Fabrication and Properties of Axially Compressed Isotropic Epoxy-Bonded NdFeB Magnets with Partial Rare-Earth Substitution

Abstract

This work investigates the fabrication and performance of axially compressed isotropic epoxy-bonded NdFeB-type magnets produced from melt-spun powders with partial substitution of (Nd,Pr) by (La,Ce). Four alloy compositions were synthesized and processed into bonded magnets using two powder-to-binder weight ratios (95:5 and 96.5:3.5). Structural analysis confirms that all substituted alloys retain the tetragonal Nd₂Fe₁₄B phase (up to ~95 wt%) even at high substitution levels, while the lattice parameters decrease slightly with increasing (La,Ce) content. Microscopy analysis confirms a homogeneous distribution of the binder phase around the powder particles, demonstrating uniform binder–powder integration. Thermal analysis reveals composition-dependent Curie temperatures and enhanced crystallization onset in highly substituted powders. Magnetic measurements on both powders and bonded magnets show that increasing substitution leads to a gradual reduction in remanence, coercivity, and energy product, though all samples maintain strong hard-magnetic behavior. Increasing the powder fraction to 96.5 wt.% significantly improves all magnetic parameters due to higher magnetic-phase density and enhanced interparticle coupling, yielding bonded magnets with densities up to ~80% of the theoretical value. The resulting magnets achieve competitive performance, uniform field distribution and isotropic magnetization with (BH)max values about 65 kJ/m³, a coercivity around 660 kA/m, and superior thermal stability compared with commercial bonded NdFeB magnets. Overall, partial substitution with light rare-earth elements (La,Ce) provides a cost-effective route to high-density bonded NdFeB magnets that combine strong magnetic performance, enhanced thermal stability, and suitability for lightweight, complex-shaped industrial applications. Surprisingly, the coefficients of the temperature variation of coercivity and (BH)_max are much better compared to the commercial NdFeB bonded magnets.

1. Introduction

The discovery of NdFeB permanent magnet materials in 1983 revolutionized the field of rare-earth-based magnetic materials and stimulated extensive research on high-performance permanent magnets for a wide range of applications, including electric vehicles, wind turbines, robotics, and consumer electronics [1,2,3]. These magnets offer the highest energy product ${\left(\mathrm{B}\mathrm{H}\right)}_{\mathrm{m}\mathrm{a}\mathrm{x}}$ among commercially available magnetic materials, making them indispensable for advanced electromechanical systems. However, the increasing demand for neodymium (Nd) and other critical rare-earth (RE) elements has raised significant concerns about their cost, supply stability, and environmental impact [4]. Consequently, research efforts have been directed toward developing Nd–Fe–B magnets with reduced Nd content or partial substitution by more abundant rare earths such as praseodymium (Pr), lanthanum (La), and cerium (Ce) [1,4].

Bonded Nd–Fe–B magnets, which are produced by combining magnetic powders with a polymeric binder, have emerged as an attractive alternative to conventional sintered magnets [3,5]. These materials offer several advantages, including the ability to produce complex shapes with tight dimensional tolerances, high electrical resistivity, low eddy current losses, and excellent mechanical properties. Among various binder systems, epoxy resins have been extensively used due to their good adhesion, chemical stability, and ability to form strong three-dimensional crosslinked networks [5,6]. Depending on the processing method, bonded magnets can be either isotropic or anisotropic. In isotropic magnets, the randomly oriented grains simplify fabrication and eliminate the need for magnetic field alignment, although the magnetic performance is typically lower than that of anisotropic magnets [5,7].

Compression molding is one of the most effective fabrication techniques for bonded magnets, enabling high powder loading and density—up to 80–85% of the theoretical maximum—compared to injection molding, which typically achieves around 65% [8,9]. The process involves compacting the powder–binder mixture under high pressure and subsequently curing the resin to achieve mechanical integrity. Increasing the density of the bonded magnet is crucial because magnetic properties such as remanence (${\mathrm{B}}_{\mathrm{r}}$) and energy product (${\left(\mathrm{B}\mathrm{H}\right)}_{\mathrm{m}\mathrm{a}\mathrm{x}}$) are directly related to the volume fraction of magnetic material in the composite [8]. Axial compression molding, in particular, allows the production of magnets with a high density and uniformity along the compression axis, which is essential for high-performance applications [8].

While compression molding of epoxy-bonded NdFeB magnets has been widely investigated, most previous studies have focused on Nd-rich compositions or on limited substitution levels of light rare-earth elements, typically not exceeding 30–40% of the rare-earth sublattice [10,11]. In addition, the reported bonded systems often employ moderate powder loadings (≤95 wt.%), resulting in densities and magnetic performances that remain significantly lower than those of sintered magnets, thereby limiting their applicability in demanding operating conditions [12,13]. Studies combining high rare-earth substitution levels with high-density axial compression molding are scarce, and systematic investigations of their combined effects on magnetic performance and thermal stability are still lacking.

The present work addresses this gap by demonstrating the fabrication of axially compressed isotropic epoxy-bonded NdFeB magnets with high substitution levels of (Nd,Pr) by (La,Ce), reaching 75% of the rare-earth sublattice, while preserving the dominant tetragonal Nd₂Fe₁₄B phase. By employing optimized axial compression molding with powder loadings up to 96.5 wt.%, bonded magnets with densities approaching ~80% of the theoretical value are achieved. Scanning electron microscopy (SEM) analysis confirms homogeneous phase distribution and a well-defined powder–binder interface. Furthermore, the resulting magnets exhibit unusually low temperature coefficients of coercivity and energy product, outperforming commercial bonded NdFeB magnets in thermal stability. These results demonstrate that high-density compression molding combined with extensive light rare-earth (La,Ce) substitution constitutes a viable route toward cost-effective bonded magnets with competitive performance.

2. Materials and Methods

A series of NdFeB-type alloys with the general composition (Nd,Pr)_28−x(La,Ce)_x(Co,Zr)_~2B_~1Fe_bal (weight%) were synthesized using the melt-spinning (MS) technique [13]. The compositional design incorporated gradual substitution of neodymium–praseodymium (Nd,Pr) with lanthanum–cerium (La,Ce) to investigate the effect of rare-earth substitution on the magnetic properties of the resulting powders. The substitution ratio was defined as

$${\mathrm{r}}_{\mathrm{sub}}={\displaystyle \frac{\left(\mathrm{L}\mathrm{a},\mathrm{C}\mathrm{e}\right)}{\left(\mathrm{N}\mathrm{d},\mathrm{P}\mathrm{r}\right)+\left(\mathrm{L}\mathrm{a},\mathrm{C}\mathrm{e}\right)}}$$

(1)

Four alloy compositions were used, corresponding to four different x-values as shown in

Table 1. Labeling, composition, and density of the bonded (RE)₂Fe₁₄B melt-spun (MS) magnetic powders.

Powder Name	Substitution Index
MS-Powder_1	x = 0
MS-Powder_2	x = 5
MS-Powder_3	x = 8
MS-Powder_4	x = 21

, with $\mathrm{x}=0$ signifying no substitutions, and thus representing the reference NdFeB permanent magnet. Each powder was mixed with an epoxy solution to create the axially compressed isotropic bonded magnets, following the steps depicted in the fabrication flow chart in Figure 1 and explained below.

The epoxy used in these experiments consisted of XD4447 resin and XD4448 hardener (Robnor ResinLab Ltd., Swindon, UK)), mixed in a weight ratio of 100:30 with a magnetic stirrer. The substances were stirred in a plastic cup under ambient air and in acetone for better homogeneity. The mixture was then transferred to a ceramic container and pre-dried at 100 °C for approximately 30 min, or until the acetone was completely evaporated.

Initially, a small amount of Licowax (Supelco Solutions, Darmstadt, Germany) was manually mixed with the magnetic powder, carefully coating all the grains, thereby improving the integration with the epoxy in the following step. The respective magnetic powder was then combined with the epoxy mixture in acetone [14] using a magnetic stirrer, at two carefully selected weight ratios: 95:5 and 96.5:3.5, as presented in detail in

Table 2. Labeling, composition, and density (g/cm³) of the bonded magnet samples.

Sample Name	Powder		Epoxy	Wax	Ratio	Density
	Name	(g)	(g)	(g)	Powder–Binder	(g/cm3)
A1	MS-Powder_1	14.25	0.75	0.03	95:5	6.17
A2	MS-Powder_1	14.475	0.525	0.04	96.5:3.5	6.10
B1	MS-Powder_2	14.25	0.75	0.03	95:5	6.11
B2	MS-Powder_2	14.475	0.525	0.04	96.5:3.5	6.14
C1	MS-Powder_3	14.25	0.75	0.03	95:5	6.04
C2	MS-Powder_3	14.475	0.525	0.04	96.5:3.5	6.06
D1	MS-Powder_4	14.25	0.75	0.03	95:5	5.98
D2	MS-Powder_4	14.475	0.525	0.04	96.5:3.5	6.02

. The 95:5 ratio was chosen to ensure improved mechanical properties while maintaining adequate magnetic performance, whereas the 96.5:3.5 ratio prioritized the magnetic properties over the mechanical strength [9,13,14].

The resulting mixture was left at room temperature until the acetone evaporated. An 8 mm diameter cylindrical die was then filled with the powder–epoxy mixture and placed in a hydraulic press, applying ~500 MPa for 5–10 min. Preliminary trials indicated that the pressing duration impacts both the density and surface texture of the resulting pellet. Subsequently, the pellets were placed on a ceramic plate and cured in a furnace at 150 °C for 2 h in air atmosphere, sufficient for the solidification of the epoxy.

Immediately after curing, the pellets were magnetized using a 2.2 T magnetic field. While pellets can also be magnetized after cooling—at room temperature—it was observed that maximum magnetization is achieved when the samples are magnetized while still hot. At elevated temperatures, the magnetic particles and magnetic domains within the material can more easily reorient along the applied field, leading to higher remanence and improved magnetic performance [14].

The density of the samples has been determined by two methods: the Archimedes principle [9,14,15] and by measuring the weight and the dimensions of the magnets [9,16]. Both methods gave the same results, validating the data shown in Table 2.

The structural characterization of the melt-spun magnetic powders was performed using powder X-ray diffraction (XRD) on a Rigaku (MiniFlex600, Tokyo, Japan) Diffractometer (600 W) equipped with Cu Kα radiation and operated in Bragg–Brentano geometry. The diffraction patterns were analyzed by Rietveld refinement via FULLPROF software (version January 2023) [17]. The particle size distribution of the melt-spun NdFeB powders was measured by Mastersizer Laser diffraction particle size analysis (Malvern Panalytical, Malvern, UK) under dry dispersion conditions complemented by scanning electron microscopy (SEM) for morphology verification, yielding volume-based cumulative and density distributions (D10, D50, D90), mean diameters, and specific surface areas. The surface morphology and elemental composition of the new specimens were examined using a Quanta Inspect scanning electron microscope (FEI, Hillsboro, OR, USA) and by energy-dispersive X-ray microanalysis (EDS) with a PV7760/68 ME detector (EDAX AMETEK, Warrendale, PA, USA). The analyses were conducted under high vacuum at an accelerating voltage of 15 kV in Secondary Electron (SE) imaging mode. For the elemental analysis, a 200 s acquisition time and 30–35% dead time were used. Quantitative analysis of the collected spectra was performed using the dedicated software (EDAX Genesis, V4.5, AMETEK) with ZAF (atomic number, absorbance, fluorescence) correction factors. Thermal characterization was conducted using differential scanning calorimetry (DSC) (Hunan Gonoava Instrument Co., Changsha City, China) to evaluate the thermal behavior and phase transformations of the (Nd,Pr)_28-x(La,Ce)_x(Co,Zr)_~2B_~1Fe_bal alloy powders. Magnetic measurements were carried out at room temperature and elevated temperatures using a DX-2012H150 BH-looper magnetic property measurement system (Xiamen Dexing Magnet Tech. Co., Ltd., Xiamen, China) with maximum field of 2.2 T. Additional magnetic characterization was performed using a VSM (155 PAR) vibrating sample magnetometer with a maximum field of 2 T (Lake Shore, Westerville, OH, USA), as well as a custom-built 3D magnetic mapping system. The setup employs a 3-axis Hall magnetic field transducer mounted on a computer-controlled Cartesian motion platform, enabling automated field mapping within a scanning volume of 135 × 135 × 135 mm³ with micrometer-level positioning accuracy. Data acquisition and motion control were performed through an integrated multifunction DAQ and dedicated PC software, allowing reliable spatial reconstruction of the magnetic field distribution.

3. Results and Discussion

3.1. Structural Characterization

Figure 2 presents the powder X-ray diffraction patterns of the composition (Nd,Pr)_28–x(La,Ce)_x(Co,Zr)_~2B_~1Fe_bal compound with x = 0, prepared by the melt-spinning technique of polycrystalline materials. The black dots represent the experimental data collected, while the red line shows the predicted pattern obtained from the Rietveld refinement. The blue line indicates the difference between the experimental data and the Rietveld fit. The Rietveld pattern is described using the T-C-H pseudo-Voigt function No. 7. The structural model that we used consists of the Nd₂Fe₁₄B phase, which is characterized by space group No. 186, P-42/mnm (Hermann–Mauguin symbol), in the tetragonal crystal system with a general multiplicity of 12, and the α-Fe, described by space group No. 82, (I m–3 m), in the cubic crystal system with a general position multiplicity of 24. The free refinement parameters include the lattice constants and the Lorentzian isotropic strain and size parameters. The presence of sharp and well-defined diffraction peaks in Figure 2 indicates that the sample exhibits high crystallinity and a well-ordered atomic structure. A small amount of amorphous phase is also present at lower degrees, which does not affect the peak positions and is a common characteristic for sample developed with the melt-spinning method [18]. The weight fraction of the Nd₂Fe₁₄B-type phase is 96.5%, suggesting that the sample is composed almost entirely of one phase, with a small presence of α-Fe phase. The Rietveld analysis determined that the lattice constants of the Nd₂Fe₁₄B phase are a = 8.7882(1) Å and c = 12.1958(1) Å, corresponding to a cell volume of 941.91(1) ų. Additionally, the isotropic strain parameter is X = 0.5664(1), and the isotropic size parameter is Y = 0.0087(1), which yields an average apparent size of 1060(10) Å. The refinement converged with a weighted profile R-factor Rwp = 9.20% for the Nd₂Fe₁₄P phase.

Figure 3 shows the Rietveld plots of the sample with no additional elemental substitution (x = 0), along with the samples with x = 5–21. As the substitution level (x-values) increases, no additional phases are detected, indicating the persistence of a robust 2:14:1 tetragonal crystal structure. Furthermore, increasing x leads to a slight reduction in the weight fraction of the NdFeB phase, up to approximately 95%, accompanied by a corresponding increase in the cubic α-Fe phase, as summarized in

Table 3. Comparison of the weight fraction (%), of the Nd₂Fe₁₄B-type phase, the lattice constants (Å), and the cell volume (${Å}^3$), between the (Nd,Pr)_28–x(La,Ce)_x(Co,Zr)_~2B_~1Fe_bal samples.

Compound (x)	Nd2Fe14 B-Type ph. w. fr. (%)	$\mathbf{a}\left(\mathit{Å}\right)$	$\mathbf{c}\left(\mathit{Å}\right)$	$\mathbf{V}\mathbf{o}\mathbf{l}\mathbf{u}\mathbf{m}\mathbf{e}\left({\mathit{Å}}^3\right)$
0	96.5	8.7882(1)	12.1958(1)	941.91(1)
5	96.4	8.7859(1)	12.1816(1)	940.33(1)
8	95.5	8.7777(2)	12.1714(2)	938.68(1)
21	94.8	8.7787(2)	12.1666(2)	937.63(1)

. The introduction of elemental substitution also induces changes in the lattice constants, resulting in a subtle shift of the Bragg peaks toward higher diffraction angles.

Table 3 summarizes the weight fraction (%) of the Nd₂Fe₁₄B phase, together with the lattice constants (Å) and cell volumes (ų) for the series of samples. It is noteworthy that both the a and c lattice parameters vary with elemental substitution. Specifically, as the substitution level increases up to x = 21, both lattice constants decrease, resulting in a corresponding reduction in the overall unit-cell volume. Changes in rare-earth cations in the host lattice might be the reason for the result. Specifically, a part of the positions of Nd³⁺ were occupied by both Ce³⁺ (radius~1.14 Å) and Ce⁴⁺ (radius~0.97 Å) ions, and the radius of Ce⁴⁺ ions is small compared with that of Nd³⁺ (radius~1.15 Å), which led to the decrease of the lattice constants [19,20,21]. Meanwhile, the theoretical density of a Nd₂Fe₁₄B magnet without substitutions (ρ = 7.605 g/cm³) was also obtained by XRD refinement.

The particle size distribution (PSD) of the substituted NdFeB powder, shown in Figure 4, reveals the overall uniformity and granularity of the powder particles. The cumulative and density distribution curves indicate that the powder exhibits a relatively narrow particle size range, with most particles falling between approximately 20 µm and 150 µm. The median particle diameter (x₅₀) is 75.23 µm, suggesting that half of the powder mass consists of particles smaller than this value. The distribution also shows characteristic values of x₁₀ = 22.60 µm, x₁₆ = 33.14 µm, x₈₄ = 119.94 µm, and x₉₀ = 133.67 µm, reflecting moderate size dispersion. The Sauter mean diameter (SMD) of 39.22 µm represents the surface area to weighted mean size, which is important for surface-related phenomena such as sintering and oxidation behavior. The volume mean diameter (VMD) of 77.55 µm corresponds to the dominant particle fraction and is consistent with the primary mode observed in the density distribution curve. The specific surface area (S_v = 0.15 m²/cm³) and specific surface (S_m = 201.27 cm²/g) values indicate that the powder has a moderately developed surface, suitable for efficient compaction during the magnet fabrication process. Overall, the PSD analysis confirms that the powder possesses a well-balanced particle size profile, providing favorable packing density and sinterability for the subsequent processing of NdFeB-based magnetic materials [9].

The microstructure of the (Nd,Pr)_28–x(La,Ce)_x(Co,Zr)_~2B_~1Fe_bal with x = 5 bonded magnet is presented in Figure 5a by scanning electron microscopy (SEM) images. The observations reveal a densely packed particulate microstructure in which elongated substituted NdFeB particles are embedded within a continuous binder phase. At higher magnification, seen in Figure 5b, a well-defined interfacial coating surrounding the particles becomes clearly visible, indicating good binder wetting and coverage of the particle surfaces. This conformal layer suggests effective particle encapsulation, which is expected to influence both interparticle coupling and the overall mechanical integrity of the composite.

Local EDS analysis performed at the particle–matrix interface (Figure 5b) shows that Fe is the dominant elemental contribution from the particles, accompanied by characteristic peaks from rare-earth elements (La, Ce, Pr, and Nd). A pronounced C peak is detected in the interfacial regions and is attributed to the organic binder, confirming the presence of a carbon-rich coating surrounding the particles. Minor amounts of Zr and Co are also observed. The spatial association of the C signal with the interparticle regions supports the formation of a distinct binder-derived coating layer rather than a homogeneous carbon distribution throughout the particles.

3.2. Thermal Characterization

Differential scanning calorimetry (DSC) was performed to evaluate the thermal behavior and phase transformations of the (Nd,Pr)_28–x(La,Ce)_x(Co,Zr)_~2B_~1Fe_bal alloy powders, as in Figure 6. The DSC curves acquired under a nitrogen atmosphere reveal two characteristic transitions: the Curie temperature T_C, and the crystallization of the amorphous matrix. A clear endothermic step is observed at lower temperatures, which corresponds to the Curie temperature (T_C) of the crystalline NdFeB matrix. The ${\mathrm{T}}_{\mathrm{C}}=\left(580, 546, 545, 504\right)\mathrm{K}$ for each substitution rate $\mathrm{x}=0, 5, 8, 21$, respectively, shifts slightly depending on the substitution level of (La,Ce), reflecting the effect of rare-earth replacement on magnetic ordering. At higher temperatures, exothermic peaks are evident, indicating crystallization of the amorphous phase into crystalline Nd₂Fe₁₄B and secondary phases. This amorphous phase is also detected in the XRD diagrams, Figure 2 and Figure 3, and is now validated by the thermal study of the alloys. The crystallization onset temperature decreases with higher (La,Ce) content, suggesting enhanced thermal stability of the substituted alloys. These results confirm that (La,Ce) substitution influences both the magnetic transition temperature and the crystallization behavior, thereby affecting the overall structural and thermal stability of the powders.

3.3. Magnetic Characterization of Melt-Spun Powders

The magnetic hysteresis loops of the substituted NdFeB powders, presented in Figure 7, reveal the influence of compositional modification on their magnetic performance. All samples exhibit typical ferromagnetic behavior with well-defined coercivity (${\mathrm{H}}_{\mathrm{c}}$), remanence (${\mathrm{M}}_{\mathrm{r}}$), and saturation magnetization (M_s), confirming the preservation of the Nd₂Fe₁₄B magnetic phase. A clear trend is observed with an increasing substitution level x, both M_s and M_r gradually decrease, accompanied by a reduction in the coercive field. The reference powder (MS-Powder_1, x = 0) shows the highest magnetic parameters ${\mathrm{H}}_{\mathrm{c}}=712 \mathrm{k}\mathrm{A}/\mathrm{m}$, ${\mathrm{M}}_{\mathrm{r}}=507 \mathrm{k}\mathrm{A}/\mathrm{m}$ and ${\mathrm{M}}_{\mathrm{S}}=670 \mathrm{k}\mathrm{A}/\mathrm{m}$, indicating the strongest magnetocrystalline anisotropy and domain-wall pinning. In contrast, the rest of the powders display progressively narrower hysteresis loops and a noticeable shift toward lower ${\mathrm{H}}_{\mathrm{c}}$ values. This behavior is due to the anisotropy constant and possible disruption of exchange interactions introduced by the substituent elements. The most heavily substituted sample (MS-Powder_4, x = 21) exhibits the weakest magnetic response with ${\mathrm{H}}_{\mathrm{c}}=480 \mathrm{k}\mathrm{A}/\mathrm{m}$, ${\mathrm{M}}_{\mathrm{r}}=450 \mathrm{k}\mathrm{A}/\mathrm{m}$ and ${\mathrm{M}}_{\mathrm{S}}=660 \mathrm{k}\mathrm{A}/\mathrm{m}$, confirming that excessive substitution reduces magnetic ordering and anisotropy energy, but even then, it is sufficiently close to the reference powder. Overall, the hysteresis behavior confirms that the substituted NdFeB powders retain strong ferromagnetic properties suitable for high-performance permanent magnet applications.

To further investigate the magnetic powders, Figure 8 illustrates the variations in the remanent magnetization (${\mathrm{B}}_{\mathrm{r}}$), coercivity (${\mathrm{H}}_{\mathrm{c}}$) and maximum energy product $(\mathrm{B}\mathrm{H}{)}_{\mathrm{m}\mathrm{a}\mathrm{x}}$ of the (Nd,Pr)_28–x(La,Ce)_x(Co,Zr)_~2B_~1Fe_bal powders as a function of (La,Ce) substitution level (x = 0, 5, 8, 21) at 300 K. All three magnetic parameters show a distinct decreasing trend with an increasing substitution r_sub ratio up to 75%—using Equation (1)—indicating a gradual deterioration of magnetic performance. The reference composition (x = 0) exhibits the highest ${\mathrm{B}}_{\mathrm{r}}$, H_C and (BH)_max values, confirming the superior intrinsic magnetic strength of the NdPrFeB system. As the content of (La,Ce) increases, both remanence and coercivity decrease significantly, which can be attributed to the lower magnetocrystalline anisotropy and reduced exchange interaction upon substitution of (Nd,Pr) with (La,Ce). The replacement of rare-earth elements with lighter, less anisotropic atoms weakens domain-wall pinning and lowers the anisotropy field, leading to the observed decline in coercivity and energy product [4]. At the highest substitution level (x = 21), the sample exhibits the lowest magnetic performance, demonstrating that excessive (La,Ce) incorporation compromises the hard-magnetic characteristics of NdFeB-type powders. These results confirm that while partial substitution can reduce material cost and improve thermal stability, it inevitably leads to a trade-off in magnetic strength.

3.4. Magnetic Characterization of Bonded Magnets

Figure 9 presents a three-dimensional magnetic field distribution map (M_z component) obtained from 3D magnetometry measurements of the bonded NdFeB sample. The surface plot represents the spatial variation in the normal magnetic flux density (Mz, in µT) across the measured area. The symmetric, bell-shaped profile centered on the Z-axis indicates uniform magnetization and well-aligned magnetic domains within the bonded structure. The high-intensity region (red zone) corresponds to the maximum magnetic flux density near the central axis of the magnet, while the gradual decrease toward the edges (blue–green regions) reflects the expected field decay due to geometric demagnetization effects and limited particle alignment near the boundaries.

This 3D mapping confirms that the bonded magnet exhibits a high degree of isotropy and consistent magnetic field distribution, implying effective powder compaction and uniform curing of the binder during fabrication. Any deviation from perfect symmetry would suggest inhomogeneous powder dispersion or uneven magnetic orientation; however, the smooth, continuous surface of the measured field demonstrates the structural integrity and magnetic uniformity of the composite. Such analyses provide valuable insight into the local field characteristics of bonded magnets and complement the hysteresis and B–H loop results by visualizing field homogeneity and intensity distribution across the sample.

Figure 10 represents the demagnetization (B–H) curve of the bonded NdFeB sample B₂ with for the substitution level (x = 5) and powder-to-binder ratios 96.5–3.5% (similar figures were obtained for all the samples A–D). The loop illustrates the typical behavior of hard-magnetic materials, showing a wide coercive field (H_cj) and (H_cb), characteristic of NdFeB systems [13]. Based on the data from

Table 4. Summary of magnetic properties of flux density remanence ${\mathrm{B}}_{\mathrm{r}}$ (T), maximum energy product BH_max (kJ/m³), and coercivity H_C (kA/m) of bonded NdFeB sample magnets measured with DX-2012H150 magnetometer at room temperature.

Sample	Powder	Ratio of Powder To Binder (%)	Br (T)	(BH)max (kJ/m3)	Hc (kA/m)
A1	MS-Powder_1	95–5%	0.63	57.1	659
A2	MS-Powder_1	96.5–3.5%	0.65	60.1	658
B1	MS-Powder_2	95–5%	0.60	54.7	591
B2	MS-Powder_2	96.5–3.5%	0.63	62.2	664
C1	MS-Powder_3	95–5%	0.59	55.5	675
C2	MS-Powder_3	96.5–3.5%	0.63	59.6	551
D1	MS-Powder_4	95–5%	0.54	43.6	527
D2	MS-Powder_4	96.5–3.5%	0.55	45.4	522

it is clearly demonstrated that both the substitution level and the binder fraction influence the magnetic response. As the substitution of (La,Ce) increases from x = 5 to x = 21, the slope of the second quadrant becomes steeper, and the (BH)_max value decreases, indicating reduced magnetic energy storage capability and weakened anisotropy. Samples with lower substitution levels retain higher remanence (B_r) and coercivity (H_C). Overall, the B–H curves confirm that compositional and microstructural factors control the bonded NdFeB magnets’ performance. The resulting magnetic parameters for each bonded sample (A–D) at room temperature are summarized in Table 4.

The measured remanence (B_r), coercivity (H_C), maximum energy product (BH)_max, saturation magnetization (${\mathrm{M}}_{\mathrm{s}}$), and remanent magnetization (${\mathrm{M}}_{\mathrm{r}}$) all demonstrate a consistent dependence on both powder composition and powder-to-binder ratio. The stable trends and small sample-to-sample variations confirm the reproducibility of the fabrication process.

As shown in Table 4, the substitution level x significantly affects the magnetic performance of the bonded NdFeB samples. Samples B₁ and B₂ (MS-Powder_2, x = 5) exhibit the best properties, reaching B_r = 0.63 T and (BH)_max = 62.2 kJ/m³. This enhanced behavior is attributed to the relatively low substitution level, which preserves the strong Fe–Fe exchange interactions and high magnetocrystalline anisotropy of the Nd–Fe–B phase [22]. In contrast, samples D₁ and D₂ (MS-Powder_4, x = 21) demonstrate the lowest magnetic performance, indicating that excessive substitution weakens magnetic coupling and reduces anisotropy. The intermediate C-series samples (MS-Powder_3, x = 8) exhibit magnetic values between those of the B and D compositions, showcasing the gradual impact of increasing substitution.

The powder-to-binder ratio also plays a significant role [15,23]. Increasing the powder content from 95–5% to 96.5–3.5% improves all magnetic parameters across all compositions. This enhancement arises from the higher magnetic phase fraction and improved magnetic connectivity between particles, which reduces demagnetizing effects associated with the non-magnetic binder. This trend is consistent with earlier studies on epoxy-bonded and compression-molded Nd–Fe–B magnets, which report that higher powder loading increases the magnetic phase fraction and enhances particle–particle interaction, resulting in improved flux continuity and reduced demagnetizing fields [13,14,23].

Despite the clear influence of substitution level on (BH)_max and Hc, the magnetization-related parameters (B_r, ${\mathrm{M}}_{\mathrm{s}}$, ${\mathrm{M}}_{\mathrm{r}}$) remain relatively stable across all samples. Even at the highest substitution level (x = 21), samples maintain magnetization values comparable to those of the low-substitution compositions (x = 5). This robustness indicates that, although high substitution reduces anisotropy and energy product, it does not markedly compromise the overall magnetization of the bonded NdFeB magnets.

The temperature variation of ${\mathrm{B}}_{\mathrm{r}}$ and ${\mathrm{H}}_{\mathrm{c}}$ of the bonded magnets was measured with a DX-2012H150 magnetometer, equipped with a temperature controller from T₀ = 25 °C (room temperature) to 100 °C. At each temperature step, full demagnetization (B–H) curves were recorded after thermal equilibration. The parameters extracted from the demagnetization curves were the intrinsic coercivity (${\mathrm{H}}_{\mathrm{c}\mathrm{j}}$), defined as the reverse field at which the magnetization becomes zero, and the coercivity of magnetic induction (${\mathrm{H}}_{\mathrm{c}\mathrm{b}}$), defined as the reverse field at which the magnetic induction reaches zero [4,5]. The remanence (B_r) and maximum energy product ((BH)_max) were also determined at each temperature.

To quantify the thermal stability, the reversible temperature coefficients α and β were determined by applying Formulas (2) and (3).

$$\mathsf{\alpha }={\displaystyle \frac{{\mathrm{B}}_{\mathrm{r}}\left(\mathrm{T}\right)-{\mathrm{B}}_{\mathrm{r}}\left({\mathrm{T}}_0\right)}{{\mathrm{B}}_{\mathrm{r}}\left({\mathrm{T}}_0\right)\left(\mathrm{T}-{\mathrm{T}}_0\right)}}\times 100\%$$

(2)

$$\mathsf{\beta }={\displaystyle \frac{{\mathrm{H}}_{\mathrm{c}\mathrm{j}}\left(\mathrm{T}\right)-{\mathrm{H}}_{\mathrm{c}\mathrm{j}}\left({\mathrm{T}}_0\right)}{{\mathrm{H}}_{\mathrm{c}\mathrm{j}}\left({\mathrm{T}}_0\right)(\mathrm{T}-{\mathrm{T}}_0)}}\times 100\%$$

(3)

where T₀ is the reference temperature (25 °C). Negative values of α and β indicate the expected decrease of remanence and coercivity with increasing temperature [24].

Based on the literature, the values of the α and β parameters are in the range of −0.10 to −0.12 and −0.35 to −0.46, respectively [25,26]. The calculated median temperature coefficients $\mathsf{\alpha }=-0.0166\%/°\mathrm{C}$ and $\mathsf{\beta }=-0.115\%/°\mathrm{C}$ of the samples A–D are smaller in magnitude compared to those typically reported for commercial pure NdFeB magnets, indicating superior thermal stability. The median values from samples A–D for each parameter and at each thermal step are shown in Figure 11.

The temperature dependence of B_r is mainly governed by the thermal decay of the spontaneous magnetization of the Fe(Co) sublattice [18]. The relatively weak temperature sensitivity (small |α|) observed here is attributed to improved high-temperature magnetization retention due to Co addition, as well as enhanced densification and phase uniformity introduced by multi-rare-earth substitution, which reduces the contribution of weakly coupled regions that typically amplify remanence degradation with temperature.

The temperature dependence of coercivity is generally stronger, as ${\mathrm{H}}_{\mathrm{c}\mathrm{j}}$ is highly sensitive to thermally activated magnetization reversal processes and to microstructural weak points that become increasingly effective at elevated temperatures. The relatively low median β coefficient value (−0.115%/°C) obtained for the present samples is attributed to the reduced temperature sensitivity of the coercive field combined with microstructural stabilization effects [22]. Multi-rare-earth substitution, together with Zr addition and melt-spinning processing, refines the microstructure and stabilizes grain boundaries, reducing the effectiveness of thermally activated weak links and domain-wall depinning [13,16]. As a result, the decrease of ${\mathrm{H}}_{\mathrm{c}\mathrm{j}}$(T) becomes more gradual, leading to enhanced coercivity stability.

The thermal dependence of the magnetic properties shown in Figure 11 demonstrates a gradual and consistent decrease of B_r, ${\mathrm{H}}_{\mathrm{c}\mathrm{j}}$, and (BH)_max with increasing temperature, without abrupt degradation up to 100 °C. This confirms that the bonded NdFeB magnets maintain excellent magnetic performance over a wide temperature range, outperforming standard commercial bonded NdFeB materials in terms of temperature stability [26].

4. Conclusions

The development and properties of axially compressed isotropic epoxy-bonded Nd–Fe–B magnets with partial rare-earth substitution were investigated. Substitution of up to 75% La–Ce at the RE site was achieved, substantially reducing the dependence on Nd–Pr. A scalable workflow was established for producing high-density (ρ up to 80%) NdFeB bonded magnets, encompassing powder preparation, mixing, high-pressure compaction, polymer bonding with powder loadings up to 96.5 wt.%, curing at 150 °C, and final magnetization. Structural characterization confirmed the predominance of the pure tetragonal 2–14–1 Nd₂Fe₁₄B phase, reaching up to 95% weight fraction even with increasing (La,Ce) substitution. Microscopy analysis indicates a uniform phase distribution and a well-formed interface between the powder and the binder. The resulting magnets demonstrated competitive magnetic performance and density, achieving (BH)max values of ~65 kJ/m³ and a coercivity of ~660 kA/m, along with superior thermal stability as verified by thermal analysis. These results were fully benchmarked against commercial counterparts. Despite these advancements, important challenges remain in optimizing the balance between magnetic performance and cost-effectiveness. Partial substitution of (Nd,Pr) with light rare-earth elements (LREs)—notably, La, and Ce—offers a promising strategy for reducing raw-material cost and mitigating supply-chain risks; however, the associated effects on the magnetic and microstructural behavior are complex and, in some cases, inconsistently reported in the literature.