Magnetic susceptibility ($\chi$)

Magnetic susceptibility describes how readily a magnetic field changes a material's magnetization. The field variable and normalization matter: volume magnetic susceptibility in the International System of Units is dimensionless, whereas a derivative of total magnetic moment with respect to $B$ has units of $\mathrm{J}{\mathrm{T}}^{-2}$. A slope measured against the applied magnetic field intensity also depends on the sample's demagnetizing field.

For an isotropic, reversible linear response through the origin, writeHere $\mathbf{M}$ is magnetization, or magnetic moment per volume; ${\mathbf{H}}_{\mathrm{int}}$ is the internal magnetic field intensity; ${\mathbf{B}}_{\mathrm{int}}$ is the internal magnetic flux density; and ${\mu }_0$ is the vacuum permeability. Both $M$ and $H$ have units $\mathrm{A}{\mathrm{m}}^{-1}$. The magnetic permeability is ${\mu }_0(1+{\chi }_v)$. ^[ref]

The independent-moment magnetic response below shows why one number cannot describe every field: magnetization approaches a saturation magnetization, while its differential magnetic susceptibility approaches zero. Anisotropy requires a magnetic susceptibility tensor; hysteresis requires specifying the preparation and sweep branch as well as the bias magnetic field, temperature, and timescale. The Gouy balance turns a small response into a measurable force without presuming that its readout is already the intrinsic volume magnetic susceptibility.

The magnetization caused by an applied magnetic field strength. The displayed scalar linear law assumes isotropic susceptibility and neglects magnetic self-interaction; remanent magnetization is an additional component that need not align with the present field.

The product ${\chi }_{\mathrm{mol}}=V_m{\chi }_v$ of volume magnetic susceptibility and molar volume. In SI it has units ${\mathrm{m}}^3\mathrm{mo}{\mathrm{l}}^{-1}$. For one localized magnetic ion per formula unit, its Curie–Weiss law is ${\chi }_{\mathrm{mol}}=C_{\mathrm{mol}}/(T-\Theta )$, with $C_{\mathrm{mol}}={\mu }_0{N}_A{\mu }_{\mathrm{eff}}^2/(3k_B)$. Normalization per mole of formula units and per mole of magnetic ions differ when there is more than one such ion per formula unit.

For a scalar quartic Landau free energy with $a=a_{0}t$ and positive quartic coefficient, the zero-field susceptibility is $1/(a_{0}t)$ above the transition and $1/(2a_0|t|)$ within a selected ordered phase below it. Their singular amplitudes have ratio two. Below the transition this is a one-sided pure-phase response: the globally minimizing magnetization jumps between opposite phases when the field crosses zero. The mean-field ratio is not asserted to be exact in every fluctuation-dominated system.

The coefficient in the high-temperature Curie Magnetic susceptibility.

One noninteracting spatial orbital with two spin energies $ϵ\pm {\mu }_{B}B$, in the grand canonical ensemble with mean occupation one, has chemical potential $\mu =ϵ$ for every positive temperature. Its spin population difference is $-\tanh [{\mu }_{B}B/(2k_{B}T)]$, giving total-moment response $\chi ={\mu }_B^2/(2k_{B}T)$ at zero field. Empty and doubly occupied orbital states are permitted. Fixing exactly one electron instead leaves only the two singly occupied states and gives $\chi ={\mu }_B^2/(k_{B}T)$. The factor of two is an ensemble effect in this small system.

For $r,b>0$, minimization of the homogeneous Landau free energy gives $2rm+2b{m}^3=B^3$. The unique minimum near zero has $m=B^3/(2r)+O(B^9)$. Therefore its first two derivatives with respect to $B$ vanish at zero, whereas its third derivative is $3/r$. With $r=a(T-T_c)$ the first nonzero response diverges as $3/[a(T-T_c)]$. This is response to the variable $B$ through a cubic source, so it differs from the conventional third-order coefficient for a linearly coupled magnetic field. The missing factor $3!$ would give an incorrect susceptibility if one confused a series coefficient with a derivative.

For independent moments of angular momentum $J$ at small ${\mu }_B|B|/(k_{B}T)$, the zero-field variance is $⟨m^2⟩=J(J+1)/3$. Differentiating the canonical mean gives the magnetic susceptibilityThus susceptibility is inversely proportional to temperature. The small-field or high-temperature condition distinguishes this law from the exponentially vanishing fixed-field susceptibility of a saturated paramagnet at low temperature.

For independent localized angular-momentum-$J$ ions at fixed field, the weak-polarization mean moment is $⟨{\mu }_z⟩=C_{\mu }/T$. Unlike the volume-susceptibility Curie constant, this coefficient includes the magnetic field and no ion number density. The expansion requires $g{\mu }_B|B|J\ll k_{B}T$; the moment saturates toward $g{\mu }_{B}J$ beyond that regime.

Volume magnetic susceptibility is the dimensionless ratio of longitudinal magnetization to internal magnetic field intensity in a linear regime through the origin. Positive response gives paramagnetism. Always distinguish magnetization $M=\mathcal{M}/V$ from total magnetic moment $\mathcal{M}$: dividing by volume changes the normalization. Mass-normalized and mole-normalized values likewise require the mass density or the volume per amount of substance before comparison with ${\chi }_v$.

In the existing Pauli paramagnetism formula, $M$ denotes total magnetic moment, and $g(E_F)$ is the total spin-summed density of states. Relabel that response asHere ${\mu }_B$ is the Bohr magneton. If its $B$ is a prescribed vacuum/applied magnetic flux density $B_0={\mu }_0{H}_{\mathrm{app}}$, the dimensionless apparent slope is ${\mu }_0{\chi }_B^{\mathrm{vol}}$. It agrees approximately with intrinsic ${\chi }_v$ only when the demagnetizing field correction and magnetic feedback are negligible. The existing Curie's law formula also uses a total-magnetic moment, $B$-derivative convention; its $N$ counts moments rather than giving a number density.

Differential magnetic susceptibility is the local slope at a specified bias magnetic field intensity, holding temperature and other stated constraints fixed. It equals volume magnetic susceptibility only for a linear response through the origin. For example, $M=M_s\tanh (H/H_s)$ has $M/H=(M_s/H_s)\tanh (x)/x$, but $dM/dH=(M_s/H_s)/{\cosh }^2(x)$, where $x=H/H_s$. At $x=1$ these are about $0.762$ and $0.420$ times their common zero-field slope. A finite sweep chord is another quantity, not automatically this derivative.

If the differentiated field is instead the internal magnetic flux density, the longitudinal chain rule applied to $B_{\mathrm{int}}={\mu }_0(H_{\mathrm{int}}+M)$ givesThis differs from the vacuum/applied-$B_0$ conversion in volume magnetic susceptibility. It assumes a differentiable single branch with locally invertible $B_{\mathrm{int}}(H_{\mathrm{int}})$.

A magnetic susceptibility tensor maps a small change of internal magnetic field intensity to a change of magnetization:This tensor reduces to $\chi {\delta }_{ij}$ for isotropy. With anisotropy, even a linear response need not point along the applied magnetic field intensity. As a stipulated example, $\mathbf{\chi }=\mathrm{diag}(1,1,3)\times 10^{-4}$ sends $(H,0,H)$ to $(H,0,3H)\times 10^{-4}$. State the axes and bias field when reporting its components. A constant tensor describes linear response; in a nonlinear material it is the Jacobian matrix of the response at the chosen operating point.

In vacuum, magnetization is zero and magnetic flux density obeys $\mathbf{B}={\mu }_0\mathbf{H}$. The applied quantities ${\mathbf{H}}_{\mathrm{app}}$ and ${\mathbf{B}}_0={\mu }_0{\mathbf{H}}_{\mathrm{app}}$ mean the field of the external sources at the sample position with the sample absent. The internal magnetic field intensity includes the sample's demagnetizing field; its internal magnetic flux density additionally includes ${\mu }_0\mathbf{M}$. A field quoted in the tesla is a $B$ quantity, even when an instrument informally calls it “$H$”.

For a uniformly magnetized ellipsoid, along a principal axis with demagnetizing factor $N_d$ in the International System of Units,For a sphere, $N_d=1/3$. The axial value approaches zero in the long slender ellipsoid limit and the normal value approaches one in the thin flat ellipsoid limit. Thus a stipulated intrinsic volume magnetic susceptibility ${\chi }_v=3$ would yield apparent values $1.5$ in a sphere and approximately $3$ in a long axial sample. These are calculated geometry effects, not measured material values. For fixed $N_d$, implicit differentiation gives the same conversion for local differential magnetic susceptibility, $dM/d{H}_{\mathrm{app}}={\chi }_{\mathrm{diff}}/(1+N_d{\chi }_{\mathrm{diff}})$.

The demagnetizing field is the contribution to magnetic field intensity produced by the sample's own magnetization. Inside a uniformly magnetized ellipsoid, it is uniform and has the form ${\mathbf{H}}_d=-\mathsf{N}\mathbf{M}$; the demagnetizing factor along a principal axis gives its longitudinal component. These conditions make the scalar correction in field conventions for magnetic susceptibility useful.

A cylinder generally has spatially varying magnetization and demagnetizing field. A volume-sensitive instrument and a short central pickup coil consequently need different effective corrections: one averages over the sample, while the other samples central flux. A shape factor copied from an ellipsoid is not an exact correction for every cylinder, especially with nonlinear response or hysteresis. ^[ref]

For a homogeneous linear isotropic sphere in a uniform applied field, the internal field is ${\mathbf{H}}_i={\mathbf{H}}_0-\mathbf{M}/3$. Combining this geometric demagnetizing field with $\mathbf{M}=\chi {\mathbf{H}}_i$ gives the title expression. Matching the scalar magnetic potentials $-H_{i}r\cos \theta$ inside and $-H_{0}r\cos \theta +C\cos \theta /r^2$ outside provides an independent boundary-value derivation. Tangential $H$ is continuous without free surface current and normal $B$ is continuous. A negative susceptibility makes magnetization antiparallel, not parallel in sense, to the applied field.

For an ellipsoidal sample in a uniform applied magnetic field intensity $H_0$ along a principal axis, its uniform demagnetizing field is $-N_{d}M$. The SI demagnetizing factor obeys $0\le N_d\le 1$. Linear magnetic susceptibility $\chi ={\mu }_r-1$ gives $M=\chi (H_0-N_{d}M)$, hence $M=\chi H_0/(1+N_d\chi )$. An axially slender prolate limit has $N_d\to 0$, a sphere has $N_d=1/3$, and a thin plate with field normal to it has $N_d\to 1$. Thus $M=({\mu }_r-1)H_0/{\mu }_r$ belongs to the plate limit, not the slender axial-rod limit. A finite cylinder generally has nonuniform magnetization; an effective ellipsoidal factor is an approximation rather than an exact universal correction.

A demagnetizing factor relates a component of demagnetizing field to magnetization. The three principal factors of a uniformly magnetized ellipsoid sum to one in the International System of Units. Factors quoted in a different unit convention need conversion; state the convention and measurement geometry before applying them.

Consider a stipulated dilute collection of independent two-level systems, each with longitudinal magnetic moment $s\mu$, $s=\pm 1$, and energy $E_s=-s\mu B_0$. The prescribed vacuum/applied magnetic flux density is $B_0={\mu }_0{H}_{\mathrm{app}}$. Ignore interactions, orbital magnetic response, and the small sample-field correction. This is the magnetic version of the existing independent spin-one-half two-level system, with alignment probabilities rather than an interacting Curie–Weiss model.

For temperature $T>0$, the canonical ensemble gives the one-moment partition function $Z_1=2\cosh x$, with $x=\mu B_0/(k_{B}T)$. Taking the expected value over its two Boltzmann weights givesHere $n$ is the number density, $k_B$ the Boltzmann constant, and $\tanh$ the hyperbolic tangent. For $|x|\ll 1$, its small-argument expansion of the hyperbolic tangent gives $M\simeq {\mu }_{0}n{\mu }^2{H}_{\mathrm{app}}/(k_{B}T)$, a Curie's law response in the applied-$H$ convention. For large $|x|$, $M$ approaches the saturation magnetization $n\mu$ with sign set by the field, and its differential magnetic susceptibility falls towards zero.

Both plots use stipulated $n=10^{26}{\mathrm{m}}^{-3}$, $\mu =9.2740101\times 10^{-24}\mathrm{J}{\mathrm{T}}^{-1}$, $k_B=1.380649\times 10^{-23}\mathrm{J}{\mathrm{K}}^{-1}$, and ${\mu }_0=1.25663706\times 10^{-6}\mathrm{T}\mathrm{m}{\mathrm{A}}^{-1}$. The moment is approximately one Bohr magneton; the chosen vacuum permeability is a rounded model input. At $T=2\mathrm{K}$, $M_s=927.401\mathrm{A}{\mathrm{m}}^{-1}$ and ${\chi }_{\mathrm{model}}(0)=3.91410\times 10^{-4}$. The linear approximation overestimates $M$ by less than $3\%$ for $|x|\le 0.3$, corresponding to $|{\mu }_0{H}_{\mathrm{app}}|\le 0.89324\mathrm{T}$ or $|H_{\mathrm{app}}|\le 0.71082\mathrm{MA}{\mathrm{m}}^{-1}$. These parameters keep the modeled response weak; neglected magnetic feedback is of relative order ${\chi }_{\mathrm{model}}(0)$, rather than an explanation of a material's ordered phase.

At $B_0=0$, cooling increases the zero-field slope as $1/T$. At fixed $B_0\ne 0$, cooling instead gives ${\chi }_{\mathrm{model}}\sim 4{\mu }_{0}n{\mu }^2{e}^{-2\mu |B_0|/(k_{B}T)}/(k_{B}T)$: a saturated moment can hardly change further. The order of the zero-field and zero-temperature limits matters. This independent model has no finite-temperature collective phase transition; the diverging zero-field slope alone does not establish one.

A freely orientable classical magnetic dipole moment of fixed magnitude $m$ in a uniform magnetic field $B$ has angular canonical partition function $Z=4\pi \sinh x/x$, where $x=mB/(k_{B}T)$. Its internal energy is $U=-mB\coth x+k_{B}T$, and its fixed-field heat capacity is $C_B=k_B[1-x^2/{\sinh }^{2}x]$. At low temperature, two locally Cartesian tangent coordinates on the orientation sphere each contribute $k_{B}T/2$ by the equipartition theorem, giving $C_B\to k_B$. A discrete quantum spin in a nonzero magnetic field instead has a spectral gap and its heat capacity vanishes exponentially at sufficiently low temperature.

Saturation magnetization is the limiting magnetization when the responding magnetic moments are fully aligned within the stated model or experimental regime. In independent-moment magnetic response, $M_s=n\mu$. Reaching a large magnetization does not imply a large differential magnetic susceptibility: the remaining slope can be almost zero.

A Gouy balance detects the force exerted by a nonuniform magnetic field on a sample of uniform cross-section extending between high- and low-field regions. TeachSpin's implementation reverses the usual arrangement: a rare-earth permanent-magnet assembly sits on a one-pan electronic balance, while a separate holder brings the sample into the field. The magnet's apparent mass change measures the opposite of the sample's vertical force, by Newton's third law. The described pole faces have diameter $25\mathrm{mm}$ and separation $15\mathrm{mm}$; samples occupy cuvettes of cross-section $(10\mathrm{mm}{)}^2$ and height about $32\mathrm{mm}$. ^[ref]

The same balance calibrates magnetic flux density by measuring the reaction force on the magnet when a wire of known horizontal length $L$ carries known electric current $i$ perpendicular to the field: $|F_z|=iL{B}_0$. The manufacturer describes water being repelled by the high-field region, identifying negative magnetic susceptibility. This observed sign supplies a qualitative test, not a measured Curie's law curve. ^[ref]

For weak reversible linear response, a uniform sample displacing a weakly magnetic surrounding medium has, to leading order, downward forceThis follows by integrating the induced-dipole force density $({\chi }_v-{\chi }_{\mathrm{medium}}){\partial }_z(B_0^2)/(2{\mu }_0)$ along a downward coordinate. $A$ is the cross-sectional area, and the two $B_0$ values are applied magnetic flux densities. The force depends on the endpoint fields; the entire field-gradient profile need not be measured under these assumptions. In the reversed apparatus, $F_{\downarrow }=-g\Delta m_{\mathrm{magnet}}$ after subtracting the holder background, where $g$ is gravitational acceleration. This sign must be reversed before inferring volume magnetic susceptibility.

Useful controls are an empty-holder reading at the same position, a distant-sample baseline, repeated wire-current reversals during field calibration, and stable temperature. They separate holder response, balance drift, force offsets, and thermal changes; these are proposed checks for the stated inference. Finite endpoint fields, nonuniform cross-section, uneven powder packing, stray forces and a demagnetizing field limit the simple formula. If magnetization is nonlinear, the weak-response force involves $A{\int }_{B_{0,\mathrm{top}}}^{B_{0,\mathrm{bottom}}}M(B_0)d{B}_0$ instead of a constant-${\chi }_v$ endpoint square. A balance reading then measures an integrated response, not directly the local differential magnetic susceptibility.