Microstructure formation and electrical resistivity behavior of rapidly solidified Cu-Fe-Zr immiscible alloys

1. Introduction

Alloys with a metastable miscibility gap in the range of the undercooled liquid state have been of increasing interest [[1], [2], [3], [4], [5], [6]]. Many of them are excellent candidates for industrial applications. The Cu-Fe alloy, as a kind of high strength and high electric conductivity materials [[7], [8], [9], [10], [11]], is a typical peritectic system. From the Cu-Fe binary alloy diagram [12,13], it has a nearly flat liquidus line and a retrograde solidus line and hence displays a thermodynamic tendency toward the formation of a miscibility gap in undercooled liquid state. The effect of the cooling rate, undercooling degree and gravity level on the microstructure evolution of the Cu-Fe system was studied through experiments such as gas atomization [[14], [15], [16]], containerless solidification [17,18] and drop tube [19]. Novel structure with self-organized core/shell was detected in these powders, which are promising solder balls for modern electronic packaging technology. Much attention has been paid to the microstructure formation to clarify the kinetics of metastable phase separation in the undercooled binary liquid Cu-Fe alloys [[20], [21], [22], [23], [24], [25]]. It was indicated that the formation of such structure are attributed to the droplets spacial migration and coagulation and the wetting behavior [[14], [15], [16],26,27]. More recently, Li and co-worker [22,23] analyzed the mechanism of the secondary liquid-liquid phase separation (LLPS) in the Cu₈₀Fe₂₀ alloy subjected different cooling conditions, and examined soft ferromagnetic characteristics of the Cu₈₀Fe₂₀ alloy.

The metastable miscibility gap of the binary Cu-Fe system may be extended into some ternary and multicomponent systems [28,29]. For some ternary Cu-Fe-X (X = Pb, Si, C, etc.) systems, the third addition can enhance the LLPS [30,31]. The liquid-liquid hierarchical separation phenomenon was detected in the ternary Cu-Fe-Pb alloy, having new potential applications in the separation and recovery of the secondary complex metal resource of printed circuit boards [32,33]. In contrast, a metastable LLPS into Fe-rich and Cu-rich liquids was examined for the Cu-Fe-Z (Z = Ni, Co, Cr, Mn, etc.) alloy [34]. The phenomenon of phase separation was utilized to enhance their mechanical properties of some multicomponent systems (i.e., high-entropy alloys [35]). The elements Cu, Fe and Zr are the basic constituents for some engineering materials [[36], [37], [38], [39]] such as high-strength and high-conductivity copper alloys, glass-forming alloy systems, etc. As known, enthalpies of mixing (ΔH_mix) of Cu-Fe, Cu-Zr, and Fe-Zr are +12 kJ mol^-1, -23 kJ mol^-1 and -25 kJ mol^-1, respectively [40]. The negative and positive enthalpies mean strong attractive and repulsive interactions between atoms, and may lead to changes of local structure [[41], [42], [43], [44]], which influence the properties of such materials. In this work, the microstructure formation of the rapidly solidified Cu-Fe-Zr ternary alloys was studied. The effect of Zr content, atomic ratio of Cu to Fe, and cooling rate on the structure of the Cu-Fe-Zr ternary system has been clarified. What’s more, an abnormal change of electrical resistivity upon temperature in the as-quenched Cu-Fe-Zr alloys was detected for the first time. The crystallization behavior has been discussed in detail. It is of significance in understanding the property of complex multicomponent Cu-Fe-Zr-based metal materials.

2. Material and methods

The ingots of master alloys were prepared from Cu, Fe and Zr metals with purity better than 99.95% by arc melting under a Ti-gettered purified argon atmosphere in water-cooled copper crucible. The alloy ingots were melted at least four times to ensure homogeneous. The 5-10 g master alloy was remelted in a quartz-glass tube with a 0.7 mm orifice and ejected through a nozzle onto the copper wheel with an overpressure of 50 kPa. Rapidly quenched ribbon was fabricated by single-roller melting-spinning method with the surface velocity of the copper roller is about 27 m s^-1. The cylindrical specimens with 2 mm in diameter were prepared by remelting the master alloy in quartz-glass tubes and subsequent ejecting of the melt with an overpressure of 50 kPa through a nozzle into copper mold.

X-ray diffraction (XRD) analysis was performed for the samples by using CuKa radiation for a 2θ range of 20°-80°. The microstructure of the rapidly solidified samples was characterized by scanning electron microscopy (SEM), transmission electron microscopy (TEM) linked with an STEM detector. At heating and cooling rates of 20 K min^-1, the thermal properties of the samples were analyzed by differential scanning calorimeter (DSC) under argon atmosphere. The change of the electrical resistivity upon temperature for the sample with dimension of about 1.5 mm × 1.0 mm × 0.2 mm was measured using a standard four-probe technique with a heating rate of 10 K min^-1.

3. Results and discussion

3.1. Correlations between microstructure and composition

3.1.1. Effect of atomic ratio of Cu to Fe

Fig. 1 shows the microstructures of the as-quenched (Cu_1-xFe_x)₉₀Zr₁₀ (x = 0.45, 0.5, 0.6) alloys. Obviously, the microstructure is correlated to the atomic ratio of Cu to Fe. When the Fe content x in (Cu_1-xFe_x)₉₀Zr₁₀ alloys were 0.45 and 0.5, the two-phase interconnected structure was observed, as shown by the bright and dark regions in Fig. 1(a) and (b). The EDS analysis reveals that the dark and bright regions are Fe-rich and Cu-rich phases, respectively. Average compositions Cu_32.13Fe_56.98Zr_10.89 and Cu_68.48Fe_23.53Zr_7.98 for the two regions in as-quenched (Cu_0.5Fe_0.5)₉₀Zr₁₀ alloys were detected, respectively, indicating that the solute element Zr dissolves in both Cu-rich and Fe-rich regions. From their microstructural characteristics, it can be concluded that the LLPS into Cu-rich and Fe-rich liquids occurs during the rapid cooling. In contrast, a particle-isolated structure forms in the as-quenched (Cu_0.4Fe_0.6)₉₀Zr₁₀ alloy sample, as shown in Fig. 1(c). The characteristic size and volume fraction estimated by standard stereological methods of Cu-rich particles are 0.2‒0.8 μm and 17.58% ± 0.2% (Fig. 1(d)). The miscibility gap of the binary Cu-Fe alloy can be extended to the ternary Cu-Fe-Zr alloy system. The LLPS can occur by two mechanisms, i.e., by liquid-state spinodal decomposition or by liquid-state nucleation and growth. The liquid-state spinodal decomposition normally results into the interconnected structure shown in Fig. 1(a) and (b). The particle-isolated microstructure in Fig. 1(c) is attributed to the LLPS by nucleation and growth. Comparing Fig. 1(a), (b) with Fig. 1(c) suggests that the microstructure of the as-quenched (Cu_1-xFe_x)₉₀Zr₁₀ alloys is tunable by changing the atomic ratio of Cu to Fe.

View original graphic| Download| PPT slide

Fig. 1.   SEM images of the as-quenched (a) (Cu_0.55Fe_0.45)₉₀Zr₁₀, (b) (Cu_0.5Fe_0.5)₉₀Zr₁₀, and (c) (Cu_0.4Fe_0.6)₉₀Zr₁₀ alloys, (d) size distribution of the particles in the as-quenched (Cu_0.4Fe_0.6)₉₀Zr₁₀ alloy.

3.1.2. Effect of Zr content

Fig. 2 shows the XRD patterns and DSC curves of the as-quenched (Cu_0.5Fe_0.5)_100-xZr_x (x = 10, 20, 40, 60, 80) alloys. For x = 10, the alloy sample has a completely crystalline structure and is mainly composed of Cu and Fe₂Zr phase. With the further increase of Zr content, an obvious broad diffraction peak appears in the XRD patterns of as-quenched (Cu_0.5Fe_0.5)₆₀Zr₄₀ alloys. Sharp peaks of crystalline phases superimpose on the broad scattering hump characteristic of an amorphous phase. When the Zr content x reaches 60, the XRD pattern only exhibits a broad diffraction maximum without any crystalline reflection, indicating that the structure of the as-quenched (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloy sample is fully amorphous. It indicates that an increase of Zr content can obviously improve the glass-forming ability (GFA) of (Cu_0.5Fe_0.5)_100-xZr_x system. The DSC scans of the as-quenched (Cu_0.5Fe_0.5)_100-xZr_x alloys are shown in Fig. 2(b). During the melting stage, it is clear that there are two endothermic reactions, i.e., the melting of Cu-rich phase and that of Fe-rich phase, as marked by the dotted frames in Fig. 2(b). The DSC curves of the as-quenched (Cu_0.5Fe_0.5)_100-xZr_x (x = 40, 60, 80) alloys exhibit obvious crystallization exothermic peak. It further verifies the existence of amorphous phase in these samples. The crystalline peaks have a left shift with the increase of the Zr content, which means a decrease of thermal stability.

View original graphic| Download| PPT slide

Fig. 2.   (a) XRD patterns and (b) DSC heating scans of as-quenched (Cu_0.5Fe_0.5)_100-xZr_x alloys.

Fig. 3(a) reveals the STEM image of the as-quenched (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloys. The (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloys exhibits a heterogeneous structure with the white Cu-rich nanoparticles embedded in the gray Fe-rich matrix. The characteristic size of the Cu-rich nanoparticle is 2-8 nm, approximating to a log-normal distribution (Fig. 3(b)). The population density and volume fraction of nanoparticles are estimated by standard stereological methods to be 5.37 × 10²⁴ m^-3 and 10.4% ± 0.2%. The selected-area diffraction pattern (SAED) shown by an inset in Fig. 3(a) indicates that the structure of the as-quenched (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloy is fully glassy. The HRTEM images indicate no evidence for nanocrystals (NCs) in any of the samples and it also shows that long strips of microstructure exists in addition to particle-isolated microstructure (see Fig. 3(c) and (d)).

View original graphic| Download| PPT slide

Fig. 3.   (a) STEM image of the as-quenched (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloy (the inset is the SAED pattern), (b) size distribution of the glassy nanoparticles in the as-quenched (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloy, (c) and (d) HRTEM images of the typical nanoparticles in the as-quenched (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloy, exhibiting different appearances.

Considering that the (Cu_1-xFe_x)₄₀Zr₆₀ alloy system has a better GFA, we further studied the effect of the atomic ratio of Cu to Fe on the structure of the as-quenched (Cu_1-xFe_x)₄₀Zr₆₀ alloys. Fig. 4 shows the XRD patterns of the as-quenched (Cu_1-xFe_x)₄₀Zr₆₀ (x = 0.3, 0.5, 0.6) alloys. Obviously, the GFA of the (Cu_1-xFe_x)₄₀Zr₆₀ alloy is preferable when atomic ratio of Cu to Fe is between 1:1 and 7:3. For (Cu_0.4Fe_0.6)₄₀Zr₆₀ alloys, sharp peaks of crystalline Fe₂Zr phases superimpose on the broad scattering hump characteristic of an amorphous phase. The position of broad peak in single-phase Zr-Cu and Zr-Fe amorphous alloys is almost identical, and therefore, there is only one broad peak in XRD patterns of Cu-Fe-Zr system, as shown in Fig. 4.

View original graphic| Download| PPT slide

Fig. 4.   XRD patterns of as-quenched (Cu_1-xFe_x)₄₀Zr₆₀, Zr₆₀Cu₄₀, and Zr₇₅Fe₂₅ alloys.

The TEM image of the as-quenched (Cu_0.7Fe_0.3)₄₀Zr₆₀ alloy is shown in Fig. 5. The as-quenched (Cu_0.7Fe_0.3)₄₀Zr₆₀ alloy exhibits an structure with the white Fe-rich spheres embedded in the gray Cu-rich matrix. The phase separation into the Cu-rich and Fe-rich liquids took place during the rapid cooling. The SAED presents halo diffraction intensity with a few diffraction spots, indicating the as-quenched (Cu_0.7Fe_0.3)₄₀Zr₆₀ alloy contains NCs. The examination indicates that the NCs are orignated from the black nanospheres shown in the Fig. 5. The diameter of the white Fe-rich spheres is 20-50 nm. Comparing the structure of the (Cu_0.7Fe_0.3)₄₀Zr₆₀ alloy with that of (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloy, we can conclude that the scale of phase separation and the GFA of the (Cu_1-xFe_x)₄₀Zr₆₀ system could be tunable by changing the atomic ratio of Cu to Fe.

View original graphic| Download| PPT slide

Fig. 5.   TEM image of as-quenched (Cu_0.7Fe_0.3)₄₀Zr₆₀ alloys.

3.2. Microstructure formation

From the viewpoint of thermodynamics, the liquid-solid phase transformation (LSPT) is possible if the difference Gibbs free energy (ΔG_SL) between the final solid state and initial liquid state is negative:

$\Delta G_{SL}=G_{S}-G_{L}<0$ (1)

where G_S and G_L are Gibbs free energy of the solid solution and liquid and can be described by:

$G_{S}=\sum_{i} 0_{G}x_{i}+RT \sum_{i} x_{i} ln x_{i}+E_{G}+\Delta G_{mag}$ (2)

$G_{L}=\sum_{i} 0_{G}x_{i}+RT \sum_{i} x_{i} ln x_{i}+E_{G}$ (3)

where R is the gas constant, T is the absolute temperature (K), x_i is the mole fraction of component i, ΔG_mag is the magnetic contribution to the free energy and ^EG_m is the excess free energy. Also, the difference in the Gibbs free energy between homogeneous liquid and the completely separated liquid can be considered as a driving force for LLPS. The magnitude of driving force determines the probability of phase transformation. Fig. 6 reveals the Gibbs free energy of the undercooled Cu-Fe melt at 1500 K. Although the Gibbs free energy of γ-Fe is lower than that of the melt when Fe content is higher than 41% (Fig. 6(a)), the Gibbs free energy change on primary solidification and LLPS (Fig. 6(b)) explicitly shows that LLPS occurs preferentially in the melt before its solidification when Fe content is lower than 54%. Otherwise, the γ-Fe phase resulting from a liquid-solid transition may primarily form.

View original graphic| Download| PPT slide

Fig. 6.   Gibbs free energies of the undercooled Cu-Fe melt at T = 1500 K: (a) Gibbs free energies G_L and G_γ-Fe of the melt and γ-Fe, (b) Gibbs free energy difference for phase separation in liquid and solidification of γ-Fe.

The miscibility gap of the (Cu_1-xFe_x)₉₀Zr₁₀ system was calculated using thermodynamic data [13,[45], [46], [47]]. The liquid miscibility gap exhibits a pronounced asymmetry (Fig. 7(a)), whose critical point temperature (T_c) and composition are approximately 1379 K and Cu₅₄Fe₃₆Zr₁₀, respectively. The liquidus temperatures for the (Cu_1-xFe_x)₉₀Zr₁₀ (x = 0, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, and 1) alloys were determined by DSC, and the results are shown in Fig. 7. Similar to the binary Cu-Fe system, there exists the competitive behavior of phase formation in the as-quenched (Cu_1-xFe_x)₉₀Zr₁₀ alloys. The compositions (Cu_0.55Fe_0.45)₉₀Zr₁₀ and (Cu_0.5Fe_0.5)₉₀Zr₁₀ are close to the critical point composition Cu₅₄Fe₃₆Zr₁₀. Combining DSC analyses with thermodynamic calculations shows the undercooling degree required for LLPS is about 110 K for the Cu₅₄Fe₃₆Zr₁₀ alloy. During rapid quenching, the spinodal decomposition is more possible for these alloys, leading to two-phase interconnected structure (Fig. 1(a) and (b)). The composition (Cu_0.4Fe_0.6)₉₀Zr₁₀ is far away from the critical point composition, exhibiting a larger undercooling degree required for the spinodal decomposition. The LLPS may occur by the liquid-state nucleation and growth, and thus result in the particle-isolated microstructure (Fig. 1(c)). If the melt was not undercooled into the liquid miscibility gap of the (Cu_1-xFe_x)₉₀Zr₁₀ alloys, the LSPT may primarily occur during cooling. Fig. 7(b) and (c) shows the microstructure of the (Cu_0.4Fe_0.6)₉₀Zr₁₀ and (Cu_0.8Fe_0.2)₉₀Zr₁₀ alloy samples after the DSC circle at a cooling rate of 20 K s^-1. The Fe dendrites primarily precipitate at about 1525 K and then Fe₂Zr phase forms at about 1485 K during the (Cu_0.4Fe_0.6)₉₀Zr₁₀ alloy cooling, as shown in Fig. 7(a). The residual liquid was enriched by Cu solidifies at about 1328 K. For the (Cu_0.7Fe_0.3)₉₀Zr₁₀ and (Cu_0.8Fe_0.2)₉₀Zr₁₀ alloys, besides Fe₂Zr and Cu phases, the formation of Cu₅Zr phase corresponding to the exothermic reaction at 1250 K was detected, as shown Fig. 7(a) and (c).

View original graphic| Download| PPT slide

Fig. 7.   (a) Calculated metastable miscibility gap and DSC curves of the (Cu_1-xFe_x)₉₀Zr₁₀ system, (b) and (c) SEM images of the (Cu_0.4Fe_0.6)₉₀Zr₁₀ and (Cu_0.8Fe_0.2)₉₀Zr₁₀ alloy samples after a cooling DSC circle.

As mentioned above, the microstructure of the as-quenched (Cu_1-xFe_x)₄₀Zr₆₀ alloys may be composed of glassy matrix, glassy particles, and crystalline precipitates. It means that the competition between LLPS, LSPT and liquid-glass transition (LGT) plays a decisive role in the final phase formation. Here, the competitive mechanism of phase formation in the as-quenched (Cu_1-xFe_x)₄₀Zr₆₀ alloys is analyzed. Fig. 8(a) shows the miscibility gap and the liquidus temperature of the (Cu_1-xFe_x)₄₀Zr₆₀ alloys. For as-quenched (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloys, LLPS takes place when alloy melt is undercooled into the miscibility gap during rapid quenching and then LGT occurs at T_g = 615 K, leading to the formation of a microstructure with glassy matrix embedded by glassy particles (Fig. 3(a)).

View original graphic| Download| PPT slide

Fig. 8.   (a) Calculated metastable miscibility gap of the (Cu_1-xFe_x)₄₀Zr₆₀ system, (b) schematic diagram of viscosity changing with temperature. T_L and T_s present the liquidus temperature and onset temperature of LLPS, respectively.

For the (Cu_0.7Fe_0.3)₄₀Zr₆₀ alloy, the LLPS into Cu-rich and Fe-rich liquids takes place during the rapid quenching. The driving force for LSPT is higher than that for LGT because the Cu content of the Fe-rich droplets formed by LLPS was obviously lower than that of Cu-rich matrix in the periphery of Fe-rich spheres. As shown in the rectangular area in Fig. 5, the black Cu-rich NCs generated from LSPT prefer to be conjoined with of the white Fe-rich glassy sphere, which indicates that the interface between the Fe-rich droplets and Cu-rich liquid matrix may act as nucleation sites for the Cu-rich NCs. This phenomenon also were detected in the rapidly solidified binary Cu-Co and Cu-Fe alloys [48,49]. Consequently, at this moment, the as-quenched (Cu_0.7Fe_0.3)₄₀Zr₆₀ system consists of the Cu-rich liquid matrix, Fe-rich droplets, and Cu-rich NCs. The Cu-rich and Fe-rich liquids enriched by Zr have relatively good glass-forming ability. When the temperature approaches T_g, the two liquids subsequently undergo LGT. Finally, the microstructure containing two amorphous phases and NCs is obtained, as shown in Fig. 5. The alloy system with the composition at the CuZr-rich side in Fig. 8(a) has a good glass-forming ability [50]. The glass-transition-temperature (T_g) line has an intersection with the binodal line. As a result, there exists a competition between the LLPS and LGT. A monolithic glassy structure may be obtained if the LGT prior to the LLPS.

Concerning the formation of nanoscale structure in (Cu_1-xFe_x)₄₀Zr₆₀ system, it can be attributed to the decrease of the metastable miscibility gap of the (Cu_1-xFe_x)₄₀Zr₆₀ alloy in contrast to that of the (Cu_1-xFe_x)₉₀Zr₁₀ system. As the schematic diagram (Fig. 8(b)), the nucleated droplets, owe to the narrow temperature region H_s-g (=T_s-T_g), have no enough time to grow by diffusion before LGT. On the other hand, the high viscosity (10⁶‒10⁷ poise [51]) and small diffusion coefficient of solute lead to a very small growth rate of the nucleated droplets when the melt temperature is deeply undercooled to approach the glass transition temperature.

3.3. Resistivity behaviour

Due to electical resitivity is sensitive to structural change, measuring resistance change is an effective method to study crystallization. Bulk metallic glasses (BMGs) generally exhibit negative temperature coefficient of resistance (TCR) because the appearance of grain changes the disorder arrangement of atoms as well as the band structure [[52], [53], [54]]. Variation of normalized resistivity (ρ_T/ρ_373K) with temperature of the as-quenched (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloy at a heating rate of 10 K min^-1 are shown in Fig. 9(a). Unlike other Zr-based metallic glasses [52,53], which show continuous decrease of resistivity during crystallization, the resistivity of the (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloys shows an abnormal increase at the initial stage before the normal declining stage. In addition, the value of resistance relative change is pretty big and this phenomenon is detected for the first time.

View original graphic| Download| PPT slide

Fig. 9.   (a) Variation of nomalized resistivity with temperature and DSC heating curve of (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloys, (b) Variation of nomalized resistivity with temperature of Zr₆₀Cu₄₀ and Zr₇₅Fe₂₅ alloys.

The DSC curve of as-quenched (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloys exhibits two exothemic peaks at a heating rate of 20 K min^-1 (Fig. 9(a)), which implies a two-step crystallization process. Combine the resistivity and DSC curves, the onset temperature of abnormal increase (693 K) and normal declining (759 K) in resistivity curve are almost consistent with the temparature of crystalliztion (T_x1 = 695 K and T_x2 = 770 K) in DSC curve, which indicates that the change of resistivity is related with first crystallization.

For its structural particularity of the (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloys, the variation of normalized resistivity (ρ_T/ρ_373K) with temperature of as-quenched Zr₆₀Cu₄₀ and Zr₇₅Fe₂₅ alloys were investigated (Fig. 9(b)). Two distinct resistivity drops occur on the resistance curve of the Zr₆₀Cu₄₀ alloys, which shows a two-step crystallization process. However, the resistance curve of Zr₇₅Fe₂₅ alloys also exihibits an abnormal increase during crystallization. It can be deduced that the abnormal increase in as-quenched (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloys is related with the crystallization of FeZr-rich matrix.

The crystallization behavior can be evaluated by comparing the activation energies. The most frequently used method for the kinetics of the glass transition and crystallization is based on the Kissinger equation [55,56]:

$ln\frac{T^{2}}{\beta}=\frac{E}{RT}+C$ (4)

where β represents the heating rate (K min^-1) and C refers to the constant. R is the gas constant and T stands for T_x or T_p (K). Plots of ln(T²/β) versus 1000/T yield an approximate straight line, whose slope provides the activation energy E. Onset crystallization activation energy E_x1 and crystallization peak activation energy E_p represent the energy of nucleation and grain growth [57,58], respectively. Fig. 10 shows that plots of as-quenched Zr₆₀Cu₄₀, Zr₇₅Fe₂₅ and (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloys. From the calculated results, it shows that the E_x value of Zr₇₅Fe₂₅ alloys is much smaller, meaning that the crystallization takes place in the FeZr-rich matrix prior to the CuZr-rich particles duing heating the as-quenched (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloys. What’s more, for the as-quenched Zr₇₅Fe₂₅ alloys, the activation energy E_x is smaller than crystallization peak activation energy E_p, showing that the crystallization of as-quenched Zr₇₅Fe₂₅ alloys is characterized by high population of nucleation and low growth rate. In contrast, for the as-quenched Zr₆₀Cu₄₀ alloys, the E_x value is larger than crystallization peak activation energy E_p.

View original graphic| Download| PPT slide

Fig. 10.   Relationship between ln(T²/β) and 1/T. The subscripts 1, 2 and 3 present Zr₇₅Fe₂₅, Zr₆₀Cu₄₀, and (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloys, respectively.

To prove the above analysis, the phase evolution during the crystallization was obtained from the XRD patterns after the (Cu_0.5Fe_0.5)₄₀Zr₆₀ samples were isothermally annealed at 673 K, 713 K and 813 K for 0.5 h as shown in Fig. 11. The diffraction peaks broaden when annealing at 673 K (before abnormal increase of resistivity) for 0.5 h indicates the formation of NCs. There are obvious lattice fringes in the HRTEM image (Fig. 12(a)), which shows that NCs precipitate from FeZr-rich matrix. The formation of NCs result in the formation of nanometer grain boundaries in amorphous matrix and complete long-range ordered structure has been not formed. Finally, the strong electron scattering at the amorphous/ NCs interfaces will result in the increase of resistivity [55].

View original graphic| Download| PPT slide

Fig. 11.   XRD patterns of as-quenched (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloys annealed at 673 K, 713 K, and 813 K for 0.5 h, respectively.

View original graphic| Download| PPT slide

Fig. 12.   (a) and (b) HRTEM images of (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloys annealed at 673 K for 0.5 h.

At 713 K (resistivity rising stage) annealed for 0.5 h, the diffraction peaks in XRD patterns obviously broaden and no new phases form, which shows a large of NCs start to nucleate and grow slowly compared with XRD patterns of annealing at 673 K for 0.5 h. This is consistent with calculated activation energy. Nucleation of a large of NCs with slow growth will bring lots of nanometer grain boundaries. Therefore, the resistivity continues to increase and the maximum of the resistance relative change was about 12.5%, which is bigger than that in literature [59] and Zr₇₅Fe₂₅ alloys. It could be that there is an interface between the amorphous nanoparticles and amorphous matrix (as shown in Fig. 12(b)) in addition to amorphous/NCs interfaces. It may also enhance the electron scattering, which results in a large increase of resistivity. At 813 K (after normal decline), peaks belonging to NCs vanish and new peaks appear (ie., Zr₂(Cu,Fe) phase). It indicates that NCs finally grow up and residual amorphous phase transforms to a new stable phase completely. Because NCs grew up and the stable crystal structure formed, nanometer grain boundaries vanished and resistivity decreased finally.

4. Conclusion

The microstructural evolution and competitive mechanism of phase formation in the rapidly solidified Cu-Fe-Zr system were studied experimentally and by thermodynamic calculations. The liquid-liquid phase separation into the Cu-rich and Fe-rich liquids may occur during the Cu-Fe-Zr alloy melt cooling in the metastable miscibility gap. The microstructure from interconnected type to particle-isolated type is tunable by modifying the atomic ratio of Cu to Fe in the (Cu_1-xFe_x)₉₀Zr₁₀ alloy. Combining the thermodynamic calculations with DTA analyses indicates that the dome height of the miscibility gap decreases with the increase of Zr content. The structure of the rapidly solidified (Cu_1-xFe_x)₄₀Zr₆₀ alloy is determined by the competition among the liquid-liquid, liquid-solid, and liquid-glass transitions. For the (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloy, the nanoscale liquid-liquid phase separation and then liquid-glass transition can take place during the rapid solidification and result in a heterogeneous structure with glassy Fe-rich matrix embedded by glassy Cu-rich nanoparticles. The characteristic size of the Cu-rich nanoparticle is 2-8 nm, and the population density and volume fraction of the particles are estimated to be 5.37 × 10²⁴ m^-3 and 10.4% ± 0.2%. Moreover, the electrical property of the as-quenched (Cu_0.5Fe_0.5)₄₀Zr₆₀ alloy samples was investigated. It is of interest in an abnormal increase of electrical resistivity with the increase of temperature at the stage of crystallization. This may be attributed that the formation of high population density nanocrystals with slow growth in the glassy Fe-rich matrix results in the enhancement of electron scattering.

Acknowledgment

This work was supported by the National Natural Science Foundation of China (Nos. 51774264, 51574216, 51974288 and 51374194) and the Natural Science Foundation of Liaoning Province of China (No. 2019-MS-332).

Reference

View Option

• By original order
• By published year
• By cited within times
• By Impact factor