Analysis of the Tensile Deformation Behaviors and Microstructure Characterization under Various Temperatures of MarBN Steel by EBSD

The uniaxial tensile behavior of MarBN steel with a constant strain rate of 5 × 10−5 s−1 under various temperatures ranging from room temperature to 630 °C was analyzed. This study aimed to identify the effect of the temperature on the tensile behavior and to understand the microstructure deformation by electron backscatter diffraction. The tensile results showed that the yield and ultimate tensile strength decreased with increasing temperature. Serrated flow was observed from 430 °C to 630 °C. The electron backscatter diffraction analysis showed that the low-angle grain boundaries decreased at the medium deformation and increased at the maximum deformation. In contrast, they decreased with increasing temperatures. In addition, the number of voids increased with the increasing plastic strain. As the strain increased, the voids joined together, and the tiny cracks became larger and failed. Three mechanisms were responsible for the tensile deformation failure at various temperatures: grain rotation, the formation and rearrangement of low angle grain boundaries, and void nucleation and propagation. Finally, the formation of the low-angle grain boundaries and voids under different degrees of deformation is discussed.


Introduction
In order to reduce CO 2 emissions, ultra-supercritical (USC) plants should be used to improve power generation efficiency in the 42-46% range, in which the steam temperature is at least 600 • C at 300 bar work pressure [1][2][3], which imposes a high strength requirement on the engineering materials used in this field. For the pressure vessel and piping materials applied in USC projects, the capability to withstand tensile loads in elevated temperatures is vital. MarBN steel has been chosen as a candidate material and designed for USC applications above 600 • C, in which the boron and nitrogen stabilize the martensite laths to improve the mechanical strength in high-temperature low-cycle fatigue (LCF) [4][5][6][7].
The low-cycle fatigue properties of MarBN steel and their tensile behaviors at low and medium temperatures have been reported. Barrett et al. [8,9] presented a dislocation-based constitutive model for high-temperature microstructure degradation to predict the LCF behavior at 400-600 • C. Zhang et al. [10,11] reported that cyclic softening during LCF was related to the laths and grain rotation size and indicated that the material exhibited a non-mashing behavior, and a coarsening microstructure was observed at high strain ratios. Verma et al. [12] observed the serrated flow from 250 to 350 • C at a strain rate of 10 −4 s −1 for modified 9Cr-1Mo steel, which was related to the decohesion of the interface of the carbide particles and the austenite grain boundaries. However, few investigations have focused on the tensile behavior at different temperatures and the microstructure evolution, especially using electron backscatter diffraction (EBSD) for the systematic analysis of MarBN steel. Thus, in this work, the following three issues concerning the tensile behavior are investigated with the microstructure characterization using EBSD: (1) the tensile mechanical behavior of MarBN steel at various elevated temperatures; (2) the evolution of the microstructure at high temperatures; and (3) the analysis of the failure mechanism of the material from its microscopic characteristics.
The current study examined the tensile behavior with a constant strain rate of 5 × 10 −5 s −1 at room temperature (RT), 430 • C, and 630 • C. Then, the different degrees of deformation of the microstructure were observed and analyzed by EBSD. The paper is organized as follows. In Section 2, the material and experimental procedures are described. In Section 3, the tensile behavior at RT, 430 • C, and 630 • C is presented and analyzed. In addition, the deformation of the microstructure at different temperatures observed by EBSD is presented. Then, the relationship between the tensile properties and the microstructure is discussed. Furthermore, the formation of the geometry for necessary dislocations (GNDs) and voids is deduced and analyzed. Finally, the conclusion is presented in Section 4.

Materials and Methods
Martensitic refractory steel (named MarBN steel) was used in the current study. It has a nominal composition (wt. %) of 0.10 C, 9.16 Cr, 0.06 Si, 0.20 Mn, 0.20 Mo, 0.40 Ni, 0.08 Nb, 2.95 W, 2.82 Co, 0.20 V, and Fe as balance [10]. To obtain a stable microstructure, the material was processed with a heat treatment procedure, including solution treatment, i.e., 1250 • C/2 h + 1300 • C/2 h + air cooling (AC), and an aging treatment, i.e., I: 1150 • C/6 h + AC; II: 850 • C/20 h + AC. A uniaxial tensile test (SHIMADZU Corp., Japan) was conducted with a constant strain rate of 5 × 10 −5 s −1 under different temperatures, including RT, 430 • C, and 630 • C. A cylindrical sample with a gauge length of 5 mm in diameter and 41 mm in length [10] was used to analyze the tensile property, as shown in Figure 1. At the same time, the resistance furnace (SHIMADEN Corp., Japan) was used to heat the specimens with the temperature fluctuation at ±5 • C. The test was repeated three times, and the average values were used for each specimen. EBSD was employed using an FEI Quanta 450F field emission scanning electron microscope (SEM, EDAX Corp., USA) coupled with an Oxford HKL system to obtain the crystallographic information and microstructure. Samples at every temperature, every 1000 µm × 1000 µm in size, were selected along the loading direction. The acceleration voltage of the EBSD was 15 kV, and the scan step size range was 0.05 µm to 0.2 µm. Before measuring, the samples for EBSD observation were prepared as follows: Firstly, a thin sheet of 5 mm in length, 5 mm in width, and 1 mm in thickness was taken by a cutting machine (JUNMA machinery factory, Taizhou, China). Mechanical grinding was carried out using 400#, 1000#, 2000#, and 3000# SiC water-grinding papers. Then, electron polishing was carried out at room temperature. The polishing solution composition was 7% perchloric acid and 83% ethanol, and the current density was 450 mA/cm 2 for about 1 min. Several statistical analyses were calculated from the EBSD for each sample, including the image quality (IQ), grain boundaries (GBs), grain misorientation, and kernel average misorientation (KAM). The IQ describes the quality of the Kikuchi patterns and evaluates the stored energy or dislocation density [13,14] reflected by the strain distribution. GBs represent the different grain boundaries, such as the low angle grain boundaries (LAGBs, <15 • ) and the high angle grain boundaries (HAGBs, >15 • ), which are related to the dislocation distribution and movement. KAM is the critical parameter for evaluating the plastic deformation behavior using local misorientation, indicating the dislocation density [15]. Moreover, before the tensile tests, the specimens characterized by transmission electron microscopy (TEM, Thermo Fisher Scientific Inc., Waltham, MA, USA) with energy-dispersive X-ray spectroscopy (EDS, Thermo Fisher Scientific Inc., USA). In order to determine the fraction of phase in the MarBN steel, X-ray diffraction (XRD, Bruker Corp., Bremen, Germany) tests were performed using a micro-area XRD device, Bruker D8 Focus. Co was used as the target, the wavelength (λ) was set to 0.178897 nm, the voltage was 35 kV, and the current was 40 mA.

Initial Microstructure Characterization
The microstructure of the MarBN steel before the tests was examined by EBSD and TEM, as shown in Figure 2. Figure 2a is the inverse pole figure (IPF), which contains the prior austenite grain boundaries (PAGBs) and the parallel martensite laths. The average grain size, shown in Figure 2a, was about 13.2 µm based on the statistical results of the Image-Pro software. Moreover, the fractions of HAGBs and LAGBs were approximately 61.31% and 38.69%, respectively. The 100% body-centered cubic (BCC) phase was also observed by XRD analysis performed on the base material, as shown in Figure 2b. In addition, the TEM microstructures are presented in Figure 2c,d. As seen in Figure 2c, the width of the martensite laths ranged from 11.4 nm to 21.4 nm. It was a nano lath responsible for strengthening the mechanical properties of the material. Furthermore, dislocations entangled along the PAGBs were observed. Precipitations were observed within and along the grain boundaries. Figure 2d is the enlarged view of the yellow rectangle in Figure 2c, where two types of precipitations were analyzed by EDS, the Cr-Ferich MX carbides and the Fe-rich M 3 C carbides, respectively.

Tensile Behavior and Temperature Sensitivity
The stress-strain curves with a constant strain rate, i.e., 5 × 10 −5 s −1 at various temperatures, are presented in Figure 3. The tensile properties, yield (σ s ), and ultimate tensile strength (σ ult ) decreased with the temperature increase. The previous references reported similar results at different strain rates of 5 × 10 −3 s −1 [10]. More details related to the tensile properties of the MarBN steel are presented in Table 1. In addition, to check the hardening capacity at various temperatures, the value of the ratio of ultimate tensile strength to the yield strength minus one was computed, giving 0.28, 0.37, and 0.18 at RT, 430 • C, and 630 • C, respectively. This indicated that softening behavior was apparent at 630 • C. In Figure 3a, three deformation stages are presented along the stress-strain curves at different temperatures, the elastic region fitted by a linear function, the oscillation region between the yield strength and about a 0.02 strain at 430 • C and 630 • C, softening, and finally, the failure region. The work hardening rate curves based on the true stress-true strain curves at various temperatures are presented in Figure 3b, in which the work hardening rate was significantly larger when the strain was below 0.002. Then, the oscillation behavior appeared at 430 • C and 630 • C. However, the rate became constant above the 0.02 strain, and softening and necking occurred. This behavior also depended on the temperature and agreement with the stress-strain curves, as shown in Figure 3a.  Moreover, to understand the strain hardening behavior of the MarBN steel during the plastic deformation region at various temperatures, the strain hardening exponent was used based on the Hollomon [16] and Ludwik [17] relationship as follows: where n 1 and n 2 are the strain hardening exponents; k 1 and k 2 are the material parameters; and σ and σ y are the true stress and yield strength, respectively. The double logarithmic curves fitted by Equations (1) and (2) are provided in Figure 3c,d, respectively. The slope of the curve in the stable region is defined as the strain hardening exponent, where the curves at 630 • C were the lowest using Equations (1) and (2). However, there was a difference in the curves at RT and 430 • C, resulting from the different fitting equations, as Equation (1) contained both the elastic and plastic deformation, while Equation (2) only had the plastic deformation. The trend of the strain hardening exponent fitted by Equation (1) was similar to the hardening capacity. Furthermore, the strain hardening exponent fitted by Equation (2) had better regularity, as shown in Table 1. Therefore, the strain hardening exponent n 2 was sensitive to the temperature of MarBN steel.
Furthermore, the yield strength vs. temperature is presented in Figure 4, which was fitted by the quadratic polynomial as follows: Generally, the yield strength is related to several contributions, such as the solid solution [18], the grain size [19], and the thermal activation [20], which can be neglected at RT. For the same material and experimental conditions, the yield strength at RT can reflect the solid solution and the grain size contributions. Therefore, coupling Equation (3), the thermally activated contribution is computed as follows: where σ thermal s (T) and σ s (RT) are the yield strength of the thermal activation and RT, respectively, and σ s (T) is the yield strength at a defined temperature. As shown in Figure 4, to obtain the yield strength at high temperatures, the thermally activated contribution must be negative, resulting from the disappearance of the low angle boundaries caused by the dislocation annihilation [21] and the decreasing dislocation density at high temperatures [7]. A detailed analysis is presented in the following sections.
Additionally, Figure 3a shows the apparent oscillation behavior between the yield strength and about a 0.02 strain, which was defined as the serrated plastic flow [22,23]. Clearly, there was no serrated flow at RT for the MarBN steel at the tensile strain rate of 5 × 10 −5 s −1 . It depended on the temperature from 430 • C to 630 • C at the strain rate of 5 × 10 −5 s −1 , wherein the type of serrated flow was the same, indicating that the formation mechanism at both temperatures was the same. In addition, the serrations in the stress-strain curves appeared almost from the beginning of the yield strength and disappeared at about a 0.02 strain before the ultimate tensile strength at both temperatures. Previous references for Cr-based steel [24][25][26][27] observed the dynamic strain aging (DSA) at the temperature range from 225 • C to 420 • C, which was proven by the serrated flow and negative strain hardening index [28,29]. However, in the current investigation, the serrated flow was shown at 430 • C and 630 • C, respectively. Figure 5 shows the strain hardening index at various temperatures based on the relationship σ = kε m , where σ and ε are the true stress and the true strain, respectively. As shown in Figure 5a, there was a transparent oscillation region between the yield strength and the 0.02 strain, consisting of the serrated flow region. Above the 0.02 strain, the strain hardening index presented a continuous downward trend at all temperatures. In addition, Figure 5b is the enlarged view for the yellow rectangle in Figure 5a, where the strain hardening index at RT retained a positive constant, while it presented positive and negative trends at 430 • C and 630 • C, respectively. This indicates that the DSA behavior appeared under the serrated flow region at both temperatures, resulting from the interaction between the diffusion of the solute atoms and the gliding dislocation at high temperatures, which led to mobile dislocation and tangling along the slip path [30,31]. In summary, the tensile properties of MarBN steel were temperature dependent, with the yield and tensile strength decreasing at elevated temperatures, and the work hardening rates obtained by the equation fitting showed the same trend. The fitted curve in accordance with the Hollomon equation included the entire tensile stage, i.e., the elastic and plastic deformation, while the fitted equation by Ludwik focused only on the plastic deformation. The appearance of the serrated oscillations also correlated with the effect of temperature, as shown by the dramatic fluctuations of the strain hardening index at high temperatures, while no similar serrations were observed at RT. This is indicative of the DSA behavior at elevated temperatures, where the diffusion of solute atoms was intensified and interacted with the dislocations due to the thermal activation, leading to dislocation entanglements and causing the serrated oscillations.

EBSD Analysis at Various Temperatures
The IQ map presents the Kikuchi patterns related to the lattice distortion [14,32]. Therefore, the void can be described by the black color, which lacks the Kikuchi patterns. Figure 6 shows the IQ maps of the different degrees of deformation (positions 1 and 2) at various temperatures. For the medium deformation (position 2) at different temperatures, as shown in Figure 6a,c,e, the different phases and laths had apparent sharpness and boundaries. The red arrows in Figure 6 represent the voids distributed along the PAGBs, while the green arrows are the voids along the boundaries of the laths, agreeing with previous reports [33][34][35]. Moreover, the number and size of the voids increased and decreased with increased temperatures. At 630 • C, the voids were mainly along the parallel martensite laths. However, at the maximum degree of deformation (position 1), as shown in Figure 6b,d,f, the phases seriously deformed, especially the parallel martensite laths, resulting from the gradient effect on the plastic deformation [36]. This means that the degree of deformation for the martensite laths intensified with the increasing temperatures. In addition, the number and size of the voids increased compared to those at medium deformation, suggesting that the generation and propagation of voids were associated with energy dissipation. Inhomogeneous deformation was observed in different phases. The voids regions were possible at the high KAM and GNDs because of the increased strain. The voids initiated at the martensite laths and PAGBs boundaries, forming the planar arrays of voids that led to the delamination of phases [33]. As the strain increased, the voids were joined together, indicated by the yellow rectangles, which means that the small cracks became the larger ones and eventually failed. Temperature was another crucial factor in improving the atoms diffusion and reinforcing the martensite laths decomposition, leading to the high stress concentration and the martensite lath and PAGBs boundaries. Figure 7 presents the microstructure according to the IPF and PF maps for the medium and maximum degree deformation at RT, 430 • C, and 630 • C. For the medium deformation illustrated in Figure 7a,c,e, the PAGBs and laths retained a stable geometry. Furthermore, the microstructure had no clear texture at RT. However, the texture was more evident at 430 • C and 630 • C, which means the texture of MarBN steel was related to the temperature, resulting from the decrease in the number of LAGBs, as shown in Table 2. Compared with the initial microstructure presented in Figure 2, the number of LAGBs after the medium deformation dropped significantly, which means the dislocations interacted and annihilated each other during the medium plastic deformation. The diffusion of the solute atoms increased as the temperature increased, which kept up with the movement of the dislocations and reacted with the dislocations to pin them. This explains why the serrated flow appeared at high temperatures during the tensile tests with a constant strain rate. As seen from the maximum deformation presented in Figure 7b,d,f, the PAGBs and laths did not retain a stable geometry, where the PAGBs were divided into several elongated grains due to severe deformation. As the temperature increased, the larger the number of elongated grains, and the greater the tendency to deform. In addition, the apparent texture along the {101} slip plane was observed, which indicated that the grain orientation gradually unified after the maximum plastic deformation at different temperatures. This behavior was independent of the temperatures. Figure 8 shows the Schmid factor with the slip system at RT, 430 • C, and 630 • C, respectively, using the MTEX [37] toolbox in Matlab, where the small grain size (≤5 µm) was removed to define the slip plane (red arrow) and slip direction (blue arrow). As the Schmid factor at the maximum deformation was larger than that at the medium deformation, the cracks more easily initiated under the maximum deformation. In addition, different slip systems (slip plane and direction) were active in the different phases, such as PAGBs and martensite laths, leading to grain rotation to maintain the deformation continuity under an inhomogeneous microstructure during the plastic strain [38]. It seems that the slip systems increased with the increasing temperatures and strain. Moreover, the grain rotation was responsible for the texture changes [39] under the maximum deformation at various temperatures, as shown in Figure 7b,d,f.
In conclusion, the maps of the IQ, IPF, PF, and Schmid factor were analyzed at various temperatures by means of the EBSD, with samples taken at position 1 near the fracture and position 2 away from the fracture, as shown in Figure 6. The conclusions that can be drawn are as follows: (1) The number of the voids formed inside the material increased with the rise in the temperature, and the size reduced; (2) the increasing temperature led to a reduction in the number of LAGBs, which made the texture more evident; (3) the slip systems within the material increased under the effect of thermal activation; (4) the number of the voids formed at position 1 was greater and larger in size compared to position 2; the geometry of the grains at position 1 was unstable, showing an elongated and twisted morphology, indicating that the grains were rotated; moreover, the Schmid factor at position 1 was also considerably larger. Therefore, the tensile deformation mechanism of MarBN steel is temperature dependent, and the difference in position also influences the magnitude of the deformation.

Formation of the GNDs and Voids
In addition, the KAM measures the local grain misorientation, which effectively describes the deformation regions [40]. The GNDs are a critical indicator of the geometric continuity in the deformation state [41], resulting from the strain gradient, which rotates the continuum field from point to point [42]. Figures 9-11 show the KAM and GND maps under the medium (a,c) and maximum (b,d) degrees of deformation at RT, 430 • C, and 630 • C, respectively. The KAM at the medium deformation was lower than that at the maximum deformation at all temperatures. The GNDs had a similar distribution. The GND density decreased with increasing temperature due to the increased annihilation, which agreed with the previous references [7,10]. The KAM and GNDs at the medium deformation (Figures 9-11a,c) distributed along the grain boundaries (GBs) and martensite laths due to the inhomogeneous grain structure. This indicates that the GBs were the stress concentration location due to the dislocation pile-up during the plastic strain. Furthermore, as the strain increased, the strain gradient increased and arrived at the maximum before the final failure, as shown in Figures 9-11b,d. The KAM increased faster and distributed within all the grains, especially at the elongated PAGBs and laths, which indicated that the grains were inhomogeneously deformed to continue the geometry, and the plastic strain incompatibility of the PAGBs and laths was also shown, resulting from the different soft PAGBs and hard martensite laths [34]. The GNDs had similar rules. However, the GNDs were mostly generated at the small laths and elongated grains, where high LAGBs were observed, as shown in Table 2. Thus, the number of LAGBs was proportional to the GND density [43]. As discussed above, with the increase in the temperature, the percentage of the LAGBs inside the material decreased, and the density of the dislocations reduced due to the intensification of dislocation annihilation at high temperature, resulting in a decrease in the density of GND. Furthermore, by comparing the KAM and GND maps at two different positions, it can be concluded that the KAM and GND at position 2 were mainly distributed at the GBs, where the dislocations stacked up during the deformation and led to stress concentration. However, the KAM and GND at position 1 were distributed along the entire grains, indicating that as the inhomogeneous deformation intensified, the grains began to rotate and distort to maintain the geometric continuity.   As seen in Figure 2, the microstructure of the MarBN steel included soft PAGBs and hard martensite laths. When the uniaxial tensile loading was applied in the different phases, the inhomogeneous deformation for different phases appeared due to the misorientation, resulting in the formation of GNDs along the PAGBs and the GBs of laths, as presented in Figures 9-11. As shown in Table 2, compared with an initial number of LAGBs, the LAGBs at the medium deformation decreased, resulting from the dislocation annihilation caused by the dynamic recrystallization behavior (DRX). This can lead to the serrated flow behavior reported by the nickel-based superalloy during compression tests [22]. Therefore, the initial work hardening at the low strain, shown in Figure 3, resulted from a balance between the GNDs and DRX. Temperature played another role in decreasing the dislocation density. However, its effect was weak due to the small gap in the LAGBs at the medium deformation, as shown in Table 2. In addition, the GNDs were rearranged according to the distribution of the LAGBs. Then, a high GND density was formed at the maximum deformation, resulting from the high LAGB density caused by the small and elongated GBs. The dislocations glided across the PAGBs and martensite laths. Then, the GNDs concentrated on the GBs, as shown in Figures 9-11b,d. This behavior resulted in more work-hardening before failure, as shown in Figure 3, controlled by the gradient of strain and temperature. With increasing GNDs, the strain gradient was more critical in controlling the KAM and GNDs than the temperature for MarBN steel during the uniaxial tensile process.
To sum up, by comparing the results of the EBSD analysis at various temperatures under two different strain gradients, the following conclusions can be drawn: (1) The number and size of the voids formed inside the material increased as the deformation intensified, and the microcracks formed by the interconnection of the voids eventually resulted in failure; (2) in the region near the fracture, the grains were elongated, and the geometry presented a torsional bending morphology, indicating that the degree of deformation was intense; (3) with the elongation of the grains, the percentage of the LAGBs in the maximum deformation region increased, causing the increase in the GNDs in the whole grains, which ultimately generated stress concentration and led to failure. In general, three mechanisms were responsible for the tensile deformation failure at various temperatures, the void nucleation and propagation, the grain rotation, and the formation and rearrangement of the GNDs, as illustrated in Figure 12. Moreover, the temperature effectively decreased the number of the LAGBs and increased the solute atoms diffusion, which further forced the DXR and martensite laths to decompose, resulting in the high stress concentration and the martensite lath and PAGBs boundaries during the tensile deformation.

Conclusions
The present study investigated the uniaxial tensile deformation behavior of the MarBN steel with a constant strain rate under RT, 430 • C, and 630 • C using EBSD. The temperature dependence on the microstructure and deformation mechanism was observed and studied systematically to obtain the following conclusions: (1) The tensile behavior was affected by the temperature of the MarBN steel. As the temperature increased, the yield and ultimate tensile strength decreased. The serrated flow was only observed in the temperature range from 430 • C to 630 • C, related to the DSA behavior at high temperatures. (2) Three mechanisms were responsible for the tensile deformation failure at various temperatures, the grain rotation, the formation and rearrangement of the GNDs, and the void nucleation and propagation. (3) The KAM and GNDs density had a similar distribution, resulting from the strain incompatibility between different phases. In addition, the temperature played another crucial role in decreasing the number of the LAGBs during the tensile deformation. (4) The tensile behavior at the medium deformation resulted from a balance between the GNDs and DRX, resulting from the disappearance of the LAGBs caused by the dislocation annihilation. However, the tensile behavior at the maximum deformation was determined by the gradient of strain and temperature. The strain gradient mainly controlled the KAM and GNDs, rather than the temperature in this process. (5) The number of voids increased with the increasing plastic strain. As the strain increased, the voids were joined together, and the small cracks became larger cracks and finally failed. The temperature was crucial to improving the atoms diffusion, reinforcing the martensite laths' decomposition, resulting in a high stress concentration along the martensite lath and PAGB boundaries.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data that support the findings of this work are available from the corresponding author upon reasonable request.