity under different strain rates when the Mn content is less than 20 mass%. However, when the Mn content exceeded 25 mass%, the TWIP steel exhibited an obvious change in its mechanical behavior and the flow stress increased with increasing strain rate. Xu Shanqing et al.²⁰⁾ investigated the dynamic tensile behavior of Fe-23.7Mn-2.3Si-2.7Al-0.01C (mass%) TWIP steel under medium and low strain rates (10⁻³ to 4 × 10² s⁻¹) and demonstrated a significant decrease in the tensile strength at low strain rates (10⁻³ to 10⁻² s⁻¹). However, the tensile strength and hardening rate increased with increasing strain rate in the range of 10⁻²–4 × 10² s⁻¹. Curtze and Kuokkala²¹⁾ researched the correlation between the SFE (20.5–42 mJ·m⁻²) and the mechanical behavior of TWIP steels with strain rates between 10⁻³ and 1250 s⁻¹. At a high strain rate, the elongation of the steel decreased evidently, which was ascribed to the increase in SFE due to adiabatic heating that essentially caused the dislocation slip displace the twinning.

However, the rapid development of the automobile industry demands in-depth research on the deformation behavior of TWIP steel with high yield strength under different strain rates. In this study, the mechanical properties and deformation mechanism of Fe-20Mn-0.6C TWIP steel have been systematically investigated under different strain rates.

2. Experimental

TWIP steel having the composition Fe-20.12Mn-0.59C-1.53Al-0.04Nb-0.47V-0.03Ti (mass%) was prepared in a vacuum induction furnace and cast into ingots. The ingot was sectioned into rectangular pieces (80 mm × 80 mm × 70 mm) for forging and the as-forged TWIP steel was heated at 1180°C for 1.5 h before rolling. The initial rolling temperature was 1100°C, and the thickness of tested steel was 4 mm after seven hot rolling passes. When the rolling temperature was reduced to 620–650°C, the final rolling process was carried out, and the thickness of 1.6 mm was attained. Then, the plates were heated to 750°C at a heating rate of 10°C/s, and held for a duration of 7 min. Finally, they were cooled down to 20–25°C at a cooling rate of 10°C/s. The annealed microstructure of the TWIP steel is shown in Fig. 1, and it is evident that some annealing twins are formed inside the equiaxed austenite grains.

Fig. 1.

Microstructural morphology of ultra-high strength TWIP steel (RD: rolling direction; TD: transverse direction; A: austenite; TW: twin). (Online version in color.)

The samples for tensile testing were prepared according to the national standards; GB/T 228.1-201 and GB/T 30069.1-2013. Subsequently, a SANS-CMT5105-type microcomputer-controlled electronic universal testing machine and Zwick HTM16020-type high-speed tensile testing machine were used to study the tensile behavior of TWIP steel under different strain rates, ranging from 10⁻⁴ to 10³ s⁻¹. A transmission electron microscope (TEM, JEOL JEM 2000 FX) was used to characterize the microstructure morphology. TEM specimens of 50 μm thickness were prepared by mechanical grinding and electro-polishing with a twin jet electro-polisher at −20°C using a solution containing 80 vol.-% C₂H₅OH and 20 vol.-% HClO₄. The X-ray diffraction (XRD) was conducted based on the Bragg–Brentano geometry (2θ as 10°–150°) using the Rigaku DMAX-RB with Cu Kα radiation.

3. Results and Discussion

3.1. Mechanical Properties of Ultra-high Strength TWIP Steel

Figure 2(a) present the engineering stress-strain and true stress-strain curves of the TWIP steel at different strain rates. The stress–strain curves under different strain rates are almost coincident with each other during elastic deformation, and the stress increases rapidly with the strain. The yield strength (YS), ultimate tensile strength (UTS), yield ratio, and elongation after fracture of the TWIP steel at the quasi-static tensile process (the strain rate of 10⁻⁴) were 620 MPa, 1039 MPa, 0.6, and 49%, respectively.

Fig. 2.

Mechanical properties of TWIP steel under different strain rates: (a) engineering stress-strain curve and true stress-strain curve (b) mechanical properties (YS: yield strength, UTS: ultimate tensile strength, FE: fracture elongation, UE: uniform elongation). (Online version in color.)

Figure 2(b) shows the plot of ultimate tensile strength (UTS) and yield strength (YS) versus the strain rate. It is worth noting that the increase in both UTS and YS of TWIP steel were relatively low under low strain rates (10⁻⁴–10⁻¹ s⁻¹) and rapidly increased under medium strain rates (10⁻¹–10² s⁻¹). Figure 2(b) shows that the UTS apparently decreased from 1168 MPa to 1141 MPa at the strain rate of 10³ s⁻¹. The uniform elongation (UE) is the percentage plastic elongation at maximum force and fracture elongation (FE) is the percentage total extension at fracture. It can be found that, initially, the UE and FE increased slowly with increasing strain rate, followed by a rapid increase. Thus, in quasi-static deformation, the TWIP steel has negative strain rate sensitivity.²²⁾ Nevertheless, it exhibits positive strain rate sensitivity at an intermediate and higher strain rate.²³⁾

3.2. Strain Hardening Behavior

The strain hardening behavior during the tensile tests can be deduced from the plots of the strain hardening rate (dσ/dε) as the true strain. The differential dσ/dε was obtained by differentiating the true stress–true strain curves in Fig. 3, which indicates that the TWIP steel possesses three distinct strain hardening stages with the increase in true strain.

Fig. 3.

Strain hardening rate of the TWIP steels: (a) 10⁻³–10⁻¹ s⁻¹, (b) 10⁰–10³ s⁻¹. (Online version in color.)

During stage I (ε<0.07), the strain hardening rate of TWIP declines rapidly with the increasing degree of deformation under the conditions of quasi-static tensile deformation or high strain rate deformation. Furthermore, the amount of deformation is very small due to the generation of dislocation slip in the initial stage of plastic deformation with lower dislocation density and fewer interactions of the dislocations between different slip systems.

In stage II (ε=0.07–0.35), the strain hardening rate of TWIP changed slowly with the increase in the degree of deformation within the strain rate of 10⁻⁴–10⁻² s⁻¹, presenting a plateau until the true strain is approximately 0.36. Compared with the strain rate of 10⁻⁴–10⁻² s⁻¹, the strain hardening rate shows a slight upward trend at the strain rate of 10⁻¹ s⁻¹. As the strain rate continued to increase, the strain hardening rate showed an obvious upward trend in the region below 10³ s⁻¹, and then decreased. Literature^7,14,22) indicated that the change in strain hardening rate was related to the initiation of twins and the activation of new twin systems. In the process of the tensile deformation of steel, local stress concentrations will be caused due to the hindrance of grain boundaries, annealing twin boundaries, or precipitates to the moving dislocation. When the local stress reaches the critical twin stress, the nucleation of deformation twins will be induced, and the critical twin stress will decrease with the increase in strain rate. Therefore, with the increase in strain and strain rate, the critical twin stress may decrease, and the number of deformation twins increases. The number of deformation twins increases, the free sliding distance of the movable dislocation decreases, and a large number of dislocations plug between the deformation twins due to the obstruction of the deformation twin boundary, resulting in an increase in the strain hardening rate.

In stage III (ε>0.35), the strain hardening rate of TWIP declines rapidly at different strain rates with the increase in the degree of deformation, which is due to the fact that a large number of twins and dislocations have been produced in stage II when the deformation continues to increase, hindering the formation of new twins with a rapid decline in the strain hardening rate until fracture.

It can also be seen from Fig. 3, at each strain rate, the strain hardening rate curve and true stress-true strain curve of the TWIP steels have an intersection. According to the Considère instability theory, when the strain hardening rate of the material is equal to the true stress in the process of tensile deformation, the plastic instability will occur. Therefore, the intersections are the instability point in the partial enlargement of Figs. 3(a) and 3(b). The instable strain (the true strain corresponding to the instable point) increase with the increase of the strain rate.

3.3. Fracture Features

Figure 4 shows the fractograph of TWIP steel subjected to tensile deformation at different strain rates. From the strain rate of 10⁻⁴ s⁻¹ up to 10³ s⁻¹, the fracture surface of the TWIP steel is distinct ductile zone. A large number of dimples are distributed in the two zones, and the inside of the large dimple also contains a large number of small dimples. This is a typical ductile fracture. The slow strain rate (10⁻⁴ s⁻¹) has led to the formation of very fine micro-dimples and large individual dimples (Fig. 4(a)). As the strain rate increases, the size difference of the dimples decreases gradually, and the dimples get deeper. The particles of (Ti, Nb, V)C were also observed in some of the dimples as shown in Figs. 4(b) and 4(e). According to the analysis of TEM-EDS, the ratio of Nb:V:Ti atoms is about 4:9:10, and the precipitates belongs to NaCl (B1) type atomic structure. The large-size precipitates may be the complex precipitates of Ti, Nb and V precipitated at high temperature with abnormal growth. These indicates that inclusions or precipitates in TWIP steel play an important role in the formation of dimples, and those precipitates were observed all in the samples deformed by different strain rate. For the TWIP steel that deforms at the strain rate of 10² s⁻¹, the fractograph appears to be multilayered. There are “ridges” around the large and deep dimples, marked as “red circles” in Fig. 4(c), which indicates that significant plastic deformation has occurred before the tensile fracture. At the strain rate of 10³ s⁻¹, the obtained fracture surface (Fig. 4(d)) is found to be similar to that of the one obtained at 10² s⁻¹. However, the coarser and deeper dimples in the TWIP steel increased significantly.

Fig. 4.

SEM fractographs of TWIP steel at different strain rates. (Online version in color.)

3.4. Deformation Mechanism at the High Strain Rate

Lee²⁴⁾ studied the plastic deformation behavior of Fe-12Mn-0.6C TWIP steel at the strain rate of 10⁻⁵ s⁻¹–10⁻¹ s⁻¹. The deformed microstructure exhibited the formation of ε-martensite (TRIP effect), and as the strain rate increased, the ε-martensite content decreased. XRD tests were performed on the TWIP steel before and after the tensile deformation under the quasi-static (10⁻⁴ s⁻¹), medium strain rate (10⁰ s⁻¹ and 10² s⁻¹), and high strain rate (10³ s⁻¹) effects, and the results are shown in Fig. 5. No other phases were found in the annealed TWIP steel, which attested that the microstructure of TWIP steel was composed of single austenite. At the strain rate of 10⁻⁴ s⁻¹, there was no phase transition. Combined with the effect of the adiabatic temperature rise under different strain rates, the stacking fault energy of TWIP steel increased by approximately 5 mJ/m²–9 mJ/m² than that under quasi-static conditions. Therefore, in this work, the TWIP steel will not undergo martensitic transformation at the strain rate of 10²–10³ s⁻¹ due to the increase in the stacking fault energy, which is similar to the conclusions in the literature.^25,26,27)

Fig. 5.

XRD of high strength TWIP steel before deformation and under quasi-static, medium, and high strain rates. (Online version in color.)

Figure 6 presents the TEM images of TWIP steel under different strain rates. It can be seen that deformation twins formed in TWIP steels during the process of tensile deformation at different strain rates. Most of the deformation twins appear entangled with dislocations so that some deformation twin boundaries became unclear. In the quasi-static (10⁻³ s⁻¹) tensile deformation condition, the number of deformation twins was small and the film was thick, as shown in Fig. 6(b). Figure 6(b) shows at low strain rate, almost all of the deformation twins in one austenite grain share the same orientation relation (parallel to each other) along with certain crystal orientation. The secondary deformation twin differed in orientation from the initial deformation twin at the strain rate of 10² s⁻¹. A large number of dislocations piled up around the initial deformation twin at high strain rates without sufficient migration time, which led to an increase in localized stress in the vicinity of the initial deformation twin and facilitated the occurrence of the secondary deformation. The secondary deformation twin intersected with the initial deformation twin at an angle, dividing the austenite was into many small quadrilateral regions, as shown in Fig. 6(d). It was equivalent to grain refinement and both the strength and plasticity of TWIP steel increased during the deformation process. With the further increase in strain rate, the number of secondary deformation twins decrease (In Fig. 6(e)). Secondary deformation twins have not been observed and the initial deformation twin dwindled (Fig. 6(f)). The reduction in the occurrence of deformation twins decreases the tensile strength of the TWIP steel, which is consistent with the measured mechanical properties (Fig. 2(c)). However, at this strain rate, the uniform elongation and total elongation of TWIP steel increase, which seems to contradict the decrease in the number of deformation twins, indicating that the increase in plasticity must be caused by factors other than the TWIP effect. The plastic deformation of the material under quasi-static condition can be considered as an isothermal process, while the tensile deformation under dynamic loading can be considered as an adiabatic or quasi-adiabatic process. The complete dissipation of strain energy generated in the deformed sample to the atmosphere takes time, resulting in the rise of local temperature of the material. Benziing²⁸⁾ found that the mechanical properties of TWIP steel are seriously affected by the adiabatic temperature rise (∆T) during the dynamic deformation:   

$$\mathrm{\Delta }T=\frac{h}{\rho C}\int \sigma d\epsilon$$(1)

Where, h=0.9 is the coefficient of transformation of plastic work into heat energy, ρ= 7.8 g/cm³ is the density of steel, C=0.46 kJ/(kg∙K) is the specific heat capacity, σ is the true stress corresponding to the true strain.

Fig. 6.

TEM images of TWIP steel under different strain rates. (a) undeformed, (b) 10⁻³ s⁻¹ (c) 10⁰ s⁻¹ (d) 10² s⁻¹ (e) 5×10² s⁻¹ (f) 10³ s⁻¹. (Online version in color.)

According to the above formula, the ΔT value of the TWIP steel at different strain rates can be calculated, as shown in Table 1. Curtze and Kuokkala²⁹⁾ The stacking fault energies of the steels varied from ~20.5 to 42 mJ·m⁻² at room temperature, causing differences in their mechanical behavior which became even more distinct when the testing conditions, i.e., temperature and/or strain rate, were varied. The SFE of the TWIP steel used in this study is approximately 21.1 mJ/m² before deformation, and the SFE will increase to approximately 26–30 mJ/m² in the dynamic loading process. The change in SFE value will cause the deformation mechanism of the TWIP steel to change.³⁰⁾

Table 1. Adiabatic temperature rise (ΔT, °C) of TWIP steel under different strain rates.

Strain rate, s−1	10−3	10−2	10−1	100	101	102	103
ΔT, °C	114.67	117.69	120.93	126.41	133.43	143.92	148.63

The SFE plays a key role in determining the deformation mechanism. According to author’s previous work,⁷⁾ the value of the SFE can be expressed by the following equation: Г=2ρΔG^γ→ε+2σ^γ/ε, where Г is the SFE of austenite; ρ is the atomic density of the {111} plane in moles per unit area, here $\rho \text{=}\frac{4}{\sqrt{3}}\frac{1}{a^{2}N}$ = 2.944 × 10⁻⁵ mol·m⁻²; a is the lattice constant of austenite and its value is 0.361 nm; N is Avogadro constant); σ^γ/ε is the interfacial energy between γ (austenite) and ε (martensite) phases (here the value is 9 mJ/m²); ΔG^γ→ε relies on both chemical composition and temperature, and its value can be calculated with a regular solution model as follows:.   

$$\begin{array}{l}\mathrm{\Delta }{G}^{\gamma \to \epsilon }=x_{Fe}\mathrm{\Delta }{G}_{Fe}^{\gamma \to \epsilon }+x_{Mn}\mathrm{\Delta }{G}_{Mn}^{\gamma \to \epsilon }\\ \hspace{1em}\hspace{1em}\hspace{1em}+x_C\mathrm{\Delta }{G}_C^{\gamma \to \epsilon }+x_{Fe}{x}_{Mn}\mathrm{\Delta }{\mathrm{\Omega }}_{FeMn}^{\gamma \to \epsilon }\\ \hspace{1em}\hspace{1em}\hspace{1em}+x_{Fe}{x}_{\text{C}}\mathrm{\Delta }{\mathrm{\Omega }}_{FeC}^{\gamma \to \epsilon }+x_{Mn}{x}_C\mathrm{\Delta }{\mathrm{\Omega }}_{MnC}^{\gamma \to \epsilon }\end{array}$$(2)

where $\mathrm{\Delta }{G}_{Fe}^{\gamma \to \epsilon }$, $\mathrm{\Delta }{G}_{Mn}^{\gamma \to \epsilon }$, $\mathrm{\Delta }{G}_C^{\gamma \to \epsilon }$ are the molar free energy changes of Fe, Mn, C between γ and ε phases, respectively. x_Fe, x_Mn and x_C are atomic fractions of Fe, Mn, and C, respectively. $\mathrm{\Delta }{\mathrm{\Omega }}_{FeMn}^{\gamma \to \epsilon }$, $\mathrm{\Delta }{\mathrm{\Omega }}_{FeC}^{\gamma \to \epsilon }$, $\mathrm{\Delta }{\mathrm{\Omega }}_{MnC}^{\gamma \to \epsilon }$ are the interaction parameter differences between γ and ε phases. The values of parameters are referred in paper.⁷⁾

Therefore, the deformation mechanism of TWIP steel at different strain rates can be interpreted and summarized as schematically illustrated in Fig. 7. The existence of deformation twins not only reduces the size of austenite grain and the average free path of dislocation to a great extent, but also increases the resistance of dislocation across the twin boundary, inhibiting the annihilation of dislocation, and improving the work hardening rate. However, the original austenite grains can be divided into lamellar regions by the deformation twins, which can effectively limit the free movement of dislocations like grain boundaries, thus realizing the strengthening effect, which is similar to the grain refinement called “dynamic Hall-Petch effect”.³¹⁾ It can be seen that the plastic deformation of TWIP steel is a coexistence process of the deformation twin mechanism and dislocation slip mechanism. However, a large number of dislocation plugs accumulated around the deformation twins at high strain rate and there was insufficient time to transfer and release the stress, which leads to an increase in local stress near the deformation twins, and the local stress can only be released by the formation of secondary twins. When the strain rate increases further, the adiabatic temperature rise caused by the high-speed strain can be higher than 140°C (as shown in Table 1), which increases the mobility of the dislocation and inhibits the occurrence of secondary twinning. In addition, the lamination energy of TWIP steel increased to approximately 30 mJ/m² due to the adiabatic temperature rise, which increased the difficulty of formation of deformation twins. Therefore, the tensile strength decreased and the elongation continued to increase due to the softening effect of the adiabatic temperature rise.

Fig. 7.

Effect of strain rate on the mechanical properties and stacking fault energy of TWIP steel (UTS: ultimate tensile strength; FE: fracture elongation; UE: uniform elongation; SFE: stacking fault energy; (dσ/dε)av: the average value of dσ/dε at true strain of 0.1–0.35). (Online version in color.)

4. Conclusions

In summary, in order to better characterize the mechanical properties of cars during collision, the deformation mechanism of Fe-20Mn-0.6C TWIP steels under high strain rates have been systematically investigated. Several important conclusions have been drawn, as shown below:

(1) Both UTS and YS of TWIP steel is relatively low under low strain rates (10⁻⁴–10⁻¹ s⁻¹) and increased rapidly under medium strain rates (10⁻¹–10² s⁻¹). However, the UTS decreased from 1168 MPa to 1141 MPa. when the strain rate was increased from 10² to 10³ s⁻¹. Initially, the UE and FE increased slowly with increasing strain rate, followed by a rapid increase.

(2) TWIP steel possesses three distinct strain hardening stages when the true strain increases. Only in stage II, dσ/dε plateaus off at strain rates of 10⁻⁴ to 10⁻¹ s⁻¹, while in the other stages, dσ/dε decreases continuously with increasing the strain rate.

(3) The deformation mechanisms of TWIP steel under high strain rates was one in which the deformation twin and dislocation slip promoted and restricted each other. When the strain rate is higher than 10² s⁻¹, the adiabatic heating temperature could rise above 143°C, thereby suppressing secondary twinning and enhancing the softening effect.

Acknowledgments

Funded by the National Natural Science Foundation of China (No. U1860112). State Key Laboratory of Marine Equipment made of Metal Material and Application (No. SKLMEA-USTL-201907). The Key Project of Liaoning Education Department (No. 2019FWDF03).

References

• 1)   P.  Matteis,  G.  Scavino,  F.  D’Aiuto and  D.  Firrao: J. Iron Steel Res. Int., 83 (2012), 950.
• 2)   Y. Z.  Tian,  Y.  Bai,  L. J.  Zhao,  S.  Gao,  H. K.  Yang,  A.  Shibata,  Z. F.  Zhang and  N.  Tsuji: Mater. Charact., 126 (2017), 74.
• 3)   H. L.  Yi,  L.  Sun and  X. C.  Xiong: Mater. Sci. Technol., 34 (2018), 1112.
• 4)   J. K.  Kim,  Y.  Estrin and  B. C.  De Cooman: Metall. Mater. Trans. A, 49 (2018), 5919.
• 5)   M.  Madivala,  A.  Schwedt,  U.  Prahl and  W.  Bleck: Int. J. Plast., 115 (2019), 178.
• 6)   D.  Barbier,  N.  Gey,  S.  Allain,  N.  Bozzolo and  M.  Humbert: Mater. Sci. Eng. A, 500 (2009), 196.
• 7)   M.  Xu,  Z. L.  Mi,  H.  Li,  D.  Tang and  H. T.  Jiang: Mater. Sci. Technol., 34 (2018), 242.
• 8)   H.  Cai,  R.  Song,  L.  Jiang,  R.  Wei,  Y.  Li and  W.  Niu: Mech. Mater., 122 (2018), 104.
• 9)   R.  Bobbili and  V.  Madhu: Mater. Lett., 218 (2018), 103.
• 10)   J. C.  Álvarez Hostos,  A. D.  Bencomo,  E. S.  Puchi Cabrera,  J. D.  Guérin and  L.  Dubar: Int. J. Plast., 103 (2018), 119.
• 11)   Q. F.  Dai,  R. B.  Song,  H. J.  Cai,  S. C.  Yu and  Z.  Gao: Chin. J. Mater. Res., 27 (2013), 25 (in Chinese).
• 12)   H. W.  Yen,  M. X.  Huang,  C. P.  Scott and  J. R.  Yang: Scr. Mater., 66 (2012), 1018.
• 13)   B. C.  De Cooman,  Y.  Estrin and  S. K.  Kim: Acta Mater., 142 (2018), 283.
• 14)   O.  Grässel and  G.  Frommeyer: Mater. Sci. Technol., 14 (1998), 1213.
• 15)   S.  Lee,  Y.  Estrin and  B. C.  De Cooman: Metall. Mater. Trans. A, 45 (2014), 717.
• 16)   J. K.  Kim,  L.  Chen,  H. S.  Kim,  S. K.  Kim,  Y.  Estrin and  B. C.  De Cooman: Metall. Mater. Trans. A, 40 (2009), 3147.
• 17)   O.  Bouaziz,  S.  Allain,  C. P.  Scott,  P.  Cugy and  D.  Barbier: Curr. Opin. Solid State Mater. Sci., 15 (2011), 141.
• 18)   O.  Grässel,  L.  Krüger,  G.  Frommeyer and  L. W.  Meyer: Int. J. Plast., 16 (2000), 1391.
• 19)   G.  Frommeyer,  U.  Brüx and  P.  Neumann: ISIJ Int., 43 (2003), 438.
• 20)   S. Q.  Xu,  D.  Ruan,  J. H.  Beynon and  Y. H.  Rong: Mater. Sci. Eng. A, 573 (2013), 132.
• 21)   S.  Curtze and  V. T.  Kuokkala: Acta Mater., 58 (2010), 5129.
• 22)   X. K.  Ma,  F. G.  Li,  J.  Cao,  J. H.  Li,  Z. K.  Sun,  G.  Zhu and  S. S.  Zhou: Mater. Sci. Eng. A, 710 (2018), 1.
• 23)   G. T.  Gray III: Annu. Rev. Mater. Res., 42 (2012), 285.
• 24)   S.  Lee,  Y.  Estrin and  B. C.  De Cooman: Metall. Mater. Trans. A, 45 (2014), 717.
• 25)   D. T.  Pierce,  J. A.  Jiménez,  J.  Bentley,  D.  Raabe and  J. E.  Wittig: Acta Mater., 100 (2015), 178.
• 26)   J. K.  Kim and  B. C.  De Cooman: Mater. Sci. Eng. A, 676 (2016), 216.
• 27)   H.  Wu,  G.  Niu,  J.  Cao and  M.  Yang: Mater. Sci. Technol., 33 (2017), 480.
• 28)   J. T.  Benzing,  W. A.  Poling,  D. T.  Pierce,  J.  Bentley,  K. O.  Findley,  D.  Raabe and  J. E.  Wittig: Mater. Sci. Eng. A, 711 (2018), 78.
• 29)   S.  Curtze and  V. T.  Kuokkala: Acta Mater., 58 (2010), 5129.
• 30)   J.  Chen,  J. J.  Wang,  G.  Yuan and  C. M.  Liu: Mater. Sci. Eng. A, 759 (2019), 47.
• 31)   O.  Bouaziz: Scr. Mater., 66 (2012), 982.