Dark-Strong Coupling Phenomena

• Dark-Strong Coupling is a regime where nominally dark modes (with weak direct radiative coupling) become active through strong hybridization with bright states.
• It spans multiple domains such as photonics, molecular polaritonics, and optomechanics, where experiments reveal features like normal-mode splitting and modified emission profiles.
• Mechanisms like mediated coupling, symmetry breaking, and interference enable control over light–matter interactions and chemical dynamics in these systems.

Dark-Strong Coupling denotes a family of phenomena in which nominally dark modes, states, or sectors participate in a strong-coupling regime, or are qualitatively reshaped by it. In photonics and plasmonics, the term commonly refers to subradiant plasmonic or cavity modes that acquire strong hybridization through mediated or symmetry-enabled pathways; in molecular polaritonics, it can denote the kinetic and entropic importance of dark polariton manifolds or the cavity-induced modification of nominally dark molecular states; in optomechanics and mechanical analogs, it designates dark superpositions whose existence depends on strong single-photon or cavity-mediated coupling; in excitonic materials, strong coupling can either suppress radiative decay or invert bright–dark level ordering; and in several cosmological works the phrase is repurposed for strong late-time interactions within the dark sector [(Meng et al., 2023); (Borges et al., 29 Apr 2025); (Xu et al., 2012); (Song et al., 28 Aug 2025); (Marra, 2015)].

1. Conceptual scope and core definitions

A recurring distinction across the literature is between bright and dark excitations. In plasmonic metamaterials, a bright mode is a resonant current or dipole oscillation that couples strongly to free-space radiation, whereas a dark mode has a vanishing or very weak net electric dipole moment for the incident polarization and therefore weak radiative coupling (Meng et al., 2023). In Dicke- or Tavis–Cummings-type molecular ensembles, only the fully symmetric collective exciton is bright, while the orthogonal subspaces are dark in the ideal model (Borges et al., 29 Apr 2025). In central-spin language, dark states are bath states annihilated by collective operators such as $A^\pm=\sum_i g_i s_i^\pm$, so that the central spin decouples from the interacting bath sector (Dimo et al., 2021). In optomechanics, darkness can mean complete decoupling of the cavity from the external drives through destructive interference, with the cavity left empty while the mechanical subsystem occupies a coherent superposition (Xu et al., 2012).

Strong coupling is likewise context-dependent. In cavity and plasmonic settings it is typically identified by resolvable normal-mode splitting or avoided crossings whose splittings exceed relevant linewidths, for example $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$ or, for collective molecular coupling, $g\sqrt{N}\gtrsim(\kappa+\gamma)/4$ (Meng et al., 2023, Borges et al., 29 Apr 2025). In dissipative non-Hermitian descriptions, the distinction between internal level splitting and observable spectral splitting becomes essential, so strong coupling in the level structure need not coincide with a visible doublet in every measured channel (Liu et al., 2020).

Taken together, these usages show that the expression is not tied to one universal mechanism. The unifying idea is that states regarded as dark because of symmetry, weak dipole moment, collective cancellation, or forbidden transition pathways can nevertheless become central once a strong-coupling channel, a mediated interaction, or a strongly hybridized basis is established.

2. Mediated and symmetry-enabled dark coupling in electromagnetic structures

A direct and highly explicit realization appears in a terahertz platform consisting of an EIT-like metamaterial integrated with a one-dimensional photonic-crystal cavity. In that system, the incident $x$-polarized field excites the cavity’s bright eigenmode, which couples directly to a bright split-ring resonator. A dark split-ring resonator is then excited primarily by magnetic near-field coupling to the bright resonator, and once excited it radiates into the cavity’s orthogonally polarized dark eigenmode. Because the cavity supports two degenerate orthogonal fundamental eigenmodes, the resulting interaction involves four oscillators—plasmonic bright and dark modes and photonic bright and dark modes—and produces four polariton branches in transmission, observed near $387$, $423$, $483$, and $520\ \mathrm{GHz}$ (Meng et al., 2023). The dynamics are reproduced by a four-coupled-harmonic-oscillator model with mode amplitudes $a_b$, $a_d$, $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$0, and $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$1 and symmetry-allowed couplings $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$2, $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$3, and $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$4 (Meng et al., 2023). This is a paradigmatic mediated form of dark-mode strong coupling: the dark plasmon does not directly couple efficiently to the incident radiation, yet it reaches the strong-coupling regime with cavity photons through the bright plasmon.

Earlier terahertz metamaterial work on two closely spaced split-ring resonators demonstrated the same principle in a simpler bright–dark dimer. Broken symmetry allows indirect excitation of an otherwise symmetry-forbidden dark plasmonic eigenmode. When the resonances are tuned into near degeneracy and the resonators touch, the spectra show a pronounced avoided crossing and an Autler–Townes-like doublet near $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$5; when the resonators are separated by $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$6, the interaction weakens and the splitting collapses (0901.0365). In that work, the transparency feature is attributed to mode splitting rather than to a narrow EIT-like window, because the bright and dark linewidths are comparable (0901.0365).

A related but conceptually distinct regime was formulated for single-emitter quantum plasmonics. There, a quantum emitter with a high-energy transition couples strongly to a dark plasmon pseudomode centered near $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$7–$2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$8, while a bright dipolar plasmon lies near $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$9. The dark emitter–plasmon interaction forms dark polaritons; the lower dark polariton can then hybridize with the bright mode and induce an observable vacuum Rabi splitting in the visible. In the model, the emitter–dark coupling $g\sqrt{N}\gtrsim(\kappa+\gamma)/4$0 is much larger than the emitter–bright coupling $g\sqrt{N}\gtrsim(\kappa+\gamma)/4$1, and the bright splitting is

$g\sqrt{N}\gtrsim(\kappa+\gamma)/4$2

with the dark-mode mixing encoded in $g\sqrt{N}\gtrsim(\kappa+\gamma)/4$3 (Rousseaux et al., 2019). The observable splitting is therefore a bright readout of an underlying dark strong-coupling process.

Lateral plasmonic crystals provide yet another variant. A gate-modulated two-dimensional electron gas supports Bloch plasmon modes for which the $g\sqrt{N}\gtrsim(\kappa+\gamma)/4$4 spectrum factorizes into bright and dark conditions, $g\sqrt{N}\gtrsim(\kappa+\gamma)/4$5 and $g\sqrt{N}\gtrsim(\kappa+\gamma)/4$6. Under homogeneous excitation only the bright solutions appear in transmission, while dark modes require inhomogeneous excitation or finite in-plane photon momentum (Gorbenko et al., 2024). Increasing the density modulation depth drives a weak-to-strong coupling transition between sub-resonators inside the unit cell, and increasing the quality factor drives a resonant-to-super-resonant transition (Gorbenko et al., 2024). Here darkness is selected by excitation geometry rather than by the absence of hybridization.

3. Dark polariton manifolds in molecular ensembles and chemistry

In molecular strong coupling, the ideal single-excitation Tavis–Cummings picture contains two bright polaritons and $g\sqrt{N}\gtrsim(\kappa+\gamma)/4$7 dark states. For $g\sqrt{N}\gtrsim(\kappa+\gamma)/4$8 identical two-level molecules,

$g\sqrt{N}\gtrsim(\kappa+\gamma)/4$9

and the one-excitation eigenenergies are

$x$0

with $x$1 (Borges et al., 29 Apr 2025). In that limit the dark manifold is symmetry-protected and cavity-inactive.

The multi-excitation problem changes this picture. For excitation number $x$2, additional hybrid states appear in reduced-symmetry sectors $x$3. These “dark polaritons” have finite photonic character, smaller collective splitting, and large degeneracy (Borges et al., 29 Apr 2025). Their relative abundance compared with purely dark states scales as

$x$4

where $x$5 (Borges et al., 29 Apr 2025). A central consequence is entropic: allowing more than one excitation increases the number of cavity-participating microstates, so polaritonic reaction pathways can become entropically competitive. In a three-level extension relevant to photochemistry, with an additional optically dark state $x$6 coupled to $x$7, tuning the lower polariton into resonance with $x$8 accelerates relaxation through photon leakage under suitable conditions (Borges et al., 29 Apr 2025).

A different route by which dark states cease to be inert is cavity-induced modification of molecular rovibrational structure. In a first-principles molecular model including electronic and nuclear degrees of freedom, dark states that would be exactly decoupled in a two-level Tavis–Cummings picture acquire nontrivial nuclear correlations and energy shifts under collective strong coupling (Galego et al., 2015). For two molecules, the photonically excited surface indirectly mixes the nominally dark singly excited surfaces, and expanding the resulting lower-polariton, dark-state, and upper-polariton potential-energy surfaces near their minima yields

$x$9

For realistic couplings with $387$0, $387$1 is of order a few percent on the lower polariton and smaller on the dark-state and upper-polariton surfaces (Galego et al., 2015). Dark states therefore remain weakly photonic but are no longer dynamically trivial.

Experimental photochemistry on azopyrrole uses the term in a kinetic sense. Both $387$2 and $387$3 isomers couple collectively to a single cavity mode through the visible $387$4–$387$5 transition, and the Rabi splitting collapses during the photoisomerization process as the oscillator strength changes. In the $387$6 series, $387$7 decreases to $387$8 at $387$9, crossing the strong-to-weak boundary defined by the molecular FWHM of about $423$0 (Garg et al., 15 Mar 2025). Under upper-polariton or molecular-band excitation, the apparent rate increases and shows a sharp slope change correlated with increased dark-state spectral overlap, whereas lower-polariton excitation shows the opposite trend; in the ultrastrong-coupling series, with $423$1 and still $423$2 after $423$3, only a single slope is observed (Garg et al., 15 Mar 2025). Here the dark manifold functions as a dense reaction reservoir rather than a spectroscopically isolated state.

4. Optomechanical and mechanical realizations

In single-photon optomechanics, dark states arise from destructive interference between two sideband-selective driving pathways. For a radiation-pressure Hamiltonian

$423$4

single-photon strong coupling means that the displacement $423$5 is large enough to make Franck–Condon structure resolvable (Xu et al., 2012). Choosing detunings $423$6 and $423$7 and imposing a Franck–Condon node $423$8 confines the dynamics to a finite phonon manifold and yields a cavity-dark mechanical superposition

$423$9

with recurrence coefficients set by the two drive amplitudes and Franck–Condon ratios (Xu et al., 2012). Cavity decay then optically pumps the system into $483$0. For $483$1, $483$2, $483$3, and $483$4, numerical fidelities reach $483$5 at $483$6 (Xu et al., 2012). Darkness here is literal cavity emptiness enabled by strong nonlinear single-photon coupling.

A closely related interference mechanism appears in levitated optomechanics. Two nanospheres trapped at distinct sites are indirectly coupled through coherent scattering into a single optical cavity mode. Strong coupling is established because the observed splittings $483$7 exceed the largest measured linewidth $483$8 by more than an order of magnitude (Pontin et al., 12 Feb 2025). The cavity couples to the bright coordinate

$483$9

while the orthogonal combination

$520\ \mathrm{GHz}$0

is dark when the bare mechanical frequencies are degenerate and the coherent-scattering phase is near a conservative value (Pontin et al., 12 Feb 2025). Experimentally the upper branch dims at bare degeneracy in the spectrograms, identifying the dark mode as a weakly read-out normal mode within an otherwise strongly coupled mechanical doublet (Pontin et al., 12 Feb 2025).

5. Excitonic dark states: radiative quenching, brightening, and nonlinear collective effects

In exciton–polariton systems, strong coupling can either suppress radiation or turn a nominally dark material bright. A recent theoretical framework for monolayer excitons coupled to photonic-crystal slabs goes beyond the conventional weighted-linewidth picture and treats polariton radiation as a coherent sum of photonic and excitonic emission channels:

$520\ \mathrm{GHz}$1

so that

$520\ \mathrm{GHz}$2

Destructive interference can therefore drive the radiative linewidth to zero even while the upper and lower polariton branches remain strongly split (Song et al., 28 Aug 2025). In a MoSe$520\ \mathrm{GHz}$3/InGaP photonic-crystal slab, the calculated Rabi splitting at $520\ \mathrm{GHz}$4 is about $520\ \mathrm{GHz}$5, and both branches become symmetry-protected polaritonic bound states in the continuum with radiative linewidths vanishing at $520\ \mathrm{GHz}$6; the total linewidth is then limited by the nonradiative exciton linewidth $520\ \mathrm{GHz}$7 (Song et al., 28 Aug 2025). This is the “bright yet dark” form of dark-strong coupling: a strongly hybridized state that is radiatively quenched.

The converse effect occurs when strong coupling pushes a bright-derived polariton below a dark exciton. In monolayer WSe$520\ \mathrm{GHz}$8, the bright exciton lies at $520\ \mathrm{GHz}$9 and the dark exciton is lower by $a_b$0. Coupling to a dielectric microcavity with $a_b$1 and Rabi splitting $a_b$2 yields a lower polariton near $a_b$3, about $a_b$4 below the dark exciton (Shan et al., 2022). The usual low-temperature photoluminescence quench is then reversed: the control sample decreases by a factor of about $a_b$5 from $a_b$6 to $a_b$7, whereas the cavity-coupled lower-polariton ground state brightens strongly upon cooling (Shan et al., 2022). A similar level-reordering logic was proposed for carbon nanotubes, where the lower polariton must satisfy $a_b$8 to suppress dark-exciton trapping; for a $a_b$9 tube with $a_d$0, zero-detuning inversion requires $a_d$1, while negative detuning reduces the threshold (Shahnazaryan et al., 2017).

Dark states can also modulate strong coupling dynamically. In silver–TIPS-PEN microcavities, dark excitons generated by singlet fission change the effective bright-oscillator population, so the time-dependent collective coupling follows

$a_d$2

Transient polaritonic spectra display Fano-like gain–loss lineshapes, while hot electrons in the mirrors add a Drude loss channel with two-temperature dynamics (Kolesnichenko et al., 2024). Because the polaritons themselves are short-lived, the observed transient signal reflects time-varying strong-coupling conditions rather than long-lived polariton populations (Kolesnichenko et al., 2024).

Strong light–matter coupling can also support genuinely dark nonlinear waves. In a two-component photon–exciton model,

$a_d$3

exact dark–gray and gray–gray solitons bifurcate from the upper polariton branch (Yulin et al., 2022). The half-topological dark–gray solution has a $a_d$4 phase jump in the exciton field while the photon field remains a gray dip, and stability persists below the modulational-instability threshold $a_d$5 (Yulin et al., 2022). Here darkness is encoded in the nonlinear field profile rather than in a nonradiative discrete state.

6. Conceptual distinctions, disputed cases, and extensions beyond photonics

Several works emphasize that not every apparently dark branch is a genuine dark strong-coupling eigenstate. In asymmetric double-quantum-well microcavities, simplified three-level models predicted “dark dipolaritons,” but a microscopic treatment found finite cavity coupling to the indirect exciton, $a_d$6, because finite exciton size, realistic wavefunction overlap, and higher exciton states prevent a strictly dark polariton. At fields where the direct-exciton fraction is minimized, the state remains predominantly photonic with a finite indirect-exciton fraction, not a truly dark mixed state (Sivalertporn et al., 2013). The result is a field-tunable dipolariton rather than dark strong coupling in the strict sense.

A second distinction concerns what is actually measured. In dissipative coupled plasmon–exciton systems, the internal complex eigenfrequencies can be split even when neither measured subsystem shows a resolvable doublet. At resonance, the level-splitting threshold is

$a_d$7

while channel-specific spectral-doublet thresholds are

$a_d$8

When $a_d$9 and $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$00, the authors identify a dark-strong regime: level splitting exists, but both measured spectra remain single-peaked (Liu et al., 2020). In this usage, darkness is not about a subradiant eigenstate but about the invisibility of strong coupling in subsystem-resolved spectroscopy.

The phrase is also used outside light–matter physics. In one cosmological formulation, dark energy couples phenomenologically to dark-matter inhomogeneities through the linear growth function,

$2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$01

so that the effective interaction is negligible early and becomes $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$02 per Hubble time late, with a fiducial viable example $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$03 yielding approximately $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$04 and $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$05 (Marra, 2015). A related running-coupling scenario parameterizes

$2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$06

and current data favor a sign change from negative to positive at around $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$07–$2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$08 at about $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$09 confidence level (Li et al., 2011). In warm-dark-matter freeze-in, “stronger coupling” refers to Higgs-portal couplings much larger than in conventional freeze-in because a low maximal SM temperature suppresses production; there the Lyman-$2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$10 bound excludes masses below about $2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$11–$2g_{\mathrm{eff}}>(\Gamma_i+\Gamma_j)$12 (Feiteira et al., 23 Feb 2026). These cosmological usages are semantically related only by the coexistence of “dark” and “strong coupling,” not by a shared microscopic mechanism.

Across the literature, then, Dark-Strong Coupling is best understood as a domain-specific umbrella expression. In some contexts it means mediated access to a nominally dark mode; in others it denotes dark-manifold control of kinetics, radiative quenching of hybrid modes, strong-coupling-enabled brightening, the reemergence of cavity-dark or mechanically dark states at large coupling, or even strong interactions inside the cosmological dark sector. What unifies these otherwise disparate cases is the failure of the naive expectation that dark states are merely spectators: under sufficiently strong hybridization, they become organizing degrees of freedom of the spectrum, dynamics, or observables.