Formation of Bulk

Abstract

the optimization of compositions to obtain BMGs or BMG-based composites with desired properties.

Metallic Glasses and Their Composites

Kinetic Analysis of Glass Formation

The GFA of an alloy depends on the kinetics of evolution of the thermodynamically viable phases in the undercooled melt. The composition ranges where glasses may occur under given processing conditions are ultimately controlled by the competing crystallization of many possible phases, including not only the solid solutions but also the intermediate phases, either stable or metastable. For the GFA of a solidified alloy, the systems negative heat of mixing maintains the stability of both the undercooled melt and the solid solution, but more importantly, it reduces the atomic mobility in the melt. These conditions are not necessarily sufficient for glass formation; the kinetics of both partitionless (single-phase melt forming a single-phase crystal) and primary/eutectic crystallization of the other competing phases, either stable or metastable, ultimately controls the GFA. The primary crystallization of all possible phases from the non-phase-separated liquid can be considered. Assuming homogeneous nucleation without preexisting nuclei and following the simplest

Y. Li, S.J. Poon, G.J. Shiflet, J. Xu, D.H. Kim, and J.F. Lffler
A great expansion in the number of alloy compositions known to give bulk metallic glasses (BMGs) has occurred in recent years. This progress is reviewed, and factors contributing to glass-forming ability are discussed. Practical strategies for pinpointing compositions with optimum glass-forming ability are presented, with examples of their use. Consideration is also given to the wide range of possibilities for BMG-based composites.

Introduction
The formation of metallic glasses by direct quenching from the melt was first observed in 1960 in a Au-25at.%Si alloy.1 A Pd-based metallic glass in bulk form (diameter, >1 mm) was reported in 1969, and a 1-cm-diameter ingot of fluxed Pd40Ni40P20 glass was reported in 1984.2 Beginning in 1988, interest in the development of bulk metallic glasses (BMGs) was revived by Inoue,3 carrying out systematic searches for bulk glass formation in multicomponent alloys. These were succeeded by a report of a Zr-based Be-bearing alloy.4 Today, scores of centimeter-sized BMGs in a variety of alloy systems have been documented (see Table I).519 All of these alloys are multicomponent systems with at least three elements, and the critical cooling rates for glass formation can be as low as 1 K/s. Understanding glass formation, particularly in multicomponent systems, is a complex task involving multiple intertwined issues. Both qualitative and quantitative methods were developed during the past decades to analyze and predict glass-forming ability (GFA, expressed in terms of critical cooling rate) and glassforming range (GFR, expressing the range of composition), and to search for new glass-formers. Typical considerations involve thermodynamic driving force for crystallization, kinetic constraints to prevent nucleation and/or growth of the competing crystalline intermetallics, and dense efficient atomic packing that stabilizes the amorphous structure. These issues are addressed elsewhere in this issue,20,21 whereas this article focuses on

Table I: Summaries of Bulk Metallic Glass Alloys with Critical Size 10 mm.
System Pd-based Pd40Ni40P20 Pd40Cu30Ni10P20 Zr-based Zr65Al7.5Ni10Cu17.5 Zr41.2Ti13.8Cu12.5Ni10Be22.5 Cu-based Cu46Zr42Al7Y5 Cu49Hf42Al9 RE-based Y36Sc20Al24Co20 La62Al15.7Cu11.15Ni11.15 Mg-based Mg54Cu26.5Ag8.5Gd11 Mg65Cu7.5Ni7.5 Zn5Ag5Y5Gd5 Fe-based Fe48Cr15Mo14Er2C15B6 (Fe44.3Cr5Co5Mo12.8Mn11.2 C15.8B5.9)98.5Y1.5 Fe41Co7Cr15Mo14C15B6Y2 Co-based Ti-based Ca-based Pt-based Co48Cr15Mo14C15B6Er2 Ti40Zr25Cu12Ni3Be20 Ca65Mg15Zn20 Pt42.5Cu27Ni9.5P21 Alloy Critical Size, Dc (mm) 10 72 16 25 10 10 25 11 25 14 12 12 16 10 14 15 20 Method Fluxing Water quenching Water quenching Copper mold casting Copper mold casting Copper mold casting Water quenching Copper mold casting Copper mold casting Copper mold casting Copper mold casting Copper mold casting Copper mold casting Copper mold casting Copper mold casting Copper mold casting Water quenching Year 1984 1997 1993 1996 2004 2006 2003 2003 2005 2005 2004 2004 2005 2006 2005 2004 2004 Ref. 2 5 6 4 7 8 9 10 11 12 13 14 15 16 17 18 19

624

MRS BULLETIN VOLUME 32 AUGUST 2007 www/mrs.org/bulletin

Formation of Bulk Metallic Glasses and Their Composites

treatment22 based on JohnsonMehl Avrami transformation kinetics, the time needed for formation of any phase with a minimal volume fraction fv (say, 106) can be estimated. A reduced-time variable t = t/tmin may be introduced, where tmin represents the minimum time needed for the transformation at a given composition. Using thermodynamic driving forces computed by standard means, the composition dependence of t can be calculated for alloys at selected temperatures.23 For the Al-Gd-Fe ternary system,24 the reduced times t for formation of fcc Al, Al13Fe4, Al5Fe2, Al2Fe, Al10Fe2Gd, Al8Fe4Gd, M3Gd, or M2Gd from the undercooled liquid of Al-Fe-Gd were calculated and show overlapping contours at 600 K (Figure 1). The relatively large values of t, indicating the slowest primary crystallization, clearly occur at compositions that match the easy GFR found experimentally. Similar results have been obtained for Al-Ni-Gd systems. The slowest primary crystallizations occur at hypereutectic compositions far from the eutectic compositions, generally matching the GFRs measured in melt-spun alloys. Thus, the compositional ranges of the slowest primary crystallization kinetics define where metallic glasses may be most easily formed. The nucleation and growth of competing crystalline phases can be impeded in several ways. In principle, the favorable glass-forming region in a multicomponent phase diagram may be located by examin0.35 0.30 M Gd 3 0.25 x(Gd) 0.20 0.15 0.10 0.05 GFR Al8Fe4Gd Al10Fe2Gd Al5Fe2

ing the crystallization kinetics of the various crystalline phases on cooling the melt. The highest GFA is likely to occur in a compositional region where the crystallization rate is slowest, as reported for glass-forming Al-Gd-Ni alloys.23

Optimizing GFA
Glass formation via suppression of crystallization can be achieved in two ways: one is to avoid nucleation altogether; the other (often overlooked) is the suppression of growth. The latter is worth consideration, because even when (as in most practical cases) heterogeneous nucleation occurs, a glass can still form if the growth of the nuclei is suppressed. A glass forms if its glass-transition temperature isotherm is higher than the growth temperature of any of the possible crystalline phases. On cooling, the phase with the highest growth temperature is kinetically favored and is observed in the solidified microstructure. By applying this competitive-growth principle and treating the glass as another competing phase, complete suppression of crystal growth is expected when the temperature of full glass formation is higher than the growth temperature of any of the crystalline phases. Based on the above assumption, an explicit relationship between the GFA and composition has been derived (Figure 2).8,34,35 Two kinds of glass-forming zones, corresponding to two types of eutectic systems, were predicted. In the regular eutectic system, the best glass-forming range includes the eutectic composition (Figure 2a), whereas in an irregular eutectic, the easy glass-forming range would be outside the eutectic composition (Figure 2b). In this case, the composition range of highest GFA will not be centered on the eutectic temperature. Based on this model, a practical strategy for pinpointing the alloy composition with the optimum GFA experimentally has been developed. This makes use of the glass-forming diagrams and takes advantage of the hints provided by the microstructure. Figure 2 shows that in both cases, a glass-forming region is enclosed by composite-forming regions. Thus, as a function of alloy composition, the microstructure in a cast sample of an appropriate size would change from a composite (a primary phase in the amorphous matrix), to fully amorphous, and then to another composite (a different primary phase in the glass matrix). This is also evident for ternary systems. Thus, by monitoring the microstructure evolution with composition and eliminating the primary phases, one can locate the best glass-former in both binary and ternary systems. Successful implementation of this strategy was demonstrated in an asymmetrical binary (Cu-Zr) system near the eutectic reaction of liquid Cu8Zr3 + Cu10Zr7,36 identifying the best BMGforming composition of Cu64.5Zr35.5, and also in a ternary Zr-Al-Cu system.37

Effects of Atomic Size on Glass-Forming Ability

Glass formation requires the stabilization of the amorphous structure and suppression of crystallization. For multicomponent alloys composed of atoms of different sizes, a dense random-packed structure is energetically favorable.23 Increased atomic packing efficiency also decreases the average atomic free volume, which reduces the atomic mobility that mediates crystallization. Miracle and Senkov reported that GFA is favored by a more uniform separation in the atomic sizes, as well as a wider range of size.25,26 Earlier, Monte Carlo simulation of random close packing of polydisperse spheres produced similar findings.27 The combined effects of atomic size distribution, chemical compatibility, and liquid-phase stability on GFA were investigated at the University of Virginia.2830 The interplay of atomic size and chemical interaction in GFA, which can be readily demonstrated in multicomponent alloy systems, is seldom addressed because of the experimental difficulty involved. One system that has been studied in considerable detail is based on the near-eutectic Y56Al24(Co,Ni)20 alloy.29 The atomic radii (ra) of the alloying elements decrease in the order ra(Y > Al > Ni/Co) = (0.180 nm > 0.143 nm > 0.124/0.125 nm). To achieve a more uniform atomic size distribution, Zr, Mg, and Sc with similar ra 0.16 nm but different chemical interactions with the host Y and the solute components (Al, Ni, and Co) were selected as substitutes for Y to evaluate the GFA. Scandium clearly possesses the best combination of effects. Not surprisingly, it is the most effective substitute for Y, leading to the formation of an amorphous (Y,Sc)56Al24(Co,Ni)20 ingot with diameter of 30 mm by water quenching.9 Meanwhile, there also exist crystalline phases that can become unstable upon the addition of certain alloying elements.31 For example, some Fe-based BMGs such as Fe48Cr15Mo14(Ln,Y)2C15B6, known as amorphous steel, exhibit a significantly enhanced GFA with additions of yttrium or a lanthanide element (Ln).13,32 Whereas the large Y and Ln atoms can be accommodated in the melt, their presence in the competing Fe23C6 crystalline phase significantly decreases its stability; the need for these solutes to be partitioned for crystallization to proceed enhances the GFA.33

M2Gd

fcc 0.00 0.00 0.05 0.10 0.15 0.20 0.25 0.30 x(Fe) Figure 1. Overlapping plots of the reduced transformation time t for the primary crystallizations from undercooled Al-Fe-Gd liquid at 600 K. The black area denotes the glassforming range (GFR) for melt-spun glassy samples ~20 m thick. The variable x is in atomic percent. M is mixed Al and Fe in the sublattice M. (Adapted from Reference 23.)

MRS BULLETIN VOLUME 32 AUGUST 2007 www/mrs.org/bulletin

625

Formation of Bulk Metallic Glasses and Their Composites

Increasing the number of components in a given system is a popular and effective approach for enhancing GFA. For a quaternary system, all of the alloy compositions are distributed within a 3D composition space, which can be represented by a component tetrahedron. It is a challenge to locate the best glass-forming composition using a minimum number of alloys, especially since GFA can be strongly composition-dependent. In examining the alloying effects of a component, a simple approach is to partially substitute an existing component, for example, Ag substitution for the Cu in Mg-Cu-Y without any changes in the Mg and Y fractions. However, the metallurgical features (eutectic composition, shape of the liquidus surface, etc.) of the Agcontaining pseudo-ternary and eventually quaternary system differ from those of the starting ternary one. These factors strongly influence the GFA of the alloys. It is evident that the best glass-forming composition is not necessarily on the route of simple substitution. In a systematic study of the Mg-Cu-AgY quaternary as a model system, an approach to searching a 3D compositional space was established.11,38,39 The quaternary system was treated as pseudoternary, such as Mg-(Cu,Ag)-Y with respect to ternary Mg-Cu-Y. Then the system was decomposed into a few consecutive compositional planes, each with a fixed Ag-to-Cu ratio, expressed as Cu1x Agx (such as, x = 0, 0.1, 0.2, 0.3). This reduces the 3D search to several manageable 2D problems, because one can pinpoint the best GFA on each plane to set up reference points. Figure 3 shows the contour for BMG formation with a critical diameter Dc = 8 mm in copper mold casting. The x = 0.2 plane gives the largest GFA, with Dc = 16 mm (the red zone in Figure 3). The locus connecting the compositions of highest GFA (Dmax) found on each plane delineates the pathway, now a 1D curved line, for locating the best glassforming alloy in the 3D compositional space. This orange arrow in Figure 3 points, as expected, in a direction very different from that of the vertical green arrow of the common substitution exercise. After pinpointing it on several planes, it can be concluded that the largest Dc is 16 mm for this system, at composition Mg54Cu28Ag7Y11. It was also verified, as seen in Figure 3, that a simple Ag substitution (green arrow) in the best glassforming alloy of the base ternary system would miss the best BMG formers. The power of the 3D pinpointing approach lies in locating the best glass-forming composition in a given system. By extending this

Figure 2. Phase-formation maps including the glass- and composite-forming regions for the two kinds of eutectic system. (a) In a regular eutectic system, the best glass-forming range includes the eutectic (Eu) composition. (b) In an irregular eutectic system, the easy glassforming range would be outside the eutectic composition. In this case, the composition . range of highest glass-forming ability will not be centered on the eutectic temperature. T is the cooling rate and C is the composition. (Adapted from Reference 34.)

Figure 3. BMG-forming composition region in the Mg-(Cu,Ag)-Y system. Within the blue region, the critical diameter of the glasses exceeds 8 mm. The orange arrow connecting the best compositions deviates significantly from a simple addition of Ag (green arrow). (Adapted from Reference 11.)

approach to several systems, new BMGs with large Dc, such as Cu44.25Ag14.75Zr36Ti5 (Dc = 10 mm),40 Mg57Cu31Y6.6Nd5.4 (Dc = 14 mm),41 and 25-mm-sized11 Mg54Cu26.5Ag8.5 Gd11 were also discovered. Another systematic approach to developing new BMGs is high-temperature centrifugal processing.42,43 In this procedure, a multicomponent alloy is melted in a centrifuge, and the resulting melt is slowly cooled during continuous centrifugation at gravitational accelerations of up

to 100,000 g. Upon cooling, a series of primary phases forms, grows, and spatially separates by sedimentation. This process changes the liquids composition until it finally solidifies in multiphase eutectic microstructures. The method thus yields the sequence of crystallization and the deep eutectic compositions that solidify last upon cooling. Producing alloys of these deep multiphase eutectic compositions leads to straightforward development of new BMGs.

626

MRS BULLETIN VOLUME 32 AUGUST 2007 www/mrs.org/bulletin

X, Ag-to-Cu Ratio

Formation of Bulk Metallic Glasses and Their Composites

BMG-Based Composites
At ambient temperatures where inhomogeneous deformation prevails, monolithic metallic glasses show consistently higher strength than crystalline materials of the same density.3,4446 However, although the strength of metallic glasses is very high, they exhibit very limited plastic deformation without strain-hardening. Indeed, a tendency toward work-softening leads to localization of plastic flow into shear bands, and metallic glasses tend to fail spontaneously on one or a few dominant shear bands. To solve the problem of low plastic strain, it is thus often important to develop metallic glass composites instead of monolithic metallic glasses. Thus, apart from the search for new singlephase metallic glasses, the production of two-phase composites is of interest. One of the earliest approaches was the development of in-situformed BMG composites.47 In this in situ approach, the original composition of the BMG is altered toward the composition of the primary crystallizing (ductile) phase. On cooling the melt, this ductile phase crystallizes first and shifts the composition of the remaining liquid toward the original glass so that this liquid solidifies as a BMG matrix (Figure 2). The resulting two-phase structure (often composed of dendrites in the glassy matrix) is very effective in increasing shear-band population and hindering the propagation of shear bands. This has led to a drastic increase in compressive plastic strain without a significant loss in strength for various metallic glass systems.35,4851 The in situ method generates a homogeneous crystal distribution, but the microstructure of the two-phase composite is very sensitive to casting conditions,52 such that upscaling of the technique may be difficult. Another approach to producing twophase composites is the introduction of foreign particles into the BMG matrix. In this case, microstructure and mechanical properties can be easily tailored via adjustments in the type, shape, size, and volume fraction of the reinforcement particles. In many cases, high levels of reinforcement content were required to improve plastic strain, which drastically reduced the yield strength of the material.53,54 However, it was recently recognized that soft particles are much more effective in increasing plastic strain. In fact, a plastic strain of more than 18% was achieved without sacrificing the high yield strength of the glassy matrix for a composite containing only 3.5 vol% graphite.55,56 A slightly different approach to increasing plastic deformation is to introduce a substantial volume fraction of micrometersized pores into the material.57 Such

microstructures exhibit excellent compressive plastic strain combined with a substantial increase in specific strength because of their reduced density. A further approach is to develop foam structures based on metallic glasses. Several methods of producing such metallic foams have been applied and are also discussed in this issue.58 Further, it should be mentioned that in rare cases, monolithic glasses can show high compressive plastic strain. This is the case when the Poisson ratio exceeds a critical value of approximately 0.310.32.5962 Unfortunately, the elements with the highest Poisson ratios (e.g., = 0.39 for Pt and = 0.42 for Au) tend to be precious ones, and large plastic strain in monolithic BMGs has only been observed once, in a costly Pt-based alloy.59 Finally, apart from forming composites of amorphous and crystalline phases, one can also develop a hierarchical microstructure entirely in the glassy state, for example, in all-metal BMGs such as the Zr-Cu-Ni-Ti-Be, Ni-Zr-Ti-Si, and Cu-Zr-TiAg alloy systems.4,40,63 These heterogeneous BMGs are of potential interest in terms of shear-band propagation as well as for the development of additional functional properties. In the multicomponent systems exhibiting high GFA, the differences in the heat of mixing between some binary combinations can be large, possibly leading to phase separation into two glassy phases. Silicate glasses provide wellknown examples of phase separation and are widely studied for scientific understanding and engineering applications.6466

In the case of metallic glasses, there have been recent demonstrations of phase separation in Zr-La-Al-Cu-Ni alloy,67 Ti-Y-AlCo,68 and Ni-Nb-Y.69 In particular, the Ti-Y-Al-Co glasses exhibit the entire spectrum of microstructural possibilities expected from a phase-separating system, ranging from a novel coreshell structure of spherical glassy particles embedded in a glassy matrix (due to the interplay between the critical wetting behavior of the phaseseparating system and the glass transition) to novel hierarchical arrangements of glassy spheres in a glassy matrix. The Zr-YAl-Co system gives a bulk phase-separating metallic glass system, indicating that such systems offer a unique opportunity to design composites with hierarchical microstructures of different length scales.70 The characteristic size scale of the inhomogeneity due to primary phase separation is highly dependent on the local cooling rate as well as on the alloy system; for example, the inhomogeneity scale is 25250 nm in as-melt-spun Ti28Y28Al24Co20 ribbon (Figure 4)68 and 0.11 m in as-melt-spun Ni58.5Nb20.25Y21.25 ribbon.69 In as-melt-spun La27.5Zr27.5Al25Cu10Ni10 ribbon,67 the upper size limit of the inhomogeneity is about 20 m. The addition of a small amount of an alloying element with a positive enthalpy of mixing with the constitutive elements of the metallic glasses can produce improvements in plasticity within a limited composition range, even though the microstructure is composed of an amorphous structure without obvious structural ordering. Enhanced plastic

100 nm

100 nm

Figure 4. (a) Bright-field transmission electron microscopy (TEM) image obtained from the region near the air-side of as-melt-spun Ti28Y28Al24Co20 alloy; (inset) the corresponding selected-area diffraction pattern. (b) A dark-field TEM image obtained using the inner diffuse halo marked by the white arrow in the inset in (a). (From Reference 68.)

MRS BULLETIN VOLUME 32 AUGUST 2007 www/mrs.org/bulletin

627

Formation of Bulk Metallic Glasses and Their Composites

strains to failure of 4.5%, 6.1%, and 4.2% have been reported in as-cast Zr57Ta5 Cu18Ni8Al10, Ni59Zr16Ti13Si3 Sn2Nb7, and Cu47Ti33Zr7Nb4Ni8Si1 BMGs, respectively;7173 Ta and Nb have positive heats of mixing with the constituent elements of Zr and Ti. The as-cast BMGs without Ta and Nb exhibit much lower levels of plastic strain: 1.1%, 2.1%, and 1.5% in as-cast Zr59Ti5Cu20Ni8Al10, Ni59Zr20Ti16Si2Sn3, and Cu47Ti33Zr11Ni8Si1, respectively. In contrast, when the alloy composition is such that two-phase glasses are formed by liquid-state phase separation, the plasticity decreases dramatically, resulting in extreme brittleness. Because of the positive enthalpy of mixing between Y and Zr in the Cu-Zr-Al-Y alloy system,74 phase separation can occur with increasing Y content, strongly indicating that local chemical inhomogeneity can exist even in alloy compositions that solidify into a single amorphous phase from the liquid state. Therefore the plasticity of the metallic glasses can be enhanced by introducing chemical inhomogeneity, but within a limited composition range. Very recently, coextrusion with polymers to form composites has also been exploited.75 With the development of metallic glasses with a glass-transition temperature in the range of that of polymers, the two can be processed at temperatures where they have the same viscosity. Thus, a new class of composite materials can be produced by replacing one of the metallic glass phases by a polymer phase. The resulting metallic glass/polymer composites promise to combine the complementary properties of metallic glasses (high strength, metallic conductivity, etc.) and polymers (high ductility, electrical insulation, etc.), if the two phases can be combined on an appropriate length scale.

Summary
Wider exploitation of bulk metallic glasses, and of the range of composites that can be based on them, awaits the development of low-cost materials with well-optimized property combinations. Although many new BMG-forming alloy compositions have been located in recent years, there is enormous potential for new compositions to be identified and then optimized using some of the approaches outlined in this article. There are also many possibilities for the development of BMG-based composites of different kinds.

References
1. W. Klement, R.H. Willens, P. Duwez, Nature 187, 869 (1960). 2. H.W. Kui, A.L. Greer, D. Turnbull, Appl. Phys. Lett. 45, 615 (1984). 3. A. Inoue, Acta Mater. 48, 279 (2000). 628

4. A. Peker, W.L. Johnson, Appl. Phys. Lett. 63, 2342 (1993). 5. A. Inoue, N. Nishiyama, H. Kimura, Mater. Trans., JIM 38, 179 (1997). 6. A. Inoue, T. Zhang, N. Nishiyama, K. Ohba, T. Masumoto, Mater. Trans., JIM 34, 1234 (1993). 7. D.H. Xu, G. Duan, W.L. Johnson, Phys. Rev. Lett. 92, 245504 (2004). 8. P. Jia, H. Guo, Y. Li, J. Xu, E. Ma, Scripta Mater. 54, 2165 (2006). 9. F.Q. Guo, S.J. Poon, G.J. Shiflet, Appl. Phys. Lett. 83, 2575 (2003). 10. H. Tan, Y. Zhang, D. Ma, Y.P. Feng, Y. Li, Acta Mater. 51, 4551 (2003). 11. H. Ma, L.L. Shi, J. Xu, Y. Li, E. Ma, Appl. Phys. Lett. 87, 181915 (2005). 12. E.S. Park, D.H. Kim, J. Mater. Res. 20, 1465 (2005). 13. V. Ponnambalam, S.J. Poon, G.J. Shiflet, J. Mater. Res. 19, 1320 (2004). 14. Z.P. Lu, C.T. Liu, J.R. Thompson, W.D. Porter, Phys. Rev. Lett. 92, 245503 (2004). 15. J. Shen, Q.J. Chen, J.F. Sun, H.B. Fan, G. Wang, Appl. Phys. Lett. 86, 151907 (2005). 16. H. Men, S.J. Pang, T. Zhang, J. Mater. Res. 21, 958 (2006). 17. F.Q. Guo, H.J. Wang, S.J. Poon, G.J. Shiflet, Appl. Phys. Lett. 86, 091907 (2005). 18. E.S. Park, D.H. Kim, J. Mater. Res. 19, 685 (2004). 19. J. Schroers, W.L. Johnson, Appl. Phys. Lett. 84, 3666 (2004). 20. R. Busch, J. Schroers, W.H. Wang, MRS Bull. 32 (8) (2007) p. 620. 21. D.B. Miracle, T. Egami, K.F. Kelton, K.M. Flores, MRS Bull. 32 (8) (2007) p. 629. 22. D.R. Uhlmann, J. Non-Cryst. Solids 7, 337 (1972). 23. A. Zhu, S.J. Poon, G.J. Shiflet, Scripta Mater. 50, 1451 (2004). 24. R.E. Hackenberg, M.C. Gao, L. Kaufman, G.J. Shiflet, Acta Mater. 50, 2245 (2002). 25. O.N. Senkov, D.B. Miracle, J. Non-Cryst. Solids 317, 34 (2003). 26. D.B. Miracle, Nature Mater. 3, 697 (2004). 27. D. He, N.N. Ekere, L. Lai, Phys. Rev. B: Condens. Matter 60, 7098 (1999). 28. S.J. Poon, G.J. Shiflet, F.Q. Guo, V. Ponnambalam, J. Non-Cryst. Solids 317, 1 (2003). 29. F.Q. Guo, S.J. Poon, G.J. Shiflet, J. Appl. Phys. 97, 013512 (2004). 30. F.Q. Guo, S.J. Poon, G.J. Shiflet, Appl. Phys. Lett. 84, 37 (2004). 31. W.H. Wang, Prog. Mater. Sci., 52 (4), 540 (2007). 32. V. Ponnambalam, S.J. Poon, G.J. Shiflet, J. Mater. Res. 19, 3046 (2004). 33. J. Wang, G.J. Shiflet, S.J. Poon, Phys. Rev. B (2007) submitted. 34. D. Ma, H. Tan, D. Wang, Y. Li, E. Ma, Appl. Phys. Lett. 86, 191906 (2005). 35. M.L. Lee, Y. Li, C.A. Schuh, Acta Mater. 52, 4121 (2004). 36. D. Wang, Y. Li, B.B. Sun, M.L. Sui, K. Lu, E. Ma, Appl. Phys. Lett. 84, 4029 (2004). 37. D. Wang, H. Tan, Y. Li, Acta Mater. 53, 2969 (2005). 38. H. Ma, Q. Zheng, J. Xu, Y. Li, E. Ma, J. Mater. Res. 20, 2252 (2005). 39. H. Ma, L.L. Shi, J. Xu, Y. Li, E. Ma, J. Mater. Res. 21, 2204 (2006). 40. C.L. Dai, H. Guo, Y. Shen, Y. Li, E. Ma, J. Xu, Scripta Mater. 54, 1403 (2006). 41. Q. Zheng, H. Ma, E. Ma, J. Xu, Scripta Mater. 55, 541 (2006).

42. J.F. Lffler, W.L. Johnson, Intermetallics 10, 1167 (2002). 43. J.F. Lffler, S. Bossuyt, A. Peker, W.L. Johnson, Philos. Mag. 83, 2797 (2003). 44. A.L. Greer, E. Ma, MRS Bull. 32 (8) (2007) p. 617. 45. A.R. Yavari, J.J. Lewandowski, J. Eckert, MRS Bull. 32 (8) (2007) p. 635. 46. J.F. Lffler, Z. Metallkd. 97, 225 (2006). 47. C.C. Hays, C.P. Kim, W.L. Johnson, Phys. Rev. Lett. 84, 2901 (2000). 48. T.C. Hufnagel, C. Fan, R.T. Ott, J. Li, S. Brennan, Intermetallics 10, 1163 (2002). 49. J. Das, M.B. Tang, K.B. Kim, R. Theissmann, F. Baier, W.H. Wang, J. Eckert, Phys. Rev. Lett. 94, 205501 (2005). 50. A. Inoue, W. Zhang, T. Tsurui, A.R. Yavari, A.L. Greer, Philos. Mag. Lett. 85, 221 (2005). 51. H. Ma, J. Xu, E. Ma, Appl. Phys. Lett. 83, 2793 (2003); Y. Xu, H. Ma, J. Xu, E. Ma, Acta Mater. 53, 1857 (2005). 52. W. Lser, J. Das, A. Gth, H.J. Klau, C. Mickel, U. Khn, J. Eckert, S.K. Roy, L. Schultz, Intermetallics 12, 1153 (2004). 53. H. Choi-Yim, R. Busch, U. Kster, W.L. Johnson, Acta Mater. 47, 2455 (1999). 54. H. Choi-Yim, R.D. Conner, F. Szuecs, W.L. Johnson, Acta Mater. 50, 2737 (2002). 55. M.E. Siegrist, J.F. Lffler, Scripta Mater. 56, 1079 (2007). 56. J.F. Lffler, A.A. Kndig, F.H. Dalla Torre, in Materials Processing Handbook, J.R. Groza, J.F. Shackelford, E.J. Lavernia, M.T. Powers, Eds. (CRC Press, 2007), p. 17-1. 57. T. Wada, A. Inoue, A.L. Greer, Appl. Phys. Lett. 86, 251907 (2005). 58. A.H. Brothers, D.C. Dunand, MRS Bull. 32 (8) (2007) p. 639. 59. J. Schroers, W.L. Johnson, Phys. Rev. Lett. 93, 255506 (2004). 60. J.J. Lewandowski, W.H. Wang, A.L. Greer, Philos. Mag. Lett. 85, 77 (2005). 61. X.J. Gu, A.G. McDermott, S.J. Poon, G.J. Shiflet, Appl. Phys. Lett. 88, 211905 (2006). 62. A. Castellero, D.I. Uhlenhaut, B. Moser, J.F. Lffler, Philos. Mag. Lett. 87, 383 (2007). 63. S. Yi, J. K. Lee, W.T. Kim, D.H. Kim, J. NonCryst. Solids 291, 132 (2001). 64. J. Zarzycki, Discuss. Faraday Soc. 50, 122 (1970). 65. J.S. Langer, Ann. Phys. 65, 53 (1971). 66. D.R. Uhlmann, A.G. Kolbeck, Phys. Chem. Glasses 17, 146 (1976). 67. A.A. Kndig, M. Ohnuma, H.H. Ping, T. Ohkubo, K. Hono, Acta Mater. 52, 2441 (2004). 68. B.J. Park, H. J. Chang, W.T. Kim, D.H. Kim, Appl. Phys. Lett. 85, 6353 (2004). 69. N. Mattern, U. Khn, A. Gebert, T. Gemming, M. Zinkevich, H. Wendrock, L. Schultz, Scripta Mater. 53, 271 (2005). 70. B.J Park, H.J. Chang, D.H. Kim, W.T. Kim, K. Chattopadhyay, T.A. Abinandanan, S. Bhattacharyya, Phys. Rev. Lett. 96, 245503 (2006). 71. L.-Q. Xing, Y. Li, K.T. Ramesh, J. Li, T.C. Hufnagel, Phys. Rev. B: Condens. Matter 64, 180201 (2001). 72. M.H. Lee, J.Y. Lee, D.H. Bae, W.T. Kim, D.J. Sordelet, D.H. Kim, Intermetallics 12, 1133 (2004). 73. E.S. Park, D.H. Kim, T. Ohkubo, K. Hono, J. Non-Cryst. Solids 351, 1232 (2005). 74. E.S. Park, D.H. Kim, Acta Mater. 54, 2597 (2006). 75. A.A. Kndig, T. Schweizer, E. Schafler, J.F. Lffler, Scripta Mater. 56, 289 (2007).

MRS BULLETIN VOLUME 32 AUGUST 2007 www/mrs.org/bulletin