Magnon Kerr Nonlinearity

• Magnon Kerr nonlinearity is the quartic self-interaction in magnon modes that produces a frequency shift proportional to the magnon population, primarily emerging from magnetocrystalline anisotropy in YIG.
• It enables tunable hybrid interactions in cavity magnonics, supporting phenomena like photon blockade, squeezing, and parametric amplification through both intrinsic anisotropy and dispersive qubit coupling.
• Experimental studies report Kerr shifts from nHz to kHz, with sensitivity to sample geometry, magnetization orientation, and drive power, offering a versatile platform for exploring nonlinear quantum dynamics.

Magnon Kerr nonlinearity is the quartic self-interaction of a magnon mode, usually the Kittel mode of a yttrium-iron-garnet (YIG) sample, appearing in effective Hamiltonians as $K(m^\dagger m)^2$ or $-K_m\,m^\dagger m^\dagger m m$. In cavity magnonics and magnomechanics it produces a magnon-population-dependent frequency shift, modifies hybrid magnon–photon–phonon spectra, and, under strong driving, generates effective parametric and squeezing terms. The nonlinearity is ordinarily traced to magnetocrystalline anisotropy in YIG, but later work also realized an effective magnon self-Kerr by dispersively coupling magnons to a superconducting qubit, bringing the Kerr scale into the quantum regime (Fan et al., 2024, Wang et al., 2016, Qi et al., 2024, Weng et al., 23 Feb 2026).

1. Microscopic origin and model forms

A representative cavity-magnon Hamiltonian in the rotating frame is

$$H_{\rm sys} = \sum_{j=1,2} \Big[ \Delta_m\,m_j^\dagger m_j + g_j\,(m_j^\dagger c+c^\dagger m_j) + K_j\,(m_j^\dagger m_j)^2 \Big] + \Delta_c\,c^\dagger c + \Omega\,(c+c^\dagger),$$

where $c$ is the cavity mode, $m_j$ are magnon modes, $g_j$ are photon–magnon couplings, and $K_j(m_j^\dagger m_j)^2$ is the magnon Kerr term. In the single-sphere case the same structure is commonly written as

$$H=\omega_c a^\dagger a+\omega_m m^\dagger m+K(m^\dagger m)^2+g(a^\dagger m+a m^\dagger)+\Omega_d(m^\dagger e^{-i\omega_d t}+m e^{i\omega_d t}),$$

so the quartic term directly shifts the magnon resonance by an amount proportional to magnon occupation (Fan et al., 2024, Wang et al., 2016).

The microscopic source in YIG is magnetocrystalline anisotropy. In the macrospin description one starts from an anisotropy contribution quadratic in $S_z$, applies the Holstein–Primakoff transformation $S_z=S-m^\dagger m$, and obtains a quartic bosonic term. Two equivalent expressions quoted in the literature are

$-K_m\,m^\dagger m^\dagger m m$0

and

$-K_m\,m^\dagger m^\dagger m m$1

which make explicit the inverse-volume scaling. In the thin-film formulation the anisotropy energy is written as $-K_m\,m^\dagger m^\dagger m m$2, and expansion to fourth order yields

$-K_m\,m^\dagger m^\dagger m m$3

so both shape anisotropy and magnetization orientation enter the effective Kerr coefficient (Fan et al., 2024, Wang et al., 2016, Petrosyan et al., 29 Jan 2026).

A distinct route to magnon Kerr nonlinearity uses dispersive coupling to a superconducting qubit. After Schrieffer–Wolff elimination, the effective magnon Hamiltonian becomes

$-K_m\,m^\dagger m^\dagger m m$4

with $-K_m\,m^\dagger m^\dagger m m$5 the effective qubit–magnon coupling and $-K_m\,m^\dagger m^\dagger m m$6 the dispersive detuning. Here the quartic term is not attributed to YIG anisotropy alone, but to virtual qubit processes that generate a sizable self-Kerr in the magnon mode (Weng et al., 23 Feb 2026).

2. Sign control, scaling, and reported magnitudes

Reported Kerr quantities depend strongly on sample geometry, orientation, and whether one quotes a bare single-magnon coefficient or a drive-induced nonlinear shift.

Platform	Reported Kerr quantity	Note
1 mm YIG sphere in a 3D cavity	$-K_m\,m^\dagger m^\dagger m m$7	Pump-induced shift observable when $-K_m\,m^\dagger m^\dagger m m$8 (Wang et al., 2016)
0.25 mm diameter YIG sphere	$-K_m\,m^\dagger m^\dagger m m$9, $$H_{\rm sys} = \sum_{j=1,2} \Big[ \Delta_m\,m_j^\dagger m_j + g_j\,(m_j^\dagger c+c^\dagger m_j) + K_j\,(m_j^\dagger m_j)^2 \Big] + \Delta_c\,c^\dagger c + \Omega\,(c+c^\dagger),$$0	$$H_{\rm sys} = \sum_{j=1,2} \Big[ \Delta_m\,m_j^\dagger m_j + g_j\,(m_j^\dagger c+c^\dagger m_j) + K_j\,(m_j^\dagger m_j)^2 \Big] + \Delta_c\,c^\dagger c + \Omega\,(c+c^\dagger),$$1, $$H_{\rm sys} = \sum_{j=1,2} \Big[ \Delta_m\,m_j^\dagger m_j + g_j\,(m_j^\dagger c+c^\dagger m_j) + K_j\,(m_j^\dagger m_j)^2 \Big] + \Delta_c\,c^\dagger c + \Omega\,(c+c^\dagger),$$2 (Fan et al., 2024)
0.28 mm YIG sphere in cavity magnomechanics	$$H_{\rm sys} = \sum_{j=1,2} \Big[ \Delta_m\,m_j^\dagger m_j + g_j\,(m_j^\dagger c+c^\dagger m_j) + K_j\,(m_j^\dagger m_j)^2 \Big] + \Delta_c\,c^\dagger c + \Omega\,(c+c^\dagger),$$3, $$H_{\rm sys} = \sum_{j=1,2} \Big[ \Delta_m\,m_j^\dagger m_j + g_j\,(m_j^\dagger c+c^\dagger m_j) + K_j\,(m_j^\dagger m_j)^2 \Big] + \Delta_c\,c^\dagger c + \Omega\,(c+c^\dagger),$$4	Self-Kerr and magnon–phonon cross-Kerr extracted from bistability data (Shen et al., 2022)
200 nm YIG film in a 3D loop-gap cavity	$$H_{\rm sys} = \sum_{j=1,2} \Big[ \Delta_m\,m_j^\dagger m_j + g_j\,(m_j^\dagger c+c^\dagger m_j) + K_j\,(m_j^\dagger m_j)^2 \Big] + \Delta_c\,c^\dagger c + \Omega\,(c+c^\dagger),$$5, $$H_{\rm sys} = \sum_{j=1,2} \Big[ \Delta_m\,m_j^\dagger m_j + g_j\,(m_j^\dagger c+c^\dagger m_j) + K_j\,(m_j^\dagger m_j)^2 \Big] + \Delta_c\,c^\dagger c + \Omega\,(c+c^\dagger),$$6	$$H_{\rm sys} = \sum_{j=1,2} \Big[ \Delta_m\,m_j^\dagger m_j + g_j\,(m_j^\dagger c+c^\dagger m_j) + K_j\,(m_j^\dagger m_j)^2 \Big] + \Delta_c\,c^\dagger c + \Omega\,(c+c^\dagger),$$7 near $$H_{\rm sys} = \sum_{j=1,2} \Big[ \Delta_m\,m_j^\dagger m_j + g_j\,(m_j^\dagger c+c^\dagger m_j) + K_j\,(m_j^\dagger m_j)^2 \Big] + \Delta_c\,c^\dagger c + \Omega\,(c+c^\dagger),$$8 (Petrosyan et al., 29 Jan 2026)

The sign of the Kerr coefficient is an experimentally tunable quantity. For YIG spheres, rotating the bias field relative to the crystal axes yields $$H_{\rm sys} = \sum_{j=1,2} \Big[ \Delta_m\,m_j^\dagger m_j + g_j\,(m_j^\dagger c+c^\dagger m_j) + K_j\,(m_j^\dagger m_j)^2 \Big] + \Delta_c\,c^\dagger c + \Omega\,(c+c^\dagger),$$9 and $c$0, while the magnitude scales as $c$1. In the thin-film system, $c$2 changes sign continuously, crosses near $c$3, and reaches extrema for out-of-plane and in-plane magnetization. This suggests that geometry and field orientation are coequal control parameters for Kerr engineering rather than secondary perturbations (Fan et al., 2024, Petrosyan et al., 29 Jan 2026).

The same point appears in comparative anisotropy estimates. In a bulk YIG sphere $c$4, giving $c$5, whereas in a thin film the demagnetizing field yields $c$6, so $c$7. Together with the $c$8 scaling, this provides a direct route to larger $c$9 without abandoning strong magnon–photon coupling (Petrosyan et al., 29 Jan 2026).

3. Frequency shifts, nonlinear response, and experimental extraction

In the dispersive cavity-magnon experiment of Wang et al., the observable effect of the Kerr term was a pump-induced magnon-frequency shift and a smaller cavity-frequency shift. In the dispersive limit $m_j$0, the steady-state shift obeys

$m_j$1

equivalently

$m_j$2

At very small $m_j$3, $m_j$4 is linear in $m_j$5; at high power, $m_j$6. The cavity shift is

$m_j$7

For $m_j$8 the cavity central frequency shifted by $m_j$9, while the magnon shift was $g_j$0 (Wang et al., 2016).

In cavity magnomechanics, the Kerr coefficient was extracted by fitting the reflection spectrum and tracking a power-dependent magnon detuning

$g_j$1

The mechanical-frequency shift was then decomposed into the conventional radiation-pressure spring shift and a magnon–phonon cross-Kerr term $g_j$2. Pump sweeps upward and downward produced hysteresis in both $g_j$3 and $g_j$4, identifying a Kerr-modified bistable regime rather than a purely linear magnomechanical response (Shen et al., 2022).

In the thin-film experiment, the coupled equations

$g_j$5

were fitted to power-dependent $g_j$6. At $g_j$7, $g_j$8 and the induced magnon shift reached several MHz. In the qubit-assisted quantum-regime experiment, the dispersive coefficients were determined by direct spectroscopy and by fitting Kerr-induced phase-space shearing: $g_j$9, $K_j(m_j^\dagger m_j)^2$0, and therefore $K_j(m_j^\dagger m_j)^2$1 (Petrosyan et al., 29 Jan 2026, Weng et al., 23 Feb 2026).

4. Interference physics, blockade, and nonreciprocity

One of the most direct uses of magnon Kerr nonlinearity is unconventional photon blockade. In the weak-drive truncation,

$K_j(m_j^\dagger m_j)^2$2

and photon antibunching follows from destructive interference between a direct two-photon path, $K_j(m_j^\dagger m_j)^2$3, and a magnon-assisted path, $K_j(m_j^\dagger m_j)^2$4. Because the intermediate states shift by $K_j(m_j^\dagger m_j)^2$5 for $K_j(m_j^\dagger m_j)^2$6 and by $K_j(m_j^\dagger m_j)^2$7 for $K_j(m_j^\dagger m_j)^2$8, the cancellation condition depends on the Kerr sign: $K_j(m_j^\dagger m_j)^2$9 Solving this gives

$$H=\omega_c a^\dagger a+\omega_m m^\dagger m+K(m^\dagger m)^2+g(a^\dagger m+a m^\dagger)+\Omega_d(m^\dagger e^{-i\omega_d t}+m e^{i\omega_d t}),$$0

with the sign set by $$H=\omega_c a^\dagger a+\omega_m m^\dagger m+K(m^\dagger m)^2+g(a^\dagger m+a m^\dagger)+\Omega_d(m^\dagger e^{-i\omega_d t}+m e^{i\omega_d t}),$$1. In the one-sphere proposal this produces nonreciprocal unconventional photon blockade; in the two-sphere version, symmetric $$H=\omega_c a^\dagger a+\omega_m m^\dagger m+K(m^\dagger m)^2+g(a^\dagger m+a m^\dagger)+\Omega_d(m^\dagger e^{-i\omega_d t}+m e^{i\omega_d t}),$$2 and $$H=\omega_c a^\dagger a+\omega_m m^\dagger m+K(m^\dagger m)^2+g(a^\dagger m+a m^\dagger)+\Omega_d(m^\dagger e^{-i\omega_d t}+m e^{i\omega_d t}),$$3 yield reciprocal blockade, whereas asymmetric couplings or asymmetric Kerr strengths produce nonreciprocity (Fan et al., 2024).

The same quartic nonlinearity also drives high-order frequency conversion. In bichromatically driven cavity–magnon systems the output field acquires sidebands at $$H=\omega_c a^\dagger a+\omega_m m^\dagger m+K(m^\dagger m)^2+g(a^\dagger m+a m^\dagger)+\Omega_d(m^\dagger e^{-i\omega_d t}+m e^{i\omega_d t}),$$4, with exact spacing $$H=\omega_c a^\dagger a+\omega_m m^\dagger m+K(m^\dagger m)^2+g(a^\dagger m+a m^\dagger)+\Omega_d(m^\dagger e^{-i\omega_d t}+m e^{i\omega_d t}),$$5. The effective nonlinear shift is $$H=\omega_c a^\dagger a+\omega_m m^\dagger m+K(m^\dagger m)^2+g(a^\dagger m+a m^\dagger)+\Omega_d(m^\dagger e^{-i\omega_d t}+m e^{i\omega_d t}),$$6, and the sideband amplitudes scale approximately as $$H=\omega_c a^\dagger a+\omega_m m^\dagger m+K(m^\dagger m)^2+g(a^\dagger m+a m^\dagger)+\Omega_d(m^\dagger e^{-i\omega_d t}+m e^{i\omega_d t}),$$7. Raising the microwave drive to an optimal $$H=\omega_c a^\dagger a+\omega_m m^\dagger m+K(m^\dagger m)^2+g(a^\dagger m+a m^\dagger)+\Omega_d(m^\dagger e^{-i\omega_d t}+m e^{i\omega_d t}),$$8 populates many sidebands, while changing $$H=\omega_c a^\dagger a+\omega_m m^\dagger m+K(m^\dagger m)^2+g(a^\dagger m+a m^\dagger)+\Omega_d(m^\dagger e^{-i\omega_d t}+m e^{i\omega_d t}),$$9 directly tunes the comb spacing. In the nonreciprocal two-cavity configuration, the effective Kerr scale is $S_z$0, and the reported isolation ratios near optimal detuning $S_z$1 are $S_z$2 for the first-order sideband, $S_z$3 for the second-order sideband, and $S_z$4 for the third-order sideband (Liu et al., 2018, Wang et al., 2021).

In hybrid magnomechanics, Kerr-induced frequency shifts reshape transparency and dispersion. The effective susceptibility

$S_z$5

moves the transparency windows differently for opposite propagation directions, producing asymmetric Fano line shapes, multiple transparency dips, and large directional group-delay contrasts. Under the stated strong-coupling and mechanical-$S_z$6 conditions, forward and backward group delays can differ by $S_z$7 (Amghar et al., 11 Jun 2026).

Kerr nonlinearity can also be leveraged through non-Hermitian engineering. In the $S_z$8-symmetric cavity-magnonic proposal, balancing gain and loss near the exceptional point strongly amplifies the effective nonlinearities, so that perfect magnon blockade is obtained even for $S_z$9, and photon blockade appears simultaneously although there is no bare photonic Kerr term. In a polaritonic formulation, the Kerr-shifted spectrum

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

supports $S_z=S-m^\dagger m$1–$S_z=S-m^\dagger m$2 polariton blockade with $S_z=S-m^\dagger m$3 for $S_z=S-m^\dagger m$4, $S_z=S-m^\dagger m$5, $S_z=S-m^\dagger m$6–$S_z=S-m^\dagger m$7, and $S_z=S-m^\dagger m$8–$S_z=S-m^\dagger m$9 (Ebrahimi et al., 2022, Yang et al., 17 Mar 2026).

5. Kerr dressing, squeezing, entanglement, and engineered couplings

Under strong driving, the magnon Kerr term does more than shift a resonance: it generates effective parametric interactions. In the photon–phonon squeezing protocol, linearization around a large magnon amplitude and the squeezing transformation

$-K_m\,m^\dagger m^\dagger m m$00

convert the Kerr term into a dressed detuning $-K_m\,m^\dagger m^\dagger m m$01 and renormalized couplings. After eliminating magnon fluctuations, the effective photon–phonon Hamiltonian becomes

$-K_m\,m^\dagger m^\dagger m m$02

with

$-K_m\,m^\dagger m^\dagger m m$03

Numerically, moderate $-K_m\,m^\dagger m^\dagger m m$04–$-K_m\,m^\dagger m^\dagger m m$05 maximizes $-K_m\,m^\dagger m^\dagger m m$06, and realistic parameters allow $-K_m\,m^\dagger m^\dagger m m$07 of two-mode squeezing for $-K_m\,m^\dagger m^\dagger m m$08 and $-K_m\,m^\dagger m^\dagger m m$09 (Qi et al., 2024).

A closely related construction appears in magnomechanical cooling. Two-tone driving generates a squeezing rate $-K_m\,m^\dagger m^\dagger m m$10, and the Bogoliubov mode

$-K_m\,m^\dagger m^\dagger m m$11

diagonalizes the magnon sector. The optical damping rates

$-K_m\,m^\dagger m^\dagger m m$12

then give

$-K_m\,m^\dagger m^\dagger m m$13

In the sideband-unresolved regime, the analytic optimum is

$-K_m\,m^\dagger m^\dagger m m$14

and the representative YIG-sphere parameters yield $-K_m\,m^\dagger m^\dagger m m$15 with $-K_m\,m^\dagger m^\dagger m m$16 (Xu et al., 8 Apr 2025).

Kerr nonlinearity also mediates steady-state entanglement. In the two-magnon cavity proposal, strong blue-detuned driving creates effective parametric couplings

$-K_m\,m^\dagger m^\dagger m m$17

and the resulting quadratic fluctuation Hamiltonian supports magnon–photon squeezing and cavity-mediated magnon–magnon entanglement. For the quoted realistic parameters, numerical solutions of the Lyapunov equation yield logarithmic negativities $-K_m\,m^\dagger m^\dagger m m$18–$-K_m\,m^\dagger m^\dagger m m$19 at $-K_m\,m^\dagger m^\dagger m m$20 (Zhang et al., 2019). In a single ferrimagnetic crystal containing a Kittel mode and a higher-order magnetostatic mode, self-Kerr and cross-Kerr coexist through

$-K_m\,m^\dagger m^\dagger m m$21

and strong drives raise the effective nonlinearities to $-K_m\,m^\dagger m^\dagger m m$22, $-K_m\,m^\dagger m^\dagger m m$23, and $-K_m\,m^\dagger m^\dagger m m$24, with entanglement surviving up to $-K_m\,m^\dagger m^\dagger m m$25–$-K_m\,m^\dagger m^\dagger m m$26 (Yang et al., 2022).

More elaborate hybridizations use Kerr magnons as squeezed intermediaries. In the NV-center proposal, the squeezed-basis tripartite coupling

$-K_m\,m^\dagger m^\dagger m m$27

gives, for $-K_m\,m^\dagger m^\dagger m m$28,

$-K_m\,m^\dagger m^\dagger m m$29

and perfect magnon blockade emerges when $-K_m\,m^\dagger m^\dagger m m$30 is sufficiently large, numerically for $-K_m\,m^\dagger m^\dagger m m$31 (Chen et al., 2 Sep 2025). In the driven spin–magnon problem, the Kerr-induced squeezing parameter $-K_m\,m^\dagger m^\dagger m m$32 enhances the coupling as $-K_m\,m^\dagger m^\dagger m m$33, invalidates the single-Kittel-mode approximation under strong pumping, and leads to bound-state-induced population trapping or persistent Rabi-like oscillation (Ji et al., 2023).

The quantum-regime experiment provides an explicit demonstration that a sufficiently large magnon Kerr can directly generate nonclassical magnon states. There, a 1-mm YIG sphere coupled dispersively to a superconducting qubit exhibited $-K_m\,m^\dagger m^\dagger m m$34, and Kerr evolution produced quadrature variances of $-K_m\,m^\dagger m^\dagger m m$35, corresponding to $-K_m\,m^\dagger m^\dagger m m$36 of squeezing, with mean magnon number less than one (Weng et al., 23 Feb 2026).

6. Bistability, soft modes, sensitivity enhancement, and conceptual boundaries

Beyond antibunching and squeezing, magnon Kerr nonlinearity reorganizes nonlinear dynamics at the semiclassical level. In cavity magnomechanics, mechanical bistability arises when magnetostriction, magnon self-Kerr, and magnon–phonon cross-Kerr act simultaneously; the experiment identified hysteresis in the magnon resonance, the mechanical resonance, and the mechanical linewidth under $-K_m\,m^\dagger m^\dagger m m$37–$-K_m\,m^\dagger m^\dagger m m$38 pumping (Shen et al., 2022). In easy-axis ferromagnets coupled to a microwave cavity, the Kerr term stabilizes the soft mode by opening a finite gap,

$-K_m\,m^\dagger m^\dagger m m$39

and, in the simplest zero-drive picture,

$-K_m\,m^\dagger m^\dagger m m$40

The same analysis predicts chaos and comb-like behavior near the mode crossing, with comb spacing

$-K_m\,m^\dagger m^\dagger m m$41

and reports agreement with $-K_m\,m^\dagger m^\dagger m m$42 sidebands (Chiba, 13 Feb 2026).

Sensitivity to very small Kerr shifts can be increased through exceptional-point physics. In the coherent-perfect-absorption proposal, imposing the CPA condition yields an effective non-Hermitian Hamiltonian that can host an EP3. Near that point, a tiny Kerr-induced shift $-K_m\,m^\dagger m^\dagger m m$43 produces cube-root eigenvalue splitting,

$-K_m\,m^\dagger m^\dagger m m$44

so the observable dip separation obeys

$-K_m\,m^\dagger m^\dagger m m$45

The paper’s stated implication is that a Kerr shift far below the linewidth can still be converted into a much larger spectral splitting in the cavity transmission (Zhang et al., 2022).

A recurrent boundary condition in this literature is the distinction between bare Kerr strength and effective nonlinear response. Bulk YIG can have an intrinsic single-magnon Kerr that is extremely small, yet the same platform can display large pump-induced shifts, strong blockade, or measurable squeezing once one exploits large occupations, shape anisotropy, $-K_m\,m^\dagger m^\dagger m m$46-symmetric enhancement, exceptional-point amplification, or qubit-mediated dispersive engineering (Wang et al., 2016, Petrosyan et al., 29 Jan 2026, Ebrahimi et al., 2022, Zhang et al., 2022, Weng et al., 23 Feb 2026). This suggests that “magnon Kerr nonlinearity” in current research denotes not only a microscopic quartic coefficient, but also a design principle for converting weak intrinsic anharmonicity into experimentally resolvable nonlinear magnon dynamics.