Curie Temperature Tensor in Anisotropic Magnets

• Curie temperature tensor is defined as T_C(n)= n^T T n, representing direction-dependent magnetic criticality in anisotropic magnets.
• It originates from Callen’s theory where a high anisotropy-to-exchange ratio leads to different critical temperatures along easy and hard crystallographic axes.
• Fe7S8 exemplifies this phenomenon with distinct critical points, enabling potential applications in anisotropic magnonics and spin transport.

The Curie temperature tensor is a proposed tensorial representation of direction-dependent magnetic criticality in crystalline magnets whose loss of long-range order is not governed by a single scalar Curie point. In this usage, motivated by Callen’s theory of anisotropic Curie temperatures and by experiments on monoclinic Fe$_7$S$_8$, different crystallographic directions can become paramagnetic at different temperatures; the tensor then summarizes the directional critical temperatures through a quadratic form $$T_C(\hat{\mathbf n})=\hat{\mathbf n}^{\mathsf T}\mathbf T\hat{\mathbf n}$$ (Armstrong et al., 2012). The term is, however, not universal: in other parts of magnetism, $T_C$ is treated as a scalar even when the exchange, anisotropy, and susceptibility are tensorial, so the expression “Curie temperature tensor” remains conceptually specific rather than standard across the entire literature (Tiwari et al., 2021).

1. Callen’s anisotropic Curie-temperature regime

In the conventional high-symmetry, strong-exchange limit, the Curie point is effectively a scalar. Exchange is nearly isotropic, magnetocrystalline anisotropy is small by comparison, and the uniform susceptibility follows the ordinary Curie–Weiss law with a single Weiss temperature. Callen’s 1961 quantum-mechanical internal-field theory identified a different regime: if magnetic anisotropy is sufficiently large relative to isotropic exchange, the spontaneous moment can collapse along a hard crystallographic direction at a lower temperature while remaining ordered along an easy direction to a substantially higher temperature (Armstrong et al., 2012).

The physical mechanism is the competition between exchange and anisotropy. Magnetocrystalline anisotropy and exchange anisotropy confine spins to a narrow cone along easy axes but spread them along hard axes. When the anisotropy-to-exchange ratio becomes large, thermal fluctuations destabilize long-range order more readily along hard directions, whereas easy directions remain stabilized to higher temperature. In Callen’s formulation this effect is intrinsically nonperturbative in the anisotropy-to-exchange ratio.

A standard dimensionless parameter is $\lambda$, proportional to $K/J$, where $K$ is a magnetocrystalline anisotropy constant and $J$ measures isotropic exchange. Within Callen’s internal-field formalism, the hard-axis collapse becomes appreciable when $\lambda \gtrsim 0.462$. For $\lambda>0$, corresponding to a hard axis, the spontaneous moment vanishes at a reduced critical temperature below the easy-axis value; for sufficiently strong easy-axis anisotropy, the easy-direction transition is pushed upward toward the exchange-energy scale. This directional splitting of criticality is the core phenomenon from which the Curie temperature tensor construction is derived.

2. Susceptibility, Landau theory, and tensor construction

A concise language for direction-dependent criticality is the Curie–Weiss framework generalized to tensors. In the scalar case,

$_8$0

where $_8$1 is the Curie constant and $_8$2 is the Weiss temperature. In the anisotropic case, the inverse uniform susceptibility is written as

$_8$3

with $_8$4 a Weiss-temperature tensor encoding anisotropic molecular-field couplings (Armstrong et al., 2012).

For a field applied along a unit vector $_8$5, the measured susceptibility is

$_8$6

Directional criticality is associated with the vanishing of the smallest eigenvalue of $_8$7, equivalently with the condition

$_8$8

In this picture, the largest eigenvalue of the Weiss-temperature tensor corresponds to the highest Curie point, while smaller eigenvalues can define lower critical temperatures for hard directions.

This motivates the practical definition

$_8$9

where $$T_C(\hat{\mathbf n})=\hat{\mathbf n}^{\mathsf T}\mathbf T\hat{\mathbf n}$$0 is the Curie temperature tensor. In mean-field language,

$$T_C(\hat{\mathbf n})=\hat{\mathbf n}^{\mathsf T}\mathbf T\hat{\mathbf n}$$1

An equivalent formulation arises in a Landau expansion for the uniform magnetization $$T_C(\hat{\mathbf n})=\hat{\mathbf n}^{\mathsf T}\mathbf T\hat{\mathbf n}$$2,

$$T_C(\hat{\mathbf n})=\hat{\mathbf n}^{\mathsf T}\mathbf T\hat{\mathbf n}$$3

with

$$T_C(\hat{\mathbf n})=\hat{\mathbf n}^{\mathsf T}\mathbf T\hat{\mathbf n}$$4

The instability then occurs when the smallest eigenvalue of $$T_C(\hat{\mathbf n})=\hat{\mathbf n}^{\mathsf T}\mathbf T\hat{\mathbf n}$$5 changes sign.

This construction is explicitly conditional. It presumes uniform magnetization, a single relevant irreducible representation of the magnetic point group, and a linear-response quadratic form whose anisotropy is only weakly temperature dependent until near criticality. Under those assumptions, $$T_C(\hat{\mathbf n})=\hat{\mathbf n}^{\mathsf T}\mathbf T\hat{\mathbf n}$$6 summarizes the symmetry of the directional instability. In monoclinic crystals it can be fully anisotropic and need not be diagonal in a laboratory frame.

3. Fe$$T_C(\hat{\mathbf n})=\hat{\mathbf n}^{\mathsf T}\mathbf T\hat{\mathbf n}$$7S$$T_C(\hat{\mathbf n})=\hat{\mathbf n}^{\mathsf T}\mathbf T\hat{\mathbf n}$$8 as a crystallographic realization

Fe$$T_C(\hat{\mathbf n})=\hat{\mathbf n}^{\mathsf T}\mathbf T\hat{\mathbf n}$$9S$T_C$0 (pyrrhotite) provides the clearest experimental realization of Callen’s prediction. The crystal is pseudohexagonal but slightly monoclinic, with eight formula units per cell and ordered vacancies in alternate iron layers normal to the pseudohexagonal $T_C$1-axis. Neutron diffraction shows that the magnetic moments lie within the $T_C$2 planes, with antiparallel orientation in adjacent Fe planes; the ordered vacancies make the two Fe sublattices inequivalent and produce ferrimagnetism (Armstrong et al., 2012).

The basal plane is effectively the easy magnetic plane. Within experimental resolution it shows near-isotropic Curie behavior, despite a small triaxial magnetocrystalline anisotropy, and all basal-plane directions remain magnetically ordered up to about $T_C$3–$T_C$4 K. By contrast, the pseudohexagonal $T_C$5-axis is a hard direction. Interlayer exchange is weak along $T_C$6 because of sulfur-layer spacing and vacancy ordering, and the $T_C$7-axis magnetization at room temperature is small and linear in field up to very high fields.

The basal-plane spontaneous magnetization $T_C$8 persists to about $T_C$9–$\lambda$0 K. This was established by magnetization isotherms and by calorimetry: extrapolating the steepest part of $\lambda$1 versus $\lambda$2 gives $\lambda$3–$\lambda$4 K, and differential scanning calorimetry on heating shows a Curie transition at $\lambda$5 K. Along the $\lambda$6-axis, magnetization–field curves are strictly linear up to at least $\lambda$7 T at elevated temperature, consistent with paramagnetic behavior; on cooling, a spontaneous magnetization emerges on top of the linear background, with increasing hysteresis at lower temperature.

The $\lambda$8-axis transition was initially estimated from Arrott plots of $\lambda$9 versus $K/J$0, whose extrapolated asymptotes indicated a transition near $K/J$1 K. The decisive thermodynamic determination came from high-sensitivity ac calorimetry, which revealed a sharp lambda anomaly in the heat capacity at $K/J$2 K, less than $K/J$3 K wide, and from both ac and dc susceptibilities measured along $K/J$4, which peak at the same temperature. Combining these probes yields

$K/J$5

In the principal-axis frame and within the reported experimental resolution, Fe$K/J$6S$K/J$7 therefore admits the approximate representation

$K/J$8

where the first two axes span the basal plane and the third is the pseudohexagonal $K/J$9-axis. Fits of the $K$0-axis order parameter to Callen’s hard-axis solutions place Fe$K$1S$K$2 in the regime $K$3, consistent with a substantially depressed hard-axis ordering temperature.

4. Experimental extraction and interpretational caveats

A Curie temperature tensor, if adopted operationally, must be reconstructed from orientation-resolved critical data rather than inferred from a single magnetization curve. In the general triclinic or monoclinic case, a symmetric $K$4 tensor has six independent components, so at least six non-collinear directional measurements of $K$5 are required unless crystal symmetry reduces the number of free parameters (Armstrong et al., 2012).

The directional critical temperature can be identified by combining magnetic and thermodynamic criteria. On the magnetic side, one may use divergence of $K$6 or Arrott-plot criteria. On the thermodynamic side, one may use sharp anomalies in heat capacity or magnetocaloric signatures. The fitted quadratic form $K$7 is then constrained by the symmetry of the magnetic point group.

Several caveats are essential. Precise alignment is critical, especially in Fe$K$8S$K$9: above its own $J$0, the $J$1-axis is paramagnetic, so even slight misalignment can project basal-plane magnetization onto the nominal $J$2-axis signal and mask the true directional behavior. Demagnetization fields must be treated properly in Arrott analysis; using applied rather than internal field tends to underestimate the transition temperature, which explains why the $J$3 K Arrott estimate lies below the $J$4 K thermodynamic transition. Domain effects further complicate low-field data, so susceptibilities should be extracted from the linear high-field portions of the curves. Finally, small in-plane anisotropies may require off-diagonal tensor elements if the experimental resolution is high enough to resolve them.

5. Terminological status, limits, and misconceptions

The principal conceptual limitation of the expression “Curie temperature tensor” is that it is not standard across all magnetic theory. In work on two-dimensional ferromagnets with exchange anisotropy, the critical temperature is treated explicitly as a scalar, while the tensorial objects are the microscopic exchange couplings, the single-ion anisotropy, the $J$5-tensor, and the susceptibility tensor (Tiwari et al., 2021).

That distinction is stated particularly sharply in the formulation

$J$6

where anisotropy is encoded in the exchange tensor $J$7. For easy-axis two-dimensional ferromagnets, these tensorial anisotropies open a spin-wave gap and thereby permit a finite scalar $J$8, consistent with the Mermin–Wagner theorem. The paper then computes $J$9 by Green’s-function, renormalized-spin-wave, and Monte Carlo methods as a scalar function of the anisotropic couplings.

This indicates that the Curie temperature tensor should not be regarded as a universally accepted thermodynamic observable. A more precise reading is that it is a phenomenological or mean-field summary of direction-dependent instabilities in systems displaying Callen’s anisotropic Curie-temperature phenomenon. In that restricted sense it is useful: it organizes crystallographic symmetry, susceptibility anisotropy, and critical temperatures into a single quadratic form. Outside that setting, the established convention remains that the Curie temperature of a given ferromagnetic phase is a scalar even when the underlying couplings are tensorial.

6. Relation to tensor Curie–Weiss models and broader significance

A separate source of ambiguity is the phrase tensor Curie–Weiss, which in statistical mechanics denotes higher-order mean-field interactions rather than a tensor-valued critical temperature. In the $\lambda \gtrsim 0.462$0-tensor Curie–Weiss Potts model, the distribution is

$\lambda \gtrsim 0.462$1

and the relevant critical parameter is a scalar inverse temperature $\lambda \gtrsim 0.462$2. Recent analysis derives Berry–Esseen-type convergence rates for the magnetization vector: $\lambda \gtrsim 0.462$3 at regular and critical points, $\lambda \gtrsim 0.462$4 at type-I special points, and $\lambda \gtrsim 0.462$5 at the type-II special point occurring only for $\lambda \gtrsim 0.462$6 (Bhowal et al., 2024).

Likewise, in the $\lambda \gtrsim 0.462$7-spin Curie–Weiss Ising model,

$\lambda \gtrsim 0.462$8

the critical object is the scalar threshold

$\lambda \gtrsim 0.462$9

which separates the paramagnetic and ferromagnetic phases. In that setting, $\lambda>0$0 is also the identifiability threshold for estimating the interaction order $\lambda>0$1 from a single sample when $\lambda>0$2 is known, while joint estimation of $\lambda>0$3 is impossible when $\lambda>0$4 is unknown (Mukherjee, 2024).

These usages are mathematically and physically distinct from the Curie temperature tensor proposed for anisotropic magnets. In the anisotropic-crystal setting, the objective is to represent directional criticality in real space; in tensor Curie–Weiss models, “tensor” refers to $\lambda>0$5-body interaction structure in mean-field probability models. The two ideas share the language of anisotropy or higher-order coupling, but not the same notion of critical temperature.

The broader significance of the Curie temperature tensor concept lies in materials such as Fe$\lambda>0$6S$\lambda>0$7, where a wide temperature interval separates an ordered easy plane from a hard direction that is already paramagnetic. Such materials act as intrinsic thermal magnetic switches and suggest anisotropic magnonics, direction-selective spin-transport elements, and sensing or logic components based on orientation-dependent phase transitions (Armstrong et al., 2012). Within that domain, the tensor formalism offers a compact way to connect symmetry, experiment, and exchange–anisotropy physics.