1 Introduction

Sm2Co17-type sintered magnets have been widely applied in aerospace, military technology and electric cars fields due to the large magnetocrystalline anisotropies, high Curie temperature and excellent thermal stability [1,2,3]. It is generally known that solid solution treatment is critical for the Sm2Co17-type magnets to obtain high performance. Through an appropriate solid solution treatment process, a single-phase state 1:7H is formed, and it evolves into a microstructure consisting of 2:17R cells surrounded by 1:5H cell boundaries coexisting with a Zr-rich lamellar phase after aging. These typical microstructures make the magnets obtain high intrinsic coercivity (Hcj) and magnetic field at knee-point (Hknee) (a magnetic field at which the corresponding magnetization is equal to 0.9 times remanence) [4,5,6,7,8,9]. Recently, Zhang et al. [10] reported that the solid solution temperature has a far-reaching impact on the phase structure and microstructure, which determines the magnetic properties of the magnets. Xu et al. [11] also studied the correlation of microstructure and magnetic properties in Sm2Co17 magnets solution-treated at various temperatures. It is found that an inhomogeneity of the cellular structure is discovered in the magnets solution-treated at a higher or lower temperature. This inhomogeneity results in weak pinning field, which is accountable for the deterioration of the magnetic properties. However, as well known, the coercivity of Sm2Co17-type sintered magnets mostly originates from the strength of domain wall pinning. The enrichment of Cu in the cell boundaries increases the magnetic field needed for the wall pinning [12,13,14]. This is the key factor for that the Sm2Co17-type magnets show a much higher Hcj and Hknee. Therefore, more information about Cu distribution at the cell boundaries needs to be studied to reveal the difference of magnetic properties influenced by the solid solution temperature. Moreover, as reported in Ref. [10], there mainly exist two kinds of phase regions with obvious different contrasts in the magnet solution-treated at a higher temperature, and the corresponding microstructures should be different in the two regions, which further influences the magnetic properties of the magnet. The difference of cell structure and Cu distribution at the cell boundaries in the two regions also needs to be studied in detail.

Moreover, it was reported that Fe content of the Sm2Co17-type magnets also affected the morphology and elemental distribution of their microstructure [15,16,17]. To prepare the Sm2Co17-type magnets with high maximum energy product ((BH)max), more iron is added (at the expense of cobalt) to increase the saturation magnetization of the Sm2(Co, Fe)17 matrix phase [18, 19]. An appropriate solution process should be employed to obtain the optimal magnetic performance of the magnets with higher Fe content. However, the correlation between Fe content and solid solution temperature needs to be investigated in detail to better control the magnetic properties.

In this paper, the optimum solid solution temperature range for the magnets with various Fe contents were investigated, and the influences of solid solution temperature on the microstructures were further studied. These results provide a clue to achieve better control of the microstructures to obtain high-performance Sm2Co17-type sintered magnets, especially for different Fe contents.

2 Experimental

The magnets with nominal compositions of Sm(CobalFe0.213Cu0.073Zr0.024)7.6 (Magnet A, with lower Fe content) and Sm(CobalFe0.262Cu0.073Zr0.024)7.6 (Magnet B, with higher Fe content) were prepared by a conventional powder metallurgical technology. The ingots prepared by induction melting were crashed and jet-milled into powders with a mean particle size of 3–5 μm. Then, the powders were aligned and pressed in a magnetic field of about 2 T and followed by an isostatic pressing process. The green compacts were sintered at 1478–1488 K for 1 h and then were solid solution treated at 1363–1473 K for 4–8 h and quenched. The subsequent isothermal aging of all the magnets were at 1103–1123 K for 5–10 h, followed by slowly cooling to 673 K, and finally quenched to room temperature directly.

A Hirst 14 pulsed field magnetometer (PFM) with a maximum magnetic field of 10 T was used for magnetic properties measurements, and the test specimens with dimensions of (10 ± 0.01) mm in diameter and (10 ± 0.01) mm in height were cut from the aged magnets. The crystal structure of the phases was measured by a Rigaku Smartlab X-ray diffraction (XRD) with a Cu Kα radiation. Phase transitions of the as-solution-treated magnets were measured by a STA 449 F3 differential scanning calorimetry (DSC). A Phenom ProX scanning electron microscopy (SEM) equipped with an energy-dispersive spectroscopy (EDS) was used to observe the microstructure of the magnets. The focus ion beam (FIB) technique was employed for the preparation of transmission electron microscope (TEM) specimen. Detailed cell structure and elemental distribution analysis of the aged magnets were carried out by a FEI Tecnai G2 F20 TEM equipped with EDS.

3 Results and discussion

3.1 Magnetic properties and optimal solid solution temperature range of Magnets A and B

Figure 1 shows the dependence of typical magnetic properties, remanence (Br), Hcj and Hknee on the solid solution temperature (Ts) for Magnets A and B. As shown in Fig. 1a, b, Br is almost constant with Ts increasing for the two types of magnets. For Magnet A, Hcj and Hknee increase from 2196.96 kA·m−1 and 971.12 kA·m−1, to 3048.68 kA·m−1 and 1711.4 kA·m−1, respectively, with Ts increasing from 1393 to 1413 K, and then, they remain constant with Ts further increasing to 1463 K, after that Hcj and Hknee drop sharply to 772.12 kA·m−1 and 573.12 kA·m−1 with Ts further increasing up to 1473 K, as shown in Fig. 1a. While, for Magnet B, Hcj and Hknee increase from 2499.44 kA m−1 and 979.08 kA m−1, to 2793.96 kA m−1 and 1592 kA m−1, respectively, with Ts increasing from 1383 to 1403 K, and then, they keep constant as Ts continues to rise, and when Ts is higher than 1453 K, Hcj and Hknee begin to drop sharply, as shown in Fig. 1b. It is found that 1413–1463 K is the optimum Ts range for Magnet A to obtain relatively high Hcj and Hknee, while it is 1403–1453 K for Magnet B. And the temperature range of optimal Ts for Magnet B shifts to relatively low temperature compared with that of Magnet A. More impressively, the optimal Ts range for magnets is both about 50 K though Fe content is different.

Fig. 1
figure 1

Dependence of Br, Hcj and Hknee of magnets with different Fe contents on solid solution temperature (Ts): a Sm(CobalFe0.213Cu0.073Zr0.024)7.6 (Magnet A) and b Sm(CobalFe0.262Cu0.073Zr0.024)7.6 (Magnet B)

Full size image

To reveal the difference of the magnetic properties of the magnets, the phase constitution should be checked first. The powders XRD patterns of the as-solution-treated and aged Magnets A and B are shown in Fig. 2. As a whole, the 1:7H phase with a TbCu7 structure is the main phase, which is found in all as-solution-treated magnets. As shown in Fig. 2b, c, g, h, the as-solution-treated Magnets A and B with proper Ts exhibit single phase structure with 1:7H phase, while 2:17H phase and 1:5H phase can be indexed for the as-solution-treated Magnets A and B with other Ts, as shown in Fig. 2a, d, and f, i, respectively. Except the as-solution-treated magnets, the powder XRD patterns of the aged Magnets A and B with various Ts were also measured. XRD patterns are almost the same, and the aged magnets consist of 2:17R and 1:5H phases, as shown in Fig. 2e, j. Combined with the magnetic properties and XRD results, it can be found that the magnets solution-treated at proper Ts obtain 1:7H single phase which is the key to obtaining optimal magnetic properties of the magnets.

Fig. 2
figure 2

Powder XRD patterns of as-solution-treated magnets with various solution-treated temperatures (Ts): Magnet A, a Ts = 1393 K, b Ts = 1433 K, c Ts = 1463 K, and d Ts = 1473 K; Magnet B, f Ts = 1383 K, g Ts = 1433 K, h Ts = 1453 K and i Ts = 1463 K; powder XRD patterns of aged e Magnet A and j Magnet B

Full size image

To further reveal the origin of the difference of optimal Ts range for Magnets A and B, their DSC curves were measured, as shown in Fig. 3. It can be found that there are two obvious endothermic peaks in both DSC curves. The relatively low peak corresponds to a phase transition process, while the high peak corresponds to the melting point of the main phase. For Magnet A, the onset and peak temperatures of phase transition are 1468.6 and 1479.0 K, respectively. And the corresponding temperatures are 1455.4 and 1460.7 K for Magnet B. It is obvious that both the onset and peak temperatures for Magnet B are lower than those of Magnet A. Based on the XRD results, the magnets solution-treated at proper Ts have 1:7H single phase, while the 2:17H and 1:5H phases appear when the magnets solution-treated at a higher Ts. It means that some 1:7H phase decomposes into 2:17H and 1:5H phases when the Ts is higher than a critical temperature. Therefore, the endothermic peaks corresponding to phase transition are found in the DSC curves. The largest optimal Ts = 1463 K is lower than the phase transition temperature of Magnet A (1468.6 K) and higher than that of Magnet B (1455.4 K). This is consistent with the results that, when solution-treated at 1463 K, good magnetic properties are obtained for Magnet A, while inferior magnetic properties for Magnet B.

Fig. 3
figure 3

DSC curves of as-solution-treated Magnets A and B

Full size image

3.2 Effects of solid solution temperature on microstructures of Magnets A and B

In order to reveal the microstructures, the magnets were characterized by SEM equipped with EDS. Typical backscattered electron (BSE) images of the as-solution-treated Magnets A and B with various Ts are shown in Fig. 4. The “white” phases are Sm2O3, which have been reported in many other literatures [10, 11, 20]. As shown in Fig. 4a, there are a few precipitated phases (Region I) in the matrix phase for Magnet A solution-treated at 1393 K, and the precipitated phases (Region V) are also found in Magnet B solution-treated at 1383 K shown in Fig. 4d. The precipitation phases in Magnet A are obviously more than that in Magnet B. When Ts increases to 1433 K, the precipitation phases decrease in number in Magnet A and disappear in Magnet B, as shown in Figs. 4b, e, respectively. Figure 4c, f shows that there exist two different regions: light gray region (Regions III and VI) and dark region (Regions IV and VII) in Magnet A solution-treated at 1473 K and Magnet B solution-treated at 1463 K, respectively.

Fig. 4
figure 4

Typical BSE images of as-solution-treated magnets with different Ts: Magnet A, a Ts = 1393 K, b Ts = 1433 K, and c Ts = 1473 K; Magnet B, d Ts = 1383 K, e Ts = 1433 K, and f Ts = 1463 K

Full size image

Table 1 shows the compositions determined by EDS of the phases that appear in the magnets. It can be found that the matrix phase is close to the nominal composition of the corresponding magnet. Regions I, II and V are Zr-rich phases which prefer to precipitate at a low solution temperature [10, 11], so the number of Zr-rich phases decreases significantly as the solution temperature goes up. As it is reported in Refs.[17, 21], Fe atoms have a strong preference for dumbbell Co–Co sites in 1:7H phase and promote the transformation from ordered 1:7H phase to partially ordered 2:17R phase, while the metallurgical function of Zr is to displace Fe atoms pairs by Zr-vacancy pairs from the dumbbell sites, and thus Fe atoms must occupy the planar sites. And the magnet with higher Fe content needs more Zr-vacancy pairs to displace Fe pairs and stabilize the 1:7H phase. Therefore, Magnet B has fewer Zr-rich precipitation phases. Regions III and VI have more Sm, more Cu and less Fe, which is close to the composition of 1:5H phase, while Regions IV and VII have more Fe, less Sm and less Cu, which is close to the composition of 2:17H phase. The presence of light gray phases and dark phases at a higher solution temperature is consistent with the XRD results. These microstructure characteristics do not change after aging process [22, 23].

Table 1 Sm, Co, Fe, Cu and Zr contents of matrix and Regions I-VII shown in Fig. 4 (at%)
Full size table

The effects of Ts on the microstructures of Magnets A and B are consistent, although the optimal solution temperature range is different. To get the correlation between the magnetic properties and microstructures, the detailed nanoscale microstructure analyses on aged Magnet B were taken by a TEM. Figure 5 shows bright-field (BF) images and elemental distribution with Ts = 1383 and 1433 K. As shown in Fig. 5a, b, a few cell boundaries marked by red arrows are incomplete in the magnet with Ts = 1383 K, while the cell structure in the magnet solution-treated at 1433 K is intact. The average cell sizes are 96.3 and 94.6 nm for the magnets with Ts = 1383 and 1433 K, respectively. As well known, the cell structure is closely correlated to the phase component of as-solid solution magnets. It is revealed from the XRD results that the as-solution-treated magnet obtains 1:7H single phase when solution-treated at 1433 K. Therefore, after aging treatment, the magnet obtains complete cell structure. While 2:17H phase exists in the as-solution-treated magnet with Ts = 1383 K, which is not prone to form the ideal cell structure after aging [11, 13, 15], so there is incomplete cell structure in the aged magnet with Ts = 1383 K. As shown in Fig. 5c, d, the lamellar phases are observed clearly in the aged Magnets B with Ts = 1383 and 1433 K, and the lamellar phase densities are calculated to be 0.036 and 0.044 nm−1, respectively. It is found that the density of lamellar phase of the magnet with Ts = 1433 K is higher than that of the magnet with Ts = 1383 K. It is due to the presence of Zr-rich precipitate phase in the magnet with Ts = 1383 K observed by SEM (Fig. 4d). As reported previously [4], Zr-rich phase may lead to the reduction of the lamellar phase. So the magnet solution-treated at 1383 K shows a relatively low density of lamellar phase.

Fig. 5
figure 5

Typical TEM bright-field images of Magnet B with different Ts: a Ts = 1383 K and b Ts = 1433 K with c-axis perpendicular to imaging plane; c Ts = 1383 K and d Ts = 1433 K with c-axis parallel to imaging plane; composition profiles of cells and cell boundaries marked with e white line in a and f white line in b

Full size image

Figure 5e, f shows elemental distribution along the cells and cell boundaries of the magnets marked by white lines in Fig. 5a, b, respectively. As a whole, TEM-EDS patterns show that both Co and Fe are lean in cell boundaries; however, Cu enriches in cell boundaries. For the cell boundaries of the magnets with Ts = 1383 and 1433 K, the average peak Cu concentrations are about 14.96 at% and 23.12 at%, respectively. As well known, the Zr-rich lamellar phase acts as the elemental diffusion paths during the aging process [24,25,26]. And the larger density of lamellar phase may promote the enrichment of Cu in the cell boundary. Therefore, the magnet (Ts = 1433 K) with larger density of lamellar phase has a higher average peak Cu concentration at the cell boundary. Higher Cu concentration in the cell boundary may result in an increased domain wall energy density difference between the 2:17R cell and 1:5H cell boundary phases, which makes the Sm2Co17-type sintered magnets present high coercivity [12,13,14, 27]. In the experiment, the magnet with Ts = 1433 K has complete cell structure and shows high Cu concentration in the cell boundary. As a result, high Hcj and Hknee are obtained in the magnet solution-treated at 1433 K. However, incomplete cell structure exists in the magnet with Ts = 1383 K, which is not beneficial for co-pinning the domain walls motion during demagnetization process, and the magnet has lower Cu concentration in the cell boundary compared to the magnet with Ts = 1433 K. Therefore, the magnet with Ts = 1383 K shows lower Hcj and Hknee.

Larger elemental component difference between Regions VI and VII of Magnet B with Ts = 1463 K does not change after aging. In situ FIB technique was employed to prepare TEM specimens to reveal the microstructures in the two different regions of the aged Magnet B. As shown in Fig. 6b, c, the cell is small in the light gray region, while the cell structure is coarse and irregular, and cell boundaries are incomplete in the dark region of the aged magnet. The average cell sizes of light gray and dark regions are about 58.2 and 153.45 nm, respectively. As reported previously [18, 28,29,30], the magnets with higher Sm content or lower Fe content have smaller cell size, while it is contrary for the magnets with higher Fe content or lower Sm content. As shown in Fig. 4f and listed in Table 1, Sm and Fe contents of the light gray region (Region VI) are, respectively, about 15.1 at% and 17 at% for the as-solution-treated magnet with Ts = 1463 K, and Sm and Fe contents of about 10.4 at% and 25.9 at% are obtained at the dark region (Region VII). As a result, the large difference of Sm and Fe contents between the two regions leads to the cell size variation after aging.

Fig. 6
figure 6

a Typical BSE image of aged Magnet B with Ts = 1463 K (c-axis perpendicular to imaging plane); TEM bright-field images of b light gray area and c dark area in a: b light gray area and c dark area; composition profiles of cells and cell boundaries marked with d white line in b and e white line in c

Full size image

Figure 6d, e shows elemental distribution of the cell and cell boundaries in the two regions of the aged magnets marked by white lines in Fig. 6b, c, respectively. In the light gray region of the aged magnet, the cell boundaries have a higher average peak Cu concentration of about 24.74 at%, but the gradient of Cu concentration is very low. And for the dark region, the average peak Cu concentration is only about 10.23 at% in the cell boundaries. As the light gray region has higher Cu content than the dark region, the higher peak Cu concentration is obtained in cell boundaries in light gray region. But higher Cu content also results in a wider distribution of Cu, so the cell boundaries in the light gray region have lower Cu concentration gradient. As reported in Ref. [14], the feature of low gradient of Cu concentration may result in the lower gradient magnetocrystalline anisotropy, which reduces the pinning field of the cell boundary phase. The low pinning field in the two regions and incomplete cell structure leads to inferior magnetic properties for the magnet with Ts = 1463 K.

4 Conclusion

The effects of solid solution temperature on the microstructures and magnetic properties of different Fe content magnets were investigated. It is found the optimum magnetic properties can be obtained in the solid solution temperature range of 1413–1463 K for the magnet with x = 0.213 (low Fe content magnet), while it is 1403–1453 K for the magnet with x = 0.263 (high Fe content magnet). The 1:7H single phase is found in the magnet solution-treated at proper temperature, while 2:17H and 1:5H phases occur at a higher or lower solution temperature. There are also more Zr-rich precipitation phases appearing in the magnet with a lower solid solution temperature. The 2:17H, 1:5H and Zr-rich precipitation phases in the as-solution-treated magnet are against the complete cell structure and high Cu concentration in the cell boundaries, leading to the reduced magnetic properties. Accordingly, the magnet solution-treated at proper temperature obtains 1:7H single phase, and after aging process, the magnet has complete cell structure and high Cu concentration (23.12 at%) in the cell boundary, showing high Hcj and Hknee.