Frequency Properties of Polymer Bonded Compacts Obtained from Ball Milled Permalloy Powders with Mo and Cu Additions

Nanocrystalline powders from the Permalloy family, Ni75Fe25, Ni79Fe16Mo5, and Ni77Fe14Cu5Mo4, were obtained by mechanical alloying starting from elemental powders. All compositions were milled for up to 24 h in a high-energy planetary ball mill. The powders were single phase and nanocrystalline as determined by X-ray diffraction studies, with larger flatted particle sizes for Ni75Fe25 (about 400 μm) and Ni77Fe14Cu5Mo4 (about 470 μm), and smaller particle sizes for Ni79Fe16Mo5 (about 170 μm). The homogeneity of the samples was verified by energy-dispersive X-ray spectroscopy (EDX). Soft magnetic composites were obtained by adding 3% of Araldite to the powders, followed by compaction at 700 MPa, and then polymerization. A very good powder covering by the polymer layer was proven by EDX elementals maps. The influence of composition change on the electrical resistivity of the compacts was studied. Hysteresis measurements in static and dynamic fields of up to 10 kHz were recorded, showing the influence of composition and particle size on the compact properties.


Introduction
The Permalloy alloy family is one of the most-used classes of materials in electrotechnical applications, their development being a constant research topic [1]. This is largely due to their wide magnetic behavior tailored by composition/microstructure and annealing adjustments. One way of developing the Permalloy family consists in obtaining the ternary and quaternary alloys to enhance magnetic properties and extend the frequency use of the alloys to higher frequency ranges [2].
Although binary Permalloys have excellent magnetic properties (high permeability) their use is limited to low frequencies where the eddy currents are small. For higherfrequency applications, a reduction in the electrical conductivity of the materials is required. In the case of Permalloy, this can be achieved by the introduction of nonmagnetic elements such as molybdenum. The introduction of nonmagnetic elements leads to resistivity and, more importantly, permeability increases [3].
Of these techniques, mechanical alloying (MA) represents an interesting synthetic method since it implies elemental powder processing in a high-energy ball mill. By milling, stresses are induced in the powders and, by repeated cold welding and fracture events on elemental powder particles, atomic diffusion is promoted between elements, leading to alloy formation [20][21][22]. Mechanical alloying is a technique that allows nanocrystalline materials to be obtained in powder form, which is essential for easy shape forming by powder metallurgy process for industrial applications.

Materials and Methods
Three alloy compositions (Ni 75 Fe 25 , Ni 79 Fe 16 Mo 5 , and Ni 77 Fe 14 Cu 5 Mo 4 ) were obtained by high energy mechanical alloying starting from elemental powders: Ni-carbonyl, Fe-NC100.24, Mo-produced by chemical reduction and Cu powder. The samples were first homogenized for 15 min in a spatial homogenizator and then milled in a homemade planetary ball mill, alongside 59 hardened steel balls. The speed of the main disk was 290 rpm, and the relative speed of the vial was 240 rpm. The powders were introduced in vials filled with argon gas to avoid oxidation. The filling factor of the vial was 60%, and the ball/powder mass ratio was 8:1. The samples were milled for up to 24 h, based on the previous studies conducted on these compositions (more details about the obtaining of the Permalloy alloy by mechanical alloying are given in references [15,16,27]). The powders were then covered by 3 wt.% polymer-Araldite. The covering process can be described as follows: Araldite particles were dissolved in acetone, then the powder was added and mixed until the acetone had evaporated. This mixture was consolidated at 700 MPa in ring shapes and then polymerized in an oven at 150 • C for 3 h.
Structural analysis was performed by X-ray diffraction using a Siemens D 5000 diffractometer (Bruker, Billerica, MA, US); the patterns were recorded in the angular range of 35-103 • using Cu Kα radiation (λ = 1.5406 Å). The mean crystallite size and lattice strain were computed using Williamson-Hall plots obtained from the representation of as a function of d* = 4·sin(θ)/λ (2) where β represents the full width at half maximum (FWHM) for the sample, after the subtraction of the instrumental broadening, according to the formula where β meas is the FWHM measured in an XRD pattern, and β inst represents the FWHM of an annealed nickel sample used as reference). Using the linear plot fit, from the intercept, the mean crystalline size was calculated, and from the slope, the lattice strain was computed [44,45]. For the computation of the crystallite mean size and lattice strain, Rietveld analysis of the diffraction patterns was performed using the Fullprof software (3.00, version June 2015) [46]. As the instrumental reference, a sample of unmilled nickel powder was used. For estimating the particle size, ImageJ software (1.53t/24 August 2022) was used [47]. Hysteresis measurements were performed in DC and AC (up to 10 kHz) conditions on a Remagraph-Remacomp C-705 hysteresisgraph produced by Magnet Physik Dr. Steingroever GmbH (Köln, Germany).
For the electrical resistivity measurements, an adapted four-probe technique was used. The density was determined using the Archimedes water immersion method.
The scanning electron microscopy images were recorded using a JEOL JSM 5600 LV (Tokyo, Japan) microscope, and the chemical homogeneity was investigated by energy dispersive X-ray spectroscopy (Oxford Instruments, INCA 200 software, Aztech 4.2 software, High Wycombe, UK).

Mechanical Alloyed Powders
For all prepared systems, milled for up to 24 h, the recorded X-ray diffraction patterns are presented in Figure 1. For comparison, in each case, the diffraction pattern of an unmilled (0 h) sample is also presented.
All the 0 h samples present only the diffraction peaks of the elements used. After 24 h of milling, in all samples is recorded a single phase, confirming the complete disappearance of elemental Ni, Fe, Mo, and Cu peaks and the formation of the desired phase by mechanical alloying.
Recorded diffraction patterns for all milled samples present broad peaks, due to small crystallite size and internal stresses induced by milling. Using the Williamson-Hall method, mean crystallite size and internal stress values were computed, and they are presented in Table 1. All the 0 h samples present only the diffraction peaks of the elements used. After 24 h of milling, in all samples is recorded a single phase, confirming the complete disappearance of elemental Ni, Fe, Mo, and Cu peaks and the formation of the desired phase by mechanical alloying.
Recorded diffraction patterns for all milled samples present broad peaks, due to small crystallite size and internal stresses induced by milling. Using the Williamson-Hall method, mean crystallite size and internal stress values were computed, and they are presented in Table 1.  [19,48]. Analyzing the values, a correlation between grain size and lattice strain can be found: smaller grain size samples have smaller lattice strains. This behavior could be related to the fact that the strains induced by milling in the grains can be released by grain fragmentation.
The influence of the alloying elements on the Ni-Fe alloy morphology is best seen from the SEM images presented in Figure 2. For all samples, the chemical composition is also presented.   ) leads to a higher mean crystallite size. The obtained values for the mean crystallite size are comparable with other studies, but the strain is lower in our case [19,48]. Analyzing the values, a correlation between grain size and lattice strain can be found: smaller grain size samples have smaller lattice strains. This behavior could be related to the fact that the strains induced by milling in the grains can be released by grain fragmentation.
The influence of the alloying elements on the Ni-Fe alloy morphology is best seen from the SEM images presented in Figure 2. For all samples, the chemical composition is also presented.
In Figure 2, the SEM images show large Ni 75 Fe 25 particles for the Ni-Fe system. If 5% of Mo is added to the composition, a more brittle character for the powders is induced, leading, after 24 h, to smaller powder particles. Conversely, if Cu and Mo are added, the ductile behavior is promoted, and large particles are again obtained for the Ni 77 Fe 14 Cu 5 Mo 4 alloy milled for 24 h. Using ImageJ software for analyzing the particle size, we obtained for the Ni 75 Fe 25  The chemical analysis obtained by EDX indicates that the milled powders possess the desired composition and also a very good homogeneity, as proved by the elements distribution maps recorded for each element ( Figure 3). In Figure 2, the SEM images show large Ni75Fe25 particles for the Ni-Fe system. If 5% of Mo is added to the composition, a more brittle character for the powders is induced, leading, after 24 h, to smaller powder particles. Conversely, if Cu and Mo are added, the ductile behavior is promoted, and large particles are again obtained for the Ni77Fe14Cu5Mo4 alloy milled for 24 h. Using ImageJ software for analyzing the particle size, we obtained for the Ni75Fe25 powder, a mean value of about 400 μm, for the Ni79Fe16Mo5 powder, a mean value of about 170 μm, and for the Ni77Fe14Cu5Mo4 powder, a mean value of about 470 μm. The larger powders have a flake shape.
The chemical analysis obtained by EDX indicates that the milled powders possess the desired composition and also a very good homogeneity, as proved by the elements distribution maps recorded for each element (Figure 3).

Polymerized Compacts
Using the milled powders, compacts were fabricated. The ring sample size and aspect after polymerization are presented in Figure 4. A very smooth surface is obtained for Ni79Fe16Mo5 samples, due to the smaller particle size of the powder.  In Figure 2, the SEM images show large Ni75Fe25 particles for the Ni-Fe system. If 5% of Mo is added to the composition, a more brittle character for the powders is induced, leading, after 24 h, to smaller powder particles. Conversely, if Cu and Mo are added, the ductile behavior is promoted, and large particles are again obtained for the Ni77Fe14Cu5Mo4 alloy milled for 24 h. Using ImageJ software for analyzing the particle size, we obtained for the Ni75Fe25 powder, a mean value of about 400 μm, for the Ni79Fe16Mo5 powder, a mean value of about 170 μm, and for the Ni77Fe14Cu5Mo4 powder, a mean value of about 470 μm. The larger powders have a flake shape.
The chemical analysis obtained by EDX indicates that the milled powders possess the desired composition and also a very good homogeneity, as proved by the elements distribution maps recorded for each element (Figure 3).

Polymerized Compacts
Using the milled powders, compacts were fabricated. The ring sample size and aspect after polymerization are presented in Figure 4. A very smooth surface is obtained for Ni79Fe16Mo5 samples, due to the smaller particle size of the powder.

Polymerized Compacts
Using the milled powders, compacts were fabricated. The ring sample size and aspect after polymerization are presented in Figure 4. A very smooth surface is obtained for Ni 79 Fe 16 Mo 5 samples, due to the smaller particle size of the powder.
The electrical resistivities of the toroidal composite compacts were measured and are presented along with their densities in Table 2. In the case of Ni 75 Fe 25 , the electrical resistivity is four orders of magnitude higher than that of a cast sample [4]. The electrical resistivity increase relates to the insulating layer on the individual powder particles and the high amount of grain boundary naturally present in milled powders. As Mo and Cu are added, the electrical resistivity further increases (by a factor of two if only Mo is added, and by a factor of four if both Mo and Cu are added). Although for the cast samples, there is only a small difference between Ni 79 Fe 16 Mo 5 and Ni 77 Fe 14 Cu 5 Mo 4 , in the case of the milled samples, there is a higher difference. The difference could be the effect of the supplementary disorder induced in the binary alloy by the extra atoms and, as found in reference [49], adding nonmagnetic elements by lowering the long-range order. Electrical resistance increases with the change of the system from binary to ternary were reported also in the case of thin films [6].  The electrical resistivities of the toroidal composite compacts were measured and are presented along with their densities in Table 2. In the case of Ni75Fe25, the electrical resistivity is four orders of magnitude higher than that of a cast sample [4]. The electrical resistivity increase relates to the insulating layer on the individual powder particles and the high amount of grain boundary naturally present in milled powders. As Mo and Cu are added, the electrical resistivity further increases (by a factor of two if only Mo is added, and by a factor of four if both Mo and Cu are added). Although for the cast samples, there is only a small difference between Ni79Fe16Mo5 and Ni77Fe14Cu5Mo4, in the case of the milled samples, there is a higher difference. The difference could be the effect of the supplementary disorder induced in the binary alloy by the extra atoms and, as found in reference [49], adding nonmagnetic elements by lowering the long-range order. Electrical resistance increases with the change of the system from binary to ternary were reported also in the case of thin films [6]. The density of the composite compacts is similar for the three alloys, with a small increase for the sample containing Cu. The increase, which can be related to Cu ductility, transferred to the alloy, leading to a higher compaction ratio.
Since the toroidal composite compacts were obtained by polymerization, the covering layer of the particles by polymer was studied by EDX analysis, and the results are shown in Figure 5. In Figure 5, for consistency, are shown only the maps of Ni, and polymer components (C and O).  The density of the composite compacts is similar for the three alloys, with a small increase for the sample containing Cu. The increase, which can be related to Cu ductility, transferred to the alloy, leading to a higher compaction ratio.
Since the toroidal composite compacts were obtained by polymerization, the covering layer of the particles by polymer was studied by EDX analysis, and the results are shown in Figure 5. In Figure 5, for consistency, are shown only the maps of Ni, and polymer components (C and O).
In Figure 5, the elements C and O are included since they are the representative elements of the polymer used. The C and O distribution maps show the presence of the polymer around alloy particles. For the limitation of eddy current development, an insulating layer is beneficial for the samples. The polymerization of the milled particles, as exemplified by the EDX maps, shows a very good covering, promising good results.

AC Magnetic Properties of the Composites
On the above-obtained compacts, AC magnetic properties in frequencies up to 10 kHz were measured. The evolution of the total losses (hysteresis losses and eddy currents) for the three materials at an induction field of 0.01 T are presented in Figure 6. From the total losses, the hysteresis losses can be obtained by extrapolating the evolution of losses at zero frequency, since they do not depend on frequency [51].
The obtained values indicate that smaller core losses are obtained for the higher resistive sample, Ni 77 Fe 14 Cu 5 Mo 4 . A direct comparison between Ni 75 Fe 25 and Ni 77 Fe 14 Cu 5 Mo 4 particles shows that an electrical resistivity increase leads to a decrease in overall losses. In addition, if the particles are reduced in size, the losses are reduced further. A similar conclusion was drawn in [39,52]. The lower power losses of the Ni 75 Fe 25 compacts compared with the values reported in other studies could be due to better insulating layer coverage and particle size control [53]. In Figure 5, the elements C and O are included since they are the representative elements of the polymer used. The C and O distribution maps show the presence of the polymer around alloy particles. For the limitation of eddy current development, an insulating layer is beneficial for the samples. The polymerization of the milled particles, as exemplified by the EDX maps, shows a very good covering, promising good results.

AC Magnetic Properties of the Composites
On the above-obtained compacts, AC magnetic properties in frequencies up to 10 kHz were measured. The evolution of the total losses (hysteresis losses and eddy currents) for the three materials at an induction field of 0.01 T are presented in Figure 6. From the total losses, the hysteresis losses can be obtained by extrapolating the evolution of losses at zero frequency, since they do not depend on frequency [51]. The obtained values indicate that smaller core losses are obtained for the higher resistive sample, Ni77Fe14Cu5Mo4. A direct comparison between Ni75Fe25 and Ni77Fe14Cu5Mo4 particles shows that an electrical resistivity increase leads to a decrease in overall losses. Concerning the initial relative permeability versus frequency, there are no visible variations with frequency in the studied range, as can be seen in Figure 7. This indicates that the cutting frequency is higher than 10 kHz, and the possibility of using the compacts in applications at higher frequencies.
sistive sample, Ni77Fe14Cu5Mo4. A direct comparison between Ni75Fe25 and Ni77Fe14Cu5Mo4 particles shows that an electrical resistivity increase leads to a decrease in overall losses. In addition, if the particles are reduced in size, the losses are reduced further. A similar conclusion was drawn in [39,52]. The lower power losses of the Ni75Fe25 compacts compared with the values reported in other studies could be due to better insulating layer coverage and particle size control [53].
Concerning the initial relative permeability versus frequency, there are no visible variations with frequency in the studied range, as can be seen in Figure 7. This indicates that the cutting frequency is higher than 10 kHz, and the possibility of using the compacts in applications at higher frequencies. The lowest permeability values are obtained for the Ni79Fe16Mo5 sample, probably in connection with the smaller particle size and the polymerization process itself. According to [39], the polymer creates air gaps, and the process is more pronounced for small particles. As shown in Figure 7, the increase in the permeability values from the Ni75Fe25 compact to the Ni77Fe14Cu5Mo4 compact can be attributed to the higher density of the Ni77Fe14Cu5Mo4 compact, and, therefore, to a smaller number of air gaps present in the composite compact. Also, it is known that the simultaneous cancelation of the magnetocrystalline anisotropy and the magnetostriction of Ni-Fe alloys is reached by the addition of Cu and Mo. This will also lead to a higher magnetic permeability of the compacts based The lowest permeability values are obtained for the Ni 79 Fe 16 Mo 5 sample, probably in connection with the smaller particle size and the polymerization process itself. According to [39], the polymer creates air gaps, and the process is more pronounced for small particles. As shown in Figure 7, the increase in the permeability values from the Ni 75 Fe 25 compact to the Ni 77 Fe 14 Cu 5 Mo 4 compact can be attributed to the higher density of the Ni 77 Fe 14 Cu 5 Mo 4 compact, and, therefore, to a smaller number of air gaps present in the composite compact. Also, it is known that the simultaneous cancelation of the magnetocrystalline anisotropy and the magnetostriction of Ni-Fe alloys is reached by the addition of Cu and Mo. This will also lead to a higher magnetic permeability of the compacts based on Ni 77 Fe 14 Cu 5 Mo 4 powders. The stability of permeability versus applied frequency is a consequence of the good insulating layer on the surface of the particles [54]. A good insulating layer will hinder the excessive development of eddy currents in the samples and will reduce the skin effect (the penetration depth will be higher) that predominates at higher frequencies.
DC measurements indicate a shift in the hysteresis tilt, due to the difference in the permeabilities values, as a consequence of larger or smaller particles existing in the sample, as presented in Figure 8.
The hysteresis loops are practically identical for the Ni 75 Fe 25 and Ni 77 Fe 14 Cu 5 Mo 4 samples-samples with similar particle size distribution. For the Ni 79 Fe 16 Mo 5 compact, a more tilted curve is recorded, in connection with smaller particle size. Smaller particle sizes will automatically induce a larger number of air gaps in the compacts and, consequently, larger demagnetizing fields. This will finally lead to a horizontal tilt of the hysteresis loop corresponding to the Ni 79 Fe 16 Mo 5 compact, indicating a lower magnetic permeability of the compact. A similar conclusion was observed for NiFe-based compacts in [55]. The induction and the coercive field for each compact are given in Table 3.
Thus, the obtained polymerized compacts follow the rule and a large induction is obtained for larger particles and a smaller induction for fine particles [23]. It is worth noting that the samples presented in this study did not reach saturation, as can be observed in Figure 8. However, comparing the induction values of the compacts based on Ni 79 Fe 16 Mo 5 and Ni 77 Fe 14 Cu 5 Mo 4 powders, it can be seen that the compact based on Ni 77 Fe 14 Cu 5 Mo 4 powders presents the largest value for the saturation induction. This is somewhat unusual since the amount of non-magnetic elements is larger in the case of the Ni 77 Fe 14 Cu 5 Mo 4 powders compared to the Ni 79 Fe 16 Mo 5 powders. To explain this fact, once again particle size must be considered, and, consequently, the number of air gaps present in the samples. In the case of the compact based on Ni 79 Fe 16 Mo 5 powders, due to the smaller particle size, a large number of air gaps are created in the composite compact. It is known that air gaps create demagnetizing fields that oppose the magnetizing field. In such a case, even if the magnetizing field is 9 kA/m in both cases (compacts based on Ni 79 Fe 16 Mo 5 and Ni 77 Fe 14 Cu 5 Mo 4 powders), the magnetic field experienced by the sample based on Ni 79 Fe 16 Mo 5 powder is lower leading to a smaller magnetic induction. It seems that the mean crystallite size does not influence magnetic induction. Concerning the coercive field, compacts based on smaller particles have a smaller coercive field. The same explanation concerning the difference that exists between the applied magnetic field and the real field experienced by the sample (difference induced by the air gaps and the demagnetizing fields) can explain the measured values. It is known that the minor hysteresis loops (measured at an induction inferior to the saturation induction) present a coercive field that increases as the induction value approaches saturation induction. As the real value of the applied magnetizing field (the field experienced by the sample) is lower in the case of the sample based on Ni 79 Fe 16 Mo 5 powders, it is reasonable to expect lower values of the coercive field of this sample. This hypothesis is confirmed, in a way, by the fact that the coercive field of the compacts based on large particles (Ni 75 Fe 25  on Ni77Fe14Cu5Mo4 powders. The stability of permeability versus applied frequency is a consequence of the good insulating layer on the surface of the particles [54]. A good insulating layer will hinder the excessive development of eddy currents in the samples and will reduce the skin effect (the penetration depth will be higher) that predominates at higher frequencies.
DC measurements indicate a shift in the hysteresis tilt, due to the difference in the permeabilities values, as a consequence of larger or smaller particles existing in the sample, as presented in Figure 8. The hysteresis loops are practically identical for the Ni75Fe25 and Ni77Fe14Cu5Mo4 samples-samples with similar particle size distribution. For the Ni79Fe16Mo5 compact, a more tilted curve is recorded, in connection with smaller particle size. Smaller particle sizes will automatically induce a larger number of air gaps in the compacts and, consequently, larger demagnetizing fields. This will finally lead to a horizontal tilt of the hysteresis loop corresponding to the Ni79Fe16Mo5 compact, indicating a lower magnetic permeability of the compact. A similar conclusion was observed for NiFe-based compacts in [55]. The induction and the coercive field for each compact are given in Table 3. Thus, the obtained polymerized compacts follow the rule and a large induction is obtained for larger particles and a smaller induction for fine particles [23]. It is worth not-

Conclusions
The Permalloy alloys (Ni 75 Fe 25 , Ni 79 Fe 16 Mo 5, and Ni 77 Fe 14 Cu 5 Mo 4 ) were obtained by mechanical alloying, and after 24 h milling, as single-phase materials with nanocrystalline structure. The obtained powders have a very good chemical homogeneity. By polymerization with 3% Araldite, soft magnetic compacts were obtained.
The electrical resistivity of the compacts is in all cases four times larger than for cast equivalents. As additional elements (Cu and/or Mo) are added to the Ni-Fe powders, the electrical resistivity increases; the quaternary alloy has a four-times larger resistivity than the binary alloy.
In the AC conditions, the lowest power losses and highest permeability are obtained for the compacts with the highest resistivity (Ni 77 Fe 14 Cu 5 Mo 4 ). The losses are influenced by the particle size; for smaller particle sizes, the losses are smaller.
The permeability of the samples is dependent on the air-gap content of the samples; as the number of air gaps increases (for small particle sizes), the permeability decreases. Therefore, the Ni 79 Fe 16 Mo 5 sample has the smallest permeability.
Hysteresis loops recorded in DC condition have a sensitivity toward particle size, leading to a more tilted curve for Ni 79 Fe 16 Mo 5 samples, with smaller particle sizes. The induction field is more sensitive to the particle size than to mean crystallite size.