Micro Plastic Deformation Behavior of Carbide-Free Bainitic Steel during Tensile Deformation Process

The microstructure of carbide-free bainitic steel has been widely reported because of its excellent mechanical properties. However, no work has explored the deformation behavior of each phase in the microstructure, namely, retained austenite (RA) and bainitic ferrite (BF). In this paper, the deformation behavior of RA and BF during tension was investigated through the in situ electron backscatter diffusion method. The results showed that the RA surrounded by BF with a high Schmid factor (SF) was accelerated towards being transformed into martensite, but the surrounding BF with low SF value retarded the transformation. The kernel average misorientation (KAM) findings revealed that the strain of the RA and BF rapidly increased initially, slowed down after the critical points of 0.25 and 0.54, respectively, and then increased again. The calculated result for the KAM fraction indicated that the contribution of RA to the total plastic deformation gradually decreased throughout the total tensile process, whereas that of BF increased in the presence of RA.


Introduction
Carbide-free bainitic steel is mainly composed of bainitic ferrite (BF) and carbon-rich retained austenite (RA) [1,2]. RA is distributed between BF lathes and/or sheaves, possessing different stabilities, and the two phases coordinate with each other during deformation, thereby guaranteeing an excellent balance of strength and toughness, especially when the coarse, blocky RA is eliminated. In bainitic steels, RA is a metastable phase that can transform into martensite because of stress concentration [3][4][5][6]. The stability of RA is influenced by several factors, such as morphology, grain size, carbon content, distribution, and crystalline orientation [7,8]. Knijf et al. modelled the martensite transformation process of 11 kinds of RA with different orientations during uniaxial tensile testing [8], and revealed that transformation potential varies linearly with transformation rate, yielding a correlation coefficient of approximately 0.75. However, the effect of the crystal orientation of the surrounding phase on the transformation behavior of RA was not considered. The evolution of the soft RA in bainitic steel is the main focus of researchers, with minimal concern for harder phases, such as BF or martensite [3,7,8].
In dual-phase steel, deformation is mainly borne by the soft phase. For carbide-free bainitic steel, RA is the soft phase, but this phase only comprises a small portion of the microstructure. Therefore, deformation should mainly occur in the harder phase BF. However, the distribution of strain and stress between BF and RA in bainitic microstructure during deformation has not been studied in detail.

Experimental Methods
The chemical composition of the carbide-free bainitic steel is listed in Table 1. The studied steel was smelted in a 45 kg vacuum induction furnace and forged into a round bar (Φ 65 mm). The martensite start transformation temperature (M s ) of the specimen was 315 • C. Specimens were austenitized at 930 • C for 45 min, subjected to isothermal transformation at 350 • C for 2 h to obtain the carbide-free bainitic microstructure, and then air cooled to room temperature. Finally, all specimens were tempered at 320 • C for 1 h. The specimens were cut into plates with a thickness of 2 mm for tensile testing; the specimen size is shown in Figure 1. To analyze the same area under different strains, we limited the examined deformed area to a parallel section of 0.5 mm length in the middle of the specimens. The tensile strain of the specimen was determined by Equation (1): where ε is the strain of the specimen, g is the gauge of the extensometer, l is the undeformed length of the analyzed area, and d is the displacement of the extensometer. The tensile rate is 2 mm·min −1 . The accumulative strain of each step was 0.025, 0.054, and 0.104, respectively. Figure 2 shows the continuous tensile curve of the specimen. An ion figuring system was used to produce a high-quality surface of the specimen for EBSD observation. The mechanically polished sample was jetted by an ion beam for 15 min. The real deformed area was analyzed using EBSD, attached to a scanning electron microscope. EBSD data were acquired on a hexagonal scan grid under an accelerating voltage of 30 kV, tilt angle of 70°, and step size of 80 nm. Raw orientation data obtained at each loading step were post-processed using the OIM Analysis 7.3 software. Orientation, strain, and stress were diverse at different grains and complicated the research. Therefore, the analyzed region was limited to one austenite grain, as indicated by the encircled red polygon of Figure 3. The volume fraction of the retained austenite in specimens with different strain was analyzed by a D/max-2500/PC X-ray diffractometer (XRD, Rigaku, Tokyo, Japan) with CuKα radiation.  The tensile rate is 2 mm·min −1 . The accumulative strain of each step was 0.025, 0.054, and 0.104, respectively. Figure 2 shows the continuous tensile curve of the specimen. The tensile rate is 2 mm·min −1 . The accumulative strain of each step was 0.025, 0.054, and 0.104, respectively. Figure 2 shows the continuous tensile curve of the specimen. An ion figuring system was used to produce a high-quality surface of the specimen for EBSD observation. The mechanically polished sample was jetted by an ion beam for 15 min. The real deformed area was analyzed using EBSD, attached to a scanning electron microscope. EBSD data were acquired on a hexagonal scan grid under an accelerating voltage of 30 kV, tilt angle of 70°, and step size of 80 nm. Raw orientation data obtained at each loading step were post-processed using the OIM Analysis 7.3 software. Orientation, strain, and stress were diverse at different grains and complicated the research. Therefore, the analyzed region was limited to one austenite grain, as indicated by the encircled red polygon of Figure 3. The volume fraction of the retained austenite in specimens with different strain was analyzed by a D/max-2500/PC X-ray diffractometer (XRD, Rigaku, Tokyo, Japan) with CuKα radiation.  An ion figuring system was used to produce a high-quality surface of the specimen for EBSD observation. The mechanically polished sample was jetted by an ion beam for 15 min. The real deformed area was analyzed using EBSD, attached to a scanning electron microscope. EBSD data were acquired on a hexagonal scan grid under an accelerating voltage of 30 kV, tilt angle of 70 • , and step size of 80 nm. Raw orientation data obtained at each loading step were post-processed using the OIM Analysis 7.3 software. Orientation, strain, and stress were diverse at different grains and complicated the research. Therefore, the analyzed region was limited to one austenite grain, as indicated by the encircled red polygon of Figure 3. The volume fraction of the retained austenite in specimens with different strain was analyzed by a D/max-2500/PC X-ray diffractometer (XRD, Rigaku, Tokyo, Japan) with CuKα radiation. The tensile rate is 2 mm·min −1 . The accumulative strain of each step was 0.025, 0.054, and 0.104, respectively. Figure 2 shows the continuous tensile curve of the specimen. An ion figuring system was used to produce a high-quality surface of the specimen for EBSD observation. The mechanically polished sample was jetted by an ion beam for 15 min. The real deformed area was analyzed using EBSD, attached to a scanning electron microscope. EBSD data were acquired on a hexagonal scan grid under an accelerating voltage of 30 kV, tilt angle of 70°, and step size of 80 nm. Raw orientation data obtained at each loading step were post-processed using the OIM Analysis 7.3 software. Orientation, strain, and stress were diverse at different grains and complicated the research. Therefore, the analyzed region was limited to one austenite grain, as indicated by the encircled red polygon of Figure 3. The volume fraction of the retained austenite in specimens with different strain was analyzed by a D/max-2500/PC X-ray diffractometer (XRD, Rigaku, Tokyo, Japan) with CuKα radiation.    Figure 4 shows the phase maps of the specimen at different strains. Two types of RA, namely, film-like and block-like, exist within the microstructure. These types possess relatively high and low stabilities, respectively. The amount of RA gradually decreased with increasing strain. Statistical results in Figure 5a reveal two stages for the variation in volume fraction of RA. For stage I, a sharp decrease from 7.5% to 3.0%, with increasing strain from 0 to 0.054, was noted. For stage II, a slow decline from 3.0% to 2.2%, with further increasing strain to 0.104, was observed. The lost RA was transformed into martensite through the strain/stress-induced transformation process [18,19]. Therefore, the mechanical stability of RA is the factor that chiefly affects the variation tendency shown in Figure 5a. In general, the mechanical stability of RA is influenced by carbon content, shape, and size, and the condition of the surrounded phase [7]. The variation of mean volume fraction of the retained austenite in the microstructure, with increasing strain, was estimated via XRD test, as shown in Figure 5b; which shows a similar variation tendency to the tested region. The higher value of the XRD result, as compared with the statistical result from the phase map, is mainly due to the resolution limitations of EBSD, resulting in filmy retained austenite with ultrafine thickness which cannot be revealed in the EBSD phase map.

Strain-Induced Martensitic Transformation of RA
Materials 2020, 13, x 4 of 11 Figure 4 shows the phase maps of the specimen at different strains. Two types of RA, namely, film-like and block-like, exist within the microstructure. These types possess relatively high and low stabilities, respectively. The amount of RA gradually decreased with increasing strain. Statistical results in Figure 5a reveal two stages for the variation in volume fraction of RA. For stage I, a sharp decrease from 7.5% to 3.0%, with increasing strain from 0 to 0.054, was noted. For stage II, a slow decline from 3.0% to 2.2%, with further increasing strain to 0.104, was observed. The lost RA was transformed into martensite through the strain/stress-induced transformation process [18,19]. Therefore, the mechanical stability of RA is the factor that chiefly affects the variation tendency shown in Figure 5a. In general, the mechanical stability of RA is influenced by carbon content, shape, and size, and the condition of the surrounded phase [7]. The variation of mean volume fraction of the retained austenite in the microstructure, with increasing strain, was estimated via XRD test, as shown in Figure 5b; which shows a similar variation tendency to the tested region. The higher value of the XRD result, as compared with the statistical result from the phase map, is mainly due to the resolution limitations of EBSD, resulting in filmy retained austenite with ultrafine thickness which cannot be revealed in the EBSD phase map.    Figure 4 shows the phase maps of the specimen at different strains. Two types of RA, namely, film-like and block-like, exist within the microstructure. These types possess relatively high and low stabilities, respectively. The amount of RA gradually decreased with increasing strain. Statistical results in Figure 5a reveal two stages for the variation in volume fraction of RA. For stage I, a sharp decrease from 7.5% to 3.0%, with increasing strain from 0 to 0.054, was noted. For stage II, a slow decline from 3.0% to 2.2%, with further increasing strain to 0.104, was observed. The lost RA was transformed into martensite through the strain/stress-induced transformation process [18,19]. Therefore, the mechanical stability of RA is the factor that chiefly affects the variation tendency shown in Figure 5a. In general, the mechanical stability of RA is influenced by carbon content, shape, and size, and the condition of the surrounded phase [7]. The variation of mean volume fraction of the retained austenite in the microstructure, with increasing strain, was estimated via XRD test, as shown in Figure 5b; which shows a similar variation tendency to the tested region. The higher value of the XRD result, as compared with the statistical result from the phase map, is mainly due to the resolution limitations of EBSD, resulting in filmy retained austenite with ultrafine thickness which cannot be revealed in the EBSD phase map.   The Schmid factor (SF) can be used to reflect the potential to activate the slip system in a crystalline. A higher SF value indicates greater ease in activating the slip system than a lower SF value [20]. Figure 6 shows the distribution of SF, where the slip systems of RA were selected as (111)<110>, and those of BF were (110)<111>, (112)<111>, and (123)<111>. The phase in a black irregular closed curve is RA, and the change in SF values are represented by the color gradient.

Strain-Induced Martensitic Transformation of RA
Materials 2020, 13, x 5 of 11 The Schmid factor (SF) can be used to reflect the potential to activate the slip system in a crystalline. A higher SF value indicates greater ease in activating the slip system than a lower SF value [20]. Figure 6 shows the distribution of SF, where the slip systems of RA were selected as (111)<110>, and those of BF were (110)<111>, (112)<111>, and (123)<111>. The phase in a black irregular closed curve is RA, and the change in SF values are represented by the color gradient. The SF factor of RA was uniform and high, with a value of about 0.5, which indicated that RA deformation was commenced easily and simultaneously. However, the SF value of BF ranged from 0.2 to 0.5 across different BF plates. In area A, the RA was surrounded by the BF with a relatively high SF, indicating that both RA and BF deformed easily during tension. Stress and strain were easily transferred from BF to RA. Certain studies have also revealed that dislocation within BF can be absorbed by RA [19,21], which not only reserved the soft state of BF, but also further hardened RA. Therefore, the dislocation density within RA easily increased sufficiently to allow martensitic transformation. Then, RA in area A rapidly disappeared when the true strain reached a low value of 0.025 (Figure 6a,b). The condition of RA in area B markedly differed from that in area A. Although RA exhibited a high SF value, that of the surrounding BF was low, at ~0.2. Moreover, the SF value of the surrounding BF in this region maintained a low level during the total tensile process, although a small increase was noted with increasing strain, because of coordination of the neighboring phase. This finding indicated that the BF remained at a hard orientation for dislocation slip, and that increased stress was needed to deform the BF plates. Thus, the stress and strain become difficult to transfer from BF into RA, because RA is shielded by the surrounding hard BF. The RA in area B also became difficult to transform into martensite (Figure 4a-d), and some of the RA was retained when the true strain reached 0.104. Therefore, we conclude that the martensitic transformation of RA was affected by the orientation of the surrounding BF. The low SF value of the surrounding BF retards, whereas the high SF value accelerated, the martensitic transformation.

Microscale Deformation Behavior in RA and BF
The deformation behaviors of RA and BF were dissimilar because of their different properties. The strain at the RA and BF can be well characterized by kernel average misorientation (KAM) [22][23][24][25]. A high KAM value represents a large deformation. KAM is the average value of misorientation of measured points and all their neighbors (Figure 7). Normally, the neighbor points within the third The SF factor of RA was uniform and high, with a value of about 0.5, which indicated that RA deformation was commenced easily and simultaneously. However, the SF value of BF ranged from 0.2 to 0.5 across different BF plates. In area A, the RA was surrounded by the BF with a relatively high SF, indicating that both RA and BF deformed easily during tension. Stress and strain were easily transferred from BF to RA. Certain studies have also revealed that dislocation within BF can be absorbed by RA [19,21], which not only reserved the soft state of BF, but also further hardened RA. Therefore, the dislocation density within RA easily increased sufficiently to allow martensitic transformation. Then, RA in area A rapidly disappeared when the true strain reached a low value of 0.025 (Figure 6a,b). The condition of RA in area B markedly differed from that in area A. Although RA exhibited a high SF value, that of the surrounding BF was low, at~0.2. Moreover, the SF value of the surrounding BF in this region maintained a low level during the total tensile process, although a small increase was noted with increasing strain, because of coordination of the neighboring phase. This finding indicated that the BF remained at a hard orientation for dislocation slip, and that increased stress was needed to deform the BF plates. Thus, the stress and strain become difficult to transfer from BF into RA, because RA is shielded by the surrounding hard BF. The RA in area B also became difficult to transform into martensite (Figure 4a-d), and some of the RA was retained when the true strain reached 0.104. Therefore, we conclude that the martensitic transformation of RA was affected by the orientation of the surrounding BF. The low SF value of the surrounding BF retards, whereas the high SF value accelerated, the martensitic transformation.

Microscale Deformation Behavior in RA and BF
The deformation behaviors of RA and BF were dissimilar because of their different properties. The strain at the RA and BF can be well characterized by kernel average misorientation (KAM) [22][23][24][25]. A high KAM value represents a large deformation. KAM is the average value of misorientation of Materials 2020, 13, 4703 6 of 10 measured points and all their neighbors (Figure 7). Normally, the neighbor points within the third adjacent point are included into the calculation process, and the upper tolerance value of misorientation is set as 5 • to eliminate the influence of grain boundaries and subgrain boundaries [25][26][27][28][29].
Materials 2020, 13, x 6 of 11 adjacent point are included into the calculation process, and the upper tolerance value of misorientation is set as 5° to eliminate the influence of grain boundaries and subgrain boundaries [25][26][27][28][29]. The KAM distributions for the RA and BF are shown in Figures 8 and 9, respectively. The various KAM values are represented by a color gradient. In Figures 8a and 9a, the KAM values at the phase boundaries are higher than the value interior of the plate. These findings indicate the existence of a strain field around the phase boundaries prior to the tensile process; and this strain field is introduced by the shear transformation and volume expansion of bainite phase transformation [30]. With increasing strain, the KAM values in the inner phase of RA and BF gradually increased. The KAM value was relatively uniform in RA, as shown in the rectangular area in Figure 8. By comparison, the KAM value in BF was asymmetrical, because of the presence several high-KAM-value regions. The high KAM value regions indicated a high microstrain at these regions. The number of high-strain regions increased with increasing macrostrain, as shown in the red ellipses in Figure 9. Previous studies revealed the presence of numerous C-clusters in BF, because of the low solubility of the C atom in body-centered cubic (bcc)-structured BF [31][32][33]. These C-clusters pin the dislocations, and hinder lippage, resulting in the strain concentration around the C-clusters, and the formation of the high-KAM-value regions.  The KAM distributions for the RA and BF are shown in Figures 8 and 9, respectively. The various KAM values are represented by a color gradient. In Figures 8a and 9a, the KAM values at the phase boundaries are higher than the value interior of the plate. These findings indicate the existence of a strain field around the phase boundaries prior to the tensile process; and this strain field is introduced by the shear transformation and volume expansion of bainite phase transformation [30]. With increasing strain, the KAM values in the inner phase of RA and BF gradually increased. The KAM value was relatively uniform in RA, as shown in the rectangular area in Figure 8. By comparison, the KAM value in BF was asymmetrical, because of the presence several high-KAM-value regions. The high KAM value regions indicated a high microstrain at these regions. The number of high-strain regions increased with increasing macrostrain, as shown in the red ellipses in Figure 9. Previous studies revealed the presence of numerous C-clusters in BF, because of the low solubility of the C atom in body-centered cubic (bcc)-structured BF [31][32][33]. These C-clusters pin the dislocations, and hinder lippage, resulting in the strain concentration around the C-clusters, and the formation of the high-KAM-value regions.
Materials 2020, 13, x 6 of 11 adjacent point are included into the calculation process, and the upper tolerance value of misorientation is set as 5° to eliminate the influence of grain boundaries and subgrain boundaries [25][26][27][28][29]. The KAM distributions for the RA and BF are shown in Figures 8 and 9, respectively. The various KAM values are represented by a color gradient. In Figures 8a and 9a, the KAM values at the phase boundaries are higher than the value interior of the plate. These findings indicate the existence of a strain field around the phase boundaries prior to the tensile process; and this strain field is introduced by the shear transformation and volume expansion of bainite phase transformation [30]. With increasing strain, the KAM values in the inner phase of RA and BF gradually increased. The KAM value was relatively uniform in RA, as shown in the rectangular area in Figure 8. By comparison, the KAM value in BF was asymmetrical, because of the presence several high-KAM-value regions. The high KAM value regions indicated a high microstrain at these regions. The number of high-strain regions increased with increasing macrostrain, as shown in the red ellipses in Figure 9. Previous studies revealed the presence of numerous C-clusters in BF, because of the low solubility of the C atom in body-centered cubic (bcc)-structured BF [31][32][33]. These C-clusters pin the dislocations, and hinder lippage, resulting in the strain concentration around the C-clusters, and the formation of the high-KAM-value regions.   Figure 10 displays the variations in KAM value and GND density with increasing true strain. The KAM value of the BF quickly increased from 0.84° to 1.20° during early tensile process; the increase rate of the KAM value notably decreased. Correspondingly, the GND density in BF increased rapidly from 1.15 × 10 14 to 1.62 × 10 14 , with increasing strain to 0.025, and then the increasing tendency of GND density slowed down. This pattern indicates that large deformation occurred in the BF at the early stage of tensile deformation, and increased dislocation density and hardness in BF. The mobility of dislocation in the bcc-structured BF was harder than that in the face-centered cubic (fcc)-structured RA. Unlike that of BF, the KAM value of RA indicated a constantly increasing rate within the true strain range of 0-0.054; beyond this range, the KAM value remained relative stable (Figure 10a). The variation tendency of GND density in the RA was similar to that of the KAM value, within 0-0.054. The increase rate of GND density markedly decreased when the true strain exceeded 0.054 ( Figure  10b). Fcc-structured RA contains more slip systems than bcc-structured BF, resulting in a strong deformability of RA. Moreover, RA possesses a stronger storage capacity for GNDs compared with  The KAM value of the BF quickly increased from 0.84 • to 1.20 • during early tensile process; the increase rate of the KAM value notably decreased. Correspondingly, the GND density in BF increased rapidly from 1.15 × 10 14 to 1.62 × 10 14 , with increasing strain to 0.025, and then the increasing tendency of GND density slowed down. This pattern indicates that large deformation occurred in the BF at the early stage of tensile deformation, and increased dislocation density and hardness in BF. The mobility of dislocation in the bcc-structured BF was harder than that in the face-centered cubic (fcc)-structured RA. Unlike that of BF, the KAM value of RA indicated a constantly increasing rate within the true strain range of 0-0.054; beyond this range, the KAM value remained relative stable (Figure 10a). The variation tendency of GND density in the RA was similar to that of the KAM value, within 0-0.054. The increase rate of GND density markedly decreased when the true strain exceeded 0.054 (Figure 10b).  Figure 10 displays the variations in KAM value and GND density with increasing true strain. The KAM value of the BF quickly increased from 0.84° to 1.20° during early tensile process; the increase rate of the KAM value notably decreased. Correspondingly, the GND density in BF increased rapidly from 1.15 × 10 14 to 1.62 × 10 14 , with increasing strain to 0.025, and then the increasing tendency of GND density slowed down. This pattern indicates that large deformation occurred in the BF at the early stage of tensile deformation, and increased dislocation density and hardness in BF. The mobility of dislocation in the bcc-structured BF was harder than that in the face-centered cubic (fcc)-structured RA. Unlike that of BF, the KAM value of RA indicated a constantly increasing rate within the true strain range of 0-0.054; beyond this range, the KAM value remained relative stable (Figure 10a). The variation tendency of GND density in the RA was similar to that of the KAM value, within 0-0.054. The increase rate of GND density markedly decreased when the true strain exceeded 0.054 ( Figure  10b). Fcc-structured RA contains more slip systems than bcc-structured BF, resulting in a strong deformability of RA. Moreover, RA possesses a stronger storage capacity for GNDs compared with BF. On the one hand, RA generates numerous GNDs because of its own deformation. On the other hand, RA absorbs the GNDs from the adjacent BF [19,21]. These two reasons result in a rapid increase of KAM value during 0-0.054. The notable hardening behavior of RA reduces further deformability. The absorption of dislocation by RA also decreases the increase rate of dislocation density in BF, indirectly reducing the increase rate of the KAM value. Figure 5 shows the rapid decrease in RA at the early stage of deformation, owing to transformation into martensite. RA, which presents a high dislocation density, should be initially transformed into martensite during deformation. This effect indirectly decreases the increase rate of GND density in RA and is responsible for the lower increase rate of GND density in RA than in BF at the early stage of deformation.
The microstrain in each phase during deformation is inevitably heterogeneous, owing to different features of the phases. To analyze the variation tendency of the strain fraction for each phase, we calculated the KAM fraction of two phases through Equation (2): where KAM RA is the KAM value of RA, and KAM BF is the KAM value of BF. The variation tendencies of the KAM fraction in the two phases are shown in Figure 10. Notably, the KAM fraction cannot represent the absolute strain fraction of materials, but its change reveals the development of strain fraction. Figure 11 shows the variation of KAM fraction for each phase with increasing true strain. The KAM fraction in RA was higher than that in the BF at the initial stage. With increasing true strain, the KAM fraction of the RA decreased, whereas the KAM fraction of the BF increased. The decreased KAM fraction of RA is due to two causes. First, the high dislocation density resulting from the deformation of itself, and the absorption from the neighboring BF, reduces the deformability of RA. Second, martensite transformation also strengthens RA, and therefore reduces the latter's deformability. BF is the major phase in the microstructure, and inevitably absorbs additional strain from the specimen. Furthermore, the RA continually absorbs the dislocations of the adjacent BF and coordinates the deformation of the adjacent BF. These phenomena result in the increased KAM fraction of BF. The results also illustrate that the RA markedly contributes to the total deformation at the early deformation stage; however, this contribution weakens with increasing strain. BF, the major phase, increasingly contributes to intensifying strain under RA coordination.
BF. On the one hand, RA generates numerous GNDs because of its own deformation. On the other hand, RA absorbs the GNDs from the adjacent BF [19,21]. These two reasons result in a rapid increase of KAM value during 0-0.054. The notable hardening behavior of RA reduces further deformability. The absorption of dislocation by RA also decreases the increase rate of dislocation density in BF, indirectly reducing the increase rate of the KAM value. Figure 5 shows the rapid decrease in RA at the early stage of deformation, owing to transformation into martensite. RA, which presents a high dislocation density, should be initially transformed into martensite during deformation. This effect indirectly decreases the increase rate of GND density in RA and is responsible for the lower increase rate of GND density in RA than in BF at the early stage of deformation.
The microstrain in each phase during deformation is inevitably heterogeneous, owing to different features of the phases. To analyze the variation tendency of the strain fraction for each phase, we calculated the KAM fraction of two phases through Equation (2): where KAMRA is the KAM value of RA, and KAMBF is the KAM value of BF. The variation tendencies of the KAM fraction in the two phases are shown in Figure 10. Notably, the KAM fraction cannot represent the absolute strain fraction of materials, but its change reveals the development of strain fraction. Figure 11 shows the variation of KAM fraction for each phase with increasing true strain. The KAM fraction in RA was higher than that in the BF at the initial stage. With increasing true strain, the KAM fraction of the RA decreased, whereas the KAM fraction of the BF increased. The decreased KAM fraction of RA is due to two causes. First, the high dislocation density resulting from the deformation of itself, and the absorption from the neighboring BF, reduces the deformability of RA. Second, martensite transformation also strengthens RA, and therefore reduces the latter's deformability. BF is the major phase in the microstructure, and inevitably absorbs additional strain from the specimen. Furthermore, the RA continually absorbs the dislocations of the adjacent BF and coordinates the deformation of the adjacent BF. These phenomena result in the increased KAM fraction of BF. The results also illustrate that the RA markedly contributes to the total deformation at the early deformation stage; however, this contribution weakens with increasing strain. BF, the major phase, increasingly contributes to intensifying strain under RA coordination.

Conclusions
In-situ EBSD was used to study the microscopic plastic deformation of RA and BF in medium carbon carbide-free bainitic steel. The martensitic transformation of RA and strain distribution between two phases were analyzed. The following conclusions can be drawn:

Conclusions
In-situ EBSD was used to study the microscopic plastic deformation of RA and BF in medium carbon carbide-free bainitic steel. The martensitic transformation of RA and strain distribution between two phases were analyzed. The following conclusions can be drawn:

1.
The transformation from RA into martensite occurs more quickly at the early stage (0-0.054) of deformation than at the later stage (0.054-0.104). The martensitic transformation of RA is notably affected by the orientation of the surrounding BF. The low SF value of the surrounding BF retarded martensitic transformation, whereas high SF value accelerated it. 2.
The KAM value of RA was more symmetrical than that of the BF during tension. The true strain of 0.025 was the critical point for BF, because the KAM value and GND density increased rapidly before 0.025 and slowed down after 0.025. The variation tendency of KAM in RA was similar to that in BF, but the critical point was 0.054.

3.
With increasing true strain, the KAM fraction of the RA decreased, whereas the KAM fraction of BF increased. This result reveals that the RA provided a higher contribution than BF at the early stage of tension to the whole deformation of the specimen, and that the contribution gradually decreased. As the major phase in the microstructure, BF contributed increasingly with intensifying strain under coordination with RA.

Conflicts of Interest:
The authors declare no conflict of interest.