- Email: [email protected]
Abnormal devitrification behavior and mechanical response of cold-rolled Mg-rich Mg-Cu-Gd metallic glasses
Acta Materialia 116 (2016) 238e249
Contents lists available at ScienceDirect
Acta Materialia journal homepage: www.elsevier.com/locate/actamat
Full length article
Abnormal devitrification behavior and mechanical response of cold-rolled Mg-rich Mg-Cu-Gd metallic glasses J.I. Lee a, J.W. Kim a, H.S. Oh a, J.S. Park b, E.S. Park a, * a b
Research Institute of Advanced Materials, Department of Materials Science and Engineering, Seoul National University, Seoul 08826, Republic of Korea Department of Materials Science & Engineering, Hanbat National University, Daejeon 34158, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history: Received 4 January 2016 Received in revised form 19 April 2016 Accepted 12 June 2016
Abnormal devitrification behavior and mechanical response of Mg75Cu15Gd10 (relatively strong glass former with higher structural stability) and Mg85Cu5Gd10 (relatively fragile glass former with lower structural stability) metallic glasses, fabricated by repeated forced cold rolling, have been investigated. When metallic glasses were cold-rolled up to a thickness reduction ratio of ~33%, the heat of relaxation (DHrelax.) below Tg of the cold-rolled specimens was reduced, which indicates the formation of local structural ordering via cold rolling due to stress-induced relaxation. The local structural ordering results in abnormal devitrification behavior, such as higher resistance of glass-to-supercooled liquid transition and delayed growth, in the following heat treatment due to increased nuclei density and pinning site. In particular, the fragility index, m, could assist in understanding structural stability and local structural variation by mechanical processing as well as compositional tuning. Indeed, we examine the shear avalanche size to rationalize the variation of the deformation unit size depending on the structural instability before and after cold rolling. The deformation mode in Mg85Cu5Gd10 metallic glass might change from self-organized critical state to chaotic state by cold rolling, which results in unique hardening behavior under the condition for coexisting well distributed local structural ordering and numerous thinner shear deformed areas. These results would give us a guideline for atomic scale structural manipulation of metallic glasses, and help develop novel metallic glass matrix composites with optimal properties through effective mechanical processing as well as heat treatment. © 2016 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Metallic glass Cold rolling Structural instability Abnormal devitrification Mechanical response
1. Introduction Metallic glasses (MGs) exhibit local plastic deformation, suggesting that strong metallic bonding remains even though there is no long-range order (or translational periodicity) of constituting elements. It is of interest to know whether the atomic arrangement during deformation undergoes a change towards more crystallinity by relaxation phenomena or towards greater randomness by shear deformed area. The general consensus is that elemental process of deformation in the MGs is a local rearrangement of atoms to accommodate local strains incurred through the redistribution of free volume and the operation of shear transformation zone, which can lead to nucleation and propagation of shear bands [1], although there has been considerable debate as to whether shear localization in MGs is mainly due to thermal effects or shear-induced
* Corresponding author. E-mail address: [email protected] (E.S. Park). http://dx.doi.org/10.1016/j.actamat.2016.06.026 1359-6454/© 2016 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
disordering such as dilation. On the other hand, the roomtemperature mechanical deformation of MGs with structural instability can induce local structural ordering with relatively lower energy levels by promoting a so-called structural relaxation due to annihilation of frozen-in “defect”, especially in the shear bands with increased atomic mobility [2]. Recent investigations of the structural variation in MGs via severe deformation process [3e7] have shown that partially devitrified MGs with fine nanocrystals embedded in the shear bands or the amorphous matrix exhibit controllable microstructures and mechanical properties. Thus, it has been understood that deformation of MGs may be a way of generating unique nanostructures with denser and softer regions by annihilation and creation of free volume as a microstructural response, which is not a simple alternative to thermal annealing. Although it appears that the devitrification process of MGs via simple heat treatments is relatively well understood, the correlation between deformationeinduced structural change and following devitrification behavior of the MGs has not been fully understood [1,7e9].
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
Among existing previous studies of the question, deformationinduced structural changes and following devitrification behavior of Al-based MGs have been relatively well defined in severe deformation environments such as tensile/compressive tests or rolling process and the following heat treatment [8,9]. For example, the presence of quenched-in-nuclei of alpha Al in an amorphous matrix can enhance local structural ordering in severe deformation environments, especially under compressive force [8], and minimize the size of precipitates after 1st crystallization behavior in the following heat treatment. However, other glass-forming systems such as Zr- [5,10], Fe- [8], Cu- [11] or Mg-based [12] MGs often show different responses to applied deformation. For example, some of them did not exhibit a consistent devitrification process after coldrolling, implying that the applied deformation on MGs does not necessarily exhibit a similar structural change at room temperature for all MGs. Also, the trends of the DSC peak shift (or reduction of crystallization enthalpy) were not consistent even at a similar amount of deformation [9,12], suggesting that the crystallization behavior in the following heat treatment after cold rolling might be closely dependent on structural instability of MGs [13e16]. In this regard, it is useful to scrutinize the devitrification behavior by cold rolling and following heat treatment in various glass-forming systems, which allows manipulation of the nanostructure as well as the structural instability of MGs. However, very little attention has been focused on the local structural variations in the relaxation regime and mechanical responses of MGs with different levels of structural instability via cold rolling. In the present study, two Mg-Cu-Gd glass-forming alloy compositions [17e20] on either side of the border of bulk metallic glass (BMG)-forming alloy compositions have been selected; Mg75Cu15Gd10 with a GFA of 2 mm (relatively strong glass former) and Mg85Cu5Gd10 with a GFA of less than 1 mm (relatively fragile glass former). In order to identify the devitrification behaviors and mechanical responses of the MG ribbons depending on cold rolling and the following heat treatment, the selected MG ribbons were investigated with a Transmission Electron Microscope (TEM) equipped with heating facilities both before and after cold rolling together with DSC analyses and nanoindentation test.
239
propagation of shear bands. The thickness of the ribbon was reduced from ~30 mm to ~20 mm (~33% in thickness reduction) after cold rolling. The structures of the samples were confirmed by X-ray diffraction (XRD: Rigaku CN2301) for the as-spun and cold-rolled ribbon samples using monochromatic Cu Ka radiation. The thermal history of the as-spun and cold-rolled specimens was investigated by differential scanning calorimetry (DSC; Perkin Elmer DSC8500) using various heating rates between 5 K/min and 80 K/ min. The microstructures of the MGs were examined by TEM (JEOL 300KVJEM-3011) with single tilting heating holder (Gatan model 628). To evaluate devitrification behavior of the Mg-based MGs, the holder was quickly heated up to 423 K and was kept at the temperature for 10 min. Then the temperature was increased up to ~523 K with a 10 K interval. To stabilize the temperature at each step, we kept the temperature at each step under isothermal condition for 3 min. To minimize thermal drift, a water cooling of the outer part of the sample holder was installed. The thin foil TEM specimens were prepared by ion milling (Gatan 695 PIPS II) with liquid nitrogen cooling after mechanical thinning process. Extreme care was taken with the HRTEM analysis since Mg-based MG thin foils oxidize readily upon exposure to air. And, the nanoindentation tests were performed using a nanomechanical tester (Hysitron TI 750 TriboIndenter). Indenting force was loaded up to a maximum load of 5 mN using a conical type indenter with a 2 mm radius by load control mode at a constant loading rate of 1 mN/s.
3. Results
2. Experimental
The Mg65Cu25Gd10 alloy exhibits significantly improved GFA with at least 8 mm in diameter (Dmax) by conventional Cu-mold casting method in air atmosphere [17]. The critical cooling rate for glass formation (Rc) was estimated to be approximately 10 K/s by the following equation: log Rc ¼ 2.52 logZmax þ 3.27 [22]. Fig. 1 shows a map of Dmax with border line of BMG formation in Mg100xCuxGd10 alloys (x ¼ 5e30 at.%). As shown in Fig. 1, with increasing Mg contents up to 85 at.%, the Dmax gradually deceases down to 0.5 mm (Rc ~ 104 K/s [22]). In particular, the border of BMGforming alloy composition is located between Mg80Cu10Gd10 and
A Cu-Gd master alloy was prepared by arc melting Cu and Gd (purity > 99.9%) under a Ti-gettered argon atmosphere in a watercooled copper crucible. The master alloy was then alloyed with Mg (99.9%) in a boron nitride (BN) coated graphite crucible under a dynamic Ar atmosphere using an induction furnace. The alloy ingots were melted several times to help improve compositional homogeneity. After complete melting, the liquid alloy was poured into a Cu mold in air. The copper mold was cone-shaped, 45 mm in height, 15 mm in diameter at the top, and 6 mm in diameter at the bottom. Rapidly solidified ribbon specimens were prepared by remelting the alloys in quartz tubes with over-pressure of 50 kPa through a nozzle onto a Cu wheel rotating with a surface velocity of 40 m/s. To exclude structural relaxation of Mg-Cu-Gd MGs at roomtemperature (corresponding to a homologous temperature T/Tg of about 0.7) like Mg-Cu-Y MGs [21], all experiments were done after 10 days to exclude room temperature relaxation effect. The amorphous ribbons were cold-rolled by a twin roll in order to examine the structural bias of severe deformation. The rolling process was performed by electric motor-controlled twin roll with a diameter of 20 cm at a constant speed of ~2.0 radians/sec. A single amorphous ribbon sandwiched by two clean stainless steel plates was repeatedly cold rolled. The thicknesses of the amorphous ribbons were measured after final pass. After rolling, the specimen was elongated towards the rolling directions due to the formation and
Fig. 1. Map of maximum diameter (Dmax) for glass formation with border line of BMG formation in Mg100xCuxGd10 alloys (x ¼ 5e30 at.%).
240
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
Mg85Cu5Gd10 alloy compositions. Although it is well known that most Mg-based BMGs exhibit extreme brittleness, the ribbon sample of Mg-rich alloy compositions over 75 at.% Mg can be folded 180 , and it is possible to apply severe plastic deformation to it by cold rolling without fracture. Thus, in the present study, both Mg75Cu15Gd10 (BMG former with Dmax ¼ 2 mm and Rc ~ 3.25 102 K/s) and Mg85Cu5Gd10 (MG former with Dmax ¼ 0.5 mm and Rc ~ 104 K/s) alloy composition have been selected to scrutinize the devitrification behavior and mechanical response after cold rolling (hereafter, the Mg75Cu15Gd10 and Mg85Cu5Gd10 are called as Mg75 and Mg85, respectively). The formation of an amorphous phase was confirmed from the broad halos of XRD profiles in as-spun Mg75 and Mg85 ribbons (not shown). There are no clear differences between XRD profiles before and after cold-rolling after thickness reduction up to 33% of both ribbons. Fig. 2 shows DSC traces obtained from as-spun and coldrolled Mg75 (a, c) and Mg85 (b, d) ribbons at a constant heating rate of 20 K/min. The DSC traces of Fig. 2 (a) and (b) show that there were no clear differences in the traces and heat of crystallization (DHcryst.) after thickness reduction up to 33% by cold rolling, indicating that the crystallization enthalpies of the amorphous ribbons were not critically affected by applied deformation, which is a similar observation to the case of Cu-Zr-Al BMGs [11]. However, the enlargement of the marked area in Fig. 2 (a) and (b) shows a difference of relaxation enthalpies between as-spun and cold-rolled ribbons as shown in Fig. 2 (c) and (d), and the heat of relaxation
(DHrelax.) of Mg75 and Mg85 was reduced down to 68% and 42%, respectively. Van den Buekel et al. [23] argued that the change in relaxation enthalpies of MGs during DSC experiments at a constant heating rate is due to the variation in excess free volume, implying that a local structural ordering with low energy levels occurred during cold rolling by stress-induced relaxation in both ribbons. Although both ribbons formed multiple shear bands after cold rolling, it is also difficult to observe increase of DHrelax. due to rejuvenation behavior, which can arise due to high pressure torsion of Zr-Cu-Al BMGs [24]. The increased local structural fluctuation of both Mg75 and Mg85 after cold-rolling can be indirectly seen from the comparison of the effective activation energy for glass transition and crystallization (ET ¼ Eg, Ex and Ep) evaluated using Kissinger’s equation [25] for both as-spun and cold-rolled ribbon samples:
. ln TT2 F ¼ lnðET =kB K0 Þ þ ET =kB TT
(1)
where kB is the Boltzmann constant and K0 is the frequency factor in the Arrhenius law KT ¼ K0exp(ET/kBTT) as shown in Fig. 3. Table 1 summarizes TTs (¼ Tgs, Txs and Tps) at different heating rates (5e80 K/min) for as-spun and cold-rolled Mg75 and Mg85 ribbons. Fig. 3 shows Kissinger plot of as-spun and cold-rolled ribbons of (a, c) Mg75 and (b, d) Mg85, which exhibits a linear relationship between ln TT2 =F and 1000/TT for as-spun and cold-rolled Mg75 and Mg85. The activation energy plotted by Tg (Fig. 3 (a), (b)) for the
Fig. 2. DSC traces obtained from as-spun and cold-rolled Mg75 (a, c) and Mg85 (b, d) ribbons at a constant heating rate of 20 K/min.
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
241
Fig. 3. Kissinger plot of as-spun and cold-rolled ribbons of (a, c) Mg75 and (b, d) Mg85, which exhibits a linear relationship between ln T2T/F and 1000/TT for TT (Tg and Tx) variations depending on heating rates of 5e80 K/min.
Table 1 Glass transition temperature, Tg, crystallization onset temperature, Tx, and crystallization peak temperature, Tp at different heating rates (HR ¼ 5e80 K/min) for as-spun and cold-rolled Mg75 and Mg85 ribbons. Composition (at.%) Mg75Cu15Gd10
As-spun
Cold-rolled
Mg85Cu5Gd10
As-spun
Cold-rolled
HR (K/min)
Tg (K)
Tx (K)
Tp (K)
5 10 20 40 80 5 10 20 40 80 5 10 20 40 80 5 10 20 40 80
411.8 415.3 419.2 424.8 431.7 413.7 416.8 421.1 426.3 432.5 421.4 423.3 429.0 438.9 446.5 424.1 426.0 432.5 440.0 447.8
441.5 444.9 449.0 454.1 461.1 441.7 444.9 449.4 454.4 460.7 447.5 451.5 455.4 459.9 466.1 447.7 451.6 455.8 460.7 466.1
444.7 448.2 452.6 457.9 466.0 444.9 448.4 452.8 457.9 465.2 449.5 453.8 458.2 463.5 471.1 449.7 454.1 458.4 463.8 470.5
cold-rolled ribbon of Mg75 and Mg85 (Egs ¼ 191.0 and 143.2 kJ/mol, respectively) exhibits a higher value compared with those for the Egs of as-spun ribbons (177.2 and 128.5 kJ/mol, respectively), which means that the cold rolling process increases the resistance of glass-to-supercooled liquid transition due to structural bias from forced deformation. However, the activation energy plotted by Tx
(Fig. 3 (c), (d)) and Tp (not shown) for the Mg75 and Mg 85 ribbons shows that the Exs (216.2 and 243.8 kJ/mol) and Eps (207.0 and 215.4 kJ/mol) of both cold-rolled ribbons exhibited similar values compared with Exs (209.1 and 240.1 kJ/mol) and Eps (193.5 and 205.1 kJ/mol) of the as-spun ribbons. Indeed, the discrepancy between Eg and Ex (or Ep) might result in unique devitrification
242
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
behavior during heat treatment after cold rolling. In general, experimental and computational evidences indicated that the slowdown of glass transition is related to the growth of distinct relaxing domains. Thus, in case of MGs with heterogeneity, the relaxing domains can be easily strengthened by mechanical treatment due to an increase in the entropy of fusion in the system by the generated heterogeneity. However, the microstructural imaging by HRTEM in the present study does not give any noticeable difference between as-spun and cold-rolled ribbon samples of both compositions. Fig. 4 shows the HRTEM image obtained from coldrolled Mg85, exhibiting a homogeneous contrast fluctuation, characteristic of a fully amorphous structure. On the other hand, the fast Fourier-transformed (FFT) pattern of Fig. 4 shows some weak spots and the selected area electron diffraction (SAD) pattern has indistinct glow outside of primary amorphous ring, both of which mean the possibility of presence of local ordering, while the local ordering is not clearly discernible in the HRTEM image. Although the structure of these MGs is still amorphous phase under resolution of HRTEM, it should be noted that the structure variation of these MGs can be conspicuously revealed by thermal motion as shown in Fig. 3. Thus, we carefully evaluated the microstructure to confirm a strong influence on the degree of local structural ordering in these MGs due to stress-induced relaxation. Because of the local heterogeneity under nanometer scale in these MGs, it is necessary to get local quantitative information on the microstructural parameters. Fig. 5 shows diffraction profiles (intensity vs. scattering vector, q) as deduced from an SAD pattern of as-spun and cold-rolled Mg75 and Mg85 ribbons using a profile analysis of the SAD pattern (PASAD)-tools [26]. The PASAD is a very powerful method for quickly analyzing nanocrystalline materials quantitatively on a local scale by correlating the results with regions specifically selected by TEM images. The resulting data facilitate a complete quantitative analysis of the sample on a local scale (probe volume~0.001 mm3). The profiles of cold-rolled Mg75 and Mg85 ribbons clearly show a shoulder in the right side of the halo pattern, which is clear evidence for the local structural ordering. Although Mg85 ribbon, relatively fragile glass former
Fig. 5. Diffraction profiles converted from an electron diffraction pattern of as-spun and cold-rolled Mg75 and Mg85 ribbons using PASAD-tools.
with lower structural stability, exhibits small shoulder before cold rolling due to room temperature annealing (corresponding to a homologous temperature T/Tg of about 0.7), the shoulder’s intensity slightly increases after cold rolling due to stress-induced relaxation. These results also explain why the reduction ratio of DHrelax. in Mg85 after cold rolling is smaller than that of DHrelax. in Mg75. Various properties of MGs are closely related to their atomic structure; hence knowledge of the atomic structure and its stability is needed to understand their properties. Fig. 6 shows (a) hardness and (b) load-displacement curves of as-spun and cold-rolled Mg75 and Mg85 ribbons measured by nanoindentation test. For the Mg75, the hardness of cold-rolled ribbon (3.52 ± 0.08 GPa) exhibits lower value compared with that of as-spun ribbon (3.80 ± 0.08 GPa), which means that relatively thick shear deformed area and localized clustering in the deformed area in Mg75 mainly cause softening effect. However, for the Mg85, the hardness of cold-rolled ribbon (3.39 ± 0.07 GPa) exhibits slightly higher value compared with that of as-spun ribbon (3.34 ± 0.07 GPa), which means that relatively thin shear deformed area and homogeneously distributed clustering by plastic deformation in Mg85 mainly cause hardening effect. Under nanoindentation, plastic shearing can be stabilized due to the confinement from the surrounding material even in brittle MGs without any compressive plasticity, which can be useful for analysis of the characteristics of the intermittent shear avalanches in most MGs. The load-displacement curves of Mg75 exhibit relatively significant pop-in events in the loading stage than those of Mg85 as shown in Fig. 6 (b). The results show clear difference in the serrated flow between Mg75 and Mg85 due to different shear banding process by accumulation of elastic energy and stress relaxation. In particular, it can be understood that the serrations of Mg 85 have smaller amplitude and larger numbers than those of Mg75, which means that the size of strain bursts and shear avalanche in Mg85 can be smaller than those of Mg75 due to different structural stability depending on fragility, which will be carefully discussed in Section 4.2. 4. Discussion
Fig. 4. HRTEM image, FFT pattern (inset, left) and SAD pattern (inset, right) obtained from cold-rolled Mg85.
It was reported that anelastic deformation and annealinginduced relaxation in MGs can be separated into reversible and
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
243
relationship:
i 2 h ln 10 Tg Tg0 m ¼ D* Tg0 Tg
Fig. 6. (a) Hardness and (b) load-displacement curves of as-spun and cold-rolled Mg75 and Mg85 ribbons.
irreversible components. This has been explained in terms of chemical short-range ordering (CSRO) and topological short-range ordering (TSRO), respectively. However, such a sharp separation between TSRO and CSRO, criticized by Gibbs and Sinning [27] and Khonik et al. [28], is probably too simple to explain such a complex phenomenon, because CSRO is unlikely without an accompanying TSRO. Although it is also difficult to separate TSRO and CSRO in Mgrich Mg-based MGs via room temperature relaxation and stressinduced relaxation, it is clear that the local structural and chemical bias can result in different mechanical responses as well as devitrification behavior depending on the degree of local structural change.
4.1. Abnormal devitrification behavior via local structural ordering The stability of amorphous structure in MGs can be quantitatively assessed by the fragility index, m. The stronger the liquid becomes (lower m), generally the higher the amorphous stability is, which is related to higher GFA. Therefore, the question that arises is how both composition tailoring and severe plastic deformation application can have such marked effects on the fragility from the microstructural point of view. The kinetic fragility for as-spun and cold-rolled Mg75 and Mg85 has been measured using the following
(2)
where D* and Tg0 are the fitting parameters which can be obtained from the plot of ln F (where F is the heating rate) vs. Tg. The details of Eq. (2) are described elsewhere [29]. Fig. 7 shows the variation of Tg for as-spun and cold-rolled Mg75 and Mg85 as a function of the heating rate F. Inserted table in Fig. 7 shows ln A, D* and Tg0 in the best fit to the experimental data and m value evaluated at a heating rate of 5 K/min by using Eq. (2). The m value of as-spun Mg75 (40.9 ± 2) is smaller than that of the as-spun Mg85 (42 ± 2), which means that as-spun Mg75 is relatively strong glass former with higher amorphous stability than as-spun Mg85, reflecting higher GFA. On the other hand, the m value of cold-rolled Mg75 (40.7 ± 2) and Mg85 (36 ± 2) is smaller than that of the as-spun Mg75 and Mg85, respectively, which means that the cold rolling enhances the glassy slow dynamics with increasing anisotropy of the interatomic potential by changing combinations of bonding energy toward local structural ordering. Interestingly, Mg85 with higher m exhibits larger reduction of m after cold rolling due to higher structural bias toward local structural ordering by stress-induced relaxation compared with Mg75 with lower m. Indeed, the structural variation by plastic deformation as well as structural stability can be correlated with fragility of MGs. Especially, the m values exhibit stronger correlation with atomic structure variation by severe plastic deformation than amorphous stability variation by GFA change in Mg75 and Mg85. Thus, it can be suggested that the m value could assist in understanding local structural variation as well as amorphous stability in MGs and therefore be regarded as an indicator of the atomic scale structural variation of MGs during mechanical processing. To check the qualitative variation of crystallization behavior via local structural heterogeneity, we observed microstructural variation during heating. Fig. 8 shows the SAD patterns at different temperatures observed during heating by TEM with single tilting heating holder as well as corresponding DSC trace obtained from as-spun Mg85 ribbon at a constant heating rate of 20 K/min. Although the thermal history between TEM results and DSC trace does not exactly match (heating rate of TEM is much slower than that of DSC), SAD pattern at 300 K already shows vaguely local sharp rings with diffuse ring of amorphous matrix due to structural ordering caused by room-temperature annealing as shown in Fig. 5. However, the high resolution TEM image of as-spun Mg85 ribbon at 300 K shows a typical amorphous structure with a lattice irregularity like Fig. 4 for cold-rolled Mg85. With increasing temperature, this sharp ring pattern becomes clearer up to 473 K and the spot patterns in SAD pattern, which are related to the grain growth during crystallization, start to appear from 483 K just before the finishing temperature of first crystallization. Fig. 9 shows TEM bright field (BF) images and SAD patterns of the as-spun Mg75 and Mg85 ribbons after heating up to 503 K completely after the first crystallization peak in TEM. As shown in Fig. 9 (a) and (b), as-spun Mg85 exhibits smaller average grain size (Davg ~ 70 nm) after annealing than as-spun Mg75 (Davg ~ 180 nm), implying that the local structural ordering in Mg85 ribbon by room-temperature relaxation might result in delayed crystal growth due to relatively high nuclei density and pinning site. To confirm this idea, we have carefully observed the SAD patterns of as-spun and cold-rolled Mg75 and Mg85 ribbons during heating by TEM with single tilt heating holder (Fig. 10). The spot patterns for as-spun Mg75 and Mg85 ribbons clearly appear at 473 K and 483 K, respectively, which is just before the finishing temperature of first crystallization. However, the spot patterns for cold-rolled Mg75 and Mg85
244
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
Fig. 7. Variations of Tg for the as-spun and cold-rolled (a) Mg75 and (b) Mg85 as a function of the F. Inserted table summarizes ln A, D* and Tg0 in the best fit to the experimental data and m value evaluated at a heating rate of 5 K/min by using Eq. (2).
Fig. 8. SADP patterns obtained from as-spun Mg85 ribbon during heating at (a) 300 K, (b) 443 K, (c) 473 K, (d) 483 K, and (e) 503 K, compared with DSC trace measured at a constant heating rate of 20 K/min.
ribbons appear at 483 K and 493 K, respectively, which is delayed about 10 K compared to those of as-spun ribbons. These results are consistent with TEM BF images after annealing, which mean that local structural ordering by cold rolling as well as room temperature annealing can cause abnormal devitrification behavior of delayed crystal growth. 4.2. Plastic dynamic transition via cold-rolling The deformation of MGs generally occurs inhomogeneously by a viscous-like behavior only at limited areas. The specimen deformed by cold rolling is divided into numerous small blocks by the deformed areas, which can consist of a mixture of shear deformed area and undeformed area, but the specimen as a whole does not undergo homogeneous deformation at room temperature. Of course, the critical thickness and density of the shear deformed area are affected by structural instability of MGs, which can be
inferred from fragility [30], i.e., strong glass former with higher amorphous stability generally forms a thicker and fewer shear deformed areas during deformation than fragile glass former with lower amorphous stability. Indeed, the mixture of MG matrix and shear deformed areas generally reduces Young’s modulus, the fracture stress and the hardness due to so-called softening effect. However, Mg-rich MGs, especially out of the border of BMGforming alloy compositions, also introduced additional regularities through local structural ordering by stress-induced relaxation, which can result in higher hardness and elastic modulus. Thus, distinct effects of plastic severe deformation have been detected by comparing as-spun and cold-rolled Mg75 and Mg85 ribbons by nanoindentation test as shown in Fig. 6. The results show clear difference in the serrated flow as well as stress behavior after yielding between as-spun and cold-rolled Mg75 and Mg85 due to different shear banding process by accumulation of elastic energy and local structural ordering by stress-induced relaxation. In
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
245
Fig. 9. TEM BF images and SADP patterns of the as-spun Mg75 and Mg85 ribbons after annealing up to 503 K using in-situ heating holder.
Fig. 10. SADP patterns obtained from as-spun and cold rolled Mg75 and Mg85 ribbons during heating.
particular, it can be speculated that the serrations of Mg 85 are smaller in amplitude and larger in numbers than those of Mg75, which means that the size of strain bursts and shear avalanche in Mg85 can be smaller than those of Mg75 due to different structural
stability depending on fragility. To clarify this speculation, we carried out a statistical analysis on the strain burst size (S) by considering the distribution of the strain burst which is irregularly and stochastically changing across the as-spun and cold-rolled
246
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
Mg75 and Mg85 ribbons, attempting to acquire a better understanding of the mechanism in the intermittent pop-in events depending on local structural instability [31]. Thus, first we eliminate the influences from the increase of indentation depth (h) in the pop-in events. A polynomial function is used to fit the loading stage in load-displacement curves to get a baseline [Fig. 11 (a), (c)]. After subtraction of the baseline, the pop-in events as a function of the h value are visible. Since each pop-in event reflects a process of shear band formation and propagation, the depth drop, Dh, i.e., height difference between the peak and valley values [as marked in Fig. 11 (b), (d)], can reflect the shear step size. As shown in Fig. 11 (b), (d), the shear step size in cold-rolled Mg75 in much bigger than that of cold-rolled Mg85. However, the instrumental noise could also cause pop-in events, and needs to be removed. The pop-in events free from the background noise can be extracted from the 2s holding segment at peak load through linear fitting, which shows that the noise generates a shear step size of 0.7 nm in the present study. Thus, the pop-in events with shear step sizes less than 0.7 nm were not considered. After removing the noise, normalization of the Dh value by the h is carried out to eliminate the statistical error, which generates S (¼Dh/h). The distributions of the S value versus the h value from as-spun and cold-rolled Mg75 and Mg85 ribbons at a constant loading rate of 1 mN/s exhibit random fluctuations due to normalization process, although the size of distributions seems to decrease a little bit with increasing h. (not shown) Indeed, this deformation unit size is determined by the shear wave propagation distance. Fig. 12 (a)e(d) shows cumulative probability distributions of the S obtained from 10 different indentation curves (over 100 serrations for each sample) of as-spun
and cold-rolled Mg75 and Mg85 ribbons at a constant loading rate of 1 mN/s. The percentage of the number of pop-in events of Mg75 with the S being larger than a given value, P (>S), is clearly greater than that of Mg85 (Fig. 12 (a), (b)). And P (>S) of cold-rolled ribbons with local structural ordering as well as shear deformed area is smaller than that of as-spun ribbons (Fig. 12 (c), (d)). Interestingly, it should be noticed that the difference in P values between as-spun and cold-rolled sample in Mg85 is much higher than that in Mg75. To quantify this comparison using a Levenberg-Marquardt algorithm, the cumulative probability distributions of the as-spun and cold-rolled Mg75 and Mg85 ribbons can be predicted by an empirical relation [32]:
o n Pð > SÞ ¼ ASb exp ðS=Sc Þ2
(3)
where A is a normalization constant, b is a scaling exponent, and Sc is the cut-off of S. For an MG indented at a given loading rate, collective shear band avalanches self organize into a pattern of serrated flow characterized by intermittency with power-law distributions of avalanche sizes. The distribution of S, which is a scale free factor of serrated flow, follows a power law distribution at smaller S region, while a cut-off of this power-law scaling is observed at larger S region of cumulative curve. The fitting parameters, b and Sc, reflect the profile of the shear avalanche in MGs. b is associated with the degree of universality of the power-law relation, which characterizes the tendency of MGs to form jamming state, making cooperative motion of concordant regions [31]. Sc corresponds to the critical size (avalanche size) of strain burst,
Fig. 11. Polynomial function fitting curve of the displacement-load for the loading segment of cold-rolled (a) Mg75 and (c) Mg85 ribbons at a constant loading rate of 1 mN/s, (b) the correlation between Dh and h of cold-rolled (b) Mg75 and (d) Mg85 ribbons showing the serration events.
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
247
Fig. 12. Cumulative probability distributions of strain burst size of as-spun and cold rolled Mg75 and Mg85 ribbons. Open scattering points represent experimental results measured from the nanoindentation. Solid lines are fitting curves by Eq. (3).
Fig. 13. (a) Cut-off values of strain burst size, Sc and (b) scaling exponent, b of as-spun and cold-rolled Mg75 and Mg85 ribbons reflecting the profile of the shear avalanche in metallic glasses. Inserted table summarizes Sc and b values of as-spun and cold-rolled Mg75 and Mg85 ribbons.
overcoming the resistance of the jammed matrix. In other words, Sc can be regarded as a deformation unit size when the deformed area
grows into individual shear band showing distinct behavior from the matrix. The relationship among b, Sc and the corresponding
248
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
shear banding behavior was also investigated by analyzing compressive stress-strain curves of Zr-based MG samples having brittle to ductile transition [33]. Large b and Sc resulted in thicker and fewer shear bands, while small b and Sc resulted in thinner and numerous shear bands. Fig. 13 shows the variation of Sc and b in asspun and cold-rolled Mg75 and Mg85 ribbons. As shown in Fig. 13 (a), the Sc of as-spun Mg 85 (relatively fragile glass former with lower structural stability) is much smaller than that of as-spun Mg75 (relatively strong glass former with higher structural stability). The result implies that Mg75 has a relatively large shear avalanche size (or deformation unit size), which is closely related to its thicker shear band during plastic deformation compared with Mg85. Interestingly, there is no clear change of Sc values before and after cold-rolling in Mg 75, which means that the local structural ordering mainly in the shear bands as well as localized (thicker and fewer) shear deformed areas does not clearly influence shear avalanche size (or deformation unit size). On the other hand, the Sc value of Mg 85 distinctly decreases after cold rolling, which implies that well distributed local structural ordering as well as (thinner and many) shear deformed areas may reduce shear avalanche size (or deformation unit size). Thus, it can be understood that in coldrolled Mg85 shear bands are initiated by continuously rendering the imposed strain and cutting the elastic energy accommodation, which results in smaller amplitude of a pop-in event and numerous shear bands with smaller deformation unit size to dissipate the plastic strain. Furthermore, the cumulative probability distributions of the S in Mg75 clearly show the significant power-law relation in a wide range of S (¼ larger b), suggesting a selforganized critical state in the dynamics of the collective shear banding. This provides solid evidence to support the idea that the large rigidly packed deformation units formed a jamming state to resist small perturbations [34], approaching a stable state, i.e., a non-chaotic state in Mg75 during deformation. On the other hand, with a decrease in the deformation unit size (¼smaller Sc) in Mg 85, the rigidly packed deformation units are smaller in size (¼smaller b), which means that the jammed state of the deformation units is weakened. Thus, it is understood that the transition from jamming to unjamming states may occur in Mg85. The un-jamming state is unstable to perturbations, exhibiting a chaotic state [35], which is manifested in the lack of power-law relation in the statistical distribution of strain bursts due to the influence from smaller Sc values, where the stress perturbations make it difficult to develop jamming state and finally result in successive formation of multiple and adjacent deformation bands at lower strain rate. In particular, the decrease in b after cold rolling in Mg85 is larger than that in Mg75 and, especially, the b value in cold-rolled Mg85 is close to 0, which implies that the deformation mode might change from a complex mode between self-organized critical state and chaotic state in as-spun Mg85 to chaotic state in cold-rolled Mg85 due to well distributed local structural ordering as well as numerous thinner shear deformed areas. Indeed, Mg85 after cold rolling exhibits unique hardening behavior under the condition for coexisting well distributed local structural ordering and numerous thinner shear deformed areas. Thus, it can be concluded that effective mechanical processing of MGs with different degrees of structural instability (fragility) can manipulate properties of MGs even without clear microstructural features via variation of shear avalanche size (or deformation unit size). Indeed, there are fruitful prospects for future improvements of the performance of MGs to form a composite with unique nanostructures via effective mechanical processing as well as heat treatment. 5. Conclusions Abnormal devitrification behavior and mechanical response
have been investigated in as-spun and cold-rolled Mg75 (relatively strong glass former with higher structural stability) and Mg85 (relatively fragile glass former with lower structural stability) MG ribbons. The structural instability after cold rolling up to a thickness reduction ratio of ~33% causes abnormal devitrification behavior in the following heat treatment: higher Egs and delayed growth due to increased nuclei density and pinning site. In particular, the m values, an indicator for structural stability, exhibit relatively stronger correlation with atomic structure variation by cold rolling than that by compositional tuning in Mg75 and Mg85. Indeed, Sc of as-spun Mg85 is much smaller than that of as-spun Mg75 and distinctly decreases after cold-rolling, which implies that the increased structural instability may reduce shear avalanche size (or deformation unit size). Also, with decreasing Sc in Mg85, the rigidly packed deformation units are smaller in size (closely related to smaller b), which means that the jammed state of the deformation units is weakened. Especially, the b value in cold-rolled Mg85 is close to 0, which implies that the deformation mode may change from self-organized critical state to chaotic state due to well distributed local structural ordering as well as thinner shear deformed area. Thus, Mg85 after cold rolling exhibits unique hardening behavior due to decrease in the interparticle spacing in SRO precipitates and the higher solute concentration in the remaining amorphous matrix. This result would help deepen our understanding of the importance of atomic scale structural manipulation in MG, which can prove very useful for development of new MG composites with optimal properties through effective mechanical processing as well as heat treatment. Acknowledgments The author’s work in this area was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (Ministry of Science, ICT and Future Planning) (No. 2014K1A3A1A20034841 & 2014M3C1A8053728). One of the authors (E.S. Park) also benefited from the Center for Iron and Steel Research (RIAM) at Seoul National University. References [1] C.A. Schuh, T.C. Hufnagel, U. Ramamurty, Mechanical behavior of amorphous alloys, Acta Mater. 55 (2007) 4067e4109. [2] E.S. Park, Understanding of the shear bands in amorphous metals, Appl. Microsc. 45 (2015) 63e73. [3] D.V. Louzguine, A. Inoue, Influence of a supercooled liquid on devitrification of Cu-, Hf-and Ni-based metallic glasses, Mater. Sci. Eng. A 375 (2004) 346e350. [4] W.H. Jiang, F.E. Pinkerton, M. Atzmon, Mechanical behavior of shear bands and the effect of their relaxation in a rolled amorphous Al-based alloy, Acta Mater. 53 (2005) 3469e3477. [5] J.S. Park, H.K. Lim, J.H. Kim, H.J. Chang, W.T. Kim, D.H. Kim, E. Fleury, In situ crystallization and enhanced mechanical properties of the Zr41.2Ti13.8Cu12.5Ni10Be22.5 alloy by cold rolling, J. Non-crystalline Solids 351 (2005) 2142e2146. [6] J.W. Tian, L.L. Shaw, Y.D. Wang, Y. Yokoyama, P.K. Liaw, A study on the surface severe plastic deformation behavior of a Zr-based bulk metallic glass (BMG), Intermetallics 17 (2009) 951e957. [7] J.H. Perepezko, K.E. Kimme, R.J. Hebert, Deformation alloying and transformation reactions, J. Alloys Compd. 483 (2009) 14e19. €sner, G. Wilde, Dislocation formation during [8] R.J. Hebert, J.H. Perepezko, H. Ro deformation-induced synthesis of nanocrystals in amorphous and partially crystalline amorphous Al 88 Y 7 Fe 5 alloy, Scr. Mater. 54 (2006) 25e29. [9] H.J. Jin, F. Zhou, L.B. Wang, K. Lu, Effect of plastic deformation on thermal stability in metallic glasses, Scr. Mater. 44 (2001) 1083e1087. [10] Y. Yokoyama, K. Yamano, K. Fukaura, H. Sunada, A. Inoue, Bulk metallic glasses. III. Enhancement of ductility and plasticity of Zr55Cu30Al10Ni5 bulk glassy alloy by cold rolling, Mater. Trans. 42 (2001) 623e632. [11] Q.P. Cao, J.F. Li, Y.H. Zhou, J.Z. Jiang, Microstructure and microhardness evolutions of Cu 47.5 Zr 47.5 Al 5 bulk metallic glass processed by rolling, Scr. Mater. 59 (2008) 673e676. [12] J.S. Park, J.M. Kim, E.S. Park, Structural behaviors of single and multiple amorphous alloys deformed by forced cold rolling, Intermetallics 18 (2010) 1920e1924.
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249 [13] B.J. Lee, C.S. Lee, J.C. Lee, Stress induced crystallization of amorphous materials and mechanical properties of nanocrystalline materials: a molecular dynamics simulation study, Acta Mater. 51 (2003) 6233e6240. [14] E.S. Park, J.Y. Lee, D.H. Kim, A. Gebert, L. Schultz, Correlation between plasticity and fragility in Mg-based bulk metallic glasses with modulated heterogeneity, J. Appl. Phys. 104 (2008) 3520. [15] E.S. Park, D.H. Kim, Effect of manipulating atomic scale heterogeneity on plasticity in Mg-based bulk metallic glasses, Intermetallics 18 (2010) 1867e1871. [16] D.H. Kim, W.T. Kim, E.S. Park, N. Mattern, J. Eckert, Phase separation in metallic glasses, Prog. Mater. Sci. 58 (2013) 1103e1172. [17] H. Men, D.H. Kim, Fabrication of ternary MgeCueGd bulk metallic glass with high glass-forming ability under air atmosphere, J. Mater. Res. 18 (2003) 1502e1504. [18] F. Guo, S.J. Poon, G.J. Shiflet, Investigation of glass formability in Al-based multinary alloys, Scr. Mater. 43 (2000) 1089e1095. [19] A. Inoue, K. Ohtera, A.P. Tsai, T. Masumoto, Aluminum-based amorphous alloys with tensile strength above 980 MPa (100 kg/mm2), Jpn. J. Appl. Phys. 27 (1988) L479. [20] A. Inoue, W. Zhang, Formation, thermal stability and mechanical properties of Cu-Zr-Al bulk glassy alloys, Mater. Trans. 43 (2002) 2921e2925. € ffler, Room-tem[21] A. Castellero, B. Moser, D.I. Uhlenhaut, F. Dalla Torre, J.F. Lo perature creep and structural relaxation of MgeCueY metallic glasses, Acta Mater. 56 (2008) 3777e3785. [22] E.S. Park, C.W. Ryu, W.T. Kim, D.H. Kim, A novel parameter to describe the glass-forming ability of alloys, J. Appl. Phys. 118 (2015) 064902. [23] A. Van den Beukel, J. Sietsma, The glass transition as a free volume related kinetic phenomenon, Acta Metall. Mater. 38 (1990) 383e389. [24] F. Meng, K. Tsuchiya, I. Seiichiro, Y. Yokoyama, Reversible transition of deformation mode by structural rejuvenation and relaxation in bulk metallic
249
glass, Appl. Phys. Lett. 101 (2012) 121914. [25] H.E. Kissinger, Reaction kinetics in differential thermal analysis, Anal. Chem. 29 (1957) 1702e1706. [26] C. Gammer, C. Mangler, C. Rentenberger, H.P. Karnthaler, Quantitative local profile analysis of nanomaterials by electron diffraction, Scr. Mater. 63 (2010) 312e315. [27] J. Ding, S. Patinet, M.L. Falk, Y. Cheng, E. Ma, Soft spots and their structural signature in a metallic glass, Proc. Natl. Acad. Sci. 111 (2014) 14052e14056. [28] M. Gibbs, H.-R. Sinning, A critique of the roles of TSRO and CSRO in metallic glasses by application of the activation energy spectrum model to dilatometric data, J. Mater. Sci. 20 (1985) 2517e2525. [29] C.A. Angell, Formation of glasses from liquids and biopolymers, Science 267 (1995) 1924e1935. [30] V.A. Khonik, A.T. Kosilov, V.A. Mikhailov, V.V. Sviridov, Isothermal creep of metallic glasses: a new approach and its experimental verification, Acta Mater. 46 (1998) 3399e3408. [31] X.L. Bian, G. Wang, K. Chan, J.L. Ren, Y.L. Gao, Q.J. Zhai, Shear avalanches in metallic glasses under nanoindentation: deformation units and rate dependent strain burst cut-off, Appl. Phys. Lett. 103 (2013) 101907. [32] T. Richeton, J. Weiss, F. Louchet, Breakdown of avalanche critical behaviour in polycrystalline plasticity, Nat. Mater. 4 (2005) 465e469. [33] C. Wang, B.A. Sun, W.H. Wang, H.Y. Bai, Chaotic state to self-organized critical state transition of serrated flow dynamics during brittle-to-ductile transition in metallic glass, J. Appl. Phys. 119 (2016) 054902. [34] E.J. Banigan, M.K. Illich, D.J. Stace-Naughton, D.A. Egolf, The chaotic dynamics of jamming, Nat. Phys. 9 (2013) 288e292. [35] S. Papanikolaou, D.M. Dimiduk, W. Choi, J.P. Sethna, M.D. Uchic, C.F. Woodward, S. Zapperi, Quasi-periodic events in crystal plasticity and the self-organized avalanche oscillator, Nature 490 (2012) 517e521.
Contents lists available at ScienceDirect
Acta Materialia journal homepage: www.elsevier.com/locate/actamat
Full length article
Abnormal devitrification behavior and mechanical response of cold-rolled Mg-rich Mg-Cu-Gd metallic glasses J.I. Lee a, J.W. Kim a, H.S. Oh a, J.S. Park b, E.S. Park a, * a b
Research Institute of Advanced Materials, Department of Materials Science and Engineering, Seoul National University, Seoul 08826, Republic of Korea Department of Materials Science & Engineering, Hanbat National University, Daejeon 34158, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history: Received 4 January 2016 Received in revised form 19 April 2016 Accepted 12 June 2016
Abnormal devitrification behavior and mechanical response of Mg75Cu15Gd10 (relatively strong glass former with higher structural stability) and Mg85Cu5Gd10 (relatively fragile glass former with lower structural stability) metallic glasses, fabricated by repeated forced cold rolling, have been investigated. When metallic glasses were cold-rolled up to a thickness reduction ratio of ~33%, the heat of relaxation (DHrelax.) below Tg of the cold-rolled specimens was reduced, which indicates the formation of local structural ordering via cold rolling due to stress-induced relaxation. The local structural ordering results in abnormal devitrification behavior, such as higher resistance of glass-to-supercooled liquid transition and delayed growth, in the following heat treatment due to increased nuclei density and pinning site. In particular, the fragility index, m, could assist in understanding structural stability and local structural variation by mechanical processing as well as compositional tuning. Indeed, we examine the shear avalanche size to rationalize the variation of the deformation unit size depending on the structural instability before and after cold rolling. The deformation mode in Mg85Cu5Gd10 metallic glass might change from self-organized critical state to chaotic state by cold rolling, which results in unique hardening behavior under the condition for coexisting well distributed local structural ordering and numerous thinner shear deformed areas. These results would give us a guideline for atomic scale structural manipulation of metallic glasses, and help develop novel metallic glass matrix composites with optimal properties through effective mechanical processing as well as heat treatment. © 2016 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Metallic glass Cold rolling Structural instability Abnormal devitrification Mechanical response
1. Introduction Metallic glasses (MGs) exhibit local plastic deformation, suggesting that strong metallic bonding remains even though there is no long-range order (or translational periodicity) of constituting elements. It is of interest to know whether the atomic arrangement during deformation undergoes a change towards more crystallinity by relaxation phenomena or towards greater randomness by shear deformed area. The general consensus is that elemental process of deformation in the MGs is a local rearrangement of atoms to accommodate local strains incurred through the redistribution of free volume and the operation of shear transformation zone, which can lead to nucleation and propagation of shear bands [1], although there has been considerable debate as to whether shear localization in MGs is mainly due to thermal effects or shear-induced
* Corresponding author. E-mail address: [email protected] (E.S. Park). http://dx.doi.org/10.1016/j.actamat.2016.06.026 1359-6454/© 2016 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
disordering such as dilation. On the other hand, the roomtemperature mechanical deformation of MGs with structural instability can induce local structural ordering with relatively lower energy levels by promoting a so-called structural relaxation due to annihilation of frozen-in “defect”, especially in the shear bands with increased atomic mobility [2]. Recent investigations of the structural variation in MGs via severe deformation process [3e7] have shown that partially devitrified MGs with fine nanocrystals embedded in the shear bands or the amorphous matrix exhibit controllable microstructures and mechanical properties. Thus, it has been understood that deformation of MGs may be a way of generating unique nanostructures with denser and softer regions by annihilation and creation of free volume as a microstructural response, which is not a simple alternative to thermal annealing. Although it appears that the devitrification process of MGs via simple heat treatments is relatively well understood, the correlation between deformationeinduced structural change and following devitrification behavior of the MGs has not been fully understood [1,7e9].
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
Among existing previous studies of the question, deformationinduced structural changes and following devitrification behavior of Al-based MGs have been relatively well defined in severe deformation environments such as tensile/compressive tests or rolling process and the following heat treatment [8,9]. For example, the presence of quenched-in-nuclei of alpha Al in an amorphous matrix can enhance local structural ordering in severe deformation environments, especially under compressive force [8], and minimize the size of precipitates after 1st crystallization behavior in the following heat treatment. However, other glass-forming systems such as Zr- [5,10], Fe- [8], Cu- [11] or Mg-based [12] MGs often show different responses to applied deformation. For example, some of them did not exhibit a consistent devitrification process after coldrolling, implying that the applied deformation on MGs does not necessarily exhibit a similar structural change at room temperature for all MGs. Also, the trends of the DSC peak shift (or reduction of crystallization enthalpy) were not consistent even at a similar amount of deformation [9,12], suggesting that the crystallization behavior in the following heat treatment after cold rolling might be closely dependent on structural instability of MGs [13e16]. In this regard, it is useful to scrutinize the devitrification behavior by cold rolling and following heat treatment in various glass-forming systems, which allows manipulation of the nanostructure as well as the structural instability of MGs. However, very little attention has been focused on the local structural variations in the relaxation regime and mechanical responses of MGs with different levels of structural instability via cold rolling. In the present study, two Mg-Cu-Gd glass-forming alloy compositions [17e20] on either side of the border of bulk metallic glass (BMG)-forming alloy compositions have been selected; Mg75Cu15Gd10 with a GFA of 2 mm (relatively strong glass former) and Mg85Cu5Gd10 with a GFA of less than 1 mm (relatively fragile glass former). In order to identify the devitrification behaviors and mechanical responses of the MG ribbons depending on cold rolling and the following heat treatment, the selected MG ribbons were investigated with a Transmission Electron Microscope (TEM) equipped with heating facilities both before and after cold rolling together with DSC analyses and nanoindentation test.
239
propagation of shear bands. The thickness of the ribbon was reduced from ~30 mm to ~20 mm (~33% in thickness reduction) after cold rolling. The structures of the samples were confirmed by X-ray diffraction (XRD: Rigaku CN2301) for the as-spun and cold-rolled ribbon samples using monochromatic Cu Ka radiation. The thermal history of the as-spun and cold-rolled specimens was investigated by differential scanning calorimetry (DSC; Perkin Elmer DSC8500) using various heating rates between 5 K/min and 80 K/ min. The microstructures of the MGs were examined by TEM (JEOL 300KVJEM-3011) with single tilting heating holder (Gatan model 628). To evaluate devitrification behavior of the Mg-based MGs, the holder was quickly heated up to 423 K and was kept at the temperature for 10 min. Then the temperature was increased up to ~523 K with a 10 K interval. To stabilize the temperature at each step, we kept the temperature at each step under isothermal condition for 3 min. To minimize thermal drift, a water cooling of the outer part of the sample holder was installed. The thin foil TEM specimens were prepared by ion milling (Gatan 695 PIPS II) with liquid nitrogen cooling after mechanical thinning process. Extreme care was taken with the HRTEM analysis since Mg-based MG thin foils oxidize readily upon exposure to air. And, the nanoindentation tests were performed using a nanomechanical tester (Hysitron TI 750 TriboIndenter). Indenting force was loaded up to a maximum load of 5 mN using a conical type indenter with a 2 mm radius by load control mode at a constant loading rate of 1 mN/s.
3. Results
2. Experimental
The Mg65Cu25Gd10 alloy exhibits significantly improved GFA with at least 8 mm in diameter (Dmax) by conventional Cu-mold casting method in air atmosphere [17]. The critical cooling rate for glass formation (Rc) was estimated to be approximately 10 K/s by the following equation: log Rc ¼ 2.52 logZmax þ 3.27 [22]. Fig. 1 shows a map of Dmax with border line of BMG formation in Mg100xCuxGd10 alloys (x ¼ 5e30 at.%). As shown in Fig. 1, with increasing Mg contents up to 85 at.%, the Dmax gradually deceases down to 0.5 mm (Rc ~ 104 K/s [22]). In particular, the border of BMGforming alloy composition is located between Mg80Cu10Gd10 and
A Cu-Gd master alloy was prepared by arc melting Cu and Gd (purity > 99.9%) under a Ti-gettered argon atmosphere in a watercooled copper crucible. The master alloy was then alloyed with Mg (99.9%) in a boron nitride (BN) coated graphite crucible under a dynamic Ar atmosphere using an induction furnace. The alloy ingots were melted several times to help improve compositional homogeneity. After complete melting, the liquid alloy was poured into a Cu mold in air. The copper mold was cone-shaped, 45 mm in height, 15 mm in diameter at the top, and 6 mm in diameter at the bottom. Rapidly solidified ribbon specimens were prepared by remelting the alloys in quartz tubes with over-pressure of 50 kPa through a nozzle onto a Cu wheel rotating with a surface velocity of 40 m/s. To exclude structural relaxation of Mg-Cu-Gd MGs at roomtemperature (corresponding to a homologous temperature T/Tg of about 0.7) like Mg-Cu-Y MGs [21], all experiments were done after 10 days to exclude room temperature relaxation effect. The amorphous ribbons were cold-rolled by a twin roll in order to examine the structural bias of severe deformation. The rolling process was performed by electric motor-controlled twin roll with a diameter of 20 cm at a constant speed of ~2.0 radians/sec. A single amorphous ribbon sandwiched by two clean stainless steel plates was repeatedly cold rolled. The thicknesses of the amorphous ribbons were measured after final pass. After rolling, the specimen was elongated towards the rolling directions due to the formation and
Fig. 1. Map of maximum diameter (Dmax) for glass formation with border line of BMG formation in Mg100xCuxGd10 alloys (x ¼ 5e30 at.%).
240
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
Mg85Cu5Gd10 alloy compositions. Although it is well known that most Mg-based BMGs exhibit extreme brittleness, the ribbon sample of Mg-rich alloy compositions over 75 at.% Mg can be folded 180 , and it is possible to apply severe plastic deformation to it by cold rolling without fracture. Thus, in the present study, both Mg75Cu15Gd10 (BMG former with Dmax ¼ 2 mm and Rc ~ 3.25 102 K/s) and Mg85Cu5Gd10 (MG former with Dmax ¼ 0.5 mm and Rc ~ 104 K/s) alloy composition have been selected to scrutinize the devitrification behavior and mechanical response after cold rolling (hereafter, the Mg75Cu15Gd10 and Mg85Cu5Gd10 are called as Mg75 and Mg85, respectively). The formation of an amorphous phase was confirmed from the broad halos of XRD profiles in as-spun Mg75 and Mg85 ribbons (not shown). There are no clear differences between XRD profiles before and after cold-rolling after thickness reduction up to 33% of both ribbons. Fig. 2 shows DSC traces obtained from as-spun and coldrolled Mg75 (a, c) and Mg85 (b, d) ribbons at a constant heating rate of 20 K/min. The DSC traces of Fig. 2 (a) and (b) show that there were no clear differences in the traces and heat of crystallization (DHcryst.) after thickness reduction up to 33% by cold rolling, indicating that the crystallization enthalpies of the amorphous ribbons were not critically affected by applied deformation, which is a similar observation to the case of Cu-Zr-Al BMGs [11]. However, the enlargement of the marked area in Fig. 2 (a) and (b) shows a difference of relaxation enthalpies between as-spun and cold-rolled ribbons as shown in Fig. 2 (c) and (d), and the heat of relaxation
(DHrelax.) of Mg75 and Mg85 was reduced down to 68% and 42%, respectively. Van den Buekel et al. [23] argued that the change in relaxation enthalpies of MGs during DSC experiments at a constant heating rate is due to the variation in excess free volume, implying that a local structural ordering with low energy levels occurred during cold rolling by stress-induced relaxation in both ribbons. Although both ribbons formed multiple shear bands after cold rolling, it is also difficult to observe increase of DHrelax. due to rejuvenation behavior, which can arise due to high pressure torsion of Zr-Cu-Al BMGs [24]. The increased local structural fluctuation of both Mg75 and Mg85 after cold-rolling can be indirectly seen from the comparison of the effective activation energy for glass transition and crystallization (ET ¼ Eg, Ex and Ep) evaluated using Kissinger’s equation [25] for both as-spun and cold-rolled ribbon samples:
. ln TT2 F ¼ lnðET =kB K0 Þ þ ET =kB TT
(1)
where kB is the Boltzmann constant and K0 is the frequency factor in the Arrhenius law KT ¼ K0exp(ET/kBTT) as shown in Fig. 3. Table 1 summarizes TTs (¼ Tgs, Txs and Tps) at different heating rates (5e80 K/min) for as-spun and cold-rolled Mg75 and Mg85 ribbons. Fig. 3 shows Kissinger plot of as-spun and cold-rolled ribbons of (a, c) Mg75 and (b, d) Mg85, which exhibits a linear relationship between ln TT2 =F and 1000/TT for as-spun and cold-rolled Mg75 and Mg85. The activation energy plotted by Tg (Fig. 3 (a), (b)) for the
Fig. 2. DSC traces obtained from as-spun and cold-rolled Mg75 (a, c) and Mg85 (b, d) ribbons at a constant heating rate of 20 K/min.
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
241
Fig. 3. Kissinger plot of as-spun and cold-rolled ribbons of (a, c) Mg75 and (b, d) Mg85, which exhibits a linear relationship between ln T2T/F and 1000/TT for TT (Tg and Tx) variations depending on heating rates of 5e80 K/min.
Table 1 Glass transition temperature, Tg, crystallization onset temperature, Tx, and crystallization peak temperature, Tp at different heating rates (HR ¼ 5e80 K/min) for as-spun and cold-rolled Mg75 and Mg85 ribbons. Composition (at.%) Mg75Cu15Gd10
As-spun
Cold-rolled
Mg85Cu5Gd10
As-spun
Cold-rolled
HR (K/min)
Tg (K)
Tx (K)
Tp (K)
5 10 20 40 80 5 10 20 40 80 5 10 20 40 80 5 10 20 40 80
411.8 415.3 419.2 424.8 431.7 413.7 416.8 421.1 426.3 432.5 421.4 423.3 429.0 438.9 446.5 424.1 426.0 432.5 440.0 447.8
441.5 444.9 449.0 454.1 461.1 441.7 444.9 449.4 454.4 460.7 447.5 451.5 455.4 459.9 466.1 447.7 451.6 455.8 460.7 466.1
444.7 448.2 452.6 457.9 466.0 444.9 448.4 452.8 457.9 465.2 449.5 453.8 458.2 463.5 471.1 449.7 454.1 458.4 463.8 470.5
cold-rolled ribbon of Mg75 and Mg85 (Egs ¼ 191.0 and 143.2 kJ/mol, respectively) exhibits a higher value compared with those for the Egs of as-spun ribbons (177.2 and 128.5 kJ/mol, respectively), which means that the cold rolling process increases the resistance of glass-to-supercooled liquid transition due to structural bias from forced deformation. However, the activation energy plotted by Tx
(Fig. 3 (c), (d)) and Tp (not shown) for the Mg75 and Mg 85 ribbons shows that the Exs (216.2 and 243.8 kJ/mol) and Eps (207.0 and 215.4 kJ/mol) of both cold-rolled ribbons exhibited similar values compared with Exs (209.1 and 240.1 kJ/mol) and Eps (193.5 and 205.1 kJ/mol) of the as-spun ribbons. Indeed, the discrepancy between Eg and Ex (or Ep) might result in unique devitrification
242
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
behavior during heat treatment after cold rolling. In general, experimental and computational evidences indicated that the slowdown of glass transition is related to the growth of distinct relaxing domains. Thus, in case of MGs with heterogeneity, the relaxing domains can be easily strengthened by mechanical treatment due to an increase in the entropy of fusion in the system by the generated heterogeneity. However, the microstructural imaging by HRTEM in the present study does not give any noticeable difference between as-spun and cold-rolled ribbon samples of both compositions. Fig. 4 shows the HRTEM image obtained from coldrolled Mg85, exhibiting a homogeneous contrast fluctuation, characteristic of a fully amorphous structure. On the other hand, the fast Fourier-transformed (FFT) pattern of Fig. 4 shows some weak spots and the selected area electron diffraction (SAD) pattern has indistinct glow outside of primary amorphous ring, both of which mean the possibility of presence of local ordering, while the local ordering is not clearly discernible in the HRTEM image. Although the structure of these MGs is still amorphous phase under resolution of HRTEM, it should be noted that the structure variation of these MGs can be conspicuously revealed by thermal motion as shown in Fig. 3. Thus, we carefully evaluated the microstructure to confirm a strong influence on the degree of local structural ordering in these MGs due to stress-induced relaxation. Because of the local heterogeneity under nanometer scale in these MGs, it is necessary to get local quantitative information on the microstructural parameters. Fig. 5 shows diffraction profiles (intensity vs. scattering vector, q) as deduced from an SAD pattern of as-spun and cold-rolled Mg75 and Mg85 ribbons using a profile analysis of the SAD pattern (PASAD)-tools [26]. The PASAD is a very powerful method for quickly analyzing nanocrystalline materials quantitatively on a local scale by correlating the results with regions specifically selected by TEM images. The resulting data facilitate a complete quantitative analysis of the sample on a local scale (probe volume~0.001 mm3). The profiles of cold-rolled Mg75 and Mg85 ribbons clearly show a shoulder in the right side of the halo pattern, which is clear evidence for the local structural ordering. Although Mg85 ribbon, relatively fragile glass former
Fig. 5. Diffraction profiles converted from an electron diffraction pattern of as-spun and cold-rolled Mg75 and Mg85 ribbons using PASAD-tools.
with lower structural stability, exhibits small shoulder before cold rolling due to room temperature annealing (corresponding to a homologous temperature T/Tg of about 0.7), the shoulder’s intensity slightly increases after cold rolling due to stress-induced relaxation. These results also explain why the reduction ratio of DHrelax. in Mg85 after cold rolling is smaller than that of DHrelax. in Mg75. Various properties of MGs are closely related to their atomic structure; hence knowledge of the atomic structure and its stability is needed to understand their properties. Fig. 6 shows (a) hardness and (b) load-displacement curves of as-spun and cold-rolled Mg75 and Mg85 ribbons measured by nanoindentation test. For the Mg75, the hardness of cold-rolled ribbon (3.52 ± 0.08 GPa) exhibits lower value compared with that of as-spun ribbon (3.80 ± 0.08 GPa), which means that relatively thick shear deformed area and localized clustering in the deformed area in Mg75 mainly cause softening effect. However, for the Mg85, the hardness of cold-rolled ribbon (3.39 ± 0.07 GPa) exhibits slightly higher value compared with that of as-spun ribbon (3.34 ± 0.07 GPa), which means that relatively thin shear deformed area and homogeneously distributed clustering by plastic deformation in Mg85 mainly cause hardening effect. Under nanoindentation, plastic shearing can be stabilized due to the confinement from the surrounding material even in brittle MGs without any compressive plasticity, which can be useful for analysis of the characteristics of the intermittent shear avalanches in most MGs. The load-displacement curves of Mg75 exhibit relatively significant pop-in events in the loading stage than those of Mg85 as shown in Fig. 6 (b). The results show clear difference in the serrated flow between Mg75 and Mg85 due to different shear banding process by accumulation of elastic energy and stress relaxation. In particular, it can be understood that the serrations of Mg 85 have smaller amplitude and larger numbers than those of Mg75, which means that the size of strain bursts and shear avalanche in Mg85 can be smaller than those of Mg75 due to different structural stability depending on fragility, which will be carefully discussed in Section 4.2. 4. Discussion
Fig. 4. HRTEM image, FFT pattern (inset, left) and SAD pattern (inset, right) obtained from cold-rolled Mg85.
It was reported that anelastic deformation and annealinginduced relaxation in MGs can be separated into reversible and
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
243
relationship:
i 2 h ln 10 Tg Tg0 m ¼ D* Tg0 Tg
Fig. 6. (a) Hardness and (b) load-displacement curves of as-spun and cold-rolled Mg75 and Mg85 ribbons.
irreversible components. This has been explained in terms of chemical short-range ordering (CSRO) and topological short-range ordering (TSRO), respectively. However, such a sharp separation between TSRO and CSRO, criticized by Gibbs and Sinning [27] and Khonik et al. [28], is probably too simple to explain such a complex phenomenon, because CSRO is unlikely without an accompanying TSRO. Although it is also difficult to separate TSRO and CSRO in Mgrich Mg-based MGs via room temperature relaxation and stressinduced relaxation, it is clear that the local structural and chemical bias can result in different mechanical responses as well as devitrification behavior depending on the degree of local structural change.
4.1. Abnormal devitrification behavior via local structural ordering The stability of amorphous structure in MGs can be quantitatively assessed by the fragility index, m. The stronger the liquid becomes (lower m), generally the higher the amorphous stability is, which is related to higher GFA. Therefore, the question that arises is how both composition tailoring and severe plastic deformation application can have such marked effects on the fragility from the microstructural point of view. The kinetic fragility for as-spun and cold-rolled Mg75 and Mg85 has been measured using the following
(2)
where D* and Tg0 are the fitting parameters which can be obtained from the plot of ln F (where F is the heating rate) vs. Tg. The details of Eq. (2) are described elsewhere [29]. Fig. 7 shows the variation of Tg for as-spun and cold-rolled Mg75 and Mg85 as a function of the heating rate F. Inserted table in Fig. 7 shows ln A, D* and Tg0 in the best fit to the experimental data and m value evaluated at a heating rate of 5 K/min by using Eq. (2). The m value of as-spun Mg75 (40.9 ± 2) is smaller than that of the as-spun Mg85 (42 ± 2), which means that as-spun Mg75 is relatively strong glass former with higher amorphous stability than as-spun Mg85, reflecting higher GFA. On the other hand, the m value of cold-rolled Mg75 (40.7 ± 2) and Mg85 (36 ± 2) is smaller than that of the as-spun Mg75 and Mg85, respectively, which means that the cold rolling enhances the glassy slow dynamics with increasing anisotropy of the interatomic potential by changing combinations of bonding energy toward local structural ordering. Interestingly, Mg85 with higher m exhibits larger reduction of m after cold rolling due to higher structural bias toward local structural ordering by stress-induced relaxation compared with Mg75 with lower m. Indeed, the structural variation by plastic deformation as well as structural stability can be correlated with fragility of MGs. Especially, the m values exhibit stronger correlation with atomic structure variation by severe plastic deformation than amorphous stability variation by GFA change in Mg75 and Mg85. Thus, it can be suggested that the m value could assist in understanding local structural variation as well as amorphous stability in MGs and therefore be regarded as an indicator of the atomic scale structural variation of MGs during mechanical processing. To check the qualitative variation of crystallization behavior via local structural heterogeneity, we observed microstructural variation during heating. Fig. 8 shows the SAD patterns at different temperatures observed during heating by TEM with single tilting heating holder as well as corresponding DSC trace obtained from as-spun Mg85 ribbon at a constant heating rate of 20 K/min. Although the thermal history between TEM results and DSC trace does not exactly match (heating rate of TEM is much slower than that of DSC), SAD pattern at 300 K already shows vaguely local sharp rings with diffuse ring of amorphous matrix due to structural ordering caused by room-temperature annealing as shown in Fig. 5. However, the high resolution TEM image of as-spun Mg85 ribbon at 300 K shows a typical amorphous structure with a lattice irregularity like Fig. 4 for cold-rolled Mg85. With increasing temperature, this sharp ring pattern becomes clearer up to 473 K and the spot patterns in SAD pattern, which are related to the grain growth during crystallization, start to appear from 483 K just before the finishing temperature of first crystallization. Fig. 9 shows TEM bright field (BF) images and SAD patterns of the as-spun Mg75 and Mg85 ribbons after heating up to 503 K completely after the first crystallization peak in TEM. As shown in Fig. 9 (a) and (b), as-spun Mg85 exhibits smaller average grain size (Davg ~ 70 nm) after annealing than as-spun Mg75 (Davg ~ 180 nm), implying that the local structural ordering in Mg85 ribbon by room-temperature relaxation might result in delayed crystal growth due to relatively high nuclei density and pinning site. To confirm this idea, we have carefully observed the SAD patterns of as-spun and cold-rolled Mg75 and Mg85 ribbons during heating by TEM with single tilt heating holder (Fig. 10). The spot patterns for as-spun Mg75 and Mg85 ribbons clearly appear at 473 K and 483 K, respectively, which is just before the finishing temperature of first crystallization. However, the spot patterns for cold-rolled Mg75 and Mg85
244
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
Fig. 7. Variations of Tg for the as-spun and cold-rolled (a) Mg75 and (b) Mg85 as a function of the F. Inserted table summarizes ln A, D* and Tg0 in the best fit to the experimental data and m value evaluated at a heating rate of 5 K/min by using Eq. (2).
Fig. 8. SADP patterns obtained from as-spun Mg85 ribbon during heating at (a) 300 K, (b) 443 K, (c) 473 K, (d) 483 K, and (e) 503 K, compared with DSC trace measured at a constant heating rate of 20 K/min.
ribbons appear at 483 K and 493 K, respectively, which is delayed about 10 K compared to those of as-spun ribbons. These results are consistent with TEM BF images after annealing, which mean that local structural ordering by cold rolling as well as room temperature annealing can cause abnormal devitrification behavior of delayed crystal growth. 4.2. Plastic dynamic transition via cold-rolling The deformation of MGs generally occurs inhomogeneously by a viscous-like behavior only at limited areas. The specimen deformed by cold rolling is divided into numerous small blocks by the deformed areas, which can consist of a mixture of shear deformed area and undeformed area, but the specimen as a whole does not undergo homogeneous deformation at room temperature. Of course, the critical thickness and density of the shear deformed area are affected by structural instability of MGs, which can be
inferred from fragility [30], i.e., strong glass former with higher amorphous stability generally forms a thicker and fewer shear deformed areas during deformation than fragile glass former with lower amorphous stability. Indeed, the mixture of MG matrix and shear deformed areas generally reduces Young’s modulus, the fracture stress and the hardness due to so-called softening effect. However, Mg-rich MGs, especially out of the border of BMGforming alloy compositions, also introduced additional regularities through local structural ordering by stress-induced relaxation, which can result in higher hardness and elastic modulus. Thus, distinct effects of plastic severe deformation have been detected by comparing as-spun and cold-rolled Mg75 and Mg85 ribbons by nanoindentation test as shown in Fig. 6. The results show clear difference in the serrated flow as well as stress behavior after yielding between as-spun and cold-rolled Mg75 and Mg85 due to different shear banding process by accumulation of elastic energy and local structural ordering by stress-induced relaxation. In
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
245
Fig. 9. TEM BF images and SADP patterns of the as-spun Mg75 and Mg85 ribbons after annealing up to 503 K using in-situ heating holder.
Fig. 10. SADP patterns obtained from as-spun and cold rolled Mg75 and Mg85 ribbons during heating.
particular, it can be speculated that the serrations of Mg 85 are smaller in amplitude and larger in numbers than those of Mg75, which means that the size of strain bursts and shear avalanche in Mg85 can be smaller than those of Mg75 due to different structural
stability depending on fragility. To clarify this speculation, we carried out a statistical analysis on the strain burst size (S) by considering the distribution of the strain burst which is irregularly and stochastically changing across the as-spun and cold-rolled
246
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
Mg75 and Mg85 ribbons, attempting to acquire a better understanding of the mechanism in the intermittent pop-in events depending on local structural instability [31]. Thus, first we eliminate the influences from the increase of indentation depth (h) in the pop-in events. A polynomial function is used to fit the loading stage in load-displacement curves to get a baseline [Fig. 11 (a), (c)]. After subtraction of the baseline, the pop-in events as a function of the h value are visible. Since each pop-in event reflects a process of shear band formation and propagation, the depth drop, Dh, i.e., height difference between the peak and valley values [as marked in Fig. 11 (b), (d)], can reflect the shear step size. As shown in Fig. 11 (b), (d), the shear step size in cold-rolled Mg75 in much bigger than that of cold-rolled Mg85. However, the instrumental noise could also cause pop-in events, and needs to be removed. The pop-in events free from the background noise can be extracted from the 2s holding segment at peak load through linear fitting, which shows that the noise generates a shear step size of 0.7 nm in the present study. Thus, the pop-in events with shear step sizes less than 0.7 nm were not considered. After removing the noise, normalization of the Dh value by the h is carried out to eliminate the statistical error, which generates S (¼Dh/h). The distributions of the S value versus the h value from as-spun and cold-rolled Mg75 and Mg85 ribbons at a constant loading rate of 1 mN/s exhibit random fluctuations due to normalization process, although the size of distributions seems to decrease a little bit with increasing h. (not shown) Indeed, this deformation unit size is determined by the shear wave propagation distance. Fig. 12 (a)e(d) shows cumulative probability distributions of the S obtained from 10 different indentation curves (over 100 serrations for each sample) of as-spun
and cold-rolled Mg75 and Mg85 ribbons at a constant loading rate of 1 mN/s. The percentage of the number of pop-in events of Mg75 with the S being larger than a given value, P (>S), is clearly greater than that of Mg85 (Fig. 12 (a), (b)). And P (>S) of cold-rolled ribbons with local structural ordering as well as shear deformed area is smaller than that of as-spun ribbons (Fig. 12 (c), (d)). Interestingly, it should be noticed that the difference in P values between as-spun and cold-rolled sample in Mg85 is much higher than that in Mg75. To quantify this comparison using a Levenberg-Marquardt algorithm, the cumulative probability distributions of the as-spun and cold-rolled Mg75 and Mg85 ribbons can be predicted by an empirical relation [32]:
o n Pð > SÞ ¼ ASb exp ðS=Sc Þ2
(3)
where A is a normalization constant, b is a scaling exponent, and Sc is the cut-off of S. For an MG indented at a given loading rate, collective shear band avalanches self organize into a pattern of serrated flow characterized by intermittency with power-law distributions of avalanche sizes. The distribution of S, which is a scale free factor of serrated flow, follows a power law distribution at smaller S region, while a cut-off of this power-law scaling is observed at larger S region of cumulative curve. The fitting parameters, b and Sc, reflect the profile of the shear avalanche in MGs. b is associated with the degree of universality of the power-law relation, which characterizes the tendency of MGs to form jamming state, making cooperative motion of concordant regions [31]. Sc corresponds to the critical size (avalanche size) of strain burst,
Fig. 11. Polynomial function fitting curve of the displacement-load for the loading segment of cold-rolled (a) Mg75 and (c) Mg85 ribbons at a constant loading rate of 1 mN/s, (b) the correlation between Dh and h of cold-rolled (b) Mg75 and (d) Mg85 ribbons showing the serration events.
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
247
Fig. 12. Cumulative probability distributions of strain burst size of as-spun and cold rolled Mg75 and Mg85 ribbons. Open scattering points represent experimental results measured from the nanoindentation. Solid lines are fitting curves by Eq. (3).
Fig. 13. (a) Cut-off values of strain burst size, Sc and (b) scaling exponent, b of as-spun and cold-rolled Mg75 and Mg85 ribbons reflecting the profile of the shear avalanche in metallic glasses. Inserted table summarizes Sc and b values of as-spun and cold-rolled Mg75 and Mg85 ribbons.
overcoming the resistance of the jammed matrix. In other words, Sc can be regarded as a deformation unit size when the deformed area
grows into individual shear band showing distinct behavior from the matrix. The relationship among b, Sc and the corresponding
248
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249
shear banding behavior was also investigated by analyzing compressive stress-strain curves of Zr-based MG samples having brittle to ductile transition [33]. Large b and Sc resulted in thicker and fewer shear bands, while small b and Sc resulted in thinner and numerous shear bands. Fig. 13 shows the variation of Sc and b in asspun and cold-rolled Mg75 and Mg85 ribbons. As shown in Fig. 13 (a), the Sc of as-spun Mg 85 (relatively fragile glass former with lower structural stability) is much smaller than that of as-spun Mg75 (relatively strong glass former with higher structural stability). The result implies that Mg75 has a relatively large shear avalanche size (or deformation unit size), which is closely related to its thicker shear band during plastic deformation compared with Mg85. Interestingly, there is no clear change of Sc values before and after cold-rolling in Mg 75, which means that the local structural ordering mainly in the shear bands as well as localized (thicker and fewer) shear deformed areas does not clearly influence shear avalanche size (or deformation unit size). On the other hand, the Sc value of Mg 85 distinctly decreases after cold rolling, which implies that well distributed local structural ordering as well as (thinner and many) shear deformed areas may reduce shear avalanche size (or deformation unit size). Thus, it can be understood that in coldrolled Mg85 shear bands are initiated by continuously rendering the imposed strain and cutting the elastic energy accommodation, which results in smaller amplitude of a pop-in event and numerous shear bands with smaller deformation unit size to dissipate the plastic strain. Furthermore, the cumulative probability distributions of the S in Mg75 clearly show the significant power-law relation in a wide range of S (¼ larger b), suggesting a selforganized critical state in the dynamics of the collective shear banding. This provides solid evidence to support the idea that the large rigidly packed deformation units formed a jamming state to resist small perturbations [34], approaching a stable state, i.e., a non-chaotic state in Mg75 during deformation. On the other hand, with a decrease in the deformation unit size (¼smaller Sc) in Mg 85, the rigidly packed deformation units are smaller in size (¼smaller b), which means that the jammed state of the deformation units is weakened. Thus, it is understood that the transition from jamming to unjamming states may occur in Mg85. The un-jamming state is unstable to perturbations, exhibiting a chaotic state [35], which is manifested in the lack of power-law relation in the statistical distribution of strain bursts due to the influence from smaller Sc values, where the stress perturbations make it difficult to develop jamming state and finally result in successive formation of multiple and adjacent deformation bands at lower strain rate. In particular, the decrease in b after cold rolling in Mg85 is larger than that in Mg75 and, especially, the b value in cold-rolled Mg85 is close to 0, which implies that the deformation mode might change from a complex mode between self-organized critical state and chaotic state in as-spun Mg85 to chaotic state in cold-rolled Mg85 due to well distributed local structural ordering as well as numerous thinner shear deformed areas. Indeed, Mg85 after cold rolling exhibits unique hardening behavior under the condition for coexisting well distributed local structural ordering and numerous thinner shear deformed areas. Thus, it can be concluded that effective mechanical processing of MGs with different degrees of structural instability (fragility) can manipulate properties of MGs even without clear microstructural features via variation of shear avalanche size (or deformation unit size). Indeed, there are fruitful prospects for future improvements of the performance of MGs to form a composite with unique nanostructures via effective mechanical processing as well as heat treatment. 5. Conclusions Abnormal devitrification behavior and mechanical response
have been investigated in as-spun and cold-rolled Mg75 (relatively strong glass former with higher structural stability) and Mg85 (relatively fragile glass former with lower structural stability) MG ribbons. The structural instability after cold rolling up to a thickness reduction ratio of ~33% causes abnormal devitrification behavior in the following heat treatment: higher Egs and delayed growth due to increased nuclei density and pinning site. In particular, the m values, an indicator for structural stability, exhibit relatively stronger correlation with atomic structure variation by cold rolling than that by compositional tuning in Mg75 and Mg85. Indeed, Sc of as-spun Mg85 is much smaller than that of as-spun Mg75 and distinctly decreases after cold-rolling, which implies that the increased structural instability may reduce shear avalanche size (or deformation unit size). Also, with decreasing Sc in Mg85, the rigidly packed deformation units are smaller in size (closely related to smaller b), which means that the jammed state of the deformation units is weakened. Especially, the b value in cold-rolled Mg85 is close to 0, which implies that the deformation mode may change from self-organized critical state to chaotic state due to well distributed local structural ordering as well as thinner shear deformed area. Thus, Mg85 after cold rolling exhibits unique hardening behavior due to decrease in the interparticle spacing in SRO precipitates and the higher solute concentration in the remaining amorphous matrix. This result would help deepen our understanding of the importance of atomic scale structural manipulation in MG, which can prove very useful for development of new MG composites with optimal properties through effective mechanical processing as well as heat treatment. Acknowledgments The author’s work in this area was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (Ministry of Science, ICT and Future Planning) (No. 2014K1A3A1A20034841 & 2014M3C1A8053728). One of the authors (E.S. Park) also benefited from the Center for Iron and Steel Research (RIAM) at Seoul National University. References [1] C.A. Schuh, T.C. Hufnagel, U. Ramamurty, Mechanical behavior of amorphous alloys, Acta Mater. 55 (2007) 4067e4109. [2] E.S. Park, Understanding of the shear bands in amorphous metals, Appl. Microsc. 45 (2015) 63e73. [3] D.V. Louzguine, A. Inoue, Influence of a supercooled liquid on devitrification of Cu-, Hf-and Ni-based metallic glasses, Mater. Sci. Eng. A 375 (2004) 346e350. [4] W.H. Jiang, F.E. Pinkerton, M. Atzmon, Mechanical behavior of shear bands and the effect of their relaxation in a rolled amorphous Al-based alloy, Acta Mater. 53 (2005) 3469e3477. [5] J.S. Park, H.K. Lim, J.H. Kim, H.J. Chang, W.T. Kim, D.H. Kim, E. Fleury, In situ crystallization and enhanced mechanical properties of the Zr41.2Ti13.8Cu12.5Ni10Be22.5 alloy by cold rolling, J. Non-crystalline Solids 351 (2005) 2142e2146. [6] J.W. Tian, L.L. Shaw, Y.D. Wang, Y. Yokoyama, P.K. Liaw, A study on the surface severe plastic deformation behavior of a Zr-based bulk metallic glass (BMG), Intermetallics 17 (2009) 951e957. [7] J.H. Perepezko, K.E. Kimme, R.J. Hebert, Deformation alloying and transformation reactions, J. Alloys Compd. 483 (2009) 14e19. €sner, G. Wilde, Dislocation formation during [8] R.J. Hebert, J.H. Perepezko, H. Ro deformation-induced synthesis of nanocrystals in amorphous and partially crystalline amorphous Al 88 Y 7 Fe 5 alloy, Scr. Mater. 54 (2006) 25e29. [9] H.J. Jin, F. Zhou, L.B. Wang, K. Lu, Effect of plastic deformation on thermal stability in metallic glasses, Scr. Mater. 44 (2001) 1083e1087. [10] Y. Yokoyama, K. Yamano, K. Fukaura, H. Sunada, A. Inoue, Bulk metallic glasses. III. Enhancement of ductility and plasticity of Zr55Cu30Al10Ni5 bulk glassy alloy by cold rolling, Mater. Trans. 42 (2001) 623e632. [11] Q.P. Cao, J.F. Li, Y.H. Zhou, J.Z. Jiang, Microstructure and microhardness evolutions of Cu 47.5 Zr 47.5 Al 5 bulk metallic glass processed by rolling, Scr. Mater. 59 (2008) 673e676. [12] J.S. Park, J.M. Kim, E.S. Park, Structural behaviors of single and multiple amorphous alloys deformed by forced cold rolling, Intermetallics 18 (2010) 1920e1924.
J.I. Lee et al. / Acta Materialia 116 (2016) 238e249 [13] B.J. Lee, C.S. Lee, J.C. Lee, Stress induced crystallization of amorphous materials and mechanical properties of nanocrystalline materials: a molecular dynamics simulation study, Acta Mater. 51 (2003) 6233e6240. [14] E.S. Park, J.Y. Lee, D.H. Kim, A. Gebert, L. Schultz, Correlation between plasticity and fragility in Mg-based bulk metallic glasses with modulated heterogeneity, J. Appl. Phys. 104 (2008) 3520. [15] E.S. Park, D.H. Kim, Effect of manipulating atomic scale heterogeneity on plasticity in Mg-based bulk metallic glasses, Intermetallics 18 (2010) 1867e1871. [16] D.H. Kim, W.T. Kim, E.S. Park, N. Mattern, J. Eckert, Phase separation in metallic glasses, Prog. Mater. Sci. 58 (2013) 1103e1172. [17] H. Men, D.H. Kim, Fabrication of ternary MgeCueGd bulk metallic glass with high glass-forming ability under air atmosphere, J. Mater. Res. 18 (2003) 1502e1504. [18] F. Guo, S.J. Poon, G.J. Shiflet, Investigation of glass formability in Al-based multinary alloys, Scr. Mater. 43 (2000) 1089e1095. [19] A. Inoue, K. Ohtera, A.P. Tsai, T. Masumoto, Aluminum-based amorphous alloys with tensile strength above 980 MPa (100 kg/mm2), Jpn. J. Appl. Phys. 27 (1988) L479. [20] A. Inoue, W. Zhang, Formation, thermal stability and mechanical properties of Cu-Zr-Al bulk glassy alloys, Mater. Trans. 43 (2002) 2921e2925. € ffler, Room-tem[21] A. Castellero, B. Moser, D.I. Uhlenhaut, F. Dalla Torre, J.F. Lo perature creep and structural relaxation of MgeCueY metallic glasses, Acta Mater. 56 (2008) 3777e3785. [22] E.S. Park, C.W. Ryu, W.T. Kim, D.H. Kim, A novel parameter to describe the glass-forming ability of alloys, J. Appl. Phys. 118 (2015) 064902. [23] A. Van den Beukel, J. Sietsma, The glass transition as a free volume related kinetic phenomenon, Acta Metall. Mater. 38 (1990) 383e389. [24] F. Meng, K. Tsuchiya, I. Seiichiro, Y. Yokoyama, Reversible transition of deformation mode by structural rejuvenation and relaxation in bulk metallic
249
glass, Appl. Phys. Lett. 101 (2012) 121914. [25] H.E. Kissinger, Reaction kinetics in differential thermal analysis, Anal. Chem. 29 (1957) 1702e1706. [26] C. Gammer, C. Mangler, C. Rentenberger, H.P. Karnthaler, Quantitative local profile analysis of nanomaterials by electron diffraction, Scr. Mater. 63 (2010) 312e315. [27] J. Ding, S. Patinet, M.L. Falk, Y. Cheng, E. Ma, Soft spots and their structural signature in a metallic glass, Proc. Natl. Acad. Sci. 111 (2014) 14052e14056. [28] M. Gibbs, H.-R. Sinning, A critique of the roles of TSRO and CSRO in metallic glasses by application of the activation energy spectrum model to dilatometric data, J. Mater. Sci. 20 (1985) 2517e2525. [29] C.A. Angell, Formation of glasses from liquids and biopolymers, Science 267 (1995) 1924e1935. [30] V.A. Khonik, A.T. Kosilov, V.A. Mikhailov, V.V. Sviridov, Isothermal creep of metallic glasses: a new approach and its experimental verification, Acta Mater. 46 (1998) 3399e3408. [31] X.L. Bian, G. Wang, K. Chan, J.L. Ren, Y.L. Gao, Q.J. Zhai, Shear avalanches in metallic glasses under nanoindentation: deformation units and rate dependent strain burst cut-off, Appl. Phys. Lett. 103 (2013) 101907. [32] T. Richeton, J. Weiss, F. Louchet, Breakdown of avalanche critical behaviour in polycrystalline plasticity, Nat. Mater. 4 (2005) 465e469. [33] C. Wang, B.A. Sun, W.H. Wang, H.Y. Bai, Chaotic state to self-organized critical state transition of serrated flow dynamics during brittle-to-ductile transition in metallic glass, J. Appl. Phys. 119 (2016) 054902. [34] E.J. Banigan, M.K. Illich, D.J. Stace-Naughton, D.A. Egolf, The chaotic dynamics of jamming, Nat. Phys. 9 (2013) 288e292. [35] S. Papanikolaou, D.M. Dimiduk, W. Choi, J.P. Sethna, M.D. Uchic, C.F. Woodward, S. Zapperi, Quasi-periodic events in crystal plasticity and the self-organized avalanche oscillator, Nature 490 (2012) 517e521.