Deformation Behavior and Tensile Properties of the Semi-Equiaxed Microstructure in Near Alpha Titanium Alloy

The tensile deformation and fracture behavior of a particular semi-equiaxed microstructure (S-EM) in a near alpha titanium alloy TA19 are investigated by an in situ method. In the S-EM, the thin β lamellae grow through the equiaxed αp phase (αp), and the original αp/βtrans interface in the bimodal microstructure largely disappears, forming a blurry interface between the semi-equiaxed αp phase (equiaxed αp phase that is grew through by the thin β lamellae) and the transformed β microstructure (βtrans). The formation of dense slip bands inside the semi-equiaxed αp phase in the S-EM is inhibited by the thin β lamellae during the tensile deformation. The special characteristics of the S-EM reduce the stress concentration at the interface, and the crack initiation probability in the blurry semi-αp/βtrans interface decreased compared to the distinct αp/βtrans interface in a conventional equiaxed microstructure (EM). Moreover, the ultimate tensile strength of the S-EM is higher than that of the EM with a slight loss of plasticity.


Introduction
TA19 titanium alloy is a near alpha titanium alloy that is widely utilized in key structural components of aero-engines (e.g., compressor disks, blades, and casings) [1] due to its high strength to weight ratio, high temperature resistance, and corrosion resistance [2][3][4]. The mechanical properties of titanium alloys are largely influenced by the microstructural characteristics, such as the morphology and volume fraction of the α and β phases, and the grain size [5,6]. In particular, the interface characteristics between the α phase and β phase are fundamentally of significance during crack initiation and propagation [7][8][9]. For example, most of the cracks (approximately 50%) of TC21 titanium alloy (Ti-6Al-2Sn-2Zr-3Mo-1Cr-2Nb-0.1Si) generally initiate at the shear bands of the primary α lath, cracks also initiate at the α p /β trans interface (approximately 23%) and α s /β interface (approximately 27%) during the tensile loading [10]. In addition, when the size of α s is approximately 0.1 µm, cracks will initiate at the α p /β interface and α p /β trans interface. When the size of α s is approximately 0.5 µm, cracks also initiate at the α s /β r interface [11] (α p : equiaxed primary α phase; α s : lamellar secondary α phase; β trans : transformed β microstructure; β r : remnant β phase). The phase interface in titanium alloy can effectively prevent dislocation movement and improve strength [7]. However, inconsistencies between the stress and strain easily occur at the interface during loading, resulting in stress concentration at the phase interface due to the differences in the physical, mechanical properties, and grain orientation between the α phase and β phase. Therefore, the phase interface is the preferred site for crack nucleation and propagation under tensile [10] or fatigue loading [12], which is detrimental to the mechanical properties of the alloy [13,14]. The stress concentration caused by dislocation pile-up can be reduced by refining the α phase in titanium alloy [15], but the stress concentration at the interface has not been effectively alleviated. Different from the above-mentioned microstructural interface, the ghost α phases with blurry α p /β trans interface characteristics in the welding heat-affected zone (HAZ) of Ti6246 alloy have been reported [16,17]. The ghost α phases are the original equiaxed α phase of the base metal, which exceeds the β phase transition temperature during welding, but the temperature and time are not sufficient to reach chemical equilibrium [17]. Xu et al. [18] also revealed that Ti6242 alloy flash welded joint obtains high ductility that is equivalent to the matrix and extra strain hardening due to the presence of the semi-equiaxed microstructure (similar to the ghost α phases) in the HAZ, but the specific effect of a pure semi-equiaxed microstructure on the mechanical properties is still unclear. Furthermore, Liang et al. [19] prepared the semi-equiaxed microstructure in TA19 titanium alloy by simulating the temperature field of the welding heat-affected zone. The tensile strength of the sample of the semi-equiaxed microstructure was 1219 MPa, which was 15% higher than that of the asreceived material. However, limited efforts have been made to study the deformation and fracture characteristics of the semi-equiaxed microstructure. These studies have confirmed that the semi-equiaxed microstructure has particular microstructure and mechanical properties. Original α p /β trans interfaces disappear in the semi-equiaxed microstructure, and whether this interface's characteristics show special mechanical properties and deformation behavior requires further exploration.
In this study, the TA19 titanium alloy sample was rapidly heated (>100 • C/s) to β phase field (T > T β ) and held for 35 s, thereby inhibiting the diffusion of alloying elements, and finally, a pure semi-equiaxed microstructure was prepared. An in situ tensile test was used to explore the deformation, crack initiation, and propagation characteristics of the equiaxed microstructure and semi-equiaxed microstructure. These results were combined with the results of room temperature tensile tests to further study the differences in the tensile properties caused by the two types of microstructures. The study will provide guidance for tailoring the microstructure and enriching the mechanism of the strengthening and toughening titanium alloys.

Experimental Procedures
In this work, a TA19 bar was supplied after rolling at the α + β field to obtain a plate about 65 × 35 × 170 mm 3 ; then, it was tempered in the α + β field. The chemical composition of the TA19 bar is shown in Table 1. The microstructure consists of an equiaxed primary α phase (α p ) and transformed β microstructure (β trans ), and the transformed β microstructure contains a lamellar secondary α phase (α s ) and remnant β phase (β r ), as presented in Figure 1. The α/β phase transition temperature is approximately 1006 • C based on thermal analysis using a DIL805A/D phase analyzer (Netzsch, Selb, Germany). Two heat treatment modes were chosen to obtain the equiaxed microstructure (EM) and semi-equiaxed microstructure (S-EM), ensuring that the volume fraction of the equiaxed α p phase, the length and thickness of the α s lamellae are almost same in the two types of microstructures: (1) 970 • C/1 h, air cooling (AC) (for EM) and (2) 1015 • C/35 s, with a cooling rate of 20 • C/s (for S-EM). For optical microscopy (OM, Leica DMI5000M) (Leica, Wetzlar, Germany) and scanning electron microscopy (SEM, SUPRA40) (Carl Zeiss AG, Jena, Germany) observations, samples were ground and mechanically polished by a standard metallographic procedure (for more details, please see Ref. [20]), and they were etched with Kroll's reagent solution (V(HF):V(HNO 3 ):V(H 2 O) = 1:3:7). Moreover, microstructural parameters were counted by an Image-Pro Plus image analysis system. X-ray diffraction (XRD) analysis was carried out on a D8 Advance X-ray diffractometer with a Cu Kα source (40 kV, 200 mA) in the θ-2θ model. Electron back-scattered diffraction (EBSD) images were acquired to determine the misorientation distribution of the grains. In situ observation was performed with a SUPRA 40 scanning electron microscope at room temperature during the tensile deformation. The surfaces of the specimens were mechanically polished and strained with a crosshead speed of 0.1 µm/s. Room temperature tensile tests were carried out as standard E8/E8M Test Methods [21] on a tensile instrument (MTS 810) at a constant cross-head speed of 0.5 mm/min. Three samples with gauge dimensions of 44 × 9 × 2 mm 3 ( Figure 2) were tested to confirm the validity of the test results, and the fracture surface morphology was observed by SEM. An FEI Talos F200X field emission transmission electron microscope (TEM) (FEI, Hillsboro, OR, USA) was used to observe the microstructure after tensile fracture, thereby revealing the deformation mechanism.  Figure 2) were tested to confirm the validity of the test results, and the fracture surface morphology was observed by SEM. An FEI Talos F200X field emission transmission electron microscope (TEM) (FEI, Hillsboro, OR, USA) was used to observe the microstructure after tensile fracture, thereby revealing the deformation mechanism.    Figure 3 shows the optical micrographs and SEM micrographs of the EM and S-EM. The morphology of the EM is almost consistent with the initial microstructure, but the volume fraction of the equiaxed αp phase and αs lamellae decrease (Figure 3a), which is due to allotropic transformation (α → β) after heating to the α/β phase region. The formation of the S-EM is due to the inhomogeneous diffusion of elemental Mo in the β matrix into the equiaxed αp phase when the specimens are kept above the Tβ temperature for a short time, and in the cooling process, the thin β lamellae grow through the equiaxed αp phase [19]. A large amount of the equiaxed αp phases remain in the S-EM because after a transient β phase treatment, only the micro-zone adjacent to the original equiaxed αp phase boundary is transformed into the β phase. In this work, a relatively slow cooling rate was used to obtain the S-EM with obvious microstructural morphology, resulting in some α phases continuous precipitating along the grain boundaries (GB α) (Figure 3b). Figure 3c, d show SEM images of the EM and S-EM, respectively. Notably, there is a distinct αp/βtrans interface in the EM (Figure 3c). However, the distinct αp/βtrans interface mainly were acquired to determine the misorientation distribution of the grains. In situ observation was performed with a SUPRA 40 scanning electron microscope at room temperature during the tensile deformation. The surfaces of the specimens were mechanically polished and strained with a crosshead speed of 0.1 μm/s. Room temperature tensile tests were carried out as standard E8/E8M Test Methods [21] on a tensile instrument (MTS 810) at a constant cross-head speed of 0.5 mm/min. Three samples with gauge dimensions of 44 × 9 × 2 mm 3 ( Figure 2) were tested to confirm the validity of the test results, and the fracture surface morphology was observed by SEM. An FEI Talos F200X field emission transmission electron microscope (TEM) (FEI, Hillsboro, OR, USA) was used to observe the microstructure after tensile fracture, thereby revealing the deformation mechanism.    Figure 3 shows the optical micrographs and SEM micrographs of the EM and S-EM. The morphology of the EM is almost consistent with the initial microstructure, but the volume fraction of the equiaxed αp phase and αs lamellae decrease (Figure 3a), which is due to allotropic transformation (α → β) after heating to the α/β phase region. The formation of the S-EM is due to the inhomogeneous diffusion of elemental Mo in the β matrix into the equiaxed αp phase when the specimens are kept above the Tβ temperature for a short time, and in the cooling process, the thin β lamellae grow through the equiaxed αp phase [19]. A large amount of the equiaxed αp phases remain in the S-EM because after a transient β phase treatment, only the micro-zone adjacent to the original equiaxed αp phase boundary is transformed into the β phase. In this work, a relatively slow cooling rate was used to obtain the S-EM with obvious microstructural morphology, resulting in some α phases continuous precipitating along the grain boundaries (GB α) (Figure 3b). Figure 3c, d show SEM images of the EM and S-EM, respectively. Notably, there is a distinct αp/βtrans interface in the EM (Figure 3c). However, the distinct αp/βtrans interface mainly  Figure 3 shows the optical micrographs and SEM micrographs of the EM and S-EM. The morphology of the EM is almost consistent with the initial microstructure, but the volume fraction of the equiaxed α p phase and α s lamellae decrease (Figure 3a), which is due to allotropic transformation (α → β) after heating to the α/β phase region. The formation of the S-EM is due to the inhomogeneous diffusion of elemental Mo in the β matrix into the equiaxed α p phase when the specimens are kept above the T β temperature for a short time, and in the cooling process, the thin β lamellae grow through the equiaxed α p phase [19]. A large amount of the equiaxed α p phases remain in the S-EM because after a transient β phase treatment, only the micro-zone adjacent to the original equiaxed α p phase boundary is transformed into the β phase. In this work, a relatively slow cooling rate was used to obtain the S-EM with obvious microstructural morphology, resulting in some α phases continuous precipitating along the grain boundaries (GB α) (Figure 3b). Figure 3c,d show SEM images of the EM and S-EM, respectively. Notably, there is a distinct α p /β trans interface in the EM (Figure 3c). However, the distinct α p /β trans interface mainly disappears in the S-EM, the thin β lamellae grow through the equiaxed α p phase and exhibits a blurry interface ( Figure 3d). The thickness of the thin β lamellae gradually decreases and finally disappears within the equiaxed α p phase, as shown in the inset TEM image. The equiaxed α p phase is grew through by the thin β lamellae in the S-EM, which is defined as the semi-equiaxed α p phase (semi-α p ) in this paper. The microstructural feature results of the EM and S-EM are presented in Table 2.

Microstructural Characteristics
Materials 2021, 14, x FOR PEER REVIEW 4 of 14 disappears in the S-EM, the thin β lamellae grow through the equiaxed αp phase and exhibits a blurry interface ( Figure 3d). The thickness of the thin β lamellae gradually decreases and finally disappears within the equiaxed αp phase, as shown in the inset TEM image. The equiaxed αp phase is grew through by the thin β lamellae in the S-EM, which is defined as the semi-equiaxed αp phase (semi-αp) in this paper. The microstructural feature results of the EM and S-EM are presented in Table 2.   Figure 4 shows the XRD patterns for the initial microstructure, the EM and S-EM of TA19 titanium alloy. The phase constitutions of the three samples are the α phase and β phase, and the intensity distribution of the diffraction peak of the α phase and β phase is similar. In the XRD patterns, α(101) is the strongest among all of the diffraction peaks of both α and β phases.    Figure 4 shows the XRD patterns for the initial microstructure, the EM and S-EM of TA19 titanium alloy. The phase constitutions of the three samples are the α phase and β phase, and the intensity distribution of the diffraction peak of the α phase and β phase is similar. In the XRD patterns, α (101) is the strongest among all of the diffraction peaks of both α and β phases.
To explore the effect of the phase transition on the misorientation distribution of grains in the EM and S-EM, EBSD analysis of the microscopic regions, including the equiaxed α p phase and the semi-equiaxed α p phase of the two samples, were performed. The step size is 0.2 µm, as shown in Figure 5. For the EM, the equiaxed α p maintains the same orientation relationship, and the fine α/β lamellae in the β trans microstructure exhibit different orientation relationships (Figure 5a). For the S-EM, the semi-equiaxed α p phase is formed after the phase transition of the original equiaxed α p phase, which exhibits different orientation relationships. This phenomenon is different from the same orientation relationships of the equiaxed α p phase in the EM. The β trans microstructure in the S-EM is similar to the EM and shows multiple orientation (Figure 5b).

S-EM
41 ± 2 13 ± 1 11 ± 1 410 ± 20 1.2 850 ± 30 Figure 4 shows the XRD patterns for the initial microstructure, the EM and S-EM of TA19 titanium alloy. The phase constitutions of the three samples are the α phase and β phase, and the intensity distribution of the diffraction peak of the α phase and β phase is similar. In the XRD patterns, α(101) is the strongest among all of the diffraction peaks of both α and β phases.  To explore the effect of the phase transition on the misorientation distribution of grains in the EM and S-EM, EBSD analysis of the microscopic regions, including the equiaxed αp phase and the semi-equiaxed αp phase of the two samples, were performed. The step size is 0.2 μm, as shown in Figure 5. For the EM, the equiaxed αp maintains the same orientation relationship, and the fine α/β lamellae in the βtrans microstructure exhibit different orientation relationships (Figure 5a). For the S-EM, the semi-equiaxed αp phase is formed after the phase transition of the original equiaxed αp phase, which exhibits different orientation relationships. This phenomenon is different from the same orientation relationships of the equiaxed αp phase in the EM. The βtrans microstructure in the S-EM is similar to the EM and shows multiple orientation (Figure 5b).

Research on the Deformation Behavior
The above description of the microstructural characteristics illustrates that the S-EM has special interface microstructural characteristics compared with the EM. An in situ tensile test is designed, as it is an effective method for comparing the deformation behavior of the two types of microstructures. During the in situ tensile process, at some strain points, the test was interrupted, and the SEM images were captured. The interval between adjacent strain points is approximately two minutes. The typical Engineering stress-strain curves of the two samples in the in situ tensile test performed at room temperature are shown in Figure 6. Moreover, the representative strain points of 6%, 8.3%, 12%, and 14% (corresponding to the letters A1 (A2), B1 (B2), C1 (C2), and D1 (D2), respectively) are selected to analyze the deformation characteristics of EM and S-EM. Figure 7 shows the actual deformation behavior of the EM in the in situ tensile test, and the tensile direction of the SEM images is parallel to the horizontal direction. When the strain is 6% (corresponds to the letter A1 on the curve in Figure 6), the equiaxed αp phase deformation occurs first, and a small number of slip bands within the equiaxed αp phase (numbers 1 and 2) are observed (Figure 7a). With a further increase in strain to 8.3% (letter B1), more slip bands are observed in the equiaxed αp phase of numbers 1 and 2 (Figure 7b), indicating that the equiaxed αp phases have good deformation capacity. However, there are no slip bands in the number 3 equiaxed αp phase, and the orientation of this equiaxed αp phase might not be favorable enough to initiate deformation at an early deformation stage. In addition, a large number of slip bands are hindered by the interface between the number 4 equiaxed αp phase and βtrans microstructure, it causes local stress concentration, and a crack initiates at the αp/βtrans interface. The crack also initiates at the

Research on the Deformation Behavior
The above description of the microstructural characteristics illustrates that the S-EM has special interface microstructural characteristics compared with the EM. An in situ tensile test is designed, as it is an effective method for comparing the deformation behavior of the two types of microstructures. During the in situ tensile process, at some strain points, the test was interrupted, and the SEM images were captured. The interval between adjacent strain points is approximately two minutes. The typical Engineering stress-strain curves of the two samples in the in situ tensile test performed at room temperature are shown in Figure 6. Moreover, the representative strain points of 6%, 8.3%, 12%, and 14% (corresponding to the letters A 1 (A 2 ), B 1 (B 2 ), C 1 (C 2 ), and D 1 (D 2 ), respectively) are selected to analyze the deformation characteristics of EM and S-EM. Figure 7 shows the actual deformation behavior of the EM in the in situ tensile test, and the tensile direction of the SEM images is parallel to the horizontal direction. When the strain is 6% (corresponds to the letter A 1 on the curve in Figure 6), the equiaxed α p phase deformation occurs first, and a small number of slip bands within the equiaxed α p phase (numbers 1 and 2) are observed (Figure 7a). With a further increase in strain to 8.3% (letter B 1 ), more slip bands are observed in the equiaxed α p phase of numbers 1 and 2 (Figure 7b), indicating that the equiaxed α p phases have good deformation capacity. However, there are no slip bands in the number 3 equiaxed α p phase, and the orientation of this equiaxed α p phase might not be favorable enough to initiate deformation at an early deformation stage. In addition, a large number of slip bands are hindered by the interface between the number 4 equiaxed α p phase and β trans microstructure, it causes local stress concentration, and a crack initiates at the α p /β trans interface. The crack also initiates at the coarsening α s lamella perpendicular to the tensile direction duo to the lamella being subjected to greater tensile stress. At a strain of 12% (letter C 1 ), the crack propagates rapidly from one end along the direction of the α s lamellae and from the other end along the slip band within the equiaxed α p phase, forming a low-energy channel for crack propagation. The cross slip of the number 4 equiaxed α p phase occurs, and the direction of crack propagation changes. Moreover, a crack also initiates in the number 5 equiaxed α p phase (Figure 7c). When the strain is further increased to 14% (letter D 1 ), the microstructure is severely deformed, and the length of the number 4 equiaxed α p phase is increased by approximately 47% in the tensile direction. Moreover, the entire crack width is extended to 8 µm, which will lead to fracture failure of the alloy (Figure 7d). Notably, cracks also initiate at the other α p /β trans interface when the strain is 11.6%and 14.7%, as shown in Figure 8. coarsening αs lamella perpendicular to the tensile direction duo to the lamella being subjected to greater tensile stress. At a strain of 12% (letter C1), the crack propagates rapidly from one end along the direction of the αs lamellae and from the other end along the slip band within the equiaxed αp phase, forming a low-energy channel for crack propagation.
The cross slip of the number 4 equiaxed αp phase occurs, and the direction of crack propagation changes. Moreover, a crack also initiates in the number 5 equiaxed αp phase (Figure 7c). When the strain is further increased to 14% (letter D1), the microstructure is severely deformed, and the length of the number 4 equiaxed αp phase is increased by approximately 47% in the tensile direction. Moreover, the entire crack width is extended to 8 μm, which will lead to fracture failure of the alloy (Figure 7d). Notably, cracks also initiate at the other αp/βtrans interface when the strain is 11.6%and 14.7%, as shown in Figure  8.   For the S-EM sample, no slip bands were found in the semi-equiaxed α p phase (numbers 1, 2, and 3) when the strain is 6% (letter A 2 ), as shown in Figure 9a. Before the strain reaches 14% (letter D 2 ), the microstructural characteristics of the S-EM have not changed significantly, so SEM images with strain of 12% (letter C 2 ) are not considered. Compared to the EM, there is no obvious slip bands and cracks existing in the semiequiaxed α p phase of the S-EM when the strain is 14% (Figure 9c). The higher magnification image of Figure 9c shows that only a few and shorter slip bands occur between the thin β lamellae in the semi-equiaxed α p phase (Figure 9d). These results reveal that the thin β lamellae inhibit the formation of dense slip bands in the semi-equiaxed α p ; thereby, dislocations do not easily pile up at the blurry interface between the semi-equiaxed α p phase and β trans microstructure in a large amount. The bending and tearing deformation of the thin β lamellae within the semi-equiaxed α p phase are observed (Figure 9e). In addition, the transfer of short slip bands across the thin β lamellae was observed in the semi-equiaxed α p phase, resulting in zig-zag ledges, as shown in Figure 9f. This indicates that the thin β lamellae have a slight effect on hindering the movement of dislocations and slip transfer.    For the S-EM sample, no slip bands were found in the semi-equiaxed αp phase (numbers 1, 2, and 3) when the strain is 6% (letter A2), as shown in Figure 9a. Before the strain reaches 14% (letter D2), the microstructural characteristics of the S-EM have not changed significantly, so SEM images with strain of 12% (letter C2) are not considered. Compared to the EM, there is no obvious slip bands and cracks existing in the semi-equiaxed αp phase of the S-EM when the strain is 14% (Figure 9c). The higher magnification image of Figure  9c shows that only a few and shorter slip bands occur between the thin β lamellae in the semi-equiaxed αp phase (Figure 9d). These results reveal that the thin β lamellae inhibit the formation of dense slip bands in the semi-equiaxed αp; thereby, dislocations do not easily pile up at the blurry interface between the semi-equiaxed αp phase and βtrans microstructure in a large amount. The bending and tearing deformation of the thin β lamellae within the semi-equiaxed αp phase are observed (Figure 9e). In addition, the transfer of short slip bands across the thin β lamellae was observed in the semi-equiaxed αp phase, resulting in zig-zag ledges, as shown in Figure 9f. This indicates that the thin β lamellae have a slight effect on hindering the movement of dislocations and slip transfer. Titanium alloys exhibit many deformation behaviors, and the strain modes include planar slip, dislocation tangling, and twinning [22]. Generally, the activation of a slip system requires that a high resolved shear stress on the slip system exceeds its critical resolved shear stress (CRSS) [23]. When the resolved shear stress is higher than the critical resolved shear stress, planar slip occurs, which is the main deformation mode. Additionally, the deformation of hardly oriented grains is accompanied by dislocation tangling. During the tensile deformation, the number of slip bands and the distribution of dislocations in each grain depend on the local strain. Figure 10 shows TEM micrographs of the deformation microstructure of the EM sample after tensile fracture. The observation area is approximately 1 mm away from the fracture. As shown in Figure 10a, dislocation tangling occurs at the α p /β trans interface, which is the result of dislocation slip being hindered at the α p /β trans interface. A large number of slip bands are arranged along a specific crystal plane, which is observed in the local higher magnification images of Figure 10a. The distance between the slip bands is approximately 70 nm. From the diffraction pattern of the selected area, these slip bands belong to the {011(-)0} prismatic slip bands (Figure 10b). Slip systems with high Schmid factor values and low CRSS (basal <a> and prismatic <a> slip) can be preferentially activated under tensile stress at room temperature, and prismatic <a> slip is more easily activated than basal <a> slip [10]. The above phenomena indicate that equiaxed α p phase plays an important role in accommodating the plastic deformation of the alloy, and this result is consistent with Ref [24]. The equiaxed α p phase does not maintain the Burgess orientation relationship (BOR) with the adjacent β phase, which greatly restricts slip transmission across the α p /β trans interface [25]. So, the slip bands do not pass through the equiaxed α p phase to the β trans microstructure (Figure 10b), resulting in the stress concentration at the α p /β trans interface. Meanwhile, it can be seen from Figure 10c that the remnant β r phase is fractured in shear due to severe local deformation, and the short slips are observed in the coarsening α s lamellae, which are also recorded as shear slip bands by Ref. [26]. Moreover, there is high density dislocation tangling in the coarsening α s lamellae, which occurs because the slip motion in α s lamellae is hindered at the α s /β r interface (Figure 10d).
For the S-EM sample, no slip bands were found in the semi-equiaxed αp phase (numbers 1, 2, and 3) when the strain is 6% (letter A2), as shown in Figure 9a. Before the strain reaches 14% (letter D2), the microstructural characteristics of the S-EM have not changed significantly, so SEM images with strain of 12% (letter C2) are not considered. Compared to the EM, there is no obvious slip bands and cracks existing in the semi-equiaxed αp phase of the S-EM when the strain is 14% (Figure 9c). The higher magnification image of Figure  9c shows that only a few and shorter slip bands occur between the thin β lamellae in the semi-equiaxed αp phase (Figure 9d). These results reveal that the thin β lamellae inhibit the formation of dense slip bands in the semi-equiaxed αp; thereby, dislocations do not easily pile up at the blurry interface between the semi-equiaxed αp phase and βtrans microstructure in a large amount. The bending and tearing deformation of the thin β lamellae within the semi-equiaxed αp phase are observed (Figure 9e). In addition, the transfer of short slip bands across the thin β lamellae was observed in the semi-equiaxed αp phase, resulting in zig-zag ledges, as shown in Figure 9f. This indicates that the thin β lamellae have a slight effect on hindering the movement of dislocations and slip transfer.   Figure 11 shows the TEM micrographs of the deformation microstructure of the S-EM sample after tensile fracture, which is consistent with the observation area of the EM. The thickness of the thin β lamellae within the semi-equiaxed α p phase in the S-EM is approximately 20 nm, which is slightly smaller than that of the remnant β r phase. Different from a large number of slip bands in the equiaxed α p phase of the EM, short slip bands and dislocation tangling are produced between the thin β lamellae in the semi-equiaxed α p phase of the S-EM (Figure 11a); this is the result of dislocation slip being hindered by the thin β lamellae. The phenomenon of dislocation tangling between the thin β lamellae in the semi-equiaxed α p phase is more obvious in the local higher magnification images (Figure 11b). The diffraction pattern shows that the slip bands between the thin β lamellae in the semi-equiaxed α p belong to the {011(-)0} prismatic slip bands (Figure 11c), which is consistent with the deformation characteristics in the equiaxed α p phase of the EM.
terface [25]. So, the slip bands do not pass through the equiaxed αp phase to the βtrans microstructure (Figure 10b), resulting in the stress concentration at the αp/βtrans interface. Meanwhile, it can be seen from Figure 10c that the remnant βr phase is fractured in shear due to severe local deformation, and the short slips are observed in the coarsening αs lamellae, which are also recorded as shear slip bands by Ref. [26]. Moreover, there is high density dislocation tangling in the coarsening αs lamellae, which occurs because the slip motion in αs lamellae is hindered at the αs/βr interface (Figure 10d).   Figure 11 shows the TEM micrographs of the deformation microstructure of the S-EM sample after tensile fracture, which is consistent with the observation area of the EM. The thickness of the thin β lamellae within the semi-equiaxed αp phase in the S-EM is approximately 20 nm, which is slightly smaller than that of the remnant βr phase. Different from a large number of slip bands in the equiaxed αp phase of the EM, short slip bands and dislocation tangling are produced between the thin β lamellae in the semi-equiaxed αp phase of the S-EM ( Figure 11a); this is the result of dislocation slip being hindered by the thin β lamellae. The phenomenon of dislocation tangling between the thin β lamellae in the semi-equiaxed αp phase is more obvious in the local higher magnification images (Figure 11b). The diffraction pattern shows that the slip bands between the thin β lamellae in the semi-equiaxed αp belong to the {011(-)0} prismatic slip bands (Figure 11c), which is consistent with the deformation characteristics in the equiaxed αp phase of the EM.

Tensile Properties and Fracture Characteristics
For titanium alloys, preventing dislocation slip is the most significant strengthening mechanism, in which grain boundaries and phase interfaces act as barriers to reduce the effective slip length, i.e., the Hall-Petch effect [27,28]. The typical engineering tensile stress-strain curves of the TA19 titanium alloy with the EM and S-EM at room tempera- Figure 11. Deformation behavior of the tensile specimen with the S-EM: (a,b) dislocation tangling and planar slip bands between the thin β lamellae of the semi-equiaxed α p phase and (c) prismatic slip bands between the thin β lamellae of the semi-equiaxed α p phase.

Tensile Properties and Fracture Characteristics
For titanium alloys, preventing dislocation slip is the most significant strengthening mechanism, in which grain boundaries and phase interfaces act as barriers to reduce the effective slip length, i.e., the Hall-Petch effect [27,28]. The typical engineering tensile stress-strain curves of the TA19 titanium alloy with the EM and S-EM at room temperature are shown in Figure 12. Furthermore, the room temperature tensile properties of EM and S-EM are summarized in Table 3. It is obvious from Figure 12 and Table 3 that the yield strength (YS) and the ultimate tensile strength (UTS) of S-EM are higher 38 MPa and 46 MPa than those of EM, respectively. On the contrary, the elongation (El) and reduction in area (RA) of S-EM are slightly lower than that of EM. The characteristics of the β trans microstructure in the EM and S-EM are basically the same. Therefore, the difference in the tensile properties of EM and S-EM mainly depends on the distinct microstructural characteristics of the equiaxed α p phase and semi-equiaxed α p phase. The thin β lamellae grow through the equiaxed α p phase in the S-EM, which hinders the slip of dislocations and reduces the effective slip length in semi-equiaxed α p phase, and the strength of the S-EM is improved.   The tensile fracture surface morphology of the EM and S-EM was analyzed to further comprehend the influence of the microstructure on the properties of the TA19 titanium alloy, as displayed in Figure 13. From the low magnification photography of the fracture surface, the shearing tip and dimple zone occur in the EM sample, and the necking characteristics are relatively obvious (Figure 13a), which is a typical feature of high plasticity. In addition, the shearing tip and dimple zone can also be observed in the macroscopic fracture surface of the S-EM, but the area of the shearing tip is slightly smaller than that of the EM, and the necking characteristic is less obvious (Figure 13b). The high magnification photography of the fracture surface shows that both the EM and S-EM tensile specimens fail in a mixed fracture mode. In the fracture surface with EM, the failure induced by dimple nucleation and coalescence along the αp/βtrans interfaces is observed, and there are tearing ridges between the dimples. The cleavage facets are relatively smooth, which is attributed to the expansion of the cracks along the equiaxed αp inner slip bands or the αp/βtrans interface (Figure 13c). The dimples in the S-EM samples are small and shallow. The surfaces of the dimples are rough, and there are many small microvoids, which may be the characteristics left by the fracture of the thin β lamellae in the semi-equiaxed αp phase. Furthermore, the proportion of cleavage facets (highlighted by the red arrow) increases, and traces of fracture along the grain boundary (highlighted by the yellow arrow) are found (Figure 13d).  The tensile fracture surface morphology of the EM and S-EM was analyzed to further comprehend the influence of the microstructure on the properties of the TA19 titanium alloy, as displayed in Figure 13. From the low magnification photography of the fracture surface, the shearing tip and dimple zone occur in the EM sample, and the necking characteristics are relatively obvious (Figure 13a), which is a typical feature of high plasticity. In addition, the shearing tip and dimple zone can also be observed in the macroscopic fracture surface of the S-EM, but the area of the shearing tip is slightly smaller than that of the EM, and the necking characteristic is less obvious (Figure 13b). The high magnification photography of the fracture surface shows that both the EM and S-EM tensile specimens fail in a mixed fracture mode. In the fracture surface with EM, the failure induced by dimple nucleation and coalescence along the α p /β trans interfaces is observed, and there are tearing ridges between the dimples. The cleavage facets are relatively smooth, which is attributed to the expansion of the cracks along the equiaxed α p inner slip bands or the α p /β trans interface (Figure 13c). The dimples in the S-EM samples are small and shallow. The surfaces of the dimples are rough, and there are many small microvoids, which may be the characteristics left by the fracture of the thin β lamellae in the semi-equiaxed α p phase. Furthermore, the proportion of cleavage facets (highlighted by the red arrow) increases, and traces of fracture along the grain boundary (highlighted by the yellow arrow) are found (Figure 13d). Combined with the in situ tensile test, it is found that the approximately 3 μm crack first initiates at GB α of the S-EM when the strain is 11.8% ( Figure 14). In contrast, the crack first initiates at the αp/βtrans interface in the EM when the strain is 8.3%. Therefore, some GB α precipitates on the β grain boundary in the S-EM, which is an important factor that causes early fracture and ultimately leads to the decrease in plasticity [29]. Furthermore, based on the observation and measurement of at least 100 cracks, the crack initiation probability at different sites in the EM and S-EM was obtained during in situ tensile tests, as shown in Figure 15. The probability of crack initiation in the semiequiaxed αp phase (include the blurry semi-αp/βtrans interface) of the S-EM is lower than the equiaxed αp phase (include the αp/βtrans interface) of the EM, whereas the crack initiation probability increased in the βtrans microstructure and GB α of the S-EM in comparison to the EM. This indicates that the special interfacial microstructure of S-EM leads to the reduction of local stress concentration at the blurry interface between semi-equiaxed αp phase and βtrans microstructure compared to the αp/βtrans interface in the EM. Combined with the in situ tensile test, it is found that the approximately 3 µm crack first initiates at GB α of the S-EM when the strain is 11.8% ( Figure 14). In contrast, the crack first initiates at the α p /β trans interface in the EM when the strain is 8.3%. Therefore, some GB α precipitates on the β grain boundary in the S-EM, which is an important factor that causes early fracture and ultimately leads to the decrease in plasticity [29]. Combined with the in situ tensile test, it is found that the approximately 3 μm crack first initiates at GB α of the S-EM when the strain is 11.8% (Figure 14). In contrast, the crack first initiates at the αp/βtrans interface in the EM when the strain is 8.3%. Therefore, some GB α precipitates on the β grain boundary in the S-EM, which is an important factor that causes early fracture and ultimately leads to the decrease in plasticity [29]. Furthermore, based on the observation and measurement of at least 100 cracks, the crack initiation probability at different sites in the EM and S-EM was obtained during in situ tensile tests, as shown in Figure 15. The probability of crack initiation in the semiequiaxed αp phase (include the blurry semi-αp/βtrans interface) of the S-EM is lower than the equiaxed αp phase (include the αp/βtrans interface) of the EM, whereas the crack initiation probability increased in the βtrans microstructure and GB α of the S-EM in comparison to the EM. This indicates that the special interfacial microstructure of S-EM leads to the reduction of local stress concentration at the blurry interface between semi-equiaxed αp phase and βtrans microstructure compared to the αp/βtrans interface in the EM. Furthermore, based on the observation and measurement of at least 100 cracks, the crack initiation probability at different sites in the EM and S-EM was obtained during in situ tensile tests, as shown in Figure 15. The probability of crack initiation in the semi-equiaxed α p phase (include the blurry semi-α p /β trans interface) of the S-EM is lower than the equiaxed α p phase (include the α p /β trans interface) of the EM, whereas the crack initiation probability increased in the β trans microstructure and GB α of the S-EM in comparison to the EM. This indicates that the special interfacial microstructure of S-EM leads to the reduction of local stress concentration at the blurry interface between semi-equiaxed α p phase and β trans microstructure compared to the α p /β trans interface in the EM. Based on the above observation and discussion of the microstructural characteristic in the EM and S-EM during the tensile deformation, the relationship between slip and the formation of microvoids or cracks is schematically illustrated in Figure 16. When the EM is subjected to a large tensile load, a large number of dislocations pile up at the αp/βtrans interface, resulting in a severe stress concentration at the αp/βtrans interface. Therefore, the αp/βtrans interface is the preferred initiation site for microvoids or cracks [10,30]. Due to the different orientations and short effective slip distances in the αs lamellae and remnant βr phase, the deformation capacity is relatively poor, resulting in cracks initiating at the αs/βr interface [31], as shown in Figure 16a. For the S-EM, dense slip bands in the semi-equiaxed αp phase are suppressed by the thin β lamellae, and fewer dislocations pile up at the blurry interface between the semi-equiaxed αp phase and βtrans microstructure, which reduces the stress concentration. Then, the probability of crack initiation at the blurry semi-αp/βtrans interface decreases (Figure 16b).

Conclusions
In this work, the deformation behavior and fracture mechanism of the particular S-EM were revealed by in situ tensile tests, and the influence of the microstructural characteristics on the mechanical properties was explored. The main conclusions are summarized as follows: (1) In the S-EM, the distinct αp/βtrans interface in the bimodal microstructure primarily disappears, and the thin β lamellae grow through the equiaxed αp phase, leading to partial division of the equiaxed αp phase by the thin β lamellae. Based on the above observation and discussion of the microstructural characteristic in the EM and S-EM during the tensile deformation, the relationship between slip and the formation of microvoids or cracks is schematically illustrated in Figure 16. When the EM is subjected to a large tensile load, a large number of dislocations pile up at the α p /β trans interface, resulting in a severe stress concentration at the α p /β trans interface. Therefore, the α p /β trans interface is the preferred initiation site for microvoids or cracks [10,30]. Due to the different orientations and short effective slip distances in the α s lamellae and remnant β r phase, the deformation capacity is relatively poor, resulting in cracks initiating at the α s /β r interface [31], as shown in Figure 16a. For the S-EM, dense slip bands in the semiequiaxed α p phase are suppressed by the thin β lamellae, and fewer dislocations pile up at the blurry interface between the semi-equiaxed α p phase and β trans microstructure, which reduces the stress concentration. Then, the probability of crack initiation at the blurry semi-α p /β trans interface decreases (Figure 16b). Based on the above observation and discussion of the microstructural characteristic in the EM and S-EM during the tensile deformation, the relationship between slip and the formation of microvoids or cracks is schematically illustrated in Figure 16. When the EM is subjected to a large tensile load, a large number of dislocations pile up at the αp/βtrans interface, resulting in a severe stress concentration at the αp/βtrans interface. Therefore, the αp/βtrans interface is the preferred initiation site for microvoids or cracks [10,30]. Due to the different orientations and short effective slip distances in the αs lamellae and remnant βr phase, the deformation capacity is relatively poor, resulting in cracks initiating at the αs/βr interface [31], as shown in Figure 16a. For the S-EM, dense slip bands in the semi-equiaxed αp phase are suppressed by the thin β lamellae, and fewer dislocations pile up at the blurry interface between the semi-equiaxed αp phase and βtrans microstructure, which reduces the stress concentration. Then, the probability of crack initiation at the blurry semi-αp/βtrans interface decreases (Figure 16b).

Conclusions
In this work, the deformation behavior and fracture mechanism of the particular S-EM were revealed by in situ tensile tests, and the influence of the microstructural characteristics on the mechanical properties was explored. The main conclusions are summarized as follows: (1) In the S-EM, the distinct αp/βtrans interface in the bimodal microstructure primarily disappears, and the thin β lamellae grow through the equiaxed αp phase, leading to partial division of the equiaxed αp phase by the thin β lamellae.

Conclusions
In this work, the deformation behavior and fracture mechanism of the particular S-EM were revealed by in situ tensile tests, and the influence of the microstructural characteristics on the mechanical properties was explored. The main conclusions are summarized as follows: (1) In the S-EM, the distinct α p /β trans interface in the bimodal microstructure primarily disappears, and the thin β lamellae grow through the equiaxed α p phase, leading to partial division of the equiaxed α p phase by the thin β lamellae.
(2) The S-EM effectively suppress the formation of dense slip bands in the semi-equiaxed α p phase, so the stress concentration at the blurry semi-α p /β trans interface in the S-EM has been reduced, and the semi-equiaxed α p phase and β trans microstructure have better deformation compatibility. In addition, the tensile strength of the S-EM is higher than that of the EM due to the thin β lamellae hindering the movement of dislocations within the semi-equiaxed α p phase in the S-EM. (3) In the S-EM, the prismatic slip and dislocation tangling between the thin β lamellae are the main deformation modes of the semi-equiaxed α p phase. The deformation of the β trans microstructure in the S-EM is mainly affected by planar slip and dislocation tangling in the α s lamellae. (4) The tensile fracture failures of both EM and S-EM in TA19 alloy show a mixture fracture mode. Compared to the EM, S-EM samples have shallower dimples and more cleavage facets, which leads to the proportion of brittle fracture mechanism of S-EM is slightly larger than EM.