Effects of crystallographic texture on subsurface fatigue crack generation in Ti-Fe-O alloy at low temperature

593 ISIJ International, Vol. 62 (2022), No. 3, pp. 593–601 https://doi.org/10.2355/isijinternational.ISIJINT-2021-381 * Corresponding author: E-mail: [email protected] © 2022 The Iron and Steel Institute of Japan. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs license (https://creativecommons.org/licenses/by-nc-nd/4.0/). Effects of Crystallographic Texture on Subsurface Fatigue Crack Generation in Ti–Fe–O Alloy at Low Temperature Osamu UMEZAWA1,2)* and Weibo LI3) 1) Faculty of Engineering, Yokohama National University, 79-5 Tokiwadai, Hodogaya, Yokohama, 240-8501 Japan. 2) Center of Advanced Innovation Technologies, Vysoká Škola Báňská - Technical University of Ostrava, 17. listopadu 15, 708 33 Ostrava-Poruba, Czech Republic. 3) Graduate School of Engineering, Yokohama National University. Now at Central Research Laboratories, Nihon Parkerizing Co., Ltd., 2784 Okami, Hiratsuka, Kanagawa, 254-0012 Japan. (Received on August 18, 2021; accepted on November 8, 2021; J-STAGE Advance published date: December 29, 2021) Subsurface microcracks developed in a groove-rolled and cold-swaged Ti–Fe–O alloy were characterized to clarify the generation of subsurface fatigue crack. In addition, the effects of crystallographic texture on subsurface crack initiation and growth were discussed. A considerable number of microcracks were detected in the β grains, α grains, and at the α - β interface. The microcracks in the β grains grew negligibly into the neighboring α grains along the basal plane. This was because these grains were oriented with their c-axis almost perpendicular to the loading axis. The 10 10 {} α fiber texture prevents the formation of basal facet and its growth on the basal plane. The stress concentration around the microcrack in the β grains could assist the growth of the microcrack into neighboring α grains along the prismatic plane (which is inclined to the loading axis at a suitable angle) or occasionally at a 10 10 {} α twist boundary. The 10 10 {} α fiber texture assisted microcrack growth, and thereby, formed aligned facets and yield longer microcrack length. The combination of the shear stress and opening stress on 10 10 {} α results in a Mode II or III microcrack and causes microcrack growth on the prismatic plane in the neighboring grain. KEY WORDS: titanium alloys; high cycle fatigue; texture; subsurface crack; electron backscatter diffraction. 1. Introduction Subsurface fatigue crack initiation sites in nearα type and α - β type titanium alloys commonly consist of transgranular crack (facet) or cracks (facets) under both normal cyclic fatigue and dwell fatigue. Each facet is fitted to the α grain morphology or microstructure and the quasi-cleavage facets that are mostly formed on or near the basal plane.1–7) Their inclination to the principal stress axis revealed a wide range of variations owing to the texture of the alloys,2) although the microcrack growth of each material showed a dependence on the orientation of grain structure and principal stress axis.3–5,8) The dislocation movement in the α phase is restricted to the primary slip plane and is fairly planar so that dislocation arrays on 10 10 1120 {} are piled-up in the vicinity of the grain boundaries. The local stress concentration near the α grain boundary owing to heterogeneous slip may cause subsurface crack initiation.9,10) Furthermore, texture and microtexture (macrozone) strongly affect subsurface crack initiation and high-cycle fatigue strength.11–14) The stress redistribution between “soft” (plastically deformed and with a high Schmid factor) and “hard” (elastically deformed and with a low Schmid factor) α grains owing to strain incompatibility at their boundary is also a factor that causes the development of a facet in the hard grain under cyclic loading.10,15–19) Multiple microcracks were detected in the macrozone oriented with their main c-axis texture component. Moreover, the dominant crack grew in the macrozones owing to microcrack coalescence. A similar α -grain orientation promotes cyclic strain accumulation, because localized deformation in one grain can be straightforwardly accommodated in adjacent grains.14) Thus, the crystallographic orientations in neighboring grains and the microtexture play important roles in the generation of microcracking owing to the strain incompatibility and stress redistribution during fatigue processes.20) The microcracking on (0001) in a hard α -grain may be generated by the combination of the opening stress and shear stress components under cyclic loading. This is because the shear stress component promotes the develop- ISIJ International, Vol. 62 (2022), No. 3 © 2022 ISIJ 594 ment of the slip band on (0001). The normal direction of the (0001) α quasi-cleavage facets is inclined at 15° to 60° to the principal stress axis,1,2,5,11) because the normal stress on (0001) is exceptionally high when the facet is normal to the applied stress.10,15) However, the facet formation modes at lower stress levels differ and depend significantly on the microstructures of the alloys. The macrozones whose main c-axis texture component is almost perpendicular to the principal stress axis obstruct microcrack growth.12) The 10 10 {} α facet was detected, rather than the (0001) α facet. This was particularly so in the dwell fatigue test for highly textured alloys with <0001> perpendicular to the tensile axis.19) The prismatic crack was consistent with a high Schmid factor of prismatic slip. However, it is not evident as to which facet initiates the formation of crack.19,21) The refinement of the α -grain structure and/or grain distribution with random orientation (elimination of microtexture) can effectively lower the maximum stress concentration at a boundary. This is because a shorter slip length results in substantial improvements in fatigue strength.3,22) A Ti-6Al-4V rolled material whose α -grain structure displayed a random orientation distribution owing to a higher working ratio exhibited fatigue strength that was significantly higher than that of a forged material with a colony structure at low temperature.22,23) The nearα type Ti–Fe–O alloy which provided a characteristic structure of α phase with dispersed small amount of β phase clearly exhibited subsurface fatigue crack initiation at 77 K, where the facet was detected as a (0001) α crack.10,24,25) Cryogenic temperature fatigue could be most favorable to characterize the subsurface crack generation, because the subsurface crack initiation and the strain incompatibility prevail more at lower temperatures with higher the critical resolved shear stresses (CRSSs) of individual slip systems and their differences.26) Such heterogeneous slip deformation manner revealed that the internal stress in the α grain had accumulated normal to (0001) α by the full constraints model analysis, in which resulted in (0001) α cracking.27) Ti–Fe–O alloy rolled materials provided a microtexture (macrozone) with rather clearly divided into recovered (soft) α grain and recrystallized (hard) α grain regions. Furthermore, the microstructure of the Ti–Fe–O alloy was modified by groove-rolling and cold-swaging to refine the α -grain structure and avoid the microtexture (macrozone). A fine globular α -grain structure and a weak 10 10 {} α fiber texture were achieved. However, there was no improvement in the fatigue strength at 107 cycles.24) Although the (0001) α facet provided an origin for the subsurface crack initiation, the 10 10 {} α facets mainly covered the crack initiation site. Therefore, in the present study, the cyclic deformation structure and microcracks developed in the modified microstructure of the Ti–Fe–O alloy were characterized to discuss the subsurface crack generation. 2. Experimental 2.1. Material and Fatigue Tested Samples The fatigue tested samples of a cold-swaged and annealed nearα type Ti-0.994%Fe-0.386%O (mass%) material were examined in this study. The details of the material were described in the reference 24). The rectangular bars cut from the thick plate after hot-forged (1 273 K heating) and hotrolled (1 123 K heating) were groove-rolled (1 023 K heating, equivalent strain η =1.3) and cold-swaged ( η =0.48) into a round bar with a diameter of 20 mm. The round bar was annealed at 1 023 K for 3.6 ks followed by air cooled, and the cylindrical test pieces were cut parallel to the rolling direction (RD) direction. Fatigue tests were performed using a cryogenic servohydraulic fatigue machine under load control at a stress ratio, R of 0.01, in liquid nitrogen (77 K) and in ambient air (293 K).24) The test conditions are given in the reference 23). Table 1 lists the tensile properties of the test material. Figure 1 shows maximum cyclic stress vs. the number of cycles to failure (Nf) data of the test material.24) The samples that failed at 77 K and 293 K in higher cycles exhibited subsurface crack initiation (solid plots in Fig. 1). Herein, their maximum cyclic stress was significantly lower than 0.2% proof stress, σ 0.2. The samples that failed by subsurface crack initiation were selected for the present analyses. 2.2. Analyses The fractured samples were cut off near the subsurface crack initiation sites parallel to the direction of initial crack growth as shown in Fig. 2.21) The microstructures around microcracks beneath the crack propagation planes were analyzed in the longitudinal cross-section. The mechanically and electrochemically polished surfaces were examined by scanning electron microscopy (SEM). Electron backscatter diffraction (EBSD) pattern analysis in SEM was employed to determine the microstructure around the microcrack. The thin disk for transmission electron microscopy (TEM) analysis was sectioned from beneath the fracture surface perpendicular to the principal stress axis and was mechanically ground. Each disk diameter was measured and its maximum cyclic stress, σ max, was calibrated, because the Fig. 1. S–N data of the tested material at 77 K and 293 K. Table 1. Tensile properties of the tested material at 77 K and 293 K. Temperature (K) 0.2% proof stress, σ 0.2 (MPa) Ultimate tensile strength, σ B (MPa) Total elongation (%) 77 1 257 1 326 11.3 293 659 799 29.5 ISIJ International, Vol. 62 (2022), No. 3 © 2022 ISIJ595 Fig. 2. Schematic illustration of analysis samples cut from the failed specimens. Fig. 3. Microstructure of test material: EBSD image quality maps of the RD plane (a) and of a longitudinal section (b). Fig. 4. Dislocation structure in α and β grains of the tested material on the RD plane. Fig. 5. Dislocation structures in α grains after cyclic deformation at 293 K: (a) σ max = 549 MPa, Nf = 388 010 cycles, and (b) σ max = 597 MPa, Nf = 72 424 cycles. fatigue test specimens had an hourglass shape. TEM foils were prepared by electrochemical twinjet polishing at 243 K in a stirred solution of 6% perchloric acid, 35% butanol, and 59% methanol. A JEM-2100F electron microscope equipped with a double-tilt goniometer stage was operated at 200 keV. 2.3. Microstructure Figure 3 shows the microstructure of the material as EBSD image quality maps. The material consisted of equiaxed α grains and fine rod-like β grains, which were randomly distributed along the RD. The average α -grain size was approximately 13 μ m. Individual dislocations were distributed in both α and β grains as shown in Fig. 4, so that the equiaxed α grains could be recovered fully. The α microstructure showed a weak 10 10 {} fiber texture in which the maximum pole density of 10 10 {} was less than three.24) Furthermore, the c-axis is randomly distributed perpendicular to the principal stress axis (RD).24) 3. Results 3.1. Cyclic Deformation Structure After cyclic deformation at 293 K and 77 K, dislocation arrays were developed in the α grains at each stress level as shown in Figs. 5(a) and 6(a). Arrays were blocked at the grain boundaries or the α - β interfaces as indicated by an arrow in Fig. 6(a). During the cyclic deformation, preexisting moveable dislocations rearranged themselves and multiplied in the α grain. Certain dislocation loops were observed on the prismatic plane at low temperatures (Fig. 6(b)). When a dislocation source emits a series of dislocations all lying on the same slip plane, dislocations pile up behind the leading dislocation and interact elastically. The dominant deformation mode is via the prismatic slip system, and the glide process is markedly planar. Nearly all the dislocations in coplanar arrays observed in the α grains (Figs. ISIJ International, Vol. 62 (2022), No. 3 © 2022 ISIJ 596 5(a) and 6(a)) were of the screw type as discussed in the references 9) and 10), and the movement was restricted to their slip planes. Dislocations can pass through the boundaries and into the neighboring grains at higher stress levels. Tangled dislocations were also detected in α grains at a higher stress level at both test temperatures (see Figs. 5(b) and 6(c)). In addition, many dislocations with multiple slip systems were observed in the β grains at both test temperatures at any stress level (see Fig. 7). 3.2. Microcracks in ββ Grains Regardless of the stress level, a considerable number of microcracks and voids were detected near the fracture surface on the cross-section of the failed specimens at both test temperatures (see Fig. 8). Most of the microcracks were detected in the β grains within a depth of 100 μ m from the fracture surface. All the detected β microcracks were toughened at the α - β interface. A few of voids were also detected in the α grains and at the α - β interfaces. Figure 9 shows a microcrack generated and grown at the β grain boundary where the misorientation between the β 1 and β 2 grains was approximately 42° (Figs. 9(b) and 9(d)). The tips of the microcrack are at the triple points of α 1β 1β 2 and α 2β 1β 2, and the localized plastic deformation is generated in neighboring α 1 and α 2 grains (Fig. 9(c)). The c-axis of each α grain is nearly perpendicular to the principal stress axis (Fig. 9(b)). The Schmid factors in the individual α grains are high as M=0.46 (in α 1) and M=0.48 (in α 2) for the prismatic slip system (Table 2) and M=0.07 and 0.05, respectively, for the basal slip system. 3.3. Growth of Microcracks into αα Grains Although most of the microcracks in the β grains were arrested in the grains, a few of the microcracks were observed to have grown into the neighboring α grains. Three microcracks (labeled as A, B and C in Fig. 10(a)) were formed in individual β grains. Microcrack A grew into the neighboring grains α 3 and α 4 (Fig. 10(b)). According to the crystallographic analysis, the microcrack grew into the α 3 grain along its prismatic plane, which was inclined Fig. 6. Dislocation structures in α grains after cyclic deformation at 77 K with σ max = 978 MPa, Nf = 13 480 cycles. Fig. 7. Dislocations in β grain of CS material after cyclic deformation: (a) 293 K, σ max = 513 MPa, Nf = 2 ×107 cycles, not broken, and (b) 77 K, σ max = 1 038 MPa, Nf = 6 590 cycles. ISIJ International, Vol. 62 (2022), No. 3 © 2022 ISIJ597 Fig. 8. Microcracks (indicated by arrows) in the longitudinal section beneath the fracture surface: (a) SEI (77 K, σ max = 995 MPa, Nf = 8 455 cycles), (b), (c) magnified images indicated in (a), and (d) SEI (77 K, σ max = 1 161 MPa, Nf = 4 060 cycles). Fig. 9. Microcrack (indicated by an arrow) formed at the β -grain boundary in the longitudinal section beneath the fracture surface (77 K, σ max = 995 MPa, Nf = 8 455 cycles): (a) SEI, (b) IPF orientation map of (a), (c) KAM map of the area indicated in (b), and (d) misorientation profile of line AB. (Online version in color.) to the principal stress axis at approximately 29° (see Table 3). Both the shear and tensile stresses on the crack plane were relatively high. Microcrack A also grew along the 10 10 {} twist grain boundary between the grains α 4 and α 9, which were almost perpendicular to the principal stress axis. The grain α 5 showed a crystal orientation that was almost identical to that of grain α 4. However, microcrack B did not propagate into the neighboring grains α 5 and α 6. Deformation twinning is suppressed as oxygen content increases. However, 1122 1123 {} compressive twins and ISIJ International, Vol. 62 (2022), No. 3 © 2022 ISIJ 598 10 12 10 11 {} tensile twins were detected in the grains α 6, α 8, and α 9 near the microcrack tips, as shown in Fig. 10(b). 3.4. Localized Slips and Voids at αα -grain Boundaries Figure 11 shows a void at the α -grain boundary. The grain α 10 shown in Fig. 11(b) was oriented with a higher prismatic Schmid factor (0.49). Furthermore, several prismatic slip bands developed there. The grain α 11 was oriented with prismatic and basal Schmid factors of 0.42 and 0.25, respectively. The stress concentration at the grain boundary owing to the planar slip bands in α 10 could have generated the void. Although no microcracks were generated at the void, a localized prismatic slip band was emitted from the void in α 11. Because the triaxial stress concentration fields were developed around the voids, the localized slip bands could have been developed by shear stress components. However, no slip-off of the bands was detected under any stress level. Fig. 10. Microcrack (indicated by an arrow) that was formed at the β grain boundary and its growth into neighboring α grains in the longitudinal section beneath the fracture surface (77 K, σ max = 995 MPa, Nf = 8 455 cycles): (a) SEI, (b) IPF orientation map of (a). (Online version in color.) Fig. 11. Void formed at the α -grain boundary in the longitudinal section beneath the fracture surface (293 K, σ max = 539 MPa, Nf = 1 166 619 cycles): (a) SEI and (b) IQ map. (Online version in color.) Table 2. Schmid factors for prismatic slips of the α 1 and α 2 grains neighboring the microcrack between β grains shown in Fig. 9 ( θ : inclination of plane normal to principal stress axis). α grain Plane Slip system Schmid factor, M θ (°) α 1 P1 a 3 01 10 2110 ()   0.08 6.8 P2 a 3 10 10 12 10 ()   0.46 55.4 P3 a 3 1100 1 120 ()   0.38 74.8 α 2 P4 a 3 01 10 2110 ()   0.12 7.7 P5 a 3 10 10 12 10 ()   0.48 52.8 P6 a 3 1100 1 120 ()   0.36 67.3 Table 3. Schmid factors for prismatic slip of the grain α 3 shown in Fig. 10(b). Plane Slip system Schmid factor, M θ (°) P7 a 3 01 10 2110 ()   0.38 29.2 P8 a 3 10 10 12 10 ()   0.44 35.9 P9 a 3 1100 1 120 ()   0.06 86.4 ISIJ International, Vol. 62 (2022), No. 3 © 2022 ISIJ599 4. Discussion 4.1. Deformation Modes The dominant deformation mode in α titanium is the 10 10 1120 {} slip operation (particularly at lower temperatures), and the glide process is markedly planar. The dislocation movements are restricted to their slip planes, and cross-slip from a prism plane onto another is difficult. Heterogeneous microplasticity owing to planar slip and restricted systems causes strain incompatibility at a boundary. A process of stress distribution between weak and strong grains owing to strain incompatibility may be the cause of the generation of (0001) microcracks in titanium alloys under fatigue loading. Co-planar arrays in α grains may pass through the boundary and into the neighboring grain or be blocked at the grain boundary where strain incompatibility is developed. Because the material exhibits 10 10 {} fiber texture, a shear mode is likely to be developed on both (0001) and 10 10 {} planes by strain incompatibility. The relaxation under the simple shear mode is more straightforward than that under the tension mode. Furthermore, the localized slip on those planes under the simple shear mode may assist the growth of microcracks. Therefore, the refinement of the α -grain structure should be accompanied by a random distribution of grain orientation for an improvement in the fatigue strength of α and α - β type titanium alloys in which subsurface crack initiation is dominant. Although fewer deformation twins were detected in the α grains, 1122 1123 {} in compression and 10 12 10 11 {} in tension were operated near the microcrack tips (Fig. 10(b)). In general, twinning occurs in polycrystalline α titanium and its alloys because of an inadequate number of slip systems to accommodate an arbitrary plastic strain. Four twinning modes have been reported in α titanium. Here, 10 12 10 11 {} and 1121 1 126 {} are extension twins, and the 1122 1123 {} and 10 11 10 12 {} are contraction twins.28,29) The predominant twinning mode at room temperature is the 10 12 {} tensile twin, which corresponds to a rotation of 85° around the axis. At 77 K, the amount of 10 12 {} twinning increased considerably, and the 1121 {} tensile and 1122 {} compressive twins were also observed.30) Because the 1121 {} twin can form in regions completely constrained by surrounding grains owing to a shear stress higher than a critical value,31) the 1122 1123 {} and 10 12 10 11 {} operations were detected in this experiment. These operations relaxed the internal stress normal to the (0001). 4.2. Microcracks in ββ Grains Less attention has been paid to the role of the β phase in subsurface fatigue crack initiation. Ruppen et al.4) proposed that the piled-up dislocations in α -grain-induced slip in the neighboring β phase and the formation of a Cottrell cleavage knife {001} in the β phase provided a site for subsurface crack initiation. However, no experimental evidence of cleavage cracking has been reported. A qualitative micromechanical model was also proposed in which microcracks were generated in a specimen’s interior by the coalescence of shear-induced cavities formed at the α - β interfaces along localized slip bands.32) However, cavity nucleation was detected in a lamellar structure with thin β laths under high applied stress. Void (or microcrack) nucleation in β grains was detected at the α - β interfaces. It developed strain incompatibility between a recovered α (soft) grain and a recrystallized α (hard) grain in the cross-rolled Ti–Fe–O alloy after cyclic deformation.25) The strain incompatibility between recrystallized α grains and β grains could have induced microcrack (or void) nucleation, and the rearrangement of dislocations in the neighboring recovered α -grains enhanced the strain incompatibility. In the present study, a considerable number of microcracks in β grains were detected in highly strained regions such as near the fracture surface. This is notwithstanding that few combinations of softα -grainβ -grain-hardα -grain were formed in the material because of the fiber texture. The microcrack generated at the triple point of α - β 1β 2 could have propagated along the β grain boundary, as shown in Fig. 8(d). The density of dislocations in the β grains as shown in Fig. 7 was higher than that in the α grains.33) Because the thickness of β grains was larger than that of the cross-rolled material,25) the longer mean free slip length in the β phase could have induced a higher local stress concentration at the α - β interface, particularly at the triple point of α - β 1β 2 (see Fig. 9). This could have caused the microvoid generation at the triple point. Thereby, the opening stress on the β grain boundary that is nearly perpendicular to the loading direction could have assisted the microcrack growth from the micorvoid, as illustrated in Fig. 12. Figure 13 shows the orientations of the individual β grains in which voids were detected. The distributed β grains deviated from <111> because of the high Schmid Fig. 12. Schematic illustration of microcrack growth at β -grain boundary from the microvoid at triple point. Fig. 13. Orientation distribution of individual voids detected in β grains. (Online version in color.) ISIJ International, Vol. 62 (2022), No. 3 © 2022 ISIJ 600 factors of the {011}<111> slip. Most of the microcracks formed at the β grain boundaries revealed a misorientation of 40°–60°, as shown in Fig. 13. Furthermore, Young’s modulus of β -Ti alloys is lower than that of α -Ti (e.g. approximately 106 GPa for commercially pure titanium), although Young’s modulus depends on the alloy composition.34) At the elastic to plastic deformation stage, the prismatic slip systems in α phase were activated preferentially rather than slip systems in β phase owing to lower CRSS.35) The stress in the α phase then releases and transfers to the β phase. It also supports the local yielding in the β phase in the low plastic-strain regime. 4.3. InfluenceofTextureonMicrocrackGeneration The stress concentration by the microcracks developed at the β -grain boundary may trigger (0001) microcracking in the neighboring α grain. Although (0001) α microcracks constituted the main crack in the cross-rolled material,25) few (0001) α microcracks were detected in the present material. The subsurface crack initiation sites (particularly at 77 K) consisted mostly of 10 10 {} α facets and were rather devoid of (0001) α facets.24) The α microstructure in the present material showed a weak 10 10 {} fiber texture, and the c-axis of the α grains were randomly distributed and were nearly perpendicular to the principal stress axis. Therefore, both the shear stress and normal stress on the basal plane were low, and negligible (0001) α microcracking could be generated in the neighboring grain. Although a (0001) α microcrack provides a stress field with multiaxial modes in the neighboring grain, the combination of shear stress and opening stress on 10 10 {} α may also result in Mode II or III microcracking. The localized shear stress as well as the opening (Mode I) stress owing to the principal stress is likely to have caused the microcrack growth on the prismatic plane in the neighboring grain. In addition, Bantounas et al.12) observed that α grains whose main c-axis texture component was almost perpendicular to the loading direction functioned as barriers to faceted crack growth. Then, 10 10 {} α facets may appear,24) and the modified microstructure of small α -grains with c-axis randomly distributed normal to the loading axis prevents the growth of (0001) α facet. Both surface and subsurface cracks formed along a prismatic plane in α grains oriented with a very high Schmid factor have been reported to function as initiation sites.5,21,36) The microcrack formed along a prismatic plane was in an α grain and propagated negligibly to the neighboring grain.21) That is, the prismatic crack formation is consistent with the surface roughening mechanism rather than the facet formation in the specimen interior. Sackett et al.19) observed that prismatic cracks in the interior of a specimen were almost perpendicular to the tensile loading direction. This can be explained by the stress redistribution models such as the Evans-Bache model.19) In the previous study,24) the models forming 10 10 {} α facet or facets were proposed, where the (0001) α facet may give a trigger to generate the subsurface crack initiation site with 10 10 {} α facets. In the present work, the microcracks in the β grains propagated negligibly into their neighboring α grains, although they displayed a high Schmid factor (see the grains α 1 and α 2 in Fig. 9 and Table 2). In addition, no microcracks were detected on the prismatic plane that was nearly perpendicular to the loading axis. However, microcracks were detected at the 10 10 {} α twist boundary and on the prismatic planes inclined to the loading axis, as shown in Fig. 10. Considering the redistributed shear stress caused by the stress concentration around the β -grain microcrack and the strain incompatibility at the Fig. 14. Schematic illustration of subsurface fatigue crack initiation: (a) a model based on the subcrack observation and (b) a modified model at 77 K in ref. 24). (Online version in color.) ISIJ International, Vol. 62 (2022), No. 3 © 2022 ISIJ601 α -grain boundary, the local prismatic slip in the neighboring grain should be selected on the prismatic plane with a high Schmid factor. However, the intense prismatic slip for slipoff in the neighboring grain may have caused the microcrack growth. Therefore, a combination of high tensile stress and shear stress on the prismatic plane is necessary for β -grain microcrack propagation. Thus, a simple explanation of the 10 10 {} α facet formation model is proposed, as shown in Fig. 14(a). When the inclined 10 10 {} α microcrack grows in the α 1 grain to its boundary, the microcrack may propagate into the neighboring α 2 grain along the prismatic plane because of the cross slip. The combination of shear stress (Mode II or III) and opening stress (Mode I) on 10 10 {} α may result in microcracking on the prismatic plane, which is nearly perpendicular to the loading axis. Because deformation constraints in the α grains on the specimen surface are more flexible than the internal ones, strain incompatibility may be introduced near the specimen surface. In our previous work, we discussed a model of subsurface crack initiation based on the fractographical analysis, where the role of the β phase was omitted.23) In the model, the screw dislocations of 10 10 1120 {} pile-ups in an α -grain near the specimen surface introduce the microcracking by (0001) α slip-off, as shown in Fig. 14(b). Herein, the β microcrack may assist in the formation of the shear stress on (0001) α . Because the (0001) α microcrack also provides a stress field with multiaxial modes in the neighboring grain, the combination of shear stress and opening stress on 10 10 {} α result in Mode II or III microcracking. 5. Conclusions The characteristics of subcracks in Ti–Fe–O alloy formed by high-cycle fatigue at low temperature were investigated. The phenomenological details of microcrack generation, particularly the trigger and beginning of the microcrack, are clarified as follows: (1) Most of the microcracks were in β grains and a few propagated into α grains along the prismatic plane. A few voids were present in the α grains and at the α - β interfaces. (2) The stress concentration around the microcracks in the β grains assists microcracking on the 10 10 {} α facet. (3) The combination of shear stress and opening stress on 10 10 {} α results in Mode II or III microcracking on the prismatic plane, which is perpendicular to the loading axis. Furthermore, the 10 10 {} α texture significantly assists microcrack growth on the 10 10 {} α plane. Acknowledgment This paper was supported by project No. CZ.02.1.01/0.0/ 0.0/17_049/0008441, “Innovative Therapeutic Methods of Musculoskeletal System in Accident Surgery” within the Operational Programme Research, Development and Education financed by the European Union and by the state budget of the Czech Republic. REFERENCES 1) O. Umezawa and K. Nagai: ISIJ Int., 37 (1997), 1170. https://doi. org/10.2355/isijinternational.37.1170 2) W. Li and O. Umezawa: Key Eng. Mater., 741 (2017), 76. https://doi. org/10.4028/www.scientific.net/KEM.741.76 3) A. L. Pilchak, R. E. A. Williams and J. C. Williams: Metall. Mater. Trans. A, 41 (2010), 106. https://doi.org/10.1007/s11661-009-0064-2 4) J. Ruppen, P. Bhowal, D. Eylon and A. J. McEvily: Fatigue Mechanisms, ASTM STP 675, ed. by J. Fong, American Society for Testing and Materials, Philadelphia, (1979), 47. https://doi.org/10.1520/ STP35884S 5) I. Bantounas, D. Dye and T. C. Lindley: Acta Mater., 57 (2009), 3584. https://doi.org/10.1016/j.actamat.2009.04.018 6) M. R. Bache, M. Cope, H. M. Davies, W. J. Evans and G. Harrison: Int. J. Fatigue, 19 (1997), Suppl. 1, S83. https://doi.org/10.1016/ S0142-1123(97)00020-0 7) V. Sinha, M. J. Mills and J. C. Williams: Metall. Mater. Trans. A, 37 (2006), 2015. https://doi.org/10.1007/s11661-006-0144-5 8) Y. Mine, S. Katashima, R. Ding, P. Bowen and K. Takashima: Scr. Mater., 165 (2019), 107. https://doi.org/10.1016/j.scriptamat.2019.02.029 9) O. Umezawa, K. Nagai and K. Ishikawa: Mater. Sci. Eng. A, 129 (1990), 223. https://doi.org/10.1016/0921-5093(90)90269-9 10) H. Yokoyama, O. Umezawa, K. Nagai, T. Suzuki and K. Kokubo: Metall. Mater. Trans. A, 31 (2000), 2793. https://doi.org/10.1007/ BF02830339 11) I. Bantounas, D. Dye and T. C. Lindley: Acta Mater., 58 (2010), 3908. https://doi.org/10.1016/j.actamat.2010.03.036 12) I. Bantounas, T. C. Lindley, D. Rugg and D. Dye: Acta Mater., 55 (2007), 5655. https://doi.org/10.1016/j.actamat.2007.06.034 13) K. Le Biavant, S. Pommier and C. Prioul: Fatigue Fract. Eng. Mater. Struct., 25 (2002), 527. https://doi.org/10.1046/j.14602695.2002.00480.x 14) C. J. Szczepanski, S. K. Jha, J. M. Larsen and J. W. Jones: Metall. Mater. Trans. A, 39 (2008), 2841. https://doi.org/10.1007/s11661008-9633-z 15) M. R. Bache, W. J. Evans and H. M. Davies: J. Mater. Sci., 32 (1997), 3435. https://doi.org/10.1023/A:1018624801310 16) M. R. Bache: Int. J. Fatigue, 25 (2003), 1079. https://doi.org/10.1016/ S0142-1123(03)00145-2 17) A. N. Stroh: Proc. R. Soc. Lond. A, 223 (1954), 404. https://doi. org/10.1098/rspa.1954.0124 18) M. R. Bache, W. J. Evans, B. Suddell and F. R. M. Herrouin: Int. J. Fatigue, 23 (2001), Suppl. 1, 153. https://doi.org/10.1016/S01421123(01)00124-4 19) E. E. Sackett, L. Germain and M. R. Bache: Int. J. Fatigue, 29 (2007), 2015. https://doi.org/10.1016/j.ijfatigue.2006.12.011 20) W. J. Evans and M. R. Bache: Int. J. Fatigue, 16 (1994), 443. https:// doi.org/10.1016/0142-1123(94)90194-5 21) F. Bridier, P. Villechaise and J. Mendez: Acta Mater., 56 (2008), 3951. https://doi.org/10.1016/j.actamat.2008.04.036 22) O. Umezawa, K. Nagai and K. Ishikawa: Tetsu-to-Hagané, 76 (1990), 924 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.76.6_924 23) O. Umezawa and K. Ishikawa: Mater. Sci. Eng. A, 176 (1994), 397. https://doi.org/10.1016/0011-2275(92)90353-C 24) O. Umezawa, T. Yuasa and W. Li: ISIJ Int., 58 (2018), 1332. https:// doi.org/10.2355/isijinternational.ISIJINT-2017-673 25) W. Li, O. Umezawa and N. Koga: ISIJ Int., 58 (2018), 359. https:// doi.org/10.2355/isijinternational.ISIJINT-2017-514 26) K. Kishida, J. G. Kim, T. Nagae and H. Inui: Acta Mater., 196 (2020), 168. https://doi.org/10.1016/j.actamat.2020.06.043 27) M. Morita and O. Umezawa: Mater. Trans., 52 (2011), 1595. https:// doi.org/10.2320/matertrans.L-M2011815 28) M. H. Yoo: Metall. Trans. A, 12 (1981), 409. https://doi.org/10.1007/ BF02648537 29) L. Wang, Y. Yang, P. Eisenlohr, T. R. Bieler, M. A. Crimp and D. E. Mason: Metall. Mater. Trans. A, 41 (2010), 421. https://doi. org/10.1007/s11661-009-0097-6 30) S. Mahajan: Encyclopedia of Materials: Science and Technology (2nd ed.), ed. by K. H. J. Buschow et al., Elsevier, Amsterdam, (2002), 1. https://doi.org/10.1016/b0-08-043152-6/01809-x 31) C. J. Beevers and M. D. Halliday: Met. Sci. J., 3 (1969), 74. https:// doi.org/10.1179/msc.1969.3.1.74 32) P. Lefranc, V. Doquet, M. Gerland and C. Sarrazin-Baudoux: Acta Mater., 56 (2008), 4450. https://doi.org/10.1016/j.actamat.2008.04.060 33) W. Li and O. Umezawa: Proc. 13th World Conf. on Titanium, ed. by V. Venkatesh et al., TMS, Warrendale, PA, (2016), 843. https://doi. org/10.1002/9781119296126.ch144 34) T. Ozaki, H. Matsumoto, S. Watanabe and S. Hanada: Mater. Trans., 45 (2004), 2776. https://doi.org/10.2320/matertrans.45.2776 35) R. Shi, Z. Nie, Q. Fan and G. Li: J. Alloy. Compd., 688 (2016), 787. https://doi.org/10.1016/j.jallcom.2016.07.105 36) J. Everaerts, D. Gontcharov, B. Verlinden and M. Wevers: Int. J. Fatigue, 98 (2017), 203. https://doi.org/10.1016/j.ijfatigue.2017.01.043