The Influence of Preparation Parameters on the Morphology and Magnetic Properties of Fe-N Powders Obtained by the Gas Atomization Method

Abstract

α″-Fe₁₆N₂ materials are of increasing interest for their applications in products such as rare earth-free permanent magnets. The lack of a method of mass production for powders as raw materials delays the preparation of such magnets. Through employing the gas atomization method, we managed to prepare α″-Fe₁₆N₂ powders whose morphology and magnetic properties were tailored by the preparation parameters. As a result of optimizing the preparation parameters (ejection temperature and pressure, ejection nozzle diameter, and atomization pressure), we managed to prepare powders with a size of about 30 μm and a content of 31% α″-Fe₁₆N₂ phase. The value of the saturation magnetization (234.8 emu/g), the reasonable coercivity value (970 Oe) presented by the prepared powders, and the opportunity of scaling up approaches based on the preparation of powders via gas atomization support the feasibility of preparing α″-Fe₁₆N₂ powders at an industrial level.

1. Introduction

Permanent magnets play a crucial role in various aspects of modern life for applications in “green energy” (wind turbines, electric–hybrid cars), automation (and robotics), information technology, magnetic sensors, and magnetic microdevices, and their influence is constantly growing [1,2,3,4,5]. At the moment, rare earth-based magnets dominate the global magnet market due to their outstanding magnetic performance. However, some of these rare earths, such as neodymium (Nd), praseodymium (Pr), dysprosium (Dy), and terbium (Tb), are listed as critical materials by the US Department of Energy, the International Renewable Energy Agency, and other international institutions [6,7,8,9], and their future use for permanent magnets production is uncertain. Also, the rare earth extraction and separation process is very pollutive, although new separation and extraction methods are constantly being proposed [10,11,12,13,14]. According to the Motion Control & Motor Association (MCMA), the world’s population is expected to reach 9 billion by 2050, and by this time, every human on Earth will be served by three service robots. Furthermore, each robot is expected to contain more than 100 motors. As a result, the demand for permanent magnets is expected to increase significantly in the coming years. This, coupled with increasing concerns about environmental degradation due to the exploitation of rare earths, increasing costs, and availability problems of rare earths, has led to worldwide efforts to search for rare earth free magnetic materials that can be used for the preparation of a new generation of permanent magnets. The ideal permanent magnet should have the following characteristics: an abundant and environmentally friendly element composition, a high saturation magnetization, a high energy product, fairly high coercivity, and a high Curie temperature. The compound α″-Fe₁₆N₂ fulfills almost all of these requirements, as its constituent materials Fe and N are abundant and environmentally friendly, it possesses a giant saturation magnetization of ~2.9 T (~305 emu/g) [15,16], its theoretical energy product is estimated to be two times higher than that of the strongest rare earth magnet [17], it has a Curie temperature of about 540 °C [18], and the reported coercivity of the α″-Fe₁₆N₂ compound is about 2 kOe [19]. After the first report by Kim and Takahashi on the giant saturation magnetization of about 290 emu/g of the α″-Fe₁₆N₂ phase in the evaporated Fe-N thin film, many researchers have made efforts to reproduce these results. However, it was not until nearly 20 years later that researchers from Hitachi in Japan reported a close value for the saturation magnetization obtained using single-crystal α″-Fe₁₆N₂ films grown via molecular beam epitaxy [18,20]. Encouraged by the promising magnetic properties demonstrated by thin films, researchers are interested in preparing bulk samples with the same promising magnetic properties as potential precursors for permanent magnets. Thus, magnetic materials containing the α″-Fe₁₆N₂ phase have been prepared in the form of thin films, foils, rods, ribbons, nanocones, nanopowders, and powders [21,22,23,24,25,26,27,28,29,30]. However, the most suitable precursors for the preparation of permanent magnets are materials in the form of powders or even nanopowders. In the case of the α″-Fe₁₆N₂ compound, compaction is also a challenge due to the fact that the α″-Fe₁₆N₂ compound decomposes at temperatures over 200 °C [31]. Therefore, high-temperature sintering or spark plasma sintering, the usual methods for compacting Nd-Fe-B permanent magnets, are not feasible methods for preparing these types of permanent magnets. To date, the only feasible method is the compaction of powders or nanopowders at lower temperatures in the presence of a binder. However, in the case of nanoparticles, the presence of the binder can reduce quite a lot of the saturation flux density because of the decreased phase purity, consequently reducing the energy product, the (BH)_max of the α″-Fe₁₆N₂ magnets. In addition, due to their small size, α″-Fe₁₆N₂ nanoparticles are highly susceptible to oxidation. These problems can be mitigated by the use of powders, for which the binder–magnetic material ratio would be lower than in the approach based on nanoparticles. For this study, α″-Fe₁₆N₂ powders were prepared via two methods: (i) the mechanical ball milling of iron-based materials with a source of solid nitrogen (ammonium nitrate [30,32,33]) and (ii) by keeping the iron powders up to 660–670 °C in ammonia–hydrogen mixtures and rapidly quenching them in water with a long consecutive tempering at 120–150 °C [34] (a method inspired by the pioneering work of Jack [35]). Although preparation via ball milling is a mass preparation method used to prepare powders, the powders obtained via this method showed a high degree of oxidation that contributed to the degradation of the magnetic properties of the resulting particles. In the case of the second approach, (ii), the sample magnetization was about 189 emu/g, lower than the saturation magnetization of iron. Therefore, a mass production method for α″-Fe₁₆N₂ powders with good magnetic properties is still lacking, which is why the preparation of permanent magnets based on α″-Fe₁₆N₂ is delayed. On the other hand, it is well known one of the most used methods for the mass preparation of powders is the gas atomization method [36,37]. Therefore, in this work, we propose the preparation of Fe-N powders via the gas atomization method and study the influence of some preparation parameters on the morphology and magnetic properties of the obtained Fe-N powders.

2. Materials and Methods

Fe ingots of about 20–25 g were prepared via the arc melting technique from commercial iron lumps (Alfa Aesar) with a purity higher than 99.99%. The resulting Fe ingots were transformed into powders via the gas atomization method using a homemade atomization plant. The schematic structure of the gas atomization plant is shown in Figure 1.

The Fe ingot to be melted was placed in a quartz crucible provided in the lower part with a circular ejection nozzle and heated via induction to temperatures up to 50 °C above the liquidus temperature of Fe (1538 °C) in order to obtain appropriate viscosity and surface tension values that allowed for the steady flow of the liquid jet. A photocell located at the top of the quartz crucible was used to measure the surface temperature of the melted iron after it had been previously calibrated with a high-precision pyrometer. Both the atomizing chamber and the melting chamber were previously vacuumed to a pressure of 10⁻³ mbar and flushed 4 times with nitrogen gas and refilled with nitrogen to reduce the oxygen content. Before being ejected, the melt was maintained for five minutes under a nitrogen pressure between 1 and 3 bar. Induction melting creates currents of liquid that facilitate the continuous movement of the material, resulting in the exposure of the liquid iron to the nitrogen gas that fills the melting chamber. The flow of the liquid metal is initiated by the sudden decrease in pressure in the atomizing chamber to 0.2 bar. At the same time, the atomization gas, the pressure of which varied between 10 and 20 bars, hit the molten iron jet, resulting in the disintegration of the melt into droplets. The atomizing gas exit nozzle was in the form of a continuous annular slot with a width of 300 µm and concentric with the melt ejection nozzle (right Figure 1). The diameter of the ejection nozzle varied between 100 and 300 µm. The angle between the direction of the liquid jet and that of the gas jet was kept constant at 45° in all experiments. The distance between the ejection nozzle and the particle collector was tested and established as the optimal value of 120 cm. After atomization, the collected powders were immediately transferred to a nitrogen-filled glovebox without exposure to air to prepare the samples for structural, morphological, and magnetic measurements. The structural analysis was carried out via X-ray diffraction (XRD) in a Bruker AXS D8—Advance diffractometer over an angular range 20–90° with a step of 0.02° and a counting time of 2 s per step using conventional Cu–Kα incident radiation. The Rietveld refinements of the XRD patterns were performed using MAUD software [38]. The microstructure and morphology of the sample was investigated using a scanning electron microscope, FIB/FE—SEM Cross—Beam Carl Zeiss NEON 40 EsB (Oberkochen, Germany), equipped with an energy dispersive X-ray spectroscopy (EDS) module. The free software Image J (https://imagej.net/ij/, accessed on 24 September 2023) was used to characterize the size distribution of the particles [39]. Magnetic measurements were performed using a Vibrating Sample Magnetometer (VSM) (Lake Shore VSM 7410, Westerville, OH, USA) in a maximum applied field of 20 kOe at room temperature.

3. Results

In an attempt to investigate the influence of the parameters of the atomization process, such as the ejection temperature, the ejection pressure, the diameter of the ejection nozzle, and the atomization pressure, on the morphology and magnetic properties of the atomized powders, they were modified one by one.

3.1. Ejection Temperature

Through maintaining the ejection pressure at 2 bar, the atomization pressure at 15 bar, and the diameter of the ejection nozzle at 200 μm, atomized powders were prepared, for which the melt was superheated by 10, 20, and 50 °C, respectively, above the melting temperature of Fe (1538 °C). Figure 2 shows SEM images of the atomization results of the powders for the constant atomization parameters specified above and different ejection temperatures.

The melt ejection temperature is primarily reflected in the melt viscosity. A high melt temperature leads to a low viscosity, while a low temperature leads to a high melt viscosity. Atomization of the high viscosity melt (low temperature, 1548 °C) generally results in a large amount of fibers/ligaments and less spherical particles, as can be seen in Figure 2a. With the increase in temperature to 1568 °C, the viscosity drops; thus, it is possible to disintegrate the melt flow into particles, mostly spherical ones (Figure 2b). The further increase in the ejection temperature to 1588 °C leads to uniformity in the morphology of the formed particles; thus, for the parameters specified above, we obtained mostly spherical powders with a diameter between 20 and 120 μm (Figure 2c). Therefore, we can state that a further increase in melting/ejection temperatures would lead to finer and more uniform particles. However, such an increase in temperature is not possible due to technological limitations related to the fact that the melting temperature of iron is close to the softening temperature of the quartz tube in which it is heated, meaning that further melt heating would lead to the flow of the quartz, prohibiting the atomization process.

3.2. Ejection Pressure (P_ej)

Through maintaining the ejection temperature at 1588 °C (the optimal temperature at which the most spherical particles were obtained), the atomization pressure at 15 bar, and the diameter of the ejection nozzle at 200 μm, atomized powders were prepared for different ejection pressures between 1 and 3 bar. Figure 3 shows SEM images of the powders obtained via gas atomization at different ejection pressures whilst the rest of the atomization parameters were constant.

An analysis of the SEM images revealed that the size of the powders increases with the ejection pressure. Thus, for p_ej = 1 bar, the average particle diameter is 117 μm, and for p_ej = 2 bar, the average particle diameter is 163 μm, while for p_ej = 3 bar, the average particle diameter increases to approximately 192 μm. These results can be explained by the link between the mass flow rate of the melt and the median diameter of the particles. Lubanska [40] established the correlation between the average diameter (d_m) of the gas-atomized powders and the atomization parameters using the following equation:

$$d_m=DK{[\frac{v_m}{v_g}\frac{1}{We}(1+\frac{m_m}{m_g})]}^{1/2}$$

(1)

where $We=\frac{{\left(\Delta U\right)}^2{\rho }_{m}D}{\sigma }$ is the Weber number and represents the ratio of the product of the square of the relative velocity between the melt and the gas ($\Delta U$), the density of the metal melt (${\rho }_m$), the diameter of the ejection nozzle (D), and the surface tension of the melt (σ). K is a constant specific to each atomization system, $v_m$ and $v_g$ are the kinematic viscosities of the melt and the gas, respectively, and $m_m$/$m_g$ is the ratio between the mass flow rate of the melt and the gas flow rate. The increase in ejection pressure is reflected in the increase in the mass flow rate of the melt. According to Equation (1), the increase in mass flow leads to an increase in the diameter of the particles. Therefore, to obtain small-sized particles, the mass flow rate should be kept low.

3.3. The Diameter of the Ejection Nozzle

Figure 4 shows the distributions of the diameters of the atomized particles as a function of the diameter of the ejection nozzle. The rest of the parameters, namely the ejection temperature (1588 °C), ejection pressure (2 bar), and atomization pressure (15 bar), were kept constant.

Generally, gas atomization techniques are characterized by a broad particle size distribution. It can be observed that as the diameter of the ejection nozzle decreases, the particle size distribution curve becomes narrower and moves to the small particle size region. Median particle diameter values were extracted from the particle size distributions and plotted against the ejection nozzle diameter (D), as shown in Figure 4d. A ~50% decrease in median particle diameter is observed when the diameter of the melt nozzle is reduced from D = 300 μm to D = 100 μm, indicating that the diameter values of the ejection nozzle greatly influence the particle size distribution. However, reducing the diameter of the melt nozzle leads to an increase in the resistance to the flow of the melt through the nozzle, and further reduction leads to the possibility of clogging the ejection nozzle.

3.4. Atomizing Pressure

In Figure 5, SEM images of the atomized powders prepared at different atomization pressures while the rest of the parameters, namely the ejection temperature (1588 °C), the ejection pressure (2 bar), and ejection nozzle diameter (200 μm), were kept constant are shown.

It can be observed that the diameter of the particles decreases as the atomization pressure increases. For a constant value of the atomization nozzle section, as it is in our case (300 µm), as the atomization pressure increases, the force exerted by the gas on the melt jet also increases, which leads to its easier disintegration. According to Equation (1), the dependence between particle diameter and gas flow rate is inversely proportional; on the other hand, there is a direct proportional relationship between gas flow rate and pressure. Therefore, increasing the gas flow and the atomization pressure, respectively, leads to a decrease in the diameter of the atomized powder. Thus, for an atomization pressure of 20 bar, the average diameter of the atomized powders is 32 µm. Thus, we can conclude that atomization pressure is the most influential parameter involved in particle size reduction. This observation agrees with the results reported by Silva et al. in [41].

3.5. Structural Characterization of the Atomized Powders

It is well known that nitrogen has limited solubility in iron-based alloys, and the saturation limit depends on the temperature of the melt, as well as the partial pressure of nitrogen above the melt. According to Sievert’s law, the amount of nitrogen dissolved in the melt is directly proportional to the square root of pressure [42]. It is important to remember that the ejection pressure is the pressure at which the melt is held for 5 min in the crucible before the effective atomization. Figure 6 shows the Rietveld refined XRD diffractograms of Fe-N powders of the atomized powders prepared at an ejection temperature of 1588 °C, an atomization pressure of 20 bar, an ejection nozzle diameter of 100 μm, and different ejection pressures (between 1 and 3 bar). The reliability factors of the Rietveld refinement and the percentage concentration of the phases resulting from the XRD analysis are listed in

Table 1. The reliability factors of the Rietveld refinement and the evolution of the phase composition vs. ejection pressure of the atomized Fe-N powders.

pej (Bar)	χ2	Rwp (%)	Phase Volume Fraction (%)
			α-Fe	γ-F4N	α′-Fe8N	α″-Fe16N2
1	1.45	12.53	74.5	-	11.2	14.3
2	1.67	13.24	55	4.3	18.6	22.1
3	1.32	11.81	40.5	12.4	20.3	26.7

.

It can be observed that the reliability factors of the Rietveld refinement, such as the goodness of fit (χ²) and the residual factor of the weighted profile (Rwp), have values <2 and <15%, respectively, indicating a good agreement of refinement. The structures of the powders consist of a mixture of α-Fe, γ-F₄N, α′-Fe₈N, and α″-Fe₁₆N₂ crystalline phases. In order to explain the formation of these phases, we will refer to the Fe-N phase diagram [43]. According to the phase diagram, the formation of the austenite γ-phase optimally occurs with the combination of nitrogen potential and temperatures above 592 °C. The austenite γ-phase is a face-centered cubic (fcc) arrangement of Fe atoms, with N randomly occupying up to one in ten of the octahedral interstices [44]. When the γ-phase is quenched (what happens to the melt droplets immediately after formation), a martensitic transformation takes place: the Fe atom arrangement changes from a fcc arrangement to a body-centered cubic (bcc) arrangement [44,45]. Due to the rapid cooling of the γ-phase, part of the nitrogen atoms that do not have enough energy cannot diffuse to the desired position, remaining stuck in the previous positions, so a transformation without diffusion takes place, and the α′-Fe₈N phase is formed. The α″-Fe₁₆N₂ has a structure that is very similar to that of α′-Fe₈N, except for the arrangement of nitrogen atoms: in the α′-Fe₈N phase, the N atoms are disordered, while in the α″-Fe₁₆N₂, they are ordered. This means that during the fcc–bcc transformation, some of the N atoms have enough time and energy to move from the interstices to the desired positions, managing to arrange themselves and form the α″-Fe₁₆N₂ phase. Therefore, both nitrogen atom positions and nitrogen concentration are key points for the phases generated after quenching. The results of our XRD pattern analysis indicate that increasing the ejection pressure leads to an increase in the percentage of Fe-N-based phases (to the detriment of the α-Fe phase). Thus, increasing the ejection pressure from 1 to 3 bar led to an increase in the percentage of the α″-Fe₁₆N₂ phase from 14.3 to 26.7%. This can be explained by the fact that a higher nitrogen pressure facilitates the diffusion of a larger amount of nitrogen atoms in the iron lattice. The increased content of the γ-F₄N phase for powders obtained at 3 bar is most likely related to the insufficiently fast quenching of these larger-sized powders. The cooling rate of the large-sized atomized droplets is much slower than that of the small-sized droplets. The results from our analysis of the X-ray diffraction patterns (not shown here) corresponding to the Fe-N powders atomized at different ejection temperatures did not indicate significant modifications to the structure.

3.6. Magnetic Properties of the Atomized Fe-N Powders

Figure 7b–d show the hysteresis loops of the atomized powders prepared at an ejection temperature of 1588 °C, an atomization pressure of 20 bar, an ejection nozzle diameter of 100 μm, and different ejection pressures (between 1 and 3 bars). In order to compare the magnetic properties of the Fe-N powders atomized in nitrogen with those of some Fe powders, a reference sample of the Fe powders was atomized in argon gas, both for ejection and for atomization. Figure 7a shows the hysteresis loops corresponding to the reference powders.

The Fe reference powders (atomized in argon) showed a saturation magnetization of Ms = 214.5 emu/g and a coercivity of Hc = 29 Oe. The saturation magnetization is lower than the theoretical value of 218 emu/g, most likely due to the superficial oxidation of the powders, although the XRD investigations did not reveal the existence of oxides. Both the coercivity and saturation magnetization of the Fe-N powders are superior to those of the Fe reference powders, and they increased with increasing ejection pressure, as well as with increasing Fe-N phase content. Since the γ-F₄N phase has a lower saturation magnetization than Fe (Ms = 188 emu g [46]) and the α′-Fe₈N phase has a saturation magnetization comparable to that of Fe alone (Ms = 220 emu/g [47]), the phase responsible for increasing the saturation magnetization of the powders is the α″-Fe₁₆N₂ phase. Thus, the powders obtained at an ejection pressure of 1 bar, with a α″-Fe₁₆N₂ phase content of 14.3%, show a 9.2 emu/g increase in saturation magnetization compared to the Fe powders, and the coercivity also significantly increases from 29 Oe to 568 Oe. For the powders prepared at 2 bar, for which the phase content was increased to 22.1% (according to Table 1), the saturation magnetization increased to 232.3 emu/g, and the coercivity increased to 675 Oe. However, for the powders obtained at 3 bar, the saturation magnetization decreased to 229.4 emu/g due to the significant increase in the γ-F₄N phase content. To overcome this impediment, after maintaining a pressure of 3 bar above the melt for 5 min before atomization, the pressure was reduced, and ejection was performed at 1 bar. This approach allowed for the preparation of Fe-N powders with a size of about 30 μm, an increased α″-Fe₁₆N₂ content of 31% (without the γ-F₄N phase), and a superior saturation magnetization (Ms) of about 234.8 emu/g and a coercivity (Hc) of about 970 Oe.

4. Conclusions

For this work, we prepared iron–nitride powders containing α-Fe, γ-F₄N, α′-Fe₈N, and α″-Fe₁₆N₂ crystalline phases via the gas atomization method. We experimentally demonstrated that by tuning the preparation parameters (ejection temperature and pressure, ejection nozzle diameter, and atomization pressure), we managed to increase the content of the α″-Fe₁₆N₂ phase in the atomized powders and improve their magnetic properties. It was also found that the preparation parameters significantly influence the morphology and dimensions of the powders. The best magnetic properties (Ms = 234.8 emu/g and Hc = 970 Oe) were obtained in Fe-N powders with a content of 31% α″-Fe₁₆N₂ and a size of approximately 30 μm. Although this experimental work verified the feasibility of using the gas atomization method to produce powders containing α″-Fe₁₆N₂ phase, overcoming some of the existing limitations at our home facility by increasing the melting temperature (using a crucible with high-temperature softening) or increasing the nitrogen pressure in the ejection chamber (for the time preceding the actual atomization) could lead to an improvement of the magnetic properties of the powders obtained.