Hybrid Phosphate-Alumina Iron-Based Core-Shell Soft Magnetic Composites Fabricated by Sol-Gel Method and Ball Milling Method

: Novel Fe-based soft magnetic composites (SMCs) with hybrid phosphate-alumina layers were prepared by both sol–gel and ball-milling methods. The e ﬀ ects of the fabrication methods and the addition of Al 2 O 3 particles on the microstructure and the soft magnetic performance of SMCs were studied. The formation of the hybrid phosphate-Al 2 O 3 shell not only leads to the decrease of the total core loss, but also results in the reduction of the permeability and saturation magnetization. However, the degree of decrease caused by the di ﬀ erent methods were not identical. The sample with 8% Al 2 O 3 prepared by the sol–gel method showed the best magnetic performance, exhibiting a high-amplitude permeability ( µ a ) of 85.14 and a low total core loss ( P s) of 202.3 W / kg at 50 mT and 100 kHz. The hysteresis loss factor and the eddy current loss factor were obtained by loss separation. The results showed that the samples with the same Al 2 O 3 content prepared by di ﬀ erent methods exhibited almost the same total core loss. However, the contribution of the hysteresis loss and the eddy current loss showed an obvious di ﬀ erence in behavior because of the change of the particle shapes and the reﬁnement of the particle size during the ball-milling process. by the ball-milling method is greater than that prepared by the sol–gel method. These results suggest that the hybrid phosphate–alumina coating is promising, and can be used to obtain high-performance SMCs for electromagnetic devices at high frequency.


Introduction
The eddy currents which appear in ferromagnetic materials are caused by the magnetization in an alternating current (AC) magnetic field as a consequence of Faraday's law. The ferromagnetic materials that were magnetized in an AC magnetic field would internally-generate the eddy current, a long-existing barrier that limited the applications of the devices at medium to high frequency because of the increase in eddy current loss and the reduction of the permeability [1,2].
Owing to their superior properties, including high permeability, high-frequency stability, 3D isotropy magnetic field, high productivity and, especially, the high electrical resistivity, iron-based soft magnetic composites (SMCs) with a special core-shell structure [3] have attracted broad academic interest and investigations in order to satisfy the requirements of applications in power electronics, transformers and electric motors at medium to high frequency [4,5]. The iron-based SMCs are conventionally prepared by coating insulating materials onto the pure iron matrix to block the path of closed-current loops and the compaction and annealing process [6]. Therefore, the selection of appropriate insulating materials with high electric resistivity and good adhesive strength is vital to minimize the eddy current loss of the SMC cores [7].
The coating layers can be divided into two principal types: organic materials and inorganic materials. Organic materials such as silicon resin [8] and epoxy [9] are the most popular insulating

Preparation and Characterization
Gas-atomized iron powders with a particle size ranging from 75 µm to 250 µm were conducted to achieve the desired size ratio, annealed at 700 • C for some time, twice cleaned in absolute ethyl alcohol, and dried at 60 • C for five hours before the passivation. Then, 50 g iron powders were mixed with a certain amount of ethanol-diluted orthophosphate acid with a concentration of 0.01 g/mL, and the mixture were stirred in a solution of 65 • C until the solvent completely evaporated to obtain the phosphate-coated particles.
The sol-gel method and the ball-milling method were used to prepare Fe @ phosphate particles to obtain powders with the hybrid coating. A certain amount of aluminum oxides were dissolved in 100 mL of ethanol and was magnetically stirred at room temperature for 1 h to obtain the aluminum sol. Then, some as-prepared Fe/phosphate powders were dispersed in the solution using a spiral mixer for 15 min. A mixture of distilled water and ethanol (1:3 (V/V)) was added dropwise, acetic acid was added to adjust the pH value to around four, and the mixture was continuously stirred for 1.5 h. Subsequently, the suspension was separated and washed with ethanol three times and dried. Finally, the powders were calcinated at 600 • C for 2 h under a nitrogen atmosphere.
As for the ball-milling method, the weight ratio of ball to powders was set to 10:1 and a mixture of the as-prepared Fe/phosphate powders and the α-Al 2 O 3 nanoparticles with the weight percent of 2 wt%-8 wt% were carried out at a speed of 200 rpm for 30 min. All of the above powders were mixed with the lubricant of Zinc stearate (0.5 wt%) and were pressed at an uniaxial pressure of 800 MPa at room temperature to obtain the ring-shaped compacts with an external diameter of 20 mm, an inside diameter of 12 mm and height of approximately 3.4 mm.
An X-ray diffractometer using Cu-Kα radiation was used to analyze the phases of the samples. The morphology and composition of the powders were detected by a scanning electron microscopy (SEM, Nova NanoSEM 450, FEI, USA) equipped with an energy-dispersive X-ray Spectroscope (EDS, Ultra, EDAX, USA), whose resolutions were 1.0 nm at 15 kV and 0.8 nm at 30 kV. All of the SEM images were magnified 1000 times and a metallographic analyzer produced by a Shanghai optical instruments factory was used to characterize the microstructure. The soft magnetic properties of the ring-shaped compacts were determined by a hysteresis looper (MATS 2010SA) at a frequency ranging from 5 kHz to 100 kHz and a magnetic excitation level of 0.05 T; the uncertainty of measured µ a and total loss were less than 3% and 5%, respectively. The vibrating sample magnetometer (VSM) was conducted to measure the saturation magnetization of the compacts; the accuracy and repeatability were less than 1% and 5%, respectively. Figure 1 shows the SEM images of the raw iron powders, the phosphated powders, and the EDS (energy disperse spectrum) analysis and maps of the phosphated powders. Results suggest that the coated powders keep the same irregular shape and only show a rougher surface than the raw iron powders. Figure 1c, which displays the EDS analysis of the selected area of the phosphated iron particle, indicates the existence of Fe, O and P on the surface of the phosphated particles. In addition, the EDS maps show that the iron, oxygen and phosphate elements distribute on the surface of the coated powders. Therefore, it can be concluded that the phosphate insulating layers were formed during the phosphating process. 2 wt%-8 wt% were carried out at a speed of 200 rpm for 30 min. All of the above powders were mixed with the lubricant of Zinc stearate (0.5 wt%) and were pressed at an uniaxial pressure of 800 MPa at room temperature to obtain the ring-shaped compacts with an external diameter of 20 mm, an inside diameter of 12 mm and height of approximately 3.4 mm. An X-ray diffractometer using Cu-Kα radiation was used to analyze the phases of the samples. The morphology and composition of the powders were detected by a scanning electron microscopy (SEM, Nova NanoSEM 450, FEI, USA) equipped with an energy-dispersive X-ray Spectroscope (EDS, Ultra, EDAX, USA), whose resolutions were 1.0 nm at 15 kV and 0.8 nm at 30 kV. All of the SEM images were magnified 1000 times and a metallographic analyzer produced by a Shanghai optical instruments factory was used to characterize the microstructure. The soft magnetic properties of the ring-shaped compacts were determined by a hysteresis looper (MATS 2010SA) at a frequency ranging from 5 kHz to 100 kHz and a magnetic excitation level of 0.05 T; the uncertainty of measured μa and total loss were less than 3% and 5%, respectively. The vibrating sample magnetometer (VSM) was conducted to measure the saturation magnetization of the compacts; the accuracy and repeatability were less than 1% and 5%, respectively.  Figure 1 shows the SEM images of the raw iron powders, the phosphated powders, and the EDS (energy disperse spectrum) analysis and maps of the phosphated powders. Results suggest that the coated powders keep the same irregular shape and only show a rougher surface than the raw iron powders. Figure 1c, which displays the EDS analysis of the selected area of the phosphated iron particle, indicates the existence of Fe, O and P on the surface of the phosphated particles. In addition, the EDS maps show that the iron, oxygen and phosphate elements distribute on the surface of the coated powders. Therefore, it can be concluded that the phosphate insulating layers were formed  2 reveals the influence of the phosphating on the permeability and the total core loss of the compacts at 50 mT external magnetic field. The permeability of the as-coated sample is 127.7 at 5 kHz and is not as high as that of the pure iron sample due to the magnetic dilution effect of the non-magnetic materials [19]. The presence of non-magnetic phosphate between the magnetic cores plays the same role as air gaps, which would create a demagnetizing field in the entire magnetic circuit, resulting in lower permeability at low frequency. However, owing to the greater frequency stability, the permeability of the coated sample is superior to that of the pure iron sample at 100 kHz. At high frequency, the phosphate with strong resistivity leads to a lower eddy current. The increased eddy current caused by the high-frequency AC field in the pure iron sample circulates in the opposite direction to the applied field, which results in the decrease of the excited magnetic field and permeability [17,20].  Figure 2 reveals the influence of the phosphating on the permeability and the total core loss of the compacts at 50 mT external magnetic field. The permeability of the as-coated sample is 127.7 at 5 kHz and is not as high as that of the pure iron sample due to the magnetic dilution effect of the nonmagnetic materials [19]. The presence of non-magnetic phosphate between the magnetic cores plays the same role as air gaps, which would create a demagnetizing field in the entire magnetic circuit, resulting in lower permeability at low frequency. However, owing to the greater frequency stability, the permeability of the coated sample is superior to that of the pure iron sample at 100 kHz. At high frequency, the phosphate with strong resistivity leads to a lower eddy current. The increased eddy current caused by the high-frequency AC field in the pure iron sample circulates in the opposite direction to the applied field, which results in the decrease of the excited magnetic field and permeability [17,20].

SMCs with Single Phosphate Coating
The total core loss of the as-coated sample (shown in Figure 2b) is approximately two thirds of the pure iron sample at 100 kHz. This can be explained by the blocking effect of the phosphate insulation layer on the inter-particle eddy current loss, which is the main contributor to the total core loss at high frequency. The total core loss of the as-coated sample (shown in Figure 2b) is approximately two thirds of the pure iron sample at 100 kHz. This can be explained by the blocking effect of the phosphate insulation layer on the inter-particle eddy current loss, which is the main contributor to the total core loss at high frequency.

SMCs with Hybrid Phosphate-Alumina Coating
The XRD patterns of the pure iron particles and the hybrid-coated particles are displayed in Figure 3. The diffraction peaks at 2θ = 44.667 • , 65.000 • and 82.366 • correspond to the (110), (200) and (211) planes of the α-Fe phase (JCPDS 06-0696). The XRD patterns of the coated powders are not obviously different from those of the original. Only the diffraction peak intensity of the α-Fe phases of the hybrid-coated powders via the ball-milling method decreases slightly and no other diffraction peaks could be observed. This could be explained primarily by the possibility that the content of the phosphate layer is too small to be detected, or that the phosphate layer is amorphous [18]. Previous research has reported that amorphous Al 2 O 3 would transfer to δ-Al 2 O 3 , θ-Al 2 O 3 and α-Al 2 O 3 at 750 • C, 1000 • C, 1100 • C, respectively [21]. Therefore, preparing the Al 2 O 3 particles by the sol-gel method and calcining at 600 • C could only obtain amorphous Al 2 O 3 . As for the ball-milling method, the broadening of the α-Fe peak may be due to the small crystallites and their lattice distortions [22]. The grain refinement and the plastic deformation during the ball-milling process would also decrease the intensity of the diffraction peaks.
The total core loss of the as-coated sample (shown in Figure 2b) is approximately two thirds of the pure iron sample at 100 kHz. This can be explained by the blocking effect of the phosphate insulation layer on the inter-particle eddy current loss, which is the main contributor to the total core loss at high frequency. SEM images and EDS maps of the hybrid-coated powders via ball milling are exhibited in Figure 4. A change of the particle shape from the irregular shape to a flake shape after the ball-milling process could be observed in Figures 4a and 2a. Traces of aluminum, oxygen and phosphorus are uniformly detected on the particle surface in the EDS maps (Figure 4b-d), suggesting the formation of the core-shell structured Fe @ phosphate @ Al 2 O 3 SMCs. The XRD patterns of the pure iron particles and the hybrid-coated particles are displayed in Figure 3. The diffraction peaks at 2θ = 44.667°, 65.000° and 82.366° correspond to the (110), (200) and (211) planes of the α-Fe phase (JCPDS 06-0696). The XRD patterns of the coated powders are not obviously different from those of the original. Only the diffraction peak intensity of the α-Fe phases of the hybrid-coated powders via the ball-milling method decreases slightly and no other diffraction peaks could be observed. This could be explained primarily by the possibility that the content of the phosphate layer is too small to be detected, or that the phosphate layer is amorphous [18]. Previous research has reported that amorphous Al2O3 would transfer to δ-Al2O3, θ-Al2O3 and α-Al2O3 at 750 °C, 1000 °C, 1100 °C, respectively [21]. Therefore, preparing the Al2O3 particles by the sol-gel method and calcining at 600 °C could only obtain amorphous Al2O3. As for the ball-milling method, the broadening of the α-Fe peak may be due to the small crystallites and their lattice distortions [22]. The grain refinement and the plastic deformation during the ball-milling process would also decrease the intensity of the diffraction peaks. SEM images and EDS maps of the hybrid-coated powders via ball milling are exhibited in Figure  4. A change of the particle shape from the irregular shape to a flake shape after the ball-milling process could be observed in Figure 4a and Figure 2a. Traces of aluminum, oxygen and phosphorus are uniformly detected on the particle surface in the EDS maps (Figure 4b-d), suggesting the formation of the core-shell structured Fe @ phosphate @ Al2O3 SMCs.  Figure 5a-d indicates that increasing the Al2O3 content results in thicker and more uniform insulating layers. No obvious insulating layers could be seen with the addition of 2% Al2O3. Insulating layers were evident with additional amounts of 6% Al2O3, but some parts of the matrix particles still contacted directly. Further increasing the additional amount of Al2O3 to 8% leads to complete insulating layers. However, more pores could be observed with the increasing Al2O3 content. To further analyze the microstructure: the SEM (scanning electron microscopy) images of the polished surface of the Fe @ phosphate @ Al2O3 SMC compact were characterized. In addition, the EDS maps were shown to verify the microstructure ( Figure 6). The iron matrix particles were firmly encapsulated by a thin coating with a core-shell structure via the ball-milling method. the fact that the P and Al elements only distributed around the matrix powder and the Fe element distributed in To further analyze the microstructure: the SEM (scanning electron microscopy) images of the polished surface of the Fe @ phosphate @ Al 2 O 3 SMC compact were characterized. In addition, the EDS maps were shown to verify the microstructure ( Figure 6). The iron matrix particles were firmly encapsulated by a thin coating with a core-shell structure via the ball-milling method. the fact that the P and Al elements only distributed around the matrix powder and the Fe element distributed in almost the whole visual area could strongly prove the nature of the composite interface microstructure. Figure 5a-d indicates that increasing the Al2O3 content results in thicker and more uniform insulating layers. No obvious insulating layers could be seen with the addition of 2% Al2O3. Insulating layers were evident with additional amounts of 6% Al2O3, but some parts of the matrix particles still contacted directly. Further increasing the additional amount of Al2O3 to 8% leads to complete insulating layers. However, more pores could be observed with the increasing Al2O3 content. To further analyze the microstructure: the SEM (scanning electron microscopy) images of the polished surface of the Fe @ phosphate @ Al2O3 SMC compact were characterized. In addition, the EDS maps were shown to verify the microstructure ( Figure 6). The iron matrix particles were firmly encapsulated by a thin coating with a core-shell structure via the ball-milling method. the fact that the P and Al elements only distributed around the matrix powder and the Fe element distributed in  accepted that the saturated magnetization mainly depends on the composition and the purity of the materials [1], and that introducing non-magnetic phosphate and α-Al 2 O 3 would dilute the percentage of magnetic phases and reduce the Ms as a result. The decreasing trend of the saturated magnetization in Figure 7 is due to the increasing content of Al 2 O 3 in insulating layers.   Figure 7 reveals the hysteresis loops of the matrix iron particles and the hybrid-coated particles prepared by the ball-milling method with the various additional amounts of Al2O3. The saturated magnetization of raw pure iron is approximately 205 emu/g. All the hybrid-coated powders display excellent properties of saturated magnetization of approximately 160 emu/g to 190 emu/g. It is widely accepted that the saturated magnetization mainly depends on the composition and the purity of the materials [1], and that introducing non-magnetic phosphate and α-Al2O3 would dilute the percentage of magnetic phases and reduce the Ms as a result. The decreasing trend of the saturated magnetization in Figure 7 is due to the increasing content of Al2O3 in insulating layers.   (Figure 8b) with the same additional amount of Al2O3. The amplitude permeability of the hybrid-coated SMCs prepared by both methods exhibit excellent frequency stability at the measured frequency, which could be attributed to the better integrity of the insulation layers shown in Figure 5. Increasing the content of Al2O3 causes a tendency of permeability decrement, and the permeability of the samples with 8% Al2O3 reach 77.24 (ball milling method) and 85.14 (sol-gel method), respectively, at 100 kHz, which are ~82.4% and   (Figure 8b) with the same additional amount of Al 2 O 3 . The amplitude permeability of the hybrid-coated SMCs prepared by both methods exhibit excellent frequency stability at the measured frequency, which could be attributed to the better integrity of the insulation layers shown in Figure 5. Increasing the content of Al 2 O 3 causes a tendency of permeability decrement, and the permeability of the samples with 8% Al 2 O 3 reach 77.24 (ball milling method) and 85.14 (sol-gel method), respectively, at 100 kHz, which are~82.4% and~90% of that of phosphate-coated powders (93.7). The results could be explained by the following reasons. On the one hand, the permeability has a positive correlation with the Ms and the magnetism would be diluted by the non-magnetic Al 2 O 3, as shown in Figure 6. On the other hand, a higher porosity with an increasing Al 2 O 3 content could be concluded from Figure 5. The decreasing magnetic permeability is usually attributed to the inter-particle gap effect, as a source of the demagnetizing field [1]. As a result, the density of the SMCs reduces and results in the decreasing of the permeability.   Figure 7 reveals the hysteresis loops of the matrix iron particles and the hybrid-coated particles prepared by the ball-milling method with the various additional amounts of Al2O3. The saturated magnetization of raw pure iron is approximately 205 emu/g. All the hybrid-coated powders display excellent properties of saturated magnetization of approximately 160 emu/g to 190 emu/g. It is widely accepted that the saturated magnetization mainly depends on the composition and the purity of the materials [1], and that introducing non-magnetic phosphate and α-Al2O3 would dilute the percentage of magnetic phases and reduce the Ms as a result. The decreasing trend of the saturated magnetization in Figure 7 is due to the increasing content of Al2O3 in insulating layers.   (Figure 8a) and the sol-gel method (Figure 8b) with the same additional amount of Al2O3. The amplitude permeability of the hybrid-coated SMCs prepared by both methods exhibit excellent frequency stability at the measured frequency, which could be attributed to the The difference of the permeability caused by the fabrication technique could mainly be attributed to the grain size. As it is accepted that the magnetization is mainly down to the movement and the rotation of the domain walls, the required energy of the domain wall movement is less than that of the domain wall rotation. As the grain size decreases, the number of domain walls decreases and the contribution of the domain wall rotation to the magnetization is greater than that of the domain wall movement [22]. The effect of the ball-milling process is to decrease the grain size; therefore, the permeability of the SMCs fabricated by this method is not as high as that prepared by the sol-gel method. Figure 9 depicts the influence of the Al 2 O 3 content on the total core loss of the SMCs samples prepared by the different methods. It can be seen that the effect of the preparation methods on the total core loss is slight. Compared with the phosphate-coated samples, the total core loss of hybrid-coated samples decreases by~49% (202.3 at 100 kHz by the sol-gel method) and~47% (210.8 at 100 kHz by the ball-milling method) with 8% Al 2 O 3 . To further explore the effects of the Al 2 O 3 content and the fabrication methods on the total core loss of SMCs samples, the total core loss curves were investigated by the loss separation method. The total core loss (P t ) at a weak field and a relatively low frequency could be separated and written as [4,23]:

SMCs with Hybrid Phosphate-Alumina Coating
where f is the frequency, the constant K h and K e are the hysteresis loss factor and the eddy current loss factor, respectively. K h and K e are obtained based on Equation (1) from the curves of the total core loss and the loss-separation fitting.
Metals 2019, 9, x FOR PEER REVIEW 8 of 11 ~90% of that of phosphate-coated powders (93.7). The results could be explained by the following reasons. On the one hand, the permeability has a positive correlation with the Ms and the magnetism would be diluted by the non-magnetic Al2O3, as shown in Figure 6. On the other hand, a higher porosity with an increasing Al2O3 content could be concluded from Figure 5. The decreasing magnetic permeability is usually attributed to the inter-particle gap effect, as a source of the demagnetizing field [1]. As a result, the density of the SMCs reduces and results in the decreasing of the permeability. The difference of the permeability caused by the fabrication technique could mainly be attributed to the grain size. As it is accepted that the magnetization is mainly down to the movement and the rotation of the domain walls, the required energy of the domain wall movement is less than that of the domain wall rotation. As the grain size decreases, the number of domain walls decreases and the contribution of the domain wall rotation to the magnetization is greater than that of the domain wall movement [22]. The effect of the ball-milling process is to decrease the grain size; therefore, the permeability of the SMCs fabricated by this method is not as high as that prepared by the sol-gel method.  Figure 9 depicts the influence of the Al2O3 content on the total core loss of the SMCs samples prepared by the different methods. It can be seen that the effect of the preparation methods on the total core loss is slight. Compared with the phosphate-coated samples, the total core loss of hybridcoated samples decreases by ~49% (202.3 at 100 kHz by the sol-gel method) and ~47% (210.8 at 100 kHz by the ball-milling method) with 8% Al2O3. To further explore the effects of the Al2O3 content and the fabrication methods on the total core loss of SMCs samples, the total core loss curves were investigated by the loss separation method. The total core loss (Pt) at a weak field and a relatively low frequency could be separated and written as [4,23]: (1) where f is the frequency, the constant and are the hysteresis loss factor and the eddy current loss factor, respectively. and are obtained based on equation (1) from the curves of the total core loss and the loss-separation fitting.   Table 1 displays the K h , K e and the standard error of the samples with different additional amounts of the Al 2 O 3 fabricated by different methods. The K h shows a slight increasing trend with the elevated Al 2 O 3 amount, which is attributed to the higher porosity, as shown in Fig. 5. However, the K e exhibits an opposite, decreasing trend. The K e of the sample with 8% Al 2 O 3 is only one-quarter of that of the phosphate-coated sample. The K h and K e of samples prepared by the ball-milling method is higher and lower, respectively, than that of samples by the sol-gel method at the same Al 2 O 3 content. Density, particle size and residual stress play important roles in hysteresis loss [24]. The influence of the ball-milling process on particles could be divided into two parts: particle size and particle shape. Figures 2a and 4a suggest that the morphology of the particles transforms from irregular shape to flake shape after the ball-milling process, which deteriorates the compressibility of the powders and results in the lower density and higher porosity. In addition, the smaller particles in the compacts which were caused by the impact effect of balls on particles shows more unfavorable surfaces and defects. Owing to the pinning effect of these defects, the hysteresis loss of the samples prepared by the ball-milling method is higher than that of the samples prepared by the sol-gel method.
As for the eddy current loss factor (K e ), it is dominated by the electric resistivity, which could by elevated by increasing the intrinsic resistivity, defects, residual stress and porosity or by decreasing the particle size [24]. Coating an insulating Al 2 O 3 layer on the phosphate iron powders directly increases the intrinsic resistivity, and the electric resistivity is positively correlated to the Al 2 O 3 content. However, the increased number of the pores and grain boundaries and the reduced particle size which resulted from the ball-milling process account for the different K e of various SMC samples. Therefore, compared to the sample prepared by the sol-gel method, the contribution of the eddy current loss to the total core loss for the sample prepared by the ball milling is smaller.

Conclusions
Novel hybrid phosphate-alumina iron-based soft magnetic composites (SMCs) were successfully fabricated by both the sol-gel method and the ball-milling method. The hybrid insulating layers effectively enhanced the frequency stability of the permeability and reduced the total core loss in the SMCs. The SMCs prepared by both methods retained above 85% of the permeability and showed only approximately 50% of the total core loss of the phosphated samples prepared under the same process conditions. The loss separation results show that the contribution of the P h and the P e to the total core loss is inverse. The hysteresis loss in the samples prepared by the ball-milling method is greater than that prepared by the sol-gel method. These results suggest that the hybrid phosphate-alumina coating is promising, and can be used to obtain high-performance SMCs for electromagnetic devices at high frequency.