Structural and magnetic properties of RE-Al substituted nanocrystalline hexaferrites (Sr1−xRExAl2Fe10O19)

We describe the synthesis and characterization of magnetic materials with improved magnetic properties for permanent magnet applications. Sr1−xRExAl2Fe10O19 (RE = La, Ce, Tb, and Dy; x = 0.0 and 0.1) were prepared by ball milling and sintering at 1200 °C for 6 h. The effects of (Al3+–RE3+) substitution on the structural and magnetic properties of SrM hexaferrites were investigated by x-ray diffraction, scanning electron microscopy, and magnetic measurements at low (5 K), and high temperatures (from room temperature to above Curie temperature). The sample with x = 0, and that with La substitution, consisted of a single SrM phase. The rest of the samples contained traces of secondary oxide phases, and exhibited a small reduction in lattice parameters compared with the unsubstituted SrFe12O19. The coercivity of the samples (8800 ≤ HcM ≤ 9750 Oe) was more than double the standard value of the unsubstituted compound, and the Curie temperatures was >300 °C for all samples. Also, the saturation magnetization (32.5–46.4 emu g−1) was high enough to make the overall magnetic properties of these compounds potentially important for the permanent magnet industry.


Introduction
The permanent magnet (PM) industry faces real challenges in producing high performance magnets with high operating temperature and chemical and thermal stability at a reasonably low cost. The first generation of rareearth (RE) Sm-Co high performance PM, the (1-5) with composition (SmCo 5 ), was discovered in the mid-1960s, and introduced to the market in 1970. Further improvements led to the development of the second generation, the (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17) family with the nominal commercial composition Sm(Co, Fe, Cu, Zr) 0.74 , which was introduced to the market in 1980 as a PM far superior to pre-existing ones ( [1,2] and references therein). Commercial Sm-Co PMs were produced with a remnant magnetization of ∼60-110 emu g −1 , and high intrinsic coercivity of 4.5-25 kOe. However, due to the high market price of Sm and Co, huge efforts were made to produce a new generation of RE magnets, free of Sm and Co. These efforts culminated in the discovery of the far superior Nd-Fe-B magnet in the 1980s, with remnant magnetization of 137 emu g −1 and intrinsic coercivity of 33 kOe. Nevertheless, the high production cost balanced the reduction of the raw material cost, and the market price of Nd-Fe-B magnets remained relatively high. Further, the low operating temperature (∼100°C) of the first generation of Nd-Fe-B magnets limited their use in motors and generators [3]. Better thermal stability at higher temperatures (∼200°C) was achieved by adding a RE element such as Dy. Unfortunately, the market price of Dy is extremely volatile, which imposes a new challenge to the PM industry. Under these considerations, M-type hexaferrite permanent magnets, although exhibiting lower performance, emerged as the most cost effective materials as extrapolations to the year 2022 indicated [4].
Hexaferrite materials were discovered and characterized in the early 1950s [5][6][7], and received exponentially increasing interest since then due to their practical importance [8,9]. In addition to cost effectiveness, BaM hexaferrites (BaFe 12 O 19 ) are easy to produce with a wide range of tunable magnetic properties [10][11][12], and Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence.
Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. chemical stability required for a wide range of applications [8,[13][14][15][16][17][18][19][20][21]. The magnetic properties of BaM hexaferrites can be effectively modified by different of substitutions for Ba 2+ or Fe 3+ ions. The Sr 2+ substitution for Ba 2+ resulted in improvement of the magnetic properties of M-type hexaferrite ( [11] and references therein). Also, Al 3+ was long recognized as an agent which improves the magnetic properties and limits grain growth, preventing the deterioration of the coercivity at elevated sintering temperatures [22]. The solubility of Al 3+ in the hexaferrite is high, allowing for a wide range of tunable coercivity. Specifically, it was found that it is possible to substitute up to 50% of the Fe 3+ ions in BaM, and 100% in SrM, by Al 3+ ions to produce nanoparticles promising for imaging applications [23]. Further studies confirmed the effectiveness of Al 3+ substitution for Fe 3+ for enhancing the coercivity. Specifically, 6.7% substitution resulted in a 43% enhancement of the coercivity (from 4.2 kOe to 6.0 kOe), and a 45% increase of the magnetocrystalline anisotropy field, but a reduction of the saturation magnetization from ∼72 emu g −1 to ∼50 emu g −1 [24]. Higher levels of Al 3+ substitution resulted in a significant increase of the coercivity and a larger decrease in saturation magnetization [25][26][27].
Partial substitution of RE elements for Ba or Sr was carried out by several investigators in attempts to improve the magnetic properties of M-type hexaferrites [28][29][30][31][32]. Substituting La 3+ for Sr in Sr 1−x La x Fe 12 O 19 hexaferrites improved the saturation magnetization and coercivity up to x=0.3, whereas higher substitution levels resulted in deterioration of the magnetic properties [31]. On the other hand, the substitution of Sr by Pr resulted in an improvement of the coercivity with practically constant saturation magnetization and remanence up to x∼0.11, but a decline of both parameters at x=0.2 [33]. Also, the substitution of different RE elements for Sr in Sr 0.9 RE 0.1 Fe 10 Al 2 O 19 hexaferrites prepared by autocombustion resulted in a significant increase of the coercivity, accompanied with a decrease in saturation magnetization [27]. However, the previous studies did not address the effect of substitution of the high Z end of the RE series. Due to the effectiveness of high-level Al substitution in significantly increasing the coercivity of SrM ferrite, the present study was designed to prepare Sr 1−x RE x Al 2 Fe 12 O 19 (RE=La, Ce, Tb, and Dy; x=0.0 and 0.1) hexaferrites by high energy ball milling, and investigate their structural and magnetic properties, with the objective of examining the effect of preparation method, and extending the study to include Tb and Dy at the far end of the RE series.

Experimental procedure
Ferrite samples of Sr 1−x RE x Al 12 Fe 10 O 19 (RE=La, Ce, Tb and Dy; x=0, 0.1) were prepared from spec pure SrCO 3 , Fe 2 O 3 , Al 2 O 3 and RE 2 O 3 powders (Sigma Aldrich-make). Wet milling in an acetone bath was performed in two zirconia cups, using zirconia balls, with ball-to-powder mass ratio of (12:1). The milling was carried out at a rotational speed of (250 rpm) for 16 h, and the resulting powder was left to dry in air at room temperature. The dry powder was collected, and disk-shaped pellets (approximately 1.25 cm diameter and ∼1 mm thickness) were compacted under a force of 5 tons. The disks were then sintered in air at 1200°C for 6 h in a zirconium oxide crucible.
XRD measurements were made in the angular range 20°2θ70°using Cu-Kα radiation (λ=1.5405 Å) in an XRD 7000-Shimadzo diffractometer. Data analysis included phase identification using X'Pert PRO HighScore software, and structural refinement using FULLPROF software [34]. The size and morphology of the particles in the samples were examined by SEM using FEI-Inspect F50/FEG electron microscope. The magnetic measurements in applied magnetic field strength up to 70 kOe were obtained at room temperature (305 K) and at 5 K using an MPMS 3 SQUID Quantum Design magnetometer. Thermomagnetic measurements, at a constant applied field of 100 Oe, were performed in the temperature range from room temperature to well above the Curie temperature, using a VSM Micromag 3900, Princeton Measurements Corporation vibrating sample magnetometer.

Results and discussion
3.1. XRD analysis Rietveld refinement of the XRD patterns of the Sr 1−x RE x Al 2 Fe 10 O 19 (RE=La, Ce, Tb and Dy; x=0, 0.1) hexaferrites was carried out to explore the crystallographic phases in the samples, and determine the refined structural parameters. Figure 1 shows the refined patterns of the samples with x=0 (SAF), and with La-Al substitution (SLAF), where the black solid spheres represent the experimental data, the red continuous line represents the refined theoretical pattern, and the blue line below the pattern represents the difference curve. The horizontal difference curve with small ripples only at the positions of the Bragg peaks, together with the relatively low values of the R B and R F reliability factors and goodness of fit c , 2 are indications of a reliable fit of the experimental data with the theoretical pattern for SrM hexagonal phase with the standard pattern (JCPDS: 00-033-1340). The Bragg peak positions of the standard phase are indicated by the set of small vertical (green) ticks below the patterns. The absence of any signature of other impurity phases indicated that the substituted Al 3+ ions in these two samples were completely incorporated in replacing Fe 3+ ions in the hexaferrite lattice. The refined lattice parameters and reliability factors are listed in table 1.
Rietveld refinement of the XRD patterns of the remaining samples (figure 2), however, revealed crystallization of a major SrM hexaferrite phase, in addition to traces of α-Fe 2 O 3 (whose Bragg peak positions are indicated by the second set of vertical green ticks) and other minor phases (with Bragg peak positions indicated by the third set of vertical green ticks). Specifically, the third phase is CeO 2 (JCPDS: 00-034-0394) for Ce-Al substitution (SCAF), Tb 3 Fe 5 O 12 (JCPDS: 01-071-0697) for Tb-Al substitution (STAF), and Dy 2 O 3 (JCPDS: 00-018-0475) for Dy-Al substitution (SDAF). The presence of secondary RE oxide phases was also confirmed in Nd-and Pr-substituted SrM hexaferrites, and was attributed to limited solubility of these rare earths in the hexaferrite [35]. The refined lattice parameters (a and c), cell volume (V ), reliability factors (R B and R F ), and the measure of the goodness of fit (c 2 ) of SrM phase in the samples are also listed in table 1. The results indicated   [11,36]. The observed lattice contraction with Al 3+ substitution is attributed to its smaller radius (0.535 Å) compared with that of Fe 3+ (0.645 Å) [37]. Also, the structural parameters decreased with RE substitution, which could be associated with the smaller ionic radii for the RE elements compared with Sr [37]. However, these structural parameters revealed a slight increasing tendency from La-Al→Dy-Al substitution, which is contrary to expectations based on the decrease of ionic radius in the sequence La→Ce→Tb→Dy [37]. This result could therefore be associated with lattice distortions, which increase with the increasing difference between the radius of the RE element and the Sr element. Such distortions were reported to occur when Ba 2+ ions were replaced by the smaller Sr 2+ ions in M-type hexaferrites [36]. The x-ray densities (r x ) for the samples were calculated using the relation: Here M W is the molecular mass, N A is Avogadro's number and V is the unit cell volume. The results listed in table 2 indicated that the x-ray density for all RE-Al substituted samples was 0.6%-1.0% higher than the density of the RE-free sample. The small increase is partially attributed to the increase of the molecular mass with RE substitution (RE elements La→Dy have molecular masses of 138.91 g mol −1 →162.5 g mol −1 , which are higher than 87.62 g mol −1 for Sr), and partially to the decrease of the cell volume. However, the x-ray density exhibited a decreasing tendency with substitution La→Dy, despite the increase of molecular mass (1009.09 g mol −1 →1011.45 g mol −1 ). Here, the % increase of the cell volume (674.4 Å 3 →677.9 Å 3 ) is greater than the % increase of the molecular mass in this direction, resulting in an expected net decrease of the density as suggested by equation (1). The bulk density (r b ) of all samples was measured by Archimedes method, and the results are listed in table 2. The bulk density is usually lower than the theoretical x-ray density due to the porous ceramic structure of the ferrites. The porosity (p) of the sample is defined by the relation: The results in table 2 indicated that the measured bulk density was high, approaching the theoretical density. The low porosity of all samples (<5%) makes it possible to prepare highly dense magnets (with density up to >90% of the theoretical density) from the prepared hexaferrite powders. Further, a tendency of decreasing porosity with RE substitution (except for Tb) was observed, where the lowest porosity of 2.4% was recorded in the case of Dy substitution.
The crystallite size (D) along a given crystallographic direction perpendicular to a reflecting plane (hkl) was calculated using the Stokes-Wilson formula [38]: Here λ is the wavelength of x-ray radiation (1.5406 Å), θ is the position of the peak corresponding to the (hkl) reflection, and β is the integral breadth of the diffraction peak (ratio of the peak area to peak maximum). The integral breadth of the peak was determined by fitting the peak, and then corrected for instrumental broadening using a Si standard sample. The (110) peak at q »  2 30.3 and the (107) peak at q »  2 32.2 were used to probe the crystallite sizes in the direction perpendicular to the corresponding crystallographic planes, and the results are listed in table 3. The results indicated that all samples were nanocrystalline. Also, within experimental uncertainty (±10 nm), the crystallite size in the two directions were similar for all samples except for the STAF sample, where the crystallite size along the basal plane (110) was obviously larger. This suggests that the crystallites in this latter sample grew in platelet-like shapes, whereas in the rest of the samples the crystallites grew almost uniformly in all crystallographic directions.

Magnetic measurements
The hysteresis loops of the Sr 1−x RE x Al 2 Fe 10 O 19 (RE=La, Ce, Tb and Dy; x=0 and 0.1) samples were measured at room temperature (305 K) and at 5 K, in an applied field strength up to 70 kOe. The room temperature hysteresis loops ( figure 4) indicated that all samples are magnetically hard, and full magnetic saturation was not reached even at an applied field of 70 kOe. Therefore, the law of approach to saturation was used to determine the saturation specific magnetization (σ s ) for each sample. However, the remnant specific magnetization (σ r ), and the intrinsic coercivity (H cM ) were obtained directly from the hysteresis loops, and the results are listed in table 4. Also, the B-H curves were constructed from the hysteresis loops, where the magnetic induction is (M is the magnetization in emu/cm 3 ). A representative B-H curve for the sample SAF is shown in figure 5 in the field range from-15 kOe to+15 kOe. Such curves were used to determine the  physical coercivity (H cB ) listed in table 4, and to investigate the linearity of the B-H relation in the second quadrant, which is a critical parameter for the stability of the magnet performance. Figure 6 shows that the B-H relations in the second quadrant are linear, indicating stable permanent magnet performance up to H cB . It is   worth mentioning that the sudden drop in the magnetization of the sample STAF when a small reverse field is applied could be associated with the presence of the soft garnet phase in this sample, as confirmed by the XRD analysis.
The room temperature magnetic data listed in table 4 revealed superior magnetic properties of the SAF, SLAF, and SCAF samples compared with samples prepared by the sol-gel method and sintered at 1100°C [27]. Further, the magnetic parameters for the STAF and SDAF samples were similar to the highest parameters reported for other RE substituted SrAl 2 Fe 10 O 19 samples [27]. However, in a recent study [40], it was shown that the Ca-Al substitution in SrM hexaferrites was more effective, were a coercivity of 21.3 kOe, close to the higher end of the coercivity of the Sm-Co magnets, was reported for Sr 0.67 Ca 0.33 Fe 8 Al 4 O 19 compound. The Ca-Al substituted ferrites, in spite of the rather low saturation magnetization they exhibited, demonstrated potential for the production of low cost materials with very high coercivity and sub-terahertz natural ferromagnetic resonance [40].
The saturation magnetization of the present samples, as well as that reported for other Al-substituted M-type hexaferrites, was lower than the value of ∼70 emu g −1 reported for the unsubstituted SrM hexaferrite. The observed decrease could be associated mainly with two factors, namely, the preferential substitution of Al 3+ ions at spin-up Fe 3+ sites, and spin canting due to the reduction of the superexchange interactions between spin-up and spin-down sublattices [41]. The reduction of the saturation magnetization due to partial RE substitution for Sr in M-type hexaferrites was also reported by others [30,42,43].
Further, the hysteresis loops for all samples were measured at 5 K and compared with the hysteresis loops at 305 K. Representative curves for the SAF and STAF samples are shown in figure 7. These loops exhibited the usual higher magnetization at 5 K, and the decrease in the coercivity, which is a characteristic feature of the M-type hexagonal ferrite. The afore mentioned sharp drop in the magnetization of the STAF sample under a small applied reverse field is shown in figure 7(b). No other sample exhibited such behavior, thus confirming the structural results which revealed the presence of a secondary garnet phase (which is magnetically soft), only in the STAF sample.
At 5 K, the squareness ratio ( / s s s = rs r s ) for all samples was very close to 0.5, which is characteristic of single magnetic domain uniaxial particles with random orientations of the domain magnetic moments. This result is consistent with the measured crystallite size of 103 nm, and the SEM images which indicated that the particle size distribution in all samples is dominated by sizes below the critical single domain size of ∼0.5 μm for hexaferrites. Also, the low-temperature squareness ratios indicated that all samples exhibited a collinear spin structure at 5 K, where the remnant magnetization is practically half the saturation magnetization. The squareness ratio at 305 K, however, exhibited a lower value for all samples as the results in table 5 demonstrated. This discrepancy can be associated with a canted spin structure at room temperature. We attribute the reduction of the squareness ratio at 305 K to the reduction of the remnant magnetization due to spin canting in the absence on an applied magnetic field. In contrast, the magnetic state at the maximum applied field of 70 kOe is almost collinear.

Thermomagnetic measurements
Measurements of the temperature dependence of the magnetization at a fixed applied field can be used to determine the critical transition temperature (Curie temperature) of a magnetic material. The thermomagnetic curves for the Sr 1−x RE x Al 2 Fe 10 O 19 samples are shown in figure 8. The curves exhibited a monotonic decrease of the magnetization with the increase of temperature, and then peaked just before the Curie temperature. The peak preceding the magnetic phase transition (Hopkinson peak) indicates the existence of small hexaferrite particles with superparamagnetic behavior just below the Curie temperature [44]. The Curie temperature was evaluated by extrapolating the thermomagnetic curve to zero magnetic moment, and the results are listed in table 6. Although the Curie temperature of all samples decreased significantly in comparison with the un-doped SrFe 12 O 19 , the values recorded for these compounds are still higher than the best NdFeB commercial magnets. The observed decrease in T c is attributed to the decrease of the superexchange interaction between iron magnetic sublattices as a consequence of magnetic dilution caused by Al 3+ ions replacing Fe 3+ ions.

Conclusions
High purity SrM hexaferrites with partial substitution of Fe by Al, and Sr by RE were successfully prepared by ball milling and sintering at 1200°C. The values of the saturation magnetization and remnant magnetization of all samples, and the significant increase of more than 200% of the coercivity (H cM ) compared with the unsubstituted hexaferrite, render the present ferrites potentially important for permanent magnet applications. The low temperature and room temperature magnetic data revealed a collinear magnetic structure at 5 K, and a canted spin structure at room temperature. The preferred site selectivity of Al 3+ ions for spin-up Fe 3+ sites resulted in a reduction of the saturation magnetization, while the partial RE substitution generally resulted in an enhancement of the coercivity (H cM ), but a small reduction (<10%) of the Curie temperature. The present hexaferrites are therefore suitable for cheap ceramic-based magnets with enhanced coercivity, and relatively high operating temperatures. The performance of these ferrites as permanent magnets might be further improved by enhancing the remnant and saturation magnetization, without significantly reducing the coercivity. This possibility could be tested by adopting modified cationic substitutions, or by incorporating these hexaferrites in special composites with desired properties, which will be the focus of a future research work.

Acknowledgments
One of the authors (S H Mahmood) would like to acknowledge the financial support provided by the University of Jordan during his sabbatical leave at Michigan State University, where the high-field magnetic measurements were made, and this work was completed. Also, the support provided by Prof Jack Bass and Prof Norman Birge (Michigan State University), and the critical reading of the manuscript and suggestions made by them to improve it are highly appreciated. The technical assistance of Y Abu Salha and W Fares (The University of Jordan) is also acknowledged.