In Situ Observation on the Effects of Prior Martensite Formation on Nanostructured Low – Temperature Bainite Transformation

Nanobainite transformation behavior was comparably studied using in situ observations for two heat treatments: With and without partial quenching before isothermal holding at 300 ◦C. It was found that the prior martensite formation significantly accelerated the rate of the subsequent nanobainite transformation. Bainitic laths formed adjacent to a prior martensite plate and grew up to austenites. Bainite phase also formed both at the grain boundaries of the parent phase and inside the grains. Regarding the growth mode, bainite grows along the longitudinal direction and hardly grows along the lateral direction.

In recent years, different methods have been adopted to accelerate bainite transformation.Gong et al. [10] reported that nanobainite transformation was accelerated by introducing partial dislocations in austenite during low-temperature ausforming at 300 • C. Shibata et al. [11] and Miyamoto et al. [12] observed that bainite transformation was accelerated by incorporating a small amount of martensite prior to the isothermal holding, it was posited that martensite transformation introduced dislocations in martensite grains as well as in austenites near the martensite/austenite interface.Kawata et al. [13] found that bainite transformation was accelerated by pre-formed martensite due to an increase in the nucleation sites at the martensite/austenite interface.However, to clarify the effects of prior martensite formation on bainite transformation behavior, real-time dynamic in situ observations are necessary.Thus far, in situ heating experiments have been carried out by transmission electron microscopy (TEM) and optical microscopy (OM) to observe the transformation behavior of bainite [14,15].However, it is difficult to dynamically observe the primary transformation behavior by OM and TEM, because the spatial resolution in OM is insufficient.Although the spatial resolution in TEM is sufficient, the thin film effect cannot be avoided owing to the sample shape.The present Metals 2018, 8, 818 2 of 9 work reports the in-situ observation of the bainite formation and growth by a high-temperature laser scanning confocal microscope (LSCM, VL2000DX-SVF17SP, LASERTEC, Yokohama, Japan), where the spatial resolution is higher than that in OM and the bulk material can be observed.It aims at providing direct evidence for the effects of prior martensite formation on bainite formation and growth.

Experiments
A steel ingot of composition Fe-0.81C-1.67Si-2.01Mn-1.00Cr-0.26Mo-3.70Co-1.30Al was melted in a medium frequency induction vacuum furnace to prepare a 50 mm × 50 mm hot rolled bar, which was cut into blocks and homogenized at 1100 • C for 48 h to minimize the micro-segregation of alloying elements.Each block was encased in a C-steel box during the homogenization process to prevent excessive decarburization and oxidation.Cylindrical dilatometric samples of 6 mm in diameter and 70 mm in length were machined from the blocks and used to measure the bainitic start (Bs) and the martensitic start (Ms) temperatures by using a Gleeble-3500 thermal simulator (DSI Europe GmbH, Weissenhorn, Germany).The Bs and Ms temperatures of the steel sample were 403 • C and 219 • C, respectively.Furthermore, square specimens (4 mm × 4 mm × 4 mm) were prepared by spark wire cutting for in situ observation by a high temperature laser scanning confocal microscope (LSCM, VL2000DX-SVF17SP, LASERTEC, Yokohama, Japan).
The LSCM used the purple laser scanning illumination imaging technology, the wavelength was 408 nm, the scanning speed was 120 frames per second, and the highest resolution was 0.14 mm.By using an infrared image furnace, the heating rate is achieved.By using a thermocouple on the bottom of the aluminium crucible, the temperature was controlled.The samples were set into an alumina crucible after being machine polished.The chamber was vacuum pumped and then filled with argon to prevent the sample from being oxidized during heating.The specimens were heated as shown by the schematic illustration in Figure 1.
avoided owing to the sample shape.The present work reports the in-situ observation of the bainite formation and growth by a high-temperature laser scanning confocal microscope (LSCM, VL2000DX-SVF17SP, LASERTEC, Yokohama, Japan), where the spatial resolution is higher than that in OM and the bulk material can be observed.It aims at providing direct evidence for the effects of prior martensite formation on bainite formation and growth.

Experiments
A steel ingot of composition Fe-0.81C-1.67Si-2.01Mn-1.00Cr-0.26Mo-3.70Co-1.30Al was melted in a medium frequency induction vacuum furnace to prepare a 50 mm × 50 mm hot rolled bar, which was cut into blocks and homogenized at 1100 °C for 48 h to minimize the micro-segregation of alloying elements.Each block was encased in a C-steel box during the homogenization process to prevent excessive decarburization and oxidation.Cylindrical dilatometric samples of 6 mm in diameter and 70 mm in length were machined from the blocks and used to measure the bainitic start (Bs) and the martensitic start (Ms) temperatures by using a Gleeble-3500 thermal simulator (DSI Europe GmbH, Weissenhorn, Germany).The Bs and Ms temperatures of the steel sample were 403 °C and 219 °C, respectively.Furthermore, square specimens (4 mm × 4 mm × 4 mm) were prepared by spark wire cutting for in situ observation by a high temperature laser scanning confocal microscope (LSCM, VL2000DX-SVF17SP, LASERTEC, Yokohama, Japan).
The LSCM used the purple laser scanning illumination imaging technology, the wavelength was 408 nm, the scanning speed was 120 frames per second, and the highest resolution was 0.14 mm.By using an infrared image furnace, the heating rate is achieved.By using a thermocouple on the bottom of the aluminium crucible, the temperature was controlled.The samples were set into an alumina crucible after being machine polished.The chamber was vacuum pumped and then filled with argon to prevent the sample from being oxidized during heating.The specimens were heated as shown by the schematic illustration in Figure 1.Two different types of heat treatments were carried out for in situ observation: (a) direct isothermal bainite transformation (DIT) and (b) quenching followed by isothermal bainite transformation (QBT).During QBT, the specimens were austenitized at 1000 °C for 1800 s followed by a rapid cooling to 180 °C for 60 s, then heated up to 300 °C for 2400 s, and finally, cooled to room temperature.However, during DIT, no partial quenching process was executed, the specimens were directly cooled from 1100 °C to 300 °C for isothermal bainite transformation.The live pictures were taken every fifteenth second from 240 mm × 240 mm surface areas.Two different types of heat treatments were carried out for in situ observation: (a) direct isothermal bainite transformation (DIT) and (b) quenching followed by isothermal bainite transformation (QBT).During QBT, the specimens were austenitized at 1000 • C for 1800 s followed by a rapid cooling to 180 • C for 60 s, then heated up to 300 • C for 2400 s, and finally, cooled to room temperature.However, during DIT, no partial quenching process was executed, the specimens were directly cooled from 1100 • C to 300 • C for isothermal bainite transformation.The live pictures were taken every fifteenth second from 240 mm × 240 mm surface areas.After in-situ observation, the specimens were sectioned transversely and polished using standard techniques and finally, etched in nital solution (4 vol%) for microstructure evaluation.The central area of the specimen was examined by an optical microscope (OM, Olympus BM51, Olympus Corporation, Tokyo, Japan) and a scanning electron microscope (SEM, Nova 400 Nano field-emission, FEI, Hillsboro, OR, USA).The in-situ observation and SEM images were used to determine the microstructures.The volume fraction of microstructures was measured using the standard point counting technique, using Image-Pro Premier Image Analysis Software 9.1 (Meyer, Houston, TX, USA) [16].Twenty measurements were conducted on each sample, and the corresponding average values are reported in this work.

In Situ Observation of Bainitic Transformation Behavior for DIT and QBT
Figure 2 shows LSCM images of the QBT sample obtained before and after quenching.There are legible grains of the parent phase in the image in Figure 2a.The heating experiment was performed from room temperature up to 1000 • C for 1800 s, followed by a rapid cooling to 180 • C for 60 s, at which point the sample was maintained, after which observation began.Figure 2b shows the LSCM image of the sample after quenching.A large volume fraction of prior martensite was obtained through quenching.Figure 3a-d  After in-situ observation, the specimens were sectioned transversely and polished using standard techniques and finally, etched in nital solution (4 vol%) for microstructure evaluation.The central area of the specimen was examined by an optical microscope (OM, Olympus BM51, Olympus Corporation, Tokyo, Japan) and a scanning electron microscope (SEM, Nova 400 Nano field-emission, FEI, Hillsboro, OR, USA).The in-situ observation and SEM images were used to determine the microstructures.The volume fraction of microstructures was measured using the standard point counting technique, using Image-Pro Premier Image Analysis Software 9.1 (Meyer, Houston, TX, USA) [16].Twenty measurements were conducted on each sample, and the corresponding average values are reported in this work.

In Situ Observation of Bainitic Transformation Behavior for DIT and QBT
Figure 2 shows LSCM images of the QBT sample obtained before and after quenching.There are legible grains of the parent phase in the image in Figure 2a.The heating experiment was performed from room temperature up to 1000 °C for 1800 s, followed by a rapid cooling to 180 °C for 60 s, at which point the sample was maintained, after which observation began.Figure 2b shows the LSCM image of the sample after quenching.A large volume fraction of prior martensite was obtained through quenching.Figure 3a-d      For comparison, Figure 4a-d shows the LSCM images of the DIT sample (a) at the beginning of the observation and after observation for (b) 900 s, (c) 1500 s and (d) 2400 s.At the beginning of the observation, supercooled austenites were seen and the black arrows denote austenite grain boundaries in Figure 4a.As bainite transformation required an incubation period, nucleation of bainite did not occur in a period time, and consequently, a thin bainite phase was formed, initiated on the grain boundaries after 900 s as indicated by the red arrows in Figure 4b.With isothermal time increasing, bainite plates were formed in different directions and the number of bainite plates in the matrix increased, as indicated by the red arrows in Figure 4c,d.Moreover, the dashed lines in Figure 4c,d show the bainite regions and indicate that the bainite region grows along the longitudinal direction and hardly grows along the lateral direction.Comparing the two processes, more bainite plates were formed in the QBT specimen, signifying the bainite transformation was significantly accelerated due to prior martensite formation during the QBT process.

Microstructural Characteristics of Nanobainite Obtained by DIT and QBT
Figures 5 and 6 show the OM and SEM microstructure obtained at room temperature after in situ observation.As shown in Figures 5a and 6a,b, the structure of nanobainite formed by the DIT process consisted of nanostructured bainitic ferrite and films of austenite and blocky austenite, which is the untransformed region.It was concluded that the bainite transformation was still incomplete in the DIT process.Figure 6b shows the higher magnification observation which suggests that the nucleation site of the bainite phase is located at the grain boundary, which is also confirmed by the in-situ observation displayed in Figure 4b.

Microstructural Characteristics of Nanobainite Obtained by DIT and QBT
Figures 5 and 6 show the OM and SEM microstructure obtained at room temperature after in situ observation.As shown in Figures 5a and 6a,b, the structure of nanobainite formed by the DIT process consisted of nanostructured bainitic ferrite and films of austenite and blocky austenite, which is the untransformed region.It was concluded that the bainite transformation was still incomplete in the DIT process.Figure 6b shows the higher magnification observation which suggests that the nucleation site of the bainite phase is located at the grain boundary, which is also confirmed by the in-situ observation displayed in Figure 4b.

Microstructural Characteristics of Nanobainite Obtained by DIT and QBT
Figures 5 and 6 show the OM and SEM microstructure obtained at room temperature after in situ observation.As shown in Figures 5a and 6a,b, the structure of nanobainite formed by the DIT process consisted of nanostructured bainitic ferrite and films of austenite and blocky austenite, which is the untransformed region.It was concluded that the bainite transformation was still incomplete in the DIT process.Figure 6b shows the higher magnification observation which suggests that the nucleation site of the bainite phase is located at the grain boundary, which is also confirmed by the in-situ observation displayed in Figure 4b.Figures 5b and 6c,d present the microstructure of the QBT specimen which consists of bulky lenticular prior martensite and nanobainite.Moreover, prior martensite manifested a larger size than the martensites formed from untransformed austenites upon cooling to room temperature.The higher magnification observation as shown in Figure 6d suggested that bainite laths formed adjacent to a prior martensite and grew up to austenites.Furthermore, the untransformed regions in the QBT process are obviously less than that in the DIT process as shown in Figure 5.The observation also confirmed that bainite transformation was accelerated by the prior martensite formation in the QBT process.For further understanding of this bainitic transformation, careful analysis at the growth edge of bainite around the prior martensite is required by a dedicated analytical electron which has high sensitivity.

Comparison of Bainite Transformation Kinetics between DIT and QBT
Some previous studies [18,19] claim that carbon partitioning into austenite can increase the stabilization of austenite and consequently, hinder the subsequent bainite transformation.Therefore, in order to avoid this effect, a certain volume fraction of prior martensite was introduced before isothermal bainite transformation in our work, which helped us observe the difference between the QBT and DIT processes.The Image-Pro software measured the volume fraction of martensite (VM) as 33.2% during the QBT process, which was nearly equal to the value (34.9%) calculated by the Koistinen-Marburger law [20], which is expressed by Equation (1).Figures 5b and 6c,d present the microstructure of the QBT specimen which consists of bulky lenticular prior martensite and nanobainite.Moreover, prior martensite manifested a larger size than the martensites formed from untransformed austenites upon cooling to room temperature.The higher magnification observation as shown in Figure 6d suggested that bainite laths formed adjacent to a prior martensite and grew up to austenites.Furthermore, the untransformed regions in the QBT process are obviously less than that in the DIT process as shown in Figure 5.The observation also confirmed that bainite transformation was accelerated by the prior martensite formation in the QBT process.For further understanding of this bainitic transformation, careful analysis at the growth edge of bainite around the prior martensite is required by a dedicated analytical electron microscope which has high sensitivity.

Comparison of Bainite Transformation Kinetics between DIT and QBT
Some previous studies [18,19] claim that carbon partitioning into austenite can increase the stabilization of austenite and consequently, hinder the subsequent bainite transformation.Therefore, in order to avoid this effect, a certain volume fraction of prior martensite was introduced before isothermal bainite transformation in our work, which helped us observe the difference between the QBT and DIT processes.The Image-Pro software measured the volume fraction of martensite (V M ) as 33.2% during the QBT process, which was nearly equal to the value (34.9%) calculated by the Koistinen-Marburger law [20], which is expressed by Equation (1).
Metals 2018, 8, 818 7 of 9 where f is the martensite volume fraction, Ms is the martensite transformation starting temperature, and Tq is the quenching temperature.7. The figures confirm that the bainite transformation is accelerated by the presence of prior martensite.Moreover, the time required to obtain a 10% V B was shorter for QBT in comparison to DIT: 1000 s vs. 1500 s.Gong et al. [10] and Zhu et al. [19] had used this method to compare the kinetics of bainite transformation from the different conditions of austenite in QBT.where f is the martensite volume fraction, Ms is the martensite transformation starting temperature, and Tq is the quenching temperature.7. The figures confirm that the bainite transformation is accelerated by the presence of prior martensite.Moreover, the time required to obtain a 10% VB was shorter for QBT in comparison to DIT: 1000 s vs. 1500 s.Gong et al. [10] and Zhu et al. [19] had used this method to compare the kinetics of bainite transformation from the different conditions of austenite in QBT.Table 1 reveals that there was some relationship between the number of bainitic-ferrite nucleation points and the isothermal time.This experimental finding suggests that the QBT process can provide more nucleation sites than the DIT process in the same duration time.However, two mechanisms were proposed for the acceleration of bainite transformation due to prior martensite.First, transformation strains produced by martensite formation provided the preferential nucleation sites for subsequent bainite transformation, and second, the interface between prior martensite and austenite acted as a preferential nucleation site [21].Most researchers support the first idea.Howard et al. [22] found that most of the premature bainite was formed clearly away from the martensite plates.Kawata et al. [13] claimed that interfacial energy played a key role in the acceleration of bainite transformation and found that the acceleration effects of prior martensite (with transformation strain) and prior ferrite (without transformation strain) were the same when the volume fractions of both phases were at the same level.Therefore, the mechanism behind the acceleration of bainite transformation due to prior martensite is still under debate.Table 1 reveals that there was some relationship between the number of bainitic-ferrite nucleation points and the isothermal time.This experimental finding suggests that the QBT process can provide more nucleation sites than the DIT process in the same duration time.However, two mechanisms were proposed for the acceleration of bainite transformation due to prior martensite.First, transformation strains produced by martensite formation provided the preferential nucleation sites for subsequent bainite transformation, and second, the interface between prior martensite and austenite acted as a preferential nucleation site [21].Most researchers support the first idea.Howard et al. [22] found that most of the premature bainite was formed clearly away from the martensite plates.Kawata et al. [13] claimed that interfacial energy played a key role in the acceleration of bainite transformation and found that the acceleration effects of prior martensite (with transformation strain) and prior ferrite (without transformation strain) were the same when the volume fractions of both phases were at the same level.Therefore, the mechanism behind the acceleration of bainite transformation due to prior martensite is still under debate.It is noticeable that bainites can formed at the interface between martensite and austenite as shown in Figure 6d.However, we study it just in two dimensions.It is difficult to deduce the structure of the observed surface because a bainite can touch a prior martensite beneath the observed surface [23].Therefore, the results obtained from in situ observations are not enough to infer the major factor for the acceleration of bainite transformation.Further investigation is needed to clarify how the prior martensite impacts on the bainite transformation, making it necessary to observe the three-dimensional morphology, for example, using a Focused Ion Beam Scanning Electron Microscope (FIB-SEM).

Conclusions
The in-situ observation method was employed to explore the nanobainite transformation behavior in the DIT and QBT processes.The main results obtained are summarized as follows: 1.
Prior martensite formation is found to accelerate the subsequent nanobainite transformation.
The transformation occurs more quickly in QBT than in DIT.

2.
Bainitic laths formed adjacent to a prior martensite plate and grew up to austenites.

3.
Bainite phases formed both at the grain boundaries of the parent phase and inside the grains.
Regarding the growth mode, bainite grows along the longitudinal direction and hardly grows along the lateral direction.
Figure2shows LSCM images of the QBT sample obtained before and after quenching.There are legible grains of the parent phase in the image in Figure2a.The heating experiment was performed from room temperature up to 1000 • C for 1800 s, followed by a rapid cooling to 180 • C for 60 s, at which point the sample was maintained, after which observation began.Figure2bshows the LSCM image of the sample after quenching.A large volume fraction of prior martensite was obtained through quenching.Figure3a-dshows the LSCM images of the QBT sample (a) at the start of the observation and after observation for (b) 900 s, (c) 1500 s and (d) 2400 s.At the start of the observation, a large volume fraction of prior martensite was seen, noted as blue arrows in Figure 3a.After 900 s, a certain volume fraction of bright bainite phase was formed, as indicated by the red arrows in Figure 3b.As shown in Figure 3c,d, after a longer aging time, other bainite variants rapidly grew into the matrix in different directions, until colliding with the grain boundaries (indicated by black arrows) on the other side of the matrix [17].
Figure2shows LSCM images of the QBT sample obtained before and after quenching.There are legible grains of the parent phase in the image in Figure2a.The heating experiment was performed from room temperature up to 1000 °C for 1800 s, followed by a rapid cooling to 180 °C for 60 s, at which point the sample was maintained, after which observation began.Figure2bshows the LSCM image of the sample after quenching.A large volume fraction of prior martensite was obtained through quenching.Figure3a-dshows the LSCM images of the QBT sample (a) at the start of the observation and after observation for (b) 900 s, (c) 1500 s and (d) 2400 s.At the start of the observation, a large volume fraction of prior martensite was seen, noted as blue arrows in Figure 3a.After 900 s, a certain volume fraction of bright bainite phase was formed, as indicated by the red arrows in Figure 3b.As shown in Figure 3c,d, after a longer aging time, other bainite variants rapidly grew into the matrix in different directions, until colliding with the grain boundaries (indicated by black arrows) on the other side of the matrix [17].

Figure 6 .
Figure 6.SEM microstructures of (a,b) DIT and (c,d) QBT specimens after in situ observation.(b,d) are enlarged images of the framed area in (a,c), respectively.

Figure 6 .
Figure 6.SEM microstructures of (a,b) DIT and (c,d) QBT specimens after in situ observation.(b,d) are enlarged images of the framed area in (a,c), respectively.

Figure 7
depicts a comparison of bainite transformation kinetics between DIT and QBT at 300 • C, respectively.For the DIT process, the volume fraction of bainitic ferrite (V B ) is determined by the Image-Pro software and the results are plotted by the open circles in the figure.For the QBT process, the summation of the volume fraction of V B and V M , given as (V B+M ), is plotted by the four open squares in the figure.The normalized volume fraction of bainitic ferrite f B = V B /(1 − V M ) is plotted by the solid triangles in Figure

Figure 7
depicts a comparison of bainite transformation kinetics between DIT and QBT at 300 °C , respectively.For the DIT process, the volume fraction of bainitic ferrite (VB) is determined by the Image-Pro software and the results are plotted by the open circles in the figure.For the QBT process, the summation of the volume fraction of VB and VM, given as (VB+M), is plotted by the four open squares in the figure.The normalized volume fraction of bainitic ferrite fB = VB/(1 − VM) is plotted by the solid triangles in Figure

Figure 7 .
Figure 7.Comparison of the bainite transformation kinetics between DIT and QBT at an isothermal temperature of 300 °C .

Figure 7 .
Figure 7.Comparison of the bainite transformation kinetics between DIT and QBT at an isothermal temperature of 300 • C.

Table 1 .
Relationship between the number of bainitic-ferrite nucleation points and the isothermal time.