Preparation of an Fe₈₀P₁₄B₆ Bulk Nanocrystalline Alloy via Solidification from a Molten Alloy at Deep Undercooling

Abstract

Using fluxing technology, molten Fe₈₀P₁₄B₆ alloy achieved significant undercooling (ΔT). Experimental results demonstrate that the solidified morphologies of the Fe₈₀P₁₄B₆ alloy vary considerably with ΔT. At ΔT = 100 K, the microstructure is dendritic. At ΔT = 250 K, a variety of eutectic morphologies are observed, including a network-like structure near the solidification center, attributed to liquid spinodal decomposition. At ΔT = 350 K, the microstructure exhibits a uniform, random network-like morphology with approximately 50 nm. The mechanical property of the specimens solidified at different ΔT was checked by microhardness test, indicating that the hardness of the specimens increases with the increase in ΔT, reaching a maximum value of 1151 HV_0.2.

1. Introduction

Nanostructured alloys, first proposed by Gleiter [1] and Turnbull [2], have attracted significant attention due to their promising properties [3,4]. To date, bulk nanostructured alloys can be produced using various techniques, which can be broadly categorized into two main approaches. The first approach, introduced by Gleiter and colleagues [5] involves the preparation of nanometer-sized powders via evaporation techniques, followed by high-pressure sintering to form bulk specimens. The second approach involves the preparation of bulk nanostructured alloys using annealing bulk metallic glasses at an elevated temperature [6]. In principle, bulk nanostructured alloy can also be prepared by the direct solidification of a molten alloy subjected to deep undercooling [4,7,8], and via such a route, the liquid phase spinodal decomposition (LSD) may occur in some appropriate systems [9,10]. Another way of preparing compact nanophase alloys by melt-quenching in a single step was described by Bakonyi and Cziraki [11].

Kui and Guo [7] observed the LSD phenomenon during the solidification process of an undercooled eutectic Pd₈₂Si₁₈ molten alloy. A similar phenomenon was also found in the eutectic alloy Pd_40.5Ni_40.5P₁₉ [4,12,13]. However, the heats of mixing between the main elements such as Pd-Si, Pd-P and Ni-P are all negative in these two eutectic alloys [14]. Typically, the LSD phenomena occur only in systems with a positive heat of mixing among the main elements. Due to a lack of direct evidence and supporting theory, liquid phase separation in alloys with a negative heat of mixing has become a controversial issue. Kui [15,16,17,18] proposed that there is a miscibility gap in undercooled molten eutectic alloys. When a homogeneous eutectic alloy melt is undercooled into its miscibility gap, it decomposes into multiple liquid sub-networks of a characteristic wavelength λ that depends critically on the level of undercooling at which the decomposition takes place. When the undercooling is sufficiently large, λ can be smaller than 100 nm. Surface tension then sets in, breaking up the networks into droplet shapes. If crystallization is bypassed, it becomes a bulk amorphous nanostructure. If crystallization cannot be avoided, it will take place inside the miscibility gap, to transform the system into a nanostructure with a crystalline solid of network-like morphology. Guo et al. [18] have demonstrated fully such phase evolution in the undercooled Pd₈₂Si₁₈ eutectic alloy. As a result, a nanostructured Pd₈₂Si₁₈ alloy with grain size 3~6 nm was prepared. Therefore, the LSD can afford an effective method to produce nanostructured materials with tailorable grain size.

The LSD phenomena have been reported in some iron-based systems. Ho et al. [15,16] were able to cast molten Fe-C alloys into ingots of network-like morphology, which have attractive mechanical properties. For example, Fe₈₃C₁₇ network alloys have a compressive yield strength of ~2000 MPa, a compressive strength of ~2500 MPa, a plastic strain to failure of ~18 pct, and a hardness value of ~536 HV. Chow et al. [17,19] found that an undercooled Fe_79.5B_6.5C₁₄ molten ingot can also be cast into a network ingot and its compressive strength can be as large as ~3700 MPa, a hardness value of ~800 HV.

Both the Pd₈₂Si₁₈ and Fe₈₀P₁₄B₆ alloy systems are all the “metal-metalloid” type. In addition, Fe₈₀P₁₄B₆ alloy can be regarded as a pseudo-binary eutectic alloy with the composition of Fe₈₀(P,B)₂₀, of which the equilibrium phase diagram is similar to that of Fe-P binary eutectic alloy. Thus, it can be expected that Fe₈₀P₁₄B₆ alloy may also undergo the LSD when the undercooling is deep enough. In this work, in order to obtain bulk nanostructured metals with better performance, based on Fe₈₀P₁₄B₆ alloy, we attempt to prepare bulk nanocrystalline alloys with controllable grain size and uniform distribution of grain size by metastable LSD mechanism.

2. Experiments

Fe₈₀P₁₄B₆ alloy ingots were prepared from iron powder (99.98% pure), Fe₃P powder (98% pure), and boron powder (99.95% pure) (Alfa Aesar, Shanghai, China). After the right proportion was weighed, they were put into a clean fused silica tube and alloyed by torch under an argon atmosphere. Then, the as-prepared alloy ingots were fluxed in molten B₂O₃ about 4 h under a vacuum of ~10 Pa. After the fluxing treatment, the whole system was transferred to the furnace and placed on a thermocouple. The thermocouple and computer were used to record the temperature of the molten specimen. As soon as a crystallization event occurred, the whole system was removed from the furnace and quenched into the iced water. The diameter of each sample is about 2.5~3.0 mm. There was a shrinking point that could be regarded as the solidification center when the sample had solidified. The solidification center was marked by a color pen, and then the sample was cut from the center along the diameter by a wire cutting machine. The phase structure of the samples was characterized by X-ray diffractometer (XRD) using a Bruker D8 ADVANCE (Cu Kα radiation) (XRD, Bruker, Berlin, Germany). The microstructures and the phase composition of the specimens were investigated using a SUPRA55 FE-SEM (ZEISS, Oberkochen, Germany) with energy dispersive X-ray spectrometers (EDS, X-MaxN20, Oxford, UK). For SEM analysis, the samples were etched with a solution of CH₃CH₂OH and HNO₃ in a volume ratio of 9:1. Vickers microhardness was measured under 200 g load with 10 s dwell time (HXD-1000TMC, Taiming, China, 3 samples for each case, five indents per sample).

3. Results and Discussion

3.1. The Evolution of the Solidified Morphology Under Different Levels of Undercooling

The fluxing technique was employed to undercool the molten alloy Fe₈₀P₁₄B₆ below its melting temperature in the present work. The melting temperature (T_m) of Fe₈₀P₁₄B₆ alloy was determined to be 1310 K by Differential Thermal Analysis (DTA). If T is referred to as the crystallization temperature of the specimens, then the crystallization undercooling is defined as ΔT = T_m–T.

When ΔT = 100 K, Figure 1a shows the general morphology, and a primary dendrite morphology is observed under low magnification. Figure 1b shows the primary dendrites and the matrix under high magnification. The dendrites are composed of α-Fe, and the matrix is composed of α-Fe and Fe₃(P,B), as determined by EDX.

When ΔT = 250 K, a low magnification SEM image is shown in Figure 2a. There are two zones (dash line) in the image. While Zone A is very smooth, there are some cell-like features in Zone B. In fact, there are quite different morphologies in these zones. Figure 2b–e is an FE-SEM micrograph under a high magnification, showing the patch boundaries and morphologies of each side. There are two types for the patch boundaries in Zone A. One is Figure 2b. It shows a typical morphology about lamellar eutectic, and each side of the boundary shows a strong and different direction for the lamellar. The dark lamella is composed of α-Fe and that of the bright area is composed of Fe₃(P,B) which can be checked by EDX. The composition of the precipitated phase is Fe92(PB)8 and the matrix is Fe72(PB)28. Figure 2c shows another patch boundary and morphologies for both sides. The morphology shows a boundary between the lamellar eutectic and rod eutectic. The bar is composed of α-Fe, and the bright area is composed of Fe₃(P,B). Zone B (Figure 2d) is another boundary. We can see that the morphology on the left is the lamellar with a strong direction, but on the right, it becomes a random network-like microstructure whose morphology is like liquid spinodal decomposition, as suggested by Cahn [20,21,22,23]. The network (Figure 2e) is composed of two sub-networks, and the wavelength of one network is about 200 nm and that of another one is about 250 nm. The smooth area is composed of Fe₃(P,B), and the dark area is composed of α-Fe.

When ΔT = 350 K, from the low-magnification SEM image (Figure 3a), one can see that the microstructure is very uniform. The micrograph under a high magnification FE-SEM (Figure 3b) shows that there are no eutectic morphologies, but rather a structure of interconnected networks on the whole surface of the sample. The morphology of the networks is very uniform and random. The wavelength of the network is about 50 nm. The result of EDX shows that the dark network and bright network are composed of α-Fe and Fe₃(P,B) phases, respectively.

At the same time, the XRD study (Figure 4) also suggests that the phases of the samples at ΔT = 100 K, 250 K, 350 K all consist of α-Fe and Fe₃(P,B) phases, which is consistent with the above results of EDX analysis.

The ternary Fe₈₀P₁₄B₆ alloy can be considered as a pseudo-binary eutectic alloy with the composition of Fe₈₀(P,B)₂₀. For a eutectic system, if the temperature is high, the entropy is the dominant for the Gibbs free energy, resulting in a homogeneous liquid phase. However, when the temperature is low enough, the enthalpy will become the dominant, so there will be a miscibility gap for the liquid phase at the larger enough undercooling.

From the XRD (Figure 4), we can see that the Fe₈₀P₁₄B₆ alloy specimens solidified at the different undercoolings all consisted of α-Fe and Fe₃(P,B) phases. So, a schematic phase diagram is shown in Figure 5 for the present Fe₈₀P₁₄B₆ alloy based on the above discussion. With the condition of non-equilibrium solidification, the eutectic area of this metal-metalloid alloy under the undercooling presents an asymmetric coupled zone, as shown in the dash zone in Figure 5.

When ΔT₁ = 100 K (sample A), the undercooling is far away from the miscibility gap. The skewed couple zone dictates that only the α-Fe phase can grow as a primary phase in this region for this composition, so α-Fe becomes the primary dendrite. When ΔT₂ = 250 K (sample B), the molten sample just reach the miscibility gap so we can see the random network-like in Zone A. However, the released latent heat of solidification in Zone A will transfer to Zone B, resulting in the increase in temperature in Zone B, in which the temperature may move out of the miscibility gap to the pseudo-eutectic area. As a result, the undercooling in Zone B will decrease. In addition, the heat conduction can make a temperature gradient in Zone B, which will lead to the nuclear driving force decreasing in different degrees, and the proportion of α-Fe and Fe₃(P,B) phases may also change. So, a variety of eutectic morphology can be found in Zone B. When ΔT₃ = 350 K (sample C), the undercooling is large enough. Therefore, the released latent heat at the solidification center is not enough to raise the temperature outside the spinodal line. Thus, Sample C presents a uniform random network-like microstructure for all regions as shown in Figure 3b.

3.2. The Evolution of Microhardness Under Different Levels of Undercooling

The microhardness of each sample solidified at the different undercooling was determined by a micro hardness tester. The results (Figure 6) show that the microhardness of the samples increases with the undercooling. The microhardness rises to as high as 1151 HV_0.2.

It has been considered that the grain size of the samples decreases with the undercooling. The smaller the grain size, the more the grain boundaries, which leads to greater resistance for the deformation of the material. In addition, it is more important that the samples solidified under deep enough undercooling will present a network-like microstructure, which may result in a high hardness. Based on the above reasons, it can explain why the hardness increases with the undercooling increases.

4. Conclusions

With the help of fluxing technique, the molten Fe₈₀P₁₄B₆ alloys can be solidified at different undercooling (ΔT) through isothermal undercooling experiments. The crystalline morphologies of the Fe₈₀P₁₄B₆ alloys solidified at different undercooling are quite different. When ΔT = 100 K, the traditional crystalline morphology i.e., α-Fe primary dendrite can be found. When ΔT = 250 K, a random network-like region and patch boundaries with different eutectic microstructures can be found. When ΔT = 350 K, the solidification morphology is uniform random network for all regions, which is controlled by LSD mechanism, and the average grain size is about 50 nm. The mechanical property indicated that the microhardness of the Fe₈₀P₁₄B₆ alloy increases with the increase in the undercooling.