2020 Volume 60 Issue 5 Pages 930-938
Time-resolved and in-situ observations using synchrotron radiation X-rays successfully proved that a massive-like transformation, in which the γ phase was produced through the solid–solid transformation and partitioning of substitute elements such as Mn and Si at the δ–γ interface was negligible, was selected in the unidirectional solidification of 0.3 mass% C steel at a pulling rate of 50 µm/s. The massive-like transformation produced fine γ grains near the front of the δ–γ interface. The coarse γ grains also grew behind the fine γ grains along the temperature gradient. Distance between the δ–γ front and the advancing front of coarse γ grains was as short as 200 µm. Namely, the fine γ grains disappeared within 10 s owing to growth of coarse γ grains. In addition, the observation of the δ–γ interface confirmed that a transition from the diffusion-controlled γ growth to the massive-like γ growth occurred at a growth velocity of 5 µm/s. Thus, the massive-like transformation is dominantly selected in the carbon steel during conventional solidification processes.
It is widely accepted that the peritectic solidification occurs in carbon steel with a carbon content lower than 0.5 mass%. In addition, the volume decrease owing to the transformation from the δ phase (BCC) to the γ phase (FCC) is considered as an origin of the unevenness and deformation of solidifying shell.1,2,3) Thus, the solidification and related phenomena, including formation of casting defects, in carbon steel are understood on the basis of the peritectic solidification.
Monochromatized hard X-rays with high coherency and high brilliance, produced at third-generation synchrotron radiation facilities such as SPring-8 (Hyogo, Japan), have allowed us to observe various solidification phenomena in metallic alloys.4,5,6) Solidification phenomena in metallic alloys with relatively low melting temperatures, such as Sn, Zn, and Al alloys, have been observed in-situ by several research groups.7,8,9,10,11) In addition, radiography has also been developed to observe solidification phenomena at temperatures above 1500°C for Fe-based alloys.12,13,14,15,16)
The time-resolved and in-situ observations for steel15,16) showed that the δ–γ transformation was proceeded by a massive-like transformation. The massive-like transformation was also confirmed by the observations with a confocal laser scanning microscope.17) These experimental results indicate that the massive-like transformation occurs in the δ–γ transformation during and/or after solidification of the δ phase in carbon steel.
In this paper, the peritectic solidification and the massive-like transformation are defined as follows. The former is defined as a transformation mode in which growth of the γ phase is controlled by carbon diffusion and the γ phase continuously grows from the crystallographic view point. This mode is the same as a conventional transformation mode (the peritectic solidification). Conversely, the latter is defined as a transformation mode in which sequential nucleation and growth of the γ phase occur in the vicinity of δ–γ interface. This transformation mode is essentially a solid–solid transformation and produces fine γ grains within a δ grain. The sequence of nucleation and growth has been explained by considering the relationship of interfacial energies.18,19,20)
In our previous observations,15,16) δ dendrites grew in advance and nucleation of the γ phase hardly occurred even at temperatures below the peritectic temperature when the specimens which had dimensions of 5–10 mm width and 0.1 mm thickness was cooled. As a result, the δ phase was undercooled below the peritectic temperature. Nucleation of the γ phase typically required undercooling of several 10 K from the peritectic temperature, because the interface between the liquid and the δ phases was not a preferred nucleation site for the γ phase. Nucleation of the γ phase in the undercooled δ phase triggered the massive-like transformation. These results suggested that the massive-like transformation could occur in the initial solidification region (surface region of castings) where the melt contacted the mold through the mold flux in practical solidification processes such as continuous casting.
Although the massive-like transformation can occur in the initial solidification region, the conditions leading to selection of the massive-like transformation inside the castings remains unclear. After the nucleation of the γ phase in the surface region, the γ phase exists behind δ dendrites and the nucleation of the γ phase is not required for the peritectic solidification. Thus, it is of interest to examine which transformation mode is selected in the middle regions of the castings where undercooling of the δ phase is relatively small and the γ phase exists behind the δ phase.
The morphology of γ grains has been extensively studied and a model for formation of γ grains during unidirectional solidification in carbon steels has been proposed.21,22,23) Fine γ grains ahead of coarse columnar γ grains were observed in the quenched specimens. As mentioned before, the massive-like transformation produces fine γ grains.15,16) Thus, it would be useful to examine the contribution of the massive-like transformation to the formation of γ grains reported by the quench experiments.21,22,23)
This paper reports dendritic growth of the δ phase followed by a δ–γ transformation in 0.3 mass%C steel, studied by time-resolved and in-situ X-ray imaging. Furthermore, the critical conditions required to promote the diffusion-controlled growth of γ phase or the massive-like transformation are demonstrated. The contribution of the massive-like transformation to the formation of coarse γ grains and the selection of the δ–γ transformation mode in conventional casting processes will be also discussed.
Fe - 0.18 mass% C - 0.6 mass% Mn - 0.3 mass% Si (hereafter referred to as 0.18C steel) and Fe - 0.3 mass% C - 0.6 mass% Mn - 0.3 mass% Si alloys (referred to as 0.3C steel) were made from electrolytic iron, electrolytic manganese, high-purity silicon, and high-purity carbon in an arc furnace under purified Ar atmosphere.
Time-resolved and in-situ observations were performed at beamlines BL20XU and BL20B2 at SPring-8. The setup used in the observations is shown in Fig. 1. A monochromatized X-ray beam was passed from the left to the right side of the figure. Slits for shaping the X-ray beam, absorbers for controlling intensity of the incident beam, a vacuum chamber containing a specimen and carbon heater, and an X-ray beam monitor were placed as shown in Fig. 1. An ion chamber was used to measure the intensity of the incident X-ray beam, if required. A wide view detector (panel-type detector) was also set to measure diffraction spots. In the present study, diffraction spots of (111)γ, (200)γ, (220)γ from the γ grains were detected on the panel-type detector.
Setup of in-situ observation using synchrotron radiation X-rays.
A temperature gradient of approximately 10 K/mm was achieved with the use of the designated carbon heater. The specimen was unidirectionally solidified by pulling down at a given rate or by decreasing the temperature of the furnace at a selected cooling rate.
Figure 2 shows two different specimen holders used in this study. One held a specimen of 10 mm in width, 10 mm in height and 0.1 mm in thickness for observing the transition of the δ–γ transformation mode. The other held a specimen of 8 mm in width, 25 mm in height, and 0.1 mm in thickness for unidirectional solidification. At the beginning of unidirectional solidification, the γ phase remained behind the δ phase in the specimen with a height of 25 mm. In the long specimen, it was possible to observe the δ–γ transformation behind δ dendrite growth nearly at the steady state.
Dimensions of specimen used for unidirectional solidification.
When only transmission images were recorded, an X-ray energy of 21 keV was selected to enhance the contrast resolution on transmission images. Since the diffraction angles became smaller at higher X-ray energies, 28 keV was selected to observe transmission images and diffraction spots simultaneously. Since the density of the γ phase was slightly greater than that of the δ phase, the intensity of X-rays penetrating though the γ phase was lower than those passing though the δ phase. However, the contrast resolution between the δ and γ phases was not sufficient to detect the δ–γ transformation. A number of γ grains produced by the massive-like transformation satisfied the Bragg conditions and consequently caused dark regions in the transmission images. The δ–γ interface was identified by tracing the dark spots. In addition, the diffraction spots on the panel-type detector were used to identify the δ–γ transformation.
Figure 3 shows X-ray transmission images (left) and X-ray diffraction images (right) during the unidirectional solidification of 0.3C steel at a pulling rate of 50 μm/s under a temperature gradient of 10 K/mm. Time was set to be zero when the pulling started. The liquid phase was observed in the upper part of the transmission images, and the δ and the liquid phases coexisted in the lower part at the beginning of the unidirectional solidification (0 s). Although a broad ring caused by the liquid phase was observed in the diffraction images, no diffraction spots from the δ phase were detected. Divergence of the X-ray beam at the beamline BL20B2 was 1.5 mrad in the vertical direction and 0.06 mrad in the horizontal direction.5,6) The low divergence resulted in a low probability that the grains satisfied the Bragg conditions; hence, an absence of diffraction spots from the δ phase was normally expected.
Transmission images (left) and diffraction images (right) during steady growth state in Fe-0.3C-0.6Mn-0.3Si. Pulling rate was 50 μm/s. Arrows on the left side (transmission images) indicate the growth front of the γ phase. Pixel size for transmission image and diffraction images are 2.5 μm × 2.5 μm and 50 μm × 50 μm, respectively. Frame rate of transmission images was 1 fps. X-ray energy was 21 keV.
A morphological transition of the δ phase from cells to dendrites occurred once the specimen was pulled down and dendrites started to grow from the bottom to the top in the image. The dark regions induced by the X-ray diffraction were observed behind the front of the γ phase, as indicated by the two arrows at 80 s. The dark regions of the γ grains satisfied the Bragg conditions. At the same time, the diffraction spots of (111)γ and (200)γ planes were also detected in the diffraction image (right). As mentioned before, there was a low probability of satisfying the Bragg conditions because of the low divergence of the incident X-ray beam. Thus, some dark regions and diffraction spots at 80 s indicated that fine γ grains were produced by the massive-like transformation. Furthermore, the diffraction spots distributed in the γ phase region suggested that the equiaxed γ grains were produced in the observation area.
Since the distance between the dendrite tips of the δ phase and the front of the γ phase did not change between 80 and 100 s, nearly steady state growth at a velocity of 50 μm/s was achieved. Namely, the massive-like transformation was selected even in the steady state. The distance between the tip and the front was approximately 800 μm and the difference in temperature was estimated to be approximately 10 K.
In addition to the dark regions behind the front of the γ phase, a coarse γ grain, which had a front at 500 μm from the tip, was also observed in the transmission image at 100 s. Many diffraction spots of the γ phase were also measured by the panel-type detector. Therefore, fine γ grains produced by the massive-like transformation and a coarse γ grain coexisted in the observation area.
The coarse γ grain, indicated by “Coarse γ” in the transmission image, grew unidirectionally and the distance from the tips to the front became 200 μm at 150 s. Since the dark spots were hardly observed from the bottom-center to the bottom-right corner in the transmission image, the coarse γ grain was formed during unidirectional solidification. The configuration of the fine and coarse γ grains indicated that the coarse γ grains were not simply produced by coarsening between the fine γ grains. Rather, the coarse γ grains in the low temperature side continuously grew into the fine γ grain region.
Figures 4(a) and 4(b) show the X-ray diffraction spots of the (200)γ and (111)γ planes at 80 s. Since the diffraction angle ranged from 16° to 19° and the incident beam was collimated, the shapes of the diffraction spots reflected the regions where Bragg conditions were satisfied. Although some of the spots were less than 100 μm in diameter, spots as large as 1 mm in diameter were also observed. Thus, γ grains with a diameter greater than 1 mm existed in the observation area. This result is consistent with the grain size observed in transmission images after 80 s in Fig. 3.
Close-up views of diffraction spots at t = 80 s. (a) Spots of (111)γ and (b) (200)γ.
These in-situ observations confirmed that the peritectic solidification did not occur, in which the growth of the γ phase was diffusion-controlled. Instead the massive-like transformation, in which the sequence of nucleation and growth of the γ phase was repeated, was selected even in the unidirectional solidification at a growth velocity of 50 μm/s.
Unidirectional solidification at lower growth velocities was observed in 0.18C steel to measure a critical growth velocity for the transition from the crystallographically continuous growth of γ phase to the massive-like transformation. Figure 5 shows a schematic illustration of the microstructure at the beginning of the pulling down with the Fe–C phase diagram. When the specimen was maintained at a constant temperature under a temperature gradient, smooth interfaces of liquid-δ and δ–γ were obtained. The liquid phase, the δ phase and the γ phase were aligned along the temperature gradient. The carbon content in each phase was uniform after maintaining the specimen for a sufficient duration. The carbon concentrations of the liquid, δ and γ phases at the interfaces are shown in Fig. 5. Since the temperature at the liquid-δ interface is above the peritectic temperature, the carbon concentration of the liquid phase at the interface was higher than the average carbon concentration of 0.18 mass%. As shown in the phase diagram, carbon concentrations of the δ and the γ phases at the δ–γ interface should be less than 0.1 mass% and 0.18 mass%, respectively. The average carbon concentration determines the fractions of the liquid, δ and γ phases. The concentrations at the δ–γ interface are simply determined by the phase equilibrium. It should be noted that the carbon concentration of the δ phase in the massive-like transformation that occurred at temperatures below the peritectic temperature in Fig. 3 should be comparable with the carbon concentration of the δ phase in Fig. 5, if the local equilibrium was satisfied in the transformation.
Carbon-rich portion of Fe–C phase diagram and the initial microstructure for the unidirectional growth of γ phase. (Online version in color.)
Figure 6(a) shows a transmission image around the δ–γ interface before pulling down the specimen. The white arrow indicates the front position of the δ–γ interface. The dark regions in the γ phase, which were observed in the unidirectional solidification, were also observed behind the δ–γ interface. The specimen was pulled down at a rate of 8 μm/s. When the pulling-down started, the δ–γ interface did not follow the pulling and moved downward. The shape of the δ–γ interface slightly changed, indicating some growth of the γ phase. The backward motion of the δ–γ interface was explained by the decrease in the carbon concentration and the interface temperature owing to the growth of the γ phase and the carbon partitioning at the δ–γ interface in the local equilibrium condition. The front of the δ–γ interface rapidly moved upward after 110 s and black regions behind the front were observed.
(a) Transmission image before pulling down of specimen and (b) close-up views of δ–γ interface. White arrows indicate the position of δ–γ interface.
Figure 7 shows the position of the δ–γ interface as a function of pulling time. The lines are just guides indicating pulling rates of 8, 3 and 1.6 μm/s, respectively. Since the moving velocity of the δ–γ interface, which is defined as the moving velocity in the transmission image, was −3.0 μm/s at 20 s from the beginning, the growth velocity of the γ phase was estimated to be 5 μm/s. The moving and the growth velocities were −1.6 μm/s and 6 μm/s at 100 s, respectively. Thus, the growth velocity of the γ phase at 100 s slightly increased. The δ–γ interface position rapidly moved upward and simultaneously the massive-like transformation was detected at 110 s. This observation shows that the transition from the diffusion-controlled growth to the massive-like transformation occurred even at the growth velocities in the range of 5–6 μm/s.
Vertical position of δ–γ interface as a function of pulling time. Pulling rate was 8 μm/s. Arrow indicates the transition from the diffusion-controlled continuous growth to the massive-like growth.
Carbon atoms were transported from the δ phase to the γ phase, when the γ phase grew in the δ phase. For simplicity, the steady state growth of the γ phase in the δ phase is considered in a specimen with the average concentration, C’, as defined in Fig. 5. During the transient stage to the steady state growth from the beginning of pulling down, temperature of the δ–γ interface decreases and approaches the temperature, T’, at which the carbon concentration of the γ phase at the δ–γ interface is C’. According to the experimental results, the nucleation of the γ phase triggers the massive-like transformation when a driving force for nucleation of the γ phase ahead of growth at the δ–γ interface reaches a critical value during the transient process. The fact that the massive-like transformation was selected even at a relatively low growth velocity ranging from 5 μm/s to 6 μm/s suggested that the critical driving force for the nucleation of the γ phase in the massive-like was relatively small. Details of the driving force are discussed in the next section.
As diffusion coefficients of solutes in the γ phase (FCC) are generally smaller than those in the δ phase (BCC), the growth of the γ phase in the δ phase has some similar features to solidification in terms of solute diffusion. Here, the driving force for nucleation of theγ phase ahead of the δ–γ interface is considered using the constitutional undercooling ahead of the planar interface. Figure 8(a) shows the temperature distribution and the solute profile around the planer δ–γ interface under the steady state growth with a binary phase diagram in which a solute was partitioned into the γ phase. Assuming that the steady state growth with a planar interface was achieved, the solute concentration in the γ phase at the δ–γ interface (Cγ*) is equal to the average concentration (C0). The solute concentration in the δ phase outside of the solute diffusion layer is also the average concentration (C0). Additionally, if the temperature gradient (G) is assumed to be constant around the δ–γ interface, the temperature distribution around the δ–γ interface is simply obtained as shown Fig. 8(a). In the coordination that the δ–γ interface position is defined as z = 0, the solute and the thermal profiles in the δ phase are given by Eqs. (1) and (2), respectively.
(a) Schematic illustrations of solute profile ahead of the δ–γ interface and temperature profile during steady-state growth of γ phase, (b) solute profiles of manganese in the δ phase ahead of the δ–γ interface, plotted on the phase diagram of the Fe–Mn binary system and (c) solute profiles of carbon in the δ phase ahead of the δ–γ interface, plotted on the phase diagram of Fe–C binary system. (Online version in color.)
If the local equilibrium is achieved at the δ–γ interface, the δ phase with the concentration C0/k is in equilibrium with the γ phase with the concentration, C0, at the temperature, T*. The point (C0/k, T*) on the phase diagram is indicated by the point “p” in Fig. 8(a). The specimen is a quaternary alloy; hence, the δ–γ transformation should be treated as a multicomponent system in principle. For simplicity, the transformation is analyzed by considering the different binary systems. One is a Fe–Mn system in which manganese is a substitutional element and the other is a Fe–C system in which carbon is an interstitial element.
Temperature and solute profiles of manganese and carbon ahead of the δ–γ interface were calculated with Eqs. (1) and (2). The physical properties24) listed in Table 1 were used in the calculation. The relationships between the temperature and the concentration are drawn on the binary phase diagram. Figure 8(b) shows the profile of manganese at a growth velocity of 0.5 μm/s under a temperature gradient of 2 K/mm. The positions, z = 0 mm and z = 2 mm, are the distances from the δ–γ interface, as defined in Fig. 8(a). The profile of manganese is nearly horizontal on the phase diagram and the thickness of the diffusion layer, 2D/V, is approximately 40 μm. Thus, the constitutional undercooling that drives nucleation of the γ phase exists ahead of the δ–γ interface. Since the profile nearly reaches the solubility limit line of the γ phase, a part of the δ phase ahead of the δ–γ interface exceeds the T0 curve. Thus, the transformation from the δ to the γ phase is thermodynamically allowed without a manganese partition. Under these growth conditions, the region where the driving force exists within the diffusion layer of 40 μm and γ phase can nucleate only in the vicinity of the δ–γ interface. The calculation indicates that nucleation of the γ phase can occur only within several μm distance from the δ–γ interface at growth velocities higher than the critical growth velocity of 5 μm/s, which was measured in Fig. 6.
The above explanations are based on the partition local equilibrium mode (referred to as P-LE),25,26,27) in which the chemical potential of substitutional manganese is continuous at the δ–γ interface in the δ–γ transformation. Thus, it is clear that the driving force for nucleation of the γ phase exists ahead of the δ–γ interface in the P-LE mode. In fact, it is difficult to identify the P-LE mode or the Negligible−Partition Local Equilibrium (referred to as NP-LE)25,26,27) in the massive-like transformation. However, the driving force in the NP-LE is larger than that in the P-LE mode. Therefore, the explanation based on the constitutional undercooling (P-LE mode) shows that the driving force for nucleation of the γ phase arises in the vicinity of the δ–γ interface even if manganese atoms are sufficiently partitioned at the interface in the δ–γ transformation. The discussion here also reaches the similar conclusions for silicon as a substitutional solute.
Figure 8(c) shows the carbon profiles on the binary phase diagram, which are evaluated with Eqs. (1) and (2). The physical properties listed in Table 1 were used in the calculation. The distance z from the δ–γ interface under a temperature gradient of 2 K/mm is also indicated in the figure. Since the diffusivity of carbon in the δ phase is roughly 500 times greater than that of manganese, the diffusion layer becomes thicker. As shown in Fig. 8(c), the constitutional undercooling is hardly caused at a growth velocity of 0.5 μm/s. However, the constitutional undercooling region is caused ahead of the δ–γ interface at 5 μm/s and supersaturation of carbon in the δ phase is as great as 0.025 mass%. Therefore, the driving force for nucleation of the γ phase is generated ahead of the δ–γ interface at a growth velocity of 5 μm/s even if partition of carbon (fast diffuser) occurs. As shown in the figure, the carbon profile at 5 μm/s crosses the T0 line. Thus, the transformation from the δ phase to the γ phase without carbon partition is allowed in a limited region ahead of the δ–γ interface. However, the driving force does not play a main role controlling nucleation of the γ phase in the massive-like transformation. The thickness of the diffusion layer for carbon is estimated to be as long as 2 mm at a growth velocity of 5 μm/s. According to the in-situ observations, the massive-like transformation proceeded through the sequential nucleation and growth of the γ phase at the δ–γ interface. Since the sequence did not occur at the distance of 1 or 2 mm from the δ–γ interface, the constitutional undercooling caused by the carbon partition causes the driving force but does not control the nucleation position of the γ phase.
By analogy with the morphological transition of planar, cellular, and dendritic interfaces in solidification, the continuous growth of γ phase with cellular or dendritic interface in δ phase could occur in the unidirectional solidification. However, the in-situ observations indicated that the cellular or the dendritic growth was not selected. The selection of the massive-like transformation in the unidirectional solidification indicated that the critical undercooling for nucleation events of the γ phase at the δ–γ interface could be relatively small. Further study is required to understand why the nucleation events occurs at the interface in the massive-like transformation.
The microstructure of γ grains during unidirectional solidification in steel was extensively studied by quenching a sample during the solidification. The thermal conditions of the unidirectional solidification were comparable with those of continuous casting of steel.21) The formation of coarse columnar γ grains is not simply explained by considering the continuous growth of the γ phase along the temperature gradient.21) It has been reported that: 1) coarse columnar γ grains (referred to as CCG) continuously grew from the mold side; 2) fine columnar γ grains (referred to as FCG) existed ahead of CCG; 3) FCG is located in the region where the liquid and the γ phases coexisted; 4) width of FCG was 100 μm, which is equivalent to the diameter of the primary dendrite arms; 5) the front of the CCG was located at a position where solidification was completed, in the unidirectional solidification of 0.2 mass% C steel. In addition, the formation of CCG was explained by considering the carbon content in the two different steels (0.05 mass% to 0.135 mass%).22) Relating to the items 4) and 5), the remaining liquid phase played an important role for coarsening of γ grains. Since the δ–γ transformation occurred without a liquid phase for a carbon content less than 0.1 mass%, the columnar equiaxed γ grains (CEG) were formed. CCG was formed for a carbon content between 0.1 mass% and 0.2 mass%, because the remaining liquid phase did not exist in the γ region produced by the δ–γ transformation. Since the liquid phase remained between the dendrite arms in steels with higher carbon contents, the columnar γ grains, of which growth was restricted by the remaining liquid phase, were formed. A phase field simulation also showed that the shape of the CCG was determined by the relationship between the growth velocity of the CCG and the local cooling rate.23)
The formation of γ grains, as shown in Fig. 3, was consistent with the microstructure, explained by the items 1) and 2). Since the shapes of the γ grains were not clearly observable in the transmission images, their morphology, which correspond to FCG, was not confirmed in the present study. However, the in-situ observations give a possible explanation for the formation of fine γ grains. As shown in Fig. 3, some of the γ grains disappeared within 10 s after the massive-like transformation and a number of dark regions, which were caused by diffraction, in the transmission images suggested that the fine equiaxed γ grains existed just after the massive-like transformation. The FCG might be the result of coarsening of the fine equiaxed γ grains; however, further studies are required to observe the morphology of γ grains just after the massive-like transformation to validate this proposal. The width of the FCG, as indicated by the item 4), is simply explained by considering coarsening of the γ grains in the dendrite arms. Coarsening of γ grains in a dendritic arm, which is isolated by the remaining liquid phase can result in FCG. There was a discrepancy at the front of the CCG (term 5). For example, fine γ grains remained even in the γ region without any liquid at 100 s, as shown in Fig. 3, and a coarse γ grain existed behind the fine γ grains. Namely, the front of the coarse γ grain was not located at the position where the remaining liquid between the dendrite arms disappeared. Since the distance between the front and the remaining liquid phase was only 200 μm, this discrepancy does not affect the essential mechanism of γ grain formation.
As discussed here, the formation mechanism of γ grains in steel, which were obtained by the unidirectional solidification experiments,21,22,23) was confirmed by the present in-situ observations. The finding of this study is that the fine γ grains such as FCG originate in the massive-like transformation. In other words, the massive-like transformation has an influence on the γ grain structure. Quantitative observations with higher temporal and spatial resolutions will give detailed information on the formation of γ grain structures including FGC.
Selection of the massive-like transformation in continuous casting is discussed on the basis of the in-situ observations. Compared with the in-situ observation conditions, the cooling rate (R) is smaller and the temperature gradient (G) becomes nearly zero at the central region of casting. Since the moving velocity of the isothermal plane (V) is roughly proportional to R/G, a change in growth velocity of a solidifying shell from the surface to the center is not significant. Growth velocities of a solidifying shell in the continuous casting of steel are several 100 μm/s at the surface and several 10 μm/s at the center, respectively.35) As the moving velocity of the δ–γ interface in the in-situ observation should be of the same order as the velocity of the solidifying shell in the continuous casting, the moving velocity of the δ–γ interface in the continuous casting is expected to be greater than the critical velocity for the δ–γ transformation, observed by this study. Therefore, rather than the diffusion-controlled growth (the peritectic solidification), the massive-like transformation (the sequence of nucleation and growth of γ phase) can be selected from the surface to the center in the continuous casting of steel. This is an important conclusion of this study. It is also necessary to reconsider the microstructural evolution, γ grain structure and volume change during the solidification processes on the basis of the massive-like transformation.
Time-resolved and in-situ transmission imaging and X-ray diffraction imaging, using synchrotron radiation X-rays, were performed to observe the unidirectional solidification in 0.3C steel (temperature gradient: 10 K/mm and pulling rate: 50 μm/s) and the transition from the diffusion-controlled growth (the peritectic solidification) to the massive-like transformation in 0.18C steel.
(1) The diffusion-controlled growth of the γ phase was not observed and consequently the massive-like transformation was selected at a growth velocity of 50 μm/s in 0.3C steel.
(2) Fine γ grains were produced in δ grains by the massive-like transformation. The massive-like transformation in the unidirectional solidification was essentially the same as that occurring in the largely undercooled δ phase (undercooling ranged from several 10 to 100 K).
(3) Coarse γ grains were observed behind the δ–γ interface and the distances from the front of the coarse γ grains to the δ–γ interface was as short as several 100 μm.
(4) The transition from the diffusion-controlled growth to the massive-like transformation occurred when the growth velocity of the γ phase reached 5 μm/s in 0.18C steel.
(5) The driving force for nucleation of the γ phase ahead of the δ–γ interface was considered on the basis of the constitutional undercooling. In the case of manganese as a substitutional element, the driving force was caused at a growth velocity of 5 μm/s even when manganese atoms were partitioned at the δ–γ interface. In the case of carbon as an interstitial element, the thickness of the diffusion layer was as long as 2 mm at 5 μm/s. The carbon partition caused the constitutional undercooling ahead of the δ–γ interface even though the carbon diffusivity is relatively high. The thick diffusion layer suggested that the driving force induced by the carbon partition did not simply determine the nucleation positions of the γ phase.
(6) The planar δ–γ interface is unstable owing to the constitutional undercooling in the unidirectional solidification. However, neither cellular or dendritic growth was selected but rather the massive-like transformation occurred. This selection suggests that the driving force for nucleation events of the γ phase at the δ–γ interface was relatively small.
(7) The formation of γ grains, as observed in this study, is consistent with the previous work reporting that CCG formed behind the fine γ grains.
(8) Comparing the critical velocity from the diffusion-controlled growth to the massive-like transformation with the typical growth velocities of the solidifying shell in continuous casting, the massive-like transformation should be selected from the surface to the center of the castings.
The in-situ observations and measurements using synchrotron radiation X-rays were performed as general projects at BL20B2 and BL20XU of SPring-8 (JASRI), Japan. The authors acknowledge valuable discussions in a research group “Visualization of Solidification” of the 19th Committee on Steelmaking of JSPS. The authors also acknowledge financial support from Heterogeneous Structure Control: Towards Innovative Development of Metallic Structural Materials of the Industry–Academia Collaborative R&D Program (JST). The X-ray imaging technique that was developed with a funding from a Grant-in-Aid for Scientific Research (S) (No. 17H06155) allowed the authors to observe solidification phenomena and to measure diffraction spots. We thank Andrew Jackson, PhD, from Edanz Group (www.edanzediting.com/ac) for editing a draft of this manuscript.