**3.1 Effect of chemical composition**

During the searching for strong glass-forming alloys, the effect of composition on the crystallization behavior has been extensively studied in a variety of amorphous alloys, despite of the preparation methods (Suryanarayana & Inoue, 2011). Two examples are listed in this section to show how the chemical compositions of amorphous alloys influence the crystallization mechanism and crystallization products.

Zhang et al. (Zhang, et al., 2002) has investigated the addition of Al on the glass formation and crystalliztion in the ball-milled amorphous Ti50(Cu0.45Ni0.55)44-xAlxSi4B2 (x=0, 4, 8 and 12) alloys. Al additions were introduced to simultaneously replace part of the Cu and Ni in Ti50Cu20Ni24Si4B2 (Zhang & Xu, 2002) to further reduce the density of the resulting alloys and improve the thermal stability of the supercooled liquid. The Ti-based amorphous alloy powders prepared through this solid-state process exhibit a well-defined glass transition and a supercooled liquid region. Al addition has changed the crystallization mechanims and crystallizaiton products of the amoprhous Ti50Cu20Ni24Si4B2 alloy. Fig. 3 (a) displays the differential scanning calorimetry (DSC) scans for the as-milled samples with different Al contents. In all cases, an endothermic signal associated with the glass transition is evident. As see from Fig. 3 (a), the onset of glass transition temperature (*T*g) is apparently insensitive to the change in the overall alloy composition. With increasing Al substitution, the exothermic reaction due to crystallization occurs at higher temperatures and the single-step crystallization event changes to a two-step process. X-ray diffraction (XRD) has been used to identify the structural changes associated with the exothermal events at several different temperatures, as marked by dots in the DSC traces in Fig. 3 (b). For x = 0, the XRD pattern at 777 K crystallization peak and after the crystallization event (810 K) showed that the amorphous phase transformed into the cubic NiTi phase and an unknown phase. The same products were found for x = 4 after crystallization, as shown in the XRD pattern at 820 K. Such a transition can be regarded as a eutectic crystallization, by which the amorphous phase simultaneously transforms into more than two phases in one step (as stated in *Section 2.2*). For x = 8 and x = 12, on the other hand, crystallization is completed in two steps. Fig. 3

Crystallization Behavior and Control of Amorphous Alloys 191

annealing were investigated by XRD (Fig. 4(b)). Zr62Cu20Al10Ni8 transforms into cubic NiZr2 type and tetragonal CuZr2-type compounds. Annealing the alloy with *x* = 3 leads to primary precipitation of an icosahedral quasicrystalline (QC) phase with spherical morphology and a size of about 50 to 100 nm. For *x* = 5, the diffraction peaks are weaker in intensity and broader because the precipitates are as small as about 5 nm. For *x* = 7.5, the precipitates are about 3 nm in size. At first glance the XRD pattern (Fig. 4 (b)) after annealing displays no obvious reflections but only broad amorphous-like maxima. However, careful examinations of the annealed state by high intensity synchrotron radiation and/or by transmission electron microscopy (TEM) (Eckert, et al., 2001) clearly shows differences in scattering intensity compared to the as-cast state indicates the precipitation of a metastable

cubic phase with a grain size of ~2 nm coexisting with a residual amorphous phase.

Fig. 4. (a) DSC scans and (b) corresponding XRD patterns after isothermal annealing for the Zr62-xTixCu20Al10Ni8 (x = 0, 3, 5, and 7.5) amorphous alloys: *x =* 0, annealed at 723 K for 30 min; *x* = 3, annealed at 703 K for 5 min; *x* = 5, annealed at 683 K for 30 min and *x* = 7.5 annealed at 688 K for 40 min. Reprinted from (Eckert, et al., 2001), with permission from

Although Zr-based BMGs have shown high glass-forming ability, high thermal stability and excellent mechanical properties, the glass-forming ability of these BMGs appears to be significantly affected by the contamination of oxygen either from the raw materials or from the processing (Inoue et al., 1995a; Kubler et al., 1998; Lin et al., 1997). The investigation of the influence of oxygen on the crystallization behavior of Zr-based amorphous alloys

Elsevier.

**3.2 Effect of oxygen** 

(b) indicates that in addition to the NiTi phase precipitated in the first stage of the crystallization, the second crystallization peak in the DSC traces arises from the appearance of the Ti2Ni intermetallic compounds in the final crystallization products.

Fig. 3. (a) DSC scans and (b) the corresponding XRD patterns measured at room temperature after heating to different temperatures in DSC at a heating rate of 40 K/min for the mechanically alloyed Ti50(Cu0.45Ni0.55)44-xAlxSi4B2 (x=0, 4, 8, 12) powders. Reprinted from (Zhang, et al., 2002), with permission from Cambridge University Press.

The glass formation and crystallization behavior in multicomponent Zr-based alloys have been intensively investigated. In general, multicomponent Zr-based alloys can be used for the production of fully amorphous bulk samples with dimensions up to centimeter order, or for the formation of bulk nanostructured materials. However, the phase selection upon crystallization is strongly affected by the chemical composition of the amorphous phase. In order to obtain nanostructured materials from amorphous precursors (see *Section 5.1*), amorphous specimens are typically annealed at temperatures within the supercooled liquid region (the temperature region between onset glass transition temperature, *T*g, and the onset of crystallization, *T*x,) or close to *T*x. Eckert et al (Eckert et al., 2001) has investigate the crystallization behaviors of Zr-based BMGs and produce bulk nanostructured alloys by partial crystallization of the Zr-based BMGs precursors. Fig. 4 (a) displays the DSC scans for as-cast Zr62-xTixCu20Al10Ni8 glassy alloys (*x* = 0, 3, 5, and 7.5). Zr62Cu20Al10Ni8 crystallizes via one sharp exothermic peak to form several intermetallic compounds. Upon Ti addition, the crystallization mode changes toward a double-step process. With increasing Ti content, the first DSC peak shifts to lower temperatures and the enthalpy related to the second exothermic peak decreases. The samples were isothermally annealed for different times below *T*x for further study of the crystallization process. The crystallization products after

(b) indicates that in addition to the NiTi phase precipitated in the first stage of the crystallization, the second crystallization peak in the DSC traces arises from the appearance

of the Ti2Ni intermetallic compounds in the final crystallization products.

Fig. 3. (a) DSC scans and (b) the corresponding XRD patterns measured at room

(Zhang, et al., 2002), with permission from Cambridge University Press.

temperature after heating to different temperatures in DSC at a heating rate of 40 K/min for the mechanically alloyed Ti50(Cu0.45Ni0.55)44-xAlxSi4B2 (x=0, 4, 8, 12) powders. Reprinted from

The glass formation and crystallization behavior in multicomponent Zr-based alloys have been intensively investigated. In general, multicomponent Zr-based alloys can be used for the production of fully amorphous bulk samples with dimensions up to centimeter order, or for the formation of bulk nanostructured materials. However, the phase selection upon crystallization is strongly affected by the chemical composition of the amorphous phase. In order to obtain nanostructured materials from amorphous precursors (see *Section 5.1*), amorphous specimens are typically annealed at temperatures within the supercooled liquid region (the temperature region between onset glass transition temperature, *T*g, and the onset of crystallization, *T*x,) or close to *T*x. Eckert et al (Eckert et al., 2001) has investigate the crystallization behaviors of Zr-based BMGs and produce bulk nanostructured alloys by partial crystallization of the Zr-based BMGs precursors. Fig. 4 (a) displays the DSC scans for as-cast Zr62-xTixCu20Al10Ni8 glassy alloys (*x* = 0, 3, 5, and 7.5). Zr62Cu20Al10Ni8 crystallizes via one sharp exothermic peak to form several intermetallic compounds. Upon Ti addition, the crystallization mode changes toward a double-step process. With increasing Ti content, the first DSC peak shifts to lower temperatures and the enthalpy related to the second exothermic peak decreases. The samples were isothermally annealed for different times below *T*x for further study of the crystallization process. The crystallization products after annealing were investigated by XRD (Fig. 4(b)). Zr62Cu20Al10Ni8 transforms into cubic NiZr2 type and tetragonal CuZr2-type compounds. Annealing the alloy with *x* = 3 leads to primary precipitation of an icosahedral quasicrystalline (QC) phase with spherical morphology and a size of about 50 to 100 nm. For *x* = 5, the diffraction peaks are weaker in intensity and broader because the precipitates are as small as about 5 nm. For *x* = 7.5, the precipitates are about 3 nm in size. At first glance the XRD pattern (Fig. 4 (b)) after annealing displays no obvious reflections but only broad amorphous-like maxima. However, careful examinations of the annealed state by high intensity synchrotron radiation and/or by transmission electron microscopy (TEM) (Eckert, et al., 2001) clearly shows differences in scattering intensity compared to the as-cast state indicates the precipitation of a metastable cubic phase with a grain size of ~2 nm coexisting with a residual amorphous phase.

Fig. 4. (a) DSC scans and (b) corresponding XRD patterns after isothermal annealing for the Zr62-xTixCu20Al10Ni8 (x = 0, 3, 5, and 7.5) amorphous alloys: *x =* 0, annealed at 723 K for 30 min; *x* = 3, annealed at 703 K for 5 min; *x* = 5, annealed at 683 K for 30 min and *x* = 7.5 annealed at 688 K for 40 min. Reprinted from (Eckert, et al., 2001), with permission from Elsevier.

#### **3.2 Effect of oxygen**

Although Zr-based BMGs have shown high glass-forming ability, high thermal stability and excellent mechanical properties, the glass-forming ability of these BMGs appears to be significantly affected by the contamination of oxygen either from the raw materials or from the processing (Inoue et al., 1995a; Kubler et al., 1998; Lin et al., 1997). The investigation of the influence of oxygen on the crystallization behavior of Zr-based amorphous alloys

Crystallization Behavior and Control of Amorphous Alloys 193

results showed that the *T*g increases with the addition of oxygen. The base alloy containing the lowest amount of oxygen (x = 0.14) crystallizes in a single step. The addition of oxygen significantly decreases the width of supercooled liquid region (∆*T*x) from 85 K for x = 0.14 to 58 K for x = 0.82. The decrease in ∆*T*x is partly due to the increase in *T*g with increasing oxygen and also due to the appearance of a pre-crystallization peak in the oxygencontaining alloys before the main crystallization event. The base alloy heated to 673 K (in supercooled liquid region) shows an amorphous nature. In the x = 0.82 alloy, precipitation of spherical icosahedral particles in nanocrystalline state was observed within 10 min annealing at 673 K. Then it crystallizes to Zr2(Cu,Al) when heated to 723 K. No other phase is present in the alloy heated up to 753 K. The x = 0.82 alloy heated to 673 K has a similar trend to that of the base alloy. However, the alloy heated near the pre-crystallization peak (708 K) led to the presence of an icosahedral phase along with a small amount of Zr2(Cu,Al). Only Zr2(Cu,Al) is present in the alloy heated to 723 and 753 K. The base alloy with x = 0.14 remains amorphous for up to 10 min at 673 K, after which the formation of Zr2(Cu,Al) was observed. The XRD patterns in Fig. 5 (a) clearly indicate the formation of Zr2(Cu,Al) beyond 10 min of annealing at 673 K. No other phase was identified even after a longer annealing for 60 min (Fig. 5 (a)). XRD patterns of the alloy heat treated at 673 K (Fig. 5 (b)) show that the icosahedral phase starts forming after 10 min and persists for up to 15 min, beyond

The mechanism of the oxygen-induced precipitation of metastable fcc Zr2(Cu,Al) and icosahedral quasicrystalline phases is rationalized by considering the effect of oxygen on the nucleation process. The high thermal stability of multicomponent Zr-based amorphous alloys is generally attributed to the difficulty of precipitation of crystalline compounds from the undercooled liquid. The combination of elements with significantly different atomic sizes and negative enthalpies of mixing leads to a homogeneously mixed dense random packed structure of the liquid resulting in a large liquid–solid interface energy (Inoue, 2000). If the nucleating phase has a different composition with respect to the homogeneous undercooled liquid, then the nucleation of the phase requires substantial atomic rearrangement (Eckert, et al., 1998). The driving force for the polymorphous crystallization is ∆*G*total. However, if the icosahedral phase is stabilized by oxygen addition, the driving force for the primary crystallization of the icosahedral phase can be comparable to or higher than that for the polymorphous crystallization, ∆*G*total. In such a case, icosahedral phase would initially precipitate from the amorphous matrix by the primary crystallization. The free energy reduction is accompanied with this crystallization and there is still a driving force to form Zr2(Cu,Al) from the icosahedral phase and/or the remaining amorphous phase in the second stage. The formation of the icosahedral phase would be preferable if the driving forces for the polymorphous crystallization and the primary crystallization are comparable, because it is believed that icosahedral clusters are present in the amorphous phase, and these would act as nuclei for the icosahedral phase primary crystals. If such icosahedral clusters are stabilized by the presence of oxygen, the oxygen-enriched alloy would be favour to form an icosahedral phase by primary crystallization. The differences in the sequence of the phase formation in these alloys with x = 0.14 and 0.82 are illustrated schematically in Fig. 6, in which the darkness of the gray scale corresponds to the concentration of Zr. In the base Zr–Cu–Al and Zr–Cu amorphous alloys, crystallization proceeds by polymorphous reaction without change in composition. On the other hand, in the oxygen-containing ternary alloys, the first stage of crystallization occurs by primary

which the icosahedral phase transforms to Zr2(Cu,Al).

(Altounian et al., 1987) showed that the oxygen induces the formation of metastable facecentered cubic (fcc) NiZr2, thereby reducing the thermal stability of the Zr-Ni amorphous alloys. Extensive studies have proved that oxygen enhances the crystallization reaction in Zrbased amorphous alloys. For example, Lin et al (Lin, et al., 1997) reported for undercooled Zr– Ti–Cu–Ni–Al molten liquids that oxygen addition strongly affects crystal nucleation and can dramatically increase the necessary critical cooling rate for glass formation, thus limiting bulk glass formation and reducing the maximum attainable sample thickness. Over the range of oxygen content studied (300 – 5000 at. ppm), the time–temperature-transformation curves vary roughly by two orders of magnitude along the time axis. In other words, oxygen contamination ranging up to 0.5 at.% can increase the necessary cooling rate for glass formation by two orders of magnitude (Lin, et al., 1997). Köster et al. (Köster et al., 1996; Köster et al., 1997) reported the formation of an icosahedral phase during primary crystallization in Zr65Cu17.5Ni10Al7.5 amorphous alloys, whereas such a crystallization process was not reported in the same alloy composition by Zhang et al. (Zhang et al., 1991), indicating that the formation of quasicrystals is induced by the oxygen contamination in the alloy. Eckert and his coworkers (Eckert et al., 1998; Gebert et al., 1998) also reported the strong influence of the oxygen contamination on the crystallization kinetics and products in Zr65Cu17.5Ni10Al7.5 amorphous alloy, where supercooled liquid region decreases with increasing oxygen content due to the change in crystallization sequence from a single- to a double-step process. It was also shown that an fcc NiZr2 phase is formed at a higher oxygen level in the Zr–Cu–Ni–Al system. Therefore, oxygen contamination is of primary importance for the glass formation and crystallization behavior of Zr-based amorphous alloys.

Fig. 5. XRD patterns of the Zr65-*x*Cu27.5Al7.5O*<sup>x</sup>* bulk amorphous alloys with (a) x = 0.14 and (b) x = 0.82 after annealing at 673 K for different durations. Reprinted from (Murty, et al., 2000), with permission from Elsevier.

Murty et al. (Murty, et al., 2000) investigated the influence of oxygen on the crystallization behavior of melt-spun amorphous Zr65-xCu27.5Al7.5Ox (x = 0.14, 0.43 and 0.82) ribbons. DSC

(Altounian et al., 1987) showed that the oxygen induces the formation of metastable facecentered cubic (fcc) NiZr2, thereby reducing the thermal stability of the Zr-Ni amorphous alloys. Extensive studies have proved that oxygen enhances the crystallization reaction in Zrbased amorphous alloys. For example, Lin et al (Lin, et al., 1997) reported for undercooled Zr– Ti–Cu–Ni–Al molten liquids that oxygen addition strongly affects crystal nucleation and can dramatically increase the necessary critical cooling rate for glass formation, thus limiting bulk glass formation and reducing the maximum attainable sample thickness. Over the range of oxygen content studied (300 – 5000 at. ppm), the time–temperature-transformation curves vary roughly by two orders of magnitude along the time axis. In other words, oxygen contamination ranging up to 0.5 at.% can increase the necessary cooling rate for glass formation by two orders of magnitude (Lin, et al., 1997). Köster et al. (Köster et al., 1996; Köster et al., 1997) reported the formation of an icosahedral phase during primary crystallization in Zr65Cu17.5Ni10Al7.5 amorphous alloys, whereas such a crystallization process was not reported in the same alloy composition by Zhang et al. (Zhang et al., 1991), indicating that the formation of quasicrystals is induced by the oxygen contamination in the alloy. Eckert and his coworkers (Eckert et al., 1998; Gebert et al., 1998) also reported the strong influence of the oxygen contamination on the crystallization kinetics and products in Zr65Cu17.5Ni10Al7.5 amorphous alloy, where supercooled liquid region decreases with increasing oxygen content due to the change in crystallization sequence from a single- to a double-step process. It was also shown that an fcc NiZr2 phase is formed at a higher oxygen level in the Zr–Cu–Ni–Al system. Therefore, oxygen contamination is of primary importance for the glass formation and

Fig. 5. XRD patterns of the Zr65-*x*Cu27.5Al7.5O*<sup>x</sup>* bulk amorphous alloys with (a) x = 0.14 and

Murty et al. (Murty, et al., 2000) investigated the influence of oxygen on the crystallization behavior of melt-spun amorphous Zr65-xCu27.5Al7.5Ox (x = 0.14, 0.43 and 0.82) ribbons. DSC

(b) x = 0.82 after annealing at 673 K for different durations. Reprinted from

(Murty, et al., 2000), with permission from Elsevier.

crystallization behavior of Zr-based amorphous alloys.

results showed that the *T*g increases with the addition of oxygen. The base alloy containing the lowest amount of oxygen (x = 0.14) crystallizes in a single step. The addition of oxygen significantly decreases the width of supercooled liquid region (∆*T*x) from 85 K for x = 0.14 to 58 K for x = 0.82. The decrease in ∆*T*x is partly due to the increase in *T*g with increasing oxygen and also due to the appearance of a pre-crystallization peak in the oxygencontaining alloys before the main crystallization event. The base alloy heated to 673 K (in supercooled liquid region) shows an amorphous nature. In the x = 0.82 alloy, precipitation of spherical icosahedral particles in nanocrystalline state was observed within 10 min annealing at 673 K. Then it crystallizes to Zr2(Cu,Al) when heated to 723 K. No other phase is present in the alloy heated up to 753 K. The x = 0.82 alloy heated to 673 K has a similar trend to that of the base alloy. However, the alloy heated near the pre-crystallization peak (708 K) led to the presence of an icosahedral phase along with a small amount of Zr2(Cu,Al). Only Zr2(Cu,Al) is present in the alloy heated to 723 and 753 K. The base alloy with x = 0.14 remains amorphous for up to 10 min at 673 K, after which the formation of Zr2(Cu,Al) was observed. The XRD patterns in Fig. 5 (a) clearly indicate the formation of Zr2(Cu,Al) beyond 10 min of annealing at 673 K. No other phase was identified even after a longer annealing for 60 min (Fig. 5 (a)). XRD patterns of the alloy heat treated at 673 K (Fig. 5 (b)) show that the icosahedral phase starts forming after 10 min and persists for up to 15 min, beyond which the icosahedral phase transforms to Zr2(Cu,Al).

The mechanism of the oxygen-induced precipitation of metastable fcc Zr2(Cu,Al) and icosahedral quasicrystalline phases is rationalized by considering the effect of oxygen on the nucleation process. The high thermal stability of multicomponent Zr-based amorphous alloys is generally attributed to the difficulty of precipitation of crystalline compounds from the undercooled liquid. The combination of elements with significantly different atomic sizes and negative enthalpies of mixing leads to a homogeneously mixed dense random packed structure of the liquid resulting in a large liquid–solid interface energy (Inoue, 2000). If the nucleating phase has a different composition with respect to the homogeneous undercooled liquid, then the nucleation of the phase requires substantial atomic rearrangement (Eckert, et al., 1998). The driving force for the polymorphous crystallization is ∆*G*total. However, if the icosahedral phase is stabilized by oxygen addition, the driving force for the primary crystallization of the icosahedral phase can be comparable to or higher than that for the polymorphous crystallization, ∆*G*total. In such a case, icosahedral phase would initially precipitate from the amorphous matrix by the primary crystallization. The free energy reduction is accompanied with this crystallization and there is still a driving force to form Zr2(Cu,Al) from the icosahedral phase and/or the remaining amorphous phase in the second stage. The formation of the icosahedral phase would be preferable if the driving forces for the polymorphous crystallization and the primary crystallization are comparable, because it is believed that icosahedral clusters are present in the amorphous phase, and these would act as nuclei for the icosahedral phase primary crystals. If such icosahedral clusters are stabilized by the presence of oxygen, the oxygen-enriched alloy would be favour to form an icosahedral phase by primary crystallization. The differences in the sequence of the phase formation in these alloys with x = 0.14 and 0.82 are illustrated schematically in Fig. 6, in which the darkness of the gray scale corresponds to the concentration of Zr. In the base Zr–Cu–Al and Zr–Cu amorphous alloys, crystallization proceeds by polymorphous reaction without change in composition. On the other hand, in the oxygen-containing ternary alloys, the first stage of crystallization occurs by primary

Crystallization Behavior and Control of Amorphous Alloys 195

Melt spinning 686 744 58

Melt spinning 711 754 43

Pd81Si19 (fluxed)† Air cooling 638 696 58 (Yao & Ruan, 2005) Melt spinning 633 675 42

Melt spinning 428 469 41

(6 mm∅ rod) 428 469 41

(10 mm∅ rod) 428 469 41

(Inoue, et al., 1997) Pd40Cu30Ni10P20

Melt spinning 622 749 127

(9 mm∅ rod) 738 795 57

Melt spinning 480 550 90

Table 1. Comparison of the transformation temperatures determined from DSC at heating rate is at 0.67 K/s (if not indicated) for some typical amorphous alloys prepared by different methods. *T*g: glass transition temperature; *T*x: onset crystallization temperature; ∆*T*x: the

Fig. 7 compares the DSC curves for the [(Fe0.8Co0.2)0.75B0.2Si0.05]96Nb4 bulk amorphous alloy rods with different diameters up to 2.5 mm with the data for melt-spun ribbon of the same composition. No appreciable difference is recognized in the transformation temperatures or

(30 μm thick ribbon) 805 820 15

(K)

*T*x (K)

Cu-mold casting 675 732 57 (Inoue et al., 2005)

(2.5 mm∅ rod) 714 758 44 (Jiang et al., 2003a)

(7 mm∅ rod) 576 678 102 (He et al., 1996) Melt spinning 590 671 91 (Inoue, et al., 1997)

(16 mm∅ rod) 625 750 125 (Inoue et al., 1993b)

(Illeková et al., 1997) Water quenching

casting (9 mm∅ rod) 460 527 67 (Inoue et al., 1993a)

Ball Milling 652 717 65 (Zhang, et al., 2005a) Melt spinning 675 739 64

Ball Milling 705 771 66 (Zhang, et al., 2005a) Melt spinning 721 789 68

*∆T*<sup>x</sup>

(K) Reference

(Kang et al., 2000)

Composition Synthesis route *T*<sup>g</sup>

Cu-mold casting

Water quenching

Injection casting

Squeeze casting

(fluxed) Melt spinning 572 670 98

(unfluxed) Melt spinning 572 663 91

Water quenching

Planar flow casting

High-pressure die

width of supercooled liquid region, which is equal to *T*x – *T*g.

Cu50Zr50

Cu60Zr30Ti10

Pd40Ni40P20 (fluxed)†

Mg65Cu15Y10Ag10

Pd40Cu30Ni10P20

Zr65Al7.5Ni10Cu17.5

Zr55Ni25Al20

La55Al25Ni10Cu10

Ti50Cu35Ni12Sn3

Ti50Cu18Ni22Al4Sn6

† heating rate is 0.33 K/s.

crystallization of the icosahedral phase. The icosahedral phase is enriched in Zr and O and depleted in Cu and Al. When Zr2(Cu,Al) precipitates peritectically, the concentration of the Zr2(Cu,Al) particles becomes the same as the initial alloy composition. When this reaction is complete, only the grains of single-phase Zr2(Cu,Al) remain.

Fig. 6. Schematic diagrams showing the evolution of microstructure during crystallization of Zr65-*x*Cu27.5Al7.5Ox amorphous with (a) *x* = 0.14 and (b) *x* = 0.82. Reprinted from (Murty, et al., 2000), with permission from Elsevier.

#### **3.3 Effect of sample preparation method**

In general, when the transformation temperatures (e.g. *T*g and *T*x, etc) of an amorphous alloy are measured by DSC, there is no appreciable difference between in the amorphous samples prepared by direct melt cooling from molten liquid (e.g. by melt spinning, casting, water quenching, etc). Table 1 summarizes the transformation temperatures determined from DSC for some typical amorphous alloys prepared by different routes. Furthermore, there is no difference in the transformation temperatures of the amorphous rods with different sizes. As seen from Table 1, same transformation temperatures are obtained in the Mg65Cu15Y10Ag10 amorphous rods in 6 mm diameter prepared by injection casting and in 10 mm diameter prepared by squeeze casting.

crystallization of the icosahedral phase. The icosahedral phase is enriched in Zr and O and depleted in Cu and Al. When Zr2(Cu,Al) precipitates peritectically, the concentration of the Zr2(Cu,Al) particles becomes the same as the initial alloy composition. When this reaction is

Fig. 6. Schematic diagrams showing the evolution of microstructure during crystallization of Zr65-*x*Cu27.5Al7.5Ox amorphous with (a) *x* = 0.14 and (b) *x* = 0.82. Reprinted from (Murty, et

In general, when the transformation temperatures (e.g. *T*g and *T*x, etc) of an amorphous alloy are measured by DSC, there is no appreciable difference between in the amorphous samples prepared by direct melt cooling from molten liquid (e.g. by melt spinning, casting, water quenching, etc). Table 1 summarizes the transformation temperatures determined from DSC for some typical amorphous alloys prepared by different routes. Furthermore, there is no difference in the transformation temperatures of the amorphous rods with different sizes. As seen from Table 1, same transformation temperatures are obtained in the Mg65Cu15Y10Ag10 amorphous rods in 6 mm diameter prepared by injection casting and in 10

al., 2000), with permission from Elsevier.

**3.3 Effect of sample preparation method** 

mm diameter prepared by squeeze casting.

complete, only the grains of single-phase Zr2(Cu,Al) remain.


† heating rate is 0.33 K/s.

Table 1. Comparison of the transformation temperatures determined from DSC at heating rate is at 0.67 K/s (if not indicated) for some typical amorphous alloys prepared by different methods. *T*g: glass transition temperature; *T*x: onset crystallization temperature; ∆*T*x: the width of supercooled liquid region, which is equal to *T*x – *T*g.

Fig. 7 compares the DSC curves for the [(Fe0.8Co0.2)0.75B0.2Si0.05]96Nb4 bulk amorphous alloy rods with different diameters up to 2.5 mm with the data for melt-spun ribbon of the same composition. No appreciable difference is recognized in the transformation temperatures or

Crystallization Behavior and Control of Amorphous Alloys 197

Ti50Cu18Ni22Al4Sn6 alloys prepard by ball-milling (BM) and melt-spinning (MS). The broad diffuse maximum for the amorphous phase formed by BM is determined to be 26.89 nm-1 for Ti50Cu35Ni12Sn3 and 26.64 nm-1 for Ti50Cu18Ni22Al4Sn6, respectively. They are well in agreement with the values of the amorphous alloys prepared using MS method, *Q*p = 26.81 nm-1 for Ti50Cu35Ni12Sn3 and 26.67 nm-1 for Ti50Cu18Ni22Al4Sn6, respectively. It implies that for a given alloy, the amorphous phase obtained using the different preparation methods is very similar in the all cases. However, both *T*g and *T*x of the BM alloy shift towards a lower temperature, with respect to the MS alloys, by about 20 K for Ti50Cu35Ni12Sn3 and 16-18 K for Ti50Cu18Ni22Al4Sn6, respectively, even though a very close ∆*T*x is obtained in the BM and MS amorphous alloys for each phases. Furthermore, the heat of crystallization in the BM amophous state is slightly lower than that in MS one for both alloys. The difference in the transformation temperaturees betwen the BM and MS amorphous phase is likely caused by the minor difference in the composition, oxygen content, and/or short-range order in the

Fig. 8. (a) XRD patterns and (b) DSC curves at a heating of 0.67 K/s for Ti50Cu35Ni12Sn3 and Ti50Cu18Ni22Al4Sn6 alloys prepard by ball-milling (BM) and melt-spinning (MS). Reprinted

A few work has investigated the effect of high pressure on the crystallization of amorphous alloys, e.g. see the references (Jiang, et al., 2000; Jiang, et al., 2002; Jiang, et al., 2003b; Ye & Lu, 1999; Zhuang et al., 2000). In general, the crystallization temperature of an amorphous alloy increases with increasing pressure. However, the rate and the range of such temperature increase are closely related to the alloy systems. Fig. 9 shows the pressure dependence of the crystallization temperatures (i.e. *T*x1 and *T*x2) for the Al89La6Ni5

amorphous phases formed by different processing route.

from (Zhang, et al., 2005a), with permission from Elsevier.

**3.4 Effect of pressure** 

crystallization process between the melt-spun ribbon and cast rod samples, in spite of an increase in Curie temperature (*T*c) with the increase of diameter. All samples exhibit a distinct glass transition at 830 K, followed by crystallization at 880 K, resulting in a large supercooled liquid region of 50 K. Similar results have been obtained in a number of bulk glass-forming alloy systems (Suryanarayana & Inoue, 2011).

Fig. 7. DSC curves at a heating of 0.67 K/s for [(Fe0.8Co0.2)0.75B0.2Si0.05]96Nb4 bulk amorphous alloy rods (1.5, 2 and 2.5 mm∅) as well as the melt-spun amorphous alloy ribbon of the same composition. Reprinted from (Inoue et al., 2004a) and (Suryanarayana & Inoue, 2011), with permission from Elsevier.

Note that, although no appreciable difference in transformation temperatures has been observed in the amorphous alloys prepared by direct melt cooling from molten liquid, some alloys do have shown some differences in the transformation temperatures in the amorphous ribbon and rod samples, even though they have an identical chemical composition. As seen from Table 1, the Zr55Ni25Al20 glassy alloys prepared by two different solidification methods (one is planar flow casting with cooling rate of about 105 K s-1 and the other is water quenching with a solidification rate of about 102 K s-1) showed a significant difference in the transformation temperatures, i.e. 69 K difference in *T*g and 15 K in *T*<sup>x</sup> (Illeková, et al., 1997). By comparing the enthalpy of structural relaxation in DSC curves and the full width at half maximum (FWHM) of the first diffuse peak in XRD patterns, it is concluded that the samples produced from both methods represent the same amorphous state, but the amorphous ribbon sample contains a higher degree of short-range order (SRO) (Illeková, et al., 1997).

A number of investigations have reported a distinct difference in the transformation temperatures between the amorphous alloys prepared by melt cooling and that formed by solid-state amorphization techniques (e.g. ball milling or mechanical alloying). Fig. 8 compares the structural feature and transformation temperatures for Ti50Cu35Ni12Sn3 and

crystallization process between the melt-spun ribbon and cast rod samples, in spite of an increase in Curie temperature (*T*c) with the increase of diameter. All samples exhibit a distinct glass transition at 830 K, followed by crystallization at 880 K, resulting in a large supercooled liquid region of 50 K. Similar results have been obtained in a number of bulk

Fig. 7. DSC curves at a heating of 0.67 K/s for [(Fe0.8Co0.2)0.75B0.2Si0.05]96Nb4 bulk amorphous alloy rods (1.5, 2 and 2.5 mm∅) as well as the melt-spun amorphous alloy ribbon of the same composition. Reprinted from (Inoue et al., 2004a) and (Suryanarayana & Inoue, 2011),

Note that, although no appreciable difference in transformation temperatures has been observed in the amorphous alloys prepared by direct melt cooling from molten liquid, some alloys do have shown some differences in the transformation temperatures in the amorphous ribbon and rod samples, even though they have an identical chemical composition. As seen from Table 1, the Zr55Ni25Al20 glassy alloys prepared by two different solidification methods (one is planar flow casting with cooling rate of about 105 K s-1 and the other is water quenching with a solidification rate of about 102 K s-1) showed a significant difference in the transformation temperatures, i.e. 69 K difference in *T*g and 15 K in *T*<sup>x</sup> (Illeková, et al., 1997). By comparing the enthalpy of structural relaxation in DSC curves and the full width at half maximum (FWHM) of the first diffuse peak in XRD patterns, it is concluded that the samples produced from both methods represent the same amorphous state, but the amorphous ribbon sample contains a higher degree of short-range order (SRO)

A number of investigations have reported a distinct difference in the transformation temperatures between the amorphous alloys prepared by melt cooling and that formed by solid-state amorphization techniques (e.g. ball milling or mechanical alloying). Fig. 8 compares the structural feature and transformation temperatures for Ti50Cu35Ni12Sn3 and

glass-forming alloy systems (Suryanarayana & Inoue, 2011).

with permission from Elsevier.

(Illeková, et al., 1997).

Ti50Cu18Ni22Al4Sn6 alloys prepard by ball-milling (BM) and melt-spinning (MS). The broad diffuse maximum for the amorphous phase formed by BM is determined to be 26.89 nm-1 for Ti50Cu35Ni12Sn3 and 26.64 nm-1 for Ti50Cu18Ni22Al4Sn6, respectively. They are well in agreement with the values of the amorphous alloys prepared using MS method, *Q*p = 26.81 nm-1 for Ti50Cu35Ni12Sn3 and 26.67 nm-1 for Ti50Cu18Ni22Al4Sn6, respectively. It implies that for a given alloy, the amorphous phase obtained using the different preparation methods is very similar in the all cases. However, both *T*g and *T*x of the BM alloy shift towards a lower temperature, with respect to the MS alloys, by about 20 K for Ti50Cu35Ni12Sn3 and 16-18 K for Ti50Cu18Ni22Al4Sn6, respectively, even though a very close ∆*T*x is obtained in the BM and MS amorphous alloys for each phases. Furthermore, the heat of crystallization in the BM amophous state is slightly lower than that in MS one for both alloys. The difference in the transformation temperaturees betwen the BM and MS amorphous phase is likely caused by the minor difference in the composition, oxygen content, and/or short-range order in the amorphous phases formed by different processing route.

Fig. 8. (a) XRD patterns and (b) DSC curves at a heating of 0.67 K/s for Ti50Cu35Ni12Sn3 and Ti50Cu18Ni22Al4Sn6 alloys prepard by ball-milling (BM) and melt-spinning (MS). Reprinted from (Zhang, et al., 2005a), with permission from Elsevier.

#### **3.4 Effect of pressure**

A few work has investigated the effect of high pressure on the crystallization of amorphous alloys, e.g. see the references (Jiang, et al., 2000; Jiang, et al., 2002; Jiang, et al., 2003b; Ye & Lu, 1999; Zhuang et al., 2000). In general, the crystallization temperature of an amorphous alloy increases with increasing pressure. However, the rate and the range of such temperature increase are closely related to the alloy systems. Fig. 9 shows the pressure dependence of the crystallization temperatures (i.e. *T*x1 and *T*x2) for the Al89La6Ni5

Crystallization Behavior and Control of Amorphous Alloys 199

crystallization temperature is expected, as observed for the Al89La6Ni5 amorphous alloy in

In addition to the aforementioned factors, the crystallization temperature(s) of an amorphous phase significantly depend on the heating rate used in DSC measurement (Kissinger, 1957). In contrast, the heating rate has a slight influence on the glass transition temperature. All these transformation temperatures of amorphous alloys increase with increasing the heating rate that is used in DSC. Therefore, the heating rate is usually

The kinetics of crystallization of amorphous alloys has been extensively studied by using differential scanning calorimetry (DSC) or differential thermal analysis (DTA), as the structural change in a material upon heating or cooling is indicated by a defection or peak in the DSC/DTA curve. The kinetic behavior associated with a structural change leading to an alternative metastable state in an amorphous alloy above its glass transition is a key subject since it provides new opportunities for structural control by innovative design and processing strategies. *Section 5* will show some application examples by controlling crystallization from amorphous precursors in order to tailor microstructure for excellent properties. Such crystallization control requires fundamental understanding of the specific

In general, crystallization is a thermally activated reaction, either by isothermal or

where *α* is the fraction transformed. The temperature dependent function is generally

0 *k k ( E / RT )* = − exp (3)

where *k*<sup>0</sup> is the reaction constant, *R* is the gas constant and *E* is the activation energy. In general, the reaction function *f(*α*)* is unknown. From the above equations it follows that for transformation studies by performing studies at a constant temperature *T, E* can be obtained

( ) *f i ln t E / RT c* = + (4)

where *t*f is the time needed to reach a certain fraction transformed, and *c*<sup>i</sup> is a constant, which depends on the reaction stage and on the kinetic model. Thus, *E* can be obtained from two or more experiments at different *T*. For isothermal experiments *k(T)* is constant, the determination of *f(*α*)* is relatively straightforward, and is independent of *E*. For nonisothermal experiments, the reaction rate at all times depends on both *f (*α*)* and *k (T)*, and the determination of *f(*α*), k0* and *E* (the so-called kinetic triplet) is an interlinked problem. A deviation in the determination of any of the three parameters will cause a deviation in the other parameters of the triplet. Over the past decades a variety of non-isothermal methods have been proposed. Among them, the Kissinger method (Kissinger, 1957) is widely used in

 α

= *f* ( )( ) *k T* (2)

isochronal heating. The transformation rate during a reaction could be described as

α

indicated when describing the transformation temperatures of an amorphous phase.

the pressure range of 1–4 GPa in Fig. 9.

**4. Kinetics of crystallization** 

mechanisms influencing structural transformations.

*d / dt*

as below:

assumed to follow an Arrehnius type dependency

amorphous alloy. Both *Tx*1 and *Tx*2 firstly decrease with the increase in pressure in the range of 0–1 GPa and then increase with pressure increasing up to 4 GPa. Such changes in crystallization temperature with pressure is related to the competing process between the thermodynamic potential barrier and the diffusion activation energy under pressure (Zhuang, et al., 2000).

Fig. 9. Pressure dependence of the crystallization temperatures for Al89La6Ni5 amorphous alloy. Reprinted from (Zhuang, et al., 2000), with permission from American Institute of Physics.

Crystallization of an amorphous alloy is normally regarded as a process proceeding by nucleation and subsequent growth of crystals. During the initial stage of nucleation of crystals in the amorphous phase, the effect of pressure on the crystallization kinetics is associated with the atomic diffusion process and the volume change effect. The crystallization temperature(s) of an amorphous alloy may be governed by the thermodynamic potential barrier of nucleation and diffusion activation energy. According to crystallization kinetics theory, the nucleation rate *I* can be written as,

$$I = I\_0 \exp\{- (\Delta G^\circ + Q\_n) / kT\} \tag{1}$$

where *I*0 is a constant, ∆G\* is the free energy required to form a nucleus of the critical size, i.e., the thermodynamic potential barrier of nucleation, *Q*n is the activation energy for the transport of an atom across the interface of an embryo, and *k* is the Boltzmann's constant. The sum ∆G\* + *Q*n is called the nucleation work.

In the Al89La6Ni5 alloy, ∆*G*\* is much larger than *Q*n and the dominant factor at low pressures (0–1 GPa). Thus, the nucleation work decreases with increasing pressure, leading to an enhancement of nucleation rate *I* and a reduction of the crystallization temperature with increasing pressure, as shown in Fig. 9. With increasing pressure, ∆*G*\* rapidly decreases while *Q*n increases, resulting in atomic diffusion a dominant factor in the nucleation process. Hence, the nucleation work ∆*G*\* + *Q*n increases with increasing pressure. Consequently, nucleation rate *I* decreases with the increase in pressure and an enhancement of crystallization temperature is expected, as observed for the Al89La6Ni5 amorphous alloy in the pressure range of 1–4 GPa in Fig. 9.

In addition to the aforementioned factors, the crystallization temperature(s) of an amorphous phase significantly depend on the heating rate used in DSC measurement (Kissinger, 1957). In contrast, the heating rate has a slight influence on the glass transition temperature. All these transformation temperatures of amorphous alloys increase with increasing the heating rate that is used in DSC. Therefore, the heating rate is usually indicated when describing the transformation temperatures of an amorphous phase.
