The results of the three-point bending fracture tests for the high-Nb TiAl alloy specimens are presented in Table 2. From Table 2, it can be seen that the average fracture work per unit area (W′) of the fully lamellar structured specimens is 3.479 × 10⁻³ J/mm², and the average maximum fracture toughness (K_max) is 16.337 MPa·m1/2. Among all the samples, 3 PB-03 exhibited the highest fracture work and fracture toughness. This may be related to the fact that specimen 3 PB-03 experienced more interlamellar fracture during the fracture process. Since interlamellar cracking requires greater energy for propagation, specimens with higher proportions of translamellar fracture surfaces tend to exhibit higher fracture energy and fracture toughness.

As shown in Figure 4, the load–displacement curves of specimens 3 PB-01, 3 PB-02, 3 PB-03, and 3 PB-05 have almost identical slopes, while specimen 3 PB-04 shows a similar trend but slightly lower load response. All curves are located within the elastic stage, indicating that the material underwent sudden cleavage fracture under a certain external load rather than progressive crack growth and coalescence with increasing load.

To compare the fracture toughness of the materials, the maximum fracture toughness values of the fine-grained lamellar specimens and the coarse-grained lamellar specimens were analyzed (Shen et al., 2015). It was found that the fine-grained lamellar specimens exhibited higher maximum fracture toughness values than the coarse-grained ones, indicating that the fracture toughness of the alloy decreases with the refinement of lamellar colonies and the increase of colony size.

The fracture morphology of specimen 3 PB-01 under three-point bending is shown in Figure 5. As shown in Figure 5a, the specimen exhibits a main crack initiation site (indicated by solid arrow 1) at the notch root and a distinct crack propagation trace in the form of cleavage river patterns. This main crack was generated by stress concentration at the notch root. Although the specimen contained a machined notch, the stress was relatively uniformly distributed because of the square cross-section (approximately 6 mm thick). The weakest region near the notch root still contained more than 30 lamellar colonies, whose stresses were comparable; therefore, considerable energy accumulated ahead of the notch before fracture, resulting in the formation of a primary cleavage crack. At the same time, another secondary crack initiation site appeared at a certain distance from the notch root, as indicated by solid arrow two in Figure 5a.

Compared with the tensile specimens studied by previous scholars (Wang et al., 2020), the three-point bending specimens also exhibited obvious initiation sites and river-like fracture patterns. However, the tensile specimens had a single crack initiation site, whereas the bending specimens exhibited one main and one or two secondary initiation sites. Figure 5b shows the fracture surface photographed from the region between the two initiation sites near the notch root. The river-like pattern extending from the main initiation site can be clearly observed, confirming that a dominant main crack initiated and propagated in this region. The distance between the main initiation site and the notch root was approximately 70 μm, and this crack played an important role in the final fracture. Further analysis revealed that many translamellar river-like patterns appeared on the lamellar planes of the fracture surface, as shown in Figure 5c. This indicates that the fracture process of the alloy during bending was not a gradual crack nucleation and growth process; instead, due to the concentrated main stress field localized in a narrow region ahead of the notch root, cracks preferentially initiated along weak lamellar cleavage planes and then propagated in a sudden brittle manner.

The fracture surface of specimen 3 PB-02 under three-point bending is shown in Figure 6. The crack initiation site is still located at the notch root, as indicated by the solid arrow in Figure 6a, and the size of the initiation region is approximately 200 μm × 88 μm. Once this initiation region-formed along lamellar interfaces—was activated, it triggered cleavage fracture throughout the entire specimen. The fracture surface consists of both interlamellar and translamellar facets, and a magnified view of its fracture path and river pattern direction is shown in Figure 6b. The macroscopic load–displacement curve of 3 PB-02 shows a slight deviation from the purely elastic stage (Figure 4), which is consistent with the presence of a well-defined cleavage origin at the notch root. Overall, 3 PB-02 failed in a brittle-dominated manner initiated at the notch root.

The four-point bending fracture test results for the high Nb–TiAl alloy specimens are summarized in Table 3. As shown in the table, the fracture work per unit area of the specimens ranges from 0.8 × 10⁻² to 1.0 × 10⁻² J/mm². The specimens with a single notch exhibited slightly higher fracture work per unit area than those with double notches, but the difference in overall bending performance was not significant. The corresponding load–displacement curves of the four-point bending fracture tests are shown in Figure 7, indicating that the load and displacement maintained a nearly linear relationship, and fracture occurred during the elastic stage.

The fracture morphology of the single-notched four-point bending specimen 4 PB-02 is shown in Figure 8. Analysis revealed a main crack initiation site on the fracture surface, as indicated in Figure 8a, together with distinct river-like patterns, which resulted from a relatively uniform stress distribution at the notch root. Although the four-point bending specimen contained a notch and stress concentration did occur at the notch root, the specimen’s square cross-section (about 6 mm thick) contained more than 40 lamellar colonies, whose stresses were roughly equivalent. Consequently, before fracture, substantial energy was accumulated at the notch front, leading to the formation of a dominant cleavage crack that played a critical role in the overall fracture of the specimen. As shown in Figure 8b, secondary cracks were also observed near the notch root between two main initiation sites, indicating that crack propagation started close to the notch and extended along multiple lamellar boundaries. Numerous translamellar and interlamellar river patterns were observed across the fracture surface, as illustrated in Figure 8c, suggesting that the fracture process was not a gradual nucleation-and-growth mechanism. Instead, although the main stress field was confined to a narrow region ahead of the notch root, the crack preferentially initiated in locally weaker zones, where cleavage occurred along multiple lamellar planes. The river patterns displayed well-defined propagation directions, as shown in Figure 8d. Overall, the fracture process can be summarized as follows: after the initial crack initiation, continuous loading caused energy accumulation until a critical level was reached, at which point catastrophic cleavage fracture occurred throughout the specimen.

During the analysis of the double-notched specimen 4 PB-10, it was found that the fracture surface no longer exhibited a single crack initiation site. Instead, two main crack initiation sources were formed, as indicated by the two small arrows at the notch roots in Figure 9. These cracks propagated along their respective paths independently: some propagated directly toward the specimen edges, while others intersected with each other, as shown by the long arrow in Figure 9. After intersecting, the cracks continued to propagate along their original directions, maintaining nearly constant propagation intensity and velocity, until they extended to the specimen boundaries and caused complete fracture. The fracture mechanism can thus be attributed to the competition between different crack initiation sites and their respective propagation paths, ultimately leading to a cleavage-type fracture of the specimen.

The macroscopic unloading test results of notched three-point bending specimens of the high-Nb TiAl alloy are shown in Table 4.

The SEM images of the cross-sections of partially unloaded three-point bending specimens are shown in Figure 10. The unloaded surface of specimen 3 PB-08, which was subjected to a loading force of 389 N, is presented in Figure 10a. No apparent damage or cracks can be observed on the specimen surface; however, numerous unfavorably oriented borides are distributed near the notch root of specimen 3 PB-08. These borides effectively hindered crack initiation and propagation at the notch root.

With the increase of unloading load, material damage could already be observed in specimen 3 PB-09, which was subjected to a loading force of 540 N, as indicated by the arrows and the black-outlined region in Figure 10b. The damage appeared at a certain distance from the notch root rather than directly at the root, and it could only be clearly identified under higher magnification. As shown in Figure 10b, the absence of crack initiation from the notch root was mainly attributed to the unfavorable grain orientation in that region. It can also be observed that cracks tended to accumulate and align parallel to each other within the lamellar colonies that had favorable orientations, suggesting that this phenomenon might be related to interlamellar cracking. These lamellae were aligned in parallel, and when local stress concentration occurred, parallel microcracks were generated. However, careful examination revealed that the cracks were still very small.

At an unloading load of 706 N, clear crack morphology extending from the notch root can be observed in specimen 3 PB-11, as shown in Figure 10c. In the image, the crack initiated at the center of the notch root and propagated along a favorable orientation that was nearly parallel to the loading direction. This indicates that in the notched three-point bending tests of the high-Nb TiAl alloy, when the grains at the notch root were favorably oriented, crack initiation first occurred within those grains, followed by crack propagation along favorable colony boundaries or interlamellar regions. During crack propagation, the externally applied energy was released gradually. However, when the grain orientation at the crack tip was unfavorable, the accumulated strain energy could not be efficiently released, leading to a sudden and complete brittle cleavage fracture once the stored energy reached a critical level.

When the load increased to 790 N, a dense crack zone was observed on the specimen surface, as shown in Figure 10d. However, in specimen 3 PB-12, no cracks were found near the notch root; instead, a large number of cracks appeared at a certain distance away from it. This behavior was similar to that of specimen 3 PB-08, except that the grain misorientation near the notch root in 3 PB-12 was more pronounced, and the unfavorably oriented grains were larger in size. As a result, crack initiation at the notch root was extremely difficult, and cracks could only nucleate within favorably oriented grains located farther away from the notch. Consequently, specimen 3 PB-12 exhibited higher bending resistance than the other specimens.

From Figure 10 and Table 4, it can be concluded that the number of cracks in the notched high-Nb TiAl alloy specimens during the three-point bending unloading tests generally increased with the rise of unloading load.

Numerous borides of varying sizes were also observed on the surface of the unloaded specimens. To identify these borides, an XRD analysis was performed on the notch region of specimen 4 PB-11, as shown in Figure 11. The large-sized borides tended to induce dislocation pile-ups, and when these borides—acting as ligament bridges—were pulled out or fractured, they effectively hindered the further propagation of microcracks or altered their propagation paths. This behavior significantly increased the resistance to crack initiation and growth. In addition, the small-sized borides contributed to second-phase strengthening. Therefore, the dispersed distribution of borides markedly enhanced the fracture toughness of the high-Nb TiAl alloy. However, as shown in Figure 10, many damage sites were found within approximately 100 μm from the notch root. These microcracks tended to accumulate and align parallel to each other within lamellar colonies that had favorable orientations, which was likely related to interlamellar cracking. The parallel arrangement of these lamellae led to the formation of parallel cracks under local stress concentration. Nevertheless, detailed observation revealed that these cracks were still very small. Such damage and microcracks facilitated the partial release of the accumulated strain energy in the specimen. Comparison of these experimental results reveals that the fracture mode of the notched high-Nb TiAl alloy under bending is brittle cleavage fracture. During the fracture process, different main crack initiation sites propagated along their respective directions toward the specimen boundaries, indicating a cleavage fracture controlled by crack initiation.

To investigate the crystallographic features involved in the brittle fracture process of the high-Nb TiAl alloy, TEM analyses were conducted on the microstructures in the vicinity of the fracture surfaces of the bent specimens. The microscopic morphology at the interfaces between lamellar colonies and their colony boundaries is shown in Figure 12. Figure 12a shows the bright-field TEM image of the region near the fracture surface of specimen 3 PB-07 unloaded at a bending load of 204 N. At this stage, no obvious dislocations were observed in the material. The diffraction spots corresponding to the two alternating phases within the lamellar colonies are marked in the lower-left and upper-right corners of Figure 12a. As shown in all subfigures of Figure 12, the bright regions correspond to the γ phase (TiAl), whereas the dark regions correspond to the α₂ phase (Ti₃Al). In Figure 12b, a typical dislocation network (indicated by arrow A) and twin bands (indicated by arrow B) are present on the cross-section of specimen 3 PB-09 unloaded at 540 N. Figure 12c presents the TEM micrograph of the lamellar colony region near the fracture surface of specimen 3 PB-01 after bending fracture at 750 N. Numerous twins were emitted from the γ/α₂ lamellar interfaces and propagated for a short distance within the γ lamellae before terminating at the γ/α₂ interface. Most of the twins formed an angle of approximately 45° with the γ/α₂ interface. Figure 12d shows the microstructure near the fracture surface of specimen 3 PB-02 after bending fracture at 796 N. A large number of dislocations and dense twin bands can be observed, and the deformation of the lamellar colonies remained primarily concentrated within the γ lamellae. Both twinning and dislocation formation could occur at the γ/α₂ interfaces, as illustrated in Figure 12d. The interaction between twins and the lamellar interfaces could induce dislocations to bow out from the interface.

Based on the analysis, the deformation of the high-Nb TiAl alloy during bending was primarily governed by twinning and dislocation activity. Because the deformation of the γ lamellae was constrained by the neighboring α₂ lamellae, deformation preferentially occurred at the lamellar colony boundaries. Dislocations were unable to pass through the interfaces between the lamellar colonies and the colony boundaries, leading to dislocation pile-ups at these interfaces. As the applied stress increased, dislocations and dislocation arrays bowed out from the γ/α₂ interfaces, and twins were emitted from the same interfaces. This indicates that the γ and α₂ lamellae exhibit different deformation characteristics, resulting in significant interface stresses. The twins, dislocations, and dislocation arrangements generated at the interfaces effectively alleviated local stress concentrations, thereby enhancing the plasticity and damage tolerance of the high-Nb TiAl alloy.

From the above bending test results and analysis, it can be concluded that in the notched three-point bending tests of the high-Nb TiAl alloy, when the grains at the notch root were favorably oriented, crack initiation first occurred within those grains. The cracks then propagated along favorable colony boundaries or interlamellar regions, during which the external work was released gradually. However, when the orientation at the crack tip was unfavorable, the strain energy continued to accumulate near the crack front. Once the strain energy in this region reached a critical level, the crack nucleated and rapidly propagated, leading to fracture. The fracture mode conformed to the characteristics of a completely brittle cleavage fracture, and the corresponding crack propagation path is illustrated in Figure 13. The differences in surface crack length observed in the unloaded specimens under different external loads indicate that the locations of the crack initiation regions varied. Nevertheless, this does not imply that the fracture process of this material involves crack nucleation, coalescence, and gradual propagation until final failure. Rather, it suggests that under different unloading loads, only the size of the regions responsible for initiating cleavage fracture differs.