Thermal analyses

4.6.1 Non-isothermal analyses and kinetics

In order to understand the origin of the high thermal stability and excellent glass-forming ability, it is very important to clarify the crystallization behavior of the supercooled liquid. By using the DSC instrument, it is sensitive to detect the glass transition and the nanocrystallization.

In the Mg65Cu25-xY10Bx alloy systems, from the Figs. 4.10 to 4.13, all the samples exhibit a distinct glass transition, followed by a wide supercooled region and then exothermic reactions due to crystallization. Figure 4.14 shows with increasing x from 0 to 3, the glass transition temperature is nearly constant at 410 K but the crystallization temperature increases slightly from 468 to 476 K. Therefore, the ∆Tx value also slightly increases from 58 K at x = 0 to 66 K at x = 3. With further increasing x to the 5 and 10 at%, ∆T_x becomes smaller with the addition of B content, 50 K for 5 at% B and 45 K for 10 at%. Figure 4.15 shows the variations of T_g, T_x against the B content. It is suggested that the B substitution for Cu is of no harm for the GFA of Mg-Cu-Y amorphous alloy.

In contrast, the shape of the Mg₆₅Cu₂₅Y_10-xB_x alloy systems is different, as shown in Fig.

4.16. With increasing x from 0 to 5, it is noticed that the first exothermic peak for x = 0 splits into two exothermic peaks for alloys with x＞3. Furthermore, Form Fig. 4.17, the shape of these two curves is greatly different. The Mg65Cu22Y10B3 alloy, only one sharp exothermic peak with high enthalpy of transformation is observed. This is due to the fact that this alloy is presumed to have a higher degree of dense random packed structure. It indirectly indicates the different atomic arrangement between the Mg₆₅Cu_25-xY₁₀B_x and Mg₆₅Cu₂₅Y_10-xB_x systems.

The melting behavior of Mg-based low melting alloys is also determined using the DSC at a heating rate of 20 K/min. The ratio of the glass transition temperature Trg and the offset melting temperature (liquidus) T_l, is often used as a parameter to estimate the GFA. From the former study ^[19], Mg65Cu25Y10 exhibited a single endothermic peak with a narrow melting range about 30 K. The onset and offset temperatures of the melting designated by T_m^solid and Tmliquid

are 728 and 760 K, respectively. It indicates that the Mg65Cu25Y10 alloy is close to the ternary eutectic composition.

Compared with the parent alloy, from Fig. 4.18, the T_m^solid of the x = 3 and 5 alloys decrease to about 715 and 712 K, and Tmliquid

also decrease to about 738 and 736 K. Although the x =3 alloy has two endotherm peaks, the lower melting point and small melting interval (<30 K) indicate that the replacement of Cu by B would result in the significant narrowing of supercooled liquid region and enhance the GFA of Mg-Cu-Y amorphous alloys. When further increasing the boron content x to 5 at%, the single endothermic peak and the narrow melting interval demonstrate that Mg₆₅Cu₂₀Y₁₀B₅ alloy may be very near to the quaternary eutectic composition.

4.6.2 Kissinger plots for non-isothermal analyses

In non-isothermal analyses, the dependency of crystallization temperature on heating rate can be used to estimate the activation energy of crystallization by means of the Kissinger peak shift method. Figures 4.19 to 4.20 show the Kissinger plots of ln(φ/Tx) against reciprocal Tx

taken from the dependence of crystallization temperature on various heating rates (φ = 10, 20, 30 K/min) in the DSC curves (T_x is referred to the onset crystallization temperature). The activation energy of the first crystallization for the Mg65Cu25Y10 amorphous alloy derived from the slop of Kissinger plot is 138 kJ/mol, as shown in Fig. 4.19 (a). As a comparison, the activation energy values for Mg65Cu25-xY10Bx (x = 3, 5, 10) amorphous ribbons evaluated by the slope of this Kissinger plot, as shown in Figs. 4.19 to 4.20, are 156, 149 and 150 kJ/mol, respectively. These values are about 10% higher than the parent alloy. It indicates that the substitution of Cu by B improves the thermal stability for the Mg-based amorphous alloys.

Furthermore, some researchers use the shift of crystallization temperature Tp (Tp is referred to peak temperature of the crystallization) to yield the activation energy of crystallization (E_xp) by using the Kissinger equation. Pryds et al. also used this method to determine the activation energy of crystallization for Mg₆₀Cu₃₀Y₁₀ ^[53] and Mg-Cu-Y-Al ^[54] amorphous alloys. The activation energy of the first crystallization peak is approximately 1.6 eV ( 154 kJ/mol). It is interesting to note that the activation energy for the first phase transition (Mg

≈

2Cu) in Mg-based multicomponent alloys is the same as found by Sommer et al. ^[55] for the binary amorphous alloys Mg₇₈Cu₂₂ alloy which has E_xp =1.58 eV (≈152 kJ/mol). It implies that the process of the primary crystallization is similar in the binary, ternary and quaternary Mg-based amorphous alloys.

4.6.3 Modified Kissinger plots for non-isothermal analyses

The non-isothermal DSC thermogram also has been used to ascertain the mechanism and kinetics of first crystallization by the method of modified Kissinger equation. The fraction of crystallization could be calculated by integrating the area under the exothermal peak from the non-isothermal DSC curves, namely,

∫

= _Tx

∫

onset Tx

T

onst Tx

H(T)dT H(T)dT

X , (24)

where X is the volume fraction of crystallization, H(T) is the heat as function of temperature.

For the evaluation of the parameter n, ln[-ln(1-X)] is plotted as a function of lnφ. Figures 4.21 and Fig. 4.22 show the variation of n which is temperature dependence. Theoretically, n should not exceed 4 (i.e., the bulk nucleation and three dimensional growth). In the early crystallization stage, because nuclei are in random distribution, the n values extracted from the plots for the initial transient stage will exceed 4. The similar high values of n (n = 6) also has been reported for a ternary chalcogenide glass ^[56]. Since the n value is much larger than 1, bulk nucleation should be dominant in the early stage of phase transformation. With increasing temperature, the n value decreases to 1 and surface nucleation dominates the later stage of transformation until only growth process dominates.

Figures 4.23 to 4.26 are the plots of ln[-ln(1-X)] against the 1/T for several heating rates.

From the slope, -1.052mQ/R can be obtained. Since no particular heat treatment was given to nucleate the samples before thermal analysis, the dimensionality of growth parameter m is taken to be equal to (n - 1) in the early crystallization stage and m is equal to n when a large number of nuclei exist in the final crystallization stage. From the view of the slope, it is apparent that the early crystallization stage has a nearly linear and steep slope (means higher

activation energy for crystallization) and then suddenly decreases when pass through a break in the slope.

This break in the slope above a specific volume fraction can be observed for each heating rate. From Lin and Shen’s research ^[57] infer that the slope breaks on the ln[-ln(1-X)] vs. 1/T plot is the saturation of nucleation sites. Below the break temperature, the activation energy includes the nucleation activation energy and growth activation energy. On the contrary, due to the saturation of nucleation, only the growth activation energy remains above the break temperature. This inference is also consistent with the slope observed.

Furthermore, it is noted that the slope breaks at nearly the same transformed volume fraction X≈0.6 (ln[-ln(1-X)] 0) for all the Mg-based amorphous alloys in Figs. 4.23 to 4.25.

It means the nucleation behavior is similar in all Mg-based amorphous alloys.

≈

4.6.4 Isothermal analyses and kinetics

The samples of the Mg-based amorphous alloys are annealed isothermally at 433, 435, 438, and 440 K between Tg and Tx. Figures 4.27 and 4.28 show the results of the isothermal calorimetry for the Mg₆₅Cu₂₅Y₁₀ and Mg₆₅Cu₂₂Y₁₀B₃ amorphous alloys. It cannot be considered as one primary crystallization for the Mg65Cu22Y10B3 alloy, where the multi-exothermic peaks are observed. In order to follow the JMA assumption, the multi-exothermic reaction is separated by using the PeakFit software. As shown in Figs. 4.29 to 4-30, the modified exothermic peaks all exhibit a symmetric shape.

Figure 4.31 and Fig. 4.32 show the plot that the degree of transformation as a function of time for different Mg-based amorphous alloys. With increasing annealing temperature, the

incubation time and the time for a complete crystallization would both decrease. The incubation time at the annealing temperature of 433, 435, 438, and 440 K for the Mg65Cu25Y10

amorphous alloy are 1420, 810, 580 and 373 s and for the Mg₆₅Cu₂₂Y₁₀B₃ amorphous alloys are 1568, 1031, 778 and 595 s, respectively. the latter alloy always exhibit longer incubation time as compared with the parent alloy.

Furthermore, for the evaluation of the Avrami exponent n, ln[-ln(1-X)] is plotted versus ln(t) for different annealing temperatures. Figures 4.33 to 4.34 reveal that the Avrami exponent n of Mg₆₅Cu₂₅Y₁₀ and Mg₆₅Cu₂₂Y₁₀B₃ amorphous alloys are temperature dependent during the isothermal crystallization process. Form the slopes of the straight lines, the average value of n for Mg₆₅Cu₂₅Y₁₀ is about 3.5. It means that the crystallization is mainly controlled by three dimensional nuclei with constant growth rate until the whole amorphous phase is completely crystallized. In the Mg₆₅Cu₂₂Y₁₀B₃ amorphous alloy, the average value of n is also about 3.2. The results indicate that the Mg-based amorphous alloys exhibit similar crystallization behavior .

In isothermal analyses, the activation energy can also be evaluated by the isothermal curves of crystallized volume fraction versus annealing time at different temperatures. This is carried out by using the Arrhenius equation, t = t_o exp(Q/RT). For each annealing temperature, the time to reach the integer multiple of 10% crystallization is selected and the plots of lnt versus1/T for Mg-based amorphous alloys are made, as shown in Figs. 4.35 and 4.36. From the slopes, the average activation energies determined for the first crystalline phase (Mg2Cu) of the Mg₆₅Cu₂₅Y₁₀ and Mg₆₅Cu₂₂Y₁₀B₃ amorphous alloys are 156 and 210 kJ/mol, respectively. This indicates that the boron can increase the activation energy of crystallization for the Mg₆₅Cu₂₅Y₁₀ based alloy and results in the improvement of the thermal stability against crystallization. The activation energy of the second crystalline phase for the

Mg₆₅Cu₂₂Y₁₀B₃ which are also determined by the Arrhenius equation are 200 kJ/mol, as shown in the Fig. 4.37.