Cavity Rydberg Electromagnetically Induced Transparency

• Cavity Rydberg EIT is a phenomenon where an intracavity probe and a control laser create dark and bright polariton states through quantum interference in a three-level ladder system.
• It transforms the conventional transparency window into a distinct three-peak polaritonic spectrum, revealing features like avoided crossings and compressed linewidths.
• Rydberg interactions introduce strong photon nonlinearities and blockade effects, enabling advances in quantum optics such as single-photon generation and optical bistability.

Cavity Rydberg electromagnetically induced transparency (EIT) denotes cavity-assisted EIT in which an intracavity probe mode couples the lower transition of a three-level ladder while a classical control field drives the upper transition to a Rydberg state; the cavity then probes the dressed light–matter eigenstates of the atom–cavity–EIT system through its transmission spectrum (Sheng et al., 2017). In this regime, the familiar free-space transparency window is converted into a cavity-polaritonic spectrum containing dark-state and bright-state branches, and experiments with ultracold $^{87}$Rb have observed both a three-peak transmission structure and long-lived cavity Rydberg polaritons with compressed spectra and enhanced lifetimes (Sheng et al., 2017, Ningyuan et al., 2015).

1. Core physical picture and level structure

The canonical implementation is a ladder-type three-level system. In the high-finesse Fabry–Pérot realization, the states are $|1\rangle = 5S_{1/2}(F=2)$, $|2\rangle = 5P_{3/2}(F=3)$, and $|3\rangle$ a Rydberg state such as $35S_{1/2}$; a weak probe field is resonantly coupled into a cavity mode on $|1\rangle \to |2\rangle$, and a classical coupling laser drives $|2\rangle \to |3\rangle$ (Sheng et al., 2017). In the collective cavity-polariton formulation, the corresponding basis states are $|C\rangle$ for one cavity photon, $|E\rangle$ for one symmetric collective excitation in the intermediate state, and $|R\rangle$ for one symmetric collective excitation in the Rydberg state (Ningyuan et al., 2015).

The atom–cavity coupling and control-laser dressing are commonly expressed through the interaction Hamiltonian

$|1\rangle = 5S_{1/2}(F=2)$0

with the appropriate detunings carried by the atomic and field Hamiltonians (Sheng et al., 2017). In the collective three-state description, the essential dark eigenmode is

$|1\rangle = 5S_{1/2}(F=2)$1

so the photonic and matter fractions are governed by the mixing angle $|1\rangle = 5S_{1/2}(F=2)$2 (Ningyuan et al., 2015).

The defining feature of EIT remains destructive quantum interference: population of the lossy intermediate state is suppressed, and the cavity photon hybridizes with a collective Rydberg excitation rather than being absorbed. In a cavity, however, that interference does not merely open a spectral hole in a propagating probe response. It modifies the cavity resonance condition itself, so the experimentally accessible observable is the cavity transmission spectrum of polaritonic eigenmodes rather than a free-space transmission dip (Sheng et al., 2017).

2. Susceptibility, polaritons, and cavity spectra

The cavity response is governed by the atomic susceptibility. In the steady-state treatment of the high-finesse intracavity experiment, the EIT susceptibility is

$|1\rangle = 5S_{1/2}(F=2)$3

and the cavity transmission maxima are approximately determined by

$|1\rangle = 5S_{1/2}(F=2)$4

so the transmission spectrum is effectively a map of the atom-induced phase shift of the intracavity field (Sheng et al., 2017). The real part of $|1\rangle = 5S_{1/2}(F=2)$5 shifts the cavity resonance, whereas the imaginary part reduces transmission.

Without the control field, the atoms behave as a two-level medium and the transmission exhibits the familiar normal-mode splitting, with peaks at approximately

$|1\rangle = 5S_{1/2}(F=2)$6

When the coupling laser is turned on and Rydberg EIT is established, the spectrum acquires a three-peak structure: a narrow central peak associated with the dark-state polariton and two symmetrically spaced side peaks associated with bright-state polaritons (Sheng et al., 2017). Theoretical intracavity EIT analyses recover the same normal-mode structure in polariton language, with eigenenergies

$|1\rangle = 5S_{1/2}(F=2)$7

where $|1\rangle = 5S_{1/2}(F=2)$8 is the dark-state polariton and $|1\rangle = 5S_{1/2}(F=2)$9 are bright polaritons containing the lossy intermediate-state component (Lin et al., 2013).

In cavity Rydberg polariton experiments, the dark resonance is also characterized by frequency pulling and linewidth narrowing. Near EIT resonance,

$|2\rangle = 5P_{3/2}(F=3)$0

and

$|2\rangle = 5P_{3/2}(F=3)$1

The observed slope of the dark-polariton frequency with respect to cavity detuning,

$|2\rangle = 5P_{3/2}(F=3)$2

expresses the “compressed frequency spectrum” of the dark polariton, and the generalized cavity-EIT linewidth is given to lowest order by

$|2\rangle = 5P_{3/2}(F=3)$3

(Ningyuan et al., 2015). This compressed spectrum is a distinctive cavity consequence of EIT: the dark resonance is pinned near the two-photon condition and responds more slowly than the bare cavity mode.

3. Experimental realizations and spectroscopic signatures

Several complementary cavity platforms have established the basic phenomenology of cavity Rydberg EIT.

Paper	Platform	Key observation
(Sheng et al., 2017)	High-finesse Fabry–Pérot cavity with ultracold $|2\rangle = 5P_{3/2}(F=3)$4Rb	Three-peak transmission, anti-crossing, lower-bound coherence time $|2\rangle = 5P_{3/2}(F=3)$5
(Ningyuan et al., 2015)	Running-wave bow-tie cavity with $|2\rangle = 5P_{3/2}(F=3)$6Rb	Dark-state Rydberg polaritons, compressed spectrum, minimum inverse lifetime $|2\rangle = 5P_{3/2}(F=3)$7
(Boddeda et al., 2015)	66 mm vertical Fabry–Perot cavity with cold $|2\rangle = 5P_{3/2}(F=3)$8Rb	Cavity-enhanced nonlinear response for $|2\rangle = 5P_{3/2}(F=3)$9 and $|3\rangle$0 Rydberg states

The high-finesse Fabry–Pérot experiment used cavity mirrors separated by $|3\rangle$1, radii of curvature $|3\rangle$2 mm, finesse $|3\rangle$3, and a mode waist of about $|3\rangle$4, so only a few tens of atoms participated in the interaction region. The relevant CQED parameters were $|3\rangle$5 MHz with cooperativity $|3\rangle$6, and the three-peak cavity-EIT structure was demonstrated for the $|3\rangle$7 Rydberg state with about $|3\rangle$8 atoms in the interaction region (Sheng et al., 2017).

The bow-tie cavity experiment employed a smaller waist, about $|3\rangle$9 $35S_{1/2}$0 radii, finesse about 2500, and empty-cavity linewidth about $35S_{1/2}$1 MHz. It directly resolved a narrow dark-state peak and two broader bright-polariton peaks, and it also demonstrated two-mode spectral compression: two nearly degenerate cavity modes separated by 20 MHz were each converted into dark polaritons whose splitting was only about $35S_{1/2}$2 MHz in one representative measurement (Ningyuan et al., 2015).

The longer Fabry–Perot cavity experiment on nonlinear response used a $35S_{1/2}$3 mm vertical cavity with finesse $35S_{1/2}$4, cavity linewidth $35S_{1/2}$5 MHz, and waist $35S_{1/2}$6. In the linear regime, the spectra showed an empty-cavity Lorentzian, normal-mode splitting with atoms and no control light, and then a transparency window when the control field was turned on at the two-photon EIT condition (Boddeda et al., 2015).

Anti-crossing is among the clearest coherence signatures in these systems. In the high-finesse experiment, the three peaks showed avoided crossings as cavity detuning $35S_{1/2}$7 was varied, and the side peaks also showed anti-crossing as the coupling-laser detuning $35S_{1/2}$8 was changed. The central dark-state peak remained comparatively flat because its resonance was pinned by the steep EIT dispersion (Sheng et al., 2017). Even under experimentally imperfect conditions, EIT remained observable: the optical dipole trap imposed an AC Stark shift of about $35S_{1/2}$9 MHz, yet the cloud was cold enough that the inhomogeneous broadening did not destroy the EIT dispersion, and a separate lambda-type EIT measurement gave an EIT linewidth of about $|1\rangle \to |2\rangle$0 MHz, limited mostly by laser linewidth (Sheng et al., 2017).

4. Rydberg blockade, optical nonlinearities, and few-photon physics

The specifically Rydberg contribution to cavity EIT is the interaction-induced nonlinearity. In free-space Rydberg EIT theory, the blockade criterion is

$|1\rangle \to |2\rangle$1

which defines the blockade radius

$|1\rangle \to |2\rangle$2

The medium can then be coarse-grained into superatoms, each blockade volume supporting at most one collective Rydberg excitation (Petrosyan et al., 2011). This superatom logic is the microscopic origin of cavity-level photon nonlinearity: once one polariton occupies a blockade region, subsequent polaritons are shifted or suppressed.

In intracavity EIT with a strongly interacting Rydberg ensemble, the dark-state polariton inherits both long lifetime and interaction strength. A central theoretical reduction gives the effective dark-polariton self-interaction

$|1\rangle \to |2\rangle$3

while the corresponding decay rate is

$|1\rangle \to |2\rangle$4

By tuning the control field so that $|1\rangle \to |2\rangle$5 becomes small, the dark-state polariton becomes more atom-like, $|1\rangle \to |2\rangle$6 decreases, and $|1\rangle \to |2\rangle$7 increases, enabling the strong-blockade condition $|1\rangle \to |2\rangle$8 (Lin et al., 2013). Under representative parameters $|1\rangle \to |2\rangle$9, $|2\rangle \to |3\rangle$0 MHz, $|2\rangle \to |3\rangle$1 MHz, $|2\rangle \to |3\rangle$2 MHz, and $|2\rangle \to |3\rangle$3, the theory gives $|2\rangle \to |3\rangle$4 MHz and $|2\rangle \to |3\rangle$5 MHz, with a predicted $|2\rangle \to |3\rangle$6 (Lin et al., 2013).

Experimental cavity-enhanced nonlinearities have been resolved in both $|2\rangle \to |3\rangle$7- and $|2\rangle \to |3\rangle$8-state excitation. For $|2\rangle \to |3\rangle$9 states, the dominant interaction is the isotropic van der Waals potential

$|C\rangle$0

and the measured reduction of transparency with increasing probe photon rate is well described by a semi-classical mean-field model using no free parameters once $|C\rangle$1, $|C\rangle$2, $|C\rangle$3, and $|C\rangle$4 are fixed from the linear EIT fit (Boddeda et al., 2015). For $|C\rangle$5 states, steady-state blockade alone was insufficient: the transmission showed a time-dependent decay on a timescale of about $|C\rangle$6s, and the data required a phenomenological Rydberg bubble model in which the bright Rydberg collective state $|C\rangle$7 decays into a long-lived dark Rydberg state $|C\rangle$8 with an additional nonlinear decay term characterized by $|C\rangle$9 (Boddeda et al., 2015).

Beyond mean-field transmission, non-equilibrium quantum-field methods show that intracavity Rydberg blockade produces both elastic and inelastic output. In the Schwinger–Keldysh treatment, the fourth-order transmitted spectrum contains an elastic correction at the probe frequency and an inelastic component with resonance frequencies

$|E\rangle$0

set by the three polariton eigenmodes of the single-excitation Hamiltonian (Grankin et al., 2016). In the fully resonant case only three resonances appear because two polariton branches are degenerate. This identifies a polaritonic resonance structure in the inelastic spectrum that earlier low-excitation or ad hoc blockade models had not reported (Grankin et al., 2016).

5. Coherence limits, inhomogeneous broadening, and stray fields

Cavity Rydberg EIT is unusually sensitive to environmental perturbations because Rydberg levels are highly polarizable. In the high-finesse Fabry–Pérot experiment, Stark spectroscopy of $|E\rangle$1 and $|E\rangle$2 states revealed an electric field of $|E\rangle$3 V/cm near the cavity center and a magnetic field on the order of $|E\rangle$4 G. The electric field was not attributed to the PZTs or heater, since changing heater current or PZT voltage did not alter it; instead, the field was attributed to rubidium adsorbates on the mirror surfaces. Misaligning the blue coupling beam so that it scattered from a mirror changed the field, and UV LED illumination reduced it by up to $|E\rangle$5 V/cm (Sheng et al., 2017). These observations made mirror adsorbates a concrete cavity-specific decoherence channel rather than a generic laboratory nuisance.

The same experiment extracted a lower bound on the coherence time of $|E\rangle$6 from a narrow cavity-EIT transmission peak with linewidth $|E\rangle$7 kHz after isolating a specific $|E\rangle$8 transition with an applied magnetic field. The paper explicitly states that this is a lower bound because the observed width is dominated by probe and coupling laser linewidths: the convolution of the laser linewidths derived from the locking signals was about $|E\rangle$9 kHz, essentially matching the measured width. AC Stark broadening in the trap was of similar size, so improved laser locking and turning off the dipole trap during measurement were identified as direct routes to improvement (Sheng et al., 2017).

A distinct result from cavity Rydberg polariton spectroscopy is that collective coupling suppresses losses due to inhomogeneous broadening. Rather than a linear sensitivity to the spread of local detunings, the effective Rydberg loss scales quadratically,

$|R\rangle$0

because inhomogeneity couples the intended dark collective state to a bath of orthogonal collective excitations detuned by approximately $|R\rangle$1, making the loss off-resonant when $|R\rangle$2 is sufficiently large (Ningyuan et al., 2015). The experiments verified this by varying the principal quantum number $|R\rangle$3 over $|R\rangle$4: the loss increased with polarizability, but with a quadratic dependence rather than a linear one (Ningyuan et al., 2015). Even control-field inhomogeneity did not significantly broaden the dark state, because a dark state still exists with a modified collective Rydberg wavefunction (Ningyuan et al., 2015).

6. Extensions, bistability, and applications

The principal applications identified for cavity Rydberg EIT are single-photon generation using the Rydberg blockade effect, the study of many-body physics, and the generation of novel quantum states (Sheng et al., 2017). In the bow-tie cavity realization, the broader program was framed as the creation of photonic quantum materials: cavity photons inherit atomic longevity, atomic excitations inherit cavity mode structure, and higher Rydberg levels were identified as a route to stronger photon–photon interactions, multimode condensed-matter analogs, and correlated many-body phases (Ningyuan et al., 2015).

One direct cavity extension is optical bistability. In a unidirectional optical ring cavity containing a cascade three-level Rydberg EIT medium, van der Waals interactions substantially enhance both nonlinear dispersion and nonlinear absorption. Under two-photon resonance, the probe one-photon detuning changes the phase of the third-order nonlinear coefficient $|R\rangle$5, thereby tuning the balance between dispersive and absorptive bistability (Chuang et al., 2018). The cavity boundary conditions then produce hysteresis in the input–output curve, and the response obeys explicit scaling relations with coupling Rabi frequency and atomic density, including $|R\rangle$6 and an effective $|R\rangle$7 dependence of the nonlinear response (Chuang et al., 2018). This identifies cavity Rydberg EIT as a low-light-level nonlinear optical feedback system rather than only a spectroscopic platform.

Another extension is the move beyond three-level ladders. A four-level ladder theory for Rydberg-EIT and Rydberg-EIA in cold Cs and Rb atomic ensembles gives a self-consistent mean-field algorithm based on the interaction-induced shift

$|R\rangle$8

and an explicit probe susceptibility

$|R\rangle$9

with the three-photon resonance condition

$|1\rangle = 5S_{1/2}(F=2)$00

(Oyun et al., 2021). Because that work is not cavity-based, a direct cavity claim would be inferential. A plausible implication is that such interaction-shifted susceptibilities can be inserted into cavity transmission or polariton models to study cavity-modified EIT–EIA crossovers, nonlinear cavity transmission, or bistability (Oyun et al., 2021).

Taken together, these results establish cavity Rydberg EIT as a regime in which cavity QED, coherent interference, and Rydberg interactions are simultaneously operative. The cavity converts atomic susceptibility into a directly measurable polaritonic spectrum; the Rydberg component adds blockade, strong nonlinearity, and many-body sensitivity; and the resulting platform supports linear dark-state spectroscopy, anti-crossing and spectral compression, few-photon blockade, inelastic scattering, and cavity-feedback phenomena within a single framework (Sheng et al., 2017, Ningyuan et al., 2015).