Controlling Rydberg atom-polariton interactions: from exceptional points to fast readout

TL;DR

This work experimentally dissects the dipolar interactions between propagating Rydberg polaritons and a neighboring stationary Rydberg atom, revealing three distinct dynamical regimes—polariton blockade, coherent atom–polariton exchange, and probabilistic hopping—separated by a transition through an exceptional point. By tuning inter-ensemble distance and EIT control, the authors map the regimes via polariton transmission contrasts and provide both ab-initio and phenomenological descriptions of the underlying non-Hermitian physics. They demonstrate fast, remote, and non-destructive readout of a Rydberg qubit and lay out building blocks for nonlinear photonic networks using atomic-ensemble arrays, including all-optical switching and network connectivity. The findings offer a versatile platform for rapid quantum-state readout, controlled photon routing, and potential implementations of photonic gates and quantum walks, with a clear path toward higher-fidelity detectors and scalable neutral-atom quantum processors.

Abstract

Rydberg atoms represent a platform underpinning many recent developments in quantum computation, simulation, sensing, and metrology. They further facilitate optical nonlinearity at the single-photon level when coupled to photons propagating in atomic clouds, which form collective atomic excitations called Rydberg polaritons, strongly interacting with each other. Here, we experimentally explore interactions between a Rydberg polariton in an atomic ensemble and a single, adjacent, Rydberg atom. We discover three different regimes of quantum dynamics corresponding to polariton blockade, coherent exchange, and probabilistic hopping, which are defined by their distinct transmission characteristics, with a transition through an exceptional point occurring between blockade and coherent exchange. We investigate the applications of such interactions for fast, non-destructive detection of Rydberg atoms and present proof-of-principle demonstrations for their potential application in nonlinear photonic networks.

Figures (8)

• Figure 1: Experimental setup and level scheme: a. Three-photon scheme for preparing a single Rydberg atom in the state $\left| n'P \right\rangle = \left| 75\textrm{P}_{3/2}, m_J = 3/2 \right\rangle$. The first and third legs of the three-photon transition ($780$ nm and microwave fields with Rabi frequencies $\Omega_{1}$, $\Omega_{3}$) are detuned by $\delta = +100\,$MHz from resonance of their respective transitions, while the second leg ($480$ nm, $\Omega_{2}$) is on-resonance. b. Rydberg EIT scheme couples ground-state atoms to the Rydberg state $\left| nD \right\rangle =\left| 74\textrm{D}_{5/2}, m_J = 5/2 \right\rangle$. Both the probe and control fields ($\Omega_{p}$, $\Omega_{c}$) are on-resonance with their respective transitions. c. Schematic of the setup illustrating the trap configuration, which loads small atomic ensembles using a combination of red-detuned $808\,\textrm{nm}$ SLM traps (red) and blue-detuned $770\,\textrm{nm}$ AOD sheets (purple). A site-selective $480\,\textrm{nm}$ control field array (blue) co-propagates with the SLM trap light, while the global $780$ nm probe field counter-propagates (yellow). Inset: Dipole-dipole interactions $\textrm{V}_\textrm{PD}$ between a stationary Rydberg atom in $\left| n'P \right\rangle$ and traveling Rydberg polaritons in $\left| nD \right\rangle$ lead to modulation of probe EIT transmission.
• Figure 2: Classifying Rydberg atom-polariton interaction regimes: a. Detection contrast (see main text) of probe photon transmission through the detection ensemble (blue), for a single Rydberg atom prepared in $\left| n'P \right\rangle$ within the nearby preparation ensemble (orange), as a function of the dipolar interaction strength V$_\textrm{PD}$ and inter-ensemble distance. After preparation of the Rydberg atom, the detection ensemble is probed for 10 $\mu$s, much longer than the polariton lifetime $\tau_d \approx 60\,\textrm{ns}$. Three distinct regimes, corresponding to probabilistic hopping, coherent exchange, and polariton blockade, can be identified. The vertical lines represent the interaction strengths at the hopping radius $r_h\approx 23~\mu$m and the blockade radius $r_b\approx 18~\mu$m for the EIT control Rabi frequency of $\Omega_{c}/(2\pi) =5.5\, \textrm{MHz}$, which was chosen to allow observation of all three regimes. A phenomenological theory curve (solid black line; see SM SM) closely matches the observed data. b. Ab-initio calculation of two polariton energy eigenvalues of the Hamiltonian describing the system (see Eq. \ref{['Eq:compact']} in SM SM). We plot the real (solid) and imaginary (dashed) parts of the eigenvalues (in units of the intermediate state linewidth $\Gamma$) as a function of the dipole-dipole interaction strength, $V_\textrm{PD}$. An exceptional point clearly emerges at the interaction strength corresponding to the blockade radius (gray line). At high interaction strengths (right-hand side of the gray line), the polariton eigenvalues are imaginary (dissipative). At low interaction strengths (left-hand side of the gray line), the polariton eigenvalues have the same imaginary part (i.e., dissipation still occurs), but have a non-degenerate real part, with the same magnitude and an opposite sign (signifying the dressing of the polariton state due to the interaction with the atom).
• Figure 3: Rydberg EIT as a remote detector of single Rydberg atoms: a. Single-shot histogram for 30 $\mu$s of readout, showing the difference in detected photons exiting the detector ensemble with (green) and without (purple) a Rydberg atom nearby. The measurement fidelity is $47(5)\,\%$. b. Using repeated preparation and detection (20 repetitions of $10\,\mu$s detection), we achieve a fidelity of $79(4)\,\%$. c. Following a global microwave drive, state-selective and remote detection of the Rydberg qubit is performed, showing long-lived coherent microwave Rabi oscillations.
• Figure 4: Hopping and blockade in a nonlinear photonic network: a. Detection contrast (in photons) in both ensembles following preparation of a Rydberg atom in the right ensemble, with $11.7\,\mu\textrm{m}$ of inter-ensemble distance ($V_\textrm{PD}/h = 14.8\,\textrm{MHz}$). Simultaneous readout of both ensembles is performed for two EIT control fields, $\Omega_c/(2\pi) = 23\,\textrm{MHz}$ ($r_b \approx 9\,\mu$m; green) and $\Omega_c/(2\pi) = 8\,\textrm{MHz}$ ($r_b \approx 13~\mu$m; purple). When the distance between ensembles is smaller than the blockade radius (lower purple panel), the probe absorption is the same for both ensembles, which indicates that the left ensemble is blocked by the presence of a Rydberg atom in the right ensemble. On the other hand, when the distance is larger than the blockade radius (upper green panel), the probe absorption is different, and polaritons are allowed to propagate in the left ensemble. b. Detection contrast in the middle ensemble, which reflects the local $\left| n'P \right\rangle$ Rydberg population. The initial population (green) decreases due to the presence of polaritons in the neighboring detection ensembles (purple), which indicates exchange between the prepared Rydberg atom and neighboring Rydberg polaritons. Results are normalized according to the initial $\left| n'P \right\rangle$ Rydberg population. Note that the large control field used for these measurements ($\Omega_{c}/(2\pi) \approx 20\, \textrm{MHz}$) places the blockade radius at $r_b \approx 9\,\mu$m and the hopping radius at $r_h < 14\,\mu$m. c. Detection contrast (normalized) in the leftmost ensemble as a function of distance (green) increases in the presence (purple) of an intermediate ensemble, which mediates exchange between the Rydberg atom in the rightmost ensemble to the leftmost ensemble. The blue shaded regions denote ensembles where we measure and plot the absorption of probe EIT light, which is proportional to the Rydberg population within the ensemble. Other than the geometry, experimental parameters are the same as in b.
• Figure S1: DP vs SP: An example showcasing the purity of exchange interactions for $D-P$ states in comparison to $S-P$ states (calculated with "PairInteraction" Weber2017). The states are chosen so that both the D and S states share the same Rydberg $n$ level, while the P states in each case are selected to maximize interaction strength.