2D Magnets and Ferroelectrics

An atomically thin crystal can order magnetically or develop a switchable electric polarization. The key is symmetry: easy-axis anisotropy can stabilize a two-dimensional magnet, while a polar distortion or relative sliding can create two electric states. Layer number and stacking then determine how the responses of individual sheets add or cancel.

A local magnetic moment is not yet long-range order, and a polar crystal is not necessarily ferroelectric. We will connect these distinctions to a spin-wave calculation, an antiferromagnetic bilayer and a simple model of electric switching.

Required background. Two-Dimensional Materials supplies layer and environment effects. Ferromagnetism supplies spontaneous magnetization and domain distinctions; Magnetic Anisotropy supplies easy directions, spin-wave gaps and competing energy terms. Spontaneous Symmetry Breaking supplies the thermodynamic-limit criterion separating an ordered phase from a finite-sample response.

Magnetic order in an atomically thin layer

The general theories of ferromagnetism, antiferromagnetism and magnetic anisotropy apply here. The additional questions are how a finite thickness changes symmetry, and how the substrate and neighboring layers change the interactions.

A material is more than an isotropic spin model

A useful low-energy Hamiltonian for a localized-spin monolayer is

$$H={} -\sum_{\langle ij\rangle} J_{ij}\,\mathbf S_i\cdot\mathbf S_j -K\sum_i(S_i^z)^2 +\sum_{\langle ij\rangle} \mathbf D_{ij}\cdot \left(\mathbf S_i\times\mathbf S_j\right) -g\mu_B\mathbf B\cdot\sum_i\mathbf S_i +H_{\mathrm{dip}}.$$

Here the spin operators are dimensionless, so $J_{ij}$, $K$, and the components of $\mathbf D_{ij}$ have energy units. With this convention $J_{ij}>0$ favors parallel spins and, for $S\ge1$, $K>0$ favors the $z$ axis. For $S=1/2$, $(S_i^z)^2=1/4$ is a constant; anisotropic exchange or other terms must supply the orientation dependence. The Dzyaloshinskii–Moriya vectors $\mathbf D_{ij}$, dipolar term, exchange range, and allowed anisotropies are fixed by crystal and interface symmetry. Itinerant magnets require an electronic description in which exchange splitting and collective order emerge from the band structure and interactions; fitting them to rigid spins can still be useful, but it is an approximation.

“Two-dimensional magnet” is therefore an experimental platform label. It may mean:

• a monolayer with intrinsic ferromagnetic or antiferromagnetic order;
• a few-layer crystal whose planes are internally ordered but coupled ferro- or antiferromagnetically;
• an itinerant two-dimensional electron system with spontaneous spin or valley polarization;
• a nonmagnetic layer exchange-coupled to a nearby magnet;
• a finite flake with a long but finite correlation or blocking time.

A field-induced moment or a slowly relaxing finite flake need not be a spontaneously ordered thermodynamic phase.

Representative material classes

Platform	Typical low-energy order	What the example teaches
monolayer $\mathrm{CrI}_3$	out-of-plane Ising-like ferromagnetism	spin–orbit-generated anisotropy can stabilize monolayer order
few-layer $\mathrm{Cr}_2\mathrm{Ge}_2\mathrm{Te}_6$	soft ferromagnetism	a small field can strongly modify the effective spin-wave gap
$\mathrm{Fe}_3\mathrm{GeTe}_2$	itinerant ferromagnetism	carrier density and interfaces can strongly shift magnetic behavior
$\mathrm{FePS}_3$	Ising-like antiferromagnetism	a discrete order parameter can remain ordered at monolayer thickness
$\mathrm{CrCl}_3$	weak-anisotropy layered antiferromagnetism	shape anisotropy, spin flop, and layer parity can dominate transport
$\mathrm{CrSBr}$	anisotropic A-type antiferromagnetic semiconductor	charge, excitons, and magnons can directly sense layer-resolved order

Huang and collaborators established the monolayer $\mathrm{CrI}_3$ example by Kerr microscopy, whereas Gong and collaborators studied the markedly field-sensitive ordering of thin $\mathrm{Cr}_2\mathrm{Ge}_2\mathrm{Te}_6$. These experiments illustrate different anisotropy regimes.

The listed phase is not a universal property of a chemical formula. Polytype, stacking, oxidation, strain, defects, carrier density, encapsulation, and measurement timescale can change the result.

What the probes measure

Magneto-optical Kerr rotation, reflective magnetic circular dichroism, anomalous Hall response, tunnelling magnetoresistance, Raman zone folding, second-harmonic generation, scanning magnetometry, and exciton shifts all provide useful information. None measures “magnetism” in an assumption-free way.

For example, Kerr rotation can track a time-reversal-odd optical response but may depend on resonances and layer interference. A tunnelling junction can amplify a spin-filter transition while remaining indirect about the spatial order inside each layer. Raman anomalies can identify symmetry or zone folding without directly measuring the ordered moment. Comparing a symmetry-sensitive probe with temperature dependence, domains and excitation energies helps distinguish an ordered phase from blocking or an optical resonance. A field-induced response alone does not establish spontaneous order at zero field.

Anisotropy and Mermin–Wagner Considerations

What the theorem actually excludes

The Mermin–Wagner theorem excludes spontaneous ferromagnetic or antiferromagnetic long-range order at nonzero temperature in an infinite one- or two-dimensional isotropic Heisenberg model with sufficiently short-range exchange. More generally, thermal fluctuations obstruct spontaneous breaking of suitable continuous internal symmetries in low dimensions under the theorem’s locality assumptions.

It does not say that every thin magnet must be disordered. Its assumptions change when:

• easy-axis anisotropy reduces the order-parameter symmetry to a discrete choice;
• easy-plane order supports a Berezinskii–Kosterlitz–Thouless regime with quasi-long-range order;
• dipolar interactions add long-range and shape-dependent terms;
• interlayer exchange in a bulk stack produces a three-dimensional crossover; a fixed finite number of isotropic short-range layers still has effectively two-dimensional infrared fluctuations, as explained in Low-Dimensional Quantum Matter;
• finite lateral size cuts off the longest wavelength;
• a substrate, strain, field, or proximity layer explicitly breaks spin symmetry.

These mechanisms are not interchangeable. An easy-axis monolayer can have an ordinary finite-temperature transition, while an ideal easy-plane magnet has vortex physics rather than conventional long-range order. A sixfold in-plane crystal anisotropy can further lock the phase at long distances. A uniform magnetic field explicitly selects an orientation; removing the infrared divergence in a field does not by itself demonstrate spontaneous order.

The infrared divergence in one line

For a two-dimensional ferromagnet with long-wavelength magnon dispersion

$$\varepsilon_{\mathbf q} = \Delta_{\mathrm{an}} +D_s q^2,$$

the long-wavelength magnon number per area is

$$n_{\mathrm{mag}}^{\mathrm{IR}}(T) = \int_{q_{\min}}^\Lambda \frac{d^2q}{(2\pi)^2}\, n_B(\varepsilon_{\mathbf q}).$$

In the classical low-energy part of the integral,

$$n_{\mathrm{mag}}^{\mathrm{IR}}(T) \simeq \frac{k_BT}{4\pi D_s} \ln \left( \frac{\Delta_{\mathrm{an}}+D_s\Lambda^2} {\Delta_{\mathrm{an}}+D_s q_{\min}^2} \right).$$

Here $q$ is a wavevector, $D_s$ has units of energy times length squared, $\Lambda$ bounds the long-wavelength regime, and $q_{\min}\sim2\pi/L$ for a finite flake of lateral size $L$. The displayed integral has units of inverse area. In the simple localized-spin ferromagnet, multiplying it by $g\mu_B$ gives the moment reduction per area; multiplying instead by the area per magnetic site gives the dimensionless spin reduction per site. The classical approximation requires the retained low-energy modes to satisfy $\varepsilon_{\mathbf q}\ll k_BT$. In the isotropic thermodynamic limit, $\Delta_{\mathrm{an}}=0$ and $q_{\min}\to0$, so the logarithm diverges. A nonzero anisotropy gap or finite size regularizes the integral.

This calculation is an infrared diagnosis, not a quantitative transition-temperature theory. In a weakly anisotropic renormalized-classical regime, one often finds the schematic logarithmic scale

$$k_BT_c \sim \frac{4\pi\rho_s} {\ln\!\left(C\rho_s/\Delta_{\mathrm{an}}\right)},$$

where $\rho_s$ is spin stiffness and $C$ is nonuniversal. Exchange sets the large stiffness scale, while a much smaller anisotropy can decide whether finite-temperature long-range order exists. Reliable material estimates require the actual spin, lattice, exchange tensor, dipolar terms, and fluctuation method.

Three mechanisms in atomically thin ferroic order. The schematic dispersions have the same spin stiffness and differ by an anisotropy gap, which cuts off the infrared magnon divergence. Identical ferromagnetic sheets with antiparallel neighboring moments give zero net moment for an even number of layers and an uncompensated moment for an odd number. In a suitable polar bilayer, sliding can interchange opposite out-of-plane polarizations. Filled and open circles distinguish the two sheets; their atomic positions are schematic.

Anisotropy must be measured in the same device

Magnetic anisotropy can change with thickness because surface crystal fields, interlayer hybridization, strain, carrier density, and demagnetizing fields all change. A bulk easy axis cannot simply be assigned to a monolayer. Useful cross-checks include:

• angular dependence of saturation and spin-flop fields;
• zero-momentum magnon gaps from optical or microwave spectroscopy;
• domain-wall orientation and width;
• field-dependent transition temperatures;
• layer-resolved calculations tied to the measured structure.

The phrase “spin–orbit coupling stabilizes the magnet” is incomplete until it identifies the symmetry-allowed anisotropy and its scale relative to temperature and stiffness.

Layer-Dependent Magnetism

A stack is a coupled set of magnetic sheets

When each layer is internally ferromagnetic, a macrospin model describes the relative alignment of the layers:

$$\frac{F}{A} ={} J_\perp \sum_{\ell=1}^{N-1} \mathbf m_\ell\cdot\mathbf m_{\ell+1} -K_{2D}\sum_{\ell=1}^{N} (\mathbf m_\ell\cdot\hat{\mathbf z})^2 -\mu_0M_{2D} \mathbf H\cdot \sum_{\ell=1}^{N}\mathbf m_\ell .$$

The unit vector $\mathbf m_\ell$ describes layer $\ell$; $J_\perp>0$ favors antiparallel neighboring layers in this convention; and $J_\perp$, $K_{2D}$, and $\mu_0M_{2D}H$ are energies per area. For collinear Ising layers, antiferromagnetic interlayer coupling gives a compensated even-layer stack and an uncompensated odd-layer stack.

For a bilayer in a perpendicular field, omit the common anisotropy contribution $-2K_{2D}$ from the two collinear energies. The ideal antiparallel and field-aligned energies are

$$\frac{F_{\mathrm{AF}}}{A}=-J_\perp, \qquad \frac{F_{\mathrm{F}}}{A} =J_\perp-2\mu_0M_{2D}H.$$

Their crossing occurs at

$$H_c = \frac{J_\perp}{\mu_0M_{2D}}.$$

This is a diagnostic baseline. Canting, anisotropy, domains, thermal fluctuations, unequal surface layers, and hysteresis modify the measured transition.

Layer number is not the only structural variable

Monolayer $\mathrm{CrI}_3$ is an out-of-plane ferromagnet, while commonly studied bilayers exhibit layered antiferromagnetic coupling and a field-driven metamagnetic transition. Trilayers can recover a net moment. Yet interlayer exchange in $\mathrm{CrI}_3$ is strongly stacking dependent: lateral registry, pressure, reconstruction, and twist can change its sign or create coexisting ferro- and antiferromagnetic regions. The first-principles study by Sivadas and collaborators explicitly demonstrated stacking-dependent changes of interlayer exchange. Two bilayers with different stacking can therefore have different magnetic ground states.

$\mathrm{CrCl}_3$ provides a different limit with weak in-plane anisotropy and layer-parity-dependent spin-flop signatures. CrSBr combines ferromagnetic intralayer correlations with antiferromagnetic interlayer alignment, strong in-plane anisotropy, semiconducting transport, and excitons that sense the spin configuration. These systems show why one chemical family cannot stand in for all two-dimensional magnetism.

Gates can act through several channels

An applied gate can modify:

• carrier-mediated exchange and itinerant spin polarization;
• interlayer exchange through layer-selective doping;
• magnetic anisotropy through orbital occupation and interfacial electric fields;
• structural registry through electrostatic pressure or sliding;
• optical selection rules and the probe response itself.

The control variables should therefore be separated into carrier density $n$ and displacement field $D$. In a dual-gated device these are independent linear combinations of gate voltages only after the geometric and quantum capacitances are calibrated. A changed Kerr or tunnelling signal at fixed voltage is not automatically a changed magnetic ground state.

Ferroelectricity in van der Waals Systems

Polar, piezoelectric, and ferroelectric are different claims

A polar crystal has a symmetry-allowed spontaneous electric dipole. A piezoelectric crystal develops polarization under strain. A ferroelectric has at least two stable, symmetry-related polarization states that can be reversibly selected by an electric field. A pyroelectric response tracks a temperature-dependent spontaneous polarization. Ferroelectric implies polar, but polar and piezoelectric do not imply switchable ferroelectricity.

In an atomically thin system, the natural polarization is dipole moment per area,

$$\mathbf P_{2D} = \frac{\mathbf p}{A}, \qquad [\mathbf P_{2D}] =\mathrm{C\,m^{-1}}.$$

Reporting a three-dimensional value requires an assigned effective thickness,

$$\mathbf P_{3D} = \frac{\mathbf P_{2D}}{t_{\mathrm{eff}}}, \qquad [\mathbf P_{3D}] =\mathrm{C\,m^{-2}}.$$

The latter therefore depends on the thickness convention. For in-plane polarization, the bound line charge at an edge with outward normal $\hat{\mathbf n}$ is

$$\lambda_b = \mathbf P_{2D}\cdot\hat{\mathbf n}.$$

For an insulating layer periodic in the plane, the in-plane polarization is a Berry-phase quantity defined modulo a polarization quantum; follow a continuous insulating switching path to compare branches. Berry-Phase Polarization and Charge Pumping explains this construction. The out-of-plane dipole of an isolated, neutral slab is instead obtained from its charge distribution along the nonperiodic direction. The insulating Berry-phase construction should not be assigned to a metal merely because its structure is polar.

Depolarization competes with the polar distortion

For a scalar out-of-plane sheet polarization $P$, the local Landau energy density (energy per area) can be written

$$g(P,\nabla P) ={} \frac{a}{2}P^2 +\frac{b}{4}P^4 +\frac{\kappa_P}{2} \lvert\nabla P\rvert^2 -E_zP +g_{\mathrm{dep}}.$$

Here $G=\int_A g\,d^2r$ and $b>0$. With zero applied field and negligible depolarization energy, $a<0$ gives uniform minima $P=\pm\sqrt{-a/b}$. A nonzero depolarization term changes their location and can remove the instability.

For an ideal open-circuit slab with three-dimensional polarization $P_{3D}$, thickness $t$, and background relative permittivity $\epsilon_b$, a simple unscreened estimate is

$$E_{\mathrm{dep}} \simeq -\frac{P_{3D}}{\epsilon_0\epsilon_b}, \qquad \frac{G_{\mathrm{dep}}}{A} \simeq \frac{tP_{3D}^2} {2\epsilon_0\epsilon_b}.$$

Electrodes, mobile carriers, adsorbates, substrates, and domain formation reduce or redistribute this cost. They can also create trapped-charge hysteresis that imitates switching, so screening is both a stabilizing mechanism and an experimental confounder.

Several microscopic routes exist

Route	Representative systems	Switching coordinate
intrinsic polar distortion	$\mathrm{SnTe}$, $\mathrm{CuInP}_2\mathrm{S}_6$, $\alpha\text{-}\mathrm{In}_2\mathrm{Se}_3$	atomic displacement or order–disorder coordinate within a layer
polar few-layer stacking	few-layer $\mathrm{WTe}_2$	relative registry creates polarity from nonpolar monolayers
sliding ferroelectricity	parallel bilayer hBN, rhombohedral TMD bilayers	in-plane translation changes the sign of out-of-plane polarization
spin-driven polarity	$\mathrm{NiI}_2$ and other candidate type-II multiferroics	noncollinear magnetic order breaks inversion and induces $P$

In sliding ferroelectrics, nonpolar monolayers can form a polar interface. AB and BA registries are related by a symmetry operation and carry opposite out-of-plane dipoles; an electric field moves domain walls or drives relative sliding. A small twist creates a moiré network of registries and polar domains rather than one uniform polarization. Yasuda and collaborators realized this stacking route in bilayer boron nitride, and Wang and collaborators demonstrated electrically switchable rhombohedral TMD bilayers.

The few-layer $\mathrm{WTe}_2$ experiment of Fei and collaborators demonstrates that screening by itinerant carriers does not categorically forbid switchable polarity when the sample is thin enough for gates to penetrate. It does not imply that an arbitrary polar metal is ferroelectric.

Distinguishing switching from charging

A ferroelectric switches between polar structures and retains a selected state after the writing field is removed. Structural or domain-sensitive measurements can connect the two readout states to those structures. Switching-current measurements require subtraction of leakage and ordinary capacitive current.

Piezoresponse force microscopy is useful, but electrostatic forces, mobile ions and trapped charge can also produce phase contrast or a butterfly-shaped loop. Off-field readout, retention and frequency dependence help test these alternatives; a loop alone does not identify the switching coordinate.

Coupled Spin, Valley, and Lattice Orders

Symmetry comes before a coupling constant

Let $P$ be a polar order parameter, $\mathbf M$ a magnetization, $\mathbf L$ a Néel vector, $\varepsilon_{ij}$ strain, and $\nu_v=n_K-n_{K'}$ a valley polarization. A schematic uniform free energy may contain

$$\frac{G_{\mathrm{int}}}{A} ={} \lambda_{PM}P^2\mathbf M^2 +\lambda_{PL}P^2\mathbf L^2 +q_{ij}P^2\varepsilon_{ij} +b_{ijkl}M_iM_j\varepsilon_{kl} +g_vM_z\nu_v +\cdots .$$

Every displayed term is even under time reversal; whether it is allowed by spatial symmetry depends on the crystal, stack, and definition of the order parameters. If the high-symmetry parent has inversion, the complete coupling must be inversion-even: a term odd in $P$ is allowed only when its remaining factor is also inversion-odd. Spatially varying magnetic order can supply such a factor, as in spin-driven polarization below. Once a fixed polar interface already breaks inversion, lower-order couplings can become allowed.

For an electric- and magnetic-field thermodynamic potential,

$$dg = -P_j\,dE_j -\mu_0M_i\,dH_i,$$

the equilibrium linear magnetoelectric response obeys

$$\alpha_{ij} \equiv \frac{\partial P_j}{\partial H_i} = \mu_0 \frac{\partial M_i}{\partial E_j}.$$

Static linear magnetoelectricity is forbidden by either inversion or time reversal separately, although their product may remain a symmetry. Nonlinear, finite-frequency, interfacial, and nonequilibrium responses obey different selection rules.

Coexistence and coupling are different

In a type-I multiferroic, magnetic and polar orders have largely distinct microscopic origins; coupling can be useful without being strong. In a type-II multiferroic, magnetic order itself induces polarization. The microscopic spin-current mechanism of Katsura, Nagaosa, and Balatsky gives a schematic bond contribution

$$\mathbf P_{ij} \propto \hat{\mathbf e}_{ij} \times \left(\mathbf S_i\times\mathbf S_j\right),$$

but this inverse-Dzyaloshinskii–Moriya or spin-current form is not universal. Exchange striction, spin-dependent ligand hybridization, and other symmetry-allowed mechanisms can dominate.

$\mathrm{NiI}_2$ is an important active example. Song and collaborators reported optical evidence for intertwined magnetic and polar order down to the monolayer limit. Gao and collaborators measured resonant dynamical magnetoelectric coupling in exfoliated, bulk-like flakes about $100\,\mathrm{nm}$ thick; that measurement is not itself a monolayer result. Meanwhile, the calculations of Liu and collaborators find competing spiral, striped, ferroic, and antiferroic states whose order depends on substrate constraints, strain, and layer count. These distinct results support a van der Waals multiferroic platform while leaving the exact monolayer microscopic ground state structure sensitive.

Valleys can act as magnetic sensors

In a transition-metal dichalcogenide monolayer, time reversal exchanges the $K$ and $K'$ valleys. Exchange proximity to a magnet can therefore produce a valley-contrasting energy shift. In $\mathrm{WSe}_2$/$\mathrm{CrI}_3$ stacks, Zhong and collaborators resolved different sensitivities: spin-dependent charge transfer is dominated by the interfacial $\mathrm{CrI}_3$ layer, while the proximity exchange field also depends on the stack as a whole. The exciton response can therefore reveal layer-resolved magnetic transitions, but it is not one universal local magnetometer.

This is a coupled response, not necessarily spontaneous magnetism in the semiconductor. The observed optical splitting can also depend on spin-selective charge transfer, band alignment, exciton binding, and optical pumping. A calibrated magnetic probe and field-reversal symmetry are needed before translating an exciton shift into an “effective magnetic field.”

The lattice is an active participant

In atomically thin materials, strain and stacking alter bond angles, crystal fields, exchange paths, anisotropy, polarization, and valley energies at once. This makes twist, pressure, and domain walls unusually effective control variables, but it also creates causal ambiguity. A convincing control experiment should track the structural coordinate and show that the proposed magnetic or polar response follows it reversibly.

For example, a gate-induced optical shift could arise from a changed magnetization, carrier density or exciton binding energy. Measuring the structural change and an independent magnetic response helps identify which mechanism operated.

Exercises

Exercise 1: infrared magnons

Starting from

$$I = \int_{q_{\min}}^\Lambda \frac{d^2q}{(2\pi)^2} \frac{k_BT} {\Delta+Dq^2},$$

evaluate $I$ and identify the two independent infrared cutoffs.

Solution

Using $d^2q=2\pi q\,dq$,

$$\begin{aligned} I &= \frac{k_BT}{2\pi} \int_{q_{\min}}^\Lambda \frac{q\,dq}{\Delta+Dq^2} \\ &= \frac{k_BT}{4\pi D} \ln \left( \frac{\Delta+D\Lambda^2} {\Delta+Dq_{\min}^2} \right). \end{aligned}$$

With physical wavevectors, $I$ is a number per area. The anisotropy gap $\Delta$ and finite-size wavevector $q_{\min}\sim2\pi/L$ are independent infrared cutoffs. If both vanish, the integral diverges logarithmically.

Exercise 2: finite size versus anisotropy

For a flake of size $L$, define an effective finite-size energy $\Delta_L=D(2\pi/L)^2$. Which cutoff controls the magnon integral when $\Delta\gg\Delta_L$ and when $\Delta\ll\Delta_L$? Does finite size establish a thermodynamic phase transition?

Solution

The denominator at the lowest mode is $\Delta+\Delta_L$. When $\Delta\gg\Delta_L$, anisotropy controls the infrared physics. When $\Delta\ll\Delta_L$, the finite lateral size controls it.

A finite flake can have a long correlation length and a stable moment over the measurement time, but it has no singular thermodynamic phase transition by itself. A transition is defined after a thermodynamic-limit or finite-size-scaling analysis.

Exercise 3: bilayer spin-flip field

Use the bilayer macrospin energies above to derive the field at which an antiparallel Ising bilayer becomes field aligned. What assumptions make this estimate fail?

Solution

Equating the two energies,

$$-J_\perp = J_\perp-2\mu_0M_{2D}H_c,$$

gives

$$\mu_0M_{2D}H_c = J_\perp.$$

The estimate assumes identical rigid layers, collinear Ising spins, zero temperature, a uniform single domain, and no hysteretic barrier. Canting, unequal moments, anisotropy, domains, thermal fluctuations, and stacking inhomogeneity shift or broaden the transition.

Exercise 4: sheet polarization

A monolayer has $P_{2D}=5.0\times10^{-13}\,\mathrm{C\,m^{-1}}$. Convert it to a three-dimensional polarization using $t_{\mathrm{eff}}=0.70\,\mathrm{nm}$. Repeat with $t_{\mathrm{eff}}=1.0\,\mathrm{nm}$ and explain the difference.

Solution

For $t_{\mathrm{eff}}=0.70\,\mathrm{nm}$,

$$\begin{aligned} P_{3D} &= \frac{5.0\times10^{-13}} {0.70\times10^{-9}} \\ &= 7.1\times10^{-4}\, \mathrm{C\,m^{-2}}. \end{aligned}$$

For $t_{\mathrm{eff}}=1.0\,\mathrm{nm}$,

$$P_{3D} = 5.0\times10^{-4}\,\mathrm{C\,m^{-2}}.$$

The physical sheet dipole is unchanged. The numerical three-dimensional value changes because the assigned thickness changes, which is why $P_{2D}$ is the intrinsic comparison for isolated layers.

Exercise 5: homogeneous ferroelectric spinodal

For

$$f(P) = \frac{a}{2}P^2 +\frac{b}{4}P^4 -EP,$$

with $a<0$ and $b>0$, find the field magnitude at which one homogeneous metastable minimum disappears.

Solution

At the spinodal, both derivatives vanish:

$$aP+bP^3-E=0, \qquad a+3bP^2=0.$$

Thus

$$P_{\mathrm{sp}}^2 = -\frac{a}{3b}.$$

Substitution gives the magnitude

$$\lvert E_{\mathrm{sp}}\rvert = \frac{2\lvert a\rvert^{3/2}} {3\sqrt{3b}}.$$

This is a homogeneous mean-field upper scale, not a realistic coercive field. Actual switching usually proceeds by nucleation and domain-wall motion at a smaller, geometry- and defect-dependent field.

Exercise 6: coupling shifts a magnetic transition

Consider

$$f ={} \frac{a_P}{2}P^2 +\frac{b_P}{4}P^4 +\frac{a_M}{2}(T-T_{M0})M^2 +\frac{b_M}{4}M^4 +\frac{\lambda}{2}P^2M^2.$$

Assume $P$ is already nonzero and varies slowly near the magnetic transition. Find the shifted mean-field magnetic transition temperature and interpret the sign of $\lambda$.

Solution

The quadratic coefficient of $M$ is

$$a_M(T-T_{M0}) +\lambda P^2.$$

It vanishes at

$$T_M = T_{M0} -\frac{\lambda P^2}{a_M}.$$

For $a_M>0$, positive $\lambda$ makes polarization and magnetization compete and lowers $T_M$; negative $\lambda$ makes them cooperate and raises $T_M$. This conclusion concerns a biquadratic equilibrium coupling. It does not by itself imply that reversing $P$ reverses $M$, because the term depends on $P^2$.

Where to go next

Low-Dimensional Quantum Matter develops the fluctuation and finite-size arguments. Moiré Superlattices explains how twist produces spatially varying registry, and Engineered Heterostructures develops the transfer of exchange and other interactions across an interface.

References

• Z. Fei et al., “Ferroelectric Switching of a Two-Dimensional Metal,” Nature 560, 336–339 (2018), doi:10.1038/s41586-018-0336-3.
• F. Y. Gao et al., “Giant Chiral Magnetoelectric Oscillations in a van der Waals Multiferroic,” Nature 632, 273–279 (2024), doi:10.1038/s41586-024-07678-5.
• C. Gong et al., “Discovery of Intrinsic Ferromagnetism in Two-Dimensional van der Waals Crystals,” Nature 546, 265–269 (2017), doi:10.1038/nature22060.
• B. Huang et al., “Layer-Dependent Ferromagnetism in a van der Waals Crystal down to the Monolayer Limit,” Nature 546, 270–273 (2017), doi:10.1038/nature22391.
• H. Katsura, N. Nagaosa, and A. V. Balatsky, “Spin Current and Magnetoelectric Effect in Noncollinear Magnets,” Physical Review Letters 95, 057205 (2005), doi:10.1103/PhysRevLett.95.057205.
• N. Liu et al., “Competing Multiferroic Phases in Monolayer and Few-Layer $\mathrm{NiI}_2$,” Physical Review B 109, 195422 (2024), doi:10.1103/PhysRevB.109.195422.
• N. D. Mermin and H. Wagner, “Absence of Ferromagnetism or Antiferromagnetism in One- or Two-Dimensional Isotropic Heisenberg Models,” Physical Review Letters 17, 1133–1136 (1966), doi:10.1103/PhysRevLett.17.1133.
• N. Sivadas, S. Okamoto, X. Xu, C. J. Fennie, and D. Xiao, “Stacking-Dependent Magnetism in Bilayer $\mathrm{CrI}_3$,” Nano Letters 18, 7658–7664 (2018), doi:10.1021/acs.nanolett.8b03321.
• Q. Song et al., “Evidence for a Single-Layer van der Waals Multiferroic,” Nature 602, 601–605 (2022), doi:10.1038/s41586-021-04337-x.
• X. Wang et al., “Interfacial Ferroelectricity in Rhombohedral-Stacked Bilayer Transition Metal Dichalcogenides,” Nature Nanotechnology 17, 367–371 (2022), doi:10.1038/s41565-021-01059-z.
• K. Yasuda, X. Wang, K. Watanabe, T. Taniguchi, and P. Jarillo-Herrero, “Stacking-Engineered Ferroelectricity in Bilayer Boron Nitride,” Science 372, 1458–1462 (2021), doi:10.1126/science.abd3230.
• D. Zhong et al., “Layer-Resolved Magnetic Proximity Effect in van der Waals Heterostructures,” Nature Nanotechnology 15, 187–191 (2020), doi:10.1038/s41565-019-0629-1.

Further reading

• M. Dawber, K. M. Rabe, and J. F. Scott, “Physics of Thin-Film Ferroelectric Oxides,” Reviews of Modern Physics 77, 1083–1130 (2005), doi:10.1103/RevModPhys.77.1083.
• M. Gibertini et al., “Magnetic 2D Materials and Heterostructures,” Nature Nanotechnology 14, 408–419 (2019), doi:10.1038/s41565-019-0438-6.
• R. D. King-Smith and D. Vanderbilt, “Theory of Polarization of Crystalline Solids,” Physical Review B 47, 1651–1654 (1993), doi:10.1103/PhysRevB.47.1651.
• J. M. Kosterlitz and D. J. Thouless, “Ordering, Metastability and Phase Transitions in Two-Dimensional Systems,” Journal of Physics C 6, 1181–1203 (1973), doi:10.1088/0022-3719/6/7/010.