Thermoplastic Micro-Forming of Bulk Metallic Glasses: A Review

Abstract

Bulk metallic glasses are a fascinating class of metallic alloys with an isotropic amorphous structure that is rapidly quenched from liquid melts. The absence of a crystalline micro-structure endows them with a portfolio of properties such as high strength, high elasticity, and excellent corrosion resistance. Whereas the limited plasticity and hence poor workability at ambient temperature impede the structural application of bulk metallic glasses, the unique superplasticity within the supercooled liquid region opens an alternative window of so-called thermoplastic forming, which allows precise and versatile net-shaping of complex geometries on length scales ranging from nanometers to centimeters that were previously unachievable with conventional crystalline metal processing. Thermoplastic forming not only breaks through the bottleneck of the manufacture of bulk metallic glasses at ambient temperature but also offers an alluring prospect in micro-engineering applications. This paper comprehensively reviews some pivotal aspects of bulk metallic glasses during thermoplastic micro-forming, including an in-depth understanding of the crystallization kinetics of bulk metallic glasses and the thermoplastic processing time window, the thermoplastic forming map that clarifies the relationship between the flow characteristics and the formability, the interfacial friction in micro-forming and novel forming methods to improve the formability, and the potential applications of the hot-embossed micro-patterns/components.

Introduction

Metallic glasses (i.e., amorphous alloys) are a novel class of metallic alloys that are quenched rapidly from liquid melts such that the liquid-like amorphous structure is retained. The first report of metallic glass by Klement et al.1 in 1960 inspired substantial research efforts in glass formation,2 and an important milestone in this endeavor was achieved in 1974, when Chen3 obtained Pd-T-P amorphous alloys (T = Ni, Co, Fe) with critical casting thickness on the order of 1 mm, which turned the historic wheels of bulk metallic glasses (BMGs). Led by Inoue and Johnson et al., the search for new metallic glasses has made considerable progress since the 1980s, and a variety of BMG systems with high glass-forming ability (GFA) have been developed thus far.4–10 These successful findings, together with the fundamental understanding of micro-structures, excellent mechanical properties and potential applications, significantly boost the development of BMGs. By comparison with their crystalline counterparts, the concomitant absence of long-range order and dislocation-like defects result in a plethora of outstanding properties such as near-theoretical strength, high elasticity, high hardness, appreciable toughness and superior corrosion resistance.11–14 The attractive mechanical properties predict promising engineering applications as structural materials, and extensive work has been performed in exploring new glassy alloys with excellent mechanical properties.15–19

Unlike crystalline metals in which dislocations or grain boundaries cause plastic deformation, BMGs usually undergo inhomogeneous plastic deformation at room temperature caused by high localization of shear stress, resulting in catastrophic failure with zero global plasticity under tension,14,20,21 thus severely restricting their structural applications at the macroscale. Geometric constraints imposed by reducing the sample size or feature below a critical length scale (<1 mm) have been proved to be effective in mediating this challenge and invoking more uniform plastic deformation.22–25 The potential in the application of metallic glasses into micro- and nano-engineering such as micro- and nano-electro mechanical systems (MEMS/NEMS) has thus been highlighted.26 Whereas the high strength and limited plasticity have imposed a limitation on the processing of metallic glasses at ambient temperature, the nature of the amorphous structure with the existence of a supercooled liquid regime (SCLR) between the glass transition temperature (T _g) and the crystallization temperature (T _x) allows thermoplastic forming (TPF) of BMGs.27–29 TPF provides a remarkable, economical and scalable technique for net-shaping precise and versatile geometries on broad length scales ranging from 10 nm to several centimeters that were previously unachievable with any conventional crystalline metal processing techniques.26,30–51

Because formability is a material property that reflects the maximum strain a metastable material can undergo before it crystallizes under given geometry and processing parameters,52 in the thermoplastic forming of BMGs, the formability has been proposed as a qualification to justify the workability of metallic glass in the supercooled liquid region.53–58 For a metallic glass with a certain composition, low viscosity and long processing time are usually regarded as the two crucial factors that quantitatively define formability.40,52,58 The viscosity of supercooled liquid BMGs is actually dependent on the processing parameters such as temperature, stress and strain rate59,60 and the inherent fragility of the supercooled liquid BMGs with various compositions, which altogether play a crucial role in the transfer of material flow (namely, Newtonian and non-Newtonian flow, as depicted in the deformation map14) that directly determines the formability of supercooled liquid BMGs.49

Toward an in-depth understanding of the fundamental issues in TPF of BMGs, this paper reviews some important aspects in the following six sections. The first is a brief introduction. The second section presents the crystallization kinetics and mechanism in metallic glasses because amorphous alloys face crystallization risk during TPF. In general, the crystallization is determined by both temperature and processing time, which actually affect the viscosity, material flow and formability. The third section comprehensively reviews the forming process–flow characteristics–formability relationships. An important aspect during micro-/nano-scale TPF is the conspicuous BMG/mold interfacial effect that seriously retards the material flow and results in poor formability. Therefore, the co-relationship between material flow and interfacial modes is investigated in the fourth section. To weaken this interfacial effect and improve the workability, novel methods such as vibrational forming will also be introduced there. The fifth section focuses on the possible applications of micro-components/patterns. Finally, we summarize the progress of thermoplastic micro-forming and the future outlook in this area.

Crystallization and Thermoplastic Processing Time Window

In general, cooling reduces atomic mobility. If the cooling rate is faster than the atomic reconfiguration rate into a more thermodynamically favorable crystalline state, an amorphous solid such as metallic glass is frozen directly from the liquid state. Metallic glasses with a disordered metastable structure tend to transfer into an ordered state under certain conditions such as high temperature,61,62 high strain rate63,64 and high pressure65,66; this process is called “crystallization”. Crystallization of metallic glasses imposes drastic degradation of mechanical properties. Therefore, the investigation of crystallization kinetics is of paramount importance in preventing devitrification and evaluating the thermal stability of metallic glasses,67 which is critical in designing thermoplastic forming processes to preserve the original BMGs’ properties.67,68

Numerous studies67,69–82 have investigated the crystallization kinetics of amorphous alloys such as Zr-,67,70–73 Fe-,74–76 Pt-,62 Cu-77–81 and Ti-based 82,83 BMGs. The thermo-analytical techniques mainly focus on differential scanning calorimetry (DSC) and differential thermal analysis (DTA), in which two different modes—namely, non-isothermal mode (continuous heating from low temperature) and isothermal mode (annealing at a given temperature)—are used.80,81 In non-isothermal mode with various heating rates, the crystallization kinetic parameters, such as the apparent activation energy (E _a ) and Avrami exponent (n) at different crystallized volume fractions, are analyzed by the Kissinger method.80 The nucleation-and-growth behavior during isothermal crystallization is also represented by the variation of apparent activation energy and Avrami exponent in different crystallization stages that are illustrated by the Johnson–Mehl–Avrami (JMA) model,81 which usually reveals that the crystallization process of metallic glasses is diffusion-controlled with three-dimensional growth while increasing the nucleation rate.67,81,84–86

To dissect the physical mechanism of the transfer of metallic glasses from the amorphous state to the crystalline state under thermal conditions, the nucleation rate ($I$) for continuous nucleation is determined (from a thermodynamics point of view) by the difference in Gibbs free energy ($\Delta G$) and the interfacial energy ($\gamma$) between the crystalline precipitates and the glass matrix, expressed as87:

$$ I \propto D_{C} \cdot \exp \left( { - \frac{{W^{*} }}{{K_{\text{B}} T}}} \right) $$

(1)

where $W^{*} = 16\pi \gamma^{3} /3\left( {\Delta G} \right)^{2}$ is the nucleation barrier, K _B is the Boltzmann’s constant, $T$ is the experimental temperature, $D_{C} = K_{\text{B}} T\lambda^{2} /6\left( {\Omega_{a} \eta } \right)$ is the diffusion coefficient in which $\lambda$ is the jump distance, $\varOmega_{a}$ is the atomic volume, and $\eta$ is the viscosity of the supercooled liquid metallic glasses.88 Equation 1 can then be rewritten as

$$ I \propto \frac{{K_{\text{B}} T\lambda^{2} }}{{6\left( {\Omega_{a} \eta } \right)}} \cdot \exp \left[ { - \frac{{16\pi \gamma^{3} }}{{3\left( {\Delta G} \right)^{2} K_{\text{B}} T}}} \right] $$

(2)

According to Eq. 2, $\eta$, $T$, $\Delta G$ and $\gamma$ are the four crucial parameters that determine the nucleation rate. In general, $\Delta G$ between the original and new phases per unit volume depends on the magnitude of supercooling $\Delta T_{sc} = T_{l} - T$ ($T_{l}$ is the melting temperature).73,87 Therefore, low experimental temperatures result in a small value of $\Delta G$, which leads to a large $W^{*}$ and hence a slow nucleation rate.73 The decrease in temperature also increases the viscosity that retards the diffusion of atoms and contributes to more sluggish crystallization kinetics.

The viscosity exhibits a stronger temperature dependence compared with the Gibbs free energy difference2 and is a kinetic parameter that essentially describes the time scale for structural rearrangement of the liquid atoms in an undercooled state to form a crystalline nucleus. Therefore, the crystallization kinetics of a supercooled liquid alloy are principally determined by the viscosity, which is especially meaningful in the research of thermoplastic forming of BMGs. Previous literature 40,48,49,52,58 has reported that the temperature-dependent viscosity is one of the crucial parameters for improving the thermoplastic formability of BMGs in the supercooled liquid state. However, a risk of possible crystallization may occur at high processing temperatures (corresponding to low viscosity). For metallic glasses with various compositions, the temperature dependence of the viscosity is different, reflected in the fragility ($m$) of the supercooled liquid. Figure 1 shows a “fragility plot” in the form proposed by Angell,89 in which the viscosities of various glass-forming liquids are compared in an Arrhenius plot for which the inverse temperature axis is normalized with respect to $T_{\text{g}}$ (corresponding to the viscosity of 10¹² Pa s). The fragility is then defined as90

Fig. 1

Angell plots comparing the viscosities of several metallic glass forming liquids; data are taken from Refs. 54, 58, 91, 135

$$ m = \partial \log \eta /\partial \left( {T_{\text{g}} /T} \right)\left| {_{{T = T_{\text{g}} }} } \right. $$

(3)

By comparing the value of $m$, the strong and fragile liquids are classified. The strong liquids (such as SiO₂ in Fig. 1) exhibit near-Arrhenius behavior with high viscosities, whereas the fragile liquids present a dramatic temperature dependence of viscosity just above $T_{\text{g}}$. In this case, the viscosities of some amorphous alloys are approximately eight orders of magnitude lower than those of the strongest liquids,91 but the supercooled liquids are still thermally stable to accommodate the enhanced thermoplastic formability/workability.92

To evaluate the thermoplastic processing time of the supercooled liquid BMGs, isothermal crystallization experiments are usually carried out by DSC, and the time–temperature–transformation (TTT) diagrams can then be established based on the measured incubation time. Figure 2 (a) representatively depicts the incubation time, $\tau = t_{oc} - \left( {T_{iso} /v_{scan} } \right)$, of Au₄₉Ag_5.5Pd_2.3Cu_26.9Si_16.3 BMG93 under different isothermal temperatures, where the onset time ($t_{oc}$) for crystallization was determined as the intersection of the slopes of the baseline and the crystallization peak, and $T_{iso}$ is the isothermal temperature selectively above T _g (~118°C). According to the incubation time ($\tau$) under various isothermal temperatures, the corresponding TTT diagram is constructed, as illustrated in Fig. 2b. From the TTT curve, we can readily distinguish the time–temperature window for thermoplastic processing of BMGs. Conventional heating methods often consume some fraction of this time window for uniform sample preheating, which shortens the available window for subsequent thermoplastic forming. Recently, a Joule heating approach was introduced by Johnson et al.94,95 that generates heat ‘‘volumetrically’’ through a metallic glass such that it overcomes this limitation. In this method, an intense millisecond current pulse penetrates homogeneously into the BMG sample and achieves uniform ohmic heating from T _g to temperatures spanning the undercooled liquid region at rates of ~10⁶ K/s. This rapid capacitive discharge heating method can effectively avoid the crystallization and enable near-net thermoplastic shaping, even for marginal glass-forming alloys with very limited stability against crystallization that are hardly processable by conventional heating. However, this method is facing challenges in thermoplastic micro- and nanoscale forming; for example, the amorphous phase can only be kept for 2–3 s at high temperatures, therefore a high cooling rate is necessary, but demolding is very difficult when using a copper mold, and high cooling rate is difficult to realize when using a silicon or ceramic mold.

Fig. 2

(a) The incubation time, $\tau = t_{oc} - \left( {T_{iso} /v_{scan} } \right)$, of Au₄₉Ag_5.5Pd_2.3Cu_26.9Si_16.3 BMG under different isothermal temperatures ($T_{iso}$), wherein the onset time ($t_{oc}$) for crystallization was determined as the intersection of the slopes of the baseline and the crystallization peak, as indicated at 128°C; samples were held at $T_{iso}$ until a peak in the heat flow signal was observed. (b) The corresponding TTT curves

Thermoplastic Formability

Because of the crucial importance of formability in evaluating the workability of BMGs in a supercooled liquid state,49,53–58 Saotome et al.33,55 first proposed a method to evaluate the micro-/nano-formability of BMGs in a supercooled liquid state. In this method, the formability is quantified by the percentage of the flowed area, $R_{f} = A_{f} /A_{g}$ ($A_{f}$ is the flowed area into the V-groove, $A_{g}$ is the area of the V-groove) and the curvature of a deformed specimen into the V-groove. The alloys used by Saotome et al. were Pt-,55 Pd-33 and Zr-based32 BMGs with wide temperature ranges of the supercooled liquid region, $\Delta T_{x} = T_{x} - T_{\text{g}}$. An increase in $\Delta T_{x}$ allows a BMG to be processed at lower viscosities and longer forming times. Therefore, $\Delta T_{x}$ is highly indicative of the formability,57,92,96 and this empirical rule has inspired scientists to develop new bulk-glass forming alloys with wider $\Delta T_{x}$ to improve the formability in TPF.95,97 A different approach is taken by the normalized parameter $S = \Delta T_{x} /\left( {T_{l} - T_{\text{g}} } \right)$ 40 that should better reflect the formability of a BMG, particularly when comparing different BMG alloy families. Considering that the viscosity of metallic glasses is not only temperature dependent but also composition dependent (reflected in fragility) as mentioned above, the fragility parameter ($m$) was then proposed to measure the formability of supercooled liquid BMG by Kato et al.,92 who revealed that the most fragile BMGs (such as the Pt-, Au- and Pd-based BMGs depicted in Fig. 1) are ideal candidates for near-net shape processing with fine printability. Although the fragility parameter $m$ is more effective than $\Delta T_{x}$ in describing the TPF, the negative effect of $m$ on the crystallization kinetics should not be ignored.57 Recently, Schroers57 suggested a simple and precise standard to characterize the formability of BMGs, in which a specified volume of the BMG is heated through the supercooled liquid region under a constant load, and the maximum diameter ($d$) of the thermoplastically formed disc is taken as a measure of the formability of the BMG.

It is worth noting that all of these parameters (such as $R_{f}$, $\Delta T_{x}$, $m$ and $d$) proposed in the evolution of thermoplastic formability are faced in BMGs with various alloy systems; they do not provide any information about the temperature-dependent formability for a given BMG with a certain composition.49,58 It is conceivable that an increased temperature generally favors the formability from the viewpoint of viscosity reduction. It should be noted that the viscosity of a BMG is also affected by the strain rate. For example, increasing the strain rate at a certain temperature inherently induces a dramatic reduction of the viscosity, but such a reduction of viscosity increases the difficulty of the thermoplastic forming98 owing to the transition from Newtonian flow to non-Newtonian flow,49 wherein the flow of supercooled liquid metallic glass is inhomogeneous as discussed in the following sections.

To fully understand the phenomenon described above, the relationship between the formability and the flow characteristics (i.e., Newtonian flow and non-Newtonian flow) of BMGs during thermoplastic forming was systemically investigated by the authors,49 based on a series of hot-embossing experiments (the schematics are depicted in Fig. 3) at various strain rates and temperatures in the SCLR. According to the experimental results, we constructed a TPF map that describes the correlation of formability characterized by the filling height with the strain rate and temperature, as described in Fig. 4a. This TPF map allows us to draw a comprehensive comparison with the deformation map (see Fig. 4b) obtained from the corresponding uniaxial compression under the same conditions used in the hot-embossing experiments. The results revealed an inherent relationship between the thermoplastic formability and the flow characteristics; i.e., Newtonian flow facilitates the forming, whereas TPF in a non-Newtonian flow regime tends to be difficult.

Fig. 3

Schematics of the hot-embossing process. (a) The designed 3D sketch of micro-channel patterns. (b) The manufacturing steps of silicon master mold. (c) The DRI fabricated silicon master mold; (d, e) The Zr-based BMG is hot-embossed into the silicon master mold. (f) Zr-based BMG micro-channel structure is released by etching the Si wafer in KOH. Reproduced with permission from.49 Copyright 2012, Elsevier

Fig. 4

(a) Thermoplastic forming map of Zr₃₅Ti₃₀Be_26.75Cu_8.25 BMG processed in the supercooled liquid region, which reveals unfilled, partially filled and fully filled regimes. (b) A deformation map for the Zr₃₅Ti₃₀Be_26.75Cu_8.25 BMG obtained by compression tests, describing the transition from Newtonian flow to non-Newtonian flow. Reproduced with permission from.49 Copyright 2012, Elsevier

According to the phenomenological viscosity law99 based on the free volume model of Cohen and Grest (CG model),100 the viscosity $\eta \left( {\dot{\varepsilon },T} \right)$, which is sensitive to both temperature and strain rate, can be defined as

$$ \eta \left( {\dot{\varepsilon },T} \right) = \frac{{\eta_{0} }}{{D\left( {\dot{\varepsilon },T} \right)}}\exp \left( {\frac{\alpha }{{\upsilon_{T} }}} \right) $$

(4)

where $\upsilon_{T}$ is the temperature-dependent free volume concentration (FVC) in the CG model, $\alpha$ is a geometrical factor of order unity, $\eta_{0}$ is the viscosity of the liquid at the high temperature limit, and $D\left( {\dot{\varepsilon },T} \right) = 1 + \exp \left( {\frac{\alpha }{{\upsilon_{T} }}} \right)\left( {\frac{{\eta_{0} \dot{\varepsilon }}}{{\sigma_{0} }}} \right)^{b\left( T \right)}$, where $\sigma_{0}$ is a reference stress and temperature dependent $b\left( T \right)$ is a stretching exponent fitted from the data.99 By considering the effect of strain rate on free volume, $\eta \left( {\dot{\varepsilon },T} \right) = \eta_{0} \exp \left( {\frac{\alpha }{{\upsilon \left( {\dot{\varepsilon },T} \right)}}} \right)$ is assumed to satisfy both Newtonian and non-Newtonian flow regimes.49 Therefore, the free volume concentration, $\upsilon \left( {\dot{\varepsilon },T} \right)$, can be derived from Eq. 4 and is expressed as

$$ \upsilon \left( {\dot{\varepsilon },T} \right) = \frac{1}{{{1 \mathord{\left/ {\vphantom {1 {\upsilon_{T} - {{\left[ {\ln D\left( {\dot{\varepsilon },T} \right)} \right]} \mathord{\left/ {\vphantom {{\left[ {\ln D\left( {\dot{\varepsilon },T} \right)} \right]} \alpha }} \right. \kern-0pt} \alpha }}}} \right. \kern-0pt} {\upsilon_{T} - {{\left[ {\ln D\left( {\dot{\varepsilon },T} \right)} \right]} \mathord{\left/ {\vphantom {{\left[ {\ln D\left( {\dot{\varepsilon },T} \right)} \right]} \alpha }} \right. \kern-0pt} \alpha }}}}} $$

(5)

According to Eq. 5, plots of FVC at various temperatures and strain rates can be obtained, as depicted in Fig. 5. In most cases, FVC increases with both temperature and strain rate, corresponding to the heterogeneous viscosity. The interactions among different forming zones with various viscosities (different FVCs) result in unsteady flow and degrade the formability. It is worth noting that, at high temperatures with low strain rates, a flat line before rising can be observed in Fig. 5c and d, indicating a steady equilibrium between the creation and annihilation of the free volume, which corresponds to homogeneous distribution of viscosities in a BMG. This favors continuous and steady operation of flow units and facilitates the formability accordingly.49

Fig. 5

Evolution of free volume concentration with strain rates at different temperatures. Reproduced with permission from.49 Copyright 2012, Elsevier

Based on the research described above, a series of micro-patterns and micro-parts used for MEMS and biochips have been fabricated in the Newtonian flow regime, as listed in Fig. 6. It is clear that the hot-embossed structures are integrated without obvious disfigurement, especially for the Chinese famous poetry of the tang dynasty, “Yellow Crane Tower” (see Fig. 6t; the name originates from a famous and historic tower that was first built in 223 AD and stands on Sheshan/Snake-Hill at the bank of Yangtze River in Wuchang District, Wuhan, China).49 The results have demonstrated that the Newtonian homogeneous flow facilitates the TPF formability of BMGs.

Fig. 6

The hot-embossed surface micro-patterns and micro-components used in MEMS. (a–g) Micro-spring, micro-gear, micro-motor, micro-fan, micro-honeycomb structure, micro-gyroscope and micro-accelerometer structure. (h–j) Micro-turbines. (k–q) BMG molds with various patterns used in biochips; (r–t) beautiful surface features such as micro-bats and micro-poetry of Tang Dynasty “Yellow Crane Tower”

Interfacial Friction and Novel Forming Methods

The thermoplastic forming map not only clarifies the relationship between flow features and formability but also provides the selection of the processing parameters. However, the interfacial effect between amorphous alloys and mold materials becomes prominent during micro- and nanoscale forming, which seriously hinders the forming of metallic glasses.2,26,40,42,51 When the mold dimensions approach 100 nm or smaller, the interfacial effect is dominated by the strong capillary forces, which require an unexpectedly higher pressure to fulfill the master mold. This effect of surface tension (proportionate to capillary forces) was also considered by Saotome et al.32–34,55 in elucidating the reduction of the geometrical transferability with decreasing width of V-grooved molds from the microscale (e.g., 2 µm) to the sub-microscale (e.g., 200 nm). At the microscale, the interfacial effect is controlled by the frictional force,33,101,102 and various friction coefficients have been assumed in the finite element simulations.103–106

It is worth noting that the supercooled liquid exhibits Newtonian and non-Newtonian flow107 in the homogeneous deformation regime14 in contrast to the flow features of their crystalline counterparts at high temperatures. Consequently, the physical models (such as the open and closed lubricant pocket model108,109) that were proposed to illustrate the interfacial friction in the micro-forming of crystalline metals are possibly inapplicable to the micro-forming of BMGs. To clarify the possible friction mechanism of metallic glasses in micro-processing, the authors51 carried out a series of micro-extrusion experiments at a wide range of strain rates and temperatures in SCLR, and the resulting interfacial morphologies between the steel mold and the metallic glass were observed (Fig. 7). It is clear that metallic glass detours the protuberances and fully fills the subsequent concave surface without discernible gaps when micro-extrusion is conducted in the Newtonian flow regime, as described in Fig. 7a and d. In the non-Newtonian flow regime, metallic glass sweeps over the protrusion and leaves a large cavity, as marked by the yellow arrow in Fig. 7b and c.

Fig. 7

The morphology of the interface between Zr₃₅Ti₃₀Be_26.75Cu_8.25 metallic glass and steel mold after hot-extrusion under Newtonian flow (a, d) and non-Newtonian flow (b, c) regimes. The blue arrows show the direction of flow of the metallic glass in micro-forming. Reproduced with permission from.51 Copyright 2013, Taylor & Francis

Compared with typical frictional mechanisms (mechanical engagement regime, furrow regime and adhesion regime)110–112 depicted in Fig. 8a, an interfacial friction mechanism map of the BMG under various strain rates and temperatures are constructed as described in Fig. 8b. This enables us to comprehensively understand the correlation between the interfacial friction mechanism and the flow characteristics of the Zr-based BMG; namely, Newtonian flow promotes adhesion mechanism, whereas non-Newtonian flow enhances the furrow regime.51

Fig. 8

(a) Three types of contact tribological models for the Zr-based BMG during hot extrusion. (b) Frictional mechanism map, which exhibits the correlation between flow characteristics and friction models. Reproduced with permission from.51 Copyright 2013, Taylor & Francis

The research discussed above has revealed that the conspicuous interfacial effect between materials and molds seriously impedes the thermoplastic formability of BMGs in the supercooled liquid state. To achieve high thermoplastic formability, Newtonian flow (corresponding to spatiotemporally homogeneous flow) at high temperatures and low strain rates is usually required. However, the high processing temperature and the long processing time (corresponding to a low strain rate) incur a risk of possible crystallization. To reduce or eliminate the effect of interfacial friction, rolling113 and blowing methods2,114,115 have been introduced to reduce the contact area and enhance the thermoplastic formability of metallic glasses.

Based on the improvements of formability achieved in conventional metal forming by employing ultrasonic vibration,116–118 we have introduced vibrational loading to facilitate the thermoplastic formability of metallic glasses in our recent research.48 By comparison with the original sample, the plastic strain of the Zr-based BMG increases with the increase of both loading frequency and force amplitude, as described in Fig. 9a and b, respectively. To rationalize the physical origin of this vibration-facilitated formability, the concept of the “core–shell” model119–122 is introduced, in which the “core” region refers to the soft region with a loosely packed atomic structure and higher fraction of free volumes. Therefore, the “core” region can be activated easily during the plastic deformation and act as the fertile “flow units”. The volume ($\Omega$) of these flow units is thus regarded as one of the critical parameters that determine the plastic flow of BMGs. Based on the results of dynamic mechanical analyzer (DMA) testing, the variation of $\Omega$ with temperatures under various loading frequencies can be calculated theoretically, as depicted in Fig. 10a. In general, when the temperature approaches or exceeds $T_{\text{g}}$, the elastic shells will collapse, and the flow units around them will combine to form new, larger flow units. It should be noted that the increase in $\Omega$ decelerates with elevating loading frequency owing to the strong diffusion capacity that impedes the flow unit aggregation,48 as observed in Fig. 10a. In addition, the local atomic arrangement of flow units is also affected by the relaxation time. The increase in loading frequency at a certain temperature will shorten the relaxation time and slow down the structural relaxation process (i.e., annihilation of free volume), therefore inducing a high amount of free volume (corresponding to FVC),48,123 as simulated in Fig. 10b and c. According to the foregoing analysis, it can be well imagined that the increase in FVC and the decrease in $\Omega$ with increasing loading frequency in the deformed BMG result in a more homogeneous distribution of flow units spatiotemporally, which promotes material flow and thereby thermoplastic formability.48,49

Fig. 9

(a) The displacement–temperature (time) curves of the Zr₃₅Ti₃₀Be_26.75Cu_8.25 BMG after vibrational tension under various loading frequencies; (b) the distribution of effective strain of the BMG simulated under various force amplitudes. Reproduced with permission from.48 Copyright 2013, Elsevier

Fig. 10

(a) The theoretical calculated flowing unit volume of the Zr₃₅Ti₃₀Be_26.75Cu_8.25 BMG tested under different loading frequencies. (b) and (c) The simulated free volume distribution under the frequencies of 0.1 Hz and 1.0 Hz, respectively. Reproduced with permission from.48 Copyright 2013, Elsevier

The vibrational loading-facilitated thermoplastic formability of BMG in the supercooled liquid state can be unambiguously demonstrated through both compressive and hot embossing experiments. Figure 11(a) illustrates the stress–time curves under vibrational loading and maintained loading conditions. Although the maintained loading time is longer than that under vibrational loading, the vibrational loading induces a larger diameter of the deformed specimens, as depicted in Fig. 11b. This viewpoint has been further clarified through hot-embossing BMG on the silicon mold with micro-pillar features, as described in Fig. 11c and d, which reveal a higher pillar under vibrational loading than that obtained under maintained loading.48

Fig. 11

(a) The stress–time curves under maintained loading and vibrational loading; (b) photographs of the original BMG sample, compressive samples after vibrational loading and maintained loading; (c, d) SEM micro-graphs of the micro-pillar array after maintained loading and vibrational loading, respectively. Reproduced with permission from.48 Copyright 2013, Elsevier

Potential Applications of Micro-/Nano-Fabricated Components/Patterns

As reviewed in the foregoing sections, TPF provides a promising method that breaks through the bottleneck of processing BMGs at ambient temperature and demonstrates its superiority in net-shaping precise and versatile structures comprising nano- and micro-sized features. Through TPF-based nano-embossing, metallic glass nanowires with very high aspect ratios (>200) were fabricated by Schroers et al.26,42,52 The nanowire architectures exhibited superb durability combined with high electrocatalytic activity toward CO, methanol, and ethanol oxidation, displaying vast potential in areas such as energy conversion/storage and sensors.124 These novel architectures and outstanding durability motivate the generation of first functional proton exchange membrane micro-fuel cells (MFCs) with high-surface area catalytic and gas flow field components made from BMGs. These novel MFCs have been identified as promising alternative power sources for portable electronics.125 In their review, Inoue et al.126 summarized the possible applications of the nano-imprinted patterns on BMG surfaces. They noted the potential applications of these nanoscale-imprinted BMG surfaces in anti-reflection materials, cell culture media for bio-chips, electrode materials, hologram technology, and next-generation ultrahigh-density data storage material. Recently, Hasan et al.127 built hierarchical structures by using a sequential embossing technique by integrating nano-, micro- and macrosized features in a sequential order, which endows BMG surfaces with versatile functions for potential utilization in wetting, cellular response, electrochemical activity, and optical devices.

The TPF technique has also been used to fabricate micro-patterns/components of metallic glasses, such as the hot-embossed micro-lens arrays that can be used in aspheric lenses128 and micro-channel geometries with potential applications in fuel cell interconnect plates.129 Owing to the unique mechanical properties of BMGs, the hot-embossed metallic glass parts have been used as a robust tool to replicate the micro-patterns by micro-imprinting of the amorphous polymer poly(methyl methacrylate) (PMMA).38,47,130 The complex 3D micro-structures with high aspect ratios were hot-embossed from the BMG by Bardt et al.37 They envisaged some possible applications including high-Q (lightly damped) micro-resonators for the telecom industry, high-surface-area structures, micro-wave waveguides and connectors suitable for higher-frequency operations, multi-degree-of-freedom flexure-based micro-mechanisms, micro-surgical tools and devices, microscale motors and transmission components, micro-fluidic arrays, and free-form reflective micro-optics.

The unique mechanical properties of BMGs and their excellent micro-/nano-geometrical transferability in the supercooled liquid state exhibited great potential in the fabrication of superhydrophobic surfaces with a long lifespan in service.46,50 Considering that the wettability on a solid surface is governed by both surface free energy and geometrical structure,131 the BMG surface energy of various alloys systems, such as Pd-, Mg-, Zr-, Cu-, Fe- and Ni-based BMG, were examined. The static contact angles (CA) of water droplets on the smooth BMG surfaces were measured, as depicted in Fig. 12. The maximum and minimum values of CA were found for Pd₄₀Cu₃₀Ni₁₀P₂₀ and Cu₆₀Zr₂₀Hf₁₀Ti₁₀ BMGs, corresponding to a low and high surface energy, respectively.

Fig. 12

(a) The micro-graph of the polished BMG surface with roughness of approximately 9 nm; (b) the essential contact angles of water on mirror-like BMG surfaces

To demonstrate the effect of surface energy on the wettability of BMG surfaces and directly fabricate superhydrophobic metallic surfaces through surface geometric structure control, both Pd₄₀Cu₃₀Ni₁₀P₂₀ and Zr₃₅Ti₃₀Be_26.75Cu_8.25 BMGs with various surface energies (judged from the CA difference; see Fig. 12) were selected, and honeycomb micro-structures with various pitches were hot-embossed on the predesigned silicon master molds, as described in Fig. 13. The CA on the hot-embossed BMG surfaces without any chemical modification was then measured with a water droplet, and the results are displayed in Fig. 14. With regard to the Pd-based BMG with low surface energy, the value of CA is only 98.8° on the smooth surface, but it increases sharply on the patterned surface with increasing pitches and reaches more than 150° when the pitch between adjacent cells ranges from 115.5 to 600 μm and exhibits superhydrophobicity. This phenomenon can be well rationalized in terms of the modified Cassie–Baxter theory132 by considering the surface energy gradient.46

Fig. 13

(a, b) The SEM micro-graphs of the hot-embossed honeycomb structures on Pd-based BMG surface with various pitches; (c) micrographs of the hot-embossed pattern on a Zr-based BMG surface after etching, and (d) the corresponding magnified images of the top regions. Reproduced with permission from.46,50 Copyright 2012, 2013, American Institute of Physics

Fig. 14

Comparison of the experimental CA-pitch curves for the Zr-based BMG with hot-embossed micro-patterns (red line) from the Pd-based BMG with the same size patterns (black line), and from the Zr-based BMG with micro/nano hierarchical structures after chemical etching (blue line). Reproduced with permission from.50 Copyright 2013, American Institute of Physics

For the Zr-based BMG with a relatively large surface energy, the CAs on all patterned surfaces are significantly smaller than those on the Pd-based BMG surface, demonstrating the conspicuous effect of high surface energy on the wettability. To improve this negative effect on the superhydrophobicity, the micro/nano hierarchical structures were constructed by dealloying (chemical etching) the micro-patterned surface (Fig. 13c and d). From the magnified morphology of the top surface (Fig. 13d), numerous nanoscale protrusions were formed after etching that were distributed homogeneously on the surfaces of the honeycomb micro-pattern. It is interesting that CAs on the dealloyed Zr-based BMG surfaces (see the blue line in Fig. 14) are much larger than those before etching and exhibit superhydrophobicity when the pitch ranges from 75.5 μm to 195 μm, which demonstrates that the nanoscale structures indeed facilitate hydrophobicity.50

Another interesting finding can also be observed on the Zr-based BMG surface with micro/nano hierarchical structures, as depicted in Fig. 15. Although the surface exhibits superhydrophobicity with CAs over 150° (Fig. 15a), the water droplet could not roll and was firmly pinned on the BMG surface even when it was turned upside down, as described in Fig. 15b, demonstrating a strong adhesive force that was measured by a high-sensitivity micro-electromechanical balance system.45 The measured force–distance curves when a water droplet gradually approached and retracted from the BMG surfaces are illustrated in Fig. 15c and d. Compared with the Zr-based BMG surface with microscale honeycomb structures (see Fig. 15c), the dealloyed surface with micro/nano hierarchical structures exhibits a larger adhesive force (see Fig. 15c). The micro/nano hierarchical structures formed on the Zr-based BMG surface are actually very similar to the micro-graphs of gecko’s feet, in which the micrometer-size setae split into nanoscale spatulas that induce the intermolecular capillary effect, which gives rise to the high adhesive force. This superhydrophobic BMG surface with strong adhesion toward liquids indicates its promising application as a dry adhesive and the transport of liquid micro-droplets.

Fig. 15

Shapes of the water droplets on the as-prepared Zr-based BMG surfaces with different tilt angles: (a) 0°, (b) 180°. The adhesive force–distance curves recorded before and after the water droplet contacts the BMG surface with (c) micro-/nano- and (d) micro-structures. Reproduced with permission from.50 Copyright 2013, American Institute of Physics

Conclusion and Outlook

As a frontier of metal research, BMGs have attracted enduring attention in recent decades owing to their unique properties that exhibit great potential in engineering as structural materials. However, the processing limitation at ambient temperature hinders the wide application of BMGs. Thermoplastic forming takes advantage of the superplasticity of BMGs in the supercooled liquid state and provides a promising method to manufacture metallic glass patterns/parts on length scales ranging from nanometers to centimeters, which offers alluring prospects in micro-engineering applications.

Based on the research in the micro-forming of metallic glass, this paper focuses on some crucial fundamental issues that determine the surface quality, dimensional precision, ultimate micro-structures and properties of the thermoplastically formed micro-patterns/components. The crystalline kinetics analysis clarifies the mechanism underlying the “race” against crystallization during TPF of BMGs. The various TPF parameters that determine the crystallization kinetics are discussed, and the TTT diagram that predicts the TPF window is analyzed. The processing parameters not only affect the viscosity of supercooled BMGs but also determine the material flow and formability. The thermoplastic forming map clearly illustrates the forming process–flow characteristics–formability co-relationship. The interfacial effect in micro-forming is also discussed, and the relationship between flow characteristics and the interfacial friction modes is investigated. Some clever methods to reduce interfacial friction and improve formability are introduced. Finally, a variety of hot-embossed micro-patterns/components are listed, and their possible applications are discussed.

Although TPF of BMGs has undergone rapid progress, some challenges still exist that impede practical applications. First, the risk of crystallization in the thermoplastic forming of BMGs still exists, especially for the BMGs with narrow supercooled liquid regions.133 Second, oxidization occurs for most BMGs at high temperatures, which dramatically affects the surface quality and dimensional precision in addition to the formability at the nanoscale. The rapid heating method94,95 and wetting layer to reduce the barriers134 may serve as a toolbox to resolve these problems. Third, the current TPF techniques face challenges in fabricating complicated 3D structures due to the high viscosity of supercooled liquid BMGs, and novel forming methods may be necessitated in this respect. Fourth, the material flow is seriously affected by the interfacial effect on the micro- and nanoscale, and the root physical mechanism remains vaguely understood and must be settled. And fifth, large-scale manufacture is necessary to improve productivity and reduce the cost if the market for commercial application is to be developed. Therefore, opportunities abound in further investigation of the TPF of metallic glasses, and these endeavors will ensure the future of metallic glasses in micro-engineering applications.