Correlation between Microstructure and Magnetism in Ball-Milled SmCo5/α-Fe (5%wt. α-Fe) Nanocomposite Magnets

Magnetic nanocomposites SmCo5/α-Fe were synthesized mechanically by high-energy ball milling (HEBM) from SmCo5 and 5%wt. of α-Fe powders. The X-ray diffraction analysis reveals the hexagonal 1:5 phase as the main one accompanied by the cubic α-Fe phase and 2:17 rhombohedral as the secondary phase. The content of each detected phase is modified throughout the synthesis duration. A significant decrease in crystallite size with a simultaneous increase in lattice straining is observed. A simultaneous gradual reduction in particle size is noted from the microstructural analysis. Magnetic properties reveal non-linear modification of magnetic parameters associated with the strength of the exchange coupling induced by various duration times of mechanical synthesis. The highest value of the maximum energy product (BH)max at room temperature is estimated for samples milled for 1 and 6 h. The intermediate mixed-valence state of Sm ions is confirmed by electronic structure analysis. An increase in the Co magnetic moment versus the milling time is evidenced based on the performed fitting of the Co3s core level lines.

So far, SmCo 5 -based nanocomposites have been synthesized in an argon atmosphere or by the use of the wet SA-HEBM procedure, usually applied for up to 10 h, and also vacuum at the sintering temperature 700-740 • C and pressure 10-10.5 kN for 5 min) due to stronger exchange coupling between the hard and soft magnetic phases than evidenced for the 10 wt.% α-Fe-containing sample. Nonetheless, in the synthesized samples, besides the main 1:5 phase, other phases of 2:17, 1:2 and Fe(Co) having grains in the range of 50-100 nm were found [54].
PPC was found by Zheng et al. [50] as a promising second step in the production of bulk composite magnets made of SmCo 5 /α-Fe (20 wt.% α-Fe) with high density at reduced temperature. The first step in such synthesis was low-energy ball milling and aligning the as-milled specimen in the magnetic field of 1.2 T for 5 min. As it was shown, the magnetic coupling between hard and soft phases is maintained after PPC despite the evident soft phase grain growth seemingly beyond the theoretically predicted limits. The authors also demonstrated the formation of the 2:17 phase due to a limitation in the Fe solubility in the SmCo 5 matrix. The microstructure of the synthesized nanocomposite was detected as uniformly CoFe-coated hard Sm-Co particles.
Therefore, as it was demonstrated, mechanically milled specimens may be successfully considered as a starting material in the fabrication of anisotropic nanocomposite magnets with enhanced magnetism. In addition, the variety of technical parameters in SmCo 5 /α-Fe nanocomposite synthesis leads to the ability to control the magnetic parameters by tuning the exchange coupling between soft and hard phases. One of the adjusted parameters is definitely the amount of the α-Fe phase. In our paper, we focused on its low content (5%wt.) as it has been the least analyzed so far. Our study aims to bring out a detailed investigation based on the synthesis of SmCo 5 /α-Fe nanocomposite nanopowders via modified HEBM. We analyze the microstructure and magnetism in as-milled powders as a function of the applied milling time. In addition, we present the modification within the electronic structure versus milling, which for the SmCo 5 /α-Fe-based nanocomposites is undoubtedly a novelty element.

Materials and Methods
The SmCo 5 compound was prepared by arc melting from high-purity elements under argon atmosphere. The ingot was melted several times in order to obtain homogeneity. Afterwards, the as-cast sample was wrapped in tantalum foil, placed in quartz tubes and annealed at 24 h at 1100 • C. The crystal structure was checked by means of X-ray powder diffraction (XRD) using an Empyrean PANalytical (Malvern Instruments, Malvern, UK) diffractometer equipped with a Cu X-ray source (Kα 1 of 1.54056 Å). In order to obtain the SmCo 5 nanopowder, first, the bulk compound was crushed and pre-milled with a standard agate mortar for about 10 min. Afterwards, the mixture of SmCo 5 nanopowder and 5%wt of α-Fe was ground for 0.5, 1, 1.5, 2, 4, 6, 8 and 10 h with a 30 Hz frequency (1800 rpm) using a Mixer Mill 400 (Retsch, Haan, Germany). As in the case of previously synthesized intermetallics [16,17,[34][35][36][37][38][39][40][41], an innovative wet milling method was carried out in dimethylformamide (DMF) using Eppendorf vials and 2 mm ZrO 2 balls with a balls to powder ratio of about 10:1. The structure and morphology of the as-milled specimens were investigated by high-resolution scanning electron microscopy (HR-SEM) on a JEOL JSM-7100F FEG (field emission gun) operated at 15 kV in secondary electron (SE) mode and X-ray powder diffraction (XRD) (Empyrean PANalytical, Cu X-ray source Kα 1 of 1.54056 Å). The XRD studies were carried out at room temperature using the solution containing the mixture of SmCo 5 /α-Fe nanopowders and DMF dried on the circular quartz plate.
The SEM images were recorded after each stage of grinding taken in the SE mode. All examined powders were fixed to the sample holder by the carbon double-coated conductive tape. The EDXS composition elemental maps were collected using JEOL EDX SSD and IXRF LN detectors attached to the JEOL JSM 7100F and JSM 6480 SEM (JEOL Ltd., Tokyo, Japan), respectively.
After the last stage of shredding (after 10 h), transmission electron microscopy (TEM) with a JEOL JEM 3010 (JEOL Ltd., Tokyo, Japan) was conducted. For TEM analysis, the sample was dried and purified in ethanol in the ultrasonic washer for 2 h in order to remove the excess DMF and to fracture the agglomerates. Afterwards, a few drops of the powder suspended in ethanol were attached adhesively to the carbon-coated Cu grid (400 mesh). The X-ray photoelectron spectroscopy (XPS) measurements with a PHI 5700/660 spectrometer (Physical Electronics Inc., Eden Prairie, MN, USA) were performed for samples milled for 0.5, 4 and 10 h. The XPS data were collected at room temperature using a monochromatic Al K α (1486.6 eV) X-ray source. Powders were compacted and fixed to the sample holder by using the carbon double-coated conductive tape. Subsequently, such prepared specimen was measured in three separated steps. At first, the as-milled material was studied after its storage under ultra-high vacuum for about one week. Afterwards, ion etching by an Ar + beam was applied for 1 h, and then the same specimen was measured. All the obtained spectra were calibrated using the C1s peak (EB = 284.8 eV) as an adventitious carbon usually accumulated on the surface of the sample and used as a reference for a charge correction.
The hysteresis loops were determined at 2 and 200 K based on the wide-range superconducting quantum interference device (SQUID) magnetometer MPMS XL7 (Quantum Design Inc., San Diego, CA, USA) under the change in the external magnetic field up to µ 0 H = 7 T.

XRD
The XRD measurements were carried out at room temperature for all as-milled powders. The XRD patterns collected for all specimens are depicted in Figure 1. The subsequent broadening of diffraction peaks is noticed with the increase in milling time due to the reduction in crystallite size. It is worth mentioning that in XRD patterns, there is no evidence of other impurity traces, namely, from the ZrO 2 milling media or samarium oxide. The observed broadening of diffraction peaks is associated with the gradual amorphization upon the duration of the grinding process manifested mainly as a significant decrease in crystallite size. As the next effect of mechanosynthesis, visible differences in the peak intensity are observed due to the separation of various crystal phases, the shape and size effects and the varied texture related to different orientations of the crystallites in the grains/particles. The dominant 1:5 hexagonal CaCu 5 , space group P6/mmm, crystal phase is evidenced in all samples. Obviously, the presence of cubic α-Fe/Co as the composite component is evident. In addition, the 2:17 rhombohedral Th 2 Zn 17 phase, space group hR57, is formed. The phase content was determined based on Rietveld refinement of diffraction patterns. The evolution of unit cell parameters determined based on crystal structure refinement is presented in Table 1. Examples of refined XRD patterns for 0.5 h-and 10 h-milled samples are presented in Figure 2a,b. Other refined patterns are shown in Figure S1a-f, while refinement factors are added into Table S1 in the Supplementary Materials. For the short-milled specimen, the lattice parameters of the main 1:5 phase equal a = 4.996 Å and c = 3.976 Å. The prolonged milling up to 10 h is reflected in the slight expansion of the unit cell a = 5.000 Å and c = 4.083 Å, also proved by the increase within a volume of the unit cell from 85.379 (0.5 h) to 88.410 Å 3 (10 h). The observed phenomenon can be caused by the Co/Fe interdiffusion process associated with the evident difference in the atomic radii of both constituents (Co-1.25 Å, Fe-1.26 Å) and separation of the additional 2:17 phase. However, the influence of gradual amorphization and reduction in crystallite size cannot be neglected. It is accompanied by a reduction in α-Fe and 2:17 contents. Comparing to t = 0.5 h, where the estimated phase content is about 1:5-48.63%, α-Fe-39.22%, and 2:17-12.15%, after t = 10 h of synthesis, such relation is changed into 1:5-66.91%, α-Fe-26.28%, and 2:17-6.28%. Thus, the analysis of the phase content points to the significant increase in the 1:5 phase of about 18% (see Figure 2c). The dominant 1:5 hexagonal CaCu5, space group P6/mmm, crystal phase is evidenced in all samples. Obviously, the presence of cubic α-Fe/Co as the composite component is evident. In addition, the 2:17 rhombohedral Th2Zn17 phase, space group hR57, is formed. The phase content was determined based on Rietveld refinement of diffraction patterns. The evolution of unit cell parameters determined based on crystal structure refinement is presented in Table 1. Examples of refined XRD patterns for 0.5 h-and 10 hmilled samples are presented in Figure 2a,b. Other refined patterns are shown in Figure  S1a-f, while refinement factors are added into Table S1 in the Supplementary Materials. For the short-milled specimen, the lattice parameters of the main 1:5 phase equal a = 4.996 Å and c = 3.976 Å . The prolonged milling up to 10 h is reflected in the slight expansion of the unit cell a = 5.000 Å and c = 4.083 Å , also proved by the increase within a volume of the unit cell from 85.379 (0.5 h) to 88.410 Å 3 (10 h). The observed phenomenon can be caused by the Co/Fe interdiffusion process associated with the evident difference in the atomic radii of both constituents (Co-1.25 Å , Fe-1.26 Å ) and separation of the additional 2:17 phase. However, the influence of gradual amorphization and reduction in crystallite size cannot be neglected. It is accompanied by a reduction in α-Fe and 2:17 contents. Comparing to t = 0.5 h, where the estimated phase content is about 1:5-48.63%, α-Fe-39.22%, and 2:17-12.15%, after t = 10 h of synthesis, such relation is changed into 1:5-66.91%, α-Fe-26.28%, and 2:17-6.28%. Thus, the analysis of the phase content points to the significant increase in the 1:5 phase of about 18% (see Figure 2c).  The presence of an additional 2:17 phase was already observed for other SmCo 5 /α-Fe compounds with different Fe contents [22,[48][49][50][51][52][53][54][55]. To see all numerical data for phase contents, please refer to Table S1 in the Supplementary Materials. The application of HEBM as a method of composite synthesis over the selected milling time t has a significant influence on the microstructure, demonstrated as, e.g., by the gradual decrease in the average crystallite size (d cryst ) from about 526 (bulk SmCo 5 compound) and 596.47 (α-Fe) to about 6.3 (SmCo 5 ) and 47.19 nm (α-Fe), respectively (see Figure 2d). For more details, see Table S2 in the Supplementary Materials.
As it has been evidenced so far for other ball-milled intermetallics [16,17,[32][33][34][35][36][37][38][39][40][41]58,59], the d cryst is significantly reduced in the first stages of mechanical synthesis (here, it is noted just after 0.5 h) and further grinding is reflected only in its slight variation. As it was evidenced by, e.g., Rama Rao et al. [53], the crystallite size for the same quantity of Fe is significantly reduced after 20 h of milling, starting from about 30 (4 h) to 5 nm (20 h). Further milling up to 50 h exhibits only a slight change in such parameter. Similar behavior was also noticed for 10%wt. Fe. The rapid decrease in d cryst during the first stages and its further subtle difference is an effect of the change within the deformation mechanism from plastic deformation to grain boundaries sliding, and the time point at which such deformation occurs is dependent not only on the milling parameters but also on the milled material. Please note that the final size of crystallites is also related to the duration of the synthesis process as well as being determined by the melting point, e.g., the final size after 100 h was found to be about 8 and 14 nm for individual components Fe (melting point at 1535 • C) and Co (melting point at 1495 • C), respectively [59,60].   The presence of an additional 2:17 phase was already observed for other SmCo5/α-Fe compounds with different Fe contents [22,[48][49][50][51][52][53][54][55]. To see all numerical data for phase contents, please refer to Table S1 in the Supplementary Materials. The application of HEBM as a method of composite synthesis over the selected milling time t has a significant influence on the microstructure, demonstrated as, e.g., by the gradual decrease in the average The analysis of the lattice strain at 10 h of pulverization indicates its progressive increase (see Figure 2d). Considerable lattice straining is one of the critical effects for various ball-milled intermetallics [5,16,17,[32][33][34][35][36][37][38][39][40][41] induced by the substantial lattice disorder generated during HEBM by different deformation mechanisms. The sintering of as-milled powders leads to an increase in crystallite size due to the crystallization process caused by the heat treatment [45]. The crystallite size is dependent on the sintering temperature, e.g., being raised from about 18 (973 K) to about 32 nm (1123 K) [45].

Microstructure: SEM and TEM
By analyzing SEM images, the microstructural evolution over the pulverization process is noted (see Figure 3). As one may observe, all as-milled powders are inhomogeneous and composed of irregular particles, which tend to form flakes during a longer mechanosynthesis time. For more details, see Figure S2 in the Supplementary Materials. Thus, the shape and size of as-milled powders vary across grinding times, and at the final stage (t = 10 h), the average thickness of particles is even less than 200 nm (see Figure 3d). Nonetheless, by analyzing SEM images, one may observe larger particles and finer ones, mostly agglomerated due to a tendency to reduce their surface via coalescence. As we have already shown, the synthesis of inhomogeneous R-T-based intermetallics is typical for the HEBM process [16,17,[32][33][34][35][36][37][38][39][40][41] regardless of the type of crystal structure in as-synthesized powders. Noticeably, the wet milling performed in DMF leads to the formation of irregular micro/nanoparticles, which tend to agglomerate depending on their sizes. The elemental map on the t = 0.5 h sample (see Figure 4a) indicates the non-homogenous distribution of the α-Fe powder in the SmCo5 matrix. One may notice regions with evident domination of iron powder, which is consistent with XRD data. The extended milling up to t = 10 h leads to significant homogenization of the as-milled powders. In this case, the elemental map (see Figure 4b) does not show regions with visible phase separation, characteristic for a shorter synthesis time. Thus, at the end of the synthesis, the asmilled powder is much more homogeneous. The TEM images of the SmCo5/α-Fe nanopowder ground for 10 h are shown in Figure  5. The samples milled for a shorter time were too thick for TEM imaging. The bright-field image (BF) exhibits a dark region of interest, pointing to a significant thickness of the studied material. The analysis depicting a region of interest reveals the irregular nanoflake particle with dimensions of 486 (length) and 254 nm (maximal width) as denoted by ar- The elemental map on the t = 0.5 h sample (see Figure 4a) indicates the non-homogenous distribution of the α-Fe powder in the SmCo 5 matrix. One may notice regions with evident domination of iron powder, which is consistent with XRD data. The extended milling up to t = 10 h leads to significant homogenization of the as-milled powders. In this case, the elemental map (see Figure 4b) does not show regions with visible phase separation, characteristic for a shorter synthesis time. Thus, at the end of the synthesis, the as-milled powder is much more homogeneous. The elemental map on the t = 0.5 h sample (see Figure 4a) indicates the non-homogenous distribution of the α-Fe powder in the SmCo5 matrix. One may notice regions with evident domination of iron powder, which is consistent with XRD data. The extended milling up to t = 10 h leads to significant homogenization of the as-milled powders. In this case, the elemental map (see Figure 4b) does not show regions with visible phase separation, characteristic for a shorter synthesis time. Thus, at the end of the synthesis, the asmilled powder is much more homogeneous. The TEM images of the SmCo5/α-Fe nanopowder ground for 10 h are shown in Figure  5. The samples milled for a shorter time were too thick for TEM imaging. The bright-field image (BF) exhibits a dark region of interest, pointing to a significant thickness of the studied material. The analysis depicting a region of interest reveals the irregular nanoflake particle with dimensions of 486 (length) and 254 nm (maximal width) as denoted by ar- The TEM images of the SmCo 5 /α-Fe nanopowder ground for 10 h are shown in Figure 5. The samples milled for a shorter time were too thick for TEM imaging. The brightfield image (BF) exhibits a dark region of interest, pointing to a significant thickness of the studied material. The analysis depicting a region of interest reveals the irregular nanoflake particle with dimensions of 486 (length) and 254 nm (maximal width) as denoted by arrows. However, the detailed analysis of the dark-field image (DF) exhibits the presence of an irregular and non-homogeneous grained structure. The performed analysis of the crystallite size distribution by using the log-normal function allows for an estimate of the average crystallite size. Based on the crystallite size distribution (CSD) histogram, such parameter is estimated as <d cryst > ≈ 6.4 ± 0.18 nm. Such result is in good agreement with that obtained from the XRD pattern for the 1:5 phase, which dominates in the sample synthesized for t = 10 h. The selected area electron diffraction (SAED) pattern collected for the image depicted in Figure 5b was fully indexed according to the dominant hexagonal CaCu 5 phase (red rings) and cubic α-Fe (blue rings) (see Figure 5c). The obtained electronogram is typical for polycrystalline and textured specimens. A certain degree of fortuitousness observed in the HEBM method TEM micrographs certainly confirms the polycrystalline heterogeneous nature of the SmCo5/α-Fe nanocomposite composed of inhomogeneously distributed and irregularly shaped crystallites. an irregular and non-homogeneous grained structure. The performed analysis of the crystallite size distribution by using the log-normal function allows for an estimate of the average crystallite size. Based on the crystallite size distribution (CSD) histogram, such parameter is estimated as <dcryst> ≈ 6.4 ± 0.18 nm. Such result is in good agreement with that obtained from the XRD pattern for the 1:5 phase, which dominates in the sample synthesized for t = 10 h. The selected area electron diffraction (SAED) pattern collected for the image depicted in Figure 5b was fully indexed according to the dominant hexagonal CaCu5 phase (red rings) and cubic α-Fe (blue rings) (see Figure 5c). The obtained electronogram is typical for polycrystalline and textured specimens. A certain degree of fortuitousness observed in the HEBM method TEM micrographs certainly confirms the polycrystalline heterogeneous nature of the SmCo5/α-Fe nanocomposite composed of inhomogeneously distributed and irregularly shaped crystallites.

Magnetic Properties
The hysteresis loops M(H) for all as-milled SmCo5/α-Fe (5%wt. of α-Fe) specimens are shown in Figure 6. By analyzing such curves at specific milling times, one may notice their deformation and the appearance of the characteristic kink. This may indicate the additional magnetic phase formed during HEBM, which is magnetically decoupled from

Magnetic Properties
The hysteresis loops M(H) for all as-milled SmCo 5 /α-Fe (5%wt. of α-Fe) specimens are shown in Figure 6. By analyzing such curves at specific milling times, one may notice their deformation and the appearance of the characteristic kink. This may indicate the additional magnetic phase formed during HEBM, which is magnetically decoupled from the hard phase. Apart from that, it may also be associated with some other microstructural non-uniformities. Such behavior was already evidenced in other ball-milled intermetallics and is mostly related to the exchange spring mechanism [12,30,[61][62][63]. In our case, we may observe such a deformation and characteristic kink especially at 2 K for t = 6 and 8 h, whereas at room temperature, such effect is not so clearly evident. Thus, at the low temperature for the middle t, the existence of a decoupling effect between hard SmCo 5 and soft α-Fe phases seems to be prevailing. In addition, as one may observe, the magnetization at the external magnetic field (H ext ) of 7 T is almost saturated in the low-and roomtemperature regimes. However, the variation in M 7T (t) dependence is non-linear, probably as an effect of the change within the particle size and shape and different exchange coupling between all magnetic phases present in nanopowders throughout the pulverization process. The observed non-linearity of M 7T (see Table 2) may also be due to possible agglomeration of finer particles, which tend to reduce their surface energy, especially under applying the external magnetic field. As demonstrated for other materials, the finer nanoparticles, nanoflakes, favor arrangements along an easy axis of magnetization, which is reflected as an evident increase in saturation magnetization [1,10,14,18,20,21,27,28,39]. and is mostly related to the exchange spring mechanism [12,30,[61][62][63]. In our case, we may observe such a deformation and characteristic kink especially at 2 K for t = 6 and 8 h, whereas at room temperature, such effect is not so clearly evident. Thus, at the low temperature for the middle t, the existence of a decoupling effect between hard SmCo5 and soft α-Fe phases seems to be prevailing. In addition, as one may observe, the magnetization at the external magnetic field (Hext) of 7 T is almost saturated in the low-and roomtemperature regimes. However, the variation in M7T(t) dependence is non-linear, probably as an effect of the change within the particle size and shape and different exchange coupling between all magnetic phases present in nanopowders throughout the pulverization process. The observed non-linearity of M7T (see Table 2) may also be due to possible agglomeration of finer particles, which tend to reduce their surface energy, especially under applying the external magnetic field. As demonstrated for other materials, the finer nanoparticles, nanoflakes, favor arrangements along an easy axis of magnetization, which is reflected as an evident increase in saturation magnetization [1,10,14,18,20,21,27,28,39]. The change in all magnetic parameters estimated from the hysteresis loops exhibits a non-linear variation (see Table 2). Generally, the observed modification of Hc versus t is mainly related to the microstructural modification over the milling process and different phase contents. Usually, for ball-milled materials, in the first stages of synthesis, the coercivity increases due to effective continuous particle/grain size refinement. The further reduction within HC values in HEBM is caused by the formation of flaked particles, crystal-  The change in all magnetic parameters estimated from the hysteresis loops exhibits a non-linear variation (see Table 2). Generally, the observed modification of Hc versus t is mainly related to the microstructural modification over the milling process and different phase contents. Usually, for ball-milled materials, in the first stages of synthesis, the coercivity increases due to effective continuous particle/grain size refinement. The further reduction within H C values in HEBM is caused by the formation of flaked particles, crystallite refinement and a stronger exchange coupling between them [32][33][34][35][36][37][38][39][40][41]. In our case, such a mechanism is relatively sustained. Therefore, up to the middle synthesis time t, we can observe the gradual but non-linear increase in coercivity, which is obviously higher at 2 K than at room temperature, with its maximum of H C = 1.24 T for t = 6 h. At room temperature, such a maximum is shifted to the shorter t and maxes for the sample milled for t = 4 h, where H C = 0.42 T. The estimated coercivity is relatively low comparing to other ballmilled powders [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23]. It is worth mentioning that other factors, e.g., an accumulation of defects created by grinding, larger shape anisotropy, variation within magnetic anisotropy, possible decay of the long-range magnetic order and gradual amorphization, may have a significant impact on H C (t) dependence.
The non-linear variation can also be noticed for the M R (t) dependence. However, in this case, the observed maximum value for this parameter is evident at t = 1.5 h (see Table 2) and is subsequently reduced for further milling. The observed non-linearity for remanence may be associated with (i) the change in the microstructure and consequently the shape anisotropy, (ii) the noticeable agglomeration of particles for the higher milling time, (iii) the variation within magnetic interactions between neighboring nanograins and modification within soft/hard magnetic coupling over milling or (iv) the change in the 1:5, α-Fe and 2:17 phase contents. In the research performed by Rama Rao et al. [53], it was shown that during milling, the coercivity is enhanced to about 7 kOe at 4 h and thereafter is significantly reduced for longer grinding to about 1 kOe noted for 54 h. Such a drastic decrease was explained by the authors as an evident effect of nanostructurization and amorphization of the 1:5 phase. Simultaneous modification within magnetization is caused by the probable disordering at the surface as an effect of the particle size reduction (initial decrease in magnetization) and recovery of Co moments preliminarily suppressed in the alloy from the elemental value due to localization by distortion (an increase in magnetization). The reduction in coercivity was also noted from nominal SmCo 5 (4.6 kOe) to 10%wt. Fe (1.4 kOe). Further, an increase in the iron content in nanocomposites up to 35%wt leads to an ensuing H C reduction [55]. One of the possibilities to improve coercivity is sintering the as-milled powders [45], e.g., from 973 (H C ≈ 6 kOe) to 1073 K (H C = 8.2 kOe), or optimization of milling parameters, e.g., under a magnetic field [22].
One of the utilizable tools for determining magnetic properties in nanomaterials is the analysis of the remanence-to-saturation (M R /M S ) ratio, which is strongly dependent on the alteration within the particle size versus the grinding time [13,[35][36][37][38][39][40][41]64]. As it was proposed by the Stoner-Wohlfarth (S-W) model, such ratio equals M R /M S = 0.5 at T = 0 K [65] and is typical for randomly oriented single-domain nanoparticles. Such value can be exceeded due to the increase within the exchange coupling between neighboring nanograins caused by the deviation in magnetization from the easy axes. The enhanced parameter M R /M S is generally observed for nanocrystalline hard magnets with a strong short-range exchange coupling between different nanograins [13,[64][65][66]. The lowering of the M R /M S << 0.5 ratio is usually demonstrated for non-interacting particles, which exhibit superparamagnetic behavior, generally due to the weak exchange coupling [67]. Bearing in mind the prediction of the S-W model, we may analyze the remanence-to-saturation ratio for the as-milled SmCo 5 /α-Fe powders with 5 wt. α-Fe%. In such an analysis, we took into account M 7T instead of M S . Thus, both at low temperature and room temperature, such ratios are relatively high and mostly higher or just below the limit predicted by the S-W model, which may point to relatively strong exchange coupling between nanograins slightly modified over pulverization. However, the observed phenomenon is also owing to multiple factors such as (i) a visible decrease within particle/crystallite size, (ii) an increased fraction of finer particles inclined at random orientation and incoherence in grain boundaries, (iii) an enhancement of the random anisotropy and (iv) destruction of the long-range magnetic order and its transformation into a short-range one during the milling process.
As one may notice, the hysteresis loops deteriorate with maximal deformation observed at t = 6 h (for 2 K) and t = 4 h (for 300 K). Such behavior is associated with the modification within the exchange coupling between hard and soft magnetic phases. One of the tools to analyze the hard and soft contributions of the composite magnetization process is a derivative of the remagnetization dM/dH curves versus H. For the as-milled SmCo 5 /α-Fe (5%wt. of α-Fe) powders, such curves are depicted in Figure 7. Generally, the area under curves dM/dH is related to the hard/soft magnetic species. As it was previously demonstrated by Pop et al. [52], the presence of only one thin peak placed at the low field is attributed to a narrow crystallite size distribution and stronger exchange coupling. The SmCo 5 /α-Fe composites with 20 wt.% milled for 6 and 8 h exhibit such a peak just below 0 T, and for such samples, coercivity is also very low, pointing to the vanishing of exchange coupling due to a reduction in the hard phase by milling. The subsequent annealing leads to a shift in the dM/dH maximum and a significant increase in coercivity. ratio for the as-milled SmCo5/α-Fe powders with 5 wt. α-Fe%. In such an analysis, we took into account M7T instead of MS. Thus, both at low temperature and room temperature, such ratios are relatively high and mostly higher or just below the limit predicted by the S-W model, which may point to relatively strong exchange coupling between nanograins slightly modified over pulverization. However, the observed phenomenon is also owing to multiple factors such as (i) a visible decrease within particle/crystallite size, (ii) an increased fraction of finer particles inclined at random orientation and incoherence in grain boundaries, (iii) an enhancement of the random anisotropy and (iv) destruction of the long-range magnetic order and its transformation into a short-range one during the milling process.
As one may notice, the hysteresis loops deteriorate with maximal deformation observed at t = 6 h (for 2 K) and t = 4 h (for 300 K). Such behavior is associated with the modification within the exchange coupling between hard and soft magnetic phases. One of the tools to analyze the hard and soft contributions of the composite magnetization process is a derivative of the remagnetization dM/dH curves versus H. For the as-milled SmCo5/α-Fe (5%wt. of α-Fe) powders, such curves are depicted in Figure 7. Generally, the area under curves dM/dH is related to the hard/soft magnetic species. As it was previously demonstrated by Pop et al. [52], the presence of only one thin peak placed at the low field is attributed to a narrow crystallite size distribution and stronger exchange coupling. The SmCo5/α-Fe composites with 20 wt.% milled for 6 and 8 h exhibit such a peak just below 0 T, and for such samples, coercivity is also very low, pointing to the vanishing of exchange coupling due to a reduction in the hard phase by milling. The subsequent annealing leads to a shift in the dM/dH maximum and a significant increase in coercivity. Moreover, there is also a second maximum which is a proof for a mixture of uncoupled soft and hard phases [52]. In our case, at low temperature, we can see a relatively narrow peak just up to t = 2 h of the synthesis process, where we also note a significant increase in the 1:5 phase and a decrease in the α-Fe phase. For the higher milling time, the maximum is reduced and is shifted towards 0 T, which probably indicates that some Fe crystallites are not perfectly coupled. An additional maximum for t = 6 and 10 h is also observed, which can testify to the reduction in coupling and the significant crystallite size distribution. At room temperature, the variation in the dM/dH peaks is similar. It is worth mentioning that in both cases, the peaks' width and long tails can indicate a significant percentage of crystallites which are not well coupled. Thus, the intensity of dM/dH peaks around HC is relatively weak, pointing to the low contribution of a well-exchange-coupled Moreover, there is also a second maximum which is a proof for a mixture of uncoupled soft and hard phases [52]. In our case, at low temperature, we can see a relatively narrow peak just up to t = 2 h of the synthesis process, where we also note a significant increase in the 1:5 phase and a decrease in the α-Fe phase. For the higher milling time, the maximum is reduced and is shifted towards 0 T, which probably indicates that some Fe crystallites are not perfectly coupled. An additional maximum for t = 6 and 10 h is also observed, which can testify to the reduction in coupling and the significant crystallite size distribution. At room temperature, the variation in the dM/dH peaks is similar. It is worth mentioning that in both cases, the peaks' width and long tails can indicate a significant percentage of crystallites which are not well coupled. Thus, the intensity of dM/dH peaks around H C is relatively weak, pointing to the low contribution of a well-exchange-coupled composite, whereas the domination of the highest peak around small fields reveals the main contribution of a non-coupled soft magnetic phase. A visible shoulder at higher magnetic fields may correspond to a non-coupled hard magnetic phase. The interpretation of such curves is also rather unobvious due to the possible interdiffusion Co/Fe process.
The maximum energy product (BH) max as the maximum amount of magnetic energy stored in a magnet was estimated based on demagnetization curves which are shown in Figure S3 in the Supplementary Materials. As we may notice at the low temperature, such a parameter varies in a non-linear way and maxes at t = 6 h, where (BH) max = 84.7 kJ/m 3 , consistent with the change in coercivity. At room temperature, we observe the maximum (BH) max = 35.1 kJ/m 3 at t = 1 h. The estimated values are significantly lower than those recently published by Chakraborty et al. [57] for SmCo 5 /α-Fe with 10 wt.% Fe samples, where the maximum value of about (BH) max = 145 kJ/m 3 was observed for specimens milled for 6 h and annealed at 510 • C.
Similarly, for the pure SmCo 5 compound, the maximum energy product was determined as 231 kJ/m 3 [2]. Previously, we have also shown that extended milling of SmCo 5 is reflected in a drastic reduction in (BH) max , where the maximum of 129.7 kJ/m 3 was observed at room temperature for the sample milled for t = 1 h [16,17,54]. Thus, the addition of 5 wt.% Fe leads to a significant reduction in such a parameter, probably due to the weak exchange coupling between the hard and soft phases of various grain sizes. As it was already published, the sintering of as-milled powders for different sintering temperatures leads to an improved (BH) max parameter up to almost 278 kJ/m 3 for 1073 K, which is an effect of exchange coupling enhancement [45]. The modification within the maximum energy product can also be tuned by the change in the iron content combined with heat treatment [55]. The optimal addition of such a soft phase was estimated in the range of 20-30%wt., where (BH) max attains the maximum value of about 105.8 kJ/m 3 for 25%wt. and an annealing temperature of 873 K. A slightly lower value of 84 kJ/m 3 was found by Pop et al. [52] for the sample with 20%wt. milled for 8 h and annealed at 550 • C for 1.5 h. Thus, optimization of the iron content in nanocomposites followed by the optimization of synthesis parameters is crucial for enhancing the maximum energy product.
The magnetic anisotropy constants at different temperatures and milling times may be estimated from the approach to the saturation region by using the empirical law where M(H) is the magnetization measured at the external magnetic field, and M S is the saturation magnetization. The a/ √ H term represents the strain field around dislocations by Brown and non-magnetic inclusions in voids by Néel [37], so it can be associated with pointlike defects and intrinsic magnetostatic fluctuations, in which both, in nanocompounds, are strongly dependent on the particle size. The second term b/H 2 represents various kinds of anisotropies in the cubic compound, e.g., magnetocrystalline anisotropy, shape anisotropy or strain anisotropy, and can be approximated by b ≈ 8 105 where K 1 is a magnetocrystalline anisotropy as a component of the effective anisotropy K eff . The last term χ p H represents the magnetic paraprocess caused by the partial suppression of spin waves by an external applied magnetic field. The fitting of M(H) was performed in the field range of 1-7 T, as it is shown in Figure S4, and estimated values of M S , a, b and χ p are placed in Table S3 in the Supplementary Materials. The analysis of the K 1 parameter (see Table 2) reveals its modification versus the milling time. Generally, a reduction in the grain size may increase the effective inter-grain exchange interaction and decrease the effective magnetic anisotropy. However, in this case, the variation in the K 1 parameter as a component of the effective anisotropy does not follow such a trend. Most likely, it is not only an effect of the modification within exchange interactions and the decrease in grain size but also of (i) the influence of the particle shape modified over milling, (ii) a variation in the 1:5, 2:17 and α-Fe phase contents and (iii) the subsequent Co/Fe interdiffusion process. Therefore, the change in anisotropy versus the milling time, which is typical for inhomogeneous HEBM powders, might be rather a resultant effect of all aforementioned factors than a specific trend. Nonetheless, the variation in anisotropy observed by Le Breton et al. [40] based on Monte Carlo simulations revealed a linear increase in the anisotropy constant within a Sm−Co−Fe mixed interface confirming that Fe/Co interdiffusion can lead to an increase in the coercive field.

Electronic Structure: XPS
The survey spectra measured after ion cleaning by an Ar + beam for about 1 h are presented in Figure 8a. The main photoemission lines Sm3d, Co2p, Fe2p, Sm4p, Sm4d, Co3s and Co3p and the Auger CoLMM and FeLMM lines are denoted. There is no drastic change in all spectra collected for samples synthesized for t = 0.5, 4 and 10 h. The chemical composition estimated based on survey spectra acquired before and after ion cleaning is determined for comparison (see Table 3). The evident difference between both values is related to a significant quantity of surface impurities noted before using the Ar + beam. After ion cleaning, the amount of C1s and O1s is significantly reduced. Note that the atomic concentration of iron and cobalt may be slightly disturbed due to the overlapping of 2p core lines with the Auger LMM lines.
The measured valence band spectra (VB) measured in a broad range together with the closely located core level lines are presented in Figure 8b. Around the binding energy of BE ≈ 58.6 eV, we may observe a Co3p line dominating over Fe3p located at about BE ≈ 52.1 eV. Sm5s states with significantly low intensity are observed just above 40 eV. In the range 30-16 eV, one may notice overlapped Sm4p and O2s states. The latter is observed despite surface ion cleaning as a residual oxygen impurity trace. The valence bands in the range −2 eV-12 eV are dominated by hybridized (Co/Fe)3d states with a typical 3d spectrum in the vicinity of the Fermi level (E F ). The cusp around 5.2 eV is typical for trivalent Sm 3+ (4f) states. The divalent Sm 2+ (4f) states are overlapped with 3d and therefore, due to the domination of Co/Fe, its contribution can be barely noticed. The increase in the intensity of states in VB in relation to the t milling time is evident (see insert into Figure 8b) and may be related to the change within magnetic properties and the increase in the 1:5 phase content.
The Sm3d spectra reveal splitting due to a spin-orbit interaction of ∆E = 26.7 eV into 3d 5/2 (BE ≈ 1082.6eV) and 3d 3/2 (BE ≈ 1109.3 eV) lines. Such parameters are typical for metallic Sm 3+ [68][69][70][71]. Nonetheless, below the main line, one may observe additional components visible as a shoulder with lower binging energy and lower intensity at BE ≈ 1076.1 and ≈ 1103.3 eV, respectively. Both lines are typical for Sm 2+ states. Thus, the Sm3d spectrum consists of two components related to Sm 3+ and Sm 2+ . The divalent Sm3d 5/2 component associated with the 4f 6 configuration is usually observed between 1070.0 and 1080.0 eV and the trivalent Sm3d 5/2 (4f 5 ) component is usually observed between 1080.0 and 1090.0 eV [68][69][70][71]. We do not observe any noticeable variation in or sifting of samarium lines versus the milling time. Therefore, the noted mixed-valence state is independent of the synthesis duration. change in all spectra collected for samples synthesized for t = 0.5, 4 and 10 h. The chemical composition estimated based on survey spectra acquired before and after ion cleaning is determined for comparison (see Table 3). The evident difference between both values is related to a significant quantity of surface impurities noted before using the Ar + beam. After ion cleaning, the amount of C1s and O1s is significantly reduced. Note that the atomic concentration of iron and cobalt may be slightly disturbed due to the overlapping of 2p core lines with the Auger LMM lines.  The measured valence band spectra (VB) measured in a broad range together with the closely located core level lines are presented in Figure 8b. Around the binding energy of BE ≈ 58.6 eV, we may observe a Co3p line dominating over Fe3p located at about BE ≈ 52.1 eV. Sm5s states with significantly low intensity are observed just above 40 eV. In the range 30-16 eV, one may notice overlapped Sm4p and O2s states. The latter is observed  The Co2p 3/2 (BE ≈ 777.6 eV) and Co2p 1/2 (BE ≈ 792.7 eV) core level spectra are plotted in Figure 8d. The L-S splitting is stable during sample processing and equals about ∆E ≈ 15.1 eV. The Fe2p lines (see Figure 8e) are split into Fe2p 3/2 (BE ≈ 706.6 eV) and Fe2p 1/2 (BE ≈ 719.5 eV), with a splitting parameter of about ∆E ≈ 12.9 eV typical for metallic iron. Obviously, the CoLMM Auger line as the dominant component is overlapped with Fe2p. Nonetheless, the Co2p, as well as Fe2p, lines are shifted to a higher BE of about 0.2 eV, and such energy shifts can be caused by: (i) a change in the local environment of Co atoms induced by the modification of the phase content throughout the synthesis duration and (ii) the influence of hybridization of the Co3d states with Sm 5d6s and Fe3d, as well as conduction states. Generally, such chemical shifts, when compared to the corresponding pure metals, reveal a charge transfer from the Sm rare-earth to (Co/Fe)3d band. The observed chemical shift is usually larger for Sm-based compounds due to the mixed valence of Sm ions [69,70].
Multiplet splitting of the transition metal 3s core-level photoemission spectra is generally used to determine the magnetic moment of the 3d atom [72][73][74][75][76][77][78][79]. Its value is related to the partial filling of the 3d shell. Therefore, due to the quenching of the orbital moment, the 3d moment is around 2S. Using a simple model [50][51][52][53][54], the splitting of 3s spectra with two final states in relation to uncompensated 3d states can be discussed. As it is known, the initial state may be described as n+1 L, while after photoionization, the double final state associated with the different arrangements of the 3d N spin and 3s vacancy is defined as the n+2 L high-spin term (S + 1/2) (parallel arrangement) and the n L low-spin term (S − 1/2) (antiparallel arrangement). Therefore, in the 3s photoemission spectra, one may notice two peaks, namely, the main peak with higher intensity (I A ) and the satellite peak with lower intensity (I B ), as equivalents of two final states with the ratio I B ( n L)/I A ( n+2 L) = S/(S + 1) where S represents the spin moment of the unpaired 3d electrons. The splitting parameter (∆E) between both peaks determined from the 3s fitting is proportional to the exchange overlap integral of 3s and 3d shells (J 3s-3d ) based on the relation ∆E = (2S + 1) J 3s-3d (4) In addition, the magnetic moment of the 3d atom can be ascribed as The results of the Co3s spectra fitting for powders milled for 0.5, 4 and 10 h are depicted in Figure 9. It is worth mentioning that the quantity of iron in the tested nanocomposites is too low to observe similar splitting to Co3s states, and that is why only one single peak around 90.5 eV can be noticed. Thus, all shown Co3s spectra were fitted by using an asymmetric Doniach-Sunjić (DS) type of line [55] with the same asymmetry parameter α for both I A and I B peaks as it was previously performed for other compounds [33,34,[50][51][52][53][54]. All parameters obtained from the applied fitting procedure are presented in Table 4. As we have observed, the Co3s lines for all measured powders exhibit quite significant splitting. Laslo et al. [80] already performed Co3s analysis for the series of RCo 5 compounds and the estimated ∆E parameter for the SmCo 5 compound equaled 3.5 eV, which is significantly lower than that obtained by us. However, the presented results suggest that the authors did not use the typical asymmetric line shape in the fitting procedure, which is usually applied for 3s lines [72][73][74][75][76][77][78]. Nonetheless, the ∆E parameters obtained by us are close to those previously noted for pure Co, where they were found to be about 4.5-4.7 eV [72][73][74][75]79,80]. However, the value of the estimated cobalt magnetic moment is significantly lower than for pure Co (1.7 µ B ) [79], probably due to the presence of α-Fe in the nanocomposite. Moreover, the increase in the milling time is reflected in a visible increase in (i) the Co magnetic moment and (ii) a slight increase in the 3s-3d exchange integral J 3s-3d values within the experimental uncertainty. The observed phenomena are related to the increase in the 1:5 phase and the reduction in α-Fe over the milling time as a consequence of the variation within the 3s-3d interaction due to hybridization effects within the (Co/Fe)3d band. The observed changes are also consistent with the VB, especially with the evident increase within the intensity of states at the Fermi level (see insert in Figure 8b).

Conclusions
In this study, we presented exchange-coupled SmCo5/α-Fe nanocomposites with 5%wt. of α-Fe content synthesized by wet HEBM for various periods up to 10 h. We demonstrated that the modification of the multiphase polycrystalline crystal structure over the milling process is reflected in: (i) the change in the 1:5, 2:17 and α-Fe phase contents, (ii) the decrease in crystallite size and the simultaneous increase in the lattice strain in the main hexagonal 1:5 phase and (iii) the slight expansion of the 1:5 unit cell, probably associated with the Co/Fe interdiffusion process. The microstructure of as-milled powders evolved from coarse grains/particles to finer nanoparticles/nanoflakes during the pulverization time. Generally, the magnetic properties of milled nanopowders are dependent on the microstructure. The analysis of hysteresis loops revealed the presence of exchange coupling between the hard and soft phases. The magnetic parameters vary in a non-linear way, probably due to size and shape effects, different phase contents and the subsequent modification within the exchange coupling strength. The maximum coercivity HC above 1T was attained at low temperatures after t = 6-8 h of pulverization, but at room temperature, it was noted as about 0.4 T for t = 4 h. The maximum value of the (BH)max energy product of 35.1 kJ/m 3 at room temperature was determined after t = 1 h of milling which, on account of the presence of the soft α-Fe phase with different exchange couplings, is significantly lower than that for the corresponding SmCo5 nanopowders. The electronic structure studies of valence bands, as well as core level lines, revealed the intermediate valence state of Sm ions in all measured as-milled samples associated with the 4f 6 (Sm 2+ ) and 4f 5 (Sm 3+ ) samarium configurations. The analysis of the multiplet splitting of Co3s spectra indicated the evident increase in the Co magnetic moment and a slight increase in the J3s-3d exchange integral across the milling process, which is consistent with the increase in the 1:5 phase, the reduction in the α-Fe phase and the subsequent change within magnetic interactions in relation to hybridization effects within the (Co/Fe)3d band and Co/Fe interdiffusion process.
Summarizing, our approach shows a significant potential for cost-effective synthesis of SmCo5-based nanopowders by using a modified wet HEBM method. We are aware that

Conclusions
In this study, we presented exchange-coupled SmCo 5 /α-Fe nanocomposites with 5%wt. of α-Fe content synthesized by wet HEBM for various periods up to 10 h. We demonstrated that the modification of the multiphase polycrystalline crystal structure over the milling process is reflected in: (i) the change in the 1:5, 2:17 and α-Fe phase contents, (ii) the decrease in crystallite size and the simultaneous increase in the lattice strain in the main hexagonal 1:5 phase and (iii) the slight expansion of the 1:5 unit cell, probably associated with the Co/Fe interdiffusion process. The microstructure of as-milled powders evolved from coarse grains/particles to finer nanoparticles/nanoflakes during the pulverization time. Generally, the magnetic properties of milled nanopowders are dependent on the microstructure. The analysis of hysteresis loops revealed the presence of exchange coupling between the hard and soft phases. The magnetic parameters vary in a non-linear way, probably due to size and shape effects, different phase contents and the subsequent modification within the exchange coupling strength. The maximum coercivity H C above 1T was attained at low temperatures after t = 6-8 h of pulverization, but at room temperature, it was noted as about 0.4 T for t = 4 h. The maximum value of the (BH) max energy product of 35.1 kJ/m 3 at room temperature was determined after t = 1 h of milling which, on account of the presence of the soft α-Fe phase with different exchange couplings, is significantly lower than that for the corresponding SmCo 5 nanopowders. The electronic structure studies of valence bands, as well as core level lines, revealed the intermediate valence state of Sm ions in all measured as-milled samples associated with the 4f 6 (Sm 2+ ) and 4f 5 (Sm 3+ ) samarium configurations. The analysis of the multiplet splitting of Co3s spectra indicated the evident increase in the Co magnetic moment and a slight increase in the J 3s-3d exchange integral across the milling process, which is consistent with the increase in the 1:5 phase, the reduction in the α-Fe phase and the subsequent change within magnetic interactions in relation to hybridization effects within the (Co/Fe)3d band and Co/Fe interdiffusion process. Summarizing, our approach shows a significant potential for cost-effective synthesis of SmCo 5 -based nanopowders by using a modified wet HEBM method. We are aware that the interrelated magnetism and microstructure can be strictly controlled by optimizing various HEBM parameters. Nevertheless, the approach adopted by us may provide a promising basis for the first step in the fabrication of exchange-coupled SmCo 5 /α-Fe nanocomposite magnets which can be, in the next step, annealed in order to enhance, e.g., the coercivity and the value of the maximum energy product.