Hybrid Cavity Magnomechanical System

• Hybrid cavity magnomechanical system is a tripartite bosonic platform where cavity photons, YIG-based magnons, and mechanical phonons interact via magnetic-dipole and magnetostrictive couplings.
• It employs strong drives to linearize weak magnetostrictive interactions into effective beam-splitter or parametric channels, enabling sideband physics and efficient signal transduction.
• Experimental implementations span microwave, optical, and SAW architectures, achieving phenomena like induced transparency, phonon lasing, nonreciprocity, and robust quantum entanglement.

A hybrid cavity magnomechanical system is a tripartite bosonic platform in which a cavity photon mode, a magnon mode in a magnetic medium—most commonly the Kittel mode of a yttrium iron garnet (YIG) resonator—and a mechanical phonon mode are coupled through magnetic-dipole and magnetostrictive interactions. In its canonical form, the cavity couples directly to magnons, magnons couple dispersively to phonons, and a strong cavity or magnon drive linearizes the intrinsically weak magnetostrictive interaction into beam-splitter or parametric channels, enabling cavity–magnon polaritons, sideband physics, and photon–magnon–phonon transduction (Zuo et al., 2023). The field now includes 3D microwave cavities, planar surface-acoustic-wave cavities, optical whispering-gallery platforms, cryogenic devices, and non-Hermitian or PT-symmetric extensions, all built around the same hybridization logic (Zhang et al., 2015).

1. Constituent modes and physical architecture

The standard hybrid cavity magnomechanical system contains three modes: a cavity photon mode, a collective magnon mode, and a mechanical phonon mode. In the widely used microwave implementation, a single-crystal YIG sphere is placed near the magnetic-field antinode of a microwave cavity, and a static magnetic field tunes the magnon resonance through the Kittel relation. The cavity photon and magnon are coupled by magnetic dipole interaction, while the magnon and phonon are coupled by magnetostriction. In the stationary cavity-magnomechanical model used to study irreversibility, the photon mode is denoted $a$, the magnon mode $m$, the phonon mode $b$, and there is no direct cavity–phonon coupling because the YIG sphere is much smaller than the microwave wavelength (Edet et al., 2024).

This architecture is not restricted to a single geometry. The literature includes 3D copper cavities with sub-millimeter YIG spheres, planar surface-acoustic-wave cavities embedding ferromagnetic films, and optical whispering-gallery-mode resonators interfaced with magnetic media. In the planar SAW realization, a nickel film is placed inside a high-$Q$ acoustic Fabry–Perot cavity on LiNbO$_3$, and the cavity confinement enhances the magnetoelastic interaction enough to yield a magnomechanical cooperativity $C = 1.2 \pm 0.1$ with an extracted coupling $g/2\pi = 9.9 \pm 0.2$ MHz (Hatanaka et al., 2021). In an optical hybrid realization, a BTS glass microsphere is deposited on a YIG film and read out through optical whispering-gallery modes, with the magnetostrictive response converted into optical phase modulation (Colombano et al., 2019).

The canonical YIG-sphere microwave platform remains the reference model because it simultaneously offers strong cavity–magnon coupling, tunable magnon frequency, and long-lived mechanical modes. The 2023 review on cavity magnomechanics defines this setting as a coherent interaction among microwave cavity photons $a$, magnons $m$, and mechanical phonons $b$, and places it at the intersection of cavity QED, magnonics, quantum optics, and quantum information (Zuo et al., 2023).

2. Canonical Hamiltonians and linearized dynamics

The minimal Hamiltonian of cavity magnomechanics combines three bare energies with two interaction channels. In the notation common to the review literature,

$m$0

$m$1

where $m$2 is the cavity–magnon beam-splitter coupling and $m$3 is the single-magnon dispersive magnomechanical coupling (Zuo et al., 2023). The original cavity-magnomechanics experiment used the same structure, with a 3D TE$m$4 cavity, a YIG sphere, and a spheroidal phonon mode, and introduced hybridized photon–magnon normal modes

$m$5

with mixing angle $m$6 (Zhang et al., 2015).

A strong coherent drive is ordinarily required because the bare magnetostrictive coupling is small. Linearization around the driven steady state converts the radiation-pressure-like interaction into resonant bilinear forms. On the red sideband, the effective interaction is

$m$7

whereas on the blue sideband it becomes

$m$8

The enhanced rate is set by the coherent magnon amplitude, $m$9, or equivalently $b$0 in other notations (Zuo et al., 2023). These linearized forms underpin transparency, backaction, phonon lasing, entanglement, and reservoir engineering.

The corresponding open-system description is usually written through linearized quantum Langevin equations or an effective non-Hermitian dynamical matrix. In the steady-state Gaussian treatment, the covariance matrix $b$1 is obtained from a Lyapunov equation, $b$2, with stability determined by the requirement that all eigenvalues of the drift matrix $b$3 have negative real parts (Edet et al., 2024). Notation is not uniform across the literature: in the quantum-battery model, the cavity is still $b$4, but the magnon is denoted $b$5 and the mechanical mode $b$6 (Singh et al., 23 Nov 2025). The underlying tripartite structure is, however, the same.

3. Resonances, hybridization regimes, and experimentally observed phenomena

The best-known operating regime is triply resonant cavity magnomechanics, in which the photon–magnon polariton splitting is tuned to the mechanical frequency. In the original microwave experiment, the condition

$b$7

enabled simultaneous resonant enhancement of drive and mechanically scattered sidebands, leading to magnomechanically induced transparency and absorption, parametric amplification, and phonon lasing (Zhang et al., 2015). The same paper reported a measured single-magnon–single-phonon coupling $b$8 mHz for a $b$9 YIG sphere, with a theoretical upper value $Q$0 mHz, and showed that triply resonant operation reached a cooperativity $Q$1 at $Q$2 dBm, compared with about $Q$3 dBm off resonance (Zhang et al., 2015).

The review literature organizes the observed dynamics into several standard phenomena. Zhang et al. observed magnomechanically induced transparency and absorption in a 3D copper cavity with a sub-millimeter YIG sphere, while Potts et al. reported dynamical backaction, cooling, amplification, and the magnonic spring effect under triple resonance (Zuo et al., 2023). In linear response, the transparency linewidth scales as

$Q$4

and the backaction-induced mechanical damping and frequency shift follow

$Q$5

These relations make explicit that cavity magnomechanics inherits the dynamical-backaction logic of cavity optomechanics while replacing the optical intermediate with a magnonic one (Zuo et al., 2023).

Cryogenic operation remained experimentally absent until the 2025 report of magnomechanics at low temperature. That experiment used a 3D polished OFHC copper cavity at $Q$6 GHz and a $Q$7-diameter YIG sphere mounted on a copper needle. It measured thermomechanical motion and magnon-linewidth thermometry down to $Q$8 K, with $Q$9 MHz and a mechanical linewidth $_3$0 kHz in the copper-needle geometry (Huang et al., 27 Mar 2025). The same work also quantified drive-induced heating of the YIG sphere, showing, for example, that at $_3$1 dBm a stage temperature of $_3$2 K corresponded to a fitted magnon temperature of about $_3$3 K (Huang et al., 27 Mar 2025).

Planar acoustomagnonic implementations establish that cavity magnomechanical physics is not limited to 3D microwave resonators. In the LiNbO$_3$4 SAW cavity, the transmission spectrum exhibited Fabry–Perot resonances between $_3$5 and $_3$6 GHz with spacing $_3$7 MHz and quality factors up to about $_3$8, while the cavity-enhanced magnetoelastic interaction produced acoustic absorption above $_3$9 and measurable backaction on both resonance frequency and linewidth (Hatanaka et al., 2021). This broadens the field from bulk YIG spheres toward integrated spin-acoustic devices.

4. Non-Hermitian, PT-symmetric, and nonreciprocal extensions

A major recent direction introduces explicit non-Hermiticity into cavity magnomechanics. In the PT-symmetric formulation of a YIG sphere in a microwave Fabry–Pérot cavity, a traveling field directly modifies the magnon–photon coupling by a complex term $C = 1.2 \pm 0.1$0, producing the effective matrix

$C = 1.2 \pm 0.1$1

In that model, PT symmetry and exceptional points occur only at $C = 1.2 \pm 0.1$2 and $C = 1.2 \pm 0.1$3, but stability is found only at $C = 1.2 \pm 0.1$4. The third-order exceptional point lies on the axis

$C = 1.2 \pm 0.1$5

and the eigenvalue map contains a “uni-protected” PT-unbroken axis separating two PT-broken regions described as “bi-broken” (Chengyong et al., 2024). Protected PT symmetry exists only on the exceptional-point axis, and off-axis the spectrum acquires complex eigenvalues (Chengyong et al., 2024).

Nonreciprocity can also arise from nonlinear or gain-engineered hybridization. In the passive–active two-cavity magnomechanical system, one cavity is passive and the other effectively active, giving a PT-symmetric-like dimer with $C = 1.2 \pm 0.1$6. Within the stable regime, this architecture enhances distant entanglement and supports directional EPR steering, while tuning the cavity–cavity hopping $C = 1.2 \pm 0.1$7, the magnon–photon coupling, or the detunings transfers correlations between near and distant mode pairs (Chen et al., 2020).

A distinct nonreciprocal mechanism is magnon Kerr nonlinearity. In the 2026 two-YIG-sphere proposal, direction-dependent intracavity populations generate direction-dependent Kerr shifts $C = 1.2 \pm 0.1$8, which reshape transparency windows, Fano resonances, and group delay. The transmission coefficients

$C = 1.2 \pm 0.1$9

$g/2\pi = 9.9 \pm 0.2$0

then differ in both amplitude and phase, yielding directional transparency and slow/fast-light switching (Amghar et al., 11 Jun 2026). These formulations indicate that non-Hermitian cavity magnomechanics is not a single protocol but a family of gain/loss, active-passive, and nonlinear asymmetry schemes.

5. Quantum correlations, entropy production, and coherence control

The tripartite structure naturally supports Gaussian entanglement, steering, and reservoir-engineered state preparation. In a dual-cavity scheme driven by a two-mode squeezed vacuum, red-detuned magnon drives activate state-swap interactions that transfer correlations from the cavity fields to the magnons and then to the phonons of two distant YIG spheres. The predicted steady-state phonon–phonon logarithmic negativity is about $g/2\pi = 9.9 \pm 0.2$1 at $g/2\pi = 9.9 \pm 0.2$2 mK for $g/2\pi = 9.9 \pm 0.2$3, and the entanglement survives up to approximately $g/2\pi = 9.9 \pm 0.2$4 mK (Li et al., 2020).

A complementary protocol uses the magnon as an engineered cold reservoir. For $g/2\pi = 9.9 \pm 0.2$5 and $g/2\pi = 9.9 \pm 0.2$6, the photon–magnon beam-splitter and magnon–phonon two-mode-squeezing interactions can be recast in terms of a Bogoliubov mode,

$g/2\pi = 9.9 \pm 0.2$7

which is cooled through the dissipative magnon channel. This produces strong steady photon–phonon entanglement that the paper states is robust against temperature and can remain sizable up to about $g/2\pi = 9.9 \pm 0.2$8 K with suitable parameters (Liu et al., 2022).

The passive–active two-cavity architecture extends this to directional EPR steering and correlation transfer. In the active regime $g/2\pi = 9.9 \pm 0.2$9, the distant photon–magnon entanglement $a$0 reaches about $a$1, more than twice the corresponding near entanglement in one reported operating point, and remains nonzero up to about $a$2 K (Chen et al., 2020). The same model exhibits one-way steering, such as $a$3, and near-complete transfer of steering from one pair to another by tuning $a$4 or the detunings (Chen et al., 2020).

Thermodynamic characterizations add a different perspective. In the stationary Gaussian treatment of a driven cavity magnomechanical system, the steady-state entropy production rate is

$a$5

with $a$6 at steady state. The work shows that the entropy flow between cavity photons and phonons is governed primarily by the magnon–photon coupling $a$7 and the cavity dissipation $a$8, and that for small $a$9 the magnon–phonon mutual information approximately follows $m$0 (Edet et al., 2024). This directly connects irreversibility with hybrid-mode correlations.

Phase-sensitive coherence control has also been proposed. In a squeezed-magnon-driven cavity magnomechanical system, the parametric term introduces phase-dependent renormalizations

$m$1

which shift the effective magnon detuning and linewidth. The Gaussian coherence of the cavity, magnon, and mechanical subsystems is minimized at $m$2 and maximized at $m$3, with the mechanical coherence showing the strongest dependence on the squeezing-enhanced $m$4 (Hidki et al., 9 Nov 2025). This provides a route to nonreciprocal coherence transfer rather than only nonreciprocal transmission.

6. Platforms, representative parameters, and application directions

The experimental and theoretical landscape spans several hardware classes. The following examples summarize representative implementations and design studies already reported in the literature.

Platform	Representative features	Source
3D microwave cavity + YIG sphere	$m$5 GHz, $m$6 MHz, measured $m$7 mHz, $m$8 Hz	(Zhang et al., 2015)
Cryogenic OFHC copper cavity + YIG sphere	$m$9 GHz, $b$0 YIG sphere, $b$1 MHz, $b$2 kHz, observed down to $b$3 K	(Huang et al., 27 Mar 2025)
Planar SAW cavity + Ni film	$b$4, $b$5 MHz, $b$6	(Hatanaka et al., 2021)
Optical WGM microsphere on YIG film	$b$7, peak sensitivity $b$8 pT Hz$b$9 near $m$00 MHz	(Colombano et al., 2019)
Phase-field-designed YIG/SiN cavity electromagnonic device	Triple resonance near $m$01 GHz with a chiral $m$02 TA phonon and simulated $m$03 MHz	(Zhuang et al., 2024)

Across these realizations, several constraints recur. Bare magnomechanical coupling is usually small and must be enhanced by strong driving (Zhang et al., 2015). Triple resonance and sideband resolution require careful matching of cavity, magnon, and mechanical frequencies (Zuo et al., 2023). Mechanical damping and clamping loss remain important in cryogenic devices, where improved thermal anchoring can increase $m$04 even while enabling lower-temperature operation (Huang et al., 27 Mar 2025). Gain-assisted and non-Hermitian variants must satisfy explicit stability conditions rather than only spectral matching (Chengyong et al., 2024).

Application proposals now extend beyond spectroscopy and transduction. A quantum-battery model couples two identical two-level atoms to a cavity–magnomechanical charger and finds that strong, resonant light–matter interactions enhance stored energy and ergotropy, while excessively large couplings can trap energy and reduce performance (Singh et al., 23 Nov 2025). A gyroscope proposal based on a YIG whispering-gallery resonator uses optomagnonic two-mode squeezing for sub-shot-noise rotation sensing; the authors explicitly note that the working Hamiltonian does not include a mechanical mode even though the broader platform supports magnomechanical couplings (Yang et al., 13 May 2025). This indicates that the hybrid cavity magnomechanical system now functions both as a concrete tripartite device class and as a design template for broader YIG-based hybrid quantum architectures.

Taken together, the literature defines hybrid cavity magnomechanics as a controllable many-body interface among photons, magnons, and phonons. Its core technical themes are strong magnon–photon hybridization, drive-enhanced magnetostriction, sideband-selective linearization, and increasingly, engineered dissipation or non-Hermiticity. The resulting platform supports transparency, backaction, phonon lasing, entropy-flow control, Gaussian entanglement, directional steering, nonreciprocal coherence, and device concepts for sensing, storage, and transduction (Zuo et al., 2023).