Intercavity Polaritons

• Intercavity polaritons are hybrid light–matter modes with spatially segregated photon and exciton components distributed across distinct cavities, enabling unique interference effects.
• They are modeled by a three-mode coupled oscillator that uses photon tunneling and exciton-photon coupling to yield a tunable dispersion and a heavy effective mass.
• Intercavity polaritons facilitate experimental studies of slow-light behavior, nonlinear coupling, and protected dark-state dynamics in advanced photonic systems.

Intercavity polaritons are hybrid light–matter eigenmodes of coupled-cavity systems in which the photonic and excitonic constituents are intentionally distributed across different cavities rather than co-located in a single resonator. In the canonical realization, a purely photonic cavity is tunnel-coupled to a second cavity that contains both a photon mode and an excitonic medium, so that the relevant middle branch can become a spatially indirect superposition of a photon in one cavity and an exciton in the other. This distinguishes intercavity polaritons from conventional intracavity exciton-polaritons, and it places them in direct conceptual continuity with $\Lambda$-scheme dark-state polaritons, slow-light analogues, and coupled-mode polaritonics (Jomaso et al., 2023).

1. Definition and physical scope

The defining feature of an intercavity polariton is spatial segregation of its light and matter sectors. In the room-temperature double Fabry–Pérot implementations developed for Frenkel excitons, the left cavity hosts only a photonic mode, while the right cavity hosts both a photonic mode and molecular excitons; photon tunneling through a thin intermediate mirror then produces a middle polariton that can eliminate the right-cavity photon contribution and retain only the left-cavity photon plus the right-cavity exciton (Jomaso et al., 2023). In this sense, the state is “intercavity” not merely because two cavities are present, but because the hybrid quasiparticle itself is shared across them.

This usage differs from the standard single-cavity polariton picture, where photon and exciton occupy the same optical volume and are described by an ordinary two-level anticrossing. It also differs from a purely photonic cavity supermode, because the intercavity state remains a genuine mixed light–matter excitation. A representative experimental platform uses two vertically stacked nanocavities, with PMMA in the left cavity and Erythrosin B dispersed in PVA in the right cavity, and realizes room-temperature intercavity Frenkel polaritons (Jomaso et al., 2023).

A closely related formulation emphasizes the “pure intercavity exciton-polariton” as a mixed light–matter state whose photon part resides predominantly in one cavity and whose exciton part resides in the other, with remote coupling mediated by the intermediate cavity photon. In that description, the central objective is not only to create the intercavity state at normal incidence, but also to preserve its spatial segregation and photon–exciton mixing over a wide momentum range (Sánchez-Martínez et al., 27 Jan 2025).

2. Canonical coupled-cavity model

The minimal microscopic description is a three-mode coupled-oscillator Hamiltonian. In one common notation,

$$\hat H= \begin{bmatrix} \omega_c^{L}(\theta) & -t & 0 \ -t& \omega_c^{R}(\theta) &\Omega\ 0&\Omega& \omega_X \end{bmatrix},$$

where $\omega_c^{L}(\theta)$ and $\omega_c^{R}(\theta)$ are the left- and right-cavity photon energies, $\omega_X$ is the exciton energy, $t$ is the cavity-cavity photon tunneling amplitude, and $\Omega$ is the exciton-photon coupling in the right cavity (Jomaso et al., 2023). Diagonalization yields three branches—lower polariton, middle polariton, and upper polariton—and the middle branch is the one associated with intercavity character.

At normal incidence and at the appropriate resonance condition, the middle polariton becomes a “pure intercavity polariton” containing only the left-cavity photon and the exciton, with amplitudes $\Omega/\sqrt{\Omega^2+t^2}$ and $t/\sqrt{\Omega^2+t^2}$, respectively, and no right-cavity photon contribution (Jomaso et al., 2023). The corresponding mixing angle satisfies

$\sin(\phi)=\frac{t}{\sqrt{\Omega^2+t^2}},$

so the light–matter admixture is controlled by the ratio $$\hat H= \begin{bmatrix} \omega_c^{L}(\theta) & -t & 0 \ -t& \omega_c^{R}(\theta) &\Omega\ 0&\Omega& \omega_X \end{bmatrix},$$0. This is the direct cavity-QED analogue of a dark-state construction in a $$\hat H= \begin{bmatrix} \omega_c^{L}(\theta) & -t & 0 \ -t& \omega_c^{R}(\theta) &\Omega\ 0&\Omega& \omega_X \end{bmatrix},$$1-scheme.

A driven-dissipative version of the same architecture replaces $$\hat H= \begin{bmatrix} \omega_c^{L}(\theta) & -t & 0 \ -t& \omega_c^{R}(\theta) &\Omega\ 0&\Omega& \omega_X \end{bmatrix},$$2 by $$\hat H= \begin{bmatrix} \omega_c^{L}(\theta) & -t & 0 \ -t& \omega_c^{R}(\theta) &\Omega\ 0&\Omega& \omega_X \end{bmatrix},$$3 and writes

$$\hat H= \begin{bmatrix} \omega_c^{L}(\theta) & -t & 0 \ -t& \omega_c^{R}(\theta) &\Omega\ 0&\Omega& \omega_X \end{bmatrix},$$4

Under the resonant condition $$\hat H= \begin{bmatrix} \omega_c^{L}(\theta) & -t & 0 \ -t& \omega_c^{R}(\theta) &\Omega\ 0&\Omega& \omega_X \end{bmatrix},$$5, the middle polariton operator is

$$\hat H= \begin{bmatrix} \omega_c^{L}(\theta) & -t & 0 \ -t& \omega_c^{R}(\theta) &\Omega\ 0&\Omega& \omega_X \end{bmatrix},$$6

with $$\hat H= \begin{bmatrix} \omega_c^{L}(\theta) & -t & 0 \ -t& \omega_c^{R}(\theta) &\Omega\ 0&\Omega& \omega_X \end{bmatrix},$$7 at $$\hat H= \begin{bmatrix} \omega_c^{L}(\theta) & -t & 0 \ -t& \omega_c^{R}(\theta) &\Omega\ 0&\Omega& \omega_X \end{bmatrix},$$8 (Carmona-Moreno et al., 24 Mar 2026). This formulation makes explicit that the intermediate right-cavity photon is a virtual pathway rather than a constituent of the ideal middle branch.

3. Pure-state protection, dispersion flattening, and heavy mass

The exact intercavity condition is realized at normal incidence, but the more demanding problem is preserving that character away from $$\hat H= \begin{bmatrix} \omega_c^{L}(\theta) & -t & 0 \ -t& \omega_c^{R}(\theta) &\Omega\ 0&\Omega& \omega_X \end{bmatrix},$$9. In the protected-state formulation, the middle polariton at

$\omega_c^{L}(\theta)$0

is

$\omega_c^{L}(\theta)$1

and its right-cavity photonic residue satisfies

$\omega_c^{L}(\theta)$2

(Sánchez-Martínez et al., 27 Jan 2025). The central question is then how rapidly $\omega_c^{L}(\theta)$3 grows at finite momentum.

The protection criterion is expressed as

$\omega_c^{L}(\theta)$4

This inequality defines a transparency-window condition analogous to electromagnetically induced transparency: as long as the middle branch remains within that window, it stays effectively decoupled from the right-cavity photonic mode and keeps its intercavity identity (Sánchez-Martínez et al., 27 Jan 2025). The same level-structure engineering therefore controls both the transparency-window physics and the preservation of the spatially segregated hybrid state.

Near normal incidence, the middle-polariton dispersion is

$\omega_c^{L}(\theta)$5

with

$\omega_c^{L}(\theta)$6

The branch is therefore heavier than the bare left-cavity photon, and the heavy-mass character is tunable by the ratio $\omega_c^{L}(\theta)$7 (Sánchez-Martínez et al., 27 Jan 2025). In measured detuning scans, decreasing the right-cavity detuning from $\omega_c^{L}(\theta)$8 to $\omega_c^{L}(\theta)$9 and then to $\omega_c^{R}(\theta)$0 broadens the angular range over which the right-cavity residue remains negligible, with the negative-detuning case producing a quasi-flat middle branch and essentially vanishing right-cavity contribution even at large angles (Sánchez-Martínez et al., 27 Jan 2025).

4. Nonequilibrium dynamics, lifetime effects, and collective modes

Intercavity polaritons were introduced partly as a route to heavy-polariton or slow-light-like behavior without an in-plane lattice. In angle-resolved reflectance, the middle polariton stays close to the exciton energy and becomes markedly flatter as the left-cavity detuning approaches resonance, with a dispersion reduction by a factor of $\omega_c^{R}(\theta)$1 compared to the upper and lower branches (Jomaso et al., 2023). Because the flattening arises from the three-level interference structure rather than simple exciton-like detuning, the branch can remain more balanced in its light–matter composition than a conventional two-level polariton pushed toward the exciton line.

The same architecture produces distinctive dynamical signatures. Time-correlated single-photon counting shows that the middle branch has slower short-time dynamics than the lower branch; at the pure intercavity condition $\omega_c^{R}(\theta)$2, the biexponential fit gives

$\omega_c^{R}(\theta)$3

whereas the lower polariton at the same detuning has

$\omega_c^{R}(\theta)$4

The corresponding effective short-time rate is

$\omega_c^{R}(\theta)$5

which is much smaller than the dominant lower-polariton decay rate (Jomaso et al., 2023). Under continuous-wave pumping, when the upper polariton becomes predominantly right-cavity-photon-like and the middle polariton has lost its right-photon component, the middle-polariton fluorescence is strongly suppressed, which is interpreted as drive-pathway-selective darkness rather than an optically forbidden exciton (Jomaso et al., 2023).

A later driven-dissipative treatment identifies an even sharper dynamical distinction: when the middle polariton branch is resonantly driven, coherent Rabi oscillations are suppressed and the system evolves monotonically toward its steady state, in contrast to the damped Rabi oscillations found under lower- or upper-branch driving (Carmona-Moreno et al., 24 Mar 2026). In the small-$\omega_c^{R}(\theta)$6 limit, the steady state satisfies

$\omega_c^{R}(\theta)$7

while the left-cavity and excitonic amplitudes remain finite, making the dark-state character explicit. Including exciton–exciton interactions then yields Bogoliubov excitations with

$\omega_c^{R}(\theta)$8

and low-momentum phonon-like behavior

$\omega_c^{R}(\theta)$9

so the driven middle branch supports collective modes that inherit their interaction scale from the excitonic fraction while preserving the underlying intercavity structure (Carmona-Moreno et al., 24 Mar 2026).

5. Nonlinear, ultrafast, and multimode generalizations

The coupled-cavity concept extends beyond the protected $\omega_X$0-scheme dimer. In the mid-infrared, a checkerboard matrix of adjacent Fabry–Pérot microcavities filled with W(CO)$\omega_X$1/hexane shows that pumping polaritons in one cavity can change the nonlinear response of polaritons in a neighboring cavity (Wang et al., 2019). The two bare cavity resonances are approximately $\omega_X$2 and $\omega_X$3, while the molecular vibration lies at $\omega_X$4, and the measured 2D IR spectrum exhibits cross peaks between cavity-specific polaritons that are absent when spectra from two uncoupled cavities are simply added (Wang et al., 2019).

The decisive observation is that a pump at

$\omega_X$5

produces a probe response at

$\omega_X$6

demonstrating intercavity nonlinear coupling at $\omega_X$7 (Wang et al., 2019). The accompanying model requires both intercavity photonic delocalization and molecular anharmonicity: turning off delocalization reduces the response to isolated single-cavity polaritons, while setting the molecular anharmonicity to zero removes the 2D IR signal. This establishes a broader sense in which intercavity polaritonics can denote not only protected single-particle eigenmodes, but also spatially extended nonlinear interactions between cavity-specific polaritons.

A further generalization arises in multimode nanophotonic environments, where the relevant “cavity mode” may itself be an interference-induced resonance extracted from several overlapping modes rather than a bare single-cavity pole. In that language, effective non-Hermitian modes with complex emitter couplings can hybridize with emitters to form polaritons even when the usual single-mode strong-coupling criterion is not satisfied, and the resulting branches can display imaginary Rabi splitting, long-lived polaritons, and interference-protected linewidth asymmetry (Ben-Asher et al., 13 Oct 2025). This suggests that intercavity polaritons need not be confined to simple tunnel-coupled dimers: a plausible implication is that in structured photonic environments the relevant intercavity-like hybrid can be a dressed supermode embedded in a broader multimode background.

6. Relation to neighboring polariton concepts

The term “intercavity polariton” should be distinguished from several nearby but non-equivalent constructions. A finite dipolar chain inside a single elongated cavity waveguide can develop edge-localized polaritonic Tamm states controlled by cavity geometry and long-range dipole couplings, but that system is explicitly not about two or more distinct cavities coupled to one another (Downing et al., 2020). It is relevant as a prototype of geometry-controlled photon-mediated localization, not as an intercavity polariton in the narrow sense.

Likewise, a single confined polariton cavity with two polarization modes can be modeled as two coupled nonlinear resonators with self- and cross-Kerr nonlinearities, and birefringence-induced polarization coupling can generate both conventional and interference-mediated antibunching; however, the two “resonators” are polarization degrees of freedom of one cavity rather than two separate cavities (Bleu et al., 2021). A single molecule coupled simultaneously to two degenerate orthogonally polarized Fabry–Pérot modes also realizes a genuine two-mode polariton problem and supports lower, middle, and upper vibro-polaritonic states, but it remains a two-mode single-cavity system rather than an intercavity architecture (Fischer et al., 2022).

A similar boundary applies to Floquet-engineered multimode cavity polaritons. Periodic modulation of an atomic transition can redistribute spectral weight into Floquet sidebands and simultaneously support dark polaritons in two chosen spatial modes such as TEM$\omega_X$8 and TEM$\omega_X$9, with both same-mode and cross-mode blockade (Clark et al., 2018). This is best described as an intermode analogue of intercavity polariton physics: multiple photonic channels share a common interacting matter sector, but the experiment does not involve two distinct physical cavities.

Taken together, these distinctions show that intercavity polaritons occupy a specific place within the broader polariton taxonomy. The core object is a hybrid mode or interaction channel that is genuinely distributed across coupled cavities, typically through photon tunneling and a $t$0-type level structure. Related single-cavity multimode systems can reproduce portions of the mathematics or phenomenology, but they do not by themselves satisfy that spatial definition.