Gain-Driven Magnon-Polaritons

• Gain-driven magnon-polaritons are hybrid light–matter excitations where magnon and electromagnetic modes coherently mix under effective gain and loss compensation.
• They employ non-Hermitian dynamics with feedback-controlled loss and dissipative coupling, enabling auto-oscillation, coherent emission, and lasing.
• Experimental implementations range from microwave cavity magnonics to proposed terahertz, ultrastrong, and topological platforms, illustrating versatile applications.

Gain-driven magnon-polaritons (MPs) are hybrid light–matter excitations in which a magnon mode and an electromagnetic mode remain coherently mixed while the hybrid dynamics are shaped by effective gain, compensated loss, or both. In the passive limit, MPs are described by the familiar cavity-magnonics or polaritonic strong-coupling picture; in the active limit, negative effective damping, feedback-assisted loss compensation, or dissipative coupling can drive auto-oscillation, amplification, coherent emission, exceptional-point dynamics, or lasing (Yao et al., 2023, Wang et al., 20 Nov 2025, Suzuki et al., 11 Sep 2025). The subject now spans microwave cavity magnonics, dissipatively coupled cavity–magnon systems, active van der Pol cavities, ultrastrong-coupling platforms, and proposed terahertz antiferromagnetic and magnon–plasmon realizations (Wu et al., 30 Apr 2026, Kritzell et al., 2023, Yoshii et al., 8 Jul 2025).

1. Conceptual basis and relation to conventional magnon-polaritons

Magnon-polaritons originate from coherent coupling between a magnonic excitation and a photonic or plasmonic mode. In the standard cavity-magnonics formulation, the passive system is described by

$$\hat{H} = \hbar \omega_{\rm cav} \hat{a}^\dagger \hat{a} + \hbar \omega \hat{b}^\dagger \hat{b} + \hbar g \left(\hat{a}^\dagger \hat{b} + \hat{a} \hat{b}^\dagger\right),$$

with $g$ the magnon–photon coupling strength (Kritzell et al., 2023). Earlier work established an explicit connection between cavity magnonics and bulk magnon-polaritons through an effective permeability description, showing that the cavity system inherits a polariton gap whose magnitude depends on an effective filling factor $\eta$, and that the standard bulk magnon-polariton limit is recovered for $\eta=1$ (Hyde et al., 2017).

Gain-driven MPs differ from passive MPs not by stronger probing alone, but by a qualitative change in the dynamical equations. In a gain-embedded cavity magnonics platform, the cavity frequency becomes

$$\tilde{\omega}_c = \omega_c + i\bigl(G - \kappa_c - \gamma|a|^2\bigr),$$

so that linear gain $G$ competes with cavity loss $\kappa_c$ and a nonlinear saturation term $\gamma |a|^2$ (Yao et al., 2023). In that regime the hybrid is a self-oscillating, gain-clamped magnon–photon mode rather than a passive avoided crossing driven by an external tone. A closely related, but distinct, route is dissipative coupling: a cavity mode and a magnon mode can lase collectively even when neither constituent mode is individually gainy, provided a feedback circuit partially compensates cavity loss and the dissipative cooperativity reaches unity (Wang et al., 20 Nov 2025).

The term also extends beyond microwave cavities. Strongly coupled magnon–plasmon polaritons in graphene–2D-ferromagnet heterostructures were calculated to exhibit Rabi couplings in the range of $100$ GHz, with electrical and mechanical tunability (Costa et al., 2022). Those results are passive, but they suggest a broader category of gain-driven MPs in which the electromagnetic component is plasmonic rather than cavity-photonic.

2. Non-Hermitian formulations, strong coupling, and threshold conditions

With loss included, the coupled-mode eigenfrequencies take the non-Hermitian form

$$\omega_{\pm} = \frac{1}{2}(\omega + \omega_{\rm cav}) - \frac{i(\gamma + \kappa)}{4} \pm \sqrt{ g^2 + \left[ \frac{i(\kappa - \gamma)}{4} + \frac{\omega - \omega_{\rm cav}}{2} \right]^2 },$$

where $g$0 and $g$1 are the magnon and cavity decay rates (Kritzell et al., 2023). On resonance,

$g$2

This immediately separates three regimes: strong-coupling or underdamped hybridization,

$g$3

an exceptional point,

$g$4

and weak-coupling or overdamped dynamics,

$g$5

(Kritzell et al., 2023).

Gain enters the same structure through effective linewidths,

$g$6

yielding, on resonance,

$g$7

Strong coupling then requires

$g$8

while polariton lasing or self-oscillation requires the imaginary part of at least one eigenfrequency to become negative, $g$9 (Kritzell et al., 2023). In the symmetric case $\eta$0, one has $\eta$1; the gain-driven regime requires $\eta$2 (Kritzell et al., 2023).

A second non-Hermitian formulation is specific to dissipative coupling. There the effective Hamiltonian is

$\eta$3

with $\eta$4 the feedback-reduced cavity damping and $\eta$5 the dissipative coupling strength (Wang et al., 20 Nov 2025). The associated dissipative cooperativity is

$\eta$6

and $\eta$7 is both the lasing threshold and the condition for a zero-linewidth polariton mode identified as a perfect Friedrich–Wintgen bound state in the continuum (Wang et al., 20 Nov 2025).

3. Microwave implementations: auto-oscillation, coherent emission, and lasing

The first direct microwave implementation of a gain-driven polariton employed a gain-embedded cavity magnonics platform composed of a half-wavelength microstrip line resonator and a $\eta$8 mm YIG sphere (Yao et al., 2023). The bare cavity parameters were

$\eta$9

the Kittel-mode damping was

$\eta=1$0

the coherent magnon–photon coupling was

$\eta=1$1

and the dissipative magnon–photon coupling was

$\eta=1$2

(Yao et al., 2023). At $\eta=1$3 V the one-photon gain reached

$\eta=1$4

with nonlinear saturation

$\eta=1$5

(Yao et al., 2023).

That system exhibited polariton auto-oscillations, a polariton phase singularity, self-selection of a polariton bright mode, and gain-induced magnon–photon synchronization (Yao et al., 2023). At $\eta=1$6, with no external coherent drive, it emitted at

$\eta=1$7

with peak power $\eta=1$8 mW and full width at half maximum $\eta=1$9 Hz (Yao et al., 2023). The same platform demonstrated coherent microwave amplification of $$\tilde{\omega}_c = \omega_c + i\bigl(G - \kappa_c - \gamma|a|^2\bigr),$$0 dB and coherent microwave emission with $$\tilde{\omega}_c = \omega_c + i\bigl(G - \kappa_c - \gamma|a|^2\bigr),$$1 (Yao et al., 2023).

A complementary realization used dissipative coupling rather than explicit negative damping of the hybrid mode. In a split-ring cavity coupled to a $$\tilde{\omega}_c = \omega_c + i\bigl(G - \kappa_c - \gamma|a|^2\bigr),$$2 mm YIG sphere, the experimentally extracted rates were

$$\tilde{\omega}_c = \omega_c + i\bigl(G - \kappa_c - \gamma|a|^2\bigr),$$3

(Wang et al., 20 Nov 2025). A feedback circuit reduced the effective cavity loss while retaining its dissipative character, and single-mode magnon-polariton lasing appeared precisely at $$\tilde{\omega}_c = \omega_c + i\bigl(G - \kappa_c - \gamma|a|^2\bigr),$$4, followed by amplification in the strong dissipative coupling regime (Wang et al., 20 Nov 2025). A common misconception is that gain-driven MPs require a conventional gain medium inside the cavity; this experiment showed that a cavity mode partially compensated through feedback and coupled dissipatively to a passive magnon mode is sufficient for single-mode lasing (Wang et al., 20 Nov 2025).

4. Nonlinear active MPs: attractors, combs, chaos, and field-amplified response

Active MPs also support a nonlinear fixed-point landscape that is absent in conventional passive systems at comparable effective photon numbers. In a self-oscillating microwave cavity coupled to a YIG sphere, the feedback loop generated a van der Pol cavity, while Kerr frequency pulling and Suhl-mediated magnon–magnon scattering produced an enhanced effective nonlinearity (Wu et al., 30 Apr 2026). The fitted parameters were

$$\tilde{\omega}_c = \omega_c + i\bigl(G - \kappa_c - \gamma|a|^2\bigr),$$5

and the effective Kerr coefficient was

$$\tilde{\omega}_c = \omega_c + i\bigl(G - \kappa_c - \gamma|a|^2\bigr),$$6

(Wu et al., 30 Apr 2026).

Stability analysis revealed multiple unstable-fixed-point phases and a triple-point region (Wu et al., 30 Apr 2026). By increasing gain, the system displayed the first experimental evidence of explosive growth of bistability, then transitions to multifrequency limit cycles, comb-like or fractal spectra, and broadband chaotic dynamics, all at self-oscillation powers from $$\tilde{\omega}_c = \omega_c + i\bigl(G - \kappa_c - \gamma|a|^2\bigr),$$7 to $$\tilde{\omega}_c = \omega_c + i\bigl(G - \kappa_c - \gamma|a|^2\bigr),$$8 dBm, corresponding to approximately $$\tilde{\omega}_c = \omega_c + i\bigl(G - \kappa_c - \gamma|a|^2\bigr),$$9–$G$0 (Wu et al., 30 Apr 2026). Near a critical point, magnetic-field-triggered switching between nonlinear emission states produced spectral shifts up to $G$1 times the bare gyromagnetic response (Wu et al., 30 Apr 2026). This does not merely indicate larger signal power. It indicates that internal gain reshapes the attractor topology of the MP system itself.

A related theoretical treatment of gain-driven MPs in the strong-coupling and ultrastrong-coupling regimes used an effective circuit model with magnetic flux as the magnon variable and a negative-resistance element as the gain source (Suzuki et al., 11 Sep 2025). The coherent coupling at the magnon–photon crossing was

$G$2

with $G$3 the main tuning knob (Suzuki et al., 11 Sep 2025). Two examples were emphasized: $G$4 (Suzuki et al., 11 Sep 2025). In the strong-coupling regime, self-Kerr nonlinearity generates a frequency shift and reduces the coherent magnon–photon coupling; in the ultrastrong-coupling regime, the coherent coupling overcomes the self-Kerr nonlinearity and effectively couples to gain via the imaginary part of complex eigenfrequencies, resulting in magnon-like auto-oscillations (Suzuki et al., 11 Sep 2025).

5. Exceptional points, PT symmetry, and coherent control

Non-Hermitian control of MPs was established experimentally before active auto-oscillation platforms matured. In a rectangular microwave cavity containing a $G$5 mm YIG sphere, coherent perfect absorption was used to realize an effective PT-symmetric Hamiltonian

$G$6

with eigenfrequencies

$G$7

(Zhang et al., 2017). The cavity linewidth was

$G$8

the magnon linewidth

$G$9

and the maximal coupling

$\kappa_c$0

(Zhang et al., 2017). The exceptional point occurred at $\kappa_c$1, separating a regime with real-frequency splitting from a regime with imaginary splitting (Zhang et al., 2017).

A later active platform extended exceptional-point physics into coherent control. Two active microwave resonators, each containing a $\kappa_c$2 mm YIG sphere, produced two coupled MP modes governed by

$\kappa_c$3

with

$\kappa_c$4

for the selected MP pair (Lambert et al., 5 Nov 2025). Writing

$\kappa_c$5

the eigenvalues become

$\kappa_c$6

and the exceptional point lies at

$\kappa_c$7

(Lambert et al., 5 Nov 2025). Encircling the exceptional point generated chiral population transfer between the lower and upper MP eigenmodes, while driving through and beyond the exceptional point prepared an equal superposition of eigenmodes (Lambert et al., 5 Nov 2025). An important correction to a common intuition emerged here: touching the exceptional point does not by itself erase state history, because the dynamics at the exceptional point depend on generalized eigenvectors (Lambert et al., 5 Nov 2025).

6. Terahertz, ultrastrong, topological, and plasmonic extensions

The gain-driven literature is rooted mostly in GHz cavity magnonics, but several passive platforms already supply the quantitative baselines needed for active extensions. In NiO, a $\kappa_c$8 slab acts as a terahertz Fabry–Pérot cavity strongly coupled to a $\kappa_c$9 THz antiferromagnetic magnon, with

$\gamma |a|^2$0

(Kritzell et al., 2023). The corresponding average passive loss rate is

$\gamma |a|^2$1

and the strong-coupling threshold set by the loss imbalance is

$\gamma |a|^2$2

(Kritzell et al., 2023). This suggests that magnon or photon gain on the order of $\gamma |a|^2$3–$\gamma |a|^2$4 GHz would be sufficient to enter a regime of growing polariton amplitudes while remaining deep in strong coupling (Kritzell et al., 2023).

In on-chip ultrastrong coupling, a YBCO resonator coupled to multiple Py stripes achieved

$\gamma |a|^2$5

together with a Bloch–Siegert shift of about $\gamma |a|^2$6 MHz and a diamagnetic suppression factor

$\gamma |a|^2$7

(Yoshii et al., 8 Jul 2025). The cooperativity was estimated as $\gamma |a|^2$8 for one stripe and $\gamma |a|^2$9 for $100$0 stripes (Yoshii et al., 8 Jul 2025). Although passive, that platform supplies exactly the Hamiltonian ingredients needed for gain-driven ultrastrong MPs: counter-rotating terms, reduced $100$1 contributions, and large collective coupling (Yoshii et al., 8 Jul 2025).

Two additional directions are presently inferential rather than demonstrated as full gain-driven MPs. First, parity-time-symmetric dipolarly coupled magnonic waveguides showed spin-orbit-torque-controlled exceptional points, wave-vector-dependent PT phases, nonreciprocal magnon propagation, and gain-selected modes (Wang et al., 2022). This suggests wave-vector-selective and nonreciprocal gain-driven MPs once a photonic mode is hybridized to the PT-engineered magnon sector. Second, topological magnon amplification in kagome ferromagnets showed parametric edge-mode instabilities, a topological travelling-wave magnon amplifier, and a topological magnon laser (Malz et al., 2019). This suggests topological gain-driven MPs in which polariton gain is concentrated on chiral edge channels. Graphene–2D-ferromagnet heterostructures, with Rabi couplings in the range of $100$2 GHz and electrical or mechanical control, likewise suggest gain-driven magnon–plasmon polaritons in the terahertz regime (Costa et al., 2022).

Gain-driven MPs therefore occupy a spectrum of regimes rather than a single mechanism. At one end are feedback-stabilized microwave cavity magnonics platforms that already realize coherent emission, single-mode lasing, exceptional-point control, and nonlinear attractor transitions (Yao et al., 2023, Wang et al., 20 Nov 2025, Wu et al., 30 Apr 2026, Lambert et al., 5 Nov 2025). At the other are passive terahertz, ultrastrong, topological, and plasmonic systems whose loss scales, coupling strengths, and Hamiltonian structures now make quantitative active design possible (Kritzell et al., 2023, Yoshii et al., 8 Jul 2025, Costa et al., 2022).