Fabrication of Anisotropic Nd–Fe–B Powders by Ta Sputtering

Abstract

The influence of Ta sputtering on the fabrication of anisotropic Nd–Fe–B powders was investigated. X-ray fluorescence and secondary electron microscopy analysis revealed that the measured Ta content after sputtering was 5.6 mass% and the Ta particles were located on the surfaces of the Nd–Fe–B grains. Ta coating on Nd–Fe–B grain could prevent necking between Nd–Fe–B grains after annealing at 700°C for Nd–Fe–B powders with and without Ta sputtering, and the Ta-sputtered powder could be easily crushed.

After crushing and sieving under 75 µm, the coercivity value of the powders with and without Ta sputtering were almost the same, while the anisotropy of Ta-sputtered powder was higher. Upon annealing and crushing, the degree of texture (λ), which is used to evaluate anisotropy, improved from 0.08 to 0.73 for the Ta-sputtered Nd–Fe–B powder and was higher than that for the non-Ta-sputtered Nd–Fe–B powder. The results demonstrate that Ta sputtering can be an effective technique for the fabrication of anisotropic Nd–Fe–B powders.

Fig. 9 Influence of particle size on the coercivity and DOT.

1. Introduction

Numerous studies have been conducted to improve the magnetic properties of Nd–Fe–B magnets,^1–4) leading to their use in a variety of applications owing to their high maximum energy product ((BH)_max). The coercivity mechanism of Nd–Fe–B magnets involves the nucleation of reversed magnetic domains generated at regions with low magnetocrystalline anisotropy.⁵⁾ Therefore, the coercivity can be improved by suppressing the generation of reversed magnetic domains. There are two main methods for increasing the coercivity. The first method is decreasing the grain size of the Nd₂Fe₁₄B phase,^6–8) which reduces the concentration of multi-domain particles. In other words, grain refinement decreases the demagnetization factor,⁹⁾ as expressed in the Kronmüller equation,¹⁰⁾ and makes the coercivity approach the anisotropy field. The second method is controlling the microstructure at the interface between the Nd₂Fe₁₄B phase and Nd-rich phase. During the annealing process, the Nd-rich phase liquefies and coats the Nd₂Fe₁₄B grains, affording a smoother grain surface and decreasing the number of nucleation sites for reversed magnetic domains. In addition, a homogeneous Nd-rich phase with high Nd content improves the magnetic isolation of the Nd₂Fe₁₄B main phase.

Nd–Fe–B magnets can be classified into two types; sintered magnets fabricated by powder metallurgy, and bonded magnets consisting of Nd–Fe–B powder and resin. The Nd–Fe–B sintered magnets are prepared from strip-cast (S.C.) Nd–Fe–B alloy. The S.C. Nd–Fe–B alloy has lamellar-like structure consisting of dendritic Nd₂Fe₁₄B and Nd-rich phases^11,12) illustrated in Fig. 1(a). The S.C. alloy is pulverized after hydrogen decrepitation process (Fig. 1(b)), and single crystalline powder with dispersed Nd-rich phase, which is suitable powder for sintered magnets, can be obtained by pulverization (Fig. 1(c)). The bonded Nd–Fe–B magnets are fabricated by mixing the Nd–Fe–B powder with a resin. Although bonded Nd–Fe–B magnets exhibit lower maximum energy products than sintered Nd–Fe–B magnets, they are highly versatile owing to their ability to be fabricated in different shapes and sizes. Bonded Nd–Fe–B magnets can be divided into isotropic and anisotropic magnets. Anisotropic magnets exhibit superior magnetic properties compared with isotropic magnets, including a value of (BH)_max that can be as much as twofold higher. However, few methods of preparing anisotropic powders for Nd–Fe–B magnets have been reported. It is well known that anisotropic Nd–Fe–B powders cannot be fabricated by pulverization method because of decreasing coercivity owing to the relatively low anisotropy field of Nd₂Fe₁₄B. The hydrogenation-disproportionation-desorption-recombination (HDDR) process^13–20) and the combination of hot deformation and milling have been reported as good fabrication methods for anisotropic Nd–Fe–B powders. However, the anisotropy of HDDR powders is limited (the misaligned angles between the Nd₂Fe₁₄B grains are typically up to approximately 35°¹⁸⁾), and the combination of hot deformation and milling is associated with high processing costs. Thus, improved fabrication methods for highly anisotropic Nd–Fe–B powders with high coercivity would be of great value for obtaining high-performance anisotropic bonded magnets. Figure 2 shows a schematic depiction of an ideal bonded magnet prepared using anisotropic Nd–Fe–B powder. Here, we have to note that the arrows in Fig. 2 shows easy axes of magnetization. The powder consists of single crystals of Nd₂Fe₁₄B in which the easy axes of magnetization are aligned along one direction, and the Nd₂Fe₁₄B grains are coated with Nd-rich phase to obtain high coercivity by decreasing number of nucleation sites for reversed magnetic domains. The powder is mixed with a resin under an applied magnetic field to align the easy axes of magnetization, and ideal anisotropic bonded Nd–Fe–B magnet can be obtained. Coating the Nd-rich phase on the surfaces of the Nd₂Fe₁₄B grains, annealing at temperatures from 500 to 700°C is considered effective for increasing the H_cJ. This annealing step causes the phase change of the Nd-rich phase, and the liquefied Nd-rich phase coats the surfaces of the Nd₂Fe₁₄B grains resulting in increment of H_cJ. However, the liquefied Nd-rich phase can also bind Nd₂Fe₁₄B grains in a random manner resulting in formation of magnetic chains and hindering grain alignment. Therefore, preventing the binding of the Nd₂Fe₁₄B grains during annealing is crucial for fabricating highly anisotropic Nd–Fe–B powders.

Fig. 1

Schematic depiction of microstructure of strip-cast alloy and pulverizing process.

Fig. 2

Schematic depiction of a high-performance bonded magnet based on Nd–Fe–B powder.

Because Ta possesses a high melting point (3020°C), it does not melt during annealing at 500–700°C. Ta also exhibits low solubility in Nd–Fe–B alloys. From these properties, Ta coating of the surfaces of Nd–Fe–B powders was expected to suppress the binding of Nd₂Fe₁₄B grains during annealing. It was considered that during annealing the liquefied Nd-rich phase would spread at the interfaces between the Nd₂Fe₁₄B grain and Ta coated layer and cover the surfaces of the Nd₂Fe₁₄B grains. Thus, Ta coating and subsequent annealing was expected to be an effective method for obtaining anisotropic Nd–Fe–B powders with high coercivity.

In this study, Ta coating was achieved using the sputtering method, which is capable of delivering a uniform coating of Ta on the surfaces of the Nd–Fe–B powder when the Nd–Fe–B powder is stirred during sputtering. The purpose of this study was to investigate the influence of Ta sputtering on the magnetic properties and microstructures of Nd–Fe–B powders, which were prepared by pulverization of S.C. alloy, with the goal of fabricating anisotropic Nd–Fe–B powders for use in high-performance bonded magnets.

2. Experimental Procedure

Figure 3 shows flow chart of sample preparation process and magnetic properties evaluation process in this study. The starting material was a strip-cast alloy with a composition of Fe_66.0Co_0.94Nd_27.5Pr_4.2B_1.02Al_0.26Cu_0.1 (mass%). This alloy was crushed into a coarse powder by hydrogen decrepitation and then pulverized into a fine powder with an average particle size of 4.2 µm by jet-milling. Ta was deposited on the Nd–Fe–B powder by magnetron sputtering using a base pressure of <1 × 10⁻⁶ Pa, sputtering pressure of 5.0 Pa, sputtering power of 40 W, and sputtering time of 240 min. The Nd–Fe–B powder was placed in a cup and agitated by vibration during the sputtering process. The resulting powder was annealed at 700°C for 30 min under vacuum of <9.9 × 10⁻⁵ Pa.

Fig. 3

Flow chart of (a) sample preparation and (b) measurement of magnetic properties.

The composition of the powder was determined by X-ray fluorescence (XRF) analysis, and the microstructure was examined by field-emission scanning electron microscopy (FE-SEM) with energy dispersive X-ray spectroscopy (EDS). For microstructure observation, powder was fixed by a paste containing Ag particles, and it was polished. The magnetic properties were measured using a vibrating sample magnetometer (VSM). Prior to the VSM measurements, the powder (20 ± 3 mg) was embedded in molten paraffin and the grains were aligned by applying a magnetic field of 1.6 MA·m⁻¹, and then the sample was magnetized by applying a pulsed magnetic field of 6.4 MA·m⁻¹ (Fig. 3(b)). In this study, demagnetization factor correction was not performed, and the magnetization value under an applied magnetic field of 1.6 MA·m⁻¹ was defined as the saturation magnetization (σ_1.6).

3. Results and Discussion

The composition of the Nd–Fe–B powder after Ta sputtering was analyzed by XRF, which revealed a Ta content of 5.6 mass%. Figure 4 presents secondary electron (SE) and Ta mapping images of the Nd–Fe–B powders before and after Ta sputtering, prior to the annealing step. In the SE images, no obvious changes were observed in the grain shape of the Nd–Fe–B powder after Ta sputtering, nor was aggregation of the Nd–Fe–B powder due to bonding by Ta observed. The elemental mapping image of the powder after Ta sputtering (Fig. 4(d)) revealed the presence of Ta at the surfaces of some, but not all, of the Nd–Fe–B grain. This result indicates that the Ta coating was inhomogeneous. Figure 5 shows high-magnification SE images of the Nd–Fe–B powders before and after Ta sputtering. Prior to sputtering, the grains of the Nd–Fe–B powder exhibit a smooth surface (Fig. 5(a)), whereas after sputtering, numerous tiny particles were observed (Fig. 5(b)). Considering the elemental mapping results demonstrating the presence of Ta at the surfaces of some of the Nd–Fe–B grains (Fig. 4(d)), these tiny particles were considered to be Ta. To evaluate the changes in the magnetic properties, VSM measurements were conducted for the powders before and after Ta sputtering. The measured saturation magnetization (σ_1.6) values of the powders before and after Ta sputtering were 154 and 143 A·m²·kg⁻¹, respectively, corresponding to a difference of 11 A·m²·kg⁻¹ between the two samples. The decrease magnetization of the powder after sputtering was ascribed to its Ta content of approximately 6 mass%. The measured coercivity values of the powders before and after sputtering were almost the same at 19.6 and 28.7 kA·m⁻¹, respectively. Both H_cJ values were low because these samples were prior to annealing. These results indicate that the Ta was present at the surfaces of the grains, and Ta sputtering did not cause deterioration of the magnetic properties. The Ta sputtering process was therefore considered an effective method for coating Nd–Fe–B powders.

Fig. 4

SE and Ta mapping images of the Nd–Fe–B powders (a), (b) before and (c), (d) after Ta sputtering.

Fig. 5

High-magnification SE images of the Nd–Fe–B powders (a) before and (b) after Ta sputtering.

The two powders were then subjected to annealing, and their magnetic properties and microstructures were investigated. Figure 6 shows the demagnetization curves for the annealed Nd–Fe–B powders with and without Ta sputtering. The saturation magnetization (σ_1.6), remanence (σ_r) and coercivity were evaluated from the demagnetization curves measured along the alignment direction. To evaluate the anisotropy of the powders, the degree of texture (DOT, λ) was defined as follows:^21,22)   

\begin{equation} \lambda = \frac{\sigma_{r}(\parallel) - \sigma_{r}(\perp)}{\sigma_{r}(\parallel)} \end{equation} (1)

where σ_r(//) is the remanence of the sample measured parallel to the magnetic alignment direction and σ_r(⊥) is the remanence measured perpendicular to the magnetic aligned direction, and λ = 0 and λ = 1 indicate isotropy and anisotropy, respectively. The remanence values were obtained from the demagnetization curves shown in Fig. 6.

Fig. 6

Demagnetization curves for the annealed Nd–Fe–B powders (a) without and (b) with Ta sputtering.

Based on the results presented in Fig. 6, the coercivity of the annealed powder without Ta sputtering was 1270 kA·m⁻¹, whereas that of the annealed powder with Ta sputtering was 660 kA·m⁻¹. One explanation for the decrease in H_cJ could be oxidation of the Nd–Fe–B powder during the sputtering process. The DOT (λ) values of the samples with and without Ta sputtering were 0.08 and 0.05, respectively, indicating low anisotropy. To clarify the reason for the low values of λ regardless of whether Ta sputtering had been performed, the microstructures of the powders were examined.

Figure 7 presents the backscattered electron (BSE) images and elemental mapping for the polished surfaces of the annealed Nd–Fe–B powders with and without Ta sputtering. Figure 7(c) and (d) show magnified BSE images of the regions enclosed by the yellow rectangles in Fig. 7(a) and (b), respectively. In Fig. 7(c) and (d), the elemental mapping results are also superimposed, and the elements Nd, Fe, and Ta are indicated in blue, green, and magenta, respectively. In the BSE image of the powder without Ta sputtering (Fig. 7(a)), a phase with white contrast can be observed between the particles with gray contrast. This interconnection between the gray particles indicated the formation of secondary particles. The elemental mapping results (Fig. 7(c)) revealed that the phase with white contrast at the grain boundary was the Nd-rich phase and the phase with gray contrast was Nd₂Fe₁₄B. These findings suggest that necking of Nd₂Fe₁₄B particles bonded by the Nd-rich phase (indicated by yellow arrows) occurred during annealing in the sample without Ta sputtering. In the sample with Ta sputtering (Fig. 7(b) and (d)), the Nd-rich phase was observed at the surfaces of the Nd₂Fe₁₄B particles, along with some necking Nd₂Fe₁₄B particles. Figure 7(d) reveals that Ta was also present at the surfaces of the Nd₂Fe₁₄B particles. Although Ta was deposited at the surfaces of the Nd–Fe–B grain, binding of the Nd₂Fe₁₄B particles by Nd-rich phase (indicated by yellow arrows) also occurred during annealing, which could explain the low λ. As described in Fig. 4 and 5, Ta coated surface of Nd–Fe–B powder, but not all. Thus, some Nd-rich phase, which was un-coated by Ta, liquefied and bound powders resulting in low λ.

Fig. 7

BSE and elemental mapping images of the annealed Nd–Fe–B powders (a), (c) without and (b), (d) with Ta sputtering.

It can be obvious that the area of necking Nd₂Fe₁₄B particles in the sample with Ta sputtering was smaller than that in the sample without Ta sputtering. This indicates that the binding strength between the Nd₂Fe₁₄B particles was lower in the sample with Ta sputtering, which was expected to permit easier crushing of the bound powder into a finer. Thus, the two annealed powders were crushed using an agate mortar and pestle, and the magnetic properties and microstructures were examined after sieving to size of <150 µm and <75 µm.

Figure 8 shows the demagnetization curves for the annealed Nd–Fe–B powders with and without Ta sputtering after crushing and sieving to a particle diameter of <75 µm. It has to be noted that VSM measurement conditions were same as described above. The sieved powders were aligned under applying magnetic field of 1.6 MA·m⁻¹ in molten paraffin wax, and then, it was magnetized to the aligned direction by applying pulsed magnetic field of 6.4 MA·m⁻¹ (Fig. 3(b)). DOT was evaluated by eq. (1). In the sample without Ta sputtering in Fig. 8(a), λ increased from 0.05 to 0.55, and H_cJ which was measured parallel to aligned direction decreased from 1270 to 337 kA·m⁻¹ upon crushing. The reduction of H_cJ was considered to originate from the introduction of defects at the surfaces of the Nd₂Fe₁₄B grains during crushing. In contrast, after crushing, the sample with Ta sputtering exhibited a λ of 0.73 in Fig. 8(b), which was 0.18 higher than the sample without Ta sputtering in Fig. 8(a). The H_cJ values of both samples decreased upon crushing, although the sample with Ta shows higher value of H_cJ (407 kA·m⁻¹) than the sample without Ta sputtering. These results indicate that Ta sputtered Nd–Fe–B powder was easier to pulverize than powder without Ta, thus induced damages such as defects and strain at the powder surface during crushing in agate mortar was smaller in the powder with Ta than that without Ta. It is well known that a nucleation of reversed magnetic domain coming from surface of Nd₂Fe₁₄B phase deteriorates their coercivity, therefore it is thought that the crushing in an agate mortar could induce nucleation sites at the surface of Nd₂Fe₁₄B phase resulting into decrease in coercivity.

Fig. 8

Demagnetization curves for the crushed annealed Nd–Fe–B powders (a) without and (b) with Ta sputtering.

Figure 9 shows the influence of the powder size on the magnetic properties of the annealed Nd–Fe–B powders with or without Ta sputtering. For both samples, λ increased and H_cJ decreased with decreasing powder size. It is considered that λ and coercivity exhibit a trade-off relationship. The value of λ approached that of raw powder, (λ = 0.86), with decreasing powder size due to crushing, and higher λ values were observed for the samples with Ta sputtering. The λ of sieved powder with Ta sputtering was higher than that of without sputtering, while the H_cJ of both sieved powder was almost same. Consequently, Fig. 9(b) indicates that Ta sputtering was effective for improving λ, therefore to clarify the reason for the increased anisotropy upon crushing, the microstructures of the samples were examined.

Fig. 9

Influence of particle size on the coercivity and DOT.

Figure 10 presents BSE images of the polished surfaces of the Nd–Fe–B powders with and without Ta sputtering after annealing, crushing and sieving under 75 µm. The primary particles and the secondary particles of Nd₂Fe₁₄B were observed in both samples. The secondary particle was already mentioned in Fig. 7 that liquefied Nd-rich phase bound primary particles during annealing. The size of secondary particles in the sample with Ta sputtering tended to be smaller than that in the sample without Ta sputtering, in other words, fraction of primary particles must be higher in Ta-sputtered powder. Therefore, the particle size distributions were evaluated from BSE images. The particle size distributions for the two samples are presented in Fig. 11. It was found that the particle size of the sample with sputtering was smaller than that of the sample without sputtering. This means that volume fraction of primary particles became high in powder with Ta sputtering. For the aggregated particles, it is hard to align by external magnetic field because the aggregated particles act like isotropic polycrystalline powder. On the contrary, primary particles can rotate and align by magnetic torque attribute to external magnetic field. Therefore, the increase in the ratio of magnetically alignable grains in the powders with Ta sputtering led to increased anisotropy shown in Fig. 9.

Fig. 10

BSE images of the polished surfaces of the crushed annealed Nd–Fe–B powders (a) without and (b) with Ta sputtering.

Fig. 11

Particle size distributions for the Nd–Fe–B powders with and without Ta sputtering.

It was concluded that Ta sputtering is an effective method for fabricating highly anisotropic powders.

4. Conclusion

In this study, the influence of Ta sputtering on the magnetic properties and microstructures of Nd–Fe–B powders was investigated with the aim of developing an improved method for fabricating anisotropic powders. XRF analysis revealed that the Ta content after sputtering was 5.6 mass%, and microstructural observation indicated that the Ta was located on the surfaces of Nd–Fe–B grains. The λ values of the Nd–Fe–B powders with and without Ta sputtering were 0.08 and 0.05, respectively, and that is similar and rather low. This was caused by necking of the grains in random directions relative to the easy axis of magnetization. The fineness of the annealed powder was increased by crushing, whereupon it was found that Ta sputtering increased the λ value from 0.55 to 0.73. Microstructural observations indicated that the increase in λ originated from an increase in the percentage of magnetically alignable grains. Thus, sputtering is an effective technique for coating Nd–Fe–B powders with Ta and improving their anisotropy.

Acknowledgements

The Nd–Fe–B powders were obtained by Mr. Une and Mr. Kubo from Intermetallics Co., Ltd.

This work was partially supported by the Elements Strategy Initiative Center for Magnetic Materials (ESICMM) under the outsourcing project of the Ministry of Education, Culture, Sports, Science and Technology, Japan.

REFERENCES

• 1)   M.  Nakamura,  M.  Matsuura,  N.  Tezuka,  S.  Sugimoto,  Y.  Une,  H.  Kubo and  M.  Sagawa: Appl. Phys. Lett. 103 (2013) 022404.
• 2)   M.  Nakamura,  M.  Matsuura,  N.  Tezuka,  S.  Sugimoto,  Y.  Une,  H.  Kubo and  M.  Sagawa: Mater. Trans. 55 (2014) 1582–1586.
• 3)   M.  Nakamura,  M.  Matsuura,  N.  Tezuka,  S.  Sugimoto,  Y.  Une,  H.  Kubo and  M.  Sagawa: Mater. Trans. 56 (2015) 129–134.
• 4)   A.  Iijima,  M.  Matsuura,  N.  Tezuka,  S.  Sugimoto,  Y.  Une,  H.  Kubo and  M.  Sagawa: Mater. Trans. 58 (2017) 825–828.
• 5)   M.  Sagawa,  S.  Fujimura,  N.  Togawa,  H.  Yamamoto and  Y.  Matsuura: J. Appl. Phys. 55 (1984) 2083.
• 6)   R.  Ramesh and  K.  Srikrishna: J. Appl. Phys. 64 (1988) 6406.
• 7)   R.  Ramesh,  G.  Thomas and  B.M.  Ma: J. Appl. Phys. 64 (1988) 6416.
• 8)  M. Sagawa and Y. Une: Proc. 20th Int. Workshop on Rare Earth Permanent Magnets and Their Applications, Crete, Greece, ed. by D. Niarchos, (Admore, Greece, 2008) p. 103.
• 9)   T.  Fukada,  M.  Matsuura,  R.  Goto,  N.  Tezuka,  S.  Sugimoto,  Y.  Une and  M.  Sagawa: Mater. Trans. 53 (2012) 1967–1971.
• 10)   H.  Kronmüller,  K.-D.  Durst and  M.  Sagawa: J. Magn. Magn. Mater. 74 (1988) 291–302.
• 11)   T.  Hattori,  H.  Ishihara,  S.  Sugimoto,  R.  Goto,  N.  Tezuka,  Y.  Une and  M.  Sagawa: Mater. Trans. 50 (2009) 2347–2350.
• 12)   K.  Yamamoto,  M.  Matsuura and  S.  Sugimoto: Metall. Mater. Trans. A 48 (2017) 3482–3489.
• 13)   J.M.  Cadogan and  J.M.D.  Coey: Appl. Phys. Lett. 48 (1986) 442.
• 14)  T. Takeshita and R. Nakayama: Proc. 10th Int. Workshop on REM and their Applications, (Kyoto, Japan, 1989) p. 551.
• 15)   P.J.  McGuiness,  X.J.  Zhang,  X.J.  Yin and  I.R.  Harris: J. Less Common Met. 158 (1990) 359.
• 16)   S.  Sugimoto,  H.  Nakamura,  K.  Kato,  D.  Book,  T.  Kagotani,  M.  Okada and  M.  Homma: J. Alloys Compd. 293–295 (1999) 862–867.
• 17)   T.  Horikawa,  M.  Matsuura,  S.  Sugimoto,  M.  Yamazaki and  C.  Mishima: IEEE Trans. Magn. 51 (2015) 2103904.
• 18)   C.  Mishima,  N.  Honda,  H.  Mitarai and  Y.  Honkura: IEEE Trans. Magn. 37 (2001) 2467.
• 19)   K.  Morimoto,  N.  Katayama,  H.  Akamine and  M.  Itakura: J. Magn. Magn. Mater. 324 (2012) 3723–3726.
• 20)   K.  Güth,  T.G.  Woodcock,  L.  Schultz and  O.  Gutfleisch: Acta Mater. 59 (2011) 2029–2034.
• 21)   J.  Han,  C.  Tong,  A.  Sun,  Y.  Xiao and  R.  Wang: J. Magn. Magn. Mater. 270 (2004) 136.
• 22)   S.  Sawatzki,  T.  Woodcock,  K.  Guth,  K.H.  Muller and  O.  Gutfleisch: J. Magn. Magn. Mater. 382 (2015) 219.