Effect of Dislocation Slip Mechanism under the Control of Oxygen Concentration in Alpha-Case on Strength and Ductility of TC4 Alloy

: The oxygen diffusion layer (alpha-case) is generally considered to have a negative impact on the mechanical properties and applications of titanium alloys. In this study, TC4 alloy specimens with four types of different oxygen concentrations in alpha-case were obtained by controlling the oxygen diffusion process parameters. Scanning electron microscopy and glow discharge spectrometry were employed to characterize the microstructure and oxygen concentration of alpha-case. The effect of alpha-case on strength and ductility of TC4 alloy was investigated via tensile test and new insights were provided. The results indicate that with the increase in the oxygen concentration in the alpha-case, the ductility of the TC4 alloy gradually decreased. Interestingly, the strength of TC4 alloy with the alpha-case ﬁrst increased and then decreased, resulting in the existence of a peak corresponding to a lower oxygen concentration before the decline of strength. Furthermore, a relatively good ductility match was also observed at the peak. When the oxygen concentration was relatively high, both the strength and ductility decreased. This phenomenon is attributed to the fact that dislocations in the alpha-case controlled by the oxygen concentration were modiﬁed from wavy slip to planar slip. Finally, the dislocation’s slip morphology was characterized by transmission electron microscopy.


Introduction
TC4 alloy is one of the main components employed in the aerospace field, and it accounts for more than half of the global titanium alloy production [1,2]. As an important interstitial element in titanium alloys, oxygen has a significant chemical affinity for Ti [3]. Bond energy of the Ti-O bond (2.12 eV) [4] is equivalent to that of the Ti-Ti bond (2.56 eV) [5]. Higher chemical affinity allows titanium to form a natural oxide film (usually less than 10 nm in thickness) even at room temperature [6]. At relatively high temperatures (usually higher than 480 • C), the increased oxidation of titanium leads to an increase in the thickness of surface oxide film as well as the formation of an oxygen diffusion layer (alpha-case) under the oxide film [7]. Therefore, after hot working, hot forming, and high-temperature applications, the surface layer of titanium alloys is often accompanied by the formation of alpha-case [8,9]. The alpha-case has an important impact on the mechanical properties of titanium alloys. Moreover, it can even limit the application of titanium alloys. For example, for gas turbine engines employed in the aerospace industry, the service life of titanium alloys can be affected by alpha-case [10].
Owing to solid solution strengthening of oxygen, the hardness of alpha-case can usually be significantly improved [11]. Therefore, oxygen diffusion has been considered an effective surface modification method for titanium alloys. Thus, a series of oxygen diffusion processes has been developed. However, although the high hardness of alphacase improves the surface strength of titanium alloys, it also reduces their ductility and fatigue life due to its brittleness [12,13]. This is very significant for the practical application of titanium alloys, in particular, in the aerospace field [8]. Although maximum solid solubility of oxygen in the alpha and beta phases reaches 33.9 at.% and 3.8 at.% [14], respectively, an oxygen concentration of 0.51 wt.% can increase the growth rate of long cracks by 5-10 times (∆K < 12 MPa·m 1/2 for R = 0.01) [15]. Even a relatively small oxygen concentration (0.06-0.18 wt.%) can have a negative effect on crack growth [16]. Under cyclic loading, the existence of alpha-case not only reduces the number of cycles for crack initiation, but also causes the initiation of multiple cracks [17]. Consequently, this leads to a decrease in the fatigue life of titanium alloys. The brittleness of the alpha-case readily causes microcracks on the surface of the titanium alloy during the stretching process [18]. For instance, Chan et al. [19] conducted a significant number of statistical analyses and found that the depth of the microcracks generated was approximately 0.9 times the thickness of the alpha-case. These micro-cracks are crucial locations of failure initiation and they easily propagate into the matrix during tensile test, thus reducing the load-bearing cross section of the specimens and leading to the early fracture of the specimens [20,21]. Therefore, both the strength and ductility decrease for the specimens with alpha-case. During the investigation of near-beta titanium alloys, it was found that a solid solution of oxygen in the alpha-case caused alloying elements to migrate to the matrix and coarsen the surface structure [22]. This resulted in a decrease in the hardness of the alpha-case relative to the matrix, thereby reducing the strength of the material. Most of the methods used regarding the alpha-case research are based on the highest temperature (500-600 • C) of the material during their service life [20,21,23]. When heat treatments cause alphacase formation on the surface of titanium alloys, the matrix inevitably forms precipitates (such as Ti 3 Al (α 2 ), silicide). This makes it impossible to singly explore the effect of alpha-case on the plasticity of titanium alloys. Moreover, a lack of systematic and effective research regarding the topic has been observed. At the same time, explanation for the embrittlement mechanism of the alpha-case has not been provided yet. Oxygen is an effective solid solution strengthening element for titanium. Within a certain range, an average increase of 0.1 wt.% of oxygen can increase the tensile strength of titanium by about 120 MPa [24,25]. However, there is no clear explanation for whether the alpha-case strengthened by oxygen in solid solution exhibits a strengthening effect on titanium alloys. Therefore, investigation of the effect of alpha-case on the properties of titanium alloys is highly desirable. This can provide theoretical and practical guidance for the use of titanium alloys at high temperatures.
In this study, industrially popular TC4 alloy was used as the material, and alpha-case specimens with different concentrations were obtained by changing pre-oxidation time of the oxygen diffusion process parameters. The tensile test was used to study the effect of alpha-case on the strength and ductility of TC4 alloy. Scanning electron microscopy was employed to observe the tensile fracture and analyze the fracture mechanism of specimens in different states. Finally, focused ion beam technology and transmission electron microscopy were combined to observe deformation characteristics of the alphacase and key factors leading to its embrittlement were discussed.

Materials and Methods
The initial material used in this study is an as-rolled TC4 alloy rod specimen (diameter: φ 30 mm). Its chemical composition is presented in Table 1. The β transition temperature of the material is roughly 975 • C. A cross-sectional micro-morphology of the initial material, which is shown in Figure 1, was obtained by SEM. The image exhibits that the material is composed of an equiaxed α phase and β phase, while β phase particles are dispersed between equiaxed α phases. Mechanical properties of TC4 alloy along the rolling direction are summarized in Table 2. dispersed between equiaxed α phases. Mechanical properties of TC4 alloy along the rolling direction are summarized in Table 2.   The raw material was processed by cutting wire into round bars with a diameter of 6 mm and a length of 70 mm (ϕ6 × 70 mm). Then, cylindrical tensile specimens with a diameter of 4 mm and a gauge length of 20 mm were machined with the tensile axis being parallel to the roll direction. The surface of tensile specimens was ground using abrasive papers with different mesh numbers, i.e., 800, 1200, 2000, 3000, and 5000 meshes. Moreover, the specimens were mechanically polished. Finally, the specimens were ultrasonically cleaned with acetone to prepare them for the oxygen diffusion process.
The oxygen diffusion process used to prepare the alpha-case consists of the following three steps: (Ι) in order to obtain oxide films with different oxygen content, the tensile specimens were oxidized for various durations at 700 °C in air by employing an electric resistance furnace and then air-cooled to room temperature, which made the oxide layer as an oxygen reservoir; (II) the pre-oxidized specimens with anti-oxidation coating were further diffusion treated in a tube furnace at 930 °C for 2 h with high-purity argon (99.999%) and water-cooled to room temperature; and (III) to avoid high local concentration of oxygen, the specimens were placed in a vacuum furnace (degree of vacuum approximately 1 × 10 −1 Pa) for 6 h at 500 °C for aging treatment. Further, bars of ϕ6 × 70 mm were directly heat-treated by diffusion and aging in steps (II) and (III). Then, they were processed into cylindrical tensile specimens without the alpha-case to compare the tensile properties. To remove anti-oxidation coating and machining marks sintered on the oxygen diffusion specimens, the specimens were ground using abrasive papers with different mesh numbers before the tensile test. The employed meshes were, sequentially, 800, 1200, 2000, 3000, and 5000 meshes. Finally, the specimens were mechanically polished. Heattreatment parameters of all specimens are listed in Table 3. The specimen without alpha-  The raw material was processed by cutting wire into round bars with a diameter of 6 mm and a length of 70 mm (φ6 × 70 mm). Then, cylindrical tensile specimens with a diameter of 4 mm and a gauge length of 20 mm were machined with the tensile axis being parallel to the roll direction. The surface of tensile specimens was ground using abrasive papers with different mesh numbers, i.e., 800, 1200, 2000, 3000, and 5000 meshes. Moreover, the specimens were mechanically polished. Finally, the specimens were ultrasonically cleaned with acetone to prepare them for the oxygen diffusion process.
The oxygen diffusion process used to prepare the alpha-case consists of the following three steps: (I) in order to obtain oxide films with different oxygen content, the tensile specimens were oxidized for various durations at 700 • C in air by employing an electric resistance furnace and then air-cooled to room temperature, which made the oxide layer as an oxygen reservoir; (II) the pre-oxidized specimens with anti-oxidation coating were further diffusion treated in a tube furnace at 930 • C for 2 h with high-purity argon (99.999%) and water-cooled to room temperature; and (III) to avoid high local concentration of oxygen, the specimens were placed in a vacuum furnace (degree of vacuum approximately 1 × 10 −1 Pa) for 6 h at 500 • C for aging treatment. Further, bars of φ6 × 70 mm were directly heat-treated by diffusion and aging in steps (II) and (III). Then, they were processed into cylindrical tensile specimens without the alpha-case to compare the tensile properties. To remove anti-oxidation coating and machining marks sintered on the oxygen diffusion specimens, the specimens were ground using abrasive papers with different mesh numbers before the tensile test. The employed meshes were, sequentially, 800, 1200, 2000, 3000, and 5000 meshes. Finally, the specimens were mechanically polished. Heat-treatment parameters of all specimens are listed in Table 3. The specimen without alpha-case was named as NPO, and the specimens with different alpha-case were named PO10, PO20, PO40, and P060 according to the pre-oxidation time, respectively. Table 3. Heat-treatment parameters.

Specimen Name
Step I (Pre-Oxidation) Step II (Diffusion) Step III (Aging) The phase composition of the specimen surface was analyzed by X-ray diffraction (XRD, PANalytical B.V. Almelo, The Netherlands) with Cu-Kα radiation. The diffraction angle (2θ) was between 10 • and 80 • , and the scanning step was 0.02 • for counting times of 1 s at each step. In-depth concentration profiles of main alloying elements and oxygen were studied on specimens with an alpha-case layer by GD-Profiler 2 glow discharge emission spectrometry (GDS, Zeiss, Thuringia, Germany) with the copper anode. The cylindrical specimen was cut along the diameter, ground, and then polished with standard metallographic methods. Then, it was corroded in a mixed solution consisting of HF, HNO 3 , and H 2 O (volume ratio 3:6:50) for observing its microstructure. The microstructure and fracture surface observations were conducted by the SUPRA40-type field emission scanning electron microscopy (SEM, Zeiss, Analytik Jena City, Germany). Material hardness at cross-sections of different specimens was measured using an HVS-100full automatic microhardness tester (Lunjie Motor Instrument Company, Shanghai, China) fitted with a Vickers indenter. The load of 0.98 N and holding period of 10 s were used as microhardness tester parameters. The tensile test was carried out on the Instron universal testing machine (Instron Corporation, Boston, MA, USA) at room temperature and a tensile rate of 0.5 mm·min −1 . The average value of three measurements was taken in each state to ensure the accuracy of the tensile data. After the tensile test, deformation characteristics near fracture surfaces of different tensile specimens were observed by OLS4500 laser confocal microscopy (LSM, Olympus Corporation, Tokyo, Japan). Finally, transmission electron microscopy (TEM: Talos F200X, American FEI Company, Hillsboro, OR, USA) specimens of the alpha-case were prepared by focused ion beam (FIB) lift-out technology near the tensile fracture, and specimens were observed by TEM at 200 kV. Figure 2 exhibits the XRD patterns of TC4 alloy surface without pre-oxidation and under different pre-oxidation times, respectively. Compared to with the specimens without pre-oxidation treatment (the initial material), the specimens after pre-oxidation treatment produce titanium oxides on their surface. Moreover, the diffraction peaks of α-Ti and β-Ti were also detected, which were caused by the penetration of X-rays onto the alloy matrix through the oxide film. At the initial stage of oxidation (10 min), some unstable suboxides, in addition to rutile, were also detected on the surface. With an increase in the oxidation time, only stable rutile can be detected, which has more diffraction peaks and higher peak intensities. The low-intensity α-Ti diffraction peak at 57.13 • decreases with an increase in the oxidation time, which is caused by the gradual increase in the oxide film thickness. This demonstrates that the surface oxide content gradually increases with an increase in pre-oxidation time. This provided oxygen sources with different oxygen content for the formation of different alpha-cases. increase in pre-oxidation time. This provided oxygen sources with different oxygen content for the formation of different alpha-cases.

Microstructure, Microhardness, and Distribution of Surface Layer Oxygen and Main Alloying Elements after Diffusion and Aging Heat Treatment
After diffusion in the α + β dual-phase region at a temperature of 930 °C and aging temperature of 500 °C, the matrix of all specimens was converted into a bi-modal microstructure composed of primary α phase (αp) and β phase transition structure (βtrans), such as that in the NPO specimen ( Figure 3a). The surface layer of pre-oxidized specimens forms a typical alpha-case gradient structure in which the volume fraction of αp gradually decreases and the volume fraction of βtrans gradually increases (Figure 3b-d). For the quantitative assessment of the changes in the surface structure, volume fractions of αp and βtrans in the range of 50 μm on the surface of all specimens (such as the red line in Figure 3d) were calculated by using Image Pro Plus software. The relationship between the two phases with the pre-oxidation time is presented in Table 4. In the table, the proportion of two types of structures in the range of the NPO specimen is equal to that of the matrix: αp (Vol.%) is 45.7%, βtrans (Vol.%) is 54.3%. With an increase in the pre-oxidation time, the layer volume fraction of αp increases, while that of βtrans decreases. This change is the largest in the PO60 specimen, where the volume fraction of αp increased to 65.4%, and that of βtrans decreased to 34.6%. There is no obvious dividing line between the alpha-case and the matrix; therefore, the thickness of the alpha-case cannot be accurately evaluated by considering the microstructural changes.

Microstructure, Microhardness, and Distribution of Surface Layer Oxygen and Main Alloying Elements after Diffusion and Aging Heat Treatment
After diffusion in the α + β dual-phase region at a temperature of 930 • C and aging temperature of 500 • C, the matrix of all specimens was converted into a bi-modal microstructure composed of primary α phase (α p ) and β phase transition structure (β trans ), such as that in the NPO specimen ( Figure 3a). The surface layer of pre-oxidized specimens forms a typical alpha-case gradient structure in which the volume fraction of α p gradually decreases and the volume fraction of β trans gradually increases (Figure 3b-d). For the quantitative assessment of the changes in the surface structure, volume fractions of α p and β trans in the range of 50 µm on the surface of all specimens (such as the red line in Figure 3d) were calculated by using Image Pro Plus software. The relationship between the two phases with the pre-oxidation time is presented in Table 4. In the table, the proportion of two types of structures in the range of the NPO specimen is equal to that of the matrix: α p (Vol.%) is 45.7%, β trans (Vol.%) is 54.3%. With an increase in the pre-oxidation time, the layer volume fraction of α p increases, while that of β trans decreases. This change is the largest in the PO60 specimen, where the volume fraction of α p increased to 65.4%, and that of β trans decreased to 34.6%. There is no obvious dividing line between the alpha-case and the matrix; therefore, the thickness of the alpha-case cannot be accurately evaluated by considering the microstructural changes.   Figure 4a exhibits the elemental composition distribution in the surface layer of the PO60 specimen obtained by GDS after the oxygen diffusion process. Clearly, oxygen diffusion leads to the inward migration of surface alloy elements. This makes the alloy concentration near the surface relatively low, while it gradually increases toward the matrix. Oxygen concentration is higher than that of the substrate, but it decreases with an increase in the distance from the surface. Figure 4b shows the normalized oxygen concentration ( wt.% normalised ) distribution on the surface of PO10 and PO60 specimens (normalized relative to the known oxygen concentration of the initial material ( wt.% (initial ), i.e., 0.11 wt.%. Measured oxygen concentrations ( wt.% measured ) by GDS were normalized to wt.% (initial by using the following relationship [7]: where O wt.% (average) is the average weight percentage of oxygen concentration within the specimen measured in the substrate after oxygen diffusion. It is noteworthy that, due to limitations of GDS measurement accuracy, normalized oxygen concentration value was not the absolute value of oxygen concentration diffused into the matrix. Figure 4b demonstrates that the oxygen concentration of specimens in two states decreases with an increase in the distance from the surface. The highest values of oxygen concentration near the surface of PO10 and PO60 are about 3 and 6.8 times that of the  Figure 4a exhibits the elemental composition distribution in the surface layer of the PO60 specimen obtained by GDS after the oxygen diffusion process. Clearly, oxygen diffusion leads to the inward migration of surface alloy elements. This makes the alloy concentration near the surface relatively low, while it gradually increases toward the matrix. Oxygen concentration is higher than that of the substrate, but it decreases with an increase in the distance from the surface. Figure 4b shows the normalized oxygen concentration ) by GDS were normalized to O (initial) wt.% by using the following relationship [7]: is the average weight percentage of oxygen concentration within the specimen measured in the substrate after oxygen diffusion. It is noteworthy that, due to limitations of GDS measurement accuracy, normalized oxygen concentration value was not the absolute value of oxygen concentration diffused into the matrix. Figure 4b demonstrates that the oxygen concentration of specimens in two states decreases with an increase in the distance from the surface. The highest values of oxygen concentration near the surface of PO10 and PO60 are about 3 and 6.8 times that of the substrate, respectively. Oxygen concentration distribution value was used to estimate the thickness of the alpha-case (thickness where the oxygen concentration exceeds 0.11wt.%). The thickness of the alpha-case in the two states was obtained and was approximately 110 µm. At the same position in the alpha-case, the oxygen concentration of the PO60 specimen was usually higher than that of the PO10 specimen. This indicates that oxygen concentration in the alpha-case following diffusion gradually increases with an increase in pre-oxidation time.
substrate, respectively. Oxygen concentration distribution value was used to estimate the thickness of the alpha-case (thickness where the oxygen concentration exceeds 0.11wt.%). The thickness of the alpha-case in the two states was obtained and was approximately 110 μm. At the same position in the alpha-case, the oxygen concentration of the PO60 specimen was usually higher than that of the PO10 specimen. This indicates that oxygen concentration in the alpha-case following diffusion gradually increases with an increase in pre-oxidation time. Material hardness is often affected by the elemental composition. Figure 5 exhibits the relationship between microhardness and distance from the surface of the specimens after the oxygen diffusion process. After the oxygen diffusion process, the microhardness increases significantly in the near-surface area of the pre-oxidized specimens. Compared to the matrix hardness (368 HV), the highest increase of microhardness 120 HV appeared on the near-surface region of the PO60 specimen. With an increase of depth from the surface, the hardness gradually decreased and eventually reached the matrix hardness. Figure 5 shows it can be observed that the depth of the hardened layer is around 130 μm. The thickness of this layer is roughly 20 μm thicker than the oxygen diffusion layer tested by GDS. This is related to the difference in measurement accuracy between the two instruments. At the same distance from the surface, the hardness value of the specimen with a longer pre-oxidation time is often higher than that of the specimen with a shorter oxidation time. This is related to the gradual increase in oxygen concentration in alpha-case.  Material hardness is often affected by the elemental composition. Figure 5 exhibits the relationship between microhardness and distance from the surface of the specimens after the oxygen diffusion process. After the oxygen diffusion process, the microhardness increases significantly in the near-surface area of the pre-oxidized specimens. Compared to the matrix hardness (368 HV), the highest increase of microhardness 120 HV appeared on the near-surface region of the PO60 specimen. With an increase of depth from the surface, the hardness gradually decreased and eventually reached the matrix hardness. Figure 5 shows it can be observed that the depth of the hardened layer is around 130 µm. The thickness of this layer is roughly 20 µm thicker than the oxygen diffusion layer tested by GDS. This is related to the difference in measurement accuracy between the two instruments. At the same distance from the surface, the hardness value of the specimen with a longer pre-oxidation time is often higher than that of the specimen with a shorter oxidation time. This is related to the gradual increase in oxygen concentration in alpha-case. substrate, respectively. Oxygen concentration distribution value was used to estimate the thickness of the alpha-case (thickness where the oxygen concentration exceeds 0.11wt.%). The thickness of the alpha-case in the two states was obtained and was approximately 110 μm. At the same position in the alpha-case, the oxygen concentration of the PO60 specimen was usually higher than that of the PO10 specimen. This indicates that oxygen concentration in the alpha-case following diffusion gradually increases with an increase in pre-oxidation time. Material hardness is often affected by the elemental composition. Figure 5 exhibits the relationship between microhardness and distance from the surface of the specimens after the oxygen diffusion process. After the oxygen diffusion process, the microhardness increases significantly in the near-surface area of the pre-oxidized specimens. Compared to the matrix hardness (368 HV), the highest increase of microhardness 120 HV appeared on the near-surface region of the PO60 specimen. With an increase of depth from the surface, the hardness gradually decreased and eventually reached the matrix hardness. Figure 5 shows it can be observed that the depth of the hardened layer is around 130 μm. The thickness of this layer is roughly 20 μm thicker than the oxygen diffusion layer tested by GDS. This is related to the difference in measurement accuracy between the two instruments. At the same distance from the surface, the hardness value of the specimen with a longer pre-oxidation time is often higher than that of the specimen with a shorter oxidation time. This is related to the gradual increase in oxygen concentration in alpha-case.   specimens with and without the alpha-case. After tensile stress reaches UTS, the specimen without alpha-case (NPO) exhibits a relatively long softening stage, which is a typical characteristic of ductile metal materials (Figure 6a). The specimen with low oxygen concentration alpha-case shows a relatively short softening stage (similar to PO10, PO20). With an increase of oxygen concentration in alpha-case, the softening stage becomes shorter and even disappears into a sharp brittle fracture. After heat treatment, strength and ductility indexes UTS, YS, EI, and RA of the specimen without the alpha-case are 1085 MPa, 970 MPa, 13.2% and 51.1%, respectively (Figure 6b). Compared to with the specimen without alphacase, the specimens with alpha-case show a decrease in ductility. With the increase in oxygen concentration in the alpha-case, ductility decreases monotonously and reaches the lowest level (EI 3.6%, RA 5.5%) on PO60 specimens. Interestingly, the strength of specimen with the alpha-case first increased and then decreased, resulting in the existence of peak on the PO20 specimen. For this peak, UTS and YS values are 1155 MPa and 1003 MPa, respectively. Furthermore, a good plastic match is also observed. More specifically, EI and RA are equal to 10.9% and 31.2%, respectively.  Figure 6 exhibits the tensile curve and the changing trend of tensile properties (ultimate tensile strength (UTS), yield strength (YS), reduction of area (RA), elongation (EI)) of specimens with and without the alpha-case. After tensile stress reaches UTS, the specimen without alpha-case (NPO) exhibits a relatively long softening stage, which is a typical characteristic of ductile metal materials (Figure 6a). The specimen with low oxygen concentration alpha-case shows a relatively short softening stage (similar to PO10, PO20). With an increase of oxygen concentration in alpha-case, the softening stage becomes shorter and even disappears into a sharp brittle fracture. After heat treatment, strength and ductility indexes UTS, YS, EI, and RA of the specimen without the alpha-case are 1085 MPa, 970 MPa, 13.2% and 51.1%, respectively (Figure 6b). Compared to with the specimen without alpha-case, the specimens with alpha-case show a decrease in ductility. With the increase in oxygen concentration in the alpha-case, ductility decreases monotonously and reaches the lowest level (EI 3.6%, RA 5.5%) on PO60 specimens. Interestingly, the strength of specimen with the alpha-case first increased and then decreased, resulting in the existence of peak on the PO20 specimen. For this peak, UTS and YS values are 1155 MPa and 1003 MPa, respectively. Furthermore, a good plastic match is also observed. More specifically, EI and RA are equal to 10.9% and 31.2%, respectively.

Fracture Behavior of TC4 Alloy
The fracture surface of NPO, PO20, and PO60 specimens is shown in Figure 7. Different fracture modes exist for specimens with different surface states. The fracture surface of all specimens can be divided into "A", "B", and "C" regions. These regions are defined as the crack initiation region, crack propagation region, and shear lip region, respectively. For the NPO specimen without the alpha-case, the crack originates from the center of the specimen. The crack initiation area is relatively large and the surface is very rough, as shown in Figure 7a. High-magnification analysis indicates that the initiation area is covered with a large number of dimples (Figure 7b), which is a typical ductile fracture mode. This matches the good ductility of specimens without the alpha-case. However, for specimens with the alpha-case, the crack initiation position shifts from the center of the specimen to the surface (Figure 7c,e). Quasi-cleavage fracture characteristics of PO20 specimen with low oxygen concentration alpha-case are obvious. Moreover, the crack initiation area can be seen as a river pattern (Figure 7d). From the perspective of a macroscopic fracture, the fracture of the PO60 specimen with high oxygen concentration alpha-case is relatively flat (Figure 7e). The crack initiation zone is composed of many facets (Figure 7f), with

Fracture Behavior of TC4 Alloy
The fracture surface of NPO, PO20, and PO60 specimens is shown in Figure 7. Different fracture modes exist for specimens with different surface states. The fracture surface of all specimens can be divided into "A", "B", and "C" regions. These regions are defined as the crack initiation region, crack propagation region, and shear lip region, respectively. For the NPO specimen without the alpha-case, the crack originates from the center of the specimen. The crack initiation area is relatively large and the surface is very rough, as shown in Figure 7a. High-magnification analysis indicates that the initiation area is covered with a large number of dimples (Figure 7b), which is a typical ductile fracture mode. This matches the good ductility of specimens without the alpha-case. However, for specimens with the alpha-case, the crack initiation position shifts from the center of the specimen to the surface (Figure 7c,e). Quasi-cleavage fracture characteristics of PO20 specimen with low oxygen concentration alpha-case are obvious. Moreover, the crack initiation area can be seen as a river pattern (Figure 7d). From the perspective of a macroscopic fracture, the fracture of the PO60 specimen with high oxygen concentration alpha-case is relatively flat (Figure 7e). The crack initiation zone is composed of many facets (Figure 7f), with individual facet sizes being similar to the grain size ( Figure 3). This indicates that the fracture model may be a mixture of intergranular and cleavage transgranular fractures for the PO60 specimen. Both these are brittle fracture modes that correspond to poor ductility. individual facet sizes being similar to the grain size ( Figure 3). This indicates that the fracture model may be a mixture of intergranular and cleavage transgranular fractures for the PO60 specimen. Both these are brittle fracture modes that correspond to poor ductility.

Formation of Alpha-Case under Different Oxygen Diffusion Processes
During pre-oxidation of TC4 alloy, the atmospheric oxygen atoms react with the metal through the original thin oxide film, thus forming oxides [11]. The oxides formed by pre-oxidation have a higher oxygen concentration than the matrix. Furthermore, they provide an oxygen source for diffusion [26,27]. The oxygen atoms in the surface oxide film are continuously diffused at high temperature, resulting in the formation of the alphacase with oxygen concentration gradient distribution on the surface of TC4 alloy ( Figure  3b-d). Longer pre-oxidation time leads to an increase in the oxide formed by the TC4 alloy on the surface (Figure 2). More oxides increase the oxygen potential of the oxide film, which results in a higher oxygen concentration in the alpha-case after diffusion ( Figure  4b) [26]. The oxygen in the solid solution can effectively increase the β transformation temperature of the TC4 alloy and widen the α+β two-phase region [28]. As a result, when rolled TC4 alloy diffuses in the α+β two-phase region, as the oxygen concentration increases, the α phase content of the surface layer transforming to the β phase decreases.

Formation of Alpha-Case under Different Oxygen Diffusion Processes
During pre-oxidation of TC4 alloy, the atmospheric oxygen atoms react with the metal through the original thin oxide film, thus forming oxides [11]. The oxides formed by pre-oxidation have a higher oxygen concentration than the matrix. Furthermore, they provide an oxygen source for diffusion [26,27]. The oxygen atoms in the surface oxide film are continuously diffused at high temperature, resulting in the formation of the alpha-case with oxygen concentration gradient distribution on the surface of TC4 alloy (Figure 3b-d). Longer pre-oxidation time leads to an increase in the oxide formed by the TC4 alloy on the surface (Figure 2). More oxides increase the oxygen potential of the oxide film, which results in a higher oxygen concentration in the alpha-case after diffusion (Figure 4b) [26]. The oxygen in the solid solution can effectively increase the β transformation temperature of the TC4 alloy and widen the α+β two-phase region [28]. As a result, when rolled TC4 alloy diffuses in the α+β two-phase region, as the oxygen concentration increases, the α phase content of the surface layer transforming to the β phase decreases. More specifically, with an increase in pre-oxidation time of specimen, the α p volume fraction is increased, while β trans volume fraction in the surface layer is decreased after diffusion (Table 3). Based on the existing studies, a linear relationship between the hardness of the titanium alloy and oxygen concentration is observed [29,30]. Therefore, with an increase in the pre-oxidation time, the microhardness in the alpha-case is also increased ( Figure 5).

Effect of Alpha-Case on Strength and Ductility of TC4 Alloy
Compared to TC4 alloy tensile specimens without the alpha-case, crack initiation zones of alpha-case specimens all shifted from the core of the specimen to the surface layer (Figure 7c,e). With the increase in oxygen concentration in the alpha-case, the ductility gradually decreases. Moreover, brittleness characteristics of the crack initiation area gradually become more significant (Figure 7d,f). This indicates that the ductility of the TC4 alloy is related to the plasticity of the alpha-case. To further explore the relationship among the alpha-case, the strength, and the ductility of specimen, LSM was used to observe surface deformation characteristics near the specimen fracture, as shown in Figure 8. Figure 8a presents that the NPO specimen shows rough bumpy features near the fracture. The surface roughness near the fracture was measured to be 51.0 µm (Ra). Obvious necking characteristics can be observed, which correspond to the good plasticity behavior of material. Near the fracture site of the PO20 specimen, bumpy and necking features after deformation are visible (Figure 8b). However, the surface roughness is reduced to 34.4 µm (Ra). It can be concluded that the degree of plastic deformation on the surface of the PO20 specimen is relatively small. More transverse microcracks perpendicular to the tensile direction can be observed near the fracture site of the PO60 specimen (Figure 8c). The measured roughness of this location is 9.9 µm (Ra). Furthermore, the vicinity of the fracture is relatively flat and no necking phenomenon is observed. There is almost no plastic deformation on the surface of the PO60 specimen, which is attributed to the brittle fracture of specimen.
When oxygen diffuses to the surface of the alloy to form the alpha-case, the plasticity of the surface worsens significantly more than that of the matrix (Figure 8). This causes the initiation of cracks first on the surface. As tensile deformation increases, surface cracks propagate into the matrix, leading to premature fracture of TC4 alloy and deterioration of ductility. The alpha-case with low oxygen concentration still has a certain degree of plasticity, and the crack sensitivity is relatively low. Therefore, cracks initiate on the surface after a long period of tensile deformation. Then, they propagate to the matrix in a quasicleavage fracture mode. Owing to relatively slow crack propagation speed, the specimen exhibits an obvious softening stage. At the same time, the alpha-case strengthened by oxygen in solid solution contributes to the TC4 alloy specimen strength; therefore, the overall strength of TC4 alloy is increased. The alpha-case with high oxygen concentration transforms into a brittle layer, which has higher crack sensitivity. Low deformation causes cracks to initiate and spread rapidly, resulting in a sharp drop in plasticity [19,20]. Parthasarathy et al. [18] conducted tensile experiments on near-alpha titanium alloy with the alpha-case. They found that surface cracks reached saturation when the strain was 1.36% or less. For the specimen with a high-concentration alpha-case used in this study, cracks are most likely to form before the yield point. This, in turn, reduces the load-bearing area of the specimen and causes both UTS and YS to drop.
The formation of transverse cracks is directly related to a sharp drop in ductility and fracture mode. Therefore, cracks of the PO60 specimen were cut along the tensile direction, and the crack propagation path was observed by SEM. Figure 9 exhibits the corresponding results, revealing that the crack width and depth decrease with an increase in the distance from the fracture site (Figure 9a), which represents a mode-I crack [31]. The depth of the largest microcrack near the fracture surface reaches approximately 270 µm, which is more than twice the thickness of the alpha-case (130 µm) measured by the microhardness method. This indicates that higher tensile deformation causes a more severe expansion of the crack toward the matrix. It is thus intuitively proved that the formation of surface cracks triggers the fracture of TC4 alloy. Crack initiation in the alpha-case is usually controlled by the macroscopic load, and it is not sensitive to the local structure [31]. Figure 9b illustrates that the crack propagation mode is a mixed-mode of transgranular and intergranular growth. This is also confirmed by small facets of the crack initiation zone of the PO60 specimen (Figure 7f). Affected by the tensile stress, small holes are formed at the high-angle grain boundary and the three points in the alpha-case [31]. In metal fracture, it is generally believed that the formation of pores is related to beneficial plasticity [32]. However, the formation of pores in the alpha-case occurs below the macroscopic yield stress of the alloy [31]. During the stretching process, the crack propagation is always connected with the hole by low energy and short distance direction. Therefore, when the crack is closer to the next hole in transgranular form than in intergranular form, the crack expands via transgranular growth and vice versa. Figure 9b shows that the crack propagation path is relatively flat near the surface Region 1. With an increase in depth in the direction of the substrate (Region 2), the crack propagation becomes tortuous, which may be related to a decrease in the oxygen concentration.
alloy is related to the plasticity of the alpha-case. To further explore the relationship among the alpha-case, the strength, and the ductility of specimen, LSM was used to observe surface deformation characteristics near the specimen fracture, as shown in Figure  8. Figure 8a presents that the NPO specimen shows rough bumpy features near the fracture. The surface roughness near the fracture was measured to be 51.0 μm (Ra). Obvious necking characteristics can be observed, which correspond to the good plasticity behavior of material. Near the fracture site of the PO20 specimen, bumpy and necking features after deformation are visible (Figure 8b). However, the surface roughness is reduced to 34.4 μm (Ra). It can be concluded that the degree of plastic deformation on the surface of the PO20 specimen is relatively small. More transverse microcracks perpendicular to the tensile direction can be observed near the fracture site of the PO60 specimen (Figure 8c). The measured roughness of this location is 9.9 μm (Ra). Furthermore, the vicinity of the fracture is relatively flat and no necking phenomenon is observed. There is almost no plastic deformation on the surface of the PO60 specimen, which is attributed to the brittle fracture of specimen.

Evolution of the Slip Mechanism in the Alpha-Case
In general, the plasticity of metal materials is closely related to their dislocation slip characteristics. To explore the embrittlement of the alpha-case, FIB was employed to cut the alpha-case. Moreover, slip deformation characteristics of the alpha-case are obtained via TEM. Figure 10a,b show cutting positions near the fracture of PO20 specimen with low-concentration alpha-case and PO60 specimen with high-concentration alpha-case. The surface near the fracture of PO20 specimen is covered with a large number of deformation slip characteristics (shown by the arrow in Figure 10). There is no obvious plastic deformation on the surface of PO60 specimen. Instead, some microcracks are distributed

Evolution of the Slip Mechanism in the Alpha-Case
In general, the plasticity of metal materials is closely related to their dislocation slip characteristics. To explore the embrittlement of the alpha-case, FIB was employed to cut the alpha-case. Moreover, slip deformation characteristics of the alpha-case are obtained via TEM. Figure 10a,b show cutting positions near the fracture of PO20 specimen with low-concentration alpha-case and PO60 specimen with high-concentration alpha-case. The surface near the fracture of PO20 specimen is covered with a large number of deformation slip characteristics (shown by the arrow in Figure 10). There is no obvious plastic deformation on the surface of PO60 specimen. Instead, some microcracks are distributed on the surface (Figure 10b). This corresponds to surface deformation characteristics observed by LSM (Figure 8). Figure 10c,d exhibit dislocation characteristics of the alpha-case of PO20 and PO60 specimens, respectively. Based on the result of selected area electron diffraction (SAED) and the structure of the selected area shown in figure, it can be concluded that α p phase is present. When the oxygen concentration is low in alpha-case, tensile deformation causes more dislocations in the α p within the alpha-case. Dislocations are in the form of bending and cross-entanglement (shown by the arrow in Figure 10c). When the dislocation is in the form of bending and cross-entanglement, it means that its slip can proceed in multiple directions; therefore, the slip of this kind of dislocation is named wavy slip [33]. Obviously, it can be judged from the dislocation line of the bending and cross-entanglement in Figure 10c that there is a wavy slip in the low oxygen concentration alpha-case. When the oxygen concentration is high in alpha-case, only a few dislocations in the α p within the alpha-case are present. Dislocations are in the form of parallel lines (Figure 10d), indicating a planar slip. In metals, the plane slip has shown adverse effects on plasticity [21]. The high-angle annular dark-field (HAADF) images of the corresponding areas in Figure 10c,d are obtained through the scanning transmission electron microscope mode of TEM. The results are shown in Figure 10e,f. More white fine particles are dispersed in the α p phase. This is a typical feature of α 2 precipitation in titanium alloys, and researchers have used dark field (DF) mode of the TEM and atom probe technology (APT) to describe the precipitation of α 2 in detail [34]. Other authors have confirmed that the precipitation of α 2 is orderly, and the interface relationship between α 2 and the matrix is coherent [23]. In SAED pattern, no diffraction spots of α 2 were found. This may be attributed to its too small particle diameter. Bagot et al. [35] used APT to prove that oxygen atoms in alpha-case mainly exist in two forms, both of which are dissolved in α-Ti and α 2 . These components are Ti70-O30 and Ti65-O10-Al20-V5, respectively. Compared to α-Ti, oxygen is more preferentially distributed in α 2 [28]. This makes it possible to observe the precipitation of α 2 in a shorter aging time.
Combined with TEM characterization of the alpha-case, it can be inferred that there are two possibilities for the planar slip caused by the alpha-case. On the one hand, it may be affected by ordered and coherent precipitate α 2 . One of the distinguishing features of ordering is the existence of antiphase domains (APD) and their associated anti-phase boundaries (APB) [36]. That is, there is also an APB between the ordered α 2 particle in α-Ti and α-Ti [23]. The APB affects the sliding of a dislocation pair. In other words, when the moving dislocation occurs through the ordered precipitates, it forms a narrow and strong slip band [37]. The α 2 in the titanium alloy has also been observed to have a similar effect [20]. During tensile deformation, ordered α 2 particles are cut by moving dislocations. This leads to a local reduction in the resolved shear stress along active slip planes. However, the movement of dislocations becomes more difficult on other planes, which is the reason behind the sliding of a titanium alloy plane [20,38]. On the other hand, planar slip may be caused by the influence of oxygen solid solution. Williams et al. [33] observed that unalloyed α-Ti with an oxygen content of approximately 500 ppm was deformed by a wavy slip at room temperature. At approximately 1800 ppm oxygen, the slip was observed as distinctly planar. Increasing oxygen content could change the slip behavior of dislocations in α-Ti. It is noteworthy that precipitation of α 2 was observed in both alpha-case with low and high oxygen concentrations, but no planar slip was observed in low oxygen concentration alpha-case. This indicates that the generation of a plane slip in alpha-case in this study is not dominated by α 2 . Combined with TEM characterization of the alpha-case, it can be inferred that there are two possibilities for the planar slip caused by the alpha-case. On the one hand, it may be affected by ordered and coherent precipitate α2. One of the distinguishing features of ordering is the existence of antiphase domains (APD) and their associated anti-phase boundaries (APB) [36]. That is, there is also an APB between the ordered α2 particle in α-Ti and α-Ti [23]. The APB affects the sliding of a dislocation pair. In other words, when the moving dislocation occurs through the ordered precipitates, it forms a narrow and strong slip band [37]. The α2 in the titanium alloy has also been observed to have a similar effect [20]. During tensile deformation, ordered α2 particles are cut by moving dislocations. This leads to a local reduction in the resolved shear stress along active slip planes. However, the movement of dislocations becomes more difficult on other planes, which is the reason behind the sliding of a titanium alloy plane [20,38]. On the other hand, planar slip may be caused by the influence of oxygen solid solution. Williams et al. [33] observed that unalloyed α-Ti with an oxygen content of approximately 500 ppm was deformed by a wavy slip at room temperature. At approximately 1800 ppm oxygen, the slip was observed as distinctly planar. Increasing oxygen content could change the slip behavior of dislocations in α-Ti. It is noteworthy that precipitation of α2 was observed in both alphacase with low and high oxygen concentrations, but no planar slip was observed in low For the comprehensive understanding of the solid solution of oxygen in different alphacases, XRD was conducted on specimens before tensile tests. Figure 11 exhibits the (103) crystal plane diffraction peaks of α-Ti of NPO, PO20, and PO60 specimens. Compared to NPO specimen without the alpha-case, diffraction peaks (103) crystal plane for other specimens shifted to lower angles. This is attributed to the lattice expansion caused by the solid solution of oxygen in α-Ti [39]. Notably, the (103) crystal plane diffraction peak of α-Ti in PO60 specimen shows a larger shift angle, which is related to the further lattice expansion caused by an increase in the amount of solid solution of oxygen. Therefore, MDI Jade software was used to calculate lattice parameters of different alpha-case α-Ti from the XRD data. The results are presented in Table 5, revealing that the lattice parameter of α-Ti increases due to a solid solution of oxygen. Moreover, the lattice parameter increases with an increase in the oxygen concentration within the alpha-case. This indicates that α-Ti in alpha-case with high oxygen concentration exhibits a higher dissolved amount of oxygen. Table 5 also summarizes that the lattice parameter expansion in α-Ti caused by oxygen in solid solution is mainly concentrated on the c-axis. This is attributed to the fact that when oxygen is dissolved in α-Ti, it tends to occupy the octahedral interstices [28]. Compared to the strain along the a-axis and the c-axis, this occupancy causes the lattice strain along the Z-axis in the crystal lattice to change significantly [40]. lattice expansion caused by an increase in the amount of solid solution of oxygen. Therefore, MDI Jade software was used to calculate lattice parameters of different alpha-case α-Ti from the XRD data. The results are presented in Table 5, revealing that the lattice parameter of α-Ti increases due to a solid solution of oxygen. Moreover, the lattice parameter increases with an increase in the oxygen concentration within the alpha-case. This indicates that α-Ti in alpha-case with high oxygen concentration exhibits a higher dissolved amount of oxygen. Table 5 also summarizes that the lattice parameter expansion in α-Ti caused by oxygen in solid solution is mainly concentrated on the c-axis. This is attributed to the fact that when oxygen is dissolved in α-Ti, it tends to occupy the octahedral interstices [28]. Compared to the strain along the a-axis and the c-axis, this occupancy causes the lattice strain along the Z-axis in the crystal lattice to change significantly [40]. Figure 11. XRD diffraction peaks of α-Ti (103) crystal plane of NPO, PO20, and PO60 specimens. In this study, an increase in oxygen concentration of high concentration alpha-case, compared to the low-concentration alpha-case, also resulted in an increase of oxygen solubility in α-Ti at the alpha-case. Basal and prismatic dislocations' nucleation was blocked at high oxygen solubility, while <c+a> pyramidal dislocation's emission is reduced even for low oxygen solubility [41]. In other words, with an increase in oxygen solubility, a Figure 11. XRD diffraction peaks of α-Ti (103) crystal plane of NPO, PO20, and PO60 specimens. In this study, an increase in oxygen concentration of high concentration alpha-case, compared to the low-concentration alpha-case, also resulted in an increase of oxygen solubility in α-Ti at the alpha-case. Basal and prismatic dislocations' nucleation was blocked at high oxygen solubility, while <c+a> pyramidal dislocation's emission is reduced even for low oxygen solubility [41]. In other words, with an increase in oxygen solubility, a proliferation of dislocations becomes more difficult. At higher dissolution rates, dislocation slip is even more highly localized, which results in a planar slip. When planar slips in different grains meet at the grain boundaries, cracks' initiation is very likely. This makes the cracks in the alpha-case initiate for a relatively small amount of strain, which may lead to the embrittlement of the alpha-case. At the same time, narrow and strong planar slip is confined in front of the crack tip, thus providing a strong driving force for crack propagation [31]. This allows for rapid crack propagation and adversely affects the strength and ductility of TC4 alloy.

Conclusions
In this study, the effect of different alpha-case on the tensile properties of TC4 alloy was studied. Based on the obtained results, the following conclusions can be drawn:

1.
After the rolled TC4 alloy is treated by oxygen diffusion process, the surface layer forms a typical alpha-case gradient structure in which the volume fraction of α p phase gradually decreases and the volume fraction of β trans gradually increases. The oxygen