Optomagnonic Dielectric Resonators

• Optomagnonic dielectric resonators are devices that co-localize optical, microwave, and magnonic modes within magneto-insulating media to enhance magneto-optical interactions.
• They employ diverse architectures—such as YIG whispering-gallery cavities, planar Bragg microcavities, and Mie resonators—to optimize spatial and symmetry overlap for efficient mode coupling.
• Design optimizations, including improved geometrical configurations and material quality, have been shown to boost conversion efficiency and reduce losses in these resonant systems.

Searching arXiv for the cited optomagnonic dielectric resonator papers and closely related work. arxiv.search(query="optomagnonic dielectric resonators YIG whispering gallery Mie resonators photonic crystal 1D microcavity", max_results=10, sort_by="relevance") arxiv.search(query="(Zhang et al., 2015)", max_results=5, sort_by="relevance") arxiv.search(query="(Haigh et al., 2015, Graf et al., 2020, Ozerov et al., 2021, Xia et al., 2021, Pantazopoulos et al., 2019, Demirchyan et al., 23 Sep 2025, Xue et al., 29 Jun 2026)", max_results=20, sort_by="relevance") Optomagnonic dielectric resonators are dielectric resonant structures made of, or containing, magneto-insulating media in which optical, microwave, and magnonic degrees of freedom occupy the same device volume and interact through magneto-optical coupling. In the literature, the term spans several concrete architectures: yttrium iron garnet (YIG) whispering-gallery-mode microspheres, magnetic dielectric films inside Bragg-mirror microcavities, magneto-optical Mie metasurfaces, and microscale photonic–magnonic crystals. Across these platforms, the recurring objective is to increase the spatial and symmetry overlap between confined electromagnetic fields and collective spin excitations so that magnons can mediate optical frequency conversion, selective spin-wave excitation, synchronization, or strongly enhanced magneto-optical response (Zhang et al., 2015, Haigh et al., 2015, Graf et al., 2020).

1. Platform concept and material basis

The material system most frequently associated with optomagnonic dielectric resonators is YIG. In the whispering-gallery literature, YIG is chosen because it is simultaneously an excellent magnonic resonator with long-lived spin-wave modes, a magnetic insulator with strong magneto-optical response, and low-loss enough in the telecom band to support optical WGMs (Zhang et al., 2015). In the broader dielectric-antenna perspective, YIG, BIG, spinels, and hexaferrites are treated as magneto-insulating materials from which all-dielectric resonant antennas can be built as standalone strongly coupled systems (Maksymov, 2018).

The architectural distinction from conventional metal-cavity magnonics is central. In the conventional platform, a microwave metal resonator traps photons and a magnetic object such as a YIG sphere is placed where the microwave magnetic field is maximal. The dielectric-resonator alternative instead uses a high-permittivity, low-loss dielectric body that itself supports the relevant resonances, avoids ohmic loss at high microwave frequencies, eliminates resonator-wall eddy currents, and leaves the same resonant volume accessible to optics, phonons, thermal control, and electrical control (Maksymov, 2018). This suggests that the term “optomagnonic dielectric resonator” is best understood not as a single geometry but as a design principle: the resonator body itself is both the photonic environment and the magnetic medium, or it is engineered to place the optical mode directly on top of a magneto-optical medium.

Three platform families recur. First, YIG whispering-gallery resonators use total internal reflection in a ferrimagnetic sphere, disk, or ring to localize optical WGMs while the same YIG body hosts ferromagnetic or spin-wave modes (Zhang et al., 2015, Demirchyan et al., 23 Sep 2025). Second, planar Bragg cavities and optomagnonic crystals use dielectric periodicity to co-localize optical defect modes and standing spin-wave modes inside a magnetic film or YIG defect region (Ozerov et al., 2021, Pantazopoulos et al., 2019, Graf et al., 2020). Third, high-index Mie resonators and dielectric dimers exploit magnetic dipole and quadrupole resonances, circular displacement current, or magnetic near-field enhancement to reshape the electromagnetic field topology seen by the magneto-optical tensor (Xia et al., 2021, Boudarham et al., 2013).

2. Whispering-gallery YIG resonators

The most direct early realization is the single-crystal YIG sphere used as a triply resonant optomagnonic cavity. In one implementation, the platform is a 300-$\mu$m-diameter single-crystal YIG sphere mounted on a 125-$\mu$m silica supporting fiber and interrogated optically not with a silica taper but with an integrated silicon nitride waveguide. The engineering reason is explicit: the YIG refractive index is about 2.2 in the telecom C-band, whereas silica is poorly matched, while the silicon nitride waveguide has effective index about 2.0, allowing efficient coupling to both TM and TE optical WGMs (Zhang et al., 2015).

The optical WGMs are labeled by $(q,l,n)$ and are grouped into TM-like $\pi$-polarized and TE-like $\sigma^+$-polarized families. Transmission spectroscopy shows large extinction ratios exceeding 10 dB, with measured free spectral ranges of $1.0765$ nm for $\sigma^+$ and $1.1068$ nm for $\pi$, close to the predicted WGM spacing of $1.1580$ nm. After mechanical polishing with silicon oxide slurry and chemical cleaning with buffered oxide etch, the sphere reaches $\mu$0 and $\mu$1, while the intrinsic infrared absorption of YIG is described as low enough that values up to about $\mu$2 should be possible (Zhang et al., 2015).

A complementary 2015 study established the static magneto-optical behavior of YIG WGMs in polished microspheres of radii $\mu$3, $\mu$4, and $\mu$5, coupled through a rutile prism. There the optical resonance frequencies follow the WGM relation

$\mu$6

and the magnetization direction tunes the resonance via the Voigt effect while the Faraday effect controls polarization rotation and mode mixing. The measured resonance shift amplitude is about $\mu$7, with a clear $\mu$8-like dependence for out-of-plane rotation and no systematic frequency shift for in-plane rotation (Haigh et al., 2015).

These two WGM papers fix the basic ontology of the field. The sphere is not merely a magnetic sample placed in an optical cavity; it is itself the dielectric optical cavity and the magnonic resonator. The same geometry also exposes a characteristic WGM subtlety: because the propagation direction rotates continuously around the orbit, the local optical basis rotates relative to the static magnetization, so Faraday and Voigt terms are sampled in a strongly geometry-dependent way (Haigh et al., 2015).

3. Planar dielectric cavities and optomagnonic crystals

A second major class replaces spherical confinement with one-dimensional dielectric confinement. In one-dimensional optomagnonic microcavities, a magnetic dielectric film $\mu$9 is sandwiched between two nonmagnetic dielectric Bragg mirrors made of alternating quarter-wave layers $(q,l,n)$0, forming a Fabry–Perot-type microcavity described as

$(q,l,n)$1

When the wavelength lies in the photonic band gap, the cavity produces a resonantly enhanced and strongly non-uniform optical field inside the magnetic layer. For circularly polarized light,

$(q,l,n)$2

and the inverse Faraday effect generates an effective field

$(q,l,n)$3

with spatial profile inherited from $(q,l,n)$4 (Ozerov et al., 2021).

That spatial structuring is used to excite perpendicular standing spin waves selectively. The post-pulse initial condition is an ultrafast IFE “kick,” and the excitation amplitude of the $(q,l,n)$5-th PSSW is governed by the overlap integral

$(q,l,n)$6

The paper’s selection rule is explicit: when the wavevector characterizing the IFE field profile satisfies $(q,l,n)$7, the PSSW of that order is excited most efficiently (Ozerov et al., 2021). In this platform, the optical cavity mode profile becomes a mode-selective magnetic-force profile.

A more strongly co-localized version is the “optomagnonic crystal,” a one-dimensional array of holes etched into a YIG slab with an abrupt defect at the center. The periodic hole array creates both a 1D photonic band gap and a magnonic band structure, and the defect localizes both an optical mode and a magnetic mode in the same region. In the proof-of-principle geometry, the lattice constant is around $(q,l,n)$8, the optical defect mode frequency is roughly $(q,l,n)$9 with $\pi$0, and the localized magnon mode frequency is around $\pi$1 (Graf et al., 2020).

The interaction Hamiltonian is written as

$\pi$2

with Faraday and Cotton–Mouton contributions. For the initial geometry, the numerically calculated couplings are $\pi$3, $\pi$4, and total $\pi$5. In an optimized geometry, $\pi$6 and total $\pi$7, with overlap $\pi$8 and directionality $\pi$9 (Graf et al., 2020). The explicit design lesson is that symmetry and defect engineering, not only material choice, determine whether the coupling integral is nonzero and sizable.

4. Mie resonators, dielectric antennas, and field-topology engineering

The optomagnonic dielectric-resonator concept also includes high-index resonators that operate through Mie physics rather than WGM or Fabry–Perot confinement. In a Si/Ce:YIG/YIG/$\sigma^+$0 metasurface, Si nanodisks of period $\sigma^+$1 nm, radius $\sigma^+$2 nm, and height $\sigma^+$3 nm are placed above 200 nm Ce:YIG and 50 nm YIG on double-side-polished quartz. The magneto-optical tensor is written as

$\sigma^+$4

and the decisive point is that magnetic dipole and magnetic quadrupole resonances generate a circulating electric field and therefore a circular displacement current in the Ce:YIG layer (Xia et al., 2021).

The reported magneto-optical response is not simply an enhancement of a planar-film signal. Under s-polarized incidence in the transverse magneto-optical Kerr geometry, the structure shows giant TMOKE up to $\sigma^+$5, specifically $\sigma^+$6 at 1275 nm and $\sigma^+$7 at 1170 nm around the MQ resonance, with another TMOKE peak of about $\sigma^+$8 at 1375 nm near the MD mode. At 1250 nm, $\sigma^+$9 is observed with reflectivity as high as 73%. In near-normal transmission LMOKE-T, the rotation reaches $1.0765$0 at 1320 nm, whereas the bare Ce:YIG/YIG film gives only $1.0765$1 degrees, about two orders of magnitude smaller (Xia et al., 2021). The comparison to planar films is the paper’s central claim: the intrinsic tensor is unchanged, but the resonator’s modal field topology activates channels that are absent or negligible in planar geometry.

At larger scale, dielectric dimers and gap antennas provide a complementary field-engineering route. A dimer of identical cubic dielectric resonators with relative permittivity $1.0765$2 exhibits strong magnetic near-field confinement in a subwavelength gap. For a single cube, $1.0765$3 reaches about 60 at $1.0765$4 GHz. In the dimer, the longitudinal illumination case gives a field-intensity peak of about 61 near $1.0765$5 GHz for a 5 mm gap and about 130 when the gap is reduced to 2 mm; transverse cases reach about 24 at $1.0765$6 GHz and about 55 at $1.0765$7 GHz (Boudarham et al., 2013). The dielectric-antenna perspective takes these results as scalable evidence that high-permittivity dielectric resonators can confine magnetic energy strongly while keeping the electric enhancement modest, a property directly relevant when the target interaction is magnetic or magneto-optical rather than purely electric (Maksymov, 2018).

5. Coupling mechanisms, selection rules, and dynamic theory

In YIG-based optomagnonic resonators, the core interaction is commonly derived from the magnetization-dependent dielectric tensor. For the YIG sphere optomagnonic cavity,

$1.0765$8

and a magnon-induced fluctuation gives

$1.0765$9

The optical coupling matrix element is

$\sigma^+$0

with nonzero coupling only when energy and azimuthal orbital angular momentum are conserved: $\sigma^+$1 These relations encode the triple-resonant Brillouin-scattering-like interaction between pump photon, magnon, and sideband photon (Zhang et al., 2015).

The WGM geometry imposes polarization-selective selection rules. In the YIG-sphere transduction experiment, the pump is launched as TM-polarized light and the scattered signal emerges in the TE channel. For a representative case, the pump is at 1534.599 nm and the sideband is shifted by 6.75 GHz, exactly matching the magnon frequency; the sideband appears only on one side of the pump, consistent with the angular-momentum and spin selection rules (Zhang et al., 2015). In the static YIG WGM theory, the same geometry is parsed into Voigt and Faraday contributions: the Voigt effect shifts resonance frequencies quadratically in magnetization, while the Faraday effect produces polarization mixing through the local parameter

$\sigma^+$2

which remains small because geometrical birefringence dominates but is sufficient to generate orthogonal-polarization output when the magnetization has an in-plane component (Haigh et al., 2015).

For layered cavities, the dynamic theory goes beyond the frozen-permittivity picture. In the Floquet scattering-matrix formulation, the spin wave makes the defect-layer permittivity periodic in time,

$\sigma^+$3

so the optical field expands into sidebands at $\sigma^+$4. The framework captures strong modulation, multi-magnon inelastic scattering, and the triple-resonance condition $\sigma^+$5, where one-magnon conversion between two optical defect modes is resonantly enhanced (Pantazopoulos et al., 2019). The paper also states the limitation of the quasistatic adiabatic approximation: it cannot describe true energy exchange, cannot capture the relevance of the pump frequency dynamically, and fails precisely when triple resonance is important.

6. Performance metrics, optimization routes, and networked extensions

Measured and calculated performance varies strongly with geometry. In the YIG-sphere WGM transducer, the loaded quality factor of the fundamental $\sigma^+$6 magnon mode is 1230 at 1840 Oe bias field. The measured raw system conversion efficiency is about $\sigma^+$7, whereas the estimated internal power conversion efficiency is about $\sigma^+$8; the difference is attributed to imperfect resonance coupling and insertion losses in the optical and microwave paths (Zhang et al., 2015). In the optomagnonic crystal, the coupling reaches the kHz range but the optical $\sigma^+$9 remains modest and can degrade from $1.1068$0 in the initial geometry to $1.1068$1 in an optimized one, illustrating a direct tradeoff between mode co-localization and optical loss (Graf et al., 2020).

Geometry-first optimization is therefore a major theme. Disk and ring microcavities are explicitly proposed to solve the overlap bottleneck associated with a WGM near the resonator rim and a Kittel mode filling the full sphere volume. In YIG disk and ring microcavities, the interaction volume

$1.1068$2

is increased because the magnetic mode volume is reduced without destroying WGM localization. For radii in the $1.1068$3 range, the coupling is already more than five times larger than in YIG spheres: spheres about $1.1068$4, disks about $1.1068$5, and rings up to another factor of $1.1068$6 improvement in that size range. For small disks or rings with radius $1.1068$7, the single-photon coupling reaches approximately $1.1068$8, and for the smallest disk the paper estimates that intracavity optical power $1.1068$9 is sufficient to reach unity conversion efficiency, with $\pi$0 (Demirchyan et al., 23 Sep 2025).

The same paper also identifies the new practical limits. Larger resonators require optimal powers in the mW range, where nonlinear absorption, Kerr shifting, and thermo-optic frequency drift can destroy the triple-resonance condition (Demirchyan et al., 23 Sep 2025). Earlier WGM work already emphasized surface roughness, contamination, and index mismatch as dominant engineering constraints, with residual sub-micrometer aluminum oxide polishing grit singled out as a major source of optical scattering loss (Zhang et al., 2015). A plausible implication is that fabrication quality and modal overlap, rather than the nominal Faraday activity of YIG alone, determine whether a given dielectric resonator operates as a useful transducer.

Networked optomagnonic dielectric resonators extend the same logic to collective dynamics. In a 2026 model, two spatially separated optical WGM resonators are each coupled to a YIG sphere and linked by phase-dependent single-photon hopping,

$\pi$1

Using the covariance-matrix formalism, the study finds that as $\pi$2 is varied from $\pi$3 to $\pi$4, the magnon trajectories evolve from weakly correlated motion to a highly synchronized state, with stronger hopping enhancing synchronization and thermal occupation $\pi$5 strongly suppressing all synchronization measures (Xue et al., 29 Jun 2026). This suggests that phase-engineered photonic mediation is emerging as a control primitive for remote magnon organization in WGM-based optomagnonic devices.

The field’s forward directions are stated in concrete device terms rather than as a single consensus roadmap. Proposed improvements include smaller and smoother YIG spheres, YIG microdisks, doping to enhance the Faraday effect, whispering-gallery magnon modes, optical Mie resonances in nanoparticles, more elaborate defect geometries in optomagnonic crystals, and resonances such as anapoles, Fano modes, supercavity modes, and bound states in the continuum (Zhang et al., 2015, Demirchyan et al., 23 Sep 2025, Xia et al., 2021). Across these proposals, the unifying criterion remains unchanged: stronger optomagnonic behavior follows when dielectric resonators force photons and magnons to occupy the same region of space, with compatible symmetry and resonance conditions.