Processing, Microstructures and Mechanical Properties of a Ni-Based Single Crystal Superalloy

: A second-generation Ni-based superalloy has been directionally solidiﬁed by using a Bridgman method, and the key processing steps have been investigated with a focus on their e ﬀ ects on microstructure evolution and mechanical properties. The as-grown microstructure is of a typical dendrite structure with microscopic elemental segregation during solidiﬁcation. Based on the microstructural evidence and the measured phase transformation temperatures, a step-wise solution treatment procedure is designed to e ﬀ ectively eliminate the compositional and microstructural inhomogeneities. Consequently, the homogenized microstructure consisting of γ / γ (cid:48) phases (size of γ (cid:48) cube is ~400 nm) have been successfully produced after a two-step (solid solution and aging) treatment. The mechanical properties of the resulting alloys with desirable microstructures at room and elevated temperatures are measured by tensile tests. The strength of the alloy is comparable to commercial monocrystalline superalloys, such as DD6 and CMSX-4. The fracture modes of the alloy at various temperatures have also been studied and the corresponding deformation mechanisms are discussed.


Introduction
Ni-based single crystal (SX) superalloys have been widely used in turbine blades of aero-engines because of their excellent combination of mechanical properties and corrosion resistance under a wide range of elevated temperatures. In order to design aero-engines with increased efficiency and performance, as well as with reduced carbon dioxide emission, the ultimate driving force is to increase the turbine operating temperature and therefore temperature capability of Ni-based SX superalloys [1][2][3]. Many decades of industrial and academic research have led to the development of a class of multi-component complex alloys, typically containing more than eight major alloying elements, most of which are refractory metals that provide the desirable microstructures and strengths for high temperature applications [4]. For example, a commercial superalloy, CMSX-10, has a nominal composition of 69.6Ni, 8.0Ta, 6.0 Re, 5.7Al, 5.0W, 3.0Co, 2.0Cr, 0.4Mo, 0.2Ti, 0.1Nb, and 0.03Hf (wt.%) [5]. As a general designing strategy, to push the superalloys to an even higher temperature capability, more refractory elements are being added into these alloys [6][7][8][9], which, however, cause dendritic microstructures after single crystal growth and severe elemental segregation between the dendrite arms and the interdendritic regions. These segregation sites are potential places for creep failure [10,11].
The elemental segregation problem is caused by the constitutional supercooling during solidification [12,13]. It has been found that solute segregation occurs near the liquid-solid interface, as elements with higher melting points (W, Re, etc.) solidify earlier, while elements with lower melting

Single Crystal Growth
A typical second-generation Ni-based SX superalloy is selected to investigate microstructural evolution during single crystal growth, heat treatment, and mechanical testing. Table 1 lists the measured compositions of the selected alloy, which is close to those of the commercial superalloy DD6 [31]. Several well-studied Ni-based SX superalloys with similar compositions, including CMSX-4, 7, and 8, are also included in Table 1 for comparison [32][33][34]. Cylindrical rods with <001> orientation are directionally solidified to obtain single crystals by using the traditional Bridgman method, which is widely adopted for the growth of single crystal turbine blades or experimental rods. The details of single crystal growth can be found elsewhere [4]. The diameter of our experimental cylindrical rods is about 11 mm.

. Heat Treatment
In order to design the solution heat treatment processes correctly, the phase transformation temperatures (e.g., γ solvus and melting temperatures) are measured by using the differential scanning calorimetry (DSC) technique. The DSC heating cycle is performed at a rate of 10 • C/min from room temperature (RT) to 1550 • C, using a Netzsch DSC 404 instrument (NETZSCH-GmbH, Germany) with an Ar flow of 20 mL/min.
The heat treatment experiments are carried out in a vacuum/Ar protected tube furnace, in which the heating rate, holding temperatures, and times can be program controlled. Before heating, the tube furnace is evacuated to a vacuum of <1 Pa and then followed by refilling with high purity argon. The evacuation and refilling processes repeat three times to reduce the impact of oxidation on the sample surfaces. The length of the hot zone and temperature of the furnace are routinely calibrated. The detailed heat treatment procedures are listed in Table 2. Table 2. Stepwise solid solution heat treatment procedures, as designed in this study.

Sample
Heat Treatment

Microstructural Characterization
Samples with a thickness of 1-2 mm are cut, and metallographically prepared to observe their microstructures. The metallographic procedure includes grinding down with 1200 grit SiC paper, followed by mechanical polishing. The final polishing media is a 40 nm colloidal silica suspension, which can effectively remove the damaged layer induced by the previous steps. To better reveal the microstructures, some of the polished samples are electrochemically etched in a 10% H 3 PO 4 + H 2 O solution at 12 V at room temperature.
The secondary electron (SE) and backscattered electron (BSE) images, the energy dispersive spectroscopy (EDS) results, and the electron backscattered diffraction (EBSD) patterns are all captured in a FEI quanta 650 SEM (ThermoFisher, Waltham, MA, USA) equipped with Oxford EDS and EBSD detectors, using an accelerating voltage of 15 kV and a 10 mm working distance. Crystal orientations are determined by EBSD with the test area larger than 200 × 200 µm 2 .

Mechanical Testing
Dog bone-shaped tensile specimens with a gauge section of 3.1 × 1.0 × 12.5 mm 3 are electro-discharge machined (EDM) from the SX rods after heat treatment. Before cutting, the SX rods are carefully reoriented to ensure that the tensile direction is [001], and other two side surfaces of the tensile plates are (100) and (010), respectively. All surfaces of the gauge section are carefully ground down with 800 grit SiC paper to eliminate micro-cracks that may be introduced by electro-discharge machining.
Tensile tests at room and elevated temperatures are conducted in air on a screw-driven mechanical testing machine equipped with an induction heater. A thermocouple, which is directly attached to the sample gauge section, is used to monitor testing temperature within ± 3 • C of the set temperature. All the specimens are tested at a constant displacement rate of 0.75 mm/min, which corresponds to an engineering strain rate of 1 × 10 −3 s −1 . Yield strengths are calculated using the 0.2% offset method.

As-Grown Microstructure
The as-grown Ni-based single crystals right after growth are shown in Figure 1. After removing the ceramic mold, the rods are ready for further experiments. EBSDs are conducted to determine the orientation and the quality of single crystals. One example is shown as Figure 1b (Figure 1c) reveal that the growth direction is within 8 • of the [001] direction, indicating a successful growth. Indeed, these rods are single crystals without high angle grain boundaries.
down with 800 grit SiC paper to eliminate micro-cracks that may be introduced by electro-discharge machining.
Tensile tests at room and elevated temperatures are conducted in air on a screw-driven mechanical testing machine equipped with an induction heater. A thermocouple, which is directly attached to the sample gauge section, is used to monitor testing temperature within ± 3 °C of the set temperature. All the specimens are tested at a constant displacement rate of 0.75 mm/min, which corresponds to an engineering strain rate of 1 × 10 −3 s −1 . Yield strengths are calculated using the 0.2% offset method.

As-Grown Microstructure
The as-grown Ni-based single crystals right after growth are shown in Figure 1. After removing the ceramic mold, the rods are ready for further experiments. EBSDs are conducted to determine the orientation and the quality of single crystals. One example is shown as Figure 1b Figure 2 shows the as-gown microstructures of the selected superalloy, which are typical dendrite structures, resulting from compositional segregation during solidification. Under BSE imaging mode, four-leaf clover-shaped dendrites with a bright contrast can be clearly observed, and the interdendritic regions are relatively dark (Figure 2a). Because brightness is proportional to the averaged atomic number (Z) of the sample under BSE mode, high-Z elements (e.g, W, Re, etc.) may segregate at the dendrite arm. Therefore, EDS mapping is subsequently conducted to reveal the composition distribution at both dendrite and interdendritic regions, and the results are shown in Figure 3. It is found that the dendrite arms are rich in Re, W, and Mo, and the interdendritic regions are rich in Ta, Al, and Nb, while Cr, Co, and Ni are distributed relatively uniformly in both regions.  Figure 2 shows the as-gown microstructures of the selected superalloy, which are typical dendrite structures, resulting from compositional segregation during solidification. Under BSE imaging mode, four-leaf clover-shaped dendrites with a bright contrast can be clearly observed, and the interdendritic regions are relatively dark (Figure 2a). Because brightness is proportional to the averaged atomic number (Z) of the sample under BSE mode, high-Z elements (e.g, W, Re, etc.) may segregate at the dendrite arm. Therefore, EDS mapping is subsequently conducted to reveal the composition distribution at both dendrite and interdendritic regions, and the results are shown in Figure 3. It is found that the dendrite arms are rich in Re, W, and Mo, and the interdendritic regions are rich in Ta, Al, and Nb, while Cr, Co, and Ni are distributed relatively uniformly in both regions.    The elemental segregation results from the constitutional supercooling during solidification. It has been found that solute segregation occurs near the liquid-solid interface [12], and elements with higher melting points (W, Re, etc.) solidify earlier and then segregate at dendrites, while elements with lower melting points (Al) and those eutectic forming elements (Al, Nb, Ta, etc.) segregate at the The elemental segregation results from the constitutional supercooling during solidification. It has been found that solute segregation occurs near the liquid-solid interface [12], and elements with higher melting points (W, Re, etc.) solidify earlier and then segregate at dendrites, while elements with lower melting points (Al) and those eutectic forming elements (Al, Nb, Ta, etc.) segregate at the interdendritic region. Consistent with this predication, there are coarse γ phases and eutectic microstructures observed in the interdendritic region ( Figure 2b). The darker blocks (white arrows in Figure 2b) and the white dots (dark arrows in Figure 2b) are eutectic and coarse γ phases, respectively. In additions, a shrinkage cavity near eutectic phases is also frequently observed (Figure 2c), which may result from insufficient liquid at the latest stage of solidification. The microstructure of dendrite cores is relatively simple and homogeneous. A typical example is shown in Figure 2d, which consists of γ precipitates, with an average size of about 350 nm, uniformly distributed in the γ phase.

Heat Treatment
To eliminate the compositional and microstructural inhomogeneity, the as-grown alloy should be solid solution (SS) treated. Because the SS treatment is a solid-state transformation and is diffusion controlled, the principle of the temperature selection must be well above the temperature at which γ precipitates totally dissolve into the γ phase. Moreover, in the meantime, incipient melting must be avoided. In order to establish the heat treatment processes correctly, the phase transformation temperatures (e.g., γ solvus and melting temperatures) are determined from DSC measurements. Figure 4 is a DSC heating curve of our alloy, and from this curve, the γ solvus temperature and incipient melting temperature of the as-grown single crystals can be determined as 1260 • C and 1350 • C, respectively. The SS temperature of the alloy should be above 1260 • C, but below 1350 • C, therefore, the range of 1280-1350 • C is defined as the SS window for our alloy.
avoided. In order to establish the heat treatment processes correctly, the phase transformation temperatures (e.g., γ΄ solvus and melting temperatures) are determined from DSC measurements. Figure 4 is a DSC heating curve of our alloy, and from this curve, the γ΄ solvus temperature and incipient melting temperature of the as-grown single crystals can be determined as 1260 °C and 1350 °C, respectively. The SS temperature of the alloy should be above 1260 °C, but below 1350 °C, therefore, the range of 1280-1350 °C is defined as the SS window for our alloy. Because the SS treatment is a diffusion-controlled process, higher temperatures and a longer processing time are needed to enhance the diffusion of solute elements. It is well known that increasing temperature rather than elongating the processing time is more efficient both in energy and the time needed for the heat treatment. However, the local inception of melting limits the highest temperatures that can be selected and the furnace capability becomes more critical at higher temperatures. With all those factors considered, eight stepwise SS heat treatments (see Table 2, designated as A-H) have been designed, and their corresponding microstructures are characterized.
As shown by the low-magnification BSE images in Figure 5, as the processing procedure varies from sample A to E (Figure 5a-e), the contrast of the dendrite arm decreases, and the shape of dendrites converts from four-leaf clover in sample A to the "+"shape in sample E. Meanwhile, the edge-contrast between dendrite arms and the interdendritic region is reduced, indicating that the stepwise heat treatment (HT) with a higher SS temperature and longer time is effective in alleviating the dendritic segregation. It is also found that in the final step above of 1330 °C/4 h + air cooling (AC) Because the SS treatment is a diffusion-controlled process, higher temperatures and a longer processing time are needed to enhance the diffusion of solute elements. It is well known that increasing temperature rather than elongating the processing time is more efficient both in energy and the time needed for the heat treatment. However, the local inception of melting limits the highest temperatures that can be selected and the furnace capability becomes more critical at higher temperatures. With all those factors considered, eight stepwise SS heat treatments (see Table 2, designated as A-H) have been designed, and their corresponding microstructures are characterized.
As shown by the low-magnification BSE images in Figure 5, as the processing procedure varies from sample A to E (Figure 5a-e), the contrast of the dendrite arm decreases, and the shape of dendrites converts from four-leaf clover in sample A to the "+"shape in sample E. Meanwhile, the edge-contrast between dendrite arms and the interdendritic region is reduced, indicating that the stepwise heat treatment (HT) with a higher SS temperature and longer time is effective in alleviating the dendritic segregation. It is also found that in the final step above of 1330 • C/4 h + air cooling (AC) (sample E), a further increase in temperature by 10 or 20 • C (i.e., samples F-G) becomes less effective (Figure 5f,g). In sample H (final step of 1360 • C/4 h), the incipient melting has been found, as indicated by the white arrows in Figure 5h, which is consistent with the SS window obtained by DSC. Note that there are some black dots located at the interdendritic region in Figure 5a-h. These dots are probably shrinkage cavities, a solidification defect. Because SX superalloys are generally used in solidification plus heat treatment conditions, without thermomechanical processing, the shrinkage cavities normally exist inside the materials and might serve as initiation sites of failure during deformation. Minimizing the amounts and sizes of the shrinkage cavities during solidification should be an important research topic in future studies.
The ultimate goal of heat treatment is to obtain desirable two-phase γ/γ microstructures in such Ni-based superalloys. Besides the SS treatment, aging at a relatively low temperature is also required to obtain cuboid γ phases with appropriate sizes and with uniform distributions in the γ matrix. In order to better understand the effects of SS heat treatments on the final microstructures after the full cycle of heat treatment, samples A-G are also subjected to an aging treatment (1120 • C/4 h AC + 870 • C/24 h AC), as shown by their microstructural images in Figure 6. These aged samples are slightly etched so as to slightly remove the γ phase for better phase contrast under both SE and BSE imaging modes. Because both shrinkage cavities and eutectics are formed at the final stage of solidification in the interdendritic regions, the high magnification SEM images are focused on the microstructures of regions near shrinkage cavities and eutectics (Figure 6e-h). In the aged sample A, some residual eutectics and coarse γ phases still exist, indicated by white arrows in Figure 6a,e. In contrast, eutectics are not found in the other aged samples at low magnification, as shown in Figure 6b-d. However, a further high-magnification study of Figure 6b leads to the appearance of some residual eutectics in Figure 6f, especially in the interdendritic regions. Further increasing temperature to 1315 • C and above (Figure 6c,d,g,h), neither residual eutectics nor coarse phases is observed, even in high magnificent images (Figure 6g,h), therefore, it is concluded that the final SS treatment temperature need to be higher than 1315 • C.
Crystals 2020, 10, x FOR PEER REVIEW 7 of 14 (sample E), a further increase in temperature by 10 or 20 °C (i.e., samples F-G) becomes less effective (Figure 5f,g). In sample H (final step of 1360 °C/4 h), the incipient melting has been found, as indicated by the white arrows in Figure 5h, which is consistent with the SS window obtained by DSC. Note that there are some black dots located at the interdendritic region in Figure 5a-h. These dots are probably shrinkage cavities, a solidification defect. Because SX superalloys are generally used in solidification plus heat treatment conditions, without thermomechanical processing, the shrinkage cavities normally exist inside the materials and might serve as initiation sites of failure during deformation.
Minimizing the amounts and sizes of the shrinkage cavities during solidification should be an important research topic in future studies.  Table 2).   Table 2). illustration of the three-dimensional structure of γ΄ precipitates, the γ phase has been selectively etched away, as shown in Figure 7b. Cuboid γ΄ precipitates with an average size of ~400 nm are found to be distributed uniformly in the γ phase with the volume percentage of ~70%. In particular, the γ΄ cubes are homogeneously distributed in both the previous dendrite arms and interdendritic regions, which make the entire sample have uniform microstructures. All the following mechanical testing specimens are subjected to the same newly designed heat treatment processes that lead to the microstructure in Figure 7. The dendritic microstructural evolution, along with each step of SS heat treatment, can be used to justify the choice of processing temperatures and durations. The main purpose of step one (1290 • C/1 h) is to prevent temperature overshoot and dissolve γ precipitates. The second step (1300 • C/2 h) could effectively eliminate the majority of γ phases, including eutectics. The third step (1315 • C/2 h) further eliminates residual eutectics, thus improving the microstructural homogeneity. The aim of the subsequent heat treatments is to make alloying elements diffuse quickly and sufficiently to eliminate the elemental segregations to improve compositional homogeneity over the entire single crystal. Solid solution heat treatment is a diffusion-controlled process, so it is obvious that a slight increase in SS temperature and prolonged time maybe beneficial. Thus, a final mutli-step SS heat treatment protocol (1290 • C/1 h + 1300 • C/2 h + 1315 • C/2 h + 1325 • C/4 h + 1330 • C/4 h AC) is designed for our single crystals.
Furthermore, combined with the subsequent aging treatment (1120 • C/4 h AC + 870 • C/24 h AC), the desirable microstructures of a typical γ/γ two-phase structure are obtained (Figure 7). For a better illustration of the three-dimensional structure of γ precipitates, the γ phase has been selectively etched away, as shown in Figure 7b. Cuboid γ precipitates with an average size of~400 nm are found to be distributed uniformly in the γ phase with the volume percentage of~70%. In particular, the γ cubes are homogeneously distributed in both the previous dendrite arms and interdendritic regions, which make the entire sample have uniform microstructures. All the following mechanical testing specimens are subjected to the same newly designed heat treatment processes that lead to the microstructure in Figure 7.

Tensile Properties
Tensile tests at room (RT) and various elevated temperatures are conducted to investigate the mechanical properties of our optimized superalloys with representative microstructures, as shown in Figure 7. The tensile loads are applied parallel to the [001] direction, which is the crystal growth

Tensile Properties
Tensile tests at room (RT) and various elevated temperatures are conducted to investigate the mechanical properties of our optimized superalloys with representative microstructures, as shown in Figure 7. The tensile loads are applied parallel to the [001] direction, which is the crystal growth direction. Other exposed surfaces of the tensile sample were (010) and (100) planes. Figure 8a shows the engineering stress-strain curves, and Figure 8b is a plot of the 0.2% off-set yield strength (YS) and ultimate tensile strength (UTS) as a function of the testing temperature.
When the testing temperature is below 750 • C, the samples fail after continuous work hardening. As the temperature increases to 750 • C from RT, YS and UTS both increase slightly (YS from 1010 to 1059 MPa, UTS from 1155 to 1253 MPa), and the elongation to failure decreases from 18% tõ 7%. The strength increase from RT to 750 • C may be caused by the anomalous yielding effect of the L1 2 structured γ phase, which originated from the cross-slip of the anti-phase boundary (APB) from the {111} plane to the {001} plane and thus the formation of immobile Kear-Wilsdorf locks [35]. As cross-slip is thermally activated, the strength increases. This phenomenon in Ni-based single SX alloys is the so-called "intermediate temperature brittleness".
When the temperature is above 850 • C, the alloy experiences a very short period of strain hardening (less than~4% plastic strain) in the stress-strain curve, followed by strength drops as the strain increases. This might be associated with the occurrence of large scale cross-slip, which facilitates dislocation annihilation [36]. The reduction of the cross-sectional area would also result in the drop-off of the engineering stress-strain curve before failure. For example, the YS and UTS of the alloy at 1100 • C are 356 and 413 MPa, the stress continuously drops in the stress-strain curve after 413 MPa and the elongation to fracture is~26%. In this temperature range, consistent with the reduction in the cross-sectional area, necking is normally observed. Overall, the tensile strength of our alloy is similar to the typical commercial CMSX-4 and DD6 alloys [32,37].

Fracture Analysis
Fracture analysis after the tensile testing was examined by SEM. Based on the fracture surface and microstructures near the fracture site, fracture modes can be divided into three groups, including low temperature (RT), intermediate temperatures (650-750 °C), and high temperatures (>850 °C). The microstructures of these samples fractured at RT, 650 °C, and 1100 °C are shown as examples in Figures 9-11, respectively.
The fracture surface at RT shows some ductile features, and no significant necking occurs ( Figure  9a,b), which is consistent with the fair ductility (18%) and the high work hardening (Figure 8a). A high density of zigzag slip traces along the {110} direction and ductile fracture features is clearly observed (Figure 9). High-magnification SEM images reveal a relatively brittle cleavage fracture in the γ΄ cubes, but a narrow γ phase surrounding the γ΄ cubes shows pull-out necking to a very sharp point, indicating that the γ phase is rather ductile (the inset of Figure 9a). Figure 9c shows the side view microstructure near the fracture surface. Rhomboid instead of rectangle shaped γ΄ phases and

Fracture Analysis
Fracture analysis after the tensile testing was examined by SEM. Based on the fracture surface and microstructures near the fracture site, fracture modes can be divided into three groups, including low temperature (RT), intermediate temperatures (650-750 • C), and high temperatures (>850 • C). The microstructures of these samples fractured at RT, 650 • C, and 1100 • C are shown as examples in Figures 9-11, respectively. the γ΄ cubes, but a narrow γ phase surrounding the γ΄ cubes shows pull-out necking to a very sharp point, indicating that the γ phase is rather ductile (the inset of Figure 9a). Figure 9c shows the side view microstructure near the fracture surface. Rhomboid instead of rectangle shaped γ΄ phases and shear traces through the γ΄ cubes (red arrows) are clearly observed. All these observations indicate that at room temperature, the γ phase contributes most of the plasticity and the γ΄ phase is relatively hard and brittle. The load partitioning in these two phases is responsible for the observed work hardening, and thus the less significant necking prior to ductile fracture. As compared to the RT samples, the fracture surface at 650 °C is smoother and contains fewer shear traces (Figure 10a,b), indicating a reduced plasticity of the alloy at this temperature, which is consistent with the so-called "intermediate temperature brittleness" [38][39][40]. However, ductile dimples are still observed in the γ phases, similar to the fracture surface tested at RT (inset in Figure  10a), but the center of the γ΄ phase is very smooth, indicating a cleavage fracture of the intermetallic phase. Additionally, little change from their original cuboidal shapes is observed in the γ΄ phases near the fracture surface (Figure 10c). It can be concluded that γ΄ phases have very little plasticity at this temperature. Fractures at intermediate temperatures may be caused by the rapid propagation of a crack at the initiation of shearing at an angle of about 45° to the tensile axis. The mechanism responsible for this shear band may be related to an increase in the strength ratio of these two phases, of which the validation requires an accurate measurement of the mechanical properties of individual phases. As the testing temperature increases to 850 °C, the engineering stress first slightly increases (work hardening) and then continuously decreases until the sample fractures (Figure 8a), indicating the necking of the samples, consistent with the photograph of a necked sample (the inset of Figure  11a). Figure 11a also shows the low magnification side view BSE image of the fractured sample, in which bright dendrite cores are marked out by white arrows. It is obvious that there are a large amount of micro-voids in the interdendritic region after the sample fails, which indicates that failure starts from the interdendritic regions. Different from those at RT and intermediate temperature, slip/shear traces have not been observed, which indicates that deformation might be diffusional rather than slip plasticity. The γ/γ΄ two-phase structure near the fracture surface is also completely different from the undeformed one. The γ΄ phase has transformed from squares to elongated rounded rectangles, with the long axis along the tensile direction. Moreover, the γ phase perpendicular to the tensile direction become broader, and the γ phase parallel to the tensile direction shrinks, similar to the rafting microstructure during creep [29].  phase. Additionally, little change from their original cuboidal shapes is observed in the γ΄ phases near the fracture surface (Figure 10c). It can be concluded that γ΄ phases have very little plasticity at this temperature. Fractures at intermediate temperatures may be caused by the rapid propagation of a crack at the initiation of shearing at an angle of about 45° to the tensile axis. The mechanism responsible for this shear band may be related to an increase in the strength ratio of these two phases, of which the validation requires an accurate measurement of the mechanical properties of individual phases. As the testing temperature increases to 850 °C, the engineering stress first slightly increases (work hardening) and then continuously decreases until the sample fractures (Figure 8a), indicating the necking of the samples, consistent with the photograph of a necked sample (the inset of Figure  11a). Figure 11a also shows the low magnification side view BSE image of the fractured sample, in which bright dendrite cores are marked out by white arrows. It is obvious that there are a large amount of micro-voids in the interdendritic region after the sample fails, which indicates that failure starts from the interdendritic regions. Different from those at RT and intermediate temperature, slip/shear traces have not been observed, which indicates that deformation might be diffusional rather than slip plasticity. The γ/γ΄ two-phase structure near the fracture surface is also completely different from the undeformed one. The γ΄ phase has transformed from squares to elongated rounded rectangles, with the long axis along the tensile direction. Moreover, the γ phase perpendicular to the tensile direction become broader, and the γ phase parallel to the tensile direction shrinks, similar to the rafting microstructure during creep [29].  The fracture surface at RT shows some ductile features, and no significant necking occurs (Figure 9a,b), which is consistent with the fair ductility (18%) and the high work hardening (Figure 8a). A high density of zigzag slip traces along the {110} direction and ductile fracture features is clearly observed (Figure 9). High-magnification SEM images reveal a relatively brittle cleavage fracture in the γ cubes, but a narrow γ phase surrounding the γ cubes shows pull-out necking to a very sharp point, indicating that the γ phase is rather ductile (the inset of Figure 9a). Figure 9c shows the side view microstructure near the fracture surface. Rhomboid instead of rectangle shaped γ phases and shear traces through the γ cubes (red arrows) are clearly observed. All these observations indicate that at room temperature, the γ phase contributes most of the plasticity and the γ phase is relatively hard and brittle. The load partitioning in these two phases is responsible for the observed work hardening, and thus the less significant necking prior to ductile fracture.
As compared to the RT samples, the fracture surface at 650 • C is smoother and contains fewer shear traces (Figure 10a,b), indicating a reduced plasticity of the alloy at this temperature, which is consistent with the so-called "intermediate temperature brittleness" [38][39][40]. However, ductile dimples are still observed in the γ phases, similar to the fracture surface tested at RT (inset in Figure 10a), but the center of the γ phase is very smooth, indicating a cleavage fracture of the intermetallic phase. Additionally, little change from their original cuboidal shapes is observed in the γ phases near the fracture surface (Figure 10c). It can be concluded that γ phases have very little plasticity at this temperature. Fractures at intermediate temperatures may be caused by the rapid propagation of a crack at the initiation of shearing at an angle of about 45 • to the tensile axis. The mechanism responsible for this shear band may be related to an increase in the strength ratio of these two phases, of which the validation requires an accurate measurement of the mechanical properties of individual phases.
As the testing temperature increases to 850 • C, the engineering stress first slightly increases (work hardening) and then continuously decreases until the sample fractures (Figure 8a), indicating the necking of the samples, consistent with the photograph of a necked sample (the inset of Figure 11a). Figure 11a also shows the low magnification side view BSE image of the fractured sample, in which bright dendrite cores are marked out by white arrows. It is obvious that there are a large amount of micro-voids in the interdendritic region after the sample fails, which indicates that failure starts from the interdendritic regions. Different from those at RT and intermediate temperature, slip/shear traces have not been observed, which indicates that deformation might be diffusional rather than slip plasticity. The γ/γ two-phase structure near the fracture surface is also completely different from the undeformed one. The γ phase has transformed from squares to elongated rounded rectangles, with the long axis along the tensile direction. Moreover, the γ phase perpendicular to the tensile direction become broader, and the γ phase parallel to the tensile direction shrinks, similar to the rafting microstructure during creep [29].
In principle, the γ phase is the predominant phase in volumes, and therefore appears to dominate the mechanical properties of the Ni-based SX superalloy. As a consequence, the fracture analysis in the present work shows that the γ phase is ductile at all testing temperatures, while the γ phase exhibits fair ductility at RT, limited ductility at intermediate temperatures (650-750 • C), and enhanced ductility at high temperatures (>850 • C).

Summary and Conclusions
To reveal the microstructural evolution of Ni-based SX superalloys during processing and to understand the processing-microstructure-property relationship, a second-generation Ni-based SX superalloy was successfully prepared by directional solidification along the [001] direction by using the Bridgman method. Based on microstructural observations, the main conclusions can be summarized as follows: (1) The as-grown microstructure is of a typical dendrite and there are significant compositional and microstructural differences between the dendritic arms and interdendritic regions. Refractory elements Re, W, and Mo segregate at the dendritic arms. Meanwhile, γ -forming elements, as well as eutectic-forming elements (Al, Ta, Nb), segregate at the interdendritic region, resulting in coarse γ precipitates and eutectic microstructures. A solution heat treatment procedure (1290 • C/1 h + 1300 • C/2 h + 1315 • C/2 h + 1325 • C/4 h + 1330 • C/4 h AC) has been identified to effectively eliminate eutectics at interdendritic regions and achieve compositional homogeneity without incipient melting. Combined with the subsequent aging heat treatment (1120 • C/4 h AC + 870 • C/24 h AC), homogeneous γ/γ two-phase structures (size of γ precipitate around 400 nm) have been obtained.
(3) The tensile strength of our alloy is comparable to typical commercial CMSX-4 and DD6 alloys.
The strength of the alloy increases with an increase in temperature from RT to 750 • C, 1059 MPa YS, and 1253 MPa UTS at 750 • C vs. 1010 MPa YS and 1155 MPa UTS at RT. Above this temperature, both YS and UTS decrease with the increase in temperature. (4) Subsequent fracture analysis demonstrates that the γ phase is ductile at all test temperatures but the γ phase appears to dominate the tensile properties. At RT, the γ phase exhibits a small degree of ductility, and the alloy fractures in a ductile manner. At intermediate temperatures, due to limited ductility of the γ phase, a rapid propagation of cracks by the shearing of the γ phase occurs, which is consistence with the "intermediate temperature brittleness" in Ni-based SX superalloys. At higher temperatures, necking occurs. Both the γ and γ phases become ductile, resulting in enhanced plastic deformation before fracture.