Shock-Induced Mechanical Response and Microstructure Evolution of Titanium Alloys

Abstract

The application of titanium alloys in weaponry is increasingly widespread, due to their high specific strength and excellent corrosion resistance. The weapons such as armors must be subjected to intense shock loads caused by explosion and hyper-velocity collision, etc., during service. Therefore, their service performance is closely related to the shock-induced response characteristics of materials, especially the microstructural evolution during the shock pulses and its effect on the mechanical properties. This chapter introduces the research progress on the shock response of some typical titanium alloys such as Ti-6Al-4V, Ti-10V-2Fe-3Al, and Ti-3.5Al-10Mo-8V-1Fe. The effects of alloying composition (alloy type) and stress amplitude on the shock-induced mechanical response and microstructural evolution of titanium alloys are explored through soft recovery shock experiments, quasi-static reloading tests, as well as careful multi-scale microscopic analyses.

1. Introduction

Titanium alloys have the advantages of low density, high specific strength, strong corrosion resistance, etc. Presently, titanium materials have been widely used in aerospace, medical, and chemical industries [1, 2, 3]. They will undoubtedly play an important role in military fields such as warheads and lightweight armor, due to the many advantages mentioned above [4, 5].

Weapons usually need to withstand extreme loads such as explosions and high-speed collisions during service. At this time, the material often exhibits a very unique behavior, which is different from the response under static loading conditions. Strong shock loads are accompanied by high pressure and high temperature and will cause the material to deform rapidly in a nearly adiabatic state [6]. On the other hand, different from the assumption in statics that the applied force reaches equilibrium in the material instantaneously, the impact load generally propagates in the material in the form of disturbance (wave), and the material state (including microstructures) changes are closely related to the wave propagation process [7]. Therefore, the dynamic behavior of titanium alloys has been the focus of researchers.

As an important topic in the research field of dynamic response of materials, the shock-induced microstructural evolution and its effect on the mechanical behavior of metals have received continuous attention of scholars for nearly 70 years. Although the duration of the shock pulse is short (in general 0.5–2 μs), most of the microstructural evolution processes have enough time to occur and reach an equilibrium state during this period [8]. Therefore, when the shock waves pass through the metals, high-density defects, such as dislocations, twins, will be formed, and phase transformations may also occur [7, 9, 10, 11, 12, 13]. The defects and new phases usually increase the strength of materials to some extent, while reducing their plasticity, that is, the so-called shock-induced strengthening effect. However, the strengthening behavior of metallic materials with different crystal structures (such as face-centered cubic, FCC, body-centered cubic, BCC, or hexagonal close-packed, HCP) is very different, and some metals (such as A-70 Ti) even show the phenomenon of “shock softening” under equivalent strain conditions [9, 10]. This suggests that, in addition to defect density, changes in type, morphology, and distribution of defects caused by differences in crystal structure will also have an important influence on the mechanical properties of metallic materials after shock wave loading.

The generation, movement, and interaction of dislocations in metals during shock pulses are affected by many external and intrinsic factors, among which the most important external factor is pressure (shock stress amplitude). The dislocation density increases with increasing shock pressure, which has been recognized by many researchers [7]. However, the dislocation density cannot increase infinitely, but tends to saturate when the pressure exceeds a certain value. Correspondingly, the shock-induced strengthening effect will also be saturated [11]. Another external factor is the shock pulse duration ($t_P$). In principle, the longer the pulse duration, the closer the shock-induced substructure is to an equilibrium state. However, the time required for dislocation initiation and interaction is in the sub-microsecond order. Therefore, for most metals, if $t_P$ > 1 μs, the evolution of the dislocation substructure has sufficient time to reach equilibrium and does not change significantly as the shock loading time increases [12, 13].

The main intrinsic factors that influence the shock-induced microstructure evolution of metals are stacking fault energy (SFE) and Peierls stress (PS) [14, 15]. The Peierls stress (also called Peierls–Nabarro stress) is defined as the minimum shear stress required to move a single dislocation in a perfect crystal at zero temperature, which can be considered as the intrinsic lattice resistance to the dislocation motion [16, 17, 18]. The higher the Peierls stress of a metallic material, the less mobility of dislocations in its crystal, which leads to the poor deformability and strain hardening effect [17]. For FCC metals with high SFE (>60 mJ/m²) and low PS (such as Ni (SFE=128 mJ/m² [14])), dislocations are easy to multiply and cross-slip to form mature dislocation cells during shock loading process [13, 14]. Therefore, the shock-induced strengthening effect of this type of metals is very obvious. FCC metals with low SFE (<40 mJ/m²), such as 304 stainless steel (SFE=21 mJ/m² [14, 15]), are more prone to the formation of planar slip, stacking faults, and {111} twinning [19]. On the other hand, although there are more slip systems in BCC metals, their PS is larger than that of FCC metals, so dislocations in BCC metals (such as Ta) proliferate slowly and the mobility is poor. Hence, long-and-straight screw dislocations are often formed and entangled in BCC metals under shock loading conditions. Cell-like structures, dislocation loops, or twins may also be formed [9]. In general, the dislocation density in BCC metals is relatively low after shock loading, which is the main reason for their insignificant strengthening effect.

Compared with FCC and BCC metals, the shock-induced microstructure evolution of HCP metals, such as titanium, magnesium, and their alloys, has not been extensively studied. Koul and Breedis [20] observed a large number of twins in 7 GPa shocked pure Ti and found that there are high-density dislocations in both the twins and the matrix, and the arrangement feature of dislocations is between that of FCC-structured Ni and BCC-structured Fe. Cerreta et al. [10] observed a similar substructure morphology in 11 GPa shocked high-purity Ti. While in the A-70 Ti shocked at the same pressure, planar slip and a small amount of fine twins are the main features of the substructure.

After solution in the β phase region and following rapid quenching, the microstructure of some β titanium alloys consists wholly of equiaxed β grains with BCC structure. If the stability of the β phase is high, the shock-induced substructure is also dominated by dislocations and fine twins [20, 21]. However, if the β phase is in a metastable state, stress/strain-induced phase transformation (SIM) will be activated by shock waves [20].

In addition to the SIM phase transition, titanium and its alloys may also undergo x$\leftrightarrow$ω phase transition when the shock wave amplitude is high enough [10, 22, 23, 24]. Where x represents α, α′, β or their mixture, and “$\leftrightarrow$” indicates that the phase transition process is reversible. The shock-induced x$\leftrightarrow$ω phase transition has been observed in pure Ti [10] and some titanium alloys such as Ti-64, VT-14, VT-20, and VT-23 [22, 23]. The stress threshold for the x$\leftrightarrow$ω phase transition ranges from about 4 GPa to above 35 GPa, which depends on the stress state and the type and content of alloying and interstitial elements [10, 22, 25, 26, 27]. Since the ω phase is a brittle phase, dislocations cannot move in it, and this phase mostly exists in the form of dispersed fine particles in the alloy [28, 29], so the x→ω phase transformation may improve the strength of titanium alloys [10].

In this chapter, the research progress on the shock response of some titanium alloys such as extra-low interstitial grade Ti-6Al-4V, Ti-10V-2Fe-3Al and Ti-3.5Al-10Mo-8V-1Fe is presented. The effects of alloying component (alloy type) and stress amplitude on the shock-induced mechanical response and microstructural evolution of titanium alloys are explored through soft recovery shock experiments, quasi-static reloading tests, as well as careful multi-scale microscopic analyses. The research on dynamic behavior of titanium alloy will lay a theoretical foundation for their application in extreme service environment.

2. Materials

In order to explore the influence of content, stability, morphology, and distribution of α and β phase on the high-pressure shock response characteristics of titanium alloys, shock loading experiments of three titanium alloys, one α+β alloy, the famous Ti-6Al-4V (extra-low interstitial grade, Ti-64 ELI [30]), and two metastable β alloys, Ti-10V-2Fe-3Al (Ti-1023) and Ti-3.5Al-10Mo-8V-1Fe (TB3) were carried out. The chemical compositions of three titanium alloys are listed in Table 1.

Alloys	Alloying elements				Interstitial elements
	Al	V	Fe	Mo	C	H	O	N
Ti-64 ELI	6.16	3.90	—	—	0.011	0.0022	0.057	0.007
Ti-1023	3.07	10.15	1.68	—	0.007	0.0024	0.075	0.014
TB3	3.62	8.59	1.13	10.13	0.011	0.0028	0.097	0.011

Table 1.

Chemical components of three titanium alloys used (mass fraction, wt.%).

The Ti-64 ELI was tested as received. Ti-1023 and TB3 were solution treated above β-transus temperature ($T_{\beta }$) and then water quenched to room temperature before shock loading. $T_{\beta }$ and heat treatment processes of titanium alloys are listed in Table 2.

Alloys	Molybdenum equivalent/wt.%	$T_{\beta }$/K	Heat treatment processes
Ti-64 ELI	2.8	1259	As received
Ti-1023	10.6	1087	1123 K/2 h/Water quenched
TB3	18.5	1028	1103 K/25 min/Water quenched

Table 2.

Beta-transus temperature and heat treatment processes for three titanium alloys.

Figure 1 shows the microstructures of three titanium alloys after heat treatments. Ti-64 ELI comprises of α laths (gray area) precipitated from the β matrix (black area), which is called lamellar microstructure, as shown in Figure 1a. There is a certain amount of intergranular α phase at the β grain boundaries, indicated by arrows in Figure 1a. The average thickness of the α laths is about 2.4 μm. The TEM image of the as-received Ti-64 ELI (Figure 1b) shows a low dislocation density before testing. Most of dislocations are short and arranged arbitrarily. The selected area electron diffraction (SAED) pattern (inset in Figure 1b) of α phase proves that there is no precipitation of other phases.

Figure 1.

Microstructures of three titanium alloys before shock loading: (a) and (b) Ti-64 ELI, (c) and (d) Ti-1023, and (e) and (f) TB3. (a), (c), and (e) are optical micrographs. (b), (d) and (f) are TEM images.

Compared with Ti-64 ELI, the two metastable β alloys Ti-1023 and TB3 have higher molybdenum equivalent ([Mo]_eq), which means the β phase is more stable. After solution treatment, a large amount of β phase can be retained to room temperature. Hence, both Ti-1023 and TB3 are composed of equiaxed β grains (Figure 1c and e). The average β grain sizes of the two alloys are about 302 and 254 μm, respectively. Because the [Mo]_eq of Ti-1023 is at a critical value and the diameter of the bar stock used is large (90 mm), the cooling rate of the core part of the bar stock is slow, so finely dispersed α phases (α_P) are precipitated in the β grains during water quenching, as shown in Figure 1c. The average size of these α_P is about 1.3 μm. No α_P precipitated in β grains of TB3 (Figure 1e), due to the higher [Mo]_eq (higher β phase stability) of this alloy. According to the TEM images (Figure 1d and f), the dislocation density is also low in both metastable β alloys before shock loading. There is a small amount of fine acicular martensite α″ phase (α″_th) activated in Ti-1023 by thermal stress during quenching process (Figure 1d). The needle-like α″ phase has an average width of about 41 nm and a length of about 910 nm. The corresponding SAED patterns (inset in Figure 1d and f) confirm no precipitation of other phases in both Ti-1023 and TB3.

3. Experimental method

3.1 Shock recovery experiments

Effective control of the lateral and rear release waves returning back during unloading process and consequently the residual strain of samples (less than 2%) is the key to correctly evaluate the effect of pre-shock loading on the mechanical properties and microstructural evolution of materials, which is the so-called soft recovery shock experiment [31]. The size of each component in the soft recovery target assembly, which is closely related to the shock Hugoniot parameters of materials and the experimental parameters such as the shock pressure and the pulse duration, determines the protective effect of the recovery assembly on the recovery sample [31, 32]. The shock pressure (P_H) is set to 5–14 GPa. According to Bourne [8], when the pulse duration is greater than 1 μs, most of the shock-induced microstructural evolution processes have enough time to occur and reach equilibrium. Hence, the pulse duration ($t_P$) is set to 1.5 μs. The density and shock Hugoniot parameters (C₀ and S) of three titanium alloys used to design the soft recovery target assembly are listed in Table 3. C₀ and S are material constants, which are related to the zero-pressure bulk sound velocity and the first pressure derivative of bulk modulus, respectively [35].

Alloys	Density ρ0/g·cm^−3	C0/mm·μs^−1	S	References
Ti-64 ELI	4.39	5.05	1.07	[33]
Ti-1023	4.59	4.99	1.06	[34]
TB3	4.82	4.92	1.09	[34]

Table 3.

Shock Hugoniot parameters of three titanium alloys.

In this work, the soft recovery target assembly was designed according to Ref. [31] and illustrated in Figure 2a. A 23-mm-diameter and 9-mm-thick Ti-alloy recovery sample was placed behind a 70-mm-diameter and 2.5-mm-thick cover plate. The recovery sample was protected from residual strain and spallation by surrounding an inner momentum trapping ring with outside diameter of 67 mm and supported with three 5-mm-thick spall plates. The inner momentum trapping ring was further concentrically surrounded by an outer momentum trapping ring with an outside diameter of 78 mm. The recovery sample and the inner momentum trapping ring were interference fit together. Two trapping rings, cover plate, and spall plates were assembled with slip fit. Molybdenum disulfide grease was used during assembly of two momentum trapping rings to remove air gaps in the assembly and to facilitate ring separation during recovery.

Figure 2.

(a) Schematic diagram of the soft recovery target assembly, and (b) typical macrograph of the longitudinal section of Ti-64 ELI recovery samples shock-loaded at different pressures.

Using the above-designed soft recovery target assemblies, shock wave loading experiments were carried out on three titanium alloys by a single-stage gas gun with a diameter of 80 mm and a length of 12.5 m. Shock waves with stress amplitudes ranging from ∼5 to 14 GPa were yielded in Ti-alloy recovery samples through the impact by flyer plates within the velocity range of 365–1092 ms⁻¹. The thickness of flyer plates ranged from 3.91 to 4.43 mm to ensure a constant pulse duration ($t_{\mathrm{P}}$) of 1.5 μs. All experiments were conducted in symmetrical impact, which means that the whole shock recovery assembly (containing the recovery sample, two trapping rings, cover plate, and three spall plates) and flyer plate were made of the same material. The impact surfaces of all recovery assemblies and flyer plates were mechanically polished to ensure that the surface roughness was less than 0.8 μm. The planarity of the target assembly to the flyer plate was controlled by adjusting the specimen mount to better than 1 mrad. The impact velocities were measured by a magnetic induction speed measurement system to an accuracy of about 0.5%.

After shock wave loading, the recovery samples were cut open longitudinally. The longitudinal section of the recovery samples showed that both end faces were flat and no cracks were observed although the edges of the recovery samples deformed slightly, as shown in Figure 2b. The residual plastic strain of the shock-loaded recovery samples measured using an outside micrometer is between 0.5 and 1.9%, indicating that the influence of lateral release waves on the recovery samples is deemed to be negligible. The residual strain is defined here as the starting sample thickness minus the recovered sample thickness following shock loading divided by the starting sample thickness.

3.2 Reloading tests and microscopic analyses

Several cylinders (4 mm in diameter and 8 mm in thickness) were sectioned from the center area of the recovery samples for quasi-static reloading compression tests. The reload tests were conducted at 298 K and under a strain rate of 10⁻³ s⁻¹ using an MTS 810 hydraulic servo machine. Tests were continuously conducted until the specimens fractured. The reloading direction is parallel to the shock wave loading direction. For comparison, initial materials of Ti-64 ELI, Ti-1023, and TB3 were also compressed under the same conditions. Quasi-static compression tests of all initial and shock prestrained materials were repeated three times to ensure the reliability of the data.

Blocks with dimensions of 7 mm × 5 mm × 2 mm were also cut from the center area of the recovery samples for microstructural analysis. Specimens for optical microscopy (OM) studies were prepared by electrochemically polishing in a solution of 95% ethanoic acid + 5% perchloric acid, followed by etching in solutions of 2% HF + 10% HNO₃ + 88% H₂O (for Ti-64 ELI) and 10% HF + 30% HNO₃ + 60% H₂O (for Ti-1023 and TB3) at room temperature. Then, the blocks were carefully detected using a ZEISS Axio Observer A1m optical microscope. X-ray diffraction (XRD) analyses for phase identification were conducted on a Rigaku Smartlab diffractometer using a Cu-K_α radiation operated at 45 kV and 200 mA. To show the change in the substructure of the preshocked alloy, TEM analyses were conducted using a JEOL JEM-2100 system at an accelerating voltage of 200 kV. For the TEM observations, thin samples with an initial thickness of 0.5 mm were cut from the blocks after OM and XRD examinations, firstly reduced to less than 40 μm thickness by mechanical means, then punched into several standard 3-mm-diameter TEM discs, and finally thinned by ion milling.

4. Mechanical behavior of the postshock titanium alloys

4.1 Stress-strain response

The mechanical behavior of three titanium alloys was detected before and after shock wave loading. Only compression tests were carried out because the tensile specimens were unable to be machined due to the limited size of the recovery samples (23 mm in diameter and 9 mm in thickness). Figure 3 shows the quasi-static reloading compressive stress-strain curves of the shock prestrained titanium alloys at room temperature. The curves of shock-free materials are also included in Figure 3 for comparison. For shock-free (0 GPa) materials, the yield stress of Ti-64 ELI is the highest, followed by Ti-1023, and that of TB3 is the lowest. On the other hand, the plasticity (fracture strain) of TB3 is the best, followed by Ti-64 ELI, and that of Ti-1023 is the worst. After shock wave loading, the shape of the stress-strain curves of three titanium alloys does not change significantly. However, the yield stress and flow stress of the postshock alloys all increase to a certain extent. Besides, the strain hardening rate of the materials decreases gradually with increasing shock pressure (P_H).

Figure 3.

Typical quasi-static reloading compression true stress-strain curves of three titanium alloys after shocked to different pressures: (a) Ti-64 ELI, (b) Ti-1023, and (c) TB3.

To further explore the influence of alloying component and stress amplitude on the shock-induced mechanical behavior, the yield stress (${\sigma }_y$) and fracture strain (${\epsilon }_f$) of the shock prestrained alloys were plotted versus P_H, as shown in Figure 4. The change trend of the reloading ${\sigma }_y$ value of Ti-64 ELI with P_H can be roughly divided into two stages, as shown in Figure 4a. When the shock pressure is less than 12 GPa, ${\sigma }_y$ increases steadily with increasing P_H. When P_H exceeds 12 GPa, the ${\sigma }_y$ value of the alloy rises rapidly. As P_H reaches 13.5 GPa, its ${\sigma }_y$ value increases by 34% compared with the shock-free material. On the other hand, the plasticity of Ti-64 ELI decreases after shock wave loading. The ${\epsilon }_f$ value of this alloy decreases in an approximate “parabolic” trend with increasing P_H, as shown in Figure 4b.

Figure 4.

Quasi-static reloading compression mechanical properties of three titanium alloys after shocked to different pressures: (a) yield stress vs. shock pressure and (b) fracture strain vs. shock pressure.

The change trend of the reloading ${\sigma }_y$ value of Ti-1023 with P_H can also be divided into two stages (Figure 4a). The ${\sigma }_y$ value of this alloy has increased significantly after shock loaded at low amplitudes (∼5 GPa). On the other hand, as P_H exceeds 6 GPa, the increasing trend of ${\sigma }_y$ slows down with increasing P_H, and the yield strength increases steadily. The plasticity of the shock prestrained Ti-1023 decreases significantly, and the ${\epsilon }_f$ value of the alloy decreases nearly linearly. Namely, compared with Ti-64 ELI, the shock-induced plasticity deterioration of Ti-1023 is more obvious (Figure 4b).

After shock loaded at 4.5–14.1 GPa, the reloading ${\sigma }_y$ value of TB3 increases steadily and continuously, as shown in Figure 4a. The reloading plastic strain of TB3 is still greater than 0.50 after shocked at different amplitudes (Figure 4b), indicating that, when P_H is less than 15 GPa, the shock wave propagation process has no obvious effect on the plasticity of this alloy.

4.2 Hardness

The Vickers hardness (HV) of the postshock titanium alloys was also measured using a WOLPERT 450 SVD Digital Vickers Hardness Tester. The test force and dwell time were set to 5 kgf (49 N) and 10 s, respectively. Generally, the shock prestrain process has no significant effect on the hardness of three titanium alloys, as shown in Figure 5. The HV values of Ti-64 ELI and Ti-1023 increase slowly with increasing P_H and tend to be saturated gradually. The HV value of TB3 increases approximately linearly with P_H; however, the increasing trend is not obvious. This indicates that the hardness of titanium alloys is not very sensitive to shock wave loading.

Figure 5.

Vickers hardness-shock pressure curves of three titanium alloys.

5. Shock-induced microstructure evolution

5.1 Phase analysis

Figure 6 shows the typical XRD patterns of the shock-free and postshock titanium alloys. All spectral lines were normalized. Compared with the initial material, except for peaks of α and β phases, no new peaks were detected in the slow-scanning XRD patterns of the postshock Ti-64 ELI, indicating that, when P_H is less than 14 GPa, no new phase is generated in the alloy during the shock pulse.

Figure 6.

Slow X-ray scans from titanium alloys after shocked to different pressures.

For the metastable β alloy, Ti-1023, in addition to the peaks involving α_P and β phases, there occur new peaks in the XRD spectrum of the alloy shocked at 4.8 GPa. These new peaks correspond to the α″ phase [36, 37], that is, the shock-induced martensite (SIM). It suggests that the β→α″ phase transformation has already occurred in Ti-1023 even if the shock pressure is relatively low. As P_H increases, the peaks of α″ phase always exist, and no new peaks appear.

For another metastable β alloy, TB3, when the shock pressure is relatively low (4.5 GPa), except for the β peaks, no new peaks appear in the XRD pattern. However, at a higher shock pressure, the α″ peaks were observed. This result indicates that the stability of β phase in TB3 is higher than that in Ti-1023.

5.2 Metallographic microstructure

Figure 7 presents the optical microstructures of Ti-64 ELI, Ti-1023, and TB3 preshocked at different pressures. For the α+β alloy, Ti-64 ELI, compared with the initial material (Figure 1a), the metallographic microstructure of the alloy does not change, even if P_H reaches 13.5 GPa (Figure 7a and b). For Ti-1023, after shocked at 4.8 GPa, the structure of the alloy has significantly changed: A large number of needle-like phases are produced in the equiaxed β grains (Figure 7c). According to the XRD analysis mentioned in Section 5.1, these acicular phases are SIMs. Most α″ laths fully develop to go throughout the entire β grains and intersect each other. As P_H increases, the morphology of α″ has no obvious change, but its amount (density) slightly increases (Figure 7d). For TB3 with higher [Mo]_eq, the number of needle-like α″ phases is very limited at relatively low pressure (4.5 GPa), as shown in Figure 7e. Hence, they cannot be detected by XRD (Figure 6). The density of SIM phase increases gradually with increasing P_H. Besides, at the same shock pressure, the amount of α″ phase in TB3 is not only less than that in Ti-1023 but most of the α″ laths in TB3 are arranged parallel to each other and rarely staggered (Figure 7f).

Figure 7.

Optical microstructures of the postshock titanium alloys: (a) Ti-64 ELI shocked at 5.2 GPa, (b) Ti-64 ELI shocked at 13.5 GPa, (c) Ti-1023 shocked at 4.8 GPa, (d) Ti-1023 shocked at 12.5 GPa, (e) TB3 shocked at 4.5 GPa, and (f) TB3 shocked at 14.1 GPa.

5.3 Substructure evolution

Figure 8 shows the TEM images of the substructures in the α phase of the Ti-64 ELI shock loaded at different pressures. Compared with the initial material (Figure 1b), after shocked at a relatively low pressure (5.2 GPa), the dislocations in the α plates become longer. The density and entanglement degree of the dislocations are increased. Some small dislocation cusps on the long dislocations (indicated by the arrows in Figure 8a) were also observed. When P_H increases to 8.1 GPa (Figure 8b), the dislocations in the α phase proliferate greatly, forming a substructure dominated by net-like planar slip. This dislocation substructure is very similar to that in the quasi-statically deformed commercial Ti-64 [38]. However, the distribution of planar slip dislocations formed by shock wave loading is more uniform, indicating that the shock-induced instantaneous deformation of Ti alloys is more uniform. When the shock pressure reaches 10.5 GPa (Figure 8c), the dislocation density continues to increase, and the dislocation entanglement is more serious. At this time, the dislocation substructure has lost the regular arrangement characteristic and tended to form a band/cluster structure. With the further increase of P_H to 12.3 and 13.5 GPa (Figure 8d and f), the high-density dislocations in the α phase are further entangled with each other, and dislocation clusters or slip bands are finally formed. The SAED pattern obtained from the α region in the preshocked Ti-64 ELI shows the existence only of α phase, consistent to the XRD analysis results.

Figure 8.

Bright field TEM images showing dislocation and twin substructures in the α phase within Ti-64 ELI shock loaded to different pressures: (a) 5.2 GPa, (b) 8.1 GPa, (c) 10.5 GPa, (d) and (e) 12.3 GPa, and (f) 13.5 GPa.

In addition to the entangled high-density dislocations, fine twins were also observed in the Ti-64 ELI after shocked at higher pressures (>12 GPa), as shown in Figure 8d–f. When P_H is 12.3 GPa, the number of twin laths is still small, and the average thickness of these twins is about 309 nm (Figure 8d). The SAED pattern (Figure 8e) confirms the twins are $\left\{10\overline{1}2\right\}$ twins [39]. Besides, similar to magnesium alloys [40], stacking faults (SFs) were also observed inside the twin laths (indicated by arrows in Figure 8e). When P_H further increases to 13.5 GPa, the number of twins in Ti-64 ELI increases substantially (Figure 8f). The average thickness of the twin laths is about 230 nm, which is slightly smaller than that of twins generated at 12.3 GPa.

Figure 9 presents the α″ substructure characteristics in the equiaxed β grains of the preshocked Ti-1023. According to these TEM images, one can find that the needle-like α″ phases observed in the optical micrographs (Figure 7c and d) are actually martensitic lath bundles. These SIM bundles are made up of many finer interweaving martensites. The SAED patterns in Figure 9 also reflect the complex internal structure of the α″ bundles. The bundles have a thickness of about 1.58 to 2.51 μm, and the average thickness is about 2.04 μm. As P_H increases, the thickness of the α″ bundle decreases, but its internal substructures become more regular. This implies the rotation and coalescence of α″ laths during the shock pulses with higher stress amplitudes. Besides, a large number of interlaced fine martensites were also observed, as shown in Figure 9b and e.

Figure 9.

TEM images showing α′′ substructures in the β phase within Ti-1023 shock loaded to different pressures: (a) and (b) 4.8 GPa, (c) 8.5 GPa, (d) and (e) 10.2 GPa, and (f) 12.5 GPa. (b)–(d) are dark field TEM images.

Figure 10 shows the substructural features in the TB3 shocked at different pressures. Compared with the initial material (Figure 1f), after shock loaded at a relatively low pressure (4.5 GPa), dislocation walls consisting of many short dislocations arranged in parallel are formed in the β grains, as shown in Figure 10a. The width of these dislocation walls is about 375 nm. When P_H continues to increase, the dislocation density gradually increases, which leads to an increase in the number of dislocation walls and intersection between them (Figure 10b and c). The width of the dislocation walls gradually decreases with increasing P_H. When P_H exceeds 10 GPa (Figure 10d and e), the dislocations in the β phase are severely entangled, and the dislocation walls begin to transform into slip bands, which are similar to those formed in 13.5 GPa-shocked Ti-64 ELI. Besides, no additional spots appear in the SAED patterns of the postshock alloys, suggesting that, when the shock pressure is less than about 15 GPa, the β→ω phase transition does not occur in TB3.

Figure 10.

Bright field TEM images showing dislocation and α″ substructures in the β phase within TB3 shock loaded to different pressures: (a) 4.5 GPa, (b) 6.8 GPa, (c) 9.2 GPa, (d) 12.1 GPa, (e) 14.1 GPa, and (f) 12.1 and 14.1 GPa.

The substructure of the α″ lath in TB3 is presented in Figure 10f. Compared with that in Ti-1023, the internal structure of the α″ lath in TB3 is relatively simple. The α″ laths have a thickness of about 0.71–1.57 μm, which is narrower than the SIM bundle formed in Ti-1023.

6. Discussion

The shock preloading and quasi-static reloading results indicate that the propagation of shock waves can significantly change the microstructure and further the mechanical properties of titanium alloys. For Ti-64 ELI, as the shock pressure is lower than 12 GPa, the shock-induced substructure of this alloy is dominated by dislocations, similar to those dislocation substructures generated in A-70 Ti [10] and commercial Ti-64 [38]. The shock-induced high-density dislocations (clusters or bands) prevent further proliferation and movement of dislocations in the material during subsequent reloading deformation process, resulting in an increase in the yield strength of Ti-64 ELI (Figure 4a). No dislocation cells or cell-like structures were observed, due to the low symmetry and relatively high Peierls stress of the α phase with HCP crystal structure. Therefore, when shock loaded at relatively low pressures (<12 GPa), Ti-64 ELI does not show a very obvious shock-induced strengthening effect like Ni [13, 14]. However, as the shock pressure exceeds 12 GPa, twinning is activated in α phase. The twin laths are able to split α grains, resulting in grain refinement. Hence, as a new strengthening mechanism, twinning makes the reloading yield strength of Ti-64 ELI rapidly increase, showing a more obvious shock strengthening effect (Figure 4a). The generation of a large number of micro defects during the shock pulse hinders the movement of dislocations, leading to the dislocation pile-up more serious and stress concentration in the process of reloading deformation. This makes it easier for micro damages to nucleate and expand and finally results in the decline in the plasticity of the postshock Ti-64 ELI, as shown in Figure 4b.

For Ti-1023 with low β phase stability, the SIM phase transformation has occurred at 4.8 GPa. A large number of interlaced α″ laths were formed and went throughout the entire β grains. The α″ phase possesses the orthorhombic crystal structure [41], and it has fewer slip systems than that in the matrix β phase with BCC structure [36]. Therefore, compared with the β phase, the α″ phase is more difficult to coordinate the plastic strain. Moreover, the martensite laths are mostly interlaced with each other and their internal structure is very complex (Figures 7 and 9). These substructures will greatly hinder the dislocation movement and cause the interaction between the dislocations and the martensites or cause the dislocations pile-up at the α″/β boundaries. Consequently, the deformation resistance and yield stress of the postshock material dramatically increase during the reloading compression process even if the shock pressure is relatively low (4.8 GPa), as shown in Figure 4a. The similar strengthening effect caused by SIM transition was also observed in the 7 GPa shocked Ti-12 Mo [20]. However, the number of α″ phases in Ti-1023 has tended to be saturated at 4.8 GPa, and its amount has little change with increasing P_H. Hence, the increase in yield stress of the postshock alloy tends to be gentle (Figure 4a). Compared with dislocation clusters/bands and twins, the interlaced acicular α″ martensites hinder the movement of dislocations more seriously, so the plasticity of Ti-1023 decreases sharply after shock wave loading.

When the shock pressure amplitude is similar, the amount of α″ martensite formed in TB3 is far less than that formed in Ti-1023 (Figure 7), due to the higher stability of β phase in TB3. Most of the α″ laths in TB3 are arranged in parallel and rarely interlaced. Their internal structure is also relatively simple (Figure 10f). The smaller grain size also inhibits the occurrence of β→α″ phase transition to some extent [42]. Therefore, the shock-induced α″ phase has little contribution to the shock strengthening effect of TB3; however, it also has little effect on the plasticity of the alloy (Figure 4b). On the other hand, although a certain amount of dislocation walls or slip bands are formed in TB3, the degree of dislocation proliferation and entanglement is not as good as that in Ti-64 ELI under shock loading conditions. Besides, no twinning occurred in TB3. However, shock-induced twins were observed in some other β titanium alloys such as Ti-26 [20]. The higher β phase stability (higher [Mo]_eq) of Ti-26 alloy may completely suppress the occurrence of SIM transition. Hence, twinning replaces SIM transformation to coordinate the deformation during shock loading pulse. The single type of micro defect (only dislocations) and the low density of dislocations and α″ phases make the shock-induced strengthening effect in TB3 limited, and the yield stress increases gently, as shown in Figure 4a.

The extent of the shock-induced strengthening in titanium alloys significantly depends on the type, morphology, and density of the defects that are decided by the alloying composition (alloy type). For α+β alloy, Ti-64 ELI, the stability of both α and β phases is very high due to the enrichment of α- and β-stabilizing elements in α and β phases, respectively. Hence, the shock-induced substructure evolution in Ti-64 ELI is dominated by dislocation multiplication and interaction. Twinning is activated as dislocation proliferation tends to saturate during shocked at higher pressures (>12 GPa). For metastable β alloys, Ti-1023 and TB3, the β-stabilizing elements are uniformly distributed in the β phase of the alloys after solution treatment, and the β phase is in an unstable state. Therefore, except for dislocation substructures, shock-induced α″ martensites are also generated. However, the number, morphology, and distribution of the α″ phase depend on the stability of the β phase, which is determined by the content of β-stabilizing elements. The content of β-stabilizing elements in TB3 is higher than that in Ti-1023; hence, the proliferation, growth, and interlacing of α″ martensites are weaker in TB3 due to the higher stability of β phase in this alloy.

In general, when P_H is less than 15 GPa, with the increase of [Mo]_eq (from Ti-64 ELI to TB3), the shock-induced microstructure evolution characteristics of titanium alloys can be roughly summarized as follows: planar slip (tangled dislocations, slip bands) + twins (Ti-64 ELI) → interlaced α″ martensites (Ti-1023) → paralleled α″ martensites + planar slip (dislocation walls, slip bands) (TB3). That is to say, the main deformation mechanism of the three titanium alloys during the shock pulse undergoes the evolution of “slip/twinning → β/α″ phase transition → β/α″ phase transition + slip.”

7. Conclusions

In this chapter, the effects of alloying composition and stress amplitude on the shock-induced mechanical behavior and microstructure evolution of three titanium alloys are revealed through a series of soft recovery shock experiments, quasi-static reloading tests, and multi-scale microscopic analyses. The major conclusions are as follows:

1. When P_H is less than 12 GPa, the shock-induced substructure of Ti-64 ELI is dominated by tangled dislocations. When P_H exceeds 12 GPa, dislocation entanglement is further intensified to form dislocation clusters/bands, and $\left\{10\overline{1}2\right\}$ twin is also activated. Correspondingly, the yield stress of the postshock alloy first increases slowly and then increases rapidly.

2. A large amount of α″ martensite phase is produced in Ti-1023 during the shock pulses. Most of the α″ laths have a complex internal structure, interlace with each other, and go through the entire β grains. The existence of high-density martensite laths makes Ti-1023 show a significant shock strengthening effect even if the shock pressure is relatively low. However, the plasticity of the preshocked alloy is also degraded seriously.

3. Higher β phase stability and smaller grain size inhibit the β→α″ transition in TB3. The density of the martensite phase is low in this alloy, and the α″ laths are mostly arranged in parallel with few intersections. The internal structure of the α″ laths is relatively simple. A certain amount of dislocation walls or slip bands also form in TB3, but no twins generate during the shock loading process. Hence, the shock-induced strengthening effect in TB3 is limited and the yield stress increases gently due to the single type of micro defect and the low density of defects and α″ phases.

4. When P_H is less than 15 GPa, with the increase of [Mo]_eq (from Ti-64 ELI to TB3), the main deformation mechanism of the three titanium alloys during shock pulse undergoes the evolution of “slip/twinning → β/α″ phase transition → β/α″ phase transition + slip.”

Acknowledgments

This study was funded by China Postdoctoral Science Foundation (Grant number: 2021M690012), the Young Scientists Fund of the National Natural Science Foundation of China (Grant number: 51501064), and the Fundamental Research Funds for the Central Universities (Grant number: 2019MS012).

Conflict of interest

The authors declare no conflict of interest.