Exchange coupling between soft magnetic ferrite and hard ferromagnetic Sm₂Fe₁₇N₃ in ferrite/Sm₂Fe₁₇N₃ composites Open Access

Sm₂Fe₁₇N₃ compounds are good candidates for high performance magnets because of their high saturation magnetization and strong uniaxial anisotropy.^1,2 Since Sm₂Fe₁₇N₃ compounds decompose into SmN and α-Fe when heated above 873K, it is fairly difficult to obtain fully dense magnets. Although several researchers have attempted to prepare fully dense magnets by using shock compression, aerosol deposition, hot-isostatic pressure, compression shearing, cylindrical explosive consolidation and high-pressure current sintering,^3–8 the Sm₂Fe₁₇N₃ magnets produced by those methods have not yet achieved practical application because, other than oxidation resistance, Sm₂Fe₁₇N₃ magnets do not provide any overwhelming advantages, such as high resistivity or low cost processing, over existing sintered rare-earth magnets. Thus, at present, their application is limited to resin bonded magnets.

In permanent magnets which are operated at high frequency in advanced applications such as electrical vehicle magnets, eddy current loss is a serious problem. Although Nd₂Fe₁₄B magnets are often used in this application, their resistivity is approximately 100 μΩcm at most. Thus, a ten-fold increase in resistivity is necessary. The resistivity of Sm₂Fe₁₇N₃⁹ is four times higher than that of Nd₂Fe₁₄B, but this is still not sufficient for high frequency applications. Our basic strategy is to develop a ferrite/Sm₂Fe₁₇N₃ composite magnet which provides enhanced resistivity without greatly reducing magnetic properties. In our previous paper,⁹ we successfully fabricated ferrite/Sm₂Fe₁₇N₃ composite magnets from explosive-consolidating Sm₂Fe₁₇N₃ powders which were coated in advance with a continuous iron ferrite layer in an aqueous solution. We estimated the resistivity for fully dense ferrite/Sm₂Fe₁₇N₃ magnets as being 4000 μΩcm, which is 10 times higher than the estimated resistivity of fully dense Sm₂Fe₁₇N₃ magnets, while their remanence B_r and coercivity H_cJ were only slightly reduced in comparison with sample Sm₂Fe₁₇N₃ magnets. The magnetization curves had no inflection, which suggests that the soft magnetic ferrite layer is exchange-coupled with the hard ferromagnetic Sm₂Fe₁₇N₃ particles.

Our ferrite/Sm₂Fe₁₇N₃ composites consist of coarse hard magnetic particles (diameter: 2μm, not nano-sized) which have a comparative thick (50nm) soft magnetic layer. In the well-known exchange spring magnet, the optimal thicknesses of the hard magnetic phase is 2δ (δ : domain wall thickness) and that of soft magnetic phase is 2δ_s. These thickness values are given by δ = π (A_h/K_h)^1/2 and δ_s = π (A_s/2K_h)^1/2, where A_h and A_s are the exchange stiffness constants of the hard and soft magnetic phases, respectively, and K_h is the magnetocrystalline anisotropy constant of the hard magnetic phase. The values of 2δ and 2δ_s are almost equal to each other¹⁰; therefore, these optimal thicknesses do not exceed a few 10nm, because δ of rare-earth magnet material is about 4-9nm.¹¹ Similar results for these optimal thicknesses have also been reported based on the results of three-dimensional computer simulations.¹² Exchange-coupling composites having a “large hard magnetic core and thick soft magnetic layer (LHATS)” have been designed not only by the present authors, who used ferrite/Sm₂Fe₁₇N₃, but also by several other researchers who used, for instance, α-Fe/Sm-Co or Nd-Fe-B.^13,14 LHATS magnets have a number of merits, in that the large hard core aligns along the direction of an external magnetic field and the thick soft layer results in a large volume fraction of the soft magnetic phase. Nevertheless, these designs are in principle very difficult to realize due to the thin 2δ_s and the fine critical diameter d₀ for a single domain particle of the soft magnetic phase.

The Sm₂Fe₁₇N₃ powders (particle size: 2μm) were provided by Sumitomo Metal Mining Co., Ltd..¹⁵ The Sm₂Fe₁₇N₃ powders were coated with an iron ferrite (Fe₃O₄-γFe₂O₃ intermediate) layer by ferrite plating, which is an aqueous process.¹⁶ After etching the surfaces of the Sm₂Fe₁₇N₃ powders with an acidic solution, the powders were dispersed in distilled water, to which a FeCl₂ reaction solution (pH up to 4) and a pH adjusting solution of KOH (pH up to 14) were added with continuous stirring under an air atmosphere.¹⁷ Ferrite/Sm₂Fe₁₇N₃ and Sm₂Fe₁₇N₃ powder compacting (PC) magnets were then formed at less than 1GPa in a field of 15 kOe by die pressing. The magnets were compacted by an explosive consolidation (EC) technique using water as the transmitting medium¹⁸ with no external magnetic field throughout the preparation process. Conventional magnetization measurements for variable demagnetization and recoil curves were performed using a vibrating-sample magnetometer (VSM). Resistivity (ρ) was measured by the four-probe method.

The volume ratio of ferrite to Sm₂Fe₁₇N₃ in ferrite/Sm₂Fe₁₇N₃ EC magnets was estimated as 15:85 based on the results of SEM observations (e.g., Fig. 1(B)) and the densities of the magnets. TEM observation (Fig. 1(A)) revealed that the ferrite coating phase had a grain size of about 10nm. Fig. 1(D) shows a SEM photograph of iron ferrite nano-powders (IFNP) prepared by the above-mentioned aqueous process without Sm₂Fe₁₇N₃ powders. The particle size of these IFNP is within a range of 5-50nm.

The crystalline structure of the IFNP, which corresponds to that of the ferrite coating phase in our magnets, is of the cubic spinel-type with no impurity phase, as revealed by XRD analysis. The lattice constant is smaller than that of magnetite (Fe₃O₄), since the iron ferrite prepared by ferrite plating becomes an intermediate between magnetite (Fe₃O₄) and maghemite (γ-Fe₂O₃).^9,16

Figure 2(A) shows a comparison of the demagnetization and recoil curves for Sm₂Fe₁₇N₃ and ferrite/Sm₂Fe₁₇N₃ EC magnets. These curves were all measured after saturation by a 60 kOe pulse field. The curves gave a recoil permeability (μ_r) of 1.31-1.35 for ferrite/Sm₂Fe₁₇N₃, which was slightly higher than that (1.23-1.28) for Sm₂Fe₁₇N₃, as show in Fig. 2(B). Here, recoil permeability was calculated by μ_r = 1 + 4π(M_d − M₀) / H_d, using magnetization M_d under the reverse field H_d (H_d: negative value), and magnetization M₀ under the zero field on each of the recoil curves.

The results described above prove that exchange coupling acts between the Sm₂Fe₁₇N₃ core and ferrite coating layer in ferrite/Sm₂Fe₁₇N₃ EC magnets.¹⁷ 

The detailed TEM observation results shown in Fig. 3(B) reveal that the Sm₂Fe₁₇N₃ main phase is directly coupled to the ferrite phase. Namely, there is no phase between the main phase and the coating phase. This is because the amorphous surface oxide layers (thickness: 10nm) observed in the Sm₂Fe₁₇N₃ powders^8,19 were removed by etching the surfaces of the Sm₂Fe₁₇N₃ particles prior to coating the powders with ferrite.¹⁷ This enabled exchange coupling between the ferrite and Sm₂Fe₁₇N₃ phases.

Table I gives the magnetic properties of the Sm₂Fe₁₇N₃ and ferrite/Sm₂Fe₁₇N₃ magnets in this work. Introducing the ferrite coating phase caused only slight degradation of the remanence of the EC magnets, from 6.24 kG (Sm₂Fe₁₇N₃) to 5.98 kG (ferrite/Sm₂Fe₁₇N₃), or a reduction ratio Δ B_r of about 4.2%. This value was very small in spite of the introduction of a large amount of 15 vol% ferrite phase in the Sm₂Fe₁₇N₃ EC magnets.

The models used to calculate Δ B_r are shown in Fig. 3(A). The details of the above estimation are as follows.

According to the Stoner-Wohlforth theory, the remanence of isotropic single domain particles is a half value of saturated magnetization 4πM_s. We conjecture that the reduction ratio of remanence Δ B_r of the ferrite/Sm₂Fe₁₇N₃ to Sm₂Fe₁₇N₃ EC magnets was in proportion to the reduction ratio of this intrinsic 4πM_s.

Introducing the ferrite coating phase degraded the observed remanence of the EC magnets by approximately Δ B_r obs.= 4.2% . First, therefore, we performed a calculation assuming that the ferrite phase simply covered the Sm₂Fe₁₇N₃ starting powder (see Fig. 3(A)(ii)). In this case, the reduction ratio (Δ B_r calc.1[%]) of B_r is

where, 4πM_N (=15.7 kG) and 4πM_O (= 6.0 kG) are the 4πM_s of Sm₂Fe₁₇N₃ and ferrite, ν_SO (= 0.03) is the volume fraction of the surface oxide layer of the Sm₂Fe₁₇N₃ particle, η_S and η_f are the packing fractions of the Sm₂Fe₁₇N₃ and ferrite/Sm₂Fe₁₇N₃ EC magnets and the volume ratio of ferrite to Sm₂Fe₁₇N₃ in ferrite/Sm₂Fe₁₇N₃ is v_O: v_N.

From formula (1), Δ B_r calc.1 is estimated as being approximately about 7.1%. However, this value is inconsistent with Δ B_r obs..

Next, we assumed that the surface oxide layers of the Sm₂Fe₁₇N₃ particles in the ferrite/Sm₂Fe₁₇N₃ EC magnets were removed by the etching step in the ferrite plating procedure,¹⁷ as shown Fig. 3(A)(i). Based on this assumption, formula (1) is modified to the following formula:

The corrected calculation value Δ B_r calc.2 equaled approximately 4.4%. Thus, Δ B_r calc.2 was essentially equal to Δ B_r obs., if it is assumed that the surface oxide layers of the Sm₂Fe₁₇N₃ particle were removed in the ferrite plating process.

From the calculation by formula (2), almost no surface oxide layer from the Sm₂Fe₁₇N₃ starting powder existed between the ferrite phase and the Sm₂Fe₁₇N₃ phase, revealing that the bond by the exchange interaction functioned strongly.

Figure 4 shows a comparison of the demagnetization curves for a) ferrite/Sm₂Fe₁₇N₃ PC magnet and b) Sm₂Fe₁₇N₃-iron ferrite mixed PC magnet. The starting mixed powders of the PC magnets in Fig. 4 were fabricated by mixing Sm₂Fe₁₇N₃ powders (particle size: 2 μm) and IFNP in an agate mortar with hexane. These PC magnets contain 15 vol% soft ferrite and have densities of a) 5.31 and b) 5.30 g/cm³, respectively. The demagnetization curve of a) has no inflection, which suggested that the ferrite layer is exchange-coupled with the Sm₂Fe₁₇N₃ particles. The curve of b) has an inflection point at 2 kOe, which indicates that no exchange coupling occurred in this magnet.

As shown in Table I, these PC magnets exhibit the resistivities of a) 7500 and b) 2300 μΩcm, respectively. Iron ferrite powders are present only between the grains of the Sm₂Fe₁₇N₃ powders in the Sm₂Fe₁₇N₃-magnetite mixed PC magnet, and did not form a coating on the Sm₂Fe₁₇N₃ particles. Therefore, the electric current permeates through the boundary formed by only the Sm₂Fe₁₇N₃ particles, and electrical resistivity is remarkably smaller compared with that of the ferrite/Sm₂Fe₁₇N₃ PC magnet, in which the Sm₂Fe₁₇N₃ particle surface is almost completely coated with a ferrite layer (see Fig. 1(B)).

Thus, the soft ferrite layer causes extreme deterioration of the magnetic properties of composite magnets without exchange coupling between a ferrite and a Sm₂Fe₁₇N₃ phase, while providing only slight improvement in electrical resistivity.

Both δ and d₀ of iron ferrite are about 100-200nm,^20,21 which is larger than those of other typical metallic soft magnetic materials. It is therefore thought that magnetic reversal does not occur easily from the thick soft layer (50nm) in our exchange-coupling LHATS magnets, even under a high demagnetization field of several kilo-oersted. As a result of this feature, or magnet retains exchange-spring characteristics on the demagnetization curves.

The ferrite layer in our ferrite/Sm₂Fe₁₇N₃ composite magnet retains magnetic exchange coupling among Sm₂Fe₁₇N₃ grains, and yet suppresses intergrain electrical coupling, thereby increasing resistivity.

Ferrite/Sm₂Fe₁₇N₃ composite magnets were prepared by a novel process with the aim of realizing enhanced resistivity, and their exchange-spring behavior was evidenced by recoil permeability measurements. The Sm₂Fe₁₇N₃ main phase is directly coupled to the ferrite coating phase, which was also clearly evidenced by TEM observations in combination with calculation of the reduction ratio of remanance. Due to the spring-exchange interaction, the resistivity of the soft ferrite layer can be increased with no serious reduction of magnetic properties. This result is expected to give new impetus to the development of high resistivity magnets, and will also lead to decreased eddy current loss and improved high frequency characteristics in composite magnets.