1 Introduction

Sm(Co,Fe,Cu,Zr)z magnets are widely used in aerospace, defense, military and other fields due to their high Curie temperature, excellent thermal stability and corrosion resistance [1,2,3,4,5,6,7]. The 2:17-type SmCo sintered magnets have complex microstructure, consisting of Sm2(Co,Fe)17 rhombohedral phase, Sm(Co,Cu)5 hexagonal phase and Zr-rich lamellar phase with hexagonal phase [8,9,10,11]. The Sm2(Co,Fe)17 phase is responsible for high remanence (Br) [12,13,14,15,16,17,18]. The increase in the concentration gradient of Cu content between the Sm2Co17 phase and the SmCo5 phase enhances the domain wall pinning. In addition, Zr-rich lamellar phase provides a path for Cu atoms to diffuse into the cell boundary phase, leading to the increase in the coercivity (Hcj) [19,20,21,22].

The increase in Fe content of the magnet can effectively enhance its saturation magnetization and promote the growth of 2:17R cellular phase, resulting in the decrease in content of 1:5H cell boundary phase [23]. However, if Fe content is too high, the integrity of cell boundary phase would be destroyed [24,25,26]. Thus, the content of Sm should be increased to maintain the volume of 1:5H phase [27]. In addition, the magnetic properties of the 2:17-type SmCo magnets are extremely sensitive to the variation of the composition. Inhomogeneity of element distribution can lead to the damage of squareness (Sr) and the reversal process of magnetic domain [28]. Therefore, it is necessary to investigate the elemental distribution and the magnetic domain in the magnets, which will possibly reveal the relationship between microstructure and Sr of demagnetization curve.

In this paper, Sm(CobalFe0.15Cu0.07Zr0.03)7.8 magnets and Sm(CobalFe0.28Cu0.07Zr0.03)6.6 magnets were prepared. The effect of Cu distribution on micromorphology and magnetic domain was studied by in-situ observation through electron probe micro-analyzer (EPMA) and magneto-optical Kerr optical microscope (MOKE). This in-situ observation can provide more intuitive images, which can effectively reveal the relationship between microstructure and magnetic properties.

2 Experimental

Two kinds of magnetic alloy ingots with nominal compositions of Sm(CobalFe0.15Cu0.07Zr0.03)7.8 (named MA magnet) and Sm(CobalFe0.28Cu0.07Zr0.03)6.6 (named MB magnet) were prepared by the induction melting method, and their nominal compositions are listed in Table 1. The ingots were crashed and ball-milled into single crystal powders. The obtained powders were pressed under a magnetic field of 1600 kA·m−1, and then isostatically compacted under a pressure of 230 MPa for 120 s. The green bodies of MA magnet and MB magnet were sintered at 1488 K for 1 h, then the solid solution was performed at 1423–1443 K for 2.5 h. Final isothermal aging was performed at 1083 K for 40 h followed by slow cooling down to 693 K at a rate of 0.4 K·min−1, and then, the magnets were aged at 693 K for 10 h before cooling to room temperature.

Table 1 Nominal compositions of Sm(CobalFe0.15Cu0.07Zr0.03)7.8 and Sm(CobalFe0.28Cu0.07Zr0.03)6.6 (wt%)
Full size table

Magnetic properties of the magnets were tested using NIM-500C B-H hysteresis loop tracer with magnetic field up to 2400 kA·m−1. The magnets were observed using FEI NOVA nano2000 scanning electron microscope (SEM), and equipped with energy-dispersive spectroscopy (EDS). The Cu distribution in the magnets was analyzed by EPMA-1720 with a wavelength-dispersive X-ray (WDX) detector. The magnetic domain structure and microstructure were observed by BH-786IP-PK-type MOKE and transmission electron microscope (TEM), respectively. A JEM-2010-type TEM was used in this work.

3 Results and discussion

Figure 1 shows the magnetic properties of the MA and MB magnets at different solid solution temperatures. It is found that the magnetic properties change drastically with the solid solution temperature increasing, indicating that the solid solution temperature plays a key role in determining the magnetic properties. Hcj and Br of both MA and MB magnets significantly fluctuate with the temperature, and the remanence of MB magnet is higher than that of MA magnet. The maximum energy product ((BH)max) of 185.27 kJ·m−3 for MA magnet and 179.74 kJ·m−3 for MB magnet are achieved at different temperatures of 1438 K and 1433 K, respectively. The optimal (BH)max of MB magnet is lower than that of MA magnet, which can be contributed to its lower Sr of 80.20%.

Fig. 1
figure 1

Main magnetic properties of sintered Sm(CoFeCuZr)z magnets at different solid solution temperatures

Full size image

Figure 2 shows the demagnetization curves of MA and MB magnets at their optimal temperature, and corresponding magnetic properties are listed in Table 2. In Fig. 2, a small step appears in the demagnetization curve of MA magnet at the applied field of about − 560 kA·m−1, but the curve tends to be smooth when the field exceeds − 800 kA·m−1. For MB magnet, the demagnetization curve is smooth before − 400 kA·m−1, then drops sharply and becomes zero at about − 1200 kA·m−1. Although the Br of MB magnet is slightly higher than that of MA magnet, the Sr of demagnetization curve of MA magnet is obviously better than that of MB magnet, leading to a larger (BH)max of MA magnet. As can be seen from Table 1, the contents of the remaining elements are very similar except for the different contents of Sm and Fe in these two alloys. The intrinsic magnetic properties and the microstructure of the magnets are two main factors affecting the magnetic properties. It is generally believed that increasing Fe content is an important way to increase the (BH)max at room temperature. The magnetic properties of MB magnet fluctuate greatly with the change in solution temperature, indicating that the magnet with higher Fe content is more sensitive to the heat treatment.

Fig. 2
figure 2

Demagnetization curves of MA magnet and MB magnet

Full size image
Table 2 Magnetic properties of MA and MB magnets
Full size table

To investigate the relationship between the poor Sr and inhomogeneous microstructure, MA magnet was observed by in-situ SEM, EPMA and MOKE. Figure 3a shows the back-scattered electron (BSE) image of the MA magnet, in which the grain boundaries and the Sm2O3 phase can be observed. Figure 3b shows the distribution of Cu observed by EPMA. It is found that the distribution of Cu in grains is homogeneous. Besides, the Cu-lean phenomenon at the grain boundaries is also observed, which agrees with the previously reported results by Wang et al. [29]. The image of the thermal demagnetization state (TDS) of MA magnet is shown in Fig. 3c. The width of the maze-like magnetic domain is about 1.5 μm. And the magnetic domain width decreases with the increase in Cu content. Magnetic domain width in different regions of MA magnet is similar due to the uniform distribution of Cu [30]. Figure 3d–h shows the magnetic domain images of MA magnet at reversal magnetic field of 0, − 400, − 560, − 800 and − 1120 kA·m−1, respectively. In Fig. 2, the demagnetization curve of MA magnet presents a small step from 0 to − 560 kA·m−1. When the demagnetizing field of − 400 kA·m−1 was applied, the magnetic domain at grain boundary is not deflected completely yet and no obvious change in the magnetic domain of the inner grains is observed. When the reversal magnetic field applied increases to − 560 kA·m−1, the magnetic domain at grain boundary with lower Cu content deflects basically, and there is still no obvious magnetic domain deflection in the grain, which is the direct reason for the observed step in the demagnetization curve of MA magnet. When the reverse field increases to − 800 kA·m−1, there are scattered reverse magnetic domains inside the grains. With the enlargement of the reverse magnetic field to − 1120 kA·m−1, the distribution of the reverse domains becomes uniform in the grains. It is noticed that the Hcj of the sample is 2400 kA·m−1, but the maximum magnetic field for the MOKE observation is only 1160 kA·m−1. Therefore, this indicates that the uniformity of composition between grains has an influence on the magnetic domain structure and the Sr of demagnetization curve.

Fig. 3
figure 3

In-situ observations of MA magnet: a BSE image; b Cu-mapping image (Cu content shown with right scale bar); magnetic domain images at c thermal demagnetization state (TDS) and d remanence state (Br); magnetic domain images at reversal magnetic fields of e − 400 kA·m−1, f − 560 kA·m−1, g − 800 kA·m−1 and h − 1120 kA·m−1

Full size image

Figure 4a shows BSE image of MB magnet where the observation surfaces were perpendicular to c-axis of the magnets. Three distinct phases named as gray 2:17 main phase, white Sm2O3 phase and light gray 2:17 phase can be seen. Figure 4b shows corresponding EPMA result of Fig. 4a, and the phenomenon of Cu-lean at grain boundaries is clearly observed. The elemental distribution is obviously different from that of the MA magnet. In the grain marked by dotted line, the content of Cu in the light gray 2:17 phase is much higher than that near the grain boundary, while the Cu content in the middle region of grain is less. Figure 4c shows magnetic domain of MB magnet at TDS. Many studies have reported that concentration gradient of Cu between cell boundary phase and cellular phase can obviously influence the magnetic domain patterns in the 2:17-type SmCo sintered magnets [31]. Cu content is inversely proportional to magnetic domain thickness, which suggests that domain width is determined by microstructure and microchemistry. The domain width in grains ranges from 1 to 5 μm, as shown in Fig. 4b. Therefore, the magnetic domains with different widths may not reverse at the same reversal magnetic field, as shown in Fig. 4c. Figure 4d shows the magnetic domain pattern at the remanence state. Only, part of the magnetic domain at the grain boundaries is reversed firstly, resulting from its low Cu content. When the demagnetized field applied increases to − 400 kA·m−1, the domain deflection at grain boundary is basically completed, and the demagnetization curve is still full. With the increase in the reverse magnetic field up to − 560 kA·m−1, some coarser magnetic domains in regions with low Cu content begin to deflect. When the reverse magnetic field increases to − 800 kA·m−1, the cluster-like reverse domain begins to enlarge gradually, as shown in Fig. 4g. At the reverse magnetic field of − 1120 kA·m−1, the cluster-like reverse domain enlarges further and coarser domain regions almost completely reverse. The inversion process of the magnetic domain is consistent with its corresponding demagnetization curve. For the 2:17-type SmCo magnets, the thinner the domain width is, the stronger the pinning force will be, as shown in Fig. 4b. Therefore, the high Cu content can enhance the ability of magnetic domains against the reversal magnetic field. As a result, increasing Fe content can lead to the inhomogeneous Cu distribution in the inner grain, which may be the main reason for the poor Sr of the MB magnet.

Fig. 4
figure 4

In-situ observations of MB magnet: a BSE image; b Cu-mapping image (Cu content shown with right scale bar); magnetic domain images at c thermal demagnetization state (TDS) and d remanence state (Br); magnetic domain images at reversal magnetic fields of e − 400 kA·m−1, f − 560 kA·m−1, g − 800 kA·m−1 and h − 1120 kA·m−1

Full size image

The samples with c-axis perpendicular to the surface were selected for TEM analysis, and the results are shown in Fig. 5. Figure 5a and b shows microscopic morphology of different regions in MA magnet, and Fig. 5c and d shows that in MB magnet. It is established that the size of cell structures in MA magnet is about 70 nm, which agrees with the results observed by EPMA and MOKE. However, the microstructures of different regions in MB magnet are quite different. Complete cellular and cell boundary phase with an average size of 120 nm are observed in Fig. 5c, and incomplete cellular phases with an average size of 140 nm are obtained in Fig. 5d. The average size of the cellular phase in MB magnet is much larger than that in MA magnet, which can be attributed to their different Fe contents. It is noticed that the smaller average cell size in MA magnet is one of the reasons for the higher Hcj. Furthermore, the corresponding observations of EPMA and TEM indicate that the higher Sm and Cu contents can also prove the formation of more cell boundary phase, leading to the complete cellular phase as well as narrower magnetic domain, as shown in Figs. 4b and 5c. The effect of increasing Fe content is proved to be twofold by our further investigations. On the one hand, the increase in Fe content can promote the growth of the cellular phase. And on the other hand, it can cause the inhomogeneous elemental distribution in the inner grains, leading to the incomplete cell boundary phase and coarser magnetic domain in some regions with lower Cu content.

Fig. 5
figure 5

TEM bright-field images of cellular structure along [0001]2:17 zone axis in a, b MA magnet and c, d MB magnet

Full size image

4 Conclusion

The effect of elemental distribution on microstructure of Sm(CobalFe0.15Cu0.07Zr0.03)7.8 and Sm(CobalFe0.28Cu0.07Zr0.03)6.6 magnets was investigated. Experimental results show that magnets with higher Fe content are more sensitive to the change in solid solution temperature. The completeness of cell boundary phase and corresponding magnetic domain width are determined by the local Cu content. The EPMA and MOKE results indicate that the regions with higher Cu content have narrower magnetic domains, leading to the stronger resistance during demagnetization stage. Furthermore, the in-situ observation of EPMA and MOKE provides an effective way to discuss the relationship between microstructure and magnetic properties.