The Magnetic Properties of a NdFeB Permanent Magnets Prepared by Selective Laser Sintering

Hard magnetic materials are classified among functional materials, which, in many respects, are the basis of modern technological processes, day-to-day operation devices, electrical transport, etc. The rate of improvement of the magnetic hysteretic properties of permanent magnets has steadily decreased because in the industrial production, the potential of the Nd₂Fe₁₄B compound has been realized almost completely [1]. To further improve the functional properties of articles with permanent magnets, new approaches to designing such articles should be used. The additive manufacturing of functional magnetic materials and articles based on them is among these approaches. Additive technologies have several significant advantages over subtractive (edge cutting machining) and forming (strain without moving off a material) technologies. One of the advantages consists in the possibility of preparing samples and articles of any form, which is limited by the mechanical properties of a material. The other advantage is the local tuning of the material properties at the preparation stage at the expense of varying both the chemical composition and microstructural state.

Some kinds of functional materials are prepared by additive manufacturing. A centrifugal pump for blood transfusion is shown in [2]. The pump case, pump impeller, and magnetic clutch based on permanent magnets were prepared by additive manufacturing. In [3, 4], the possibility of creating linear encoders by 3D printing of units made from magnetic materials with spatially varied magnetic properties was shown. Actuators [5], rotors for synchronous electrical engines [6], elements of neutron optics [7], and other articles are prepared by additive manufacturing. These devices were fabricated by material extrusion, which consists in the heating of a polymer binder with hard-magnetic material particles. The heating temperature usually is 370–420 К; this does not lead to microstructural changes of magnetic material and the method consists in placing magnetic material particles in a plastic and subsequent 3D printing. The presence of plastic itself leads to a decrease in the residual magnetization, temperature stability, and maximum energy product. As a result, the magnetic properties of such articles were found to be noncompetitive compared to those of permanent magnets prepared by sintering.

The fabrication of magnets without a binding polymer can be achieved using other additive technologies, in particular, binder jetting or selective laser sintering. The first of the technologies can be simply agglutination of particles [8] and fabrication of a magnet of a given shape. A more complex process is possible, which includes the agglutination of particles in the course of printing and subsequent removal of the binder and its replacement with a low-melting eutectic [9]. This approach allows one to prepare a porous magnet without using an organic binder, the coercivity of which exceeds that of the initial alloy. The production of permanent magnets by selective laser sintering or selective laser melting is much more common. The first studies were mainly related to the preparation of magnets of different shapes; in this case, the magnetic properties were not studied in detail [10]. Subsequently, the fabrication parameters of permanent magnets were studied and optimized. A coercivity equal to H_c = 8.7 kOe was achieved [11]. Further, in order to increase the coercivity, the grain boundary infiltration was used. The grain boundary infiltration process consists in “passing” the melt between nano- and microcrystalline grains. In this case, the grain boundary infiltration differs from grain boundary diffusion in the fact that the passing alloy is liquid [12]. It was shown experimentally that to realize infiltration in permanent magnets and rapidly quenched Nd–Fe–B alloys, alloys comprising a rare-earth metal, copper, and, sometimes another 3d-metal can be used [13‒15]. In [14], the grain boundary infiltration was performed after additive manufacturing of magnets, whereas in [16], the record coercivity of printer magnets was achieved in the absence of post-processing and heavy rare-earth metals. The general possibility of obtaining the anisotropy of magnetic properties of printed magnets via the formation of a temperature gradient was shown in [17]. An original approach to fast searching for the compositions for the 3D printing and estimation of potential magnetic properties was given in [18]. In recent years, there is a direction of studies related to the formation of microstructural peculiarities [19] and cracks in magnets [20].

The aim of the present study was to determine the effect of synthesis conditions for permanent magnets on the formation of magnetic hysteresis properties based on single-layer samples. Since the additive production assumes the layer-by-layer building of an object, the understanding of the potential of the magnetic properties within single layer allows one to predict the upper limit of magnetic hysteresis properties of a bulk permanent magnet sample.

For 3D printing, we used a mechanical mixture of two powders. The mixture consists of the rapidly quenched Nd_11.7Fe_77.6Co_5.4B_5.3 alloy of grade MQP-B and the Pr_75.0Cu_6.2Co_18.8 alloy, which were taken in the 80 : 20 wt % proportion. The content of 3d-elements in the Pr–Cu–Co alloy is due to the eutectic composition. The copper to cobalt ratio is related to the mechanical properties of the alloy since the lower cobalt content makes it difficult to mill the alloy to the fineness required for the additive production. According to available experimental data, to achieve the complete grain boundary infiltration in the MQP-B alloy, ~15 wt % of the Pr–Cu–Co alloy should be used. A slight excess is used in view of the possible oxidation of the rare-earth metal. The alloys were milled in ethanol using a ball mill. All milling processes were performed in ethanol in order to prevent the powder oxidation. The maximum particle size of powders used as the powder batch did not exceed 100 μm. The low melting addition was used in order to ensure (1) the liquid-phase sintering of powder particles of the main magnetic alloy; (2) the infiltration of the addition between crystallites of the main magnetic alloy, which is accompanied by the formation of paramagnetic interlayer enriched in the rare-earth metal and decreasing the intergranular exchange interaction; and (3) the protection of hard magnetic powder particles against the overheating at the expense of latent melting heat.

To perform the 3D printing process, a specially turned brass plate was placed on a building platform; the plate has a cylindrical cavity 1 mm in depth and 90 mm in diameter, in which the powder was located. The powder together with ethanol, which subsequently is used for the 3D printing, was uniformly applied manually on the plate. The powder was dried directly in the printer chamber. In the present study, we report results obtained for ~500 samples of printed single-layer magnets.

An Orlas Creator RA (Coherent, Germany) additive manufacturing machine was used for the 3D printing. The printing process was performed in an argon atmosphere; the oxygen content in the build chamber was no more than 0.5%. The following printing parameters were varied: laser power (P, W), the number of laser beam scanning over a sintered sample (N), hatch (h, μm), scanning velocity of laser beam at the building surface (v, mm/s), and laser beam diameter on the building surface (d, μm).

Prepared samples of planar magnets are rectangular parallelepipeds 10 × 10 mm in size and about 1-mm thick. Figure 1 shows the appearance of prepared samples.

The magnetic hysteresis properties were measured at room temperature in magnetic fields to H = 26 kOe using a KVANS-1 vibrating sample magnetometer and at 300 K in magnetic fields to H = 90 kOe using a PPMS DynaCool T9 equipped with a vibrating sample magnetometer (Quantum Design, United States) phase composition was studied by X-ray diffraction analysis using a D8 Advance powder diffractometer (Bruker, Germany) and CuKα radiation.

Figure 2 shows X-ray diffraction patterns of a printed sample at the surface subjected to laser beam action (surface 1) and at the opposite side (surface 2). The printing parameters are P = 52 W, v = 1500 mm/s, h = 27 μm, d = 890 μm, and N = 3.

The main phase in the sample is a Nd₂Fe₁₄B compound with a tetragonal structure (space group P4₂/mnm) and the lattice parameters a = 8.818(2) and c = 12.236(5) Å, which agree with those available in the database (COD: 1008718). According to the Bragg reflection broadening, the crystallite sizes at surfaces 1 and 2 are 55–65 and 45–50 nm, respectively. As is seen from the data given in Fig. 2, the phase composition and relationship of phases found for two surfaces differ substantially. At surface 1, the presence of α-Fe (with average crystallite sizes of 55–65 nm) is clearly seen; for surface 2, it is not clearly observed. As was noted, for example, in [1, 16], the presence of the Nd₂О₃ oxide is possible; it is difficult to identify using our X-ray diffraction data. The formation of the oxide is related to the overheating of surface 1 to higher temperatures under the action of laser beam. For surface 1, the presence of copper oxide also is possible. It was shown in [16] that different phases can be formed at surface 1; in particular, these are rare-earth hydroxides. However, in the present study, we used the carefully dried powder, and no hydroxides were formed.

The PrCo₅-based phase with the hexagonal structure (space group P6/mmm) and crystallite sizes of ~35–40 nm is found for both the surfaces. It is important to note that the lattice parameters of the phase formed upon printing are a = 5.162(2) and c = 3.856(9) Å, which differ substantially from those of the pure PrCo₅ phase; this is likely to be related to the substitution of neodymium for praseodymium and of copper and iron for cobalt. This phase, like the main magnetic phase, has uniaxial magnetic anisotropy and PrCo₅-based permanent magnets with a coercivity of more than 10 kOe can be prepared [21].

The presence of relatively large α-Fe grains, whose size exceeds 500 nm [16], allows us to expect inflection in the major hysteresis loop at low magnetic fields. The presence of PrCo₅-based crystallites also allows us to expect inflection at a demagnetizing field of 2–6 kOe.

Figure 3 shows the major hysteresis loops for the single-layer permanent magnet samples prepared by selective laser sintering and for the mixture of initial powders (before laser sintering). The synthesis parameters are P = 52 W, N = 3, h = 135 μm, and v = 1500 mm/s; the laser beam diameter on the building surface was varied from d = 160 to d = 890 μm. Instead of homogenizing annealing in a furnace, upon printing, three passes of laser beam (N = 3) were used instead of single pass in order to increase the time in which the grain boundary infiltration can occur.

The reported magnetization reversal curves are characterized by two inflections. The first inflection is observed near the demagnetizing field H = 0 kOe; the second inflection is observed near H = –10 kOe. The first inflection is due to the overheating of surface layer of particles upon action of laser radiation and formation of α-Fe crystallites. At the same printing parameters, the increase in the laser beam diameter at the build surface leads to a decrease in the surface layer temperature and, therefore, to the lower fraction of iron in the samples. As the laser beam diameter at the building surface increases, this manifests itself in the decrease in the value of inflection and an increase in the magnetization in a range of magnetic fields of –9 to 0 kOe. The second inflection is related to the magnetization reversal of the MQP-B alloy volumes, within which no grain boundary infiltration occurs, whose coercivity is 10 kOe. No clear inflection due to the magnetization reversal of PrCo₅-phase grains is observed; this can be related to small volumes of this phase. The inset in Fig. 3 shows the dependence of the ratio of specific magnetization σ_–15 measured in the field H = –15 kOe to the magnetization σ₅ measured in the field H = +5 kOe. Since the coercivity of the infiltrated alloy is higher than that of the starting alloy, the increase in the fraction of infiltrated alloy leads to the increase in the magnetization in the demagnetizing fields, which are lower than the coercivity of this alloy and higher than the coercivity of the starting alloy. Owing to the fact that the coercivity of the starting MQP-B alloy is H_c ≈ 10 kOe, and iron formed in the surface layer is not exchange-coupled with the main phase, the fields H = 5 kOe (all ferro- and ferrimagnetic phases have a positive projection on the magnetizing field direction) and H = – 15 kOe (iron crystallites and base alloy volumes, within which no intergranular infiltration occurs, have a negative projection on the magnetizing field direction, whereas the infiltrated volume of sample has a positive projection) were selected for the determination of the specific magnetization. The higher the volume of magnet in which the infiltration occurred was, the lower σ_–15/σ₅ was. For the starting hard magnetic alloy, σ_–15/σ₅ ≈ 1.

The dependence σ_–15/σ₅(d) exhibits a nonmonotonic behavior and a minimum at d = 800 μm. The decrease in the alloy temperature under the surface layer characterized by the oxidation of rare-earth metals, which takes place because of the increase in d and an decrease in the density of laser beam energy at the build surface, leads, on the one hand, to a decrease in the oxidation rate of rare-earth metals and, on the other hand, to a decrease in the rate of intergranular infiltration. The competition of variations of the rates of these processes leads to the nonmonotonic dependence shown in Fig. 3. The decrease in the oxidation rate leads to a decrease in the inflection near H = 0 and increase in the volume of sample, within which the intergranular infiltration can occur. However, the decrease in the temperature of the powder layer (within which the melting of eutectic alloy occurs) below a critical value leads to the infiltration of liquid eutectic in the main magnetic powder without the intergranular infiltration and to changing the trend in the dependence σ_–15/σ₅(d).

Figure 4 shows the dependence of the coercivity of samples prepared by printing in varying the velocity of laser beam over the build surface. The following synthesis parameters were used: P = 52 W, N = 3, h = 135 μm, and d = 890 μm. As the velocity of laser beam over the build surface increases from 675 to 2025 mm/s, the increase in the coercivity is due to the decrease in the thickness of the magnet layer, in which the oxidation of the main magnetic phase and the formation of iron crystallites take place and the increase in the layer thickness within which the grain boundary infiltration occurs. The magnetization reversal curve of the sample synthesized at the velocity of laser beam over the building surface v = 675 mm/s exhibits inflection near H = 0 kOe. At the same time, for the major hysteresis loop of the sample synthesized at v = 2025 mm/s, such an inflection almost is not observed. The further increase in the velocity v causes the decrease in the coercivity because of the low temperature, to which low-melting eutectic particles are heated, and, therefore, the decrease in the rate of infiltration and main magnetic material volume, in which the grain boundary infiltration occurs. The magnetization reversal curve of the sample synthesized at v = 3375 mm/s exhibits inflection at H ≈ 10 kOe. In this case, the coercivity becomes lower than that of the starting MQP-B alloy because of the local oxidation; this is indirectly confirmed by the inflection in the major curve. In the range of magnetization reversing fields H = –15 kOe, the ascending and descending branches of the major hysteresis loops do not coincide; this indicates the presence of particles characterized by a coercivity of more than 15 kOe. Such particles can be obtained in the case when they are surrounded on all sides by a paramagnetic interlayer and the exchange interaction between them is suppressed. One direction for further increase in the coercivity can be changing the synthesis conditions in order to increase the fraction of particles via the increase in the temperature and time of grain boundary infiltration.

Figure 5 shows the portion of the major hysteresis loop of the sample synthesized at P = 52 W, N = 3, h = 27 μm, d = 890 μm, and v = 2025 mm/s. The coercivity is H_c = 19.5 kOe; in this case, necking is observed that can be due to the presence of iron grains that are exchange coupled with main hard-magnetic phase grains, PrCo₅ phase, and exchange-coupled crystallites within starting MQP-B powder alloy particles. It is likely that within these particles, the grain-boundary infiltration has not occurred because of, for example, the absence of low-melting eutectic particles. For this sample, the 3D printing conditions differ from those for the sample characterized by the maximum coercivity (Fig. 4) only in the hatch. The five-fold decrease in this parameter, namely, from 135 to 27 μm leads to the fact that that possible time of grain-boundary infiltration also increases five times at an almost unchanged temperature. The indicated combination of the parameters leads to temperatures at which the diffusion of low-melting eutectic occurs, and the growth of main phase crystallites is suppressed. The main fraction of oxygen incorporated into the sample is fixed in the form of oxide of the rare-earth metal comprising the Pr–Cu–Co alloy. This is confirmed by X-ray diffraction data.

In the present study, we report the main regularities of the formation of magnetic hysteresis properties of the Nd–Fe–B alloys in the course of additive manufacturing. Based on the example of single-layer magnets, the general possibility of preparing the high-coercivity state of permanent magnets by selective laser melting is demonstrated.

The coercivity equal to H_c = 19.5 kOe was obtained for a material free of heavy rare-earth metals. To our knowledge, this value is the maximum one for samples prepared by selective laser melting.