Introduction

Metallic materials are usually at the mercy of the strain rate effect, i.e., the yield strength improves while the ductility declines with increasing strain rate. As the strain rate increases, the thermal activation effect for dislocations decreases and the dislocation viscous-drag effect is promoted1,2. Both would raise Peierls–Nabarro stress and then contribute to strengthening. During dynamic deformation, high-density dislocations resulting from rapid multiplication cannot glide readily, so severe dislocation pile-up happens then leads to stress concentration and crack propagation at the early deformation stage. In addition, strong shear stress and pronounced adiabatic effect prevail with increasing strain rate, leading to severe deformation localization and subsequent formation of adiabatic shear bands, which deteriorates ductility3,4. Therefore, the strength-ductility trade-off of metallic materials with increasing strain rates has been a long-standing challenge.

Recently, high-entropy alloys (HEAs), which generally exhibit solid solution phases with high configuration entropy, were proposed5,6,7. They have intrinsic characteristics such as pronounced lattice distortion8,9, local chemical orderings (LCOs)10,11 and complex chemical environment12,13,14, which brings intriguing properties, especially under extreme conditions such as at low-temperature15,16, high-temperature17,18 and dynamic loading19,20,21. Individually, in face-centered cubic (FCC) structured 3d transition HEAs (3d-HEAs), the complex chemical environment modulates stacking-fault energy and reduces the critical stress for twinning12,22,23. The formation of deformation twins alleviates stress concentration induced by dislocation entanglement and massive pile-ups. As a result, FCC 3d-HEAs generally exhibit a combination of high strength and decent ductility upon dynamic loading19,23,24. However, for body-centered cubic (BCC) structured refractory HEAs (RHEAs), which are mainly composed of refractory elements and are more promising for dynamic applications25,26,27, the high stacking-fault energy is unfavorable for twinning deformation. Moreover, the low thermal conductivity of RHEAs inherited from the constituents’ intrinsic characteristics would aggravate adiabatic shear effect28,29, which makes the corresponding ductilization upon dynamic loading more challenging and intractable. Theoretically, if the severe deformation localization and formation of adiabatic shear bands can be alleviated, enhanced ductility may be achieved upon dynamic loading in RHEAs.

We herein propose to stimulate the dislocation multiple slips by LCOs in RHEAs, thus homogenize the local strain and alleviate the deformation localization10,14,30, which may give rise to the improved dynamic ductility. We demonstrated exemplarily this design concept in a TiZrNbTa RHEA. When the strain rate surges from a quasi-static regime (10−3 s−1) to a dynamic one (6500 s−1), the yield strength increases by 89% to 1879 ± 10 MPa. More surprisingly, the ultimate true strain can reach up to 103.7 ± 1.7% and no fracture occurs at the strain rate of 6500 s−1, illustrating a superior strength-ductility synergy upon dynamic loading.

Results

Mechanical response at different strain rates

Figure 1a shows the true stress-strain curves of TiZrNbTa RHEA at strain rates ranging from 10−3 s−1 to 6500 s−1, as well as the corresponding deformed samples. When the loading transfers from a quasi-static regime (10−3 s−1) to a dynamic one (800 s−1), the yield strength increases from 996 ± 7 MPa to 1706 ± 17 MPa, i.e., an increment of 71%. As the strain rate reaches 6500 s−1, the yield strength further increases to 1879 ± 10 MPa (89% higher than the quasi-static one). Moreover, although the stress declines at ~40% strain, it remains stable at ~1800 MPa upon the strain ranging from 50% to 80%. Then the stress surges with strain larger than 80% and the ultimate strength at ~103% strain reaches 1932 ± 14 MPa. The ultimate true strain of TiZrNbTa RHEA deformed at the strain rate of 6500 s−1 is larger than 100%. The shape of deformed TiZrNbTa RHEA is symmetrical and no obvious fracture or crack appears (see inset in Fig. 1a), demonstrating the remarkable deformation capacity of TiZrNbTa RHEA upon dynamic loading. Figure 1b compares the ultimate strain under the dynamic regime of several typical metallic materials20,25,26,27,28,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55. The hollow labels indicate the strain rate at which fracture happens and the corresponding fracture strain, while the solid labels represent the ultimate strain without fracture. Most hollow labels locate at the strain rate range below 6000 s−1, revealing that most metallic materials, including 3d-HEAs31,32,33,34,35,36, steels38,39,40,41 and pure metals42,43,44,45 which possess good quasi-static ductility, would fracture under dynamic loading. Only several materials including TiZrNbTa RHEA exhibit large dynamic ductility without fracture in high strain rate scope (>6000 s−1).

Fig. 1: Mechanical properties of TiZrNbTa refractory high-entropy alloy (RHEA) at different strain rates.
figure 1

a True stress-strain curves of TiZrNbTa RHEAs at different strain rates (from 10−3 s−1 to 6500 s−1) and all the curves are incomplete but representative (as interpreted in “Methods”). The insets are pictures of samples deformed at the corresponding strain rate. TiZrNbTa RHEA exhibits high strength and large ductility under dynamic loading. b The comparison of ultimate strain under dynamic regime for the current TiZrNbTa RHEA and other reported HEAs20,25,26,27,28,31,32,33,34,35,36,37, steels38,39,40,41, pure metals42,43,44,45, tungsten alloys46,47,48, titanium alloys49,50,51,52,53, tantalum alloys54, and superalloy Inconel 71855, revealing the distinguished ultimate strain of TiZrNbTa HEA at high strain rate regime.

Microstructural characterization of TiZrNbTa RHEA

To understand the origin of the superior mechanical properties upon dynamic loading, microstructures down to the atomic scale of the as-prepared TiZrNbTa RHEA were systematically analyzed. Figure 2a shows the X-ray diffraction (XRD) pattern of TiZrNbTa RHEA. Only one set of diffraction peaks corresponding to BCC structure appears, revealing the single BCC phase structure. Also, as illustrated in the inverse pole figure (IPF) obtained from electron backscatter diffraction (EBSD) (Fig. 2b), the TiZrNbTa RHEA is composed of randomly oriented equiaxed grains. Scanning electron microscopy (SEM)-backscattered electron image (Supplementary Fig. 1a), scanning transmission electron microscopy high-angle annular dark-field (STEM-HAADF) micrograph and corresponding selected area electron diffraction (SAED) pattern (Supplementary Fig. 1b) further verifies the single BCC structure. Moreover, the SEM-energy-dispersive spectroscopy (EDS) (Supplementary Fig. 1a) and STEM-EDS maps (Supplementary Fig. 1b) also demonstrate the elemental homogeneity at the microscale and nanoscale in TiZrNbTa RHEA.

Fig. 2: Microstructural characterization of TiZrNbTa RHEA.
figure 2

a X-ray diffraction pattern and b inverse pole figure (IPF) of the as-prepared TiZrNbTa RHEA, demonstrating the single-phase body-centered cubic structure and randomly oriented equiaxed grains. c Aberration-corrected scanning transmission electron microscope (STEM) high-angle annular dark-field image of the studied alloy and d the corresponding energy-dispersive spectroscopy line profile of the yellow rectangle area in Fig. 2c. The red and bule rectangles are used as guides for the eye. Some atomic columns are enriched in Zr (see the red rectangle area) while some are enriched in Nb and Ta (see the blue rectangle area), revealing the existence of Zr-enriched and (Nb, Ta)-enriched local chemical orderings (LCOs). Three-dimensional atom probe tomographic reconstructions of the studied alloy, the threshold for the iso-composition surface is (e) 28 at.% Zr and (f) 58 at.% (Nb+Ta). g Elemental concentration profile across the iso-composition surface of Zr-enriched and (Nb, Ta)-enriched LCOs. These results further verify the Zr-enriched and (Nb, Ta)-enriched LCOs.

The aberration-corrected STEM and three-dimensional atom probe tomography (APT) were used to visualize the atomic-scale chemical fluctuation in TiZrNbTa RHEA. Figure 2c illustrates an aberration-corrected STEM-HAADF micrograph with the incident electron beam along the [111]BCC zone axis of the grain selected. Figure 2d showcases the EDS line profiles of the projected atomic columns in the yellow square area of Fig. 2c. There is one atomic column that is enriched in Zr but depleted in Nb and Ta (see the red rectangle area in Fig. 2d). Also, one atomic column which is enriched in Nb and Ta but depleted in Zr (see the blue rectangle area in Fig. 2d) can be found. These observations of the chemical deviation from the randomness in the interatomic scale indicate the formation of Zr-enriched and (Nb, Ta)-enriched LCOs in TiZrNbTa RHEA56. To further clarify the chemical composition and the three-dimensional microstructure information of LCOs, three-dimensional APT reconstruction was employed to characterize the Zr-enriched (Fig. 2e) and (Nb, Ta)-enriched (Fig. 2f) LCOs. The iso-composition surfaces of 28 at.% Zr and 58 at.% (Nb+Ta) are shown in Fig. 2e and Fig. 2f, respectively. Figure 2g exhibits the composition profiles as a function of the distance from the matrix to Zr-enriched and (Nb, Ta)-enrich LCOs, respectively. The element partitioning behavior further proves the presence of Zr-enriched and (Nb, Ta)-enriched LCOs. It should be noted that LCOs in TiZrNbTa alloy are the atomic-scale chemical fluctuations that might be formed during annealing treatments (Supplementary Fig. 2). The differences in mixing enthalpy and diffusion coefficient contribute to the formation of LCOs57.

Microstructure evolution during dynamic-deformation

To reveal the underlying mechanism of this distinguished dynamic mechanical properties of TiZrNbTa RHEA, we studied the microstructure of samples after deformation under different strain rates. Supplementary Fig. 3 shows XRD patterns of TiZrNbTa RHEA deformed at the strain rates of 800 s−1, 3200 s−1 and 6500 s−1. These deformed RHEAs retain a single BCC phase structure, revealing that no phase transformation takes place during dynamic deformation. Figure 3a(i), b(i), 3c(i) displays IPFs with Z-direction (loading direction) of deformed TiZrNbTa HEAs at the strain rates of 800, 3200 and 6500 s−1, respectively. The 11.8% strained RHEA at the strain rate of 800 s−1 shows equiaxial and randomly oriented grains (see Fig. 3a(i)). A tiny misorientation of 1–3° caused by microstrain spreads randomly within the grains, as shown in the Kernel average misorientation (KAM) mapping (Supplementary Fig. 4a). As the strain rate increases to 3200 s−1, a certain degree of texture with {001}BCC or {111}BCC orientations forms in the grains after 37.1% strain (see Fig. 3b(i)), which is similar to that in the BCC structured alloy deformed upon dynamic loading58. The corresponding KAM mapping indicates the uniformly distributed microstrain (see Supplementary Fig. 4b). Upon further increases in the strain rate (i.e., 6500 s−1), a deformed band with a width of ~30 μm and numerous bifurcations appear in the 103.7% strained alloy (Fig. 3c(i) and Supplementary Fig. 5). Moreover, most grains exhibit {001}BCC or {111}BCC texture. The microstrain is found to be enhanced but remains relatively uniform (Supplementary Fig. 4c). The intensity of (200)BCC peak in XRD patterns enhances with increasing strain rate (Supplementary Fig. 3), which also demonstrated the above texture evolution.

Fig. 3: EBSD and TEM characterized structural evolutions at different strain rates.
figure 3

a(i), b(i), c(i) are Z-IPFs at the strain rates of 800 s−1, 3200 s−1, and 6500 s−1, respectively, showing the enhancement of {001}BCC and {111}BCC crystallographic texture with increasing strain rate. Z shows loading direction. The bifurcate deformed bands appear at the strain rate of 6500 s−1. a(ii)–(iii), b(ii)–(iii), c(ii)–(iii) are TEM images at the strain rates of 800 s−1, 3200 s−1, and 6500 s−1, respectively. Dislocations in linear arrays are pinned (red dashed line) at a low strain rate. Dislocation planar slip evolves into multiple slip (red dashed circles) and multi-directional microbands (white arrows) with increasing strain rates. c(iv), c(v) are TEM bright-field and dark-field images of the area within deformed bands at the strain rate of 6500 s−1, respectively. c(vi) The high-resolution TEM image of the white square area in (c(iv)). The deformed band is comprised of fibrous nanocrystals resulting from grain boundary bowing out (red dashed polygon).

To further clarify the dynamic deformation mechanism of this material, the dislocation substructures at different strain rates and strains were investigated by TEM. Massive linear dislocation arrays on the {110}BCC lattice plane appear in the 11.8% strained alloy deformed at the strain rate of 800 s−1 (Fig. 3a(ii)). This typical morphology of dislocation planar slip can also be observed in the quasi-static deformed TiZrNbTa with a 10% strain (Supplementary Fig. 6a). Meanwhile, a small number of curved dislocations (outlined by red dashed curves in Fig. 3a(iii)), which are generally caused by pinning, are found in the linear dislocation arrays. As the strain rate increases to 3200 s−1, microbands with a width of ~100 nm on the {110}BCC plane can be seen in the 37.1% deformed alloy (Fig. 3b(ii)). Moreover, vast dislocation debris as circled by red dashed curves can be found in the regions beside microbands (Fig. 3b(iii)). Such morphology with planar and multiple slip features was not seen in the quasi-static deformed alloy with a strain of 35% (Supplementary Fig. 6b). For the alloy deformed at 6500 s−1 with a true strain of 103.7%, several microbands measuring ~200 nm wide are observed (Fig. 3c(ii)). These microbands are not parallelly arranged and also exhibit the deviation from traces of the {110}BCC slip planes, which may stem from the dislocation of multiple slips. High-density dislocation debris indicated by red dashed circles exist beside the microbands, as shown in Fig. 3c(iii). In contrast, only parallel thick microbands but no dislocation debris are observed in the quasi-static deformed alloy with a true strain of ~100% (Supplementary Fig. 6c). More importantly, plenty of elongated fibrous nanocrystals are found in the deformed bands (Fig. 3c(iv)), which can be verified by the corresponding dark-field image (Fig. 3c(v)) and diffraction rings in the SAED pattern (insert in Fig. 3c(iv)). An equiaxed subgrain (outlined by red dashed curves in Fig. 3c(vi)) forms beside a triple junction of grain boundaries. This nucleation through the grain boundary bowing out usually occurs during recrystallization.

Discussion

We note that TiZrNbTa RHEA exhibits both high strength and large ductility upon dynamic loading. When the strain rate (\(\dot{\varepsilon }\)) increases from quasi-static (10−3 s−1) to dynamic regimes (800 s−1 and 6500 s−1), the yield strengths (σy) of TiZrNbTa HEA are substantially increased by ~71% and ~89%, respectively. These increments are larger than those of almost all reported HEAs, such as CrMnFeCoNi23 (~46%) and TiZrHfNbTa29 (~31%) HEAs at the similar strain rate, indicating a notable strain rate effect. According to thermal activation theory, the strain rate effect is dependent on the ability of dislocation to overcome short-range obstacles at the atomic level with the aid of thermal activation1. Upon dynamic loading with a high strain rate, thermal activation effect decreases, the requirement of external loading for dislocations to surmount short-range barriers increases, and then the yield strength increases. For HEAs, chemical complexity induces an inhomogeneous atomic environment and a local stress field, which will further dampen the thermal activation effect and dislocation activation. Also, LCOs are a kind of strong short-range obstacle56. Thus, a remarkable increment of strength emerges in TiZrNbTa RHEA. Upon dynamic loading (strain rate at ~103 s−1), the strain rate effect of materials is mainly affected by the viscous-drag effect1,19,59, i.e., the interaction between the dislocations and phonons, then the yield strength surges with increasing strain rate as shown in Supplementary Fig. 7. Moreover, the existence of LCOs would lead to the inhomogeneous local density of electrons60,61. This promotes the frequency and energy range of lattice vibration and phonon, which reduces their energy. Due to the high atomic mass of constituents and the presence of LCOs, the strain rate sensitivity parameter m (defined by \(\frac{\partial {{{{{\rm{ln}}}}}}\,{\sigma }_{y}}{\partial {{{{{\rm{ln}}}}}}\dot{\varepsilon }}\)) under dynamic regime of TiZrNbTa RHEA is just 0.043 (see Supplementary Fig. 7), which is much lower than those of 3d-HEAs and conventional alloys without LCOs, as listed in Supplementary Table 2.

The dynamic-deformation mechanisms of TiZrNbTa RHEA are schematically illustrated in Fig. 4. Under quasi-static loading, dislocation planar slip dominate the deformation process of TiZrNbTa RHEA (Supplementary Fig. 6) and LCOs could induce the dislocation coplanar slip then decent ductility62. Upon dynamic loading, LCOs play the pinning effect for dislocations on the {110} plane (see Fig. 4a). Consequently, the curved dislocations are observed in linear arrays in the alloy deformed at 800 s−1 with an 11.8% strain (Fig. 3a(ii), a(iii)). With the increase of strain rate, the grains start to rotate, so that the normal of the active slip plane aligns along the direction of the compression load. Hence, the randomly orientated grains are realigned into the preferred orientations, such as <100> and <111> crystallographic directions (see Fig. 3b(i)). Deformation texture enhances the resolved shear stress on the primary {110} plane63, leads to a portion of dislocations pinned by LCOs be released, and then promotes dislocation slips and proliferation to form microbands (Fig. 4b). Furthermore, the Zr-enriched and (Nb, Ta)-enriched LCOs possess different local arrangements which bring different resistance to dislocation slip14. Thus, dislocations would not slip on a constant plane, instead, multiple slips are promoted (see Fig. 4b), and a large quantity of dislocation debris along with microbands appear in the alloy deformed under 3200 s−1 with a 37.1% strain (Fig. 3b(ii), b(iii)). When the strain rate further increases to 6500 s−1, the preferred orientation in almost all grains (Fig. 3c(i)) maximizes the resolved shear stress to overcome the pinning effect. Hence dislocations evolve into multi-directional microbands besides multiple slips (Fig. 3c(ii), c(iii)). The LCOs stimulate dislocation multiplication, resulting in high-density dislocation debris (Fig. 3c(iii)), microband broadening (Fig. 3c(ii)) and resultant bifurcated deformed bands (Fig. 3c(i)). In addition, the high-density dislocation entanglements inside the deformed bands transfer into elongated fibrous nanocrystals (Fig. 3c(iv), c(v)) and a few equiaxed subgrain form through the grain boundary bowing out (Fig. 3c(vi)).

Fig. 4: Schematical illustration of dynamic-deformation mechanisms in TiZrNbTa RHEA.
figure 4

ac show substructures under the strain rates of 800 s−1, 3200 s−1, and 6500 s−1. LCO represents the local chemical ordering. As the strain rate increases, planar dislocations evolve into multi-directional microbands and dislocation multiple slip is activated.

Adiabatic shear plays a decisive role in the dynamic-mechanical behavior of refractory metals since the dynamic deformation is an adiabatic process1,28,64. Strong shear force during dynamic deformation brings the localized deformation and heat concentration, then adiabatic shear bands. One type of adiabatic shear band is intrinsic, which is linearly along a 45° deviating direction from the loading direction due to the shear stress concentration. Another type of adiabatic shear band forms along the boundaries where dislocation pile-ups and local stress concentration appear3. The TiZrNbTa RHEA with a single BCC phase structure (Fig. 2a) and a homogeneous elemental distribution (Supplementary Fig. 1) has good resistance to the formation of the second type of adiabatic shear band. Adiabatic shear dissipation energy is a parameter to describe the critical energy barrier for the formation and extension of adiabatic shear band65 (detailed discussion in Supplementary Note 1). The lower the thermal conductivity, the smaller the adiabatic shear dissipation energy and the higher the adiabatic shear susceptibility. The thermal conductivity of TiZrNbTa RHEA is ~14 W−1 K−1, which is much lower than that of all its components and most conventional alloys (Supplementary Table 1). The low thermal conductivity can be attributed to the slow phonon conduction caused by the large atomic mass as well as the strong electron scattering caused by the LCOs66. Thus, the theoretical adiabatic shear dissipation energy of TiZrNbTa RHEA should be extremely low (detailed discussion in Supplementary Note 1). Also, the fitting results based on the Johnson-Cook model67 (detailed discussion in Supplementary Note 2) demonstrate that the thermal effect indeed exists in the dynamic deformed TiZrNbTa RHEA. The thermal effect could also be confirmed by the macroscopic mechanical behavior and microstructure evolution upon dynamic loading. Upon the strain rate of 6500 s−1, a slight stress decline appears at ~40% strain as a result of the 288 K adiabatic-temperature rise, which is quantified via Supplementary Eq. (5) and the curve in Fig. 1a. With strain further increases, the adiabatic-temperature rise is also enhanced, and that at ~103% strain reaches 758 K. Consequently, the recovery and incipient recrystallization indicated by respective plenty of fibrous nanocrystals and a few equiaxed subgrains (Fig. 3c(vi)) are found in deformed bands with higher temperature rise than the matrix. The obvious thermal effect indicates that the adiabatic shear band of TiZrNbTa RHEA should form at a relatively low strain rate. However, no typical signs of adiabatic shear, including stress collapse in stress-strain curves25,26,68 and single linear shear band with the fusion morphology27,29 were observed in the deformed TiZrNbTa RHEA, even when the strain rate reaches 6500 s−1 (Fig. 3c(i)). By contrast, the critical strain rates for the formation of adiabatic shear bands in single BCC structured alloys25,26,29,69 are usually lower than 3500 s−1, and those for the BCC structured pure metals64 and other metallic materials20,33,47,49,70,71,72,73 are lower than 6000 s−1 (Fig. 5). Specifically, the adiabatic shear bands in TiZrNbV26 and HfZrTiTa0.5325,74 RHEAs with sheet specimens appear at relatively low strain rates (~4800 s−1 and ~1500 s−1, respectively) according to the corresponding split Hopkinson pressure bar experiment results. This suggests that the formation of adiabatic shear bands in RHEAs upon dynamic deformation mainly depends on their intrinsic characteristics and microstructure.

Fig. 5: Critical strain rate for the formation of adiabatic shear bands in TiZrNbTa RHEA compared with those in other metallic materials.
figure 5

The reference materials include single body-centered cubic (BCC) structured alloys (TiVZrNb26, TiZrHfNbTa29, TiZrHfTa0.5325, Ti-55511 alloy69) and Ta64, single body-centered cubic (FCC) structured alloys (Al0.3CrFeCoNi70, CrMnFeCoNi71, Cu−11.14%Al72), and alloys with complex phase structures (AlCrFeCoNi2.133, FeNiMoW20, 93W-4.9Ni-2.1Fe47, Ti-6Al-4V49, ultrahigh strength steel73). TiZrNbTa RHEA shows a higher critical strain rate for the formation of adiabatic shear bands.

It can be speculated that the LCOs in TiZrNbTa RHEA contribute to the factually low adiabatic shear susceptibility. As illustrated in Fig. 4, LCOs stimulate the dislocation multiple slips (Fig. 3b(iii), c(iii)), which efficiently facilitate dislocation multiplication, disperse the stress and deformation across different grains and planes, alleviating the strain localization (Supplementary Fig. S3) and heat concentration. So thermal effect could be balanced even overcome and there exists a plateau and hardening in the later stage of the curve at 6500 s−1. The exceptional multi-directional microbands (Fig. 3c(ii)) and bifurcated deformed bands rather than adiabatic shear bands with white-etching morphology emerge in the TiZrNbTa deformed at a high strain rate of 6500 s−1 (Fig. 3c(i)). If the strain rate further increases, the larger deformation would result in severer deformation localization as well as thermal effect, leading to adiabatic shear bands finally75,76. In this sense, the critical strain rate for the appearance of adiabatic shear band and subsequent failure in TiZrNbTa HEA is delayed to higher than 6500 s−1. The absence of adiabatic shear cracks, as well as the high-density dislocations and nanocrystals which can neutralize the thermal softening effect, jointly contribute to the high ultimate strength (1932 ± 14 MPa) and large ultimate strain (≥103%) of the TiZrNbTa RHEA under high strain rate loading. It is noted that the above discussion on the effect of LCOs on the dynamic-mechanical behavior of TiZrNbTa RHEA is mainly based on the characterization of the current specimens before and after deformation. The general effects of LCOs on the dynamic-mechanical behavior of HEAs definitely need further investigation.

Conclusion

In summary, we report a combination of high strength and large ductility upon dynamic loading in TiZrNbTa RHEA. At a strain rate of 6500 s−1, TiZrNbTa RHEA shows dramatically enhanced yield strength (1879 ± 10 MPa) and striking ultimate true strain (103% ± 1.7% without fracture). The significant strain rate effect can be attributed to the amplified short-range obstacles for dislocation slips in this alloy. Furthermore, the LCOs-facilitated dislocation multiple slips alleviate the stress concentration, lead to the formation of multi-directional microbands and bifurcated deformed bands, delay the adiabatic shear failure and consequently improve the dynamic ductility. Moreover, the high-density dislocations and nanocrystals in the deformed bands could counteract the thermal softening effect, which also contributes to the improved ultimate strength and ultimate strain. This work opens up a route for optimizing dynamic mechanical properties of metallic materials.

Methods

Materials preparations

The TiZrNbTa alloy ingot was prepared by arc melting a mixture of raw metals (purity >99.9 wt%) in a Ti-gettered high-purity argon atmosphere. To ensure homogeneity, the ingot was remelted at least eight times. Then the ingot was drop-casted into a water-cooled copper mold with dimensions of 10 mm × 10 mm × 60 mm. The as-cast alloy was sealed in quartz tubes filled with high-purity argon and experienced the heat treatments at 1250 °C for 24 h, finally quenched in water.

Microstructural characterization

The crystal structure was characterized by X-ray diffractometer (Rigaku Dmax 2500). X-ray source was a Cu-Kα (λ = 0.1542 nm) ray, the range of the scanning angle (2θ) was 20–80° and the scanning speed was 5°/min. The microstructures were examined by a Hitachi S-3400N SEM equipped with an HKL EBSD system. The step size was 0.3 μm, and the scan size was 1420 μm × 1063 μm. The EBSD samples were mechanically ground using 800- to 2500-grit SiC papers and polished with diamond spray polishing compounds as well as a Fischione 1061 ion mill. The data analysis of EBSD was conducted via Channel 5. For KAM analysis, the 5th nearest neighbor pixels and a maximum misorientation angle of 5° were chosen. The fine microstructures were observed via TEM (FEI Talos F200X) and aberration-corrected STEM (Thermo Scientific Themis Z in Analytical Instrumentation Center of Hunan University). The TEM/STEM samples were mechanically ground to a thickness of 50 μm and then prepared by Gatan 695 ion milling. The 3D-APT analyses were carried out in a CAMECA Instruments LEAP 5000XR local electrode atom probe system under an ultrahigh vacuum of approximately 2.5 × 10−11 torr, the specimens were analyzed in laser mode, with a specimen temperature of 50 K, a pulse repetition rate of 200 kHz, a pulse energy of 120 pJ, and a detection rate of 0.5% ions per filed evaporation pulse. Imago Visualization and Analysis Software (IVAS) version 3.8.0 was used for 3D reconstructions, composition analysis and the creation of iso-concentration surfaces. The sharp tip specimens required for 3D-APT were fabricated by focused ion beam milling on a dual-beam FEI Helios 600.

Mechanical and physical tests

The quasi-static compression test was performed using a universal mechanical machine (America Instron-3369) at a strain rate of 10−3 s−1. The test was performed on specimens with the dimension of Φ 5 mm × 10 mm. Correspondingly, the dynamic compression test was carried out via a split Hopkinson pressure bar (SHPB) at the National University of Defense Technology, which is detailedly introduced in Supplementary Note 3. The main components of the SHPB test apparatus were a bullet, an incident bar and a transmitted bar, as illustrated in Supplementary Fig. 9. According to the one-dimensional stress wave theory, the stress-strain-strain rate relationships could be derived from stress wave propagation in the bars77. Two dimensions of specimens, i.e., Φ 8 mm × 3 mm25,74 and Φ 5 mm × 2 mm26, were selected to obtain data under different strain rates. Both dimensions could keep the stability of a one-dimensional stress wave, make the specimen in dynamic equilibrium and neglect the wave propagation effect. Due to the time duration and magnitude of the incident wave as well as the large deformation of the alloy, the stress-strain curves were not completely recorded at all dynamic strain rates. These aforementioned experiments were carried out at room temperature and repeated at least three times at each strain rate, the corresponding results of mechanical tests and the average yield strength are listed in Supplementary Table 3. The sample with a diameter of 12 mm and thickness of 2 mm was prepared for thermal diffusivity measurement through the laser flash method (Netzsch, LFA 447) with an accuracy within ±0.8%.