Efficient and high-fidelity entanglement in cavity QED without high cooperativity

The so-called state-carving protocol generates high-fidelity entangled states at an atom-cavity interface without requiring high cavity cooperativity. However, this protocol is limited to 50% efficiency, which restricts its applicability. We propose a simple modification to the state-carving protocol to achieve efficient entanglement generation, with unit probability in principle. Unlike previous two-photon schemes, ours employs only one photon which interacts with the atoms twice - avoiding separate photon detections which causes irrecoverable probability loss. We present a detailed description and performance evaluation of our protocol under non-ideal conditions. High fidelity of 0.999 can be achieved with cavity cooperativity of only 34. Efficient state-carving paves the way for large-scale entanglement generation at cavity-interfaces for modular quantum computing, quantum repeaters and creating arbitrary shaped atomic graph states, essential for one-way quantum computing.

💡 Research Summary

The paper addresses a fundamental limitation of the state‑carving (SC) protocol, a method for generating high‑fidelity atom‑photon entanglement in cavity QED without requiring large cavity cooperativity. The original SC scheme relies on two ancillary photons that must be detected sequentially; because each photon detection is probabilistic, the overall success probability is capped at 50 %. This probabilistic bottleneck severely limits scalability for applications such as modular quantum computing, quantum repeaters, and the construction of large graph states.

The authors propose a simple yet powerful modification: use a single ancilla photon that interacts with the two atoms twice. The protocol works in a two‑sided cavity (or equivalently in a ring cavity) with only passive linear optics (mirrors, a 50:50 beam splitter) and a small number of optical switches. The sequence is as follows:

1. Prepare both atoms in the separable super‑position |+⟩⊗|+⟩.
2. Send a single photon into the cavity. Depending on the joint atomic state, the photon is either transmitted (|t⟩) or reflected (|r⟩), creating an entangled atom‑photon state.
3. If the photon is reflected, a mirror routes it back into the cavity while a NOT gate flips each atom, converting the |11⟩ component into |00⟩.
4. The photon makes a second pass; now the transmitted and reflected components from the second pass interfere on a 50:50 beam splitter, producing three possible detection outcomes (two transmission ports D₂, D₃ and one reflection port D₁).
5. Each detection projects the atoms onto one of the four Bell‑type states: |00⟩±|11⟩ (from D₂/D₃) or |01⟩+|10⟩ (from D₁). Conditional single‑qubit rotations can then select any desired Bell state.

Because only one photon is ever generated and detected, the protocol eliminates the irrecoverable loss associated with the two‑photon detection in the original SC. In the ideal lossless limit the success probability is unity; the only failures arise from realistic cavity imperfections and photon loss.

The performance analysis focuses on three key cavity parameters: the input‑output coupling rates κ₁ and κ₂, the scattering loss κ_sc, and the cooperativity C = 4g²/(κγ). Using a full quantum‑optical treatment (details in the Supplementary Material), the authors derive analytic expressions for the total success probability P_tot and the average fidelity F_avg. The fidelity scales as

F_avg ≈ √(1 − ε₁²) − (17/16) C⁻², ε₁ = 1 − 2κ₁/κ,

showing an O(C⁻²) dependence, which is substantially weaker than the O(C⁻¹/²) or O(C⁻¹) scaling typical of photon‑exchange mediated gates. Consequently, a fidelity of 0.999 is achievable already at C ≈ 34, whereas comparable fidelities in other cavity‑mediated schemes would require C ≈ 10³.

Numerical simulations explore the impact of asymmetry (κ₁ ≠ κ₂) and finite scattering loss (κ_sc ≈ 0.02 κ). The average fidelity is primarily sensitive to κ₁, while the total probability depends on both κ₁ and κ₂. An “fidelity‑weighted success metric” (∑ P_i F_i) is introduced to capture the fact that a high raw detection probability does not guarantee entanglement if the photon never entered the cavity (e.g., κ₁ = 0). This metric reaches unity only near the symmetric, low‑loss regime κ₁ ≈ κ₂ ≈ κ/2, confirming the importance of balanced coupling.

The authors also discuss practical implementation. High‑extinction, low‑loss optical switches with >20 dB isolation and >10 GHz bandwidth are commercially available, making the required routing feasible. In a ring (toroidal) cavity the need for switches can be reduced to a single element or even eliminated by using an optical isolator, because forward and backward propagating modes naturally separate the reflected and transmitted photons.

Scalability is demonstrated by iterating the protocol to add new atoms and grow arbitrary 2‑D graph states. Because each carving step now succeeds with near‑unit probability, the overall success probability for an N‑node cluster scales as ~1 rather than (0.5)^{N‑1}. Simulations for C = 20–50 show that clusters with tens of nodes can be generated with >90 % fidelity, and the fidelity is limited mainly by the cooperativity, not by loss, thanks to the heralded nature of the protocol.

In comparison with other cavity‑based entangling gates, the efficient SC protocol stands out for three reasons: (i) it works at modest cooperativity, (ii) it exhibits a favorable O(C⁻²) error scaling, and (iii) it requires only passive linear optics and a single photon, avoiding the complexity of multi‑photon generation and detection. These attributes make it a strong candidate for integration into modular quantum processors, quantum repeater nodes, and large‑scale one‑way quantum computing architectures where deterministic high‑fidelity entanglement is essential.