2021 Volume 62 Issue 12 Pages 1757-1763
An amorphous powder of starting compound (Sm0.8Zr0.2)1.05–1.10(Fe0.9Co0.1)11.3Ti0.7 was directly sintered in a carbon die under a pressure of 50 MPa using the spark plasma sintering (SPS) method. This is the first result showing the possibility of using this compound to prepare bulk magnets that have comparatively high magnetic properties. Typical sintered magnets with a density of 6.2 and 6.5 g cm−3, which are about 80% and 84% of the theoretical density, exhibited a coercivity of about 430 and 370 kA m−1 and (BH)max of 50.6 kJ m−3 (= 6.32 MGOe) and 57.2 kJ m−3 (= 7.15 MGOe) at room temperature, respectively. The crystal structures of the samples with high magnetic properties prepared by the SPS method were a mixture of the 1-9 and 1-12 phases, which is the same result as reported for powder samples prepared by annealing of the same starting compound.
Fig. 7 (a), (b) Hysteresis loops, (c), (d) enlarged demagnetization and (BH)max curves of SPS sample A and B, respectively, measured using a VSM with Hmax = 12 MA m−1 (HFLSM, IMR, Tohoku University). *Demagnetizing field corrected.
Magnetic materials with a ThMn12 (1-12)-type structure have attracted interest since the 1980s.1–3) In our previous studies, we prepared ThMn12-type compounds that were almost free of the α-(Fe,Co) phase and exhibited remarkable magnetic properties, with a saturation polarization (Js) of powders exceeding 1.5 T, a magnetic anisotropy field (Ha) exceeding 8 MA m−1, and a high Curie temperature (Tc) exceeding 900 K.4–12) They also have good thermal stability.12)
We initially prepared pulverized powder samples of these compounds and obtained coercivities (Hc) of approximately 150–210 kA m−1.9) We also attempted to prepare high-coercivity samples by annealing of powders prepared by rapid quenching (RQ) of these compounds. Using this method, we successfully obtained (Sm0.8Zr0.2)1.1-(Fe0.9Co0.1)11.3Ti0.7 powders with Hc values exceeding 400 kA m−1, as well as Js and Ha values similar to those of powder samples prepared by the strip-casting method under optimized annealing conditions.13)
A few previous studies have reported the possibility of preparing comparatively high-coercivity powders and bulk magnets with a ThMn12-type structure. For example, Schultz et al. reported a Sm–Fe–V magnet powder with a coercivity of Hc = 936 kA m−1 that was prepared by mechanical alloying and subsequent reaction heat treatment.14) Pinkerton et al.15) reported a bulk magnet prepared by hot-pressing which had a coercivity of 448 kA m−1. Also, Schönhöbel et al.16) recently reported Sm–Fe–V-based 1-12 bulk magnets that were prepared by a hot-deformation technique and had high magnetic properties of Js (3 T) = 0.63 T, Jr = 0.45 T, and Hc = 704 kA m−1.
Using ordinary metallurgical methods with an additive such as Sm–Cu alloy, high density (≈95 vol% of the theoretical density) bulk magnets showing a certain saturation polarization (Js ≈ 1.1 T) can be prepared, but there has been no report of preparing magnets with sufficient coercivity such as Hc > 240 kA m−1.17,18) The preparation of bulk magnets of ThMn12 compounds to realize sufficiently high magnetic properties, especially for our new compounds containing Co and Zr, is therefore an important and pressing target for us and also for other researchers in this field.
In this study, we aimed to prepare high-density bulk magnets using the spark plasma sintering (SPS) method. Since the observed microstructure of the samples prepared by annealing of amorphous (Sm0.8Zr0.2)1.1(Fe0.9Co0.1)11.3Ti0.7 powders showed very fine primary grain sizes as about 10–20 nm,13) the SPS method was considered suitable for use.19–21)
Amorphous starting powder of (Sm0.8Zr0.2)1.05–1.10(Fe0.8Co0.1)11.3Ti0.7 composition was directly sintered in a carbon die without any pre-heat treatment. This direct sintering method gives almost same coercivity as heat-treated powder samples.13)
The crystal structure of samples was confirmed by X-ray powder diffraction (XRD; SmartLab, Rigaku Corporation, Japan; CuKα), and magnetic measurements of the samples were performed using vibrating sample magnetometers (VSMs; maximum applied field [Hmax] of 4 MA m−1 [TOEI Industry, Japan] and Hmax of 12 MA m−1 [High Field Laboratory for Superconducting Materials (HFLSM), Institute for Materials Research (IMR), Tohoku University, Japan]) and a B-H loop tracer system (Hmax of 2.56 MA m−1, Tamakawa Co., Japan).
We used our improved law of approach for ferromagnetic saturation (LAFS) method to determine the magnetic properties of saturation polarization (Js) and magnetic anisotropy field (Ha) of the magnetically isotropic samples.7,22)
2.2 Sample preparationAn amorphous powder of starting compound (Sm0.8Zr0.2)1.05–1.10(Fe0.9Co0.1)11.3Ti0.7 was prepared via the RQ method by Santoku Co., Ltd., Japan with the rotation velocity of the cooling roller set to about 40 m s−1.
The composition of this compound was selected based on previous experiments aimed at optimizing magnetic properties in which a compound of similar composition prepared by the strip-casting method exhibited an approximate Js of 1.5 T, Ha of 7.0 MA m−1, and Tc exceeding 900 K.9) The composition of the starting sample in this study was tuned to be slightly rich in Sm owing to the expected evaporation of Sm during heat treatment.
The SPS apparatus used for sintering was the Dr. Sinter Lab Jr. Series 211Lx system (Fuji Electronic Industrial Co., Ltd., Japan), and the heating pattern is shown in Fig. 1. The electric current during sintering (maximum ca. 300–400 mA) was controlled to give a rate of temperature increase of about 50 K min−1. An Ar atmosphere ($P_{\text{O}_{2}} = 10^{ - 2}$ Pa) was used for sintering after evacuation for 30 min down to 103 Pa.
Heating pattern and SPS conditions.
The ribbons prepared by the RQ method were approximately 10 mm long, 1.3 mm wide, and 10 µm thick as shown in Fig. 2(a). The XRD pattern of the starting powder prepared by the RQ method was an amorphous pattern with no obvious diffraction peak.13)
(a) Morphology of ribbons prepared by the RQ method, (b) pulverized particles prepared using a mortar and pestle, and (c) the external appearance of a sintered sample prepared using the SPS method (surface covered with carbon film).
The ribbons were pulverized into powders of approximately 5–10 µm in diameter as shown in Fig. 2(b) by using a mortar and pestle, and were directly sintered in a carbon die (inside covered with carbon film) of about 10 mm in diameter under a pressure of 50 MPa. Figure 2(c) shows a typical sintered sample prepared by the SPS method. After sintering, the surface of the heat-treated sample was covered with a carbon film as shown in Fig. 2(c). The film was removed for measurement of physical (density [ρ]) and magnetic (saturation polarization [Js], magnetic anisotropy field [Ha], and coercivity [Hc]) properties of the samples. The thermal stability of the magnetic properties was also investigated for a sample with the film removed.
The coercivities of the SPS samples were measured by using the B-H loop tracer system (Hmax of 2.56 MA m−1; Tamakawa Co., Japan) because the sintered sample size was suitable for the tracer without any shape adjustment. Some typical samples were cut into cubes with side lengths of 2–3 mm and used to verify the magnetic properties through measurements using VSMs (Hmax of 4 MA m−1, [TOEI Industry, Japan] and 12 MA m−1 [HFLSM, IMR, Tohoku University]) that can apply higher magnetic fields than the B-H loop tracer system.
Figure 3 shows the microstructures of the SPS sample of the starting compound (Sm0.8Zr0.2)1.05–1.10(Fe0.9Co0.1)11.3Ti0.7 (sample B) observed using scanning electron microscopy (SEM). Figure 3(a) is a low-magnification SEM image showing that the sintered sample are composed of secondary grains of about 20 µm in thickness and about 100 µm × 100 µm in width that are overlayed together. During SPS sintering, a uniaxial pressure of 50 MPa was applied, and the original particle shape shown in Fig. 2(b) was clearly changed to the platelet shape shown in Fig. 3(a) as a result of heat treatment.
SEM images of the microstructure of a sintered sample (sample B) prepared using the SPS method. (a) Low-magnification image of secondary grains, (b) enlarged image of the interior of a secondary grain, (c) enlarged image of the boundary region between the fine (ca. 50–100 nm size) and coarse (ca. micrometer size) primary grains.
Each secondary grain contained very fine primary grains of nanometer size and certain numbers of growing grains of micrometer size that could be observed between the plate-like secondary grains, as shown in Fig. 3(b). Figure 3(c) shows the details of the boundary region between the fine and coarse grains. The fine grain region was composed of primary grains of about 50–100 nm in size. The XRD patterns of powders prepared by the above-mentioned plate-like grains were only mixed patterns of the 1-9 and 1-12 phases, and no other crystal structures could be detected.
2.3 Density, magnetic properties, and crystal structure of the sintered samplesFigure 4(a) shows the density (ρ expressed as the percentage of the full density of 7.7 g cm−3 of the compound in this study) of typical sintered samples from which the surface carbon film was removed. The density of each sample prepared under the sintering conditions shown in the figure was calculated from its measured weight and volume (diameter and height of the sintered sample) as shown in Fig. 2(c).
(a) Densities (ρ, percentage of the full density [7.7 g cm−3]) and (b) coercivities (Hc) of sintered magnets prepared using the SPS method at various sintering temperatures (horizontal axis) and sintering times (vertical axis). The values marked with asterisks (samples A and B) were measured using the 12 MA m−1 VSM (HFLSM, IMR, Tohoku University).
The density increased with increasing sintering temperature and time. For example, in samples treated for the same time of 10 min, the density increased with increasing temperature from 78.3% (1073 K) to about 91% (average of three samples treated at 1123 K), and in samples treated at the same temperature of 1073 K, the density increased with increasing time from 78.3% (10 min) to 85.5% (30 min).
Figure 4(b) shows Hc of samples measured using the B-H loop tracer with Hmax of 2.56 MA m−1. Hc was higher in samples treated for a shorter time, exhibiting values of 407 kA m−1 after treatment for 10 min at 1073 K, but 306 kA m−1 after treatment for 30 min at the same temperature. This apparently counterintuitive result is thought to be explained by the results of heat treatment of powders of the same compound as discussed in a previous paper, in which Hc exhibited comparatively high values in samples composed of the 1-9 and 1-12 mixed phases.13) The Hc behavior of sintered samples fabricated by the SPS method exactly matched that of the heat-treated powders.
For the samples prepared by the SPS method at 1073 K for 20 min with Hc = 382 kA m−1 (sample A) and at 1098 K for 10 min with Hc = 350 kA m−1 (sample B), Hc was measured using the B-H loop tracer (Hmax of 2.56 MA m−1). Hc was re-measured using the 12 MA m−1 VSM (HFLSM, IMR, Tohoku University), which found values of 432 kA m−1 (sample A) and 368 kA m−1 (sample B), respectively (noted with asterisks in Fig. 4(b)). When the maximum applied field was increased (2.56 MA m−1 to 12 MA m−1), the Hc values also increased by about 13% (sample A) and 4% (sample B).
Figure 5 shows the variation in Hc of the powder samples13) (measured by the VSM with Hmax = 4 MA m−1) and two SPS samples (samples A and B, measured by the VSM with Hmax = 12 MA m−1). The difference in the sintering temperature of the sample with optimum Hc prepared by the SPS method from the powder samples may have originated from differences in the position of the thermocouple during SPS. Large Hc of more than about 350 kA m−1 was observed in samples composed of the 1-9 and 1-12 mixed phases which were mainly treated at 1073–1173 K.
Dependence of coercivity on heat-treatment temperature for powder samples cited from Ref. 13) and the two SPS samples (A and B).
Figure 6 shows the XRD patterns of samples A and B prepared by the SPS method. The diffraction peaks of the SmFe11Ti (1-12) phase (ICDD No. 03-065-5363) and the SmFe9 (1-9) phase (ICDD No. 00-043-1311) are shown as vertical lines at the bottom of the figure.
XRD patterns of SPS samples A and B and the standard SmFe11Ti (ICDD No. 03-065-5363) and SmFe9 (ICDD No. 00-043-1311) samples.
The XRD patterns of sample A treated at 1073 K for 20 min and sample B treated at 1098 K for 10 min exhibit typical XRD patterns for samples composed of the 1-9 and 1-12 mixed phases. However, the typical peaks of the 1-12 phase, such as the (310), (002), and (202) peaks in the SmFe11Ti standard pattern, could be more clearly observed in the latter sample. The crystallinity of the phase, judging from separation of diffraction peaks of 1-12 phase, also seemed to be better in the sample B.
Figure 7(a) and (b) show the hysteresis loops (within ±2 MA m−1) of the SPS bulk samples measured using the VSM with Hmax = 12 MA m−1 (HFLSM, IMR, Tohoku University) for samples A and B, respectively, whose coercivity (Hc) and XRD patterns are respectively shown in Figs. 5 and 6. When we applied the LAFS method to the magnetization curves,7,22) the saturation polarizations (Js) of the powder samples were 1.27 T (sample A) and 1.30 T (sample B).
(a), (b) Hysteresis loops, (c), (d) enlarged demagnetization and (BH)max curves of SPS sample A and B, respectively, measured using a VSM with Hmax = 12 MA m−1 (HFLSM, IMR, Tohoku University). *Demagnetizing field corrected.
Figure 7(c) and (d) show enlarged demagnetization curves of the SPS bulk samples in the second quadrant of the demagnetization process in the loops shown in Fig. 7(a) and (b), respectively. Remanence (Jr) and (BH)max values were determined to be Jr = 0.62 T and (BH)max = 50.6 kJ m−3 (6.32 MGOe) for sample A and Jr = 0.67 T and (BH)max = 57.2 kJ m−3 (7.15 MGOe) for sample B. For the correction of demagnetizing fields in hysteresis curves in Fig. 7, we used the factor of N = 0.33, that of a spherical sample case.
Figure 8 shows the temperature dependence of the magnetic properties of the above two SPS samples (A and B) whose hysteresis loops are shown in Fig. 7. The magnetic properties were measured using the VSM with Hmax of 12 MA m−1 (HFLSM, IMR, Tohoku University) and determined using the LAFS method.7,22)
Temperature dependences of (a) Js, (b) Ha, and (c) Hc for samples A and B, and the standard Nd2Fe14B magnet. The variation in Js for a powder sample that has a similar composition to samples A and B is also shown in panel (a).
Figure 8(a) shows the temperature dependence of Js for bulk sample B sintered at 1098 K for 10 min, which exhibited Js = 1.30 T at room temperature (RT), and that of bulk sample A sintered at 1073 K for 10 min, which exhibited Js = 1.27 T at RT. The temperature dependence of Js for a Nd–Fe–B magnet with Js of about 1.62 T at RT and a powder of a similar composition for our samples A and B13) are shown in the figure as a standard and a comparative sample, respectively.
The Curie temperatures (TC) of samples A and B are estimated to be about 600–700 K from the powder data shown above for the similar composition13) and the measured temperature dependence of Js up to about 450 K. Thus, the Js values of the samples may become larger than those of the Nd–Fe–B magnet at higher temperatures, especially above about 583 K, which is the Curie temperature of the Nd–Fe–B magnet.
Figure 8(b) shows that the magnetic anisotropy field Ha was about 8.0 MA m−1 at RT for both samples A and B. When the temperature was increased to about 423 K, the Ha values decreased to about the same value of 4.0 MA m−1, which is similar to Ha of the standard Nd–Fe–B magnet at the same temperature.23–25) In addition, Ha should disappear in the case of a Nd–Fe–B magnet at its TC of about 583 K.
The coercivities (Hc) at RT shown in Fig. 8(c) are about 368 kA m−1 for sample B and 432 kA m−1 for sample A when measured by the VSM with Hmax = 12 MA m−1 (HFLSM, IMR, Tohoku University). Whereas Hc of standard Nd–Fe–B magnet decreased from 720 kA m−1 at RT to about 190 kA m−1 at 473 K,16–18) the two ThMn12-type magnets in this study exhibited Hc of about 150 kA m−1 at 450 K using the same VSM with Hmax = 12 MA m−1 (HFLSM, IMR, Tohoku University).
The nitrogenated R = Nd compounds with the ThMn12 structure of (Nd0.8Zr0.2)(Fe0.9Co0.1)11.5Ti0.5N1.2 and Nd(Fe0.8Co0.2)11.0MoN1.0 compositions are completely decomposed when heated to approximately 873 K and 1000 K, respectively.12) On the other hand, the (Sm0.8Zr0.2)(Fe0.75Co0.25)11.5Ti0.5 compound is stable up to about 1473 K12) when heated in an atmosphere with a low-oxygen partial pressure ($P_{O_{2}} < 10^{ - 2}$ Pa). The compound does not need to be nitrogenated in order to obtain high magnetic properties. Therefore, a compound with a similar composition of R = Sm as used in this study is suitable for preparing bulk sintered magnets.
In industrial applications, high-density (ρ) bulk magnets, such as ρ > 95 vol% of the theoretical density, are necessary. The energy density of such magnets is obviously better for application in industrial devices. Since the maximum density of bonded magnets, such as those prepared by compression molding by epoxy resin binder, is about 82 vol%, the sintering of high-quality powder is necessary to prepare such high-density magnets. There have been few reports of such high-density magnets with high magnetic properties comparable with those of the ThMn12-type magnets in this study (i.e., with Hc > 350 kA m−1).17)
Even though the coercivity was limited to about Hc = 432 kA m−1, the magnets in the present study exhibited higher saturation polarizations (Js = 1.27 T [sample A] and Js = 1.30 T [sample B]) determined using the LAFS method7,22) and measured remanences (Jr = 0.62 T [sample A] and Js = 0.67 T [sample B]). These higher values are thought to originate from differences in composition, namely, the Zr and Co element substitutions in this study, which results in the naturally better thermal stability and higher Js, Jr, and Tc of the magnets in this study.13,18)
In this study, we successfully prepared sintered magnets with high magnetic properties such as approximately Js ≥ 1.3 T, Ha > 7 MA m−1 both at RT, and an estimated Tc ≈ 600–700 K, and also comparatively high coercivity of Hc ≥ 400 kA m−1 at RT, using the SPS method. We used a carbon die in which the maximum applicable pressure was limited to less than 50 MPa. There is the possibility of using a hard alloy die, with which more than 100 MPa can be applied, which is expected to be an important future task for us.
Even though the experimental conditions were limited to comparatively low applied pressure using a carbon die, the densities of the samples reached relatively high values of >84% of the theoretical density when the SPS method was performed at higher temperatures of >1098 K for short-time treatments of 10–20 min, as shown in Fig. 4(a).
Higher temperatures of >1098 K and longer treatment times of >20 min in the SPS method should form the 1-12 phase, which is the result of transformation of the 1-9 phase to the 1-12 phase, in which Hc is comparatively small value of <250 kA m−1.13) A higher Hc of >400 kA m−1 appeared in samples composed of the 1-9 and 1-12 mixed phases, as shown in this study.13,18)
Generally, high density and high coercivity for the samples in this study are contradictory targets. High sintering temperatures of >1098 K or long heating times of >30 min are necessary for preparing high-density magnets (i.e., >90% of the theoretical density). However, it should be emphasized that the appearance of a high-coercivity mixture of the 1-9 and 1-12 phases is limited in the samples treated at intermediate sintering temperatures or comparatively long heating times, such as in the sample sintered at 1073 K for 30 min (Fig. 4(a)). A more general survey of sintering conditions in the SPS method for optimizing density and coercivity is necessary in future work.
In this study, we successfully prepared sintered (Sm0.8Zr0.2)1.05–1.10(Fe0.9Co0.1)11.3Ti0.7 samples with high density (>84% of the theoretical density), high magnetic properties (Js ≥ 1.3 T, Ha > 7 MA m−1 at RT, and Tc ≈ 600–700 K), and comparatively high coercivity (Hc > 400 kA m−1) using the SPS method. The (BH)max values of the samples reached 50.6 and 57.2 kJ m−3 for samples sintered at 1073 K for 20 min and 1098 K for 10 min, respectively.
There is a possibility of preparing higher-density samples with >90% of the theoretical density of 7.7 g cm−3, which means a higher (BH)max value can be expected even for a similar Hc of about 400 kA m−1. This can be achieved by optimizing the SPS conditions using a suitable die material.
The use of the SPS method could be a solution for preparing high-density ThMn12-type magnets. Therefore, the coercivity mechanism, which exhibits an optimum value of 400 kA m−1 when the samples were composed of the 1-9 and 1-12 mixed phases, is the most important research target for the next stage in the development of the magnetic properties of ThMn12-type sintered magnets.
Part of this study is based on results obtained from the pioneering program “Development of magnetic material technology for high-efficiency motors” (MagHEM), grant number JPNP14015, commissioned by the New Energy and Industrial Technology Development Organization (NEDO). The authors also thank Mr. H. Komura and Mr. K. Hanashima of Minebea-Mitsumi Co., Ltd., Japan, for helpful discussion about the SPS method. We also thank Mr. H. Ari-izumi and Y. Hayashi of TOEI Industry Co., Ltd., Japan, for their helpful support in VSM measurements.
