2.1 Material

The Ti-6554 alloy used in this study was a 145 mm diameter hot forged bar supplied by Baoti Group Co., Ltd. with the following composition: 5.01Cr, 4.72Mo, 4.92V, 4.73Al, balance Ti (wt%). The β-transus temperature (Tα+β→β) was identified to be 810 ℃ by the metallographic observations in our previous work [37]. The specific experimental process is depicted in Figure S1 (Supplementary material).

A high-throughput gradient heat treatment method was employed to rapidly obtain the solution microstructure evolution laws [35, 36]. As illustrated in Figure 1(a), a tubular furnace was used to establish the temperature gradient distribution. A round bar of Ti-6554 alloy with a diameter of 10 mm was processed by wire-cutting, and holes with a diameter of 1 mm and a depth of 8 mm were drilled on it every 10 mm using EDM. K-type thermocouples were fixed in the holes using high-temperature conductive adhesive, and the temperature was recorded in real time using a YP5016G multi-channel temperature tester. Figure 1(b) depicts the schematic diagram of the heat treatment of a single sample at different locations, including solution annealing at 757, 774, 792, 810, 825, 840 and 857 ℃ for 2 h, followed by air cooling (AC) and subsequent aging at 500 ℃ for 4 h.

Figure 1全屏查看

(a) Temperature gradient distribution of tubular furnace and (b) schematic of heat treatment

2.2 Mechanical testing

Tensile, cyclic loading and unloading, and FCP tests were performed on Ti-6554 alloy using the MTS Landmark testing machine. The tensile performance test was conducted following the ISO 6892-1: 2019 standard, with a loading rate of 1 mm/min. The standard sample dimensions are shown in Figure 2(a). The cyclic loading and unloading test sample size was the same as the tensile sample. The loading strain rate was 1×10^-3 s^-1 and then unloaded to 50 N at a rate of 1000 N/min. The dimensions of the compact tension (CT) samples were 40 mm × 38.4 mm × 3 mm (Figure 2(b)). Prior to the FCP experiment, both surfaces of the samples were polished to eliminate the influence of surface roughness. A 2 mm pre-crack was formed on the specimens based on the ISO 12108:2018 standard. The FCP specimens were tested under sinusoidal stress control with a frequency of 10 Hz and a sinusoidal load ratio (R) of 0.1.

Figure 2全屏查看

Test sample geometry (unit: mm): (a) Tensile, cyclic loading and unloading sample; (b) FCP sample

2.3 Microstructural characterization

The microstructure of the sample was characterized using optical microscope (OM, Leica DM ILM), backscattered electron mode scanning electron microscope (SEM, JSM-7900F), electron backscattered diffraction (EBSD, Regulus8230), and transmission electron microscope (TEM, Tecnai G2F20, Talos F200X). Prior to the EBSD examination, the samples were mechanically ground followed by electrolytic polishing in a solution of 90% CH₃COOH and 10% HClO₄ (vol%). The electrolytic polishing voltage was set at 30 V, and the temperature was maintained at -30 ℃. The step size was 5 μm and the operating voltage was 20 kV. The EBSD data were post-processed using the Atex software [38]. The TEM sample was thinned to a thickness of 70 μm and then cut into 3 mm diameter circular discs. These discs were further thinned by automatic dual-jet electrolytic polishing in a solution of 5% HClO₄, 35% C₄H₉OH, and 60% CH₃OH at 20 V and -30 ℃.

3.1 Soluted and aged microstructure

Figure 3 represents the microstructure of the Ti-6554 alloy after undergoing various solution and aging treatments. As shown in Figures 3(a₁)-(c₁), the volume fraction of α_P gradually decreases from 9% at 757 ℃ to 2% at 792 ℃ with increasing solution temperature. Furthermore, the bright regions in Figures 3(b₁) and (c₁) indicate partial SRX grains in the solution microstructure at 774 and 792 ℃, respectively. When the solution temperature attains the β-transus temperature of 810 ℃, all α phases transform into the β matrix (Figures 3(d₁)-(g₁)). According to the statistical data in Table 1, the β grain size increases with increasing solution temperature.

Figure 3全屏查看

SEM and OM microstructures of the Ti-6554 alloy after (a₁-g₁) gradient solution treatment for 2 h and (a₂-g₂) aging at 500 ℃ for 4 h

The microstructures after aging at 500 ℃ for 4 h are depicted in Figures 3(a₂)-(g₂). It is evident that a substantial amount of ultra-fine α_S phases have precipitated in the microstructures of the 757 solution-aging (STA) and 774STA samples. Conversely, the higher solution temperature samples exhibit a noticeable increase in the length and thickness of the coarse α_S phases formed after aging (Figures 3(c₂)-(g₂)).

Based on the above observations, three typical microstructures of 757STA, 792STA and 825STA with different degrees of SRX were selected for detailed characterization and property analysis. Figure 4 presents the microstructures of the longitudinal sections of Ti-6554 bars after undergoing the selected three different heat treatments. The orientation map in Figure 4(a) shows that the microstructure of the 757STA sample contains numerous β-deformed grains in a fibrous morphology (β_f). This fibrous deformation structure is inherited from the forging microstructure. The β_f grains present slight color variations, indicating the presence of localized small misorientations. Figure 4(d) is the grain boundary map corresponding to Figure 4(a), where low-angle grain boundaries (LAGBs) and high-angle grain boundaries (HAGBs) are distinguished based on whether the misorientation between adjacent grains is 5°-15° or >15°. The presence of small misorientation variations within the β_f grains, where the fraction of LAGBs (F_LAGBs) is 61%, is further demonstrated in Figure 4(d).

Figure 4全屏查看

Orientation maps of Ti-6554 alloy samples of (a) 757STA, (b) 792STA and (c) 825STA; The grain boundary maps of (d) 757STA, (e) 792STA and (f) 825STA

In Figure 4(b), it can be seen that the 792STA sample undergoes approximately 40% partial SRX. The recrystallized β grains are present in an equiaxed shape (β_e) with an average grain size (β_e) of 98.7 μm, significantly smaller than the β_f grains. Notably, these β_e grains exhibit random crystal orientation without a specific texture. Figure 4(e) illustrates that the majority of the β_e grains are separated by HAGBs, while numerous LAGBs still remain in the β_f grains. Clearly, the obtained microstructure containing both β_e and β_f grains exhibits structural heterogeneity due to the partial SRX effect. The 825STA sample undergoes sufficient SRX (Figure 4(c)), with a β_e of 261.8 μm. In conjunction with Figure 4(f), only 18% of LAGBs remain; the rest disappear. For convenience, the three types of STA samples are referred to as ‘β-recovery samples’, ‘β-hetero samples’, and ‘β-fully SRX samples’ in the following sections.

Another important feature to consider is the geometrically necessary dislocation (GND) density of the three samples, as GND plays a paramount role in maintaining the continuity of plastic flow during material deformation [39]. A high GND density not only enhances the material strength but also improves its ductility [40]. Table 2 presents the results obtained from the EBSD data of the β-recovery, β-hetero, and β-fully SRX samples. The GND density of β_e grains (ρβ_e) is significantly lower than that of the β_f grains (ρβ_f). Figure 5 illustrates the GND distribution of the three specimens. The β-recovery samples exhibit the highest dislocation density due to their low solution temperature, which effectively preserves dislocations within β_f grains (Figure 5(a)). The β-fully SRX samples demonstrate the lowest GND owing to the extensive consumption of dislocations, leading to the formation of new β_e grains (Figure 5(c)). Interestingly, as shown in Figure 5(b), the β-hetero samples comprise hard domain β_f grains with high GND density and soft domain β_e grains with low GND density.

Table 2全屏查看

Statistical microstructure features from the EBSD data of the β-recovery, β-hetero, and β-fully SRX samples, including the LAGBs fraction (F_LAGBs), volume fraction of β_f (Fβ_f) and β_e (Fβ_e) grains, average grain size of β_f (β_f) and β_e (β_e) grains, and GND density of β_f (ρβ_f) and β_e (ρβ_e) grains

Sample	FLAGBs/%	Fβf/%	Fβe/%	βf/μm	βe/μm	ρβf/m-2	ρβe/m-2
β-recovery	61	100	—	468.2	—	1.75×1013	—
β-hetero	54.8	60	40	424.9	98.7	1.77×1013	5.94×1012
β-fully SRX	18	—	100	—	261.8	—	1.87×1012

展开更多

Figure 5全屏查看

GND distribution of the Ti-6554 alloy: (a) β-recovery; (b) β-hetero; (c) β-fully SRX sample

The α_S-precipitation morphologies of the three samples were observed using TEM and the thickness distribution was calculated, as shown in Figure 6. A common characteristic of three samples is the presence of three variants of α_S, intersecting at an angle of ~60° [41]. The α_S-precipitates in the β-recovery sample (Figures 6(a) and (d)), with an average thickness of 40.1 nm, were smaller and denser than those within other samples. Interestingly, Figures 6(b) and (c) present that the α_S-precipitates in β-hetero and β-fully SRX samples exhibit significant size variations. These α_S-precipitates consist of both coarse α_S and ultra-fine α_S, with increased length and thickness. The thickness distribution plot clearly shows two peaks (Figures 6(e) and (f)), indicating that α_S thickness is characterized by an ultra-fine and coarse dual-scale. The reason for such differences is that the β_f grains are rich in dislocations, providing abundant nucleation sites that facilitate the uniform formation of the ultra-fine α_S phase. Contrastingly, GNDs within the β_e grains are consumed during the SRX process, forming larger and unevenly distributed α_S phases during the subsequent aging treatment.

Figure 6全屏查看

HAADF-STEM images of the aged Ti-6554 alloy samples of (a) β-recovery, (b) β-hetero and (c) β-fully SRX samples; the α_S thickness of (d) β-recovery, (e) β-hetero and (f) β-fully SRX samples

3.2 Tensile properties and fatigue propagation

Figure 7(a) displays the tensile engineering stress-strain curves, and Table 3 presents the typical mechanical property data. The ultimate tensile strength (σ_UTS) of the β-recovery and β-fully SRX samples is 1367 MPa and 1391 MPa, respectively, while the corresponding uniform elongation (UE) is 4.6% and 3.1%, respectively. Evidently, both samples exhibit high strength but low ductility, thus experiencing a strength-ductility trade off. For the β-hetero sample, the yield strength (σ_YS) is 1403 MPa while the UE remains at 6.5%. Excitingly, compared to the β-recovery and β-fully SRX samples, the σ_YS of the β-hetero sample increased by 6.4% and 6.7%, respectively, while the corresponding UE improved by 41.3% and 109.7%. The tensile fracture characteristics of the samples are shown in Figure S2. During deformation, the soft domain β_e enters the plastic deformation stage first, leading to a significant accumulation of GNDs owing to the difficulty encountered by GNDs in surpassing the β_f /β_e interface [42]. Consequently, the dislocation accumulation generates HDI stresses [43], resulting in an enhancement of the overall σ_YS. After yielding, GNDs in the β_e phase contribute to HDI hardening and delay necking during tensile testing, thereby improving UE [28].

Figure 7全屏查看

(a) The tensile engineering stress and strain curves, and (b) FCP curves of β-recovery, β-hetero and β-fully SRX samples

Table 3全屏查看

Tensile and FCP test data of β-recovery, β-hetero and β-fully SRX samples, including yield strength (σ_YS), ultimate tensile strength (σ_UTS), uniform elongation (UE), total elongation (TE), stress concentration factor (), and critical value of of fracture ()

Sample	σYS/MPa	σUTS/MPa	UE/%	TE/%	(da/dN)/(mm·cycle-1) at =24.9	/
β-recovery	1319	1367	4.6	6.1		30.4
β-hetero	1403	1462	6.5	7.7		36.0
β-fully SRX	1315	1391	3.1	4.3		32.4

展开更多

Figure 7(b) depicts the FCP curves for the three samples, showing varying rates of crack propagation (da/dN). In comparison to the β-recovery specimen, the FCP curves of the β-hetero and β-fully SRX specimens both exhibit a rightward shift, indicating a greater critical fast fracture threshold (). The value of the β-hetero sample is 18.4% and 11.1% greater than those of the β-recovery and β-fully SRX samples respectively, indicating an enlarged steady state propagation region. At the stress concentration factor of 24.9 , the β-hetero sample demonstrates the lowest FCP rate of mm/cycle. The successful design of such a heterogeneous structure overcomes the problem of tensile brittleness in high strength metastable β titanium alloys while maintaining excellent FCP resistance.

To quantitatively measure the HDI strengthening effect, we performed cyclic loading and unloading tests (Figure 8(a)). Calculate the HDI stress (σ_HDI) [42] according to the diagram in Figure 8(b): σ_HDI=(σ_r+σ_u)/2 where σ_r and σ_u are the stresses at which the stress-strain curves of reloading and unloading deviate from the linear relationship, respectively. As shown in Figure 8(c), the σ_HDI values of β-hetero are significantly higher than those of β-recovery and β-fully SRX samples, and the rate of increment in HDI values for the former is also slightly greater. These results suggest that the β-hetero microstructure does indeed provide extra HDI strengthening due to heterogeneous yielding.

Figure 8全屏查看

(a) Cyclic loading and unloading curves, (b) schematic of calculating HDI stress and (c) HDI stress

3.3 Fatigue crack propagation path

Figure 9(a) presents the FCP path profile of the β-recovery sample, indicating the transgranular FCP mechanism. In Figures 9(b)-(d), numerous crack deflections and branching are observed, and α_S phases are distorted in the vicinity of the crack due to large stress concentration. Additionally, voids and microcracks bridging are observed at the α_P/β interface (Figures 9(e) and (f)). WANG et al [44] proposed that microcracks propagate and bridge along the α_P/β interface, facilitating easier fracture and accelerating the FCP rate, which is consistent with our results.

Figure 9全屏查看

SEM images of the FCP path in the Ti-6554 alloy for the β-recovery sample

Figure 10(a) illustrates the tortuous FCP path of the β-fully SRX sample, exhibiting a mixed mode of intergranular and transgranular fracture. Figures 10(a)-(c) reveal abundant branching and closure of cracks, which helps hinder crack propagation. In Figure 10(d) the bridging of voids and microcracks at grain boundaries can be found, accompanied by multiple deflections. The grain boundaries are considered significant barrier that inhibits the rate of FCP [45]. The crack deflection is attributed to the presence of coarse α_S phases oriented in various directions, as evidenced in Figure 10(e). In addition, numerous secondary cracks surround the main crack, and the crack tip of αS-precipitates is deformed (Figure 10(f)). The propagation behavior of these cracks decreases the da/dN, improving the crack propagation resistance of the β-fully SRX sample despite its lowest UE value.

Figure 10全屏查看

SEM images of the FCP path in the Ti-6554 alloy for the β-fully SRX sample

Figure 11 shows the FCP profile of the β-hetero sample, where the fatigue cracks alternately pass through the β_f and β_e grains. The crack propagation path is relatively tortuous (Figures 11(b) and (c)), accompanied by the wide distribution of secondary cracks surrounding the main crack (Figure 11(f)). Similarly, Figure 11(d) displays the voids and microcrack bridging. In particular, there is a pronounced contrast in α_S-precipitation within the β_f and β_e grains (Figure 11(e)), which aligns with the statistical results presented in Figure 6. This crack propagation behavior consumes more energy, resulting in a lower da/dN and higher crack propagation resistance for the β-hetero sample.

Figure 11全屏查看

SEM images of the FCP path in the Ti-6554 alloy for the β-hetero sample

4.1 Mechanism for improving damage tolerance during FCP

To further elucidate the source of the superior FCP resistance of the β-hetero specimen, the GND distribution maps near the crack surfaces of the three samples were plotted as shown in Figure 12. The β-recovery microstructure exhibits variations in GND density, which is influenced by the uneven forging deformation (Figure 12(a)). In fact, such differences do not result in significant heterogeneity effects, as supported by the density increment of GNDs in β_f grains (Δρβ_f) of only 0.60×10¹² m^-2 presented in Table 4. In contrast, the β-fully SRX sample comprises new β_e grains that demonstrate a higher capacity for dislocation accumulation and a greater work-hardening rate. Consequently, this microstructure exhibits the highest density increment of GNDs in β_e grains (Δρβ_e) of 5.19× 10¹² m^-2, as shown in Figure12(c).

Figure 12全屏查看

GND distribution of FCP for the (a) β-recovery, (b) β-hetero, and (c) β-fully SRX samples, (d) enlarged image of the yellow box in (b), and (e) GND density distribution curve along the loading direction in (d)

In the β-hetero sample, the GND value of the hard domain β_f grain is substantially higher than that of the soft domain β_e grain before FCP (Figure 5). It should be noted that inducing the HDI strengthening effect in the β_e grain results in a gradient enhancement of GND density along the grain interior towards the grain boundary, accompanied by a rapid increase in the GND density [43]. To substantiate this, the GND density distribution (Figure 12(d)) and its curve along the loading direction were plotted (Figure 12(e)). The GND density is higher at the β_f /β_e interface, and the gradient increases from the interior to the grain boundary. The cumulative incremental variation in the GNDs density of the three microstructures was counted as shown in Table 4. The Δρβ_e of the β-hetero sample (3.53×10¹² m^–2) is higher than the density increment of GNDs in β_f grains (Δρβ_f) of 0.80×10¹² m^-2, which validates the above view. Therefore, the β-hetero sample exhibits HDI strengthening effect, and the gradient structure of high-density GNDs has a significant and beneficial effect on FCP resistance.

4.2 Deformation inducing nano-scale twins during FCP

Figure 13 depicts the TEM images of the microstructure near the fatigue crack region in the β-fully SRX specimen. A large number of dislocations accumulate at the α_S/β interface, and obvious dislocations appear in the αS phase (Figures 13(a) and (b)). Figure 13(c) corresponds to the SAED pattern taken along the β zone axis. The classical BOR is maintained between α and the surrounding β grains, i.e., (0001)α//(110)β, <>α//<111>β [46]. After traversing the β matrix, the dislocations leave behind parallelly arranged debris that form dislocation walls (Figure 13(e)). Furthermore, trace analysis indicates that the dislocation walls lie on the β plane, which suggests that {}β <111>β slip system is activated. The presence of dense dislocation tangles indicates the occurrence of sufficient dislocation activity, as shown in Figure 13(f).

Figure 13全屏查看

TEM microstructure near the fatigue crack of the β-fully SRX sample after FCP test: (a, b) Dislocation interacting with the α_S and β matrix; (c) SAED pattern along the []β zone axis; (d) Band structures in the coarse α_S phase; (e) Dislocation slip arranged along the β plane; (f) Dislocation tangled in the β matrix; (g) HADDF-STEM image of band structures in the coarse α_S phase; (h) SAED pattern of the red region in (g)

Interestingly, as shown in Figure 13(d), band structures are found within the coarse α_S phase, which is absent in the β-recovery deformation microstructure (Figure S3). To further reveal the band structure characteristics, two sets of diffraction spots for the band structure interface are clearly visible in Figure 13(h). By calibrating diffraction patterns, the zone axis relationship between the band structures and α_S is obtained as [0001]//, and the relationship of overlapping spots is () band structures//()α_S. To determine the orientation relationship between the band structures and α_S, the zone axes index and overlap spots are considered as the crystal direction index (u v w) and crystal plane (h k l) of the Mille index respectively [47]. Hence, the orientation relationship can be expressed as () [0001] band structures// () []α_S, such that the axis angle relationship is <>/32°. Therefore, the band structure is a {} type nano-scale twin formed by the approximately 32° rotation around the axis. This deformation-induced twinning has been observed in other titanium alloys. CHEN et al [48] studied the deformation behavior of the Ti-55511 alloy and demonstrated that the deformation induces nano-scale twins in the lamellar α phase. LIU et al [49] pointed out that multiple deformation nano-scale twins can pile up and accommodate incoming strain, favoring stress relief by dislocation transfer and twinning-induced plasticity effect. Therefore, the plastic deformation of the coarse α_S phase is enhanced by nano-scale twins.

Figure 14 shows the typical TEM microstructure near the fatigue crack region of the β-hetero sample. The TEM bright-field image (Figure 14(a)) reveals a significant accumulation of dislocations at the α_S/β phase interface, whereas a few dislocations exist in the α_S phase. To elucidate the dislocation types, dislocation analysis was performed on the same region under the two-beam TEM condition. Based on the invisibility criterion of g·b=0 [50, 51], the dislocation indicated by the blue arrow in Figure 14(b) is invisible in g= (Figure 14(f)). Hence, it is a type of dislocation. In Figure 14(c), yellow arrows represent dislocations that are invisible in g=, indicating that they are either or type dislocations. It is interesting to note that in the heterogeneous structure we also find the same nano-scale twins described earlier, and a common feature is that all these nano-scale twins exist only within the coarse α_S phase and not in the ultra-fine α_S phase, i.e. only within the recrystallized β_e grains. Due to the HDI strengthening effect, the coarse α_S phase in the β-hetero deformed microstructure induces more nano-scale twins compared to that in the β-fully SRX microstructure, promoting the accumulation of GNDs within the soft domain β_e grains. Hence, the soft domain is strengthened such that the twinning critical shear stresses are achieved more easily. Finally, the {}<26> twinning systems are further activated within the coarse α_S phase.

Figure 14全屏查看

TEM microstructure near the fatigue crack of the β-hetero specimen after FCP test: (a) TEM bright-field image under the [011]β zone axis; (d) SAED pattern along the [011]β zone axis; TEM bright-field and dark-field images observed under two-beam condition with diffraction vectors (b, e) g= and (c, f) g=

SCHMIDT et al [52] believed that a significant ductility loss may deteriorate the FCP resistance of the titanium alloy. In this work, the β-fully SRX sample has a UE of only 3.1%, but shows a higher FCP resistance compared to the β-recovery sample with a UE of 4.6%. According to the results in Figure 7(b), the β-hetero and the β-fully SRX samples demonstrate better FCP resistance, with the deformation-induced nanoscale twins being the common deformation mechanism. In comparison, no deformation mechanism other than dislocation slip is observed in the ultra-fine α_S phase of the β-recovery microstructure. Based on this, we can infer that there is a close correlation between the formation of the nano-scale twins and the FCP resistance. Despite the occurrence of nano-scale twins within the homogeneous structure (the β-fully SRX), the HDI strengthening effect of the β-hetero microstructure could further promote the accumulation of GNDs, accelerating the achievement of the critical shear stress and increasing the nanoscale twin density or probability, thereby obviously alleviating stress concentration and improving plastic deformation in the crack tip zone. Consequently, the unique heterogeneous structure exhibits superior FCP resistance.