Loading Effect of Sol-Gel Derived Barium Hexaferrite on Magnetic Polymer Composites

Solution–processing methods were investigated as viable alternatives to produce the polymer-bonded barium hexaferrite (BaM). BaM powders were first synthesized by using the sol-gel auto-combustion method. While the ignition period in two synthesis batches varied, the morphology of hexagonal microplates and nanorods, as well as magnetic properties, were reproduced. To prepare magnetic polymer composites, these BaM powders were then incorporated into the acrylonitrile-butadiene-styrene (ABS) matrix with a weight ratio of 80:20, 70:30, and 60:40 by using the solution casting method. Magnetizations were linearly decreased with a reduction in ferrite loading. Compared to the BaM loose powders and pressed pellet, both remanent and saturation magnetizations were lower and gave rise to comparable values of the squareness. The squareness around 0.5 of BaM samples and their composites revealed the isotropic alignment. Interestingly, the coercivity was significantly increased from 1727–1776 Oe in loose BaM powders to 1874–2052 Oe for the BaM-ABS composites. These composites have potential to be implemented in the additive manufacturing of rare-earth-free magnets.

The BaM phase formation, morphology, and magnetic properties are also sensitive to the heat treatment on sol-gel derived products. Widyastuti et al. reported that the α-Fe 2 O 3 phase dominated after calcining at 750 • C and 850 • C, whereas the BaM was dominant as a result of higher temperature calcination at 950 • C. Accordingly, the highest magnetization of 63.5 emu/g and 3542 Oe were obtained after sintering at 950 • C as the particles grew from the average size around 0.2 to 0.7 µm with increasing sintering temperatures [23]. Wang and Zhang suggested that higher temperature and longer holding times promoted crystal grain growth. By sintering at 1000 • C for 5 h, hexagonal microplates of BaM with a grain size of around 1-2 µm were obtained. The Differential Scanning Calorimetry (DSC) revealed two exothermic peaks at about 380 • C related to the decomposition of remaining esters and 850 • C attributed to the formation of BaM [24]. Mali and Ataie showed that the amorphous dried gel transformed to the crystalline powder after the combustion, and the pure BaM phase was obtained after calcination at 900 • C. Besides, the increase in sintering temperature from 1000 • C to 1100 • C resulted in the agglomeration of spherical particles of 0.4 µm in diameter into the hexagonal microplates with an average size of around 2.5 µm [25].
In this research, BaM-ABS composites are fabricated for further uses in the additive manufacturing of rare-earth-free magnets. Following requirements in industrial productions, the reproducibility of magnetic properties of BaM from the sol-gel auto-combustion is firstly investigated. The solution-processing method is also used to incorporate sol-gel derived BaM powders in ABS granules. While the largest ferrite loading straightforwardly leads to the largest magnetization, the homogeneity and mechanical properties need to be considered [9,30]. The incorporation of magnetic powders affects the viscosity of the polymeric matrix, which limits the working in additive manufacturing and the loadings of 45-65% are typically used in the case of magnetic thermoplastic filaments [29]. The variations of BaM from 60% to 80% by weight are selected for this research because, from the preliminary test, these loadings led to printable magnetic polymer composites.

Synthesis and Characterization of BaM Powders
In the sol-gel auto-combustion synthesis of BaM, the molar composition of iron(III) nitrate nonahydrate (Fe(NO 3 ) 3 ·9H 2 O) (Sigma-Aldrich, Singapore: ACS reagent, 98.0-101.0%), barium nitrate (Ba(NO 3 ) 2 ) (HIMEDIA, India: ACS 99.00-102.00%) and citric acid (C 3 H 5 O(COOH) 3 was 12:1:0.75. Following the suitable condition for forming a singlephase BaM in the previous reports [17,19,23,24], the mole ratio of Ba to Fe in the precursors was selected as 12:1. The citric acid acted as a chelating agent in the bonding of anions (citrate (3−) and citrate (2−) ) to the Fe 3+ and Ba 2+ cations. The starting reagents were successively dissolved in deionized (DI) water with a mass (g) per volume (mL) ratio of 1:6. Under constant magnetic stirring, the aqueous acidic solution was obtained. The pH was adjusted to 7 by adding ammonium hydroxide (27% NH 4 OH). The solution was then heated and continuously stirred at 90-100 • C for around 3 h to evaporate the solvents. The viscous brown gel was formed and eventually dried. Afterwards, the temperature was promptly increased to 150 • C. As a result, the dried gel swelled up until reaching ignition. The auto-combustion took place and propagated throughout the gel within a minute. The obtained brown powders were ground and then heated at 450 • C for 2 h to remove residual carbon compounds [13,20]. The subsequent calcination at 1050 • C for 3 h finally allowed for the formation of BaM [10,20]. The changes from the solution to the final solid product are shown in Figure 1. To verify the reproducibility of the sol-gel auto-combustion, two batches of BaM, referred to as BaM_ex1 and BaM_ex2, were prepared by using the same procedure. It was noted that the combustion step might have occurred before the gel was completely dried. As a result of the wet gel ignition, the ignition period of 1 min for the BaM_ex2 sample was much longer than the 20 s observed in the BaM_ex1 sample. Morphology and elemental compositions of both samples were, respectively, examined by the field-emission scanning electron microscope (FESEM: Zeiss Merlin Compact, Carl Zeiss Microscopy GmbH, Germany) and the energy-dispersive X-ray spectroscopy (EDS: Oxford Aztec, Oxford Instruments, UK). The ferrite phases were identified by using a single crystal X-ray diffractometer (XRD, Rigaku SuperNova diffractometer with a HyPix 3000 detector, Rigaku Corporation, Poland) using Cuα radiation (λ = 1.54184 Å). Data reduction, scaling and absorption corrections were performed using CrysAlisPro software in the micro powder mode. A vibrating sample magnetometer (VSM: in-house developed and calibrated with Lakeshore 730,908 using a Ni sphere of 3 mm in diameter) was used to measure the magnetization (M) as a function of the external magnetic field (H) between −17,500 and 17,500 Oe.

Synthesis and Characterization of BaM-ABS Composites
To prepare the magnetic polymer composites for further uses in additive manufacturing of rare-earth-free magnets, the sol-gel derived BaM powders were incorporated into the ABS matrix via the solution casting method. Firstly, the ABS pellets (GA800, Polimaxx, IRPC Public Company Limited, Thailand) were completely dissolved in acetone under constant stirring with a magnetic bar for 2 h at room temperature. Then, the BaM powders were added to the solution, and the stirring continued for 1 h to thoroughly disperse the BaM in the ABS. 20 mL acetone was required for 2.5 g of BaM and ABS in total. After that, the residual acetone was eliminated by evaporation. The product was dried at room temperature for 24 h, resulting in the BaM in the ABS matrix. The weight ratios of BaM:ABS varied as 60:40, 70: 30, 80:20, and magnetic properties of these composites, respectively referred to as BaM60-ABS40, BaM70-ABS30, and BaM80-ABS20, were measured by the VSM. In addition, a gram of BaM powder was pressed under a uniaxial pressure at 60 kg/cm 3 for 5 min with an automatic hydraulic machine to compare the magnetic properties of this BaM pellet with those of the BaM-ABS composites. From hysteresis loops, the coercivity (H c ) and the remanent magnetization (M r ) were, respectively, determined from the x-intercept and the y-intercept.

Characterization of BaM Powders
The sol-gel derived products after calcining at 1050 • C for 3 h revealed the morphology, as shown in Figure 2a,c. Both BaM_ex1 and BaM_ex2 samples are similarly composed of hexagonal plates and nanorods. The size of hexagonal plates ranges from less than 0.3 µm to over 1 µm. Compared to the previous reports on BaM, Mali and Ataie obtained hexagonal plates with an average size of about 2.5 µm after the calcination at 1100 • C for 1 h [25] and Wang and Zhang observed the hexagonal microplate with a grain size of 1-2 µm from the powder received heat treatment at 1000 • C for 5 h [24]. The EDS spectra in Figure 2b,d reveal peaks corresponding to Fe, Ba, O, and C elements. Since an artificial carbon peak is normally visible on EDS spectra, these spectra are consistent with the composition of BaM. However, measurements at some other points do not yield a clear Ba peak, indicating inhomogeneity. The phase composition is therefore not conclusive from the EDS results and other minor phases cannot be ruled out. The calcination temperature at 1050 • C tends to promote the pure phase BaFe 12 O 19 formation [28], and the appropriate high temperature and holding time for calcination significantly affect the crystalline structure of BaM [24,25]. While the DSC, differential thermal analysis (DTA) and thermogravimetric analysis (TGA) spectra revealed the BaM phase formation at 840-860 • C [24][25][26], the calcinations at higher temperatures were reported to promote the hexagonal structures as well as the crystalline size of BaM [17,19,25].
The phase implied by the EDS results is confirmed by XRD. In Figure 3, the spectra from both BaM_ex1 and BaM_ex2 samples exhibit characteristic peaks of the M-type  Figure 4 shows similar hysteresis loops from BaM_ex1 and BaM_ex2 samples indicating the reproducibility required for productions of hard magnetic BaM. The solutionprocessing methods can be scaled up and some ferrite syntheses have been developed at lower temperatures for commercial productions [43]. The maximum magnetizations reach 62.4 and 63.6 emu/g in the applied external magnetic field of 17,500 Oe. The remanent magnetizations are 31.1 and 32.2 emu/g, as listed in Table 1. These magnetizations, as well as the coercivity of about 1750 Oe in Table 1, are functional characteristics of hard magnetic materials [30]. Magnetic properties can be further improved by the varying condition of heat treatments that influences phase purification, crystalline size, and morphology [17,19,25,26].

Magnetic Properties of BaM-ABS Composites
Hysteresis loops of BaM-ABS composites are compared to the pressed BaM pellet in Figure 5 and the corresponding magnetic parameters are listed in Table 1. Magnetizations of pressed BaM pellets are almost identical to the value measured in the form of the BaM powder with the remanent magnetization around 32 emu/g. As the magnetization of magnetic polymer composites is only attributed to magnetic filler, the reduction in magnetizations is consistent with the replacement of magnetic materials by the polymeric matrix. Nevertheless, the highest magnetization obtained from the BaM:ABS ratio of 80:20 is as high as 87% of the values in BaM loose powders and pressed pellet. Unlike magnetizations, the coercivity of the pressed BaM pellet is higher than that of the powder form. The powder compaction induced the stress that influenced the dislocations and associated the lattice distortion, enhancing the coercivity [44][45][46]. Likewise, the polymer-bonded BaM magnets have higher coercivity because of the induced stress [29]. These results are consistent with the reported by El-Sayed et al. on a marked increase in coercivity of the BaM composite in polystyrene [47]. The coercivity differs slightly with the ferrite loading and the BaM:ABS ratio of 70:30 results in the highest coercivity of about 2050 Oe. It follows that the maximum energy product compared in Table 1 is at the highest in the case of this ratio.
As all hysteresis loops in Figure 5 do not reach the saturation in the maximum field supplied, the law of approach to saturation is then used to estimate the saturation magnetizations. With the contribution of magneto-crystalline anisotropy, the change in magnetization at high magnetic fields near the saturation of hard ferrites follows Equation (1) [48][49][50][51][52][53][54]: where B is a constant. Rearranging the terms leads to a linear form in Equation (2).
The plots between M and 1/H 2 in a regime approaching 17,500 Oe, the maximum applied field, are shown in Figure 6. All graphs can be linearly fitted, consistent with Equation (2). The saturation magnetization (M s ) is then approximated from the y-intercept as listed in Table 2. The slope equals the B value, which is related to the effective anisotropy constant (K eff ) for uniaxial magnetic hexagonal crystals as in Equation (3) [48][49][50][51][52][53][54].
In Table 2, the saturation magnetizations from the linear fitting are comparable to the literature [20,26,27,[55][56][57]. From these experiments, the saturation magnetization from undoped BaM ranges from 49-70 emu/g. Remarkably high magnetizations of 88 and 89 emu/g in BaM were, respectively, obtained from the solid state reaction [58] and ball milling [59]. The K eff , also listed in Table 2, is significantly reduced with the decreased in ferrite loading. However, the reduction in M s results in the increase in H a with the decreasing loading from 100% (i.e., a pressed pellet) to 70%. However, the squareness computed from the ratio M r /M s is rather insensitive to the ferrite loading. The value around 0.5 of both BaM and BaM-ABS composites corresponds to the isotropic BaM, commonly found in randomly oriented bulk ferrite magnets [60]. The anisotropic BaM can be produced by heat treatment in the magnetic field. The magnetically oriented grains lead to the squareness close to one along the c-axis [10].
In Figure 7, both remanent and saturation magnetizations exhibit linear decreases with the decreasing BaM loading in the composites. Such variations follow the rule of mixture, which was previously applicable to the magnetization of composites with ferrite fillers in either plastic [61] or rubber matrix [62]. In cobalt ferrite-polypropylene composites, the loading of 5-45 wt.% gave rise to a graph fitting passing through the origin [61]. However, the projections of trend lines in Figure 7 predict non-zero remanent and saturation magnetizations of, respectively, 1.66 and 7.73 emu/g at the zero magnetic loading. In the case of remanent magnetization, the R 2 from the linear fitting is higher and the discrepancy from the line projection is smaller because the values are directly obtained of the y-intercept of hysteresis loops. The uncertainty is increased for the saturation magnetization calculated from the law of approach to saturation.

Conclusions
The M-type BaM were successfully synthesized by the sol-gel auto-combustion. By heating and stirring at 90-100 • C for approximately 3 h, the solution of nitrate nonahydrate, barium nitrate and citric acid transformed into viscous and eventually dried gel. By increasing the temperature to 150 • C, auto-combustion arose and was then terminated within a minute. Despite the variation in the ignition period, magnetic hysteresis loops of BaM from two synthesis batches are almost identical. The results indicated the potential of reproducibility in the sol-gel auto-combustion synthesis for commercial productions. To fabricate magnetic polymer composites, the sol-gel derived BaM was dispersed in the ABS matrix via the solution casting method. The weight ratio of BaM and ABS was varied as 80:20, 70:30, and 60:40 to examine the optimum ferrite filler loading. The excessive loading of BaM may severely deteriorate the mechanic properties desirable in the additive manufacturing, but the loading up to 80% in this research still gave rise to printable composites. All the composites exhibited significant increases in coercivity, while the magnetization values linearly decreased with the reduction in ferrite loading. The highest coercivity of about 2050 Oe was obtained in the BaM-ABS composite with a weight ratio of 70:30. The squareness of every composite was approximately 0.5, indicating the randomly oriented BaM in the ABS matrix.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.