betaFeedback

WorksPhysics and Astronomy

Including nanoparticle shape into macrospin models

Abstract

We investigate the feasibility of the macrospin approximation to account for the actual shape of soft magnetic nanoparticles (MNPs) with realistic geometries. Specifically focusing on magnetite, we use the superellipsoidal parametrisation to account for a variety of shapes, with a continuous interpolation from spherical to cubic morphologies, as well as different elongations. Our procedure consists of the direct comparison between angular-dependent hysteresis loops obtained by full micromagnetic simulations, with those produced by an extended Stoner-Wohlfarth (SW) model that incorporates both the intrinsic cubic magnetocrystalline anisotropy, and an effective uniaxial contribution arising from the particle elongation. The limits of validity of the macrospin description are approximately 10-60 nm for axial ratios r>1.5, and 20-60 nm for 1.0<r<1.5. These results establish a direct connection between nanoparticle morphology and effective macrospin parameters, demonstrating the suitability of the generalized SW model for describing the magnetic response of realistically shaped MNPs.

Introduction

The theory developed by E. C. Stoner and E. P. Wohlfarth [1] remains a cornerstone for understanding the behavior of magnetic nanoparticle (MNP) systems [2, 3, 4]. Their key aspect was to consider the magnetization of the particle to remain essentially uniform under applied and/or thermal fields. This implies that the particle size and shape lie within the single-domain range, so that the magnetization rotates coherently. In this regime, the particle’s magnetic moment can be regarded as a large macrospin. This formalism has enabled the development of advanced theoretical approaches to describe the magnetization dynamics of nanoparticle ensembles [5, 6, 7].

A major advantage of the SW model for interpreting experimental data is the direct correlation between the intrinsic particle parameters—saturation magnetization (MSM_S) and anisotropy constant (KK)—and measurable macroscopic quantities, particularly the coercive field (HCH_C) and remanence (MRM_R) obtained from magnetization–field [M(H)][M(H)] measurements. Well defined values can be obtained both for uniaxial- and cubic-anisotropy particles [8, 9]. Thus, experimental M(H)M(H) data can be used to infer the underlying anisotropy characteristics [10, 11]. Owing to its simplicity and physical transparency, the SW framework has been extensively applied in both experimental [12, 13] and theoretical [14, 15, 16] studies to interpret the magnetization behavior of MNP ensembles.

The objective of this work is to investigate the applicability of an extended SW model to account for actual particle shape of experimental samples. Thus, while Stoner and Wohlfarth considered particles with either uniaxial (KuK_u) or cubic (KcK_c) anisotropy constants in their pioneering work [1]—i.e. one type or the other—, recent works identified the need to account for both anisotropies simultaneously: the intrinsic magnetocrystalline term, always present, plus a uniaxial contribution accounting for shape effects [17, 18, 19, 20, 21]. From now on, we shall refer to such extended SW model as the Kc+KuK_c + K_u model.

Despite the reported works based on the more sophisticated Kc+KuK_c + K_u, the fact is that the behavior of MNPs is frequently analyzed within an effective uniaxial-anisotropy framework [22, 23], assuming that the shape-induced anisotropy contribution largely dominates over the intrinsic magnetocrystalline one [24, 25]. In other words, it is implicitly presumed that the particles deviate substantially from a spherical geometry. In practice, this assumption is not unfounded, as MNPs are rarely perfect spheres or cubes [26, 27, 28, 29, 30, 31, 32], as illustrated in Figure 1. The extent to which deviations from ideal geometry modify the overall magnetic response remains an open question.

Schematic representation of superellipsoidal shapes of different exponents $p$ and axial ratios $r=c/a$

Figure 1. Schematic representation of superellipsoidal shapes of different exponents pp and axial ratios r=c/ar=c/a, similar to those observed experimentally. TEM images modified from [27] with the permission of American Chemical Society.

The present study aims to establish the range of validity of the simplified uniaxial-only anisotropy approach in comparison with the more sophisticated Kc+KuK_c + K_u model. To this end, we have conducted a computational simulation of M(H)M(H) hysteresis loops of particles with different shapes and sizes, considering both a full description of the particle properties and the corresponding macrospin approximation. A micromagnetic approach is employed to accurately capture particle shape while maintaining computational feasibility for relatively large system sizes. The size dependence is analyzed to assess the applicability of the macrospin approximation. For simplicity, we restrict the analysis to spherical and cubic particles, as well as intermediate shapes described by a superellipsoidal geometry, including elongation effects.

The article is organized as follows. Sections II and III briefly describe the physical and computational frameworks employed. In Section IV A, we analyze in detail the influence of particle shape and elongation on the magnetic behavior of magnetite nanoparticles from a micromagnetic perspective, focusing on the determination of the size range for which the magnetization remains essentially uniform under an applied external field. In Section IV B, we assess the validity of the generalized Kc+KuK_c+K_u macrospin model through a direct comparison with micromagnetic simulations. This analysis includes the angular dependence of the main hysteresis parameters, orientation-averaged hysteresis loops, and a phase diagram identifying the range of applicability of the macrospin approximation. Finally, in Section IV C, we contrast these results with those obtained from the SW model considering only uniaxial or cubic anisotropy.

Physical model

In this work, we consider two complementary descriptions of the magnetic behavior of MNPs. On the one hand, we employ a full micromagnetic model, in which the particle shape is explicitly accounted for through a superellipsoidal parametrization. On the other hand, we use a macrospin model, where the particle is assumed to remain in a quasi-uniform magnetic state and the effect of elongation is incorporated effectively through the demagnetizing factors entering the energy terms. The material parameters used are listed in Table I.

ParameterSymbolValue
Saturation magnetizationMsM_s4.8×1054.8 \times10^5 A/m
Exchange stiffness constantAA1.1×10−111.1 \times10^{-11} J/m
Cubic anisotropy constantKcK_c−1.1×104-1.1 \times10^4 J/m3^3
Exchange lengthLexL_{\mathrm{ex}}8.78.7 nm

Table I. Material parameters used for magnetite (Fe3_3O4_4).

Micromagnetic model

The physical model employed in the micromagnetic simulations is the same as that used in Ref. [33]. Briefly, the micromagnetic model accounts for four energy contributions: cubic magnetocrystalline anisotropy, exchange, demagnetizing, and Zeeman energies. We consider nanoparticles exhibiting variations in both shape and elongation, modelled through the geometry of a generalized superellipsoid [34]. This parametrization enables a continuous description of morphologies ranging from spherical to cubic, as well as deviations from perfect isotropy through an axial ratio r≠1.0r \ne1.0, defined as the ratio between the longest and shortest particle axes (r=c/ar=c/a). Such variations provide a realistic representation of experimentally synthesized magnetite nanoparticles, which almost invariably display slight elongations and departures from ideal shapes, as shown in Figure 1. The superellipsoidal geometry is defined by [35, 36]

(xa)2p+(yb)2p+(zc)2p≤1,(1)\left(\frac{x}{a}\right)^{2p}+\left(\frac{y}{b}\right)^{2p}+\left(\frac{z}{c}\right)^{2p}\leq1, \tag*{(1)}

where aa, bb, cc are the particle semi-axes and pp is the shape exponent. In this study, we focus on three representative morphologies by varying the pp value: p=1p=1, p=3p=3, and p=100p=100, as illustrated in Figure 2.

Schemes of different example geometries studied in this work

Figure 2. Schemes of different example geometries studied in this work: p=1p=1 (sphere), p=3p=3, and p=100p=100 (cube), for the particular case of r=1.0r=1.0

The corresponding superellipsoidal volume (VV) is calculated from the analytical expression

Vabc=2ε1ε2B(ε12,ε1+1)B(ε22,ε2+22)=2p2B(12p,2p+12p)B(12p,p+1p),(2)\frac{V}{abc}=2\varepsilon_1\varepsilon_2 B\left(\frac{\varepsilon_1}{2},\varepsilon_1+1\right)B\left(\frac{\varepsilon_2}{2},\frac{\varepsilon_2+2}{2}\right) = \frac{2}{p^2}B\left(\frac{1}{2p},\frac{2p+1}{2p}\right)B\left(\frac{1}{2p},\frac{p+1}{p}\right), \tag*{(2)}

where ε1\varepsilon_1 and ε2\varepsilon_2 are the exponents of the general superellipsoidal parametrization, which in the present case reduce to ε1=ε2=1/p\varepsilon_1 = \varepsilon_2 = 1/p, and B(x,y)=Γ(x)Γ(y)/Γ(x+y)B(x,y) = \Gamma(x)\Gamma(y)/\Gamma(x+y) is the beta function [34, 37]. Throughout the first part of the work, the nanoparticle size is expressed as V1/3V^{1/3}. This normalization ensures that different geometries can be directly compared at equal volume, so that the observed differences in behavior arise solely from the particle shape and elongation.

Macrospin model

As introduced above, we also consider a simplified macrospin description, denoted as the Kc+KuK_c + K_u model, in which the nanoparticle is assumed to remain in a quasi-uniform magnetic state. This model combines two anisotropy contributions: (i) the cubic magnetocrystalline anisotropy intrinsic to magnetite, and (ii) an effective uniaxial shape anisotropy associated with particle elongation. In the presence of an applied field, the total energy density (J/m3^3) of the model is given by the sum of the cubic, uniaxial, and Zeeman terms:

E=Kc(α12α22+α22α32+α32α12)+Kusin⁡2θ−μ0MsHcos⁡(θH−θ),(3)E = K_c \left(\alpha_1^2\alpha_2^2 + \alpha_2^2\alpha_3^2 + \alpha_3^2\alpha_1^2\right) + K_u \sin^2\theta- \mu_0 M_s H \cos(\theta_H-\theta), \tag*{(3)}

where KcK_c and KuK_u are the cubic and uniaxial shape anisotropy constants, respectively; αi\alpha_i (i=1,2,3i = 1, 2, 3) are the direction cosines of the magnetization relative to the crystallographic axes; θ\theta and θH\theta_H are the angles that the easy axes forms with magnetization and applied field directions, respectively.

The uniaxial shape anisotropy constant is obtained as

Ku=μ02(Na−Nc)Ms2,(4)K_u = \frac{\mu_0}{2}(N_a-N_c)M_s^2, \tag*{(4)}

where NaN_a and NcN_c are the demagnetizing factors corresponding to the minor and major axes of an ellipsoid, respectively [38]. These factors depend only on the axial ratio r=c/ar = c/a of the nanoparticle (with the long axis oriented along the zz direction) and are expressed as

Nc(r)=1r2−1[rr2−1ln⁡(r+r2−1)−1],(5)N_c(r) = \frac{1}{r^2-1}\left[\frac{r}{\sqrt{r^2-1}}\ln\left(r+\sqrt{r^2-1}\right)-1\right], \tag*{(5)}
Na(r)=1−Nc(r)2.(6)N_a(r) = \frac{1-N_c(r)}{2}. \tag*{(6)}

The combined effect of the cubic and uniaxial anisotropy terms determines the energy landscape of the particle in absence of an external magnetic field. Figure 3 shows these energy surfaces in spherical coordinates for different axial ratios rr, for several particle elongations.

Schematics of particles with different elongations and corresponding energy landscapes

Figure 3. Schematics of particles with different elongations (in grey), and corresponding energy landscapes.

Computational details

To carry out the simulations, we employ the Object Oriented MicroMagnetic Framework (OOMMF) software [39], a finite-difference solver widely used for the study of static and dynamic magnetization processes in nanoscale systems. As discussed previously, we compare two complementary modelling approaches throughout this work: a full micromagnetic description and simplified macrospin models. The specific computational details of each approach are described below.

Since our aim here is to investigate how particle shape and elongation influence the magnetic response of MNPs under an applied field, we focus on the calculation of magnetic hysteresis loops. Considering that the simulations are performed for individual nanoparticles, the response of a disordered ensemble must be approximated by considering different orientations of the applied field with respect to the particle anisotropy axes. In the ideal case, this would require averaging over infinitely many field directions. Since such a procedure is not computationally feasible, we restrict the analysis to a selected set of representative field orientations, schematically shown in Figure 4. As indicated in the figure, the applied field is taken within the plane defined by φ=45∘\varphi= 45^\circ, while the polar angle θH\theta_H is varied. This choice is motivated by the cubic magnetocrystalline anisotropy of magnetite (Kc<0K_c < 0), for which the easy directions lie along the ⟨111⟩\langle111\rangle body diagonals of the cubic lattice. By selecting φ=45∘\varphi= 45^\circ and sweeping θH\theta_H, the applied-field direction is constrained to a plane that intersects these diagonal easy axes and, depending on particle elongation, also crosses the effective minimum-energy direction resulting from the competition between cubic and uniaxial anisotropies. The elongation-dependent equilibrium easy direction obtained from the combined Kc+KuK_c + K_u energy is discussed in Appendix A.

Schematics of the applied field direction regarding an elongated cubic nanoparticle

Figure 4. Schematics of the applied field direction regarding an elongated cubic nanoparticle (p=100p = 100, r=1.2r = 1.2). The field lies in the φ=45∘\varphi= 45^\circ plane, while the θH\theta_H value is varied.

For each field value, the equilibrium magnetization configuration is obtained using the OOMMF Oxs_MinDriver energy minimization module, with a convergence criterion of stopping m×H×m=0.1\mathbf{m}\times\mathbf{H}\times\mathbf{m} = 0.1. To ensure numerical stability and avoid trapping in local minima, simulations were repeated with different initial magnetization states (uniform, random, and vortex-like), consistently converging to the same equilibrium configuration.

Micromagnetic simulations

In the micromagnetic framework, each nanoparticle is discretized into cubic cells of 1 nm, a value smaller than the exchange length of magnetite (see Table I), thus ensuring that the exchange interaction is properly resolved. To verify that this discretization does not bias the results, representative simulations were repeated using finer meshes, yielding equilibrium configurations within numerical accuracy. Since no appreciable changes were observed while the computational cost increased substantially, a cell size of 1 nm was adopted for all micromagnetic calculations reported here. For this case, the field direction is varied in steps of 5∘5^\circ in θH\theta_H.

Macrospin simulations

In the macrospin approach, the nanoparticle is represented by a single uniform magnetization vector μ=MsVm\boldsymbol{\mu} = M_s V \mathbf{m}, whose behavior is governed by the different energy contributions defined in the previous section (Eq. (3)). Taking advantage of the much lower computational cost of this approach, the macrospin hysteresis loops are calculated using a finer angular resolution of 0.1∘0.1^\circ in θH\theta_H.

Results

The results are presented in three stages. We first analyze, from a micromagnetic perspective, how particle shape and axial ratio influence the magnetic response of the nanoparticles under an applied field, with particular emphasis on identifying the size range over which the magnetization remains essentially quasi-uniform. Next, within this regime, we assess the validity of the Kc+KuK_c + K_u macrospin model through direct comparison with the micromagnetic simulations. Finally, we compare this description with simpler SW-type models.

Micromagnetic simulations

To assess the feasibility of using effective macrospin models to describe particles with different shapes and elongations, the first requirement is to verify that the magnetization behaves essentially uniform. That is to say, one must examine how particle size and shape affect the magnetization behavior.

To this end, we build upon our previous work [33], where we investigated the influence of particle shape, size, and elongation on the spontaneous magnetization of MNPs using the same superspheroidal geometry. In that study, performed in the absence of an external applied magnetic field, we found that both shape and elongation play a decisive role in determining the magnetic behavior of the system, particularly around what we shall refer to as the critical volume (VcV_c)-below which the magnetization remains essentially uniform, preserving the macrospin-like character of the particle.

In the present work, we extend our previous analysis to the case where an external magnetic field is applied. By studying hysteresis loops for different geometries (p=1,3,100p = 1, 3, 100) and elongations, we aim to evaluate how the field modifies the magnetic response and whether it affects the limits of validity of the quasi-uniform regime. Since this point is not the main focus of the present work, the corresponding analysis is provided in Appendix B. As discussed there, when an external magnetic field is applied, the critical upper-size limits are found to lie in the range V1/3=50−55V^{1/3} = 50 - 55 nm for the three geometries considered and for elongations between r=1.0r = 1.0 and r=1.2r = 1.2. These values are comparable to those observed in the absence of a field [33], indicating that the applied field does not significantly alter the upper limit of the quasi-uniform regime. Therefore, we can safely select particle sizes within this range to analyze in detail the influence of small variations in shape and size on magnetic behavior.

To illustrate the variations in magnetic behavior arising from changes in particle shape and slight elongations, Figure 5 presents the hysteresis loops obtained for different directions of the applied magnetic field. Results are shown for the three geometries studied (p=1,3,100p = 1, 3, 100) and three representative elongations (r=1.0r = 1.0, 1.11.1, and 1.21.2), for particles with V1/3=40V^{1/3} = 40 nm, that is, within the range corresponding to essentially uniform behavior. For a more complete analysis including the full angular dependence, see Appendix § 1.3.3.

Example hysteresis loops for a superellipsoidal MNP with different shape parameters and elongations

Figure 5. Example hysteresis loops for a V1/3=40V^{1/3}=40 nm superellipsoidal MNP with different shape parameters (columns; p=1,3,100p=1, 3, 100) and elongations (rows; r=1.0,1.1,1.2r=1.0, 1.1, 1.2), for the field applied at representative directions θH=0∘\theta_H=0^\circ, 45∘45^\circ, and 90∘90^\circ.

In Figure 5 it is observed that, in first approximation, variations in the axial ratio of the particles have a more significant influence on the magnetic response (more pronounced for higher pp) than the particle shape in itself, which only in the transition from p=1p=1 to p=3p=3 shows a noticeable effect; otherwise, the overall form of the hysteresis loops remains largely unchanged across the different values of the shape parameter pp. In other words, within the range of morphologies considered here, the magnetic response is essentially similar regardless of whether the particle geometry is closer to a sphere or to a cube. This suggests that particle elongation is the dominant factor for the magnetic behavior, with the shape playing a second-order correction. For a more detailed analysis, see Appendix § 1.3.3.

Kc+KuK_c + K_u model vs. Micromagnetics

In this section, we assess the validity of the effective Kc+KuK_c + K_u macrospin description by directly comparing its predictions with full micromagnetic simulations. The comparison is carried out in three steps. First, we analyze the angular dependence of the main hysteresis parameters for representative quasi-uniform particles. Next, we compare orientation-averaged hysteresis loops, which provide a more realistic description of particle ensembles. Finally, we extend the analysis to the full size–elongation parameter space in order to identify the range of applicability of the macrospin approximation.

Comparison of angular dependence

As a representative case, we focus on particles with V1/3=40V^{1/3}=40 nm and r=1.1r=1.1, well below the critical threshold, and compare the resulting magnetic parameters (HcH_c, MrM_r, and hysteresis loop area) obtained from micromagnetic simulations to those obtained from the Kc+KuK_c + K_u model. The results are shown in Figure 6.

Comparison of hysteresis parameters plotted against $\theta_H$: coercive field, normalized remanent magnetization, and loop area for $p=1$, $p=3$, and $p=100$, with micromagnetic simulation symbols and $K_c+K_u$ solid lines

Figure 6. Comparison of hysteresis parameters (HcH_c, MrM_r, and loop area) obtained from micromagnetic simulations (symbols) and the Kc+KuK_c+K_u model (solid lines) for particles with p=1p=1, 33, and 100100, r=1.1r=1.1, and V1/3=40V^{1/3}=40.

The results plotted in Figure 6 indicate an excellent agreement between the micromagnetic simulations and the Kc+KuK_c + K_u model predictions for the ellipsoidal case (p=1p=1), as both models reproduce the same angular dependence of HcH_c, MrM_r, and hysteresis loop area. The small discrepancies observed between the two models arise from the finite discretization used in the micromagnetic calculations: with a spatial discretization of 1 nm, it is not always possible to construct particles with exactly the same volume and axial ratio r=1.10r=1.10. In practice, rr varies slightly between 1.08 and 1.12, which induces minor changes in the effective anisotropy field with respect to the idealized Kc+KuK_c + K_u case.

The results for the superellipsoidal geometries (p=3p=3 and p=100p=100) are also in very good agreement with the Kc+KuK_c+K_u model predictions, although the HcH_c values below 35∘35^\circ are slightly lower than those predicted by the macrospin model, again due to the discretization effects mentioned above. For a detailed analysis of the HcH_c vs. θH\theta_H features the reader is referred to Appendix A.

Overall, these results show that the Kc+KuK_c+K_u approach remains valid across the full range of shapes considered here, from nearly spherical (p=1p=1) to nearly cubic (p=100p=100) particles, reaffirming that particle shape does not play a critical role in the magnetic behavior within the quasi-uniform regime. Although we only show the case r=1.1r=1.1 here, the cases r=1.0r=1.0 and r=1.2r=1.2 exhibit the same good correspondence between micromagnetic simulations and the macrospin model, as will be shown more generally in the following subsection.

Comparison of orientation-averaged hysteresis loops

To perform a more meaningful comparison, we now compare the averaged hysteresis loops obtained from micromagnetic simulations with those predicted by the effective Kc+KuK_c+K_u macrospin model. These averaged loops provide a more representative basis for comparison, as they reflect the expected magnetic behavior of an ensemble of MNPs with a quasi-random distribution of anisotropy-axis orientations.

The normalized averaged magnetization M‾(H)\overline{M}(H), as a function of the applied field HH, is computed as:

M‾(H)=∫0π/2M(H,θH)⋅sin⁡(θH) dθH∫0π/2sin⁡(θH) dθH(7)\overline{M}(H)=\frac{\int_{0}^{\pi/2}M(H,\theta_H)\cdot\sin(\theta_H)\,d\theta_H}{\int_{0}^{\pi/2}\sin(\theta_H)\,d\theta_H} \tag*{(7)}

It is worth noting that this expression does not correspond to a fully isotropic (spherical) distribution, which would be computationally prohibitive, as explained in Sec. III. Instead, we consider a uniform distribution over the polar angle θH\theta_H, while fixing the azimuthal angle at φ=45∘\varphi=45^\circ. This simplification provides a reasonable representative approximation to the angular averaging of the hysteresis response. The results are shown in Figure 7.

Orientation-averaged hysteresis loops for axial ratios $r=1.0$, $r=1.1$, and $r=1.2$, comparing micromagnetic simulations with the $K_c+K_u$ model

Figure 7. Orientation-averaged hysteresis loops comparing micromagnetic simulations and the Kc+KuK_c+K_u model for particles with V1/3=40V^{1/3}=40 nm and axial ratios r=1.0r=1.0, 1.11.1, and 1.21.2.

In Figure 7 it is observed that the axial ratio has a pronounced influence on the hysteresis loop shape, notably affecting both coercivity and remanence. Despite these variations, the agreement between micromagnetic simulations and Kc+KuK_c + K_u model remains excellent for all cases considered. This result confirms that the proposed framework accurately captures the magnetic response of superellipsoidal nanoparticle ensembles, even in the presence of moderate elongations or departures from perfect symmetry.

Generalization: Phase Diagram

Up to this point, our analysis has been limited to a narrow range of particle sizes and axial ratios, leaving open the question of whether the correspondence between micromagnetic simulations and the effective macrospin description persists across the full parameter space. The aim of this section is therefore to assess the validity range of the Kc+KuK_c + K_u model and to identify the regions where ellipsoidal MNPs can still be accurately described within the macrospin approximation.

Based on the results of Sec. IV B, where a nearly identical behavior was observed for the different geometries considered, we restrict the following analysis to ellipsoidal particles (p=1p = 1), assuming that these results are representative of other superellipsoidal shapes as well.

To explore the full parameter space, it is not feasible to compare the complete hysteresis loops for all particle sizes and axial ratios considered. It is therefore necessary to restrict the analysis to a representative parameter. For this purpose, we focus on the coercive field, which provides a sensitive measure of the deviations between the micromagnetic simulations and the Kc+KuK_c + K_u model. This choice is further justified by the fact that other quantities, such as the remanence, exhibit only minor variations even when magnetization coherence is partially lost, whereas the coercive field is much more sensitive to the onset of non-uniform behavior (see Appendix D and Figure 16). The deviation is then evaluated using the mean absolute percentage error (MAPE) [40]:

Mean deviation $\Delta M_r$ between the micromagnetic results and the $K_c + K_u$ model as a function of the minor axis size $d$ and the axial ratio $r$

Figure 16. Mean deviation ΔMr\Delta M_r between the micromagnetic results and the Kc+KuK_c + K_u model as a function of the minor axis size dd and the axial ratio rr.

ΔHc=100×1n∑i=1n∣Hc,imicro−Hc,imacroHc,imacro∣(8)\Delta H_c = 100 \times\frac{1}{n} \sum_{i=1}^{n} \left| \frac{H_{c,i}^{\mathrm{micro}} - H_{c,i}^{\mathrm{macro}}}{H_{c,i}^{\mathrm{macro}}} \right| \tag*{(8)}

This expression allows us to quantify the discrepancy in relative terms with respect to the Kc+KuK_c + K_u model prediction. However, it presents an important limitation: when the angle approaches θH=90∘\theta_H = 90^\circ, particularly for nanoparticles with large aspect ratios, the theoretical HcH_c becomes very small. In this regime, even minor absolute differences can lead to large relative deviations, artificially amplifying the apparent error. To avoid this distortion, we excluded from the analysis the values corresponding to θH=90∘\theta_H = 90^\circ.

Throughout this section, the particle size is expressed in terms of the minor-axis length, denoted by dd. This choice is convenient because fixing dd and varying the aspect ratio rr enables a systematic construction of the phase diagram. In contrast, using the effective magnetic size V1/3V^{1/3} would introduce an additional complication: due to the spatial discretization of 1 nm31\,\mathrm{nm}^3 cells in the simulations, the represented volume changes discontinuously, making it impossible to maintain a constant V1/3V^{1/3} when modifying rr. The results are shown in Figure 8.

Phase diagram of the mean deviation in coercive fields

Figure 8. Mean deviation ΔHc\Delta H_c between coercive fields from micromagnetic simulations and the Kc+KuK_c + K_u model as a function of dd and rr. The blue contour identifies the region where deviations remain under 5%.

The diagram displayed in Figure 8 represents the mean deviation ΔHc\Delta H_c across the explored parameter space, where black dots indicate the specific (d,r)(d,r) combinations explicitly simulated. The color map is obtained by interpolation to guide the visualization of trends. The micromagnetic results deviate from the Kc+KuK_c + K_u model predictions for small sizes, particularly below d=10 nmd = 10\,\mathrm{nm}, where the discrepancy exceeds 5% (as outlined by the blue contour). This deviation does not stem from a breakdown of the single-domain assumption, but rather from numerical artifacts associated with the finite spatial discretization. As illustrated in Figure 9, when the particle size approaches d=10 nmd = 10\,\mathrm{nm}, the geometry becomes poorly resolved by the 1-nm cell size used in the mesh, leading to spurious variations in the computed coercivity. In this size regime, an atomistic description would be necessary to accurately capture the magnetic behavior.

Discretization effects in spherical and elongated nanoparticles for different sizes

Figure 9. Discretization effects in spherical (top row) and elongated (bottom row) nanoparticles for different sizes d=10,20,30d = 10, 20, 30 nm.

At the opposite end of the size range, deviations above 5% are also observed for d>60 nmd > 60\,\mathrm{nm}. In this case, the origin is physical: micromagnetic simulations indicate the onset of non-uniform magnetization states, implying that the particles no longer behave as coherent macrospins and cannot be accurately described within the Kc+KuK_c + K_u framework.

SW model vs. Micromagnetics

As demonstrated in the previous section, the Kc+KuK_c + K_u model, combining cubic magnetocrystalline and uniaxial shape anisotropies, provides an excellent description of the micromagnetic behavior of nanoparticles below 60 nm, regardless of their specific geometry.

In this section, we aim to place these results within the general context of the classical SW model, which, often formulated in terms of a uniaxial anisotropy, remains the most widely used theoretical approach in the literature for describing monodomain MNPs [41]. We therefore compare our micromagnetic results with the predictions of this classical SW framework in order to assess its suitability for describing the magnetic behavior of IONPs. For completeness, since magnetite IONPs intrinsically exhibit cubic magnetocrystalline anisotropy, we further include a comparison with a purely cubic variant of the SW model, in which the nanoparticle is assumed to be subjected exclusively to the cubic anisotropy contribution.

Uniaxial SW model vs. Micromagnetics

We construct the same type of phase diagram introduced in Sec. IV B 3, but taking now the uniaxial SW model as the theoretical reference. The resulting deviations are summarized in Figure 10. For all sizes and axial ratios considered, the deviations with respect to the SW model exceed 10%. This result underscores the non-negligible influence of cubic magnetocrystalline anisotropy on the magnetic behavior of magnetite nanoparticles, contrary to the common assumption that shape anisotropy alone suffices to describe their response [24, 25]. As in the case of the Kc+KuK_c+K_u model comparison discussed in Sec. IV B 3, deviations become more pronounced for the smallest particles (where numerical discretization effects become significant) and for the largest ones (where quasi-uniform magnetization is lost). It is also worth noting that particles with nearly spherical shapes (r≈1.0r \approx1.0) display particularly large discrepancies with respect to the uniaxial SW model. In these cases, the contribution from shape anisotropy weakens, thus allowing the cubic magnetocrystalline anisotropy to dominate the magnetic energy landscape and substantially

Mean deviation between coercive fields from micromagnetic simulations and the uniaxial SW model

Figure 10. Mean deviation ΔHc\Delta H_c between coercive fields from micromagnetic simulations and the uniaxial SW model as a function of dd and rr. The blue contour highlights the region with deviations below 15%, while all deviations are seen to remain above 10%.

affect the reversal process [42].

This behavior can also be interpreted in light of the anisotropy energy landscapes shown in Figure 3. For r=1.0r = 1.0, the anisotropy energy is purely cubic, giving easy axes along the crystallographic diagonals, as evidenced by the appearance of minima at intermediate directions. As the axial ratio increases, the contribution from uniaxial shape anisotropy becomes dominant, and the energy minima progressively align toward the elongated axis. Consequently, the system’s behavior approaches that predicted by the uniaxial SW model, explaining the gradual reduction of the coercive-field deviation to below 15% in the region highlighted in Figure 10 with a blue contour.

Cubic SW vs. Micromagnetics

After demonstrating the limited capability of the KuK_u-only SW model to describe the behavior of superellipsoidal MNPs, we now compare the micromagnetic results with the predictions of a purely cubic SW model, in which the nanoparticle is treated as a single-domain macrospin subject exclusively to cubic magnetocrystalline anisotropy. The resulting deviations between both approaches, computed according to Eq. (8), are presented in Figure 11.

Mean deviation contour plot

Figure 11. Mean deviation ΔHc\Delta H_c between coercive fields from micromagnetic simulations and the cubic SW model as a function of dd and rr. The blue contour highlights the region with deviations below 15%.

The results shown in Figure 11 reveal that nanoparticles with axial ratios differing from r=1.00r = 1.00 display deviations exceeding 30%, which increase further as elongation becomes more pronounced. Therefore, the cubic SW model is only capable of reproducing the magnetic behavior of nearly-spherical nanoparticles, for which uniaxial anisotropy is strictly absent. However, such ideal spherical systems are practically unattainable under experimental conditions [26, 27], strongly limiting the applicability of the KcK_c-only SW framework for realistic magnetite MNPs.

Conclusions

We have studied the feasibility of an extended Stoner–Wohlfarth model, denoted as the Kc+KuK_c+K_u model, to describe the magnetic behavior of magnetite nanoparticles with different shapes and axial ratios, defined through the superellipsoidal geometry. In this framework, the nanoparticle is treated as a single magnetic moment subject to two anisotropy contributions: the intrinsic cubic magnetocrystalline anisotropy of magnetite and an effective uniaxial term arising from particle elongation. In order to carry out a meaningful comparison between this reduced description and full micromagnetic simulations, it is first necessary to ensure that the nanoparticles remain within the quasi-uniform regime. For this reason, the first step of this study was to determine the critical size, which defines the boundary between quasi-uniform and non-uniform magnetic configurations, for our MNPs under the application of an external magnetic field. The results show that, for r=1.1r=1.1, the critical upper-size limits for all geometries lie in the range 50–55 nm, independently of the precise particle shape, in good agreement with previous zero-field results reported in Ref. [33].

Within this essentially-uniform regime, our micromagnetic results show that the axial ratio plays the dominant role in the magnetic response, significantly affecting the coercivity, remanence, and magnetization reversal process under the applied field. By contrast, the particle shape itself, from nearly spherical to nearly cubic morphologies, does not appear to play a relevant role in the hysteresis behavior, acting at most as a second-order correction. In agreement with this, we observe a very good correspondence between micromagnetic simulations and the Kc+KuK_c+K_u model across the full range of superellipsoidal geometries considered here. This confirms that nanoparticles in the quasi-uniform regime can be accurately described within this reduced framework, regardless of their detailed morphology.

Given the widespread use of the uniaxial Stoner–Wohlfarth description in the literature for single-moment particle systems, we compared our micromagnetic results with two simplified SW models: a uniaxial-only case (KuK_u-only) and a cubic-only case (KcK_c-only). In both limits, deviations from the micromagnetic reference are substantial. The KuK_u-only model systematically fails to capture the intrinsic contribution of cubic magnetocrystalline anisotropy, even in particles where elongation is significant, whereas the KcK_c-only model is accurate only for nearly spherical particles and rapidly becomes inaccurate as the axial ratio increases. Thus, neither simplified SW description is sufficient to account for the magnetic behavior of realistic magnetic nanoparticles. By contrast, the Kc+KuK_c+K_u model provides a minimal yet quantitatively accurate description of the magnetic response within the quasi-uniform regime.

Beyond validating the Kc+KuK_c+K_u description in the quasi-uniform regime, we establish a direct, quantitative link between particle geometry and the effective anisotropy parameters entering the macrospin model. This link provides a practical route for connecting micromagnetic simulations with analytical descriptions commonly used to interpret experimental data, particularly in realistic systems where nanoparticles rarely exhibit perfect shapes or strict monodispersity. The main contribution of the present work is therefore not simply the combination of cubic and uniaxial anisotropies, which has already been explored in previous studies, including our own work [42, 43, 44, 45]. Rather, the novelty is a systematic, quantitative correlation between particle shape and elongation and the effective parameters of the Kc+KuK_c+K_u model. By showing that these parameters can be directly linked to realistic geometrical features of the nanoparticles, the present study moves beyond a purely phenomenological use of effective anisotropy terms and provides a physically motivated reduced description of magnetite MNPs within the quasi-uniform regime. More broadly, this geometry-anisotropy correspondence opens the way to the development of hybrid spin–particle dynamic models [46, 47, 48, 49, 50, 51, 52, 53], in which geometrical constraints are expected to play a central role in determining the relative ordering and collective response of magnetic moments [35].

The present study also suggests clear directions for future research. A first step would be a more detailed investigation of the small-size regime, where the applicability of the Kc+KuK_c+K_u model becomes limited and an atomistic treatment is required [54]. Other extensions could include the incorporation of surface anisotropy [18] and its interplay with finite-size effects [55], as well as the study of branch crossing [56] or doping effects [57]. Together, these efforts would further refine our understanding of the interplay between geometry, anisotropy, and magnetization dynamics in real nanoscale systems.

Acknowledgments

We acknowledge financial support by Spanish Ministerio de Ciencia, Innovación y Universidades through projects PID2019-109514RJ-100, PID2024-157172NB-I00 and CNS2024-154574, ”ERDF A way of making Europe”, by the ”European Union”, and Catalan DURSI (2021SGR0032). Xunta de Galicia is acknowledged for the projects ED431F 2022/005 and ED431B 2023/055. AEI is also acknowledged for the Ramón y Cajal grant RYC2020-029822-I that supports the work of D.S. We acknowledge the Centro de Supercomputación de Galicia (CESGA) for computational resources.

Behavior of the Coercivity Peak

In Sec. IV B, we reported the appearance of a pronounced peak in the coercive field as a function of the applied field angle θH\theta_H. We argued that this peak is associated with the direction of the easy magnetization axis. To substantiate this interpretation, we determine here the angle corresponding to the minimum of the anisotropy energy (θmin⁡\theta_{\min}), and compare it with the angles at which the coercivity reaches its maximum. For improved resolution, this comparison is performed using the results obtained from the combined Kc+KuK_c + K_u anisotropy model.

We begin by considering the expression for the total anisotropy energy density of the system:

Eani(θ,ϕ)=Esh(θ)+Ecub(θ,ϕ)=Kusin⁡2θ+Kc(αx2αy2+αy2αz2+αz2αx2),(9)\begin{aligned} E_{\mathrm{ani}}(\theta,\phi) &= E_{\mathrm{sh}}(\theta) + E_{\mathrm{cub}}(\theta,\phi) \\ &= K_u\sin^2\theta+ K_c\left(\alpha_x^2\alpha_y^2+\alpha_y^2\alpha_z^2+\alpha_z^2\alpha_x^2\right), \tag*{(9)} \end{aligned}

where αx=sin⁡θcos⁡ϕ\alpha_x=\sin\theta\cos\phi, αy=sin⁡θsin⁡ϕ\alpha_y=\sin\theta\sin\phi, and αz=cos⁡θ\alpha_z=\cos\theta, and KuK_u is obtained from Eq. (4). Simplifying for ϕ=45∘\phi=45^\circ:

Eani(θ)=(Ku+Kc)sin⁡2θ−34Kcsin⁡4θ.(10)E_{\mathrm{ani}}(\theta)=(K_u+K_c)\sin^2\theta-\frac{3}{4}K_c\sin^4\theta. \tag*{(10)}

Taking the derivative with respect to θ\theta, the condition for the minimum is:

θmin⁡=arcsin⁡(2(Ku+Kc)3Kc).(11)\theta_{\min}=\arcsin\left(\sqrt{\frac{2(K_u+K_c)}{3K_c}}\right). \tag*{(11)}

In Figure 12(a), we present the angular dependence of HcH_c for different axial ratios, highlighting the emergence of a well-defined maximum. Figure 12(b) compares the angular positions of these HcH_c peaks with the theoretical predictions for the energy minimum obtained from Eq. (11). A very good agreement is observed between the angle of maximum HcH_c and the direction of the easy magnetization axis. Furthermore, increasing the axial ratio leads to a systematic shift of the energy-minimizing angle, θmin⁡\theta_{\min}, toward the elongation direction of the particle. This behavior reflects the increasing dominance of shape anisotropy over magnetocrystalline anisotropy, as discussed in Sec. IV A (see, in particular, Figure 3). A notable change in behavior is observed for r=1.20r=1.20, where a discontinuity appears in the angular dependence of the coercivity at θmin⁡=10.05∘\theta_{\min}=10.05^\circ. This discontinuity is associated with a qualitative change in the shape of the hysteresis loops, as illustrated in Figure 12(c). Remarkably, the angle at which this coercivity discontinuity occurs closely matches the theoretical prediction for the energy-minimizing direction, as shown in Figure 12(b).

Dependence of the coercive field on field angle and comparison of theoretical and numerical results

Figure 12. (a) Dependence of the coercive field HcH_c on the field angle θH\theta_H for different axial ratios. The values where HcH_c is maximum are highlighted. (b) Comparison between the theoretical and numerically obtained values of the angle corresponding to the minimum-energy direction, θmin⁡\theta_{\min}, as a function of the axial ratio rr. (c) Hysteresis loops for the case r=1.20r = 1.20 in the angular region where the discontinuity is observed. All results shown were obtained from the Kc+KuK_c + K_u model.

Quasi-Uniform Range

To complement the discussion presented in Sec. IV A, here we provide the detailed analysis used to determine the critical sizes of micromagnetic MNPs under an applied magnetic field.

To perform a systematic study, we represent the characteristic parameters extracted from each hysteresis loop (namely HcH_c, MrM_r, and loop area) as a function of the field orientation angle θH\theta_H. In this analysis, we focus on MNPs with an axial ratio of r=1.1r=1.1, which corresponds to a typical elongation encountered in the synthesis of this type of nanoparticles [26]. Minor variations in rr between shapes arise from discretization effects. We then investigate how the magnetic parameters of the hysteresis loops evolve with MNP size, with the goal of identifying signatures of the transition between quasi-uniform and nonuniform magnetization reversal modes.

Figure 13 shows clear changes in the trends, most notably the progressive disappearance of the HcH_c peak around 30∘30^\circ. These variations are strongly influenced by the effective particle size. For spherical particles (p=1p=1), this inflection becomes evident near V1/3=55V^{1/3}=55 nm. For p=3p=3, deviations appear already at 5555 nm and become more pronounced at 6060 nm, whereas for p=100p=100 the transition occurs around V1/3=70V^{1/3}=70 nm. Although these changes do not by themselves demonstrate a breakdown of coherent rotation, they do indicate the onset of a size-driven transition between distinct magnetic regimes.

Characteristic parameters for three nanoparticle shapes

Figure 13. Characteristic parameters, HcH_c, MrM_r, and loop area as functions of the angle θH\theta_H, for three shapes: p=1p = 1 (a), p=3p = 3 (b), and p=100p = 100 (c), all with an axial ratio r=1.1r = 1.1, and varying the nanoparticle size V1/3V^{1/3}. For visual clarity, data points are connected by lines, although this interpolation may not represent the exact trend at intermediate angles.

To assess whether the observed trends in HcH_c are associated with the onset of nonuniform magnetization, we analyze the evolution of the total magnetization modulus ∣M∣|\mathbf{M}| throughout the hysteresis cycle. In the coherent (uniform) regime, ∣M∣/Ms|\mathbf{M}|/M_s remains close to unity at all fields. A systematic reduction of this quantity below unity signals a loss of uniform alignment of the local magnetic moments and the emergence of nonuniform magnetization states. Figure 14 shows the magnetization modulus for MNPs with sizes V1/3=50V^{1/3}=50 nm (top row) and V1/3=55V^{1/3}=55 nm (bottom row).

Variation of normalized magnetization modulus along hysteresis cycles

Figure 14. Variation of the normalized magnetization modulus ∣M∣/Ms|\mathbf{M}|/M_s along the hysteresis cycle for MNPs with V1/3=50V^{1/3} = 50 nm (top row) and V1/3=55V^{1/3} = 55 nm (bottom row). Each column corresponds to a different shape: (a,d) p=1p = 1 (sphere), (b,e) p=3p = 3, and (c,f) p=100p = 100 (cube).

Overall, the results indicate that V1/3=55V^{1/3}=55 nm defines a threshold above which deviations from uniform magnetization become evident. This transition is more pronounced for more rounded particle shapes. For V1/3=50V^{1/3}=50 nm (top row of Figure 14), the system still exhibits a quasi-uniform magnetization, with ∣M∣/Ms≈1|\mathbf{M}|/M_s\approx1, justifying the restriction of the detailed analysis in the main text to this size range, where the macrospin approximation is expected to remain valid.

In our previous work performed in the absence of an external field [33], the critical volume VcV_c was found to be approximately 49 nm for spherical MNPs (p=1p = 1) and 53 nm for cubic ones (p=100p = 100), for an axial ratio r=1.1r = 1.1. The present results, obtained under an applied field, are very similar, indicating that the field does not significantly modify the critical size VcV_c.

Angular dependence of the hysteresis loops:additional results

Figure 15 shows the angular dependence of the hysteresis loops for the different particle shapes and axial ratios considered in this work. The main conclusion is that the loop morphology is governed primarily by the axial ratio rr, whereas the dependence on the shape parameter pp is comparatively weak. In practice, for a fixed value of rr, the loops obtained for p=1p = 1, 3, and 100 remain very similar, confirming that particle shape plays only a secondary role within the quasi-uniform regime. By contrast, the loop shape evolves systematically with the field angle θH\theta_H and depends strongly on particle elongation.

Angular dependence of hysteresis loops for superellipsoidal nanoparticles

Figure 15. Angular dependence of the hysteresis loops obtained from micromagnetic simulations for superellipsoidal nanoparticles with different shape parameters (p=1p = 1, 3, 100) and axial ratios (r=1.0r = 1.0, 1.1, and 1.2), all with V1/3=40V^{1/3} = 40 nm. Each row displays the hysteresis loops for different field directions θH\theta_H.

For r=1.0r = 1.0, where only cubic magnetocrystalline anisotropy is present, the angular dependence departs significantly from the uniaxial SW picture. At small θH\theta_H, the loops present intermediate steps which progressively disappear until θH∼40∘\theta_H \sim40^\circ. For larger angles, the loops become more square, with an increase in HcH_c up to θmin⁡\theta_{\min}, beyond which HcH_c decreases, as shown in Figure 15. This behavior reflects the presence of multiple equivalent easy axes and competing local minima in the anisotropy energy landscape; the switching fields result from a sequence of instabilities associated with different crystallographic directions.

For finite elongations (r=1.1,1.2r = 1.1, 1.2), the addition of a shape-induced uniaxial anisotropy term KuK_u progressively lifts the degeneracy of the cubic energy landscape and introduces a dominant easy axis. For r=1.1r = 1.1, the loops remain nearly square for θH<θc∼33∘\theta_H < \theta_c \sim33^\circ, and then develop intermediate steps whose amplitude increases as HcH_c decreases. These features indicate that the cubic contribution still perturbs the effective energy landscape.

For larger elongation (r=1.2r = 1.2), the uniaxial shape anisotropy dominates over the cubic term. However, the maximum HcH_c still occurs at Hc∼10∘H_c \sim10^\circ rather than at 0∘0^\circ, as expected for the KuK_u-only case. Moreover, the steps that appear beyond this maximum occur near H=0H = 0 and increase in magnitude with θH\theta_H, eventually leading to butterfly-like loops for θH≳75∘\theta_H \gtrsim75^\circ, where HcH_c vanishes. These features are fully consistent with the angular dependence of HcH_c discussed in Figure 13, where the persistence of a shifted maximum and the suppression of non-monotonic behavior with increasing rr reflect the progressive crossover from a regime dominated by cubic anisotropy to one that approaches an effective uniaxial response.

Comparison of MrM_r values

To complement the analysis presented in the main text, we evaluated the deviation between the remanence values obtained from micromagnetic simulations and those predicted by the Kc+KuK_c + K_u model. The same MAPE definition introduced in Eq. (8) was used, but now applied to the remanence MrM_r:

ΔMr=100×1n∑i=1n∣Mr,imicro−Mr,imacroMr,imacro∣.(12)\Delta M_r = 100 \times\frac{1}{n} \sum_{i=1}^{n} \left| \frac{M_{r,i}^{\mathrm{micro}} - M_{r,i}^{\mathrm{macro}}}{M_{r,i}^{\mathrm{macro}}} \right|. \tag*{(12)}

As shown in Figure 16, the deviation in remanence, ΔMr\Delta M_r, remains below 2% across nearly the entire range of particle sizes and elongations analyzed. This demonstrates that the remanent magnetization is essentially preserved even when the magnetization begins to lose coherence. In contrast, the coercive field exhibits a much stronger dependence on the onset of non-uniform magnetization states.

Therefore, the coercive field HcH_c serves as a more sensitive indicator of the emergence of non-uniform reversal processes, which supports its use as the primary parameter for constructing the phase diagram presented in the main text.

References

References

  1. [1]E. C. Stoner and E. P. Wohlfarth, Philosophical Transactions of the Royal Society of London. Series A, Mathematical and Physical Sciences 240, 599 (1948).
  2. [2]C. Tannous and J. Gieraltowski, European journal of physics 29, 475 (2008).
  3. [3]C. Tannous and J. Gieraltowski, Physica B 403, 3563 (2008).
  4. [4]B. Doyle, IEEE Trans. Magn. 45, 7 (2009).
  5. [5]H. Pfeiffer, Phys. Stat. Sol. (a) 118, 295 (1990).
  6. [6]R. W. R. C. R. Williamsley, J. Gore, and M. Maylin, Phys. Rev. B 63, 024410 (2000).
  7. [7]S. Ruta, R. Chantrell, and O. Hovorka, Sci. Rep. 5, 9090 (2015).
  8. [8]J. García-Otero, M. Porto, J. Rivas, and A. Bunde, J. Magn. Magn. Mater. 203, 268 (1999).
  9. [9]J. García-Otero, M. Porto, J. Rivas, and A. Bunde, J. Appl. Phys. 85, 2287 (1999).
  10. [10]W. Wernsdorfer, C. Thrion, N. Demoncy, H. Pascard, and D. Mailly, J. Magn. Magn. Mater. 242, 132 (2002).
  11. [11]S. Peterson and Y. Idzerda, AIP Adv. 11, 085111 (2021).
  12. [12]D. Cimpoesu, L. Stŏleriu, and A. Stancu, J. Appl. Phys. 114, 10.1063/1.4839895 (2013).
  13. [13]A. Winter, H. Pascher, H. Krenn, X. Liu, and J. Furdyna, J. Appl. Phys. 108, 10.1063/1.3466771 (2010).
  14. [14]S. Yan, Z. Sun, and Y. B. Bazaliy, Phys. Rev. B 88, 054408 (2013).
  15. [15]G. Amanoloaiei, A. Stancu, and L. Stŏleriu, IEEE Magn. Lett. 12, 1 (2021).
  16. [16]D. Lee, J. Park, D. Han, S. Ippei, T. Yukiko, S. Noh, J. Lee, J. Kwon, H. Lee, and S. Kim, Phys. Rev. Applied 21, 024027 (2024).
  17. [17]W. T. Coffey, Y. P. Kalmykov, and S. V. Titov, Thermal Fluctuations and Relaxation Processes in Nanomagnets (World Scientific, 2020).DOI
  18. [18]D. Garanin and H. Kachkachi, Phys. Rev. Lett. 90, 065504 (2003).
  19. [19]H. Kachkachi, J. Magn. Magn. Mater. 316, 248 (2007).
  20. [20]D. Gandia, L. Gandarias, L. Marcano, I. Orue, D. Gil-Cartón, J. Alonso, A. García-Arribas, A. Muela, and M. L. Fdez-Gubieda, Nanoscale 12, 16081 (2020).
  21. [21]V. Russier, Journal of Magnetism and Magnetic Materials 409, 50 (2016).
  22. [22]I. Conde-Leboran, D. Baldomir, C. Martinez-Boubeta, O. Chubykalo-Fesenko, M. del Puerto Morales, G. Salas, D. Cabrera, J. Camarero, F. J. Teran, and D. Serantes, J. Phys. Chem. C 119, 15698 (2015).
  23. [23]D. Niculaes, A. Lak, G. C. Anyfantis, S. Marras, O. Laslett, S. K. Avugadda, M. Cassani, D. Serantes, O. Hovor­ka, R. Chantrell, et al., ACS nano 11, 12121 (2017).
  24. [24]N. Usov, J. Appl. Phys. 107, doi.org/10.1063/1.3445879 (2010).
  25. [25]G. Vallejo-Fernandez and K. O’grady, Appl. Phys. Lett. 103, https://doi.org/10.1063/1.4824649 (2013).
  26. [26]G. Salazar-Alvarez, J. Qin, V. Sepelak, I. Bergmann, M. Vasilakaki, K. Trohidou, J. Ardisson, W. Macedo, M. Mikhaylova, M. Muhammed, et al., JACS 130, 13234 (2008).
  27. [27]J. M. Paez-Muñoz, F. Gámez, Y. Fernández-Afonso, R. Gallardo, M. P. Leal, L. Gutiérrez, J. M. de la Fuente, C. Caro, and M. L. García-Martín, Journal of Materials Chemistry B 11, 11110 (2023).
  28. [28]A. G. Roca, L. Gutiérrez, B. Gayilán, M. E. F. Brollo, S. Veintemillas-Verdaguer, and M. del Puerto Morales, Advanced drug delivery reviews 138, 68 (2019).
  29. [29]E. Navarro, Y. Luengo, S. Veintemillas, P. Morales, F. J. Palomares, U. Urdizroz, F. Cebollada, and J. M. González, AIP Advances 9, 10.1063/1.5130164 (2019).
  30. [30]H. Gavilán, G. M. Rizzo, N. Silvestri, B. T. Mai, and T. Pellegrino, Nature Protocols 18, 783 (2023).
  31. [31]G. Mandriota, S. K. Avugadda, E. Sadeghi, N. Silvestri, R. Marotta, H. Gavilán, U. Olsson, C. Giannini, Y. H. Tsai, A. C. S. Samia, et al., Nanoscale Horizons https://doi.org/10.1039/D4NH00566J (2025).
  32. [32]J. Muro-Cruces, A. G. Roca, A. López-Ortega, E. Fantechi, D. del Pozo-Bueno, S. Estradé, F. Peiró, B. Sepúlveda, F. Pineider, C. Sangregorio, et al., ACS nano 13, 7716 (2019).
  33. [33]I. López-Vázquez, D. Serantes, and Óscar Iglesias, J. Magn. Magn. Mater. 638, 173751 (2026).
  34. [34]A. Jaklič, A. Leonardis, and F. Solina, Superquadrics and their geometric properties, in Segmentation and Recou- ery of Superquadrics (Springer Netherlands, Dordrecht, 2000) pp. 13–39.
  35. [35]J. G. Donaldson, E. S. Pyanzina, and S. S. Kantorovich, ACS Nano 11, 8153 (2017).
  36. [36]D. Dresen, A. Qdemat, S. Ulusoy, F. Mees, D. Zákrutná, E. Wetterskog, E. Kentzinger, G. Salazar-Alvarez, and S. Disch, J. Phys Chem. C 125, 23356 (2021).
  37. [37]Y. Jiao, F. H. Stillinger, and S. Torquato, Phys. Rev. E 79, 041309 (2009).
  38. [38]B. D. Cullity and C. D. Graham, Introduction to magnetic materials (John Wiley & Sons, 2011).
  39. [39]M. Donahue and D. Porter, Oommf user’s guide, version 1.0, interagency report nistir 6376, national institute of standards and technology, gaithersburg, md (sept 1999), http://math.nist.gov/oommf (2018).
  40. [40]S. Makridakis, S. C. Wheelwright, and R. J. Hyndman, Forecasting methods and applications (John wiley & sons, 2008).
  41. [41]E. C. Stoner and E. Wohlfarth, Philosophical Transactions of the Royal Society of London. Series A, Mathematical and Physical Sciences 240, 599 (1948).
  42. [42]D. Faílde, V. Ocampo-Zalvide, D. Serantes, and Ò. Iglesias, Nanoscale 16, 14319 (2024).
  43. [43]N. A. Usov, M. S. Nesov, E. M. Gubanova, and N. B. Epshtein, Beilstein J. Nanotechnol. 10, 305 (2019).
  44. [44]Y. P. Kalmykov, S. Titov, W. Coffey, M. Zarifakis, and W. Dowling, Phys. Rev. B 99, 184414 (2019).
  45. [45]M. Salvador, L. Nicolao, and W. Figueiredo, Phys. Rev. B 100, 104419 (2019).
  46. [46]H. Mamiya and B. Jeyadevan, Sci. Rep. 1, 1 (2011).
  47. [47]D. Mostarac, A. A. Kuznetsov, S. Helbig, C. Abert, P. A. Sánchez, D. Suess, and S. S. Kantorovich, Phys. Rev. B 111, 014438 (2025).
  48. [48]M. Wolfschenger, A. Jaufenthaler, F. Hanser, J. Gamper, T. S. Hofer, and D. Baumgarten, Appl. Math. Model. 136, 115624 (2024).
  49. [49]N. A. Usov and B. Y. Lubimov, J. Appl. Phys. 112, 023901 (2012).
  50. [50]F. L. Durhuus, M. Beleggia, and C. Frandsen, Physical Review B 109, 054421 (2024).
  51. [51]K. D. Usadel and C. Usadel, Journal of Applied Physics 118, 10.1063/1.4937919 (2015).
  52. [52]K. D. Usadel, Phys. Rev. B 95, 104430 (2017).
  53. [53]K. Okada and A. Satoh, J. Magn. Magn. Mater., 173259 (2025).
  54. [54]R. Moreno, S. Poyser, D. Meilak, A. Meo, S. Jenkins, V. K. Lazarov, S. Vallejo-Fernandez, S. Majetich, and R. F. Evans, Sci. Rep. 10, 2722 (2020).
  55. [55]H. Kachkachi, M. Nogues, E. Tronc, and D. Garanin, J. Magn. Magn. Mater. 221, 158 (2000).
  56. [56]S. A. Mathews, A. C. Ehrlich, and N. A. Charipar, Sci. Rep. 10, 15141 (2020).
  57. [57]L. Omari, A. Lekdadrari, H. Lassri, O. Mentre, C. Minaud, E. Dhahri, and C. Jama, Mater. Today Commun. 27, 102257 (2021).

Paper details

vlnc:W_qyygye89
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents