Deformation behavior of a TiZr-based metallic glass composite containing dendrites in the supercooled liquid region

1. Introduction

Bulk metallic glasses (BMGs) exhibit superior properties, such as high strength, high corrosion resistance, and high elastic limit [[1], [2], [3], [4], [5]]. Upon heating, BMGs exhibit the supercooled liquid region (SLR), where the glass will relax into a metastable liquid and behave more like a liquid than a solid before it eventually crystallizes [[6], [7], [8], [9], [10], [11]]. However, the absence of ambient plasticity due to highly-localized shear deformation and the generation of shear bands in BMGs seriously inhibits their application [[12], [13], [14]]. To overcome this limitation, bulk metallic glass matrix composites (BMGCs) were prepared through introducing a ductile crystalline embedment into the metallic glass (MG) matrix, which induces the formation of multiple shear bands during deformation, contributing to the improvement of plasticity [[15],[16]]. Recently, some in-situ BMGCs have been prepared to present high strength, high tensile ductility as well as apparent work hardening capacity in the ZrCu- and TiZr- based alloys, due to deformation-induced martensitic transformation in the metastable crystalline phase, showing tremendous progress toward their application as structural materials [[17], [18], [19], [20], [21], [22]].

In the SLR, the rheological behaviors including superplasticity and relatively low flow stress, etc., occur in BMGs, providing an opportunity for net shaping and forming process in the macro, micro and nano scales [[23],[24]]. Although the deformation behaviors are inherited by BMGCs in the SLR due to the amorphous alloy matrix, the influence of the crystalline embedment has not been understood well so far. It is susceptible to the character, volume fraction and microstructure of crystalline embedment, the amorphous alloy matrix and testing condition, etc [[25], [26], [27], [28]]. Fu et al. found that some Zr-Cu-Al BMGCs with 0-20 vol.% in-situ brittle intermetallic reinforcements exhibited the homogeneous flow behaviors in the SLR under compressive loading and the flow stress increased with increasing the volume fraction of intermetallic. There are scarce works reported on the SLR deformation behaviors of BMGCs containing higher volume fraction intermetallic. Many studies explored the SLR deformation behaviors of BMGCs containing soft dendrites of above 50 vol.%, and reported that some Ti-Zr-Nb-Cu-Be alloy samples exhibited the steady state homogeneous flow behaviors with work hardening in compression and softening in tension [29], while some Ti-Zr-V-Cu-Be alloy samples are deformed in an inhomogeneous way [[30],[31]]. Additionally, the SLR deformation would present a significant impact on the thermal stability of the supercooled melt. Nanocrystallization is probable to be accelerated, which affects the SLR deformation behaviors again [[7],[32]]. No agreement has been reached on the influence of the crystalline phase on SLR deformation behaviors for BMGCs, and furthermore, much more work needs to be done to clarify the underlying mechanism.

As aforementioned, BMGCs containing metastable dendrites exhibit superior mechanical performance due to martensitic transformation induced plasticity at room temperature. Regarding to the deformation behaviors in the SLR, it is rarely reported for this kind of BMGCs until now. In this paper, the Ti_35.7Zr_35.6Cu_8.3Be_20.4 (in atomic percentage, at.%) alloy with metastable β-Ti dendrites is selected to investigate the deformation behavior and microstructure evolution in the SLR under tension. Homogeneous flow with superplasticity is present at the elevated temperature, and the transition from work softening to hardening will be discussed by considering the crystallization behavior and the interaction between the amorphous matrix and dendrites in the BMGC. This finding provides a fundamental guideline and reference for the formability of the BMGCs in the supercooled liquid region.

2. Material and methods

The master ingots of Ti_35.7Zr_35.6Cu_8.3Be_20.4 alloy were prepared by arc melting the four high-purity elements (purity of all elements is over 99.8 wt%) under a Ti-gettered argon atmosphere. The master ingots were re-melted four times to avoid the chemical inhomogeneity. Then the as-cast BMGC plates with 80 mm in length, 10 mm in width, and 6 mm in thickness were fabricated by arc re-melting master ingots in a water-cooled copper crucible and copper mold tilting casting technique in the high-purity argon atmosphere. The microstructures of as-cast and fracture specimens after tension were characterized by scanning electron microscopy (SEM, Zeiss Supra 55 and Quanta 600 respectively), X-ray diffraction (XRD, Philips PW 1050) with CuKα radiation and transmission electron microscopy (TEM, FEI F20). The TEM samples of the post-deformation samples were taken from the longitudinal section. Since the longitudinal width of the samples was less than 3 mm, the slices were cut to a length of 3 mm, and after mechanically grinding to a thickness of 50 μm, the slices were sandwiched with a copper ring having a diameter of 3 mm. Then the central portion of the slices was recessed to reduce the center thickness to about 10 μm and ion milled to about 100 nm using a Gatan 691 device with liquid nitrogen cooling. The glass transition temperature T_g, onset crystallization temperature T_x and crystallization enthalpy △H of the alloy were characterized by the differential scanning calorimetry (DSC, Netzsch 204F1) under the high-purity argon atmosphere at the heating rate of 20 K/min. The tensile specimens were cut into the bone-like specimens with a gauge length of 15 mm and a cross-section dimension of 2 mm × 1 mm from as-cast plates. The lateral of samples in both sides were polished to parallel to the opposite side. Uniaxial tension tests were carried out using an electronic universal testing machine (MTS E45) at the constant strain rate of 1 × 10^-3 s^-1 in the SLR without protection atmosphere. The temperature of the testing machine fluctuated ±1 K, and the samples were heated to the testing temperature at a heating rate of 40 K/min and held in the furnace for 3 min before testing to ensure that the temperature of each part of the samples was consistent. For comparison, some samples were annealed under the same experimental conditions as the tension tests, such as temperature and holding time.

3. Results and discussion

3.1. Characterization of the as-cast samples

The microstructure and thermal behaviors of the as-cast samples are shown in Fig. 1. The XRD pattern as the inset in Fig. 1(a) shows that four sharp crystalline peaks identified as β-Ti phase are superimposed in the broad diffraction peak, indicating that the BMGC consists of body-centered cubic (bcc) structure β-Ti and the amorphous phase. Fig. 1(a) shows the DSC curve of the as-cast sample, clearly indicating that T_g and T_x of the BMGC are 618 K and 698 K, respectively, and it can be derived that the width of the supercooled liquid region △T= T_x- T_g=80 K. The SEM micrograph of the as-cast alloy is shown in Fig. 1(b). Combined with the XRD pattern, it can be confirmed that the black isolated dendritic phase is the β-Ti phase and the white continuous phase is the amorphous matrix. The spatial distribution of dendrites demonstrates that fine dendrites uniformly surround the coarse dendrites [33]. The dendrites are both uniformly distributed on the glassy matrix. The volume fraction of dendrites is estimated to be about 10%, and the sizes of coarse dendrites and fine dendrites are approximately 20 μm and 1 μm, respectively. The average spacing between the fine dendrites is about 2 μm, which is comparable to the size of the fine dendrites. Since Be is hardly dissolved in the dendrite phase [34], the Ti_35.7Zr_35.6Cu_8.3Be_20.4 alloy is designed by increasing the Ti content and reducing the Cu and Be contents on the basis of the TiZrCuBe BMG, so that the volume fraction and average size of dendrites in the composite are much lower than those in the reported work [29]. The composition of the dendritic phase is comprised of Ti, Zr, and Cu, etc., which is lack of β stabilizing elements, indicating that the dendrites might exhibit martensitic phase transformation during deformation [21].

View original graphic| Download| PPT slide

Fig. 1.   Characterization of as-cast Ti_35.7Zr_35.6Cu_8.3Be_20.4 alloy: (a) the XRD pattern (the inset) and DSC curve; (b) the SEM micrograph.

3.2. Stress‒strain response

To evaluate the viscous flow of the present BMGC samples, the testing temperatures were selected from 623 K to 693 K, which is located at the supercooled liquid region. The temperature gradient is 10 K and the strain rate is a constant of 1 × 10^-3 s^-1. For comparison, the mechanical properties of the BMGC are also tested at room temperature. The alloy has no plasticity at room temperature, and it has a fracture strength of about 1400 MPa and an elongation of 2%.

Fig. 2 shows the tensile curves of the BMGC, including true stress‒strain curves, work hardening rate curves, yield strength curves, and elongation curves. It clearly shows a significant variation of the tensile behavior as the testing temperature increases.

View original graphic| Download| PPT slide

Fig. 2.   Tensile properties of Ti_35.7Zr_35.6Cu_8.3Be_20.4 alloy at different temperatures within the SLR at a strain rate of 1 × 10^-3 s^-1: (a) true stress-true strain curves and work-hardening rate curves; (b) the yield strength curves; (c) the elongation curves; (d) profile photo of the sample deformed in 643 K.

Fig. 2(a) shows the true stress‒true strain curves of the Ti_35.7Zr_35.6Cu_8.3Be_20.4 alloy tensiled at different temperatures in the SLR and the corresponding work-hardening rate curves at each testing temperature. During the rheological stage after yielding, it can be seen from the curves that the value of the work-hardening rate of the alloy gradually increases with the increase of testing temperature. That is, the value of the work-hardening rate gradually transits from less than zero (work softening) to greater than zero (work hardening). Therefore, we may divide the supercooled liquid region into three stages according to the work-hardening rate: the work-softening region (T_g -638 K), the transition region (638-658 K) and the work-hardening region (658 K‒T_x). In the work softening region, the values of work-hardening rate in the rheological stage are constantly less than zero, and the values continuously maintain at a relatively stable magnitude (about -0.5 GPa at 623 K and -0.3 GPa at 633 K) throughout the rheological stage. In the transition region, the values of work-hardening rate are less than zero after yielding and then gradually increase to greater than zero. At these temperatures, the value of the work-hardening rate initially increases the maximum (0.23 GPa at 643 K, 0.8 GPa at 653 K) and then followed by a gradual decrease until final fracture. In the work-hardening region, the values of work-hardening rate are constantly greater than zero. When the temperature is higher than 673 K (including 673 K), the values of work-hardening rate have a rapid increase before the final fracture of the alloy, which may imply the intense microstructural transformation of the BMGC during this time period. A stress overshoot phenomenon is readily observed at low temperatures, in particular at 623 K and 633 K. As the testing temperature gradually increases, the amplitude of the overshoot stress obviously decreases and eventually disappears at testing temperatures, higher than 658 K.

Fig. 2(b) shows the tensile yield strength of the alloy at different temperatures. With increasing testing temperature, the yield strength continuously drops from 638 MPa at 623 K to about 50 MPa at 673 K. When the temperature is above 673 K, the yield strength fluctuates between 30 MPa and 50 MPa. Fig. 2(c) shows the fracture elongation of the samples deformed at the corresponding temperature. The elongation maintains a stable value about 110% in the temperature range of 623 K to 633 K, and then rapidly increases to 266% in the temperature range from 633 K to 643 K. When temperatures are higher than 643 K, the elongation shows a continuous decrease with the increase of testing temperature until 26% at 693 K. At 643 K, which is at the intermediate temperature of the SLR, the alloy reaches the maximum elongation. Fig. 2(d) shows the profile photograph of the sample before and after tensile deformation at 643 K. The sample is uniformly elongated, indicating that the sample exhibits a homogeneous deformation under this experimental condition. It is evident that both the yield strength and fracture elongation are extremely sensitive to the testing temperature. For example, at a testing temperature difference of only 10 K, the yield strength decreases from 430 MPa at 633 K to about 300 MPa at 643 K, but the tensile elongation is more than twice (110% to 266%).

3.3. Characterization of post-deformation samples

3.3.1. DSC measurements

In order to explain the underlying deformation mechanism, we firstly examined the thermal stability of the amorphous matrix by analyzing the crystallization enthalpy (ΔH_crys.) of the post-deformation samples. ΔH_crys. can be used to characterize the volume fraction of the remaining amorphous phase in the samples. For comparison, ΔH_crys. of the samples, which were annealed at the same temperature and same time without tensile deformation was also measured. The corresponding DSC curves are shown in Fig. 3(a) and (b). It can be seen from the DSC curves that ΔH_crys. of the annealed samples is higher than that of the post-deformation samples at each temperature, as shown in Fig. 3(c), which implies that deformation accelerates its crystallization process. In the work-softening region, ΔH_crys. of the post-deformation samples is equivalent to that of the annealed and as-cast samples within the error range, implying that no crystallization occurs in the samples during deformation. With the temperature elevating into the transition range, ΔH_crys. of the post-deformation samples reduced more than that of the annealed samples. However, in the work-hardening region, especially at the temperature above 683 K, ΔH_crys. of the annealed samples was measured to be about 28.4 J/g while that of the post-deformation sample has dropped to 0 J/g. These results show the deformation is helpful to accelerate the crystallization of the supercooled melt. It is attributed to the generation of free volume resulted from deformation, which can enhance the atomic diffusion and further favor the nucleation [35].

View original graphic| Download| PPT slide

Fig. 3.   DSC curves of Ti_35.7Zr_35.6Cu_8.3Be_20.4 alloy: (a) samples deformed in the SLR at a strain rate of 1 × 10^-3 s^-1; (b) annealed samples; (c) enthalpy of the post-deformation samples and annealed samples.

3.3.2. XRD measurements

Fig. 4 shows the XRD patterns of the annealed and post-deformation samples. It can be seen that the crystalline phase still retains the main β-Ti phase and produced the ω-Ti phase in the annealed samples. However, in the post-deformation samples, a distinct α-Ti phase and no obvious ω-Ti phase are identified in the corresponding XRD patterns (Fig. 4(a)). This difference may be attributed to the existence of the deformation-induced phase transformation of β → α in the dendrites, which was confirmed in the other BMGCs containing the 50 vol.%‒60 vol.% dendrites at room temperature [[21],[22],[36]]. It was reported that the deformation-induced phase transformation occurred at the elastic stage of the BMGC samples in compression at room temperature [37]. By comparing the XRD curves of the post-deformation samples and the annealed samples, it can be known that the β → α phase transformation is driven by stress rather than temperature. Meanwhile, the increase of test temperature might reduce the critical stress of β → α phase transformation, so although the flow stress decreases, the β → α phase transformation can also occur at high temperatures. As shown in the XRD curves (Fig. 4(b)), such phase transformation should undergo in the SLR tension of the current BMGC samples. The formation of the α phase may increase the deformation resistance of the specimens, because of the increase of the hardness of the reinforcing phase [38]. However, this kind of phase transformation in the dendrites may be not a major factor in the change of deformation behaviors in the present work. It is believed that the change of deformation behaviors is mainly due to the amorphous phase itself. As marked by the dotted ellipses in Fig. 4, the new peaks can be clearly seen in the XRD curves of the samples deformed at temperature above 673 K. To be associated with the DSC results, it is found out that the amorphous matrix has undergone severe crystallization at these temperatures, which would result in the significant change of deformation behaviors for the current BMGC samples in the SLR.

View original graphic| Download| PPT slide

Fig. 4.   XRD patterns of Ti_35.7Zr_35.6Cu_8.3Be_20.4 alloy: (a) post-deformation samples; (b) annealed samples.

3.3.3. TEM characterization

In order to clarify the microstructural evolution during the tension deformation, TEM characterization is conducted for the samples after deformation at different temperatures. Fig. 5 shows the typical TEM images of the samples deformed at 623 K, 653 K and 683 K, respectively. The TEM result of the as-cast sample is also shown in Fig. 5(a) as a comparison. In the as-cast samples, only the dendrites of β phase and the amorphous phase are observed. The interface between the dendrites and the amorphous matrix is clear and sharp. In the post-deformation samples, a wide variety of different microstructures can be seen in the dendrites and the matrix. In the dendrites, the new α" phase is observed and indicated by selected area electron diffraction (SAED) patterns inset in Fig. 5(a), its size is fine. The corresponding morphology is very different from the lamellar one at room temperature. In the matrix, it was found that the degree of crystallization of the amorphous phase was increased due to the increase of testing temperature, as shown in Fig. 5(b)‒(d). At 623 K and almost no crystalline phase can be visible, but nanocrystals having a size of about 10 nm are seen at 653 K, complete crystallization seen at 683 K. With the occurrence of crystallization, the interface between the dendrites and the matrix varied from clear (Fig. 5(a)) to indistinct (Fig. 5(b) and (c)) to the completely indistinguishable (Fig. 5(d)). It is largely correlated to the interfacial nucleation of either deformation-induced phase transformation or the crystallization of the amorphous phase. All the TEM results are consistent with DSC, XRD ones. A good agreement is present between the microstructural evolution and mechanical behavior variation, which will be discussed in detailed in the following part.

View original graphic| Download| PPT slide

Fig. 5.   TEM images of the as-cast sample (a) and the samples deformed at different temperatures: (b) 623 K; (c) 653 K; (d) 683 K.

3.4. Deformation mechanism

As mentioned above, the present BMGC is comprised of dendritic phase and amorphous matrix, both of which are metastable thermodynamically. Accordingly, the mechanical properties of the present BMGC are significantly influenced by the stability of the constituents, which is closely dependent on the testing conditions, including temperature, time and loading, etc. In the present work, the current BMGC samples exhibit super-high homogeneous tensile strain in the SLR, and a transition from work softening to hardening as the testing temperature increases (Fig. 2). Although the microstructural characterizations indicate the binary constituents lose their stability in the deformation, the crystallization of the amorphous phase plays a more important role in the SLR deformation behavior of the current BMGC.

Firstly, the deformation-induced β → α" phase transformation indeed occurred in the samples at all the testing temperatures. It is indicated by comparing the phase composition of the annealed and post-deformation samples. In general, such phase transformation contributes to enhancing the work-hardening capacity of the BMGC alloys at room temperature. However, because the current alloy only contains ∼10 vol.% dendrites and the dendrites are fine, the enhancing effect of deformation-induced β → α" phase transformation is very limited in comparision with the softening of the amorphous matrix in the SLR. It is different from the mechanical behaviors of BMGCs containing 50 vol.%‒60 vol.% dendrites.

Secondly, it is found that there is a good agreement between the work-hardening behavior and the crystallization of the amorphous phase for the current BMGC. For example, the samples exhibit a rapid decrease in the crystallization enthalpies as the testing temperature increased from 663 K to 673 K (Fig. 3). Correspondingly, the work-hardening rate exhibits similar variation trend at 663 K and 673 K when the strain is less than 0.6, but the work-hardening rate gradually decreases at 663 K, and increases rapidly at 673 K, respectively, after the strain reaches 0.6. It can be inferred that the rapid increase in the work-hardening rate at 673 K is caused by intense crystallization of the amorphous phase. The crystallization products are usually strong and brittle, which plays as an additional reinforcement. With progressing the crystallization, the volume fraction of the crystallite phase increases, which results in an increase in flow stress, i.e., the strain hardening in the current BMG composite. A similar phenomenon in BMGs has also been discussed in Ref. [7]. Three regions are classified according to the dependence of deformation behaviors on temperature (Fig. 2). It is probably determined by the crystallization kinetics. At the relatively low temperature, the atomic diffusion is quite slow so that the crystallization hardly proceeds at the deformation time. At this time, the sample exhibits work-softening behaviors. As the testing temperature increases, i.e., in the transition region, the atomic diffusion enhances, and a small amount of crystallization occurs with the deformation progresses, leading to the work softening firstly, and then work hardening. In the work-hardening region (above 663 K), the atomic diffusion is high so that the crystallization probably occurs from the beginning of deformation. Correspondingly, the work-hardening phenomena are present throughout the deformation process, and the work-hardening rate increases rapidly with the increase of the strain. In addition to the testing temperature, the applied stress is an important factor favoring the crystallization of the amorphous phase [39].

This explanation is also confirmed by fracture morphology. Fig. 6 shows the fracture surface of the samples deformed at 623 K, 653 K and 683 K. Post-deformation samples exhibit normal tensile fractures at 623 K, 653 K and 683 K, and no obvious shear fractures. Post-deformation samples exhibit up to 98% necking at 623 K and 653 K. The vein pattern exists at 623 K and 653 K, while the vein pattern disappears and a cleavage-like pattern is revealed at 683 K. Obvious voids are observed on the tensile fracture surfaces, which are recognized as the aggregation of free volume under the triaxial stress prior to the final fracture [40]. The fracture morphology in the work-softening and transition regions is similar to that of the monolithic amorphous alloy [32]. However, the existence of dendrites results in the appearance of obvious streaks parallel to the direction of the tensile stress on the profile of the specimens, which is quite different from those of the monolithic BMGs [32] and BMGCs with a high-volume fraction of dendritic phase [29]. There are no obvious streaks on the profile for BMGs and distinct micro-cracks formed by the separation of the dendrites and the matrix for the latter one, respectively. Due to the low volume fraction and the fine dendrites in the current BMGC, the dendrites hardly block during the rheological process, so no obvious micro-cracks formed by the separation of the matrix and dendrites are observed. In the work-hardening region, the intense crystallization occurs, which leads to changes at the fracture time in tensile deformation behavior and finally produces the cleavage-like fracture morphology. It is also indicated that the specimens are very brittle in the final deformation due to the crystallization, as shown in Fig. 2(a).

View original graphic| Download| PPT slide

Fig. 6.   Macroscopic morphology of the sample tested at a strain rate of 1 × 10^-3 s^-1: (a) 623 K; (b) 653 K; (c) 683 K.

4. Conclusion

The deformation behavior and microstructure evolution of the Ti_35.7Zr_35.6Cu_8.3Be_20.4 BMGC at different temperatures in the SLR were investigated by the uniaxial tensile experiment. The current BMGC alloy exhibits a homogeneous flow and superplasticity, which are different from those of the monolithic BMG and BMGCs containing high volume fraction dendrites. The maximum elongation reaches 266% at 643 K. As the testing temperature increases, the deformation changes from work-softening mode to hardening one, which is closely related to the crystallization of amorphous phase rather than deformation induced β → α" phase transformation of dendrites. It is revealed that the tensile stress favors its crystallization during deformation. The crystallization of amorphous phase results in the work-hardening behaviors, and intense crystallization leads to the alloy becoming brittle. These findings are beneficial to designing new BMGCs with high mechanical performance as well as the good SLR formability.

Acknowledgements

This work was supported financially by the National Natural Science Foundation of China (Nos. 51790484, 51434008 and 51531005), the National Key Research and Development Program (No. 2018YFB0703402), Dong Guan Innovative Research Team Program (No. 2014607134), and Shenyang Amorphous Metal Manufacturing Co., Ltd..