1 Introduction

The application of permanent magnets starts as early as ancient China, where people made compass with the use of magnetite as a raw material. When it comes to the modern age, which is an age of electricity, permanent magnet is taking an increasingly important role in the contemporary society. In our daily life for example, majority household appliances in a family contain certain amount of permanent magnets. Moreover, electric motor, which is a crucial part in a vehicle, also includes large volume of permanent magnets. Another example lies in the field of biomedical. With the permanent magnets as the working part, the Magnetic Resonance Imaging (MRI) machine will diagnose various diseases without any harm to the human body. Therefore, the demands for permanent magnets with larger maximum energy product (BH)max, lighter weight, smaller volume, and higher working temperature are essential for the development of technologies nowadays.

The first influential finding in permanent magnets was Alnico 3(Al–Ni–Co–Fe alloy) by Honda and Mishima in early 1930s. The Alnico 3 possesses energy product about 8 kJ·m−3, which is superior than any other permanent magnet used before [1, 2]. Then, the discovery of ferrite hexagonal ferrites ((Ba/Sr)Fe12O19) increased the energy product into 24 kJ·m−3 [3]. However, these ferrite magnets hold small magnetization and low Curie temperatures. Following that, the most crucial step forward took place in 1960s, where the rare-earth-based permanent magnets were first realized as the form of RCo5 intermetallics [4]. And this significantly enlarged the energy product of permanent magnets to 240 kJ·m−3, which is ten times larger than that of ferrite magnets. Afterward, in 1980s, the discovery of Nd2Fe14B magnets further defined rare-earth-based magnets the most important permanent magnets [5–7]. The energy product of Nd2Fe14B reaches as high as 392 kJ·m−3, and large content of Fe makes the magnets comparatively cheaper than the previous SmCo5 magnets. Nevertheless, the lower Curie temperature (~573 K) is a major problem of Nd2Fe14B. In order to look for a new generation of permanent magnets which hold even higher energy product and Curie temperature, in 1990s, researchers proposed the model for the hard/soft phase exchange-coupled magnets [8]. According to the simulation result, the theoretical energy product for Nd2Fe14B/Fe exchange magnet is 960 kJ·m−3. However, results which recently reported still have big gap between the simulation ones.

Having realized its importance, scientist developed various methods to prepare rare-earth-based magnets. And among them, chemical route is considered one of the major strategies. Especially with the trends of miniaturization and integration in current industrial world, chemical synthesis provides an easy way to tune size and composition of magnets so that the as-synthesized permanent magnets will meet the requirement of application. In this review, we will describe the progress in chemical synthesis of various nanostructured rare-earth-based permanent magnets as well as their magnetic properties. And we will also provide a view for the bottlenecks as well as futures of the chemical method.

2 Chemical synthesis of Sm–Co permanent magnets

As mentioned above, Sm–Co magnets were the first investigated rare-earth-based permanent magnets. The consisting of only two elements in Sm–Co magnets make them comparatively easier to prepare than three-elements magnets such as Sm–Fe–N or Nd–Fe–B. Therefore, the chemical synthetic strategies of Sm–Co magnets have been intensively investigated over years aiming to find an approach to make Sm–Co magnets with high energy product, tunable size, and excellent stability [9–24]. In the early days, researchers attempted to synthesize Sm–Co nanoparticles (NPs) directly from wet chemical method since it is the most accepted way to control the morphology of NPs [9]. However, people soon found that the electronegativity and reactivity of Sm make it extremely difficult. Thus, a high temperature reductive annealing with Ca as reducing agent is introduced in the chemical synthesis of rare-earth-based permanent magnets [13, 20].

Hou et al. [13] synthesized SmCo5 magnets with nanoscale domains from core/shell Co/Sm2O3 NPs. According to their route, they initially prepared 8 nm-sized Co NPs via decomposition of Co2(CO)8 in mixture of oleic acid, dioctylamine, and tetralin under ambient atmosphere (Fig. 1a). The obtained 8 nm Co NPs were dispersed in hexne and subsequently injected in mixture of oleylamine and oleic acid with Sm(acac)3 dissolved in. The system was then heated to 250 °C to trigger decomposition of Sm(acac)3 on the surface of Co NPs and core/shell Co/Sm2O3 NPs were generated (Fig. 1b). Later, the SmCo5 magnets were fabricated by reductive annealing of as-synthesized Co/Sm2O3 NPs under 900 °C in Ar/H2 with metallic Ca as reducing agent and KCl as solvent. According to the HRTEM image, the acquired SmCo5 magnets are assembled by nanoscale domains with various orientations with a lattice space of 0.293 nm (Fig. 1c). The magnets exhibit coercivity of 0.8 T and remnant magnetic moment of 40–50 A·m2·kg−1 under room temperature. Moreover, the researchers showed that the same strategy can be applied in the preparation of Sm2Co17 by tune the Sm/Co ratio during the synthesis of Co/Sm2O3 NPs (Fig. 1d).

Fig. 1
figure 1

a TEM images of Co NPs with size of 8 nm, b TEM image of Co/Sm2O3 NPs, c HRTEM of as-synthesized SmCo5 magnet with assembles of nanocrystals as indicated by dashlines, and d hysteresis loops of SmCo5 under room temperature. Reproduced with permission from Ref. [13]. Copyright 2007 John Wiley & Sons

Full size image

Following the strategy developed by Hou et al., Zhang et al. [20, 25] further synthesized 6 nm monodispersed SmCo5 NPs. In their method, 7 nm monodispersed Sm–Co–O embedded in CaO matrix was first prepared by co-decomposition of Co(ac)2 and Sm(ac)3 in n-hexadecyltrimethylammonium hydroxide (HTMA–OH) with the presence of Ca(ac)2. And the decomposition of Ca(ac)2 led to the formation of CaO matrix which embraced Sm–Co–O and hence would inhibit the diffusion of SmCo5 under annealing temperature (Fig. 2a). Afterward, the Sm–Co–O@CaO composite was annealed under 960 °C for 2 h in Ar/H2 with Ca as reducing agent and KCl as solvent. According to XRD patterns and TEM images, the resultant was SmCo5 NPs embedded in CaO matrix. After removal of CaO matrix by washing in ethanol and deionized (DI) water, the as-synthesized SmCo5 NPs showed narrow size distribution and diameter of 6 nm, which was quite similar to that of Sm–Co–O NPs (Fig. 2c–e). This SmCo5 NPs exhibit reduced coercivity of 0.72 T and remnant magnetic moment of 35 A·m2·kg−1. In addition to that, the researchers also employed this method to synthesize Sm2Co17 NPs, and the coercivity as well as remnant magnetization were 0.58 T and 45 A·m2·kg−1, respectively. So far, this method is the most controllable chemical route in terms of composition and morphology. However, the as-synthesized Sm–Co NPs are so reactive that they will be rapidly oxidized and thus lose their magnetic properties if exposed to air.

Fig. 2
figure 2

a TEM image of the 7 nm SmCo3.6–O NPs in the CaO matrix, b TEM images for SmCo5 NPs in CaO matrix after annealing, c TEM images for the 6 nm SmCo5 NPs after removal of CaO matrix, d HRTEM image of the SmCo5 NPs, and e Schemes for the synthetic route of Sm–Co NPs. Reproduced with permission from Ref. [25]. Copyright 2013 IOP Publishing Ltd

Full size image

Surfactant assisted ball milling (SABM) is another popular method to synthesize nanostructured Sm–Co magnets [11, 14, 18, 19, 26]. According to the reports, oleic acid and oleylamine are the most commonly used surfactants in the high energy ball milling of Sm–Co magnets [17, 18, 25]. Surfactant will prevent the fragmentized NPs from re-welding. In addition to that, surfactant will also help the dispersion of NPs so that they will not aggregate. Poudyal et al. [18] synthesized SmCo x (x = 3.5, 4.0, 5.0, 6.0, 8.5, and 10.0) NPs via SABM technique. In their process, they prepared SmCo x magnetic powders via arc melting and the following grinding process. The as-synthesized SmCo x powders were transferred into a mixture of heptane, oleic acid, and oleylamine for ball milling. After milling for 20 h, the products were taken out under ambient atmosphere and went through a size selection process. (That is to separate NPs of various sizes by tune the sedimentary time of the mixture.) According to the TEM images, 6 nm, 20 nm, and submicron Sm2Co17 particles were received by collecting the deposition after different sedimentary time (Fig. 3a–c). The XRD diffraction peaks of 20 nm and submicron-sized Sm2Co17 NPs are broadened which indicated the reduction of the grain size in both samples. However, the XRD pattern of 6 nm NPs is one broad peak which implied the amorphous character of the 6 nm Sm2Co17 NPs (Fig. 3d). In order to investigate the magnetic properties of the SmCo x NPs made from SABM, researchers tuned the composition of SmCo x and made a series of SmCo x NPs (x = 3.5, 4.0, 5.0, 6.0, 8.5, and 10.0). With various Sm/Co ratio, researchers noticed that the coercivity of the as-synthesized SmCo x NPs vary from 0.05 to 0.3 T (Fig. 3e).

Fig. 3
figure 3

a TEM images of 6 nm Sm2Co17 NPs, b TEM images of 20 nm Sm2Co17 NPs, c SEM image of submicron Sm2Co17 particles, d XRD patterns of the as-synthesized Sm2Co17 NPs and raw material, and e demagnetization curves of the Sm–Co NPs with various composition. The inset illustrating the relationship between composition and coercivity of the SmCo NPs. Reproduced with permission from Ref. [18]. Copyright 2010 American Institute of Physics

Full size image

Yue et al. [24] also utilized SABM method and a subsequently size selection process to prepare SmCo5 NPs and nanoflakes with high coercivity and narrow size distributions. The SmCo5 NPs have average p sizes of 9.8 and 47.5 nm, and they exhibit coercivity values of 6.8 × 104 and 7.3 × 105 A·m−1 under room temperature. Moreover, the SmCo5 nanoflakes have diameter about 1.4 μm and average thickness of 75 nm. The researchers found that the SmCo5 nanoflakes present strong magnetic anisotropy. The coercivity along easy-axis is 5.5 × 105 A·m−1, while the coercivity along hard-axis is 1.6 × 106 A·m−1.

3 Chemical synthesis of Nd–Fe–B permanent magnets

As mentioned in the synthesis of Sm–Co NPs, researchers originally tried to use wet chemical route to directly generate Nd2Fe14B NPs with controllable size and pure phase [25, 27]. However, all the attempts failed due to the high negative reduction potential of Nd. Moreover, the bottom-up strategies which were described in preparation of Sm–Co NPs are unsuitable ascribe to the nature that Nd2Fe14B consisted of three elements rather than two. Therefore, sol–gel and SABM are the widely employed method to synthesize Nd2Fe14B NPs [11, 16, 19, 23, 28–32].

Deheri et al. [30] synthesized Nd2Fe14B via sol–gel-based chemical methods. In their process, NdCl3, FeCl3, and boric acid were used as the source of Nd, Fe, and B, respectively. In addition to that, citric acid and ethylene glycol were employed as crosslinking agent and solvent. The Nd–Fe–B–O was prepared by modified Pechini type sol–gel method. Then, Nd2Fe14B NPs were obtained by annealing the Nd–Fe–B–O powders under 800 °C in N2 atmosphere with CaH2 as reduction agent. According to the XRD characterization, the major phase of the resultant is Nd2Fe14B, while some impurities such as Nd2Fe14BH4.7(27.68 %) and α-Fe (2.43 %) were also detected (Fig. 4a). The TEM image suggested that the Nd2Fe14B NPs are highly aggregate with average size of 50 nm, (Fig. 4b) and the hysteresis loop characterized by VSM under room temperature showed that the obtained Nd2Fe14B NPs have saturation magnetization of 102.3 A·m2·kg−1 and decreased coercivity of 0.39 T (Fig. 4c). The big gap of the magnetic property between as-synthesized Nd2Fe14B NPs and bulk Nd2Fe14B magnet is probably caused by the impurity as well as minimization of the grain size. In a subsequent work from Deheri et al. [31], they further looked into the mechanism of the transformation from Nd–Fe–B-oxide to Nd2Fe14B NPs. And they provided the following conclusions: (1) Reduction–diffusion consisted of three steps. Initially, Fe2O3 and B2O3 will be reduced to Fe and B at 300 °C. Then, Nd2O3 and NdFeO3 will be reduced and hydrogenated to NdH2 and Fe at 620°. Finally, Nd2Fe14B phase will be formed at 692°. (2) Two parallel reactions were competing during the formation of Nd2Fe14B NPs. The first one was the direct combination of NdH2, Fe, and B to form Nd2Fe14B. The second one was the combination of NdH2 and Fe to form Nd2Fe17 followed by the reaction between Nd2Fe17 and B to form Nd2Fe14B.

Fig. 4
figure 4

a XRD pattern of as-synthesized Nd2Fe14B NPs (Inset showing that peaks corresponding to Nd2Fe14BH4.7 phase shifting to a lower 2θ compared with the XRD pattern of Nd2Fe14B.), b TEM image of as-synthesized Nd2Fe14B NPs (Inset being the SADP of the NPs), and c hysteresis loop of as-synthesized Nd2Fe14B NPs. Reproduced with permission from Ref. [30]. Copyright 2010 American Chemical Society

Full size image

Due to the difficulty of preparing a single-phased Nd–Fe–B magnet by bottom-up strategy, researchers considered SABM as a more practical way to prepare nano-sized Nd–Fe–B particles [11, 19, 23, 28, 29, 32]. Akdogan et al. [29] synthesized Nd2Fe14B alloy from arc-melting method as raw material. The Nd2Fe14B powders were pre-milled in heptane for 4 h. Then, the slurry was further milled in the mixture of heptane and oleic acid for another 6 h. Finally, the obtained Nd2Fe14B NPs were field-aligned. According to the XRD study, although the SABM leads to the broadening of the XRD peaks, the as-synthesized NPs exhibit pure phased tetragonal Nd2Fe14B phase (Fig. 5a (1) and (2)). Moreover, XRD patterns of the field-aligned samples indicated [001] out-of-plane texture in Nd2Fe14B NPs (Fig. 5a (3)). TEM image of the upper part of the slurry suggested that the generated Nd2Fe14B NPs have square morphology with average size of 12 nm (Fig. 5b). The hysteresis loop of the square Nd2Fe14B NPs showed that the coercivity of the sample is 0.18 T under room temperature and 0.4 T under 40 K (Fig. 5c).

Fig. 5
figure 5

a XRD patterns of Nd2Fe14B: (1) milled for 4 h in heptane, (2) milled for extra 6 h with OA, (3) after field-aligned for the 6 h milling; b TEM image of the upper part of the slurry; c hysteresis loop of square Nd2Fe14B NPs at 40 K and room temperature (Inset). Reproduced with permission from Ref. [29]. Copyright 2010 IOP Publishing Ltd

Full size image

Yue et al. [32] synthesized Nd2Fe14B nanoflakes through SABM method. According to their results, the nanoflakes have average thickness of several tens nm and average diameter of 500–1,000 nm, and this shape anisotropy leads to a strong c-axis texture in the as-synthesized Nd2Fe14B nanoflakes.

4 Chemical synthesis of hard/soft exchange-coupled permanent magnets

Since the first model of exchange-coupling effect was introduced by Coey et al., there have been intense interests on this effect which only takes place at interphase between hard and soft magnet in the range of nanosize. This exchange-coupled magnet can be designed according to the required properties by selecting different hard and soft phases and by tuning the phase ratio. Therefore, nanocomposite magnets have suggested new chances for the development of new generations of permanent magnets. Physical method, especially physical vapor deposition (PVD), is a more adopted method that scientists used to research the exchange-coupling effect. Through PVD method, one can easily change the phase composition as well as tune the phase ratio [33, 34]. However, PVD route cannot prepare magnets with high-yield. Therefore, based on the early results from PVD experiments as well as theoretical simulations, researchers have been trying to synthesize exchange-coupled magnets through chemical routes over decades [35–53]. However, it was never easy as the case in the PVD method.

Hou et al. [41] employed a wet chemical process and a following Ca reduction to synthesize SmCo5/Fe exchange-coupled nanocomposite. In their strategy, Fe3O4/SmCo-hydroxide composite was first generated from precipitation of Sm and Co in monodispersed Fe3O4 NPs solutions. The Fe3O4 NPs were embedded in SmCo-hydroxide matrix, and this matrix could prevent Fe NPs from aggregation in the following reductive process. Afterward, the composites were annealed at 900 °C temperature for 1 h with Ca as reducing agent and KCl as dispersion medium. According to the HRTEM image, the average grain sizes of SmCo5 and α-Fe are 29 nm and 8 nm, respectively (Fig. 6a). The researchers prepared a series of SmCo5/Fe x (x = 0–2.9) samples, and found that both saturated magnetization and coercivity are varied with Fe content (Fig. 6b). The hysteresis loop of representative SmCo5/Fe1.5 showed enhanced saturated magnetization and single-phase behavior which implied the incorporation of soft phase α-Fe into hard phase SmCo5 (Fig. 6c). The δ–M plot of the composition is initially positive, suggesting the existence of exchange coupling, but it soon drop to negative after the reversal, indicating magnetostatic interactions in the composite due to the presence of soft magnetic Fe phase (Fig. 6d).

Fig. 6
figure 6

a HRTEM of SmCo5/Fe nanocomposite, b Change of coercivity and magnetic moment with various Fe ration in the SmCo5/Fe x , c Hysteresis loop of SmCo5 and SmCo5/Fe1.5 magnets, and d δM–H plot of SmCo5/Fe1.5 magnet. Reproduced with permission from Ref. [41]. Copyright 2007 American Institute of Physics

Full size image

5 Conclusion and perspectives

In summary, we have presented various chemical synthetic strategies to prepare nanostructured rare-earth-based permanent magnets. The chemical routes not only provide convenient approaches to prepare rare-earth-based permanent magnetic NPs but also offer an opportunity to manipulate the phase and morphology of the NPs to meet the requirement of current applications. However, there are still big challenges in the future development of chemical synthesis of nanostructured rare-earth-based permanent magnets. For example, the defects and impurities are often discovered in the NPs, and the particle morphology is relatively hard to control compared to non-rare-earth-based magnetic NPs. The as-synthesized NPs are too reactive to be practically stabilized. In the future, the proper use of other middle or heavy weight rare-earth elements might reduce the cost or enhance the magnetic properties of Sm- and Nd-based nanostructured magnets. In the case of bulk magnets, although single-phased Pr-, Te-, or Dy-based magnets exhibit either low anisotropy or small moment, the incorporation of those elements in Sm- or Nd-based magnets can dramatically increase the energy product of magnets [54–59]. In addition to that, the synthesis of rare-earth-based magnetic NPs with the composition other than RCo x or RFeB was rarely reported. For example, Sm–Fe–N or Sm–Fe–C magnets also possess favorable magnetic property, while their nanostructures were seldom suggested. Moreover, the chemical synthesis of exchange-coupled magnets still remains a big challenge. The current strategy is far from the objective of thoroughly controlling the composition and ratio in both hard and soft phase. Generally, the chemical method has great importance in the development of rare-earth-based magnetic NPs, and it needs to be further investigated in the coming years.