A Tractable Protocol for Detection-Loophole-Free Bell Tests over Long Distances

TL;DR

The paper proposes an event-ready protocol that combines a two-mode squeezed vacuum at Alice with a heralded photon-number path-entangled state at Bob to herald a tunable PNPE state between distant parties. A central Bell-state measurement at Charlie induces entanglement with a loss-tolerant vacuum component, enabling post-selection-free Bell-inequality violations at the Eberhard limit and preserving a square-root scaling with channel transmittance, $\mathcal{O}(\sqrt{\eta_C})$. The authors provide analytic and numerical results showing CHSH violations down to $\eta_D=2/3$ and compare against SPPE- and polarization-based schemes, highlighting improved robustness to loss and higher reachable violations. They also discuss implications for device-independent quantum information tasks such as DI-QKD and randomness certification, and analyze how the vacuum component protects against loss while local losses limit practical performance.

Abstract

Certifying genuine nonclassical correlations over long distances is crucial for device-independent (DI) quantum information protocols. However, in photonic platforms this remains technologically challenging due to photon loss, which opens the detection-loophole, rendering violations increasingly difficult for less efficient detectors. A well-known strategy to mitigate this involves using non-maximally entangled states, which Eberhard showed can tolerate lower detection efficiencies. However, existing proposals and demonstrations have been limited to short distances, as their success rates scale linearly with channel transmittance. Here, we propose a protocol to herald a tunable entangled state between distant users that achieves a post-selection-free Bell inequality violation at the Eberhard limit. We identify the loss independence of the vacuum component amplitude of the prepared state as the source of this enhancement. Notably, our scheme requires only quantum interference at a central station, followed by the detection of a single photon, preserving the optimal square-root scaling with channel transmittance. Our approach provides greater loss-tolerance in entanglement distribution, bringing long-distance DI applications closer to practical implementation.

Figures (6)

• Figure 1: An illustration of our protocol for the heralded distribution of photon-number path-entanglement (PNPE). Alice and Bob, separated by a distance $L$, generate entangled states using sources $\text{S}_A$ and $\text{S}_B$, respectively. The modes $a_2$ and $b_2$ are directed to a middle station (Charlie), which performs a Bell-state measurement (BSM), thereby heralding entanglement between the modes $a_1$ and $b_1$. The photons travelling to Charlie experience a channel transmittance of $\sqrt{\eta_C}$, and a successful BSM occurs if exactly one photon arrives. The modes $a_1$ and $b_1$ are then interfered with weak local oscillators ($\ket{\alpha}$, $\ket{\beta}$) to implement displacement-based measurements for the Bell test. In this illustration, the sources are based on SPDC: Alice prepares a two-mode squeezed vacuum state, while Bob generates a heralded single-photon path-entangled state (SPPE).
• Figure 2: Optimal CHSH violations as a function of symmetric detection efficiency $\eta_D$. The solid green curve shows the theoretical prediction for our protocol, with numerical simulations that account for higher-order photon-number contributions (green diamonds) in excellent agreement. The gray dashed line corresponds to previous photon-number-based protocols, which require detection efficiencies above 82.6% (red cross and dropline) to observe a violation. Coloured markers denote detection-loophole-free violations from leading polarisation-based experiments. While those can operate at the Eberhard limit (66.7%), the observed violations are marginal and their distance scaling is limited. In contrast, our protocol achieves significantly stronger violations down to the Eberhard limit, demonstrating enhanced robustness to loss and improved overall performance.
• Figure 3: Success probability of performing the BSM at Charlie and heralding entanglement between Alice and Bob. The squeezing parameter $g$ is fixed at 0.33. For Ref. steffinlongo2024long and for this work, the BS transmittance is set to 0.1 and 0.8, respectively. The parameters are chosen to remain consistent with the operating assumptions and optimal regimes of each protocol.
• Figure 4: Min-entropy $\text{H}_{\min}$ (left axis, blue) and Holevo bound $\chi_{\max}$ (right axis, orange) as functions of detection efficiency $\eta_D$ for the protocols compared. Solid lines correspond to the present work; dashed lines represent previous approaches (Refs. mycroft2023proposalsteffinlongo2024long). Increasing $\text{H}_{\min}$ indicates higher certifiable randomness, while decreasing $\chi_{\max}$ reflects reduced eavesdropper information in DI-QKD.
• Figure S1: Log-negativity $\mathcal{E}_N$ of the heralded state as a function of the squeezing parameter $g$ and Bob's BS transmittance $t_b$. The colour scale ranges from $\mathcal{E}_N = 0$ (separable) to 1 (maximally entangled), demonstrating the protocol’s tunable entanglement.