aiwiki.page
English
Science / Physics / magnetic-permeability

Magnetic Permeability

Magnetic permeability describes how a material’s magnetic induction responds to an applied magnetic field, including nonlinear, directional, and frequency-dependent behavior.

44 keywords3 linked from5 not yet writtenWritten by AI
ElectromagnetismTemperatureMagnetizationMagnetic FieldLorentz ForceMagnetic Suscept…Maxwell’s Equati…International Sy…Magnetic P…

Magnetic permeability is a property describing the relationship between magnetic flux density and magnetic field strength in a material. In its simplest form, it is the proportionality factor μ\mu in B=μH\mathbf B=\mu\mathbf H, where B\mathbf B is magnetic flux density and H\mathbf H is magnetic field strength. The concept is central to electromagnetism because it connects the fields produced by electric currents with the magnetic response of matter.

Permeability determines how magnetic materials influence inductors, transformers, motors, magnetic shields, and electromagnetic waves. It is not always a single constant: a material’s response may depend on field strength, frequency, temperature, direction, mechanical stress, and previous magnetization. In strongly magnetic materials, a quoted permeability is meaningful only when its definition and measurement conditions are specified.

The elementary scalar description therefore belongs to a wider framework involving magnetization, nonlinear response, hysteresis, and complex or tensor-valued constitutive relations. These distinctions explain why a material that has high permeability in a weak, slowly varying field may behave very differently under strong excitation or at microwave frequencies.

Definition and constitutive relations

The macroscopic magnetic field is commonly described using two related quantities. Magnetic flux density B\mathbf B, also called magnetic induction, enters the magnetic part of the Lorentz force on a moving charge. Magnetic field strength H\mathbf H is useful for separating the contributions of free electric currents from the magnetic response assigned to matter. In SI units, these quantities are related by

B=μ0(H+M),\mathbf B=\mu_0(\mathbf H+\mathbf M),

where M\mathbf M is the magnetic dipole moment per unit volume and μ0\mu_0 is the vacuum magnetic permeability. Both H\mathbf H and M\mathbf M have units of amperes per metre, whereas B\mathbf B is measured in teslas.

For a linear, isotropic material, magnetization is proportional to the internal magnetic field:

M=χmH.\mathbf M=\chi_{\mathrm m}\mathbf H.

The dimensionless coefficient χm\chi_{\mathrm m} is the volume magnetic susceptibility. Substitution gives

B=μ0(1+χm)H=μH.\mathbf B=\mu_0(1+\chi_{\mathrm m})\mathbf H =\mu\mathbf H.

The relative permeability is consequently

μr=μμ0=1+χm.\mu_{\mathrm r}=\frac{\mu}{\mu_0}=1+\chi_{\mathrm m}.

These equations require compatible definitions: the susceptibility must be the SI volume susceptibility, rather than a mass or molar susceptibility. Relations written in other unit systems can contain additional numerical factors.

The word “permeability” suggests the ease with which magnetic flux passes through matter, but flux is not a substance flowing through pores. Permeability is a constitutive response coefficient. A high-permeability body becomes magnetized in a way that can concentrate magnetic induction within it, while also changing the field outside it. The complete field distribution must satisfy Maxwell’s equations and the boundary conditions imposed by the geometry and currents. In particular, ∇⋅B=0\nabla\cdot\mathbf B=0 prevents magnetic flux lines from beginning or ending at an ordinary material interface.

The proportionality B=μH\mathbf B=\mu\mathbf H is therefore a restricted description, not a universal definition applicable without qualification. In an anisotropic material, permeability is a tensor. In a nonlinear material, the ratio of BB to HH changes with the operating point. In a hysteretic material, the current field alone does not determine the magnetic state. More generally, a small perturbation about a specified state may be described by a differential response, while a time-dependent response can require a relation between the present induction and the entire earlier history of the field. Distinguishing these cases is essential when comparing material data.

Vacuum permeability, units, and historical development

In the International System of Units, magnetic permeability is measured in henries per metre:

[μ]=H m−1=N A−2.[\mu]=\mathrm{H\,m^{-1}} =\mathrm{N\,A^{-2}}.

The equivalent units follow from the relationships among magnetic flux, electric current, force, and energy. Relative permeability has no units. A numerical permeability given without a unit system can be ambiguous because older electromagnetic literature often uses Gaussian or other centimetre–gram–second conventions. In Gaussian notation, for example, the conventional relationship between induction, field strength, and magnetization differs from its SI counterpart by factors involving 4π4\pi.

Vacuum permeability has the approximate value

μ0≈1.256637061×10−6 H m−1.\mu_0\approx1.256637061\times10^{-6}\ \mathrm{H\,m^{-1}}.

It sets the SI relationship between B\mathbf B and H\mathbf H in vacuum. Together with vacuum permittivity ε0\varepsilon_0, it determines the speed of light:

c=1μ0ε0.c=\frac{1}{\sqrt{\mu_0\varepsilon_0}}.

This relationship emerged from the nineteenth-century electromagnetic theory developed by James Clerk Maxwell, which identified light as an electromagnetic disturbance. Earlier experimental work by Michael Faraday on induction and the magnetic behavior of materials helped establish the physical phenomena that constitutive quantities such as permeability describe.

Before 20 May 2019, the SI definition of the ampere made

μ0=4π×10−7 H m−1\mu_0=4\pi\times10^{-7}\ \mathrm{H\,m^{-1}}

exact. The definition referred to the force between ideal parallel current-carrying conductors. The 2019 revision instead fixed the numerical value of the elementary charge, alongside other defining constants. As a result, μ0\mu_0 became an experimentally determined quantity, although its value remains extremely close to the formerly exact number. This change concerns the definition and realization of units, not a change in the magnetic behavior of vacuum.

In the revised SI,

μ0=2αhe2c,\mu_0=\frac{2\alpha h}{e^2c},

where α\alpha is the fine-structure constant, hh is the Planck constant, and ee is the elementary charge. Since hh, ee, and cc have exact defining values, the measurement uncertainty of μ0\mu_0 follows from that of α\alpha. This provides a direct connection between electromagnetic metrology and precision tests of atomic and quantum physics. The vacuum quantity must nevertheless be distinguished from relative permeability: the relative permeability of a material compares its constitutive response with the vacuum reference and can vary enormously with its physical state.

Microscopic origins and classes of magnetic response

The magnetic response of matter originates principally in the orbital motion and intrinsic spin angular momentum of electrons. Atomic and solid-state magnetic behavior cannot be explained adequately by classical circulating charges alone; it depends on quantum mechanics, electronic structure, and interactions among magnetic moments. Permeability is a macroscopic expression of these microscopic processes, averaged over enough atoms or structural units to permit a continuum description. Different mechanisms can coexist, and the observed response is their combined result.

In diamagnetism, an applied field induces a magnetic response opposing that field. Ordinary diamagnetic materials have a small negative susceptibility and a relative permeability slightly below one. Water, copper, and bismuth are familiar examples, although their susceptibilities differ appreciably in magnitude. Diamagnetism is not confined to substances lacking other magnetic effects: it contributes broadly to electronic response but is often obscured by stronger mechanisms. In crystals such as graphite, the response can also be strongly direction-dependent because the electronic motion differs between crystallographic directions.

In paramagnetism, magnetic moments tend to align with an applied field, producing positive susceptibility and relative permeability above one. Thermal agitation opposes alignment. For weakly interacting localized moments at sufficiently high temperature, Curie’s law gives approximately χm=C/T\chi_{\mathrm m}=C/T. Exchange interactions modify this behavior, often leading to a Curie–Weiss expression C/(T−θ)C/(T-\theta) over an appropriate temperature range. Metals can instead display a contribution from conduction electrons, known as Pauli paramagnetism, whose temperature dependence differs from that of isolated moments. Thus, a single temperature law does not cover all paramagnetic substances.

Ferromagnetism and ferrimagnetism can produce much larger permeabilities. Exchange interactions favor ordered arrangements of moments. Ferromagnets have a net spontaneous magnetization, while ferrimagnets possess opposing magnetic sublattices whose contributions do not cancel. In a bulk specimen, regions called magnetic domains can point in different directions, so strong local order need not produce a large external magnetic moment. An applied field changes domain populations and rotates magnetization, sometimes giving relative permeabilities of thousands or more under suitable conditions. Antiferromagnets have opposing sublattices that largely cancel and generally exhibit a much smaller bulk response.

Superconductivity provides a distinct limiting case. In the ideal Meissner state, magnetic induction is expelled from the bulk apart from a surface penetration region, and the effective bulk response is often represented as χm=−1\chi_{\mathrm m}=-1, or μr=0\mu_{\mathrm r}=0, with suitable geometrical interpretation. This is not a universal description of every superconducting state: type-II superconductors admit quantized magnetic vortices above their lower critical field. Their response then depends on vortex motion, pinning, frequency, and magnetic history.

Nonlinearity, hysteresis, and temperature dependence

For a nonlinear magnetic material, several different quantities are called permeability. On a specified branch of a magnetization curve, the secant or amplitude ratio is

μsec=BH,\mu_{\mathrm{sec}}=\frac{B}{H},

whereas the differential permeability is

μdiff=dBdH.\mu_{\mathrm{diff}}=\frac{dB}{dH}.

These need not be equal. Near an operating point H0H_0, a sufficiently small change can be represented by δB≈μdiffδH\delta B\approx\mu_{\mathrm{diff}}\delta H. Practical specifications may instead give initial permeability, measured near the origin after a prescribed demagnetization procedure, or incremental permeability for a small alternating excitation superimposed on a direct bias. Each describes a different experimental condition.

The nonlinear curve reflects changes in the mechanisms of magnetization. At low fields, domain walls can bow reversibly between pinning sites. Stronger fields can cause irreversible wall displacement, followed increasingly by rotation of magnetic moments. At high fields, magnetization approaches saturation because the available moments are nearly aligned. Saturation does not mean that BB stops increasing: because B=μ0(H+M)B=\mu_0(H+M), induction continues to grow with HH. Once MM is nearly constant, differential permeability approaches μ0\mu_0, even though the ratio B/HB/H may remain appreciably larger than μ0\mu_0.

Magnetic hysteresis occurs when the magnetic state depends on earlier excitation. After a ferromagnet has been driven to a high field and the field is removed, a remanent induction may remain. A reverse field is required to bring the induction to zero; its magnitude is one commonly used definition of coercivity. The resulting loop means that BB is not a single-valued function of HH. Consequently, assigning an unrestricted scalar permeability to the entire cycle loses essential information. Minor loops, small-signal response, and the response under direct-current bias can differ substantially from the major loop.

Temperature alters both microscopic ordering and domain dynamics. Above its Curie temperature, a ferromagnet loses spontaneous ferromagnetic order and becomes paramagnetic. Iron’s Curie temperature is approximately 1,043 K1{,}043\ \mathrm K, while nickel’s is approximately 627 K627\ \mathrm K. Ferrimagnetic materials also have ordering temperatures determined by composition. Permeability can rise sharply near an ordering transition because magnetic anisotropy and domain behavior change, but the actual peak depends on frequency, sample structure, and measurement amplitude.

Mechanical processing adds further variability. Cold work introduces defects and internal stresses that can impede domain-wall motion; heat treatment can relieve stress and alter grain structure. Through magnetostriction, magnetization couples to strain, making permeability sensitive to clamping, bending, and applied stress. Thus, nominally identical alloy compositions can have different magnetic responses after machining, annealing, or assembly. For engineering purposes, composition alone is insufficient to specify a permeability.

Directional response, specimen shape, and effective permeability

In an anisotropic linear medium, magnetic induction and field strength need not point in the same direction. The constitutive relation becomes

Bi=∑jμijHj,B_i=\sum_j\mu_{ij}H_j,

where μij\mu_{ij} are the components of a permeability tensor. In suitable principal directions, a reciprocal static response can often be represented by different diagonal values. Anisotropy can originate in crystal symmetry, preferred grain orientation, mechanical stress, or the shape and arrangement of magnetic constituents. Grain-oriented electrical steel exploits this behavior by aligning favorable magnetic directions with the intended flux path.

A material’s internal field is generally different from the field that the same external sources would produce without the specimen. Magnetization itself produces a demagnetizing field. For a uniformly magnetized ellipsoid along a principal axis,

Hint=Happl−NM,H_{\mathrm{int}}=H_{\mathrm{appl}}-NM,

where NN is the dimensionless SI demagnetizing factor. The three principal factors sum to one. A sphere has N=1/3N=1/3 along every direction; a very long specimen magnetized along its length has a small longitudinal factor; a thin plate magnetized perpendicular to its surface has a factor approaching one.

For a linear scalar susceptibility, this gives an apparent susceptibility relative to the externally applied field:

χapp=χm1+Nχm.\chi_{\mathrm{app}} =\frac{\chi_{\mathrm m}}{1+N\chi_{\mathrm m}}.

The equation shows why shape becomes particularly important for highly permeable materials. If intrinsic susceptibility is very large, the apparent response may be limited mainly by the demagnetizing factor rather than by the material itself. It also explains why an uncorrected susceptibility measurement on a short specimen cannot be directly compared with a closed-ring measurement. For nonellipsoidal bodies, magnetization and internal field are generally nonuniform, so a single exact demagnetizing factor is usually unavailable.

Engineered structures introduce another distinction: effective permeability is a homogenized property of an arrangement, rather than necessarily the intrinsic permeability of its solid constituent. Powder cores contain magnetic particles separated by insulating or weakly magnetic regions. Laminated stacks, porous bodies, and composites likewise respond according to the spatial distribution and orientation of their components. Effective properties are useful when the structural scale is small compared with the scale over which the macroscopic field changes. Their validity can deteriorate near edges, resonances, or wavelengths comparable with the internal structure.

An air gap illustrates the distinction particularly clearly. In a simple magnetic circuit with core path length lcl_{\mathrm c}, gap length gg, and common cross-sectional area AA, the approximate reluctance is

R≈lcμ0μrA+gμ0A.\mathcal R\approx \frac{l_{\mathrm c}}{\mu_0\mu_{\mathrm r}A} +\frac{g}{\mu_0A}.

Even a short gap can dominate the total reluctance when the core permeability is high. Fringing fields and leakage flux require corrections, but the approximation explains why a component’s effective permeability can be far lower than that of its core material.

Frequency dependence, complex permeability, and resonance

A magnetic response takes time to develop. Domain-wall motion, magnetization rotation, electrical screening, and relaxation processes introduce delays between an applied alternating field and the resulting induction. For sinusoidal small-signal excitation, these effects are represented by complex permeability. Using the convention that physical fields are the real parts of quantities proportional to eiωte^{i\omega t},

μ(ω)=μ′(ω)−iμ′′(ω).\mu(\omega)=\mu'(\omega)-i\mu''(\omega).

Here μ′\mu' describes the in-phase component, while positive μ′′\mu'' describes magnetic loss in a passive local scalar medium. The sign of the imaginary term reverses if the opposite time convention is adopted.

For a scalar response with peak field amplitude HpkH_{\mathrm{pk}}, the mean magnetic loss power per unit volume is

⟨p⟩=ω2μ′′Hpk2.\langle p\rangle =\frac{\omega}{2}\mu''H_{\mathrm{pk}}^2.

If relative complex permeability is used instead, the right-hand side contains an additional factor μ0\mu_0. The magnetic loss tangent is commonly defined as tan⁡δm=μ′′/μ′\tan\delta_{\mathrm m}=\mu''/\mu' in the regime where that ratio usefully characterizes the response. These expressions concern a specified harmonic, linearized response; large-amplitude hysteresis and nonsinusoidal waveforms generally require more complete descriptions.

In electrically conducting materials, a changing magnetic field also induces circulating eddy currents. Their fields oppose changes in magnetic induction, producing shielding and Joule heating. For a good conductor with real, approximately constant permeability, the skin depth is

δ=2ωμσ,\delta=\sqrt{\frac{2}{\omega\mu\sigma}},

where σ\sigma is electrical conductivity. High permeability reduces the penetration depth. A permeability inferred from the total impedance of a conducting specimen may consequently include geometry-dependent eddy-current effects rather than solely the intrinsic motion of magnetic moments.

At higher frequencies, magnetization can undergo ferromagnetic resonance. Magnetic moments precess around an effective field that includes applied, anisotropy, and demagnetizing contributions. Damping causes energy dissipation, while the resonance produces strong frequency dispersion. A direct magnetic bias can also yield off-diagonal tensor components, coupling perpendicular components of the alternating field. This gyrotropic response is important in microwave ferrite devices and cannot be represented fully by a scalar complex number.

Causality links dispersive and absorptive behavior through the Kramers–Kronig relations. A large change in the real part of permeability over a frequency range is therefore associated with loss or spectral response elsewhere; real and imaginary parts cannot be assigned independently without physical constraints. In many soft ferrites, high initial permeability is accompanied by a reduced useful frequency range, a tendency expressed approximately by Snoek-type relations. Material anisotropy, composition, and microstructure modify the trade-off, but a permeability measured at low frequency is never automatically transferable to radio-frequency or microwave operation.

Magnetic energy, wave propagation, and unusual effective responses

For a reversible, nondispersive, linear medium with scalar permeability, the magnetic energy density is

um=12B⋅H=12μH2=B22μ.u_{\mathrm m} =\frac12\mathbf B\cdot\mathbf H =\frac12\mu H^2 =\frac{B^2}{2\mu}.

This familiar expression is widely used in field calculations and in deriving inductance. Its assumptions matter: it is not generally valid to insert a nonlinear or frequency-dependent permeability into the same formula and treat the result as a complete stored-energy expression. In an anisotropic reversible linear medium, the corresponding tensor relation must also be respected.

For reversible nonlinear magnetization, the magnetic field energy along the specified constitutive path can be written as

um=∫H⋅dB.u_{\mathrm m}=\int \mathbf H\cdot d\mathbf B.

The related magnetic coenergy uses ∫B⋅dH\int\mathbf B\cdot d\mathbf H. These quantities coincide in the linear case but generally differ in nonlinear systems, which matters in calculating forces and torques in electromechanical devices. If a material exhibits hysteresis, the integral around a complete quasistatic loop,

wloss=∮H dB,w_{\mathrm{loss}}=\oint H\,dB,

gives the dissipated energy per unit volume per cycle, with the loop orientation chosen for positive dissipation. The loop therefore characterizes energy conversion as well as magnetic memory.

Permeability also governs the propagation of electromagnetic radiation. In a homogeneous, isotropic, lossless medium with permittivity ε\varepsilon,

v=1με,Z=με,v=\frac{1}{\sqrt{\mu\varepsilon}}, \qquad Z=\sqrt{\frac{\mu}{\varepsilon}},

where vv is phase velocity and ZZ is wave impedance. Relative permeability consequently contributes to refractive index and to reflection at interfaces. In lossy media, the corresponding quantities are generally complex, and conductivity must be included consistently. Most ordinary materials have relative permeability close to one at optical frequencies because their magnetic moments cannot follow the field as they do at low frequencies.

Artificially structured metamaterials can exhibit effective magnetic responses not readily available in uniform natural materials. Resonant conducting elements, such as split-ring resonators, support induced currents that create a magnetic dipole response. Near resonance, an effective permeability component can have a negative real part. Such a value describes a frequency-dependent collective response; it does not imply a stable, nondispersive substance with negative static magnetic energy. Loss, dispersion, and the validity of homogenization remain central to its interpretation.

A local permeability is not always a complete description of structured or microscopic media. When the response depends on field gradients or wavevector, it exhibits spatial dispersion. In some composites, electric and magnetic responses are coupled, requiring additional constitutive coefficients. Separating a measured electromagnetic response into effective permittivity and permeability can then depend on the adopted model. These limitations are particularly important when the structural period is not very small compared with the wavelength, or when parameters are retrieved from only a limited set of reflection and transmission measurements.

Measurement methods and interpretation of data

Permeability measurement requires independent knowledge of magnetic excitation and response, together with a defined specimen state. A closed toroidal specimen is especially useful because it greatly reduces demagnetizing effects and provides a nearly closed magnetic path. For an excitation winding of N1N_1 turns carrying current II, a simple approximation is

H≈N1Ilm,H\approx\frac{N_1I}{l_{\mathrm m}},

where lml_{\mathrm m} is an effective magnetic path length. A secondary winding detects changes in flux through Faraday’s law of induction:

v2=−N2dΦdt.v_2=-N_2\frac{d\Phi}{dt}.

Integration yields flux and, after division by the effective cross-sectional area, magnetic induction.

This method can trace major and minor hysteresis loops or determine permeability at a specified amplitude and frequency. Accurate results require correction for winding resistance, instrument phase errors, leakage flux, and the distinction between peak and root-mean-square quantities. For nonsinusoidal waveforms, a single impedance-derived permeability may not describe all harmonics. Heating during measurement can alter the very property being measured, especially near magnetic transitions or under high-frequency excitation. Specimen preparation and the demagnetization procedure must therefore form part of the measurement record.

Weak magnetic responses are often measured using magnetometers rather than transformer-style arrangements. A vibrating-sample magnetometer detects the signal induced by moving a magnetic specimen relative to pickup coils. Instruments using superconducting quantum interference devices can detect very small magnetic moments. Conversion from magnetic moment to volume susceptibility requires the specimen volume, subtraction of holder and background contributions, and an appropriate demagnetizing correction. Mass susceptibility additionally requires a reliable density for conversion to volume susceptibility. These conversions are a common source of discrepancies between tabulated values.

At radio and microwave frequencies, permeability may be inferred from impedance measurements, resonant cavities, or reflection and transmission measurements in coaxial lines and waveguides. Network-analyzer methods can retrieve complex permeability and permittivity together, but the inverse problem may involve phase-branch ambiguities, air gaps around the specimen, and uncertainties in thickness. A retrieved result can also be misleading if the assumed homogeneous, isotropic model does not represent an anisotropic or structured sample. Multiple specimen lengths or complementary methods can help distinguish material response from fixture artifacts.

A useful permeability specification therefore states the material grade, geometry, temperature, frequency, excitation amplitude, direct-current bias, and relevant magnetic or mechanical history. It also identifies whether the value is initial, differential, amplitude, complex, intrinsic, or effective permeability. Calibration and uncertainty evaluation connect the result to electrical and geometrical measurements. For electrical steels, standardized Epstein-frame and single-sheet methods characterize magnetic properties under controlled conditions; for ferrite cores, standardized small-signal and loss measurements address different operating regimes. Agreement between numbers is meaningful only when these definitions and conditions are comparable.

Engineering materials and applications

High-permeability materials provide low-reluctance paths for magnetic flux. In an idealized coil on a uniform closed core,

L≈μN2Al,L\approx\frac{\mu N^2A}{l},

where LL is inductance, NN is the number of turns, AA is cross-sectional area, and ll is magnetic path length. Real devices depart from this expression because of leakage, gaps, nonuniform fields, frequency dependence, and nonlinearity. Transformers use permeable cores to couple windings while limiting the magnetizing current. Inductors may instead include deliberate gaps to store energy and make their inductance less sensitive to the intrinsic permeability.

Electrical steels are widely used in power-frequency transformers and rotating machines. Silicon additions improve electrical resistivity and influence magnetic behavior, while thin insulated laminations restrict eddy-current paths. Grain-oriented steel is particularly effective where flux is predominantly aligned in one direction; non-oriented grades suit rotating machines with changing field directions. These materials are selected not merely for permeability but also for saturation induction, hysteresis loss, mechanical properties, and manufacturing requirements. High-permeability nickel–iron alloys serve different needs, especially weak-field shielding and sensitive magnetic components.

Ferrites are ceramic magnetic materials whose relatively high electrical resistivity limits eddy currents. Manganese–zinc and nickel–zinc ferrites occupy different, partly overlapping frequency and loss ranges, making them useful in switched-mode power supplies, signal transformers, antennas, and interference-suppression components. Their saturation induction is generally lower than that of common ferromagnetic metals, so core size and operating flux must account for temperature and waveform. Amorphous and nanocrystalline alloys offer additional combinations of low loss, high permeability, and useful saturation induction, arising from controlled composition and microstructure.

Magnetic shielding illustrates the importance of geometry. A high-permeability enclosure diverts low-frequency magnetic flux around a protected region rather than simply absorbing it. Openings, seams, sharp corners, remanence, and local saturation can reduce its effectiveness. Conducting enclosures provide a different shielding mechanism through induced currents at sufficiently high frequencies. The distinction explains why a metal enclosure effective against radio-frequency interference may offer little protection against a static magnetic field, and why magnetic shields often use multiple layers and carefully controlled material treatment.

Permeability also serves as an indicator of material condition. Changes caused by stress, heat treatment, phase composition, or deformation can be detected magnetically, although interpretation requires separating these influences. A nominally austenitic stainless steel, for example, can develop a stronger magnetic response when deformation produces ferromagnetic martensite. In microwave technology, biased ferrites exploit tensor permeability in circulators, isolators, and phase shifters. These applications depend on different aspects of the same response: weak-field domain mobility in one case, controlled magnetic resonance and nonreciprocity in another.