Magnon Kerr Effect in Cavity Magnonics

• Magnon Kerr Effect is the nonlinear magnon self-interaction that produces a population-dependent frequency shift via a quartic anisotropy term in YIG systems.
• Experiments in YIG spheres and thin films demonstrate power-dependent magnon and cavity shifts, bistability, and hysteretic transmission in cavity-magnonic setups.
• The effect underpins advanced functionalities such as squeezed-magnon interfaces, enhanced spin–magnon coupling, and tunable nonreciprocity in hybrid quantum systems.

The magnon Kerr effect (MKE) is the Kerr-type nonlinearity of a magnon mode: the magnon resonance acquires a self-induced frequency shift that depends on magnon population through a quartic magnon self-interaction. In the cavity-magnonics literature, the effect is most explicitly developed for yttrium iron garnet (YIG), where magnetocrystalline anisotropy generates an effective $K m^\dagger m^\dagger m m$ term after bosonizing the collective spin dynamics. The resulting anharmonicity produces power-dependent magnon and cavity shifts, bistability, squeezing, enhanced spin–magnon interfaces, multimode non-Markovian dynamics, and geometry-controlled nonlinear hybrid behavior (Wang et al., 2016, Zhang et al., 2019, Xiong et al., 2021, Petrosyan et al., 29 Jan 2026).

1. Microscopic definition and origin

In the MKE literature, the starting point is a collective magnetic mode, usually the Kittel mode of a uniformly magnetized YIG body, whose energy is not strictly harmonic. For a uniformly magnetized YIG sphere, the macrospin Hamiltonian is written as

$$H_{\rm m}=-\gamma B_{0}S_{z}-\frac{\mu_{0}\gamma^{2}K_{\rm an}}{M^{2}V_{\rm m}}S_{z}^{2},$$

so that the anisotropy-induced $S_z^2$ term becomes a Kerr-type magnon interaction after the Holstein–Primakoff transformation $S_z=S-b^\dagger b$. In bosonic form, the hybrid cavity–magnon Hamiltonian contains

$$K\,b^\dagger b\,b^\dagger b, \qquad K=\frac{\mu_{0}K_{\rm an}\gamma^{2}}{M^{2}V_{\rm m}},$$

which makes the magnon frequency depend on $\langle b^\dagger b\rangle$ (Wang et al., 2016).

The same physical origin is emphasized in later theory: in a small YIG sphere inside a microwave cavity, magnetocrystalline anisotropy produces a nonlinear Dicke-type model whose Kerr coefficient scales as $K\propto 1/V_m$, so reducing magnetic volume enhances the nonlinearity while leaving the bare magnon frequency unchanged. The sign of $K$ is not universal. For a YIG sphere with the crystallographic axis $[110]$ aligned with the bias field,

$$K=-\frac{13\,\mu_0 K_{\rm an}\gamma^2}{16 M^2 V_m},$$

whereas for $$H_{\rm m}=-\gamma B_{0}S_{z}-\frac{\mu_{0}\gamma^{2}K_{\rm an}}{M^{2}V_{\rm m}}S_{z}^{2},$$0 alignment,

$$H_{\rm m}=-\gamma B_{0}S_{z}-\frac{\mu_{0}\gamma^{2}K_{\rm an}}{M^{2}V_{\rm m}}S_{z}^{2},$$1

Accordingly, $$H_{\rm m}=-\gamma B_{0}S_{z}-\frac{\mu_{0}\gamma^{2}K_{\rm an}}{M^{2}V_{\rm m}}S_{z}^{2},$$2 yields a blue shift and $$H_{\rm m}=-\gamma B_{0}S_{z}-\frac{\mu_{0}\gamma^{2}K_{\rm an}}{M^{2}V_{\rm m}}S_{z}^{2},$$3 a red shift of the magnon resonance (Zhang et al., 2019).

This microscopic definition distinguishes MKE from generic nonlinear magnetization dynamics. In this usage, the effect is specifically the occupation-dependent self-energy of a magnon mode arising from anisotropy-induced quartic terms, rather than a generic high-power shift from heating or multimagnon scattering.

2. YIG-sphere cavity experiments

An experimental demonstration of the MKE was reported in a strongly coupled cavity–magnon system formed by a 1 mm YIG sphere and a three-dimensional rectangular copper cavity operated in the $$H_{\rm m}=-\gamma B_{0}S_{z}-\frac{\mu_{0}\gamma^{2}K_{\rm an}}{M^{2}V_{\rm m}}S_{z}^{2},$$4 mode with $$H_{\rm m}=-\gamma B_{0}S_{z}-\frac{\mu_{0}\gamma^{2}K_{\rm an}}{M^{2}V_{\rm m}}S_{z}^{2},$$5 GHz. The sphere was placed at the magnetic-field antinode, the static field $$H_{\rm m}=-\gamma B_{0}S_{z}-\frac{\mu_{0}\gamma^{2}K_{\rm an}}{M^{2}V_{\rm m}}S_{z}^{2},$$6 was aligned along the hard axis $$H_{\rm m}=-\gamma B_{0}S_{z}-\frac{\mu_{0}\gamma^{2}K_{\rm an}}{M^{2}V_{\rm m}}S_{z}^{2},$$7, and a separate loop antenna directly pumped the YIG. The cavity and Kittel mode were strongly coupled with $$H_{\rm m}=-\gamma B_{0}S_{z}-\frac{\mu_{0}\gamma^{2}K_{\rm an}}{M^{2}V_{\rm m}}S_{z}^{2},$$8 MHz, while the Kerr measurements were performed in a dispersive regime with $$H_{\rm m}=-\gamma B_{0}S_{z}-\frac{\mu_{0}\gamma^{2}K_{\rm an}}{M^{2}V_{\rm m}}S_{z}^{2},$$9 MHz $S_z^2$0. The measured linewidths were $S_z^2$1 MHz for the cavity and $S_z^2$2 MHz for the Kittel mode, giving a cooperativity $S_z^2$3 (Wang et al., 2016).

When the YIG was pumped strongly enough to generate a large magnon population, both the Kittel mode and magnetostatic modes shifted to higher frequency, while the cavity central frequency also moved by a smaller amount through dispersive backaction. The cavity shift was about 0.7 MHz at 11 dBm relative to the unpumped case. In the effective dispersive Hamiltonian, the cavity and magnon shifts are

$S_z^2$4

The Kittel mode follows the analytical power law

$S_z^2$5

with linear scaling $S_z^2$6 at small drive and $S_z^2$7 at large drive. The agreement was reported as very good for the Kittel mode, while magnetostatic modes showed only approximate agreement because their magnetization is not spatially uniform (Wang et al., 2016).

The experiment established the operational signature of MKE in cavity magnonics: a power-dependent magnon blueshift, accompanied by a smaller cavity shift, quantitatively attributable to the magnon Kerr term rather than to a cavity nonlinearity.

3. Mean-field theory, bistability, and cavity transmission

A subsequent theory developed the MKE in cavity magnonics as a bistable nonlinear response problem and analyzed two drive configurations: direct pumping of the YIG sphere and direct pumping of the cavity. In mean-field form, the Kerr interaction renormalizes the magnon frequency by

$S_z^2$8

For direct YIG pumping, the steady-state Kerr shift obeys

$S_z^2$9

whereas for direct cavity pumping the corresponding equation is

$S_z=S-b^\dagger b$0

The two equations have the same nonlinear structure but different effective powers, which makes cavity pumping much less efficient in the dispersive regime when $S_z=S-b^\dagger b$1 (Zhang et al., 2019).

Bistability occurs when the $S_z=S-b^\dagger b$2–$S_z=S-b^\dagger b$3 curve develops turning points, determined by

$S_z=S-b^\dagger b$4

For $S_z=S-b^\dagger b$5, the bistable regime requires $S_z=S-b^\dagger b$6; for $S_z=S-b^\dagger b$7, it requires $S_z=S-b^\dagger b$8. The critical powers are

$S_z=S-b^\dagger b$9

for direct magnon pumping and

$$K\,b^\dagger b\,b^\dagger b, \qquad K=\frac{\mu_{0}K_{\rm an}\gamma^{2}}{M^{2}V_{\rm m}},$$0

for cavity pumping. In transmission, the effect appears through the self-energy

$$K\,b^\dagger b\,b^\dagger b, \qquad K=\frac{\mu_{0}K_{\rm an}\gamma^{2}}{M^{2}V_{\rm m}},$$1

so that the measured coefficient

$$K\,b^\dagger b\,b^\dagger b, \qquad K=\frac{\mu_{0}K_{\rm an}\gamma^{2}}{M^{2}V_{\rm m}},$$2

acquires sweep-direction-dependent hysteresis and discontinuous jumps between stable branches (Zhang et al., 2019).

Within this mean-field framework, the MKE is the basic nonlinear element that converts an otherwise linear cavity–magnon anticrossing into a foldover and hysteretic response.

4. Squeezed-magnon interfaces and long-range spin coupling

The MKE has also been repurposed as an interface-engineering resource rather than only as a frequency-pulling mechanism. In a proposal using a YIG nanosphere coupled to distant NV-center spins, the magnon sector is modeled by

$$K\,b^\dagger b\,b^\dagger b, \qquad K=\frac{\mu_{0}K_{\rm an}\gamma^{2}}{M^{2}V_{\rm m}},$$3

A microwave drive $$K\,b^\dagger b\,b^\dagger b, \qquad K=\frac{\mu_{0}K_{\rm an}\gamma^{2}}{M^{2}V_{\rm m}},$$4 is applied, and under strong driving the magnon operator is displaced as $$K\,b^\dagger b\,b^\dagger b, \qquad K=\frac{\mu_{0}K_{\rm an}\gamma^{2}}{M^{2}V_{\rm m}},$$5. The linearized Hamiltonian then acquires a two-magnon parametric term

$$K\,b^\dagger b\,b^\dagger b, \qquad K=\frac{\mu_{0}K_{\rm an}\gamma^{2}}{M^{2}V_{\rm m}},$$6

which is diagonalized by the Bogoliubov transformation

$$K\,b^\dagger b\,b^\dagger b, \qquad K=\frac{\mu_{0}K_{\rm an}\gamma^{2}}{M^{2}V_{\rm m}},$$7

The spin–magnon coupling is thereby enhanced to

$$K\,b^\dagger b\,b^\dagger b, \qquad K=\frac{\mu_{0}K_{\rm an}\gamma^{2}}{M^{2}V_{\rm m}},$$8

so that the virtual excitation of squeezed magnons in the dispersive regime produces an effective long-range spin–spin coupling

$$K\,b^\dagger b\,b^\dagger b, \qquad K=\frac{\mu_{0}K_{\rm an}\gamma^{2}}{M^{2}V_{\rm m}},$$9

Representative parameters in that analysis give $\langle b^\dagger b\rangle$0 MHz, $\langle b^\dagger b\rangle$1, and $\langle b^\dagger b\rangle$2 kHz, much larger than the quoted $\langle b^\dagger b\rangle$3 kHz spin decoherence rate (Xiong et al., 2021).

In this regime, the MKE serves as a tunable nonlinear amplifier of the spin–magnon interface. The proposed consequences include micrometer-scale spin–magnon separation, remote quantum-state transfer, and a nonlocal two-qubit iSWAP gate with high fidelity (Xiong et al., 2021).

5. Multimode regimes, bound states, and thin-film implementations

The single-Kittel-mode picture ceases to be adequate once Kerr enhancement becomes sufficiently strong. A later study considered a single spin defect near a YIG sphere coupled to multimode magnons with Kerr nonlinearity and showed that the enhanced coupling invalidates the widely used single-Kittel-mode approximation. After strong-drive linearization and a multimode Bogoliubov transformation, the effective coupling to each mode becomes exponentially enhanced by the squeezing parameter, while the entire magnon spectrum is Kerr-renormalized. The analysis explicitly retained the Kittel, quadrupolar, and octupolar resonances, with bare frequencies

$\langle b^\dagger b\rangle$4

and showed that the exact spin dynamics are non-Markovian. Depending on parameters, the long-time behavior crosses over from complete decay to population trapping and then to persistent Rabi-like oscillation because one or two spin–magnon bound states form outside the magnon continuum. In the numerical examples, reducing the spin–YIG distance from 13 nm to 9 nm to 4 nm changed the asymptotics from decay to zero, to finite trapping, to lossless oscillation (Ji et al., 2023).

The experimental platform has also expanded beyond spheres. A 2026 experiment demonstrated the MKE in a 200 nm-thick YIG film strongly coupled to a three-dimensional loop-gap microwave resonator at room temperature. In that system, strong shape anisotropy significantly enhances the MKE compared to a sphere of equivalent volume, and the magnitude and sign of the Kerr shift are continuously tunable by rotating the static magnetization. The extracted Kerr coefficients were

$\langle b^\dagger b\rangle$5

$\langle b^\dagger b\rangle$6

and, near cancellation at $\langle b^\dagger b\rangle$7,

$\langle b^\dagger b\rangle$8

The paper states that the thin-film Kerr coefficients are two orders of magnitude larger than in the smallest YIG spheres. In the shape-anisotropy-dominated out-of-plane limit,

$\langle b^\dagger b\rangle$9

so the Kerr coefficient depends only on magnetic volume (Petrosyan et al., 29 Jan 2026).

Taken together, these results show that MKE is not confined to a weakly nonlinear, single-mode YIG-sphere setting. It persists in multimode spectral densities and can be enhanced by thin-film geometry while preserving strong magnon–photon coupling.

6. Nonreciprocity, optomagnomechanics, and superradiant transitions

In hybrid cavity-magnon optomechanics, the MKE has been used as a source of nonreciprocal quantum correlations. In these models, the magnon mode carries a Kerr term $K\propto 1/V_m$0, and after linearization the dynamics contain both a Kerr-induced frequency shift and a pair-magnon term. The mean-field shift is written as

$K\propto 1/V_m$1

or, in a related formulation,

$K\propto 1/V_m$2

and the fluctuation equation contains a term proportional to $K\propto 1/V_m$3, which acts as a parametric two-magnon process. The sign of $K\propto 1/V_m$4 is controlled by the magnetic-field direction: $K\propto 1/V_m$5 for $K\propto 1/V_m$6 and $K\propto 1/V_m$7 for $K\propto 1/V_m$8. This sign reversal shifts the optimal detuning conditions in opposite directions and enables nonreciprocal enhancement of magnon–phonon, microwave–optical, and optical–magnon entanglement. Both the 2023 and 2024 studies define bidirectional contrast ratios that range from 0 to 1, with $K\propto 1/V_m$9 corresponding to ideal nonreciprocity, and both report that the nonreciprocity can be switched on and off by tuning detuning or Kerr-related parameters; the 2024 COMM study further states that the entanglement can become more robust against thermal noise (Chen et al., 2023, Liu et al., 2024).

A related extension concerns nonequilibrium phase transitions. A 2025 proposal for a cavity magnonic system with parametric cavity drive showed that the same sign change of $K$0 produces a nonreciprocal superradiant quantum phase transition. The effective Hamiltonian contains

$K$1

with $K$2 for bias along $K$3 and $K$4 for $K$5. The steady-state phases include a normal phase with $K$6, a superradiant phase with $K$7, and a bistable phase where both solutions coexist. Because the stable nonzero branch differs for $K$8 and $K$9, the critical thresholds for the transition are direction dependent. The proposal quantifies this asymmetry through a bidirectional contrast ratio $[110]$0, where $[110]$1 denotes reciprocal behavior and $[110]$2 ideal nonreciprocity (Zhang et al., 30 Sep 2025).

These developments recast the MKE as more than a local spectral anharmonicity. In current hybrid models it functions as a sign-tunable nonlinear resource for nonreciprocal dynamics, Gaussian entanglement engineering, and quantum critical behavior.

7. Relation to magneto-optical Kerr nomenclature

The term “Kerr effect” is used in at least two distinct magnetic literatures, and MKE refers specifically to the magnonic one. In the magnonic usage summarized above, “Kerr” denotes a quartic self-interaction of a magnon mode and the associated population-dependent resonance shift. By contrast, in magneto-optical Kerr effect (MOKE) studies, “Kerr” denotes the rotation and ellipticity of reflected light and is described in terms of optical conductivity or dielectric tensors. Examples include enhanced MOKE at Fe/insulator interfaces through a large $[110]$3 ratio (Gu et al., 2017), large MOKE in noncollinear antiferromagnets Mn$[110]$4 caused by spin–orbit-coupled lifting of band double-degeneracy (Feng et al., 2015), and a universal surface Kerr response in A-type antiferromagnets such as MnBi$[110]$5Te$[110]$6 (Sunko et al., 22 Apr 2025).

The distinction is not merely terminological. A 2026 theory of a Kerr-like effect induced by quantum-metric nematicity in two-dimensional nonmagnetic systems explicitly contrasts its Kerr-like optical rotation with the MKE, stating that MKE is associated with magnon dynamics in magnetic systems and is fundamentally tied to spin order, magnetization, or magnetic excitations, whereas the proposed nonmagnetic Kerr-like effect requires neither magnetic order nor spin–orbit coupling (Liang et al., 23 Feb 2026).

Accordingly, “Magnon Kerr Effect” denotes a nonlinear many-body property of magnon modes, not a magneto-optical polarization-rotation experiment. The two subjects intersect in the broad theme of Kerr physics in magnetism, but they are governed by different observables, different Hamiltonians, and different symmetry structures.