Dual-Rydberg Platform in Quantum Architectures

• Dual-Rydberg platform is a hybrid quantum architecture that integrates two Rydberg-active channels to enable functionalities absent in single-mode systems.
• It leverages coupled excitonic, photonic, and atomic interactions to facilitate advanced sensing, transduction, and quantum information processing.
• Experimental implementations range from Cu2O microstructures to dual-species neutral-atom arrays, demonstrating enhanced control and reduced crosstalk through dual-channel design.

A dual-Rydberg platform is a recurring architectural pattern in which two Rydberg-enabled subsystems are combined within a single experimental device. In current usage, the expression has been applied to integrated solid-state systems that host both Rydberg excitons and quantum-degenerate excitons, to hybrid atom–cavity systems with two bosonic modes coupled by a Rydberg superatom, to dual-ladder and dual-band receivers for radio-frequency sensing, to dual-species neutral-atom arrays for entanglement and quantum error correction, and to hybrid cavity, polaritonic, and ancilla-assisted readout architectures (Steinhauer et al., 2019, Song et al., 19 Nov 2025, Zhang et al., 2024, Oliver et al., 27 Feb 2026, Berweger et al., 2024, Elgee et al., 2023, Anand et al., 2024, Wang et al., 11 May 2026, White et al., 22 Jan 2026, Santis et al., 12 Feb 2026, Šumarac et al., 9 Jan 2026, Tebben et al., 2021, Machu et al., 29 Sep 2025). This suggests that the term functions less as the name of a single standardized hardware stack than as a family label for platforms in which two Rydberg-capable channels, species, modes, or manifolds are deliberately combined to realize functionality unavailable in single-channel settings.

1. Scope and principal meanings

The literature uses the term across several experimentally and theoretically distinct settings. What is common is the simultaneous use of two Rydberg-active components with either shared or programmable coupling: two excitonic manifolds in cuprous oxide, two cavity modes coupled by one Rydberg superatom, two optical ladders to the same Rydberg level, two RF-chip receiver modules, two atomic species, or one logical Rydberg subsystem plus an ancillary low-$\ell$ Rydberg subsystem.

Archetype	Dual element	Representative function
Cu$_2$O on silicon	Rydberg excitons + quantum-degenerate $1s$ excitons	Integrated excitonic quantum optics and BEC-proximate exciton gases
Hybrid atom–cavity transducer	Microwave mode + optical mode	Topological single-photon microwave–optical conversion
RF sensing platform	Two ladders, modules, or microwaves	Independent, dual-band, or polarization-resolved sensing
Dual-species neutral-atom array	Two atomic species or two Rydberg states	Crosstalk-free control, entanglement, QEC, QCA
Polaritonic or cavity hybrid	Rydberg atom + polariton or cavity mode	Nonlocal optical control, exceptional-point dynamics, photonic interfaces
Circular-ancilla hybrid	Circular logical atom + low-$\ell$ ancilla	QND readout and local manipulation

A common misconception is that a dual-Rydberg platform must be a dual-species neutral-atom array. The published record is broader. In (Steinhauer et al., 2019), the “dual” aspect is the coexistence of highly excited yellow $np$ Rydberg excitons and dense quantum-degenerate $1s$ exciton gases in Cu$_2$O microcrystals on silicon. In (Song et al., 19 Nov 2025), it denotes a dual-mode Jaynes–Cummings architecture in which a blocked Rydberg ensemble mediates coherent exchange between a microwave resonator and an optical cavity. In (Berweger et al., 2024) and (Oliver et al., 27 Feb 2026), the duality lies in two independent optical ladders to the same Rydberg state, enabling simultaneous and independent measurements or direct baseband $I/Q$ readout. In (Anand et al., 2024, Wang et al., 11 May 2026), and (White et al., 22 Jan 2026), the term refers to dual-species arrays whose Rydberg interactions coexist with intrinsic spectral selectivity.

2. Physical architectures

In hybrid atom–cavity transduction, a cold atomic ensemble confined in a microtrap forms a Rydberg superatom under blockade, with excitation number $\le 1$. The ensemble couples simultaneously to a superconducting coplanar waveguide microwave resonator $b$ and to an optical cavity $_2$0, while two counterpropagating lasers at $_2$1 nm and $_2$2 nm dress the atoms to produce an effective dual-mode Jaynes–Cummings interaction. The relevant collective two-level subspace is $_2$3, with $_2$4 and $_2$5 (Song et al., 19 Nov 2025).

In dual-band RF reception, the architecture consists of two independent RF-chip–integrated Rydberg-atom receiver modules operated simultaneously and in parallel via space-division multiplexing. Each module combines a spoof-surface plasmon polariton RF metawaveguide chip with a centimeter-sized rubidium vapor cell, with co-linear counter-propagating probe and coupling beams traversing the cell approximately $_2$6 mm above the chip surface. The shared RF signal is routed through both modules, while free-space local oscillators are delivered independently to each module so that one module handles higher-frequency bands and the other lower-frequency bands (Zhang et al., 2024).

In dual-ladder homodyne reception, two spatially overlapped, independent EIT systems are created in a room-temperature $_2$7Rb vapor cell by frequency-shifting and orthogonally polarizing the probe and pump beams. Arm 1 uses a $_2$8 nm probe and a $_2$9 nm pump, while arm 2 is Doppler-class shifted by $1s$0 MHz on the probe and $1s$1 MHz on the pump. Two horn antennas provide local oscillators that are $1s$2 out of phase, so the two arms yield simultaneous baseband $1s$3 and $1s$4 components and, through polarization sensitivity, angle-of-arrival information (Oliver et al., 27 Feb 2026).

In dual-species neutral-atom hardware, two interleaved species-specific tweezer systems are used. The Rb–Cs array in (Anand et al., 2024) employs $1s$5 nm tweezers for $1s$6Rb and $1s$7 nm tweezers for $1s$8Cs, with interacting pairs placed at $1s$9–$\ell$0 and a segmented in-vacuum Faraday cage for electric-field control. The Na–Cs stabilizer platform in (Wang et al., 11 May 2026) uses co-localized two-dimensional arrays with $\ell$1 nm Cs tweezers and $\ell$2 nm Na tweezers, organized into plaquettes consisting of one central Na ancilla and four surrounding Cs data qubits on a square of side length $\ell$3. The dual-species QCA processor in (White et al., 22 Jan 2026) uses interleaved $\ell$4Rb and $\ell$5Cs chains with nearest-neighbor spacing $\ell$6, species-specific trap wavelengths $\ell$7 nm and $\ell$8 nm, and independent global Rydberg control on each species.

The solid-state implementation is structurally different but conceptually parallel. Cu films of thickness $\ell$9 nm are deposited on silicon with a $np$0 nm thermal SiO$np$1 layer and a $np$2 nm Ti adhesion layer, then thermally oxidized at $np$3–$np$4 for $np$5 h or $np$6 h in synthetic air at $np$7 mbar. The resulting Cu$np$8O microcrystals are phase-pure, strongly $np$9-textured, compatible with lithographic patterning, and support both yellow $1s$0 Rydberg excitons and quantum-degenerate $1s$1 exciton gases in the same silicon-compatible material platform (Steinhauer et al., 2019).

3. Interaction models and control laws

Across these implementations, the defining structure is not merely the presence of Rydberg states but the existence of two coupled sectors with independently controllable amplitudes, detunings, or geometries. In the hybrid transducer, the on-resonance reduced Hamiltonian is

$1s$2

with collectively enhanced couplings $1s$3 and $1s$4. Conservation of total excitation number maps the dynamics to a Fock-state lattice whose hoppings are photon-number dependent,

$1s$5

and the system supports an exact zero-energy dark mode with support only on odd sites. Its topological invariant is continuous rather than quantized in SSH fashion,

$1s$6

so the zero-mode center moves smoothly across the synthetic lattice as the coupling ratio is tuned (Song et al., 19 Nov 2025).

In dual-species ladders for many-body physics, the minimal model is

$1s$7

with

$1s$8

In the specific Cs–Rb ladder of (Liu et al., 27 Apr 2026), $1s$9, so the ladder exhibits competing interaction magnitudes and signs: very strong $_2$0, moderate $_2$1, and weaker attractive $_2$2. This interaction hierarchy produces disordered phases, $_2$3, $_2$4, and $_2$5 ordered phases, floating phases with algebraically decaying correlations, and a multicritical point where Ising, chiral, and first-order lines intersect (Liu et al., 27 Apr 2026).

In dual-species processors and QEC devices, the interactions are typically expressed as blockade or resonant exchange. For the interleaved Rb–Cs QCA chain, the measured coefficient is $_2$6, giving $_2$7 MHz at $_2$8, with $_2$9 and $I/Q$0 (White et al., 22 Jan 2026). In the Rb–Cs Förster-tuned array, the near-degenerate pair states $I/Q$1 and $I/Q$2 yield $I/Q$3 and an effective blockade scale $I/Q$4 MHz at $I/Q$5 (Anand et al., 2024). In the Na–Cs stabilizer scheme, the central control issue is the finite Na–Cs interaction strength, which is compensated by choosing

$I/Q$6

so that the differential geometric phase equals $I/Q$7; the shortest useful choice is $I/Q$8, giving $I/Q$9 and $\le 1$0 (Wang et al., 11 May 2026).

The same two-sector logic also appears in polaritonic and atom–polaritons hybrids. In the stationary-light proposal, the dark-state polariton operator is

$\le 1$1

and under $\le 1$2 it acquires the quadratic dispersion

$\le 1$3

In the atom–polaritons experiment, a single adjacent Rydberg atom and a propagating Rydberg polariton are coupled by

$\le 1$4

with $\le 1$5 for the $\le 1$6–$\le 1$7 channel at $\le 1$8 (Tebben et al., 2021, Šumarac et al., 9 Jan 2026).

4. Sensing, reception, and transduction

One major branch of dual-Rydberg research uses dual architectures to separate sensing channels while retaining atomic calibration. The space-division-multiplexed dual-band receiver of (Zhang et al., 2024) demonstrated simultaneous reception around $\le 1$9 MHz and $b$0 GHz, giving $b$1 MHz, $b$2 GHz, and $b$3 octaves. The same platform reported peak sensitivity $b$4 at $b$5 GHz, off-resonant sensitivities $b$6 to $b$7 across $b$8–$b$9 GHz, a total dynamic range of $_2$00 dB with a linear region of $_2$01 dB at $_2$02 Hz RBW, and an instantaneous bandwidth of $_2$03 kHz. The demonstrated concurrent demodulation of AM at $_2$04 MHz and FM at $_2$05 GHz illustrates the utility of dual modules for wideband parallel reception (Zhang et al., 2024).

The dual-ladder receiver of (Oliver et al., 27 Feb 2026) addresses a different constraint: a conventional heterodyne Rydberg mixer yields an intermediate frequency that must be filtered and mixed down, whereas the dual-ladder homodyne platform directly produces baseband $_2$06 and $_2$07. The receiver uses the $_2$08 transition at $_2$09 GHz, two overlapped orthogonally polarized ladders, and local oscillators that are $_2$10 out of phase. It demonstrated simultaneous direct baseband detection of 16QAM, APSK, and QPSK, angle-of-arrival retrieval with average EVM $_2$11, and relative phase estimation with average error $_2$12. The key trade-off is that the dual ladder is not limited by the heterodyne detuning requirement that constrains the conventional receiver’s maximum symbol rate, but it is more sensitive to low-frequency multiplicative $_2$13 noise (Oliver et al., 27 Feb 2026).

A closely related but conceptually distinct implementation is the independent dual-ladder scheme of (Berweger et al., 2024). There, two distinct optical pathways prepare and read out the same target Rydberg level $_2$14 via different intermediate hyperfine states and separate optical fields. The RF field at $_2$15 GHz couples $_2$16, and the two ladders are cross-polarized and frequency-shifted by $_2$17–$_2$18 MHz. The result is simultaneous, independent polarization-resolved sensing at the same spatial location, rather than dual-band reception.

The dual-optical dual-microwave UHF sensor pushes the same logic into a lower-frequency range by adding an additional microwave photon to access higher-$_2$19 Rydberg states. In the reported $_2$20Rb implementation, $_2$21 optical EIT is combined with a $_2$22 GHz dressing field on $_2$23 and a $_2$24 GHz signal/LO pair on $_2$25. The resonant sensitivity at $_2$26 GHz was $_2$27, fifty times better than the measured far off-resonant scheme at the same frequency, and the system reproduced the band structure of Sirius XM satellite radio received outside the laboratory (Elgee et al., 2023).

The most explicitly transduction-oriented dual-Rydberg platform is the hybrid Rydberg superatom transducer. By adiabatically modulating

$_2$28

the zero mode moves across the Fock-state lattice while the instantaneous spectrum remains $_2$29 and $_2$30. With $_2$31 MHz, $_2$32 kHz, $_2$33 kHz, and $_2$34 kHz, the protocol achieved near-unity transduction for Fock inputs; for $_2$35, $_2$36 gave $_2$37 in the ideal Hamiltonian and overall fidelity $_2$38 for coherent and squeezed-vacuum inputs under realistic dissipation (Song et al., 19 Nov 2025).

5. Quantum information processing and many-body dynamics

In neutral-atom quantum information, the dual-Rydberg concept is especially closely linked to crosstalk-free ancilla–data separation. The dual-species Rb–Cs array of (Anand et al., 2024) combines intrinsic spectral addressability with electrically tuned heteronuclear Förster resonances. The platform demonstrated interspecies Rydberg blockade, quantum state transfer from one species to another, a hyperfine controlled-phase gate, Bell-state generation with SPAM-corrected fidelity $_2$39, and quantum non-demolition measurement of an Rb qubit using an auxiliary Cs qubit. The measured midcircuit readout discrimination fidelity for Cs was $_2$40, the QND fidelity was $_2$41, and the QND-ness was $_2$42 (Anand et al., 2024).

The dual-species Na–Cs array of (Wang et al., 11 May 2026) is oriented toward stabilizer extraction. One central Na ancilla is surrounded by four Cs data qubits, and a single global Rydberg pulse sequence measures a weight-four $_2$43 stabilizer without local addressing. The essential correction is compensation of the finite Na–Cs interaction strength by tuning the data-pulse Rabi frequency and detuning so that the differential geometric phase equals $_2$44. The compensated protocol yielded non-destructive measurement of Pauli-$_2$45 stabilizers on four-qubit Cs plaquettes with per-qubit contrast or fidelity $_2$46, demonstrating simultaneous, non-destructive, in situ stabilizer readout via global pulses alone (Wang et al., 11 May 2026).

The QCA processor of (White et al., 22 Jan 2026) uses the same species-separation principle to replace local control by staggered global operations. Independent global Rydberg excitation of interleaved Rb and Cs chains, together with strong interspecies blockade, implements Floquet quantum cellular automata capable of generating GHZ states, Bell states with $_2$47 fidelity, 17-qubit cluster states, and high-connectivity graph states. The mediated-gate layer effectively applies a full nearest-neighbor CZ layer across the chain,

$_2$48

which is why the architecture is naturally suited to measurement-based quantum computing as well as to constrained many-body dynamics (White et al., 22 Jan 2026).

The many-body role of duality is not limited to quantum computing. In the dual-species ladder studied in (Liu et al., 27 Apr 2026), the coexistence of strong $_2$49, moderate $_2$50, and attractive $_2$51 interactions produces phenomena absent in single-species chains: a smooth crossover between distinct $_2$52-ordered regimes, floating phases with incommensurate wave vectors and algebraically decaying correlations, and a multicritical point where Ising, chiral, and first-order lines intersect. In the nonequilibrium Cs–Rb chains of (Soto-Garcia et al., 14 Apr 2026), competition between intra-species repulsion and inter-species attraction induces real-space fragmentation, with frozen regions acting as emergent barriers that protect localized oscillatory sectors. The reported revival periods were $_2$53 for a single active Rb and $_2$54 for an active Rb pair, with the latter consistent with $_2$55 (Liu et al., 27 Apr 2026, Soto-Garcia et al., 14 Apr 2026).

These results clarify a second misconception: dual-Rydberg architectures are not merely engineering conveniences for multiplexed control. In several cases, the duality itself changes the phase structure, dynamical constraints, and available gate constructions. This suggests that the dual degree of freedom is often a physical resource rather than a peripheral implementation detail.

6. Solid-state, cavity, polaritonic, and ancilla-assisted extensions

The solid-state Cu$_2$56O system of (Steinhauer et al., 2019) is the clearest non-atomic example. Cu$_2$57O has a large exciton binding energy, with the effective Rydberg for the yellow series measured at $_2$58 meV in the reported microcrystals, a Bohr radius $_2$59 nm, and a well-defined hydrogenic yellow $_2$60 series obeying $_2$61 with negligible quantum defect for $_2$62-states. Photoluminescence resolved yellow $_2$63 states up to $_2$64, while the phonon-assisted $_2$65 ortho- and paraexciton gases were fit by Bose–Einstein distributions with $_2$66 at $_2$67 K for orthoexcitons and $_2$68 at $_2$69 K for paraexcitons, demonstrating proximity to the Bose–Einstein condensation boundary. The same on-chip microstructures exhibited nearly linear excitation-power dependence of total exciton luminescence, with slope $_2$70, and lithographically defined $_2$71 circular microstructures on silicon (Steinhauer et al., 2019).

In free-space atom–photon hybrids, (Šumarac et al., 9 Jan 2026) experimentally coupled a single adjacent Rydberg atom to an atomic ensemble supporting propagating Rydberg polaritons. The interaction is governed by $_2$72, and three regimes were identified: polariton blockade, coherent exchange, and probabilistic hopping. The blockade radius is defined by $_2$73, the hopping radius by $_2$74, and the transition between blockade and coherent exchange occurs at an exceptional point where two non-Hermitian eigenvalues coalesce. The same platform achieved combined preparation-plus-readout fidelity $_2$75 in a $_2$76 single-shot remote readout, increasing to $_2$77 with 20 repetitions of $_2$78 each, and measured $_2$79 on the $_2$80 transition (Šumarac et al., 9 Jan 2026).

The stationary-light proposal of (Tebben et al., 2021) embeds a Rydberg coupling into a dual-$_2$81 stationary-light scheme. Under $_2$82, the system supports a stationary Rydberg polariton with zero group velocity and quadratic dispersion, thereby increasing the interaction time relative to ordinary slow-light Rydberg EIT. In a different photonic integration direction, (Santis et al., 12 Feb 2026) realized a cavity-coupled Rydberg array in which $_2$83Rb tweezer arrays are both strongly coupled to a near-concentric optical cavity and coherently excited to $_2$84. The cavity had $_2$85 GHz and $_2$86 MHz; the average single-atom coupling was $_2$87 MHz, the measured cooperativity was $_2$88, and geometric corrections implied $_2$89–$_2$90. At the same spatial location, Rydberg blockade was observed with $_2$91 and collective Rabi scaling $_2$92 for $_2$93 MHz (Tebben et al., 2021, Santis et al., 12 Feb 2026).

A further extension appears in the circular-atom ancilla architecture of (Machu et al., 29 Sep 2025). Here the logical qubit is encoded in the circular states $_2$94 and $_2$95 of $_2$96Rb, while a nearby ancilla atom is transiently excited from $_2$97 to $_2$98. At $_2$99 V/cm, the pair $1s$00 is near-degenerate with $1s$01, producing a Förster splitting $1s$02 MHz at $1s$03. The inferred blockade radius was $1s$04, the ancilla optical $1s$05 pulse duration was $1s$06, the single-shot measurement fidelity was $1s$07, and the effective QND measurement duration was $1s$08. The same ancilla enabled a controlled phase shift of $1s$09 on the circular logical qubit and local suppression of microwave spin flips (Machu et al., 29 Sep 2025).

7. Limitations, trade-offs, and future directions

The limitations reported across dual-Rydberg platforms are diverse but structurally similar: the second Rydberg-enabled sector introduces new functionality at the cost of additional calibration, noise pathways, and interaction channels. In RF sensing, the dual-band chip-integrated receiver suffers a $1s$10 dB sensitivity reduction for free-space LO delivery because of TE/TM mode mismatch, measurable LO leakage, and multi-path reflections; its instantaneous bandwidth remains $1s$11 kHz because it is bounded by the EIT bandwidth $1s$12 MHz (Zhang et al., 2024). In dual-ladder homodyne reception, the practical ceiling is low-frequency multiplicative pink noise, which degrades EVM above $1s$13 kHz symbol rate even though the homodyne architecture is not constrained by a heterodyne intermediate frequency (Oliver et al., 27 Feb 2026).

In neutral-atom processors, the dominant limitations are often laser phase noise, differential AC Stark shifts, Doppler broadening, residual beyond-nearest-neighbor interactions, and imperfect imaging. In the Rb–Cs QND array, the largest gate error sources were idle ground–Rydberg dephasing on the control Cs qubit and blue-induced differential AC Stark dephasing of hyperfine qubits (Anand et al., 2024). In the Na–Cs stabilizer platform, finite interspecies interaction strength is itself the obstacle that necessitates geometric-phase compensation, while ancilla coherence and Na imaging survival remain critical bottlenecks (Wang et al., 11 May 2026). In the global-control QCA processor, beyond-nearest-neighbor van der Waals interactions and slow detuning drifts shorten the lifetime of idealized automaton dynamics (White et al., 22 Jan 2026).

In solid-state and photonic realizations, the key issues shift but do not disappear. Cu$1s$14O microcrystals still face luminescence broadening that limits the resolution of higher-$1s$15 states in photoluminescence, residual microscopic strain that can split the triplet orthoexciton by $1s$16 meV, and the need for cryogenic operation to reach the $1s$17–$1s$18 K exciton temperatures associated with quantum degeneracy (Steinhauer et al., 2019). Cavity-coupled arrays face mode hybridization, geometric reduction of single-atom coupling, and laser phase noise on the Rydberg excitation beams (Santis et al., 12 Feb 2026). Ancilla-based circular-atom readout remains limited by ancilla lifetime, background collisions, and thermal broadening of the Förster resonance (Machu et al., 29 Sep 2025).

The future directions stated in the cited literature are correspondingly platform-specific but conceptually aligned. Integrated silicon photonics, microcavities, and resonant excitation are natural next steps for Cu$1s$19O microstructures (Steinhauer et al., 2019). On-chip LO delivery, mode converters, and improved laser stabilization are the obvious upgrades for RF-chip receiver modules (Zhang et al., 2024). Dual-species processors explicitly target scalable midcircuit measurement, feed-forward, and error-correction cycles (Anand et al., 2024, Wang et al., 11 May 2026). QCA architectures point toward larger graph states and measurement-based quantum computing with constant-depth global operations (White et al., 22 Jan 2026). Hybrid atom–cavity and atom–polaritons devices point toward network nodes, fast non-destructive readout, and nonlinear photonic networks (Song et al., 19 Nov 2025, Šumarac et al., 9 Jan 2026, Santis et al., 12 Feb 2026). Circular-ancilla hybrids extend these ambitions to long-lived circular Rydberg simulators with mid-circuit measurements and time-correlation access over extended evolution times (Machu et al., 29 Sep 2025).

Taken together, the published record indicates that “dual-Rydberg platform” denotes a broad but coherent research direction: the systematic use of two Rydberg-active sectors to decouple control from readout, to mediate interactions between otherwise incompatible modes, or to engineer composite many-body behavior. The technical details differ sharply across semiconductors, vapor cells, tweezer arrays, cavities, and polaritonic media, but the organizing principle remains the same: two coupled Rydberg channels yield operational regimes that single-channel architectures do not natively provide.