We show two experimental realizations of Hardy's ladder test of quantum nonlocality using energy-time correlated photons, following the scheme proposed by Cabello et al.[Phys. Rev. Lett. 102, 040401 (2009)]. Unlike previous energy-time Bell experiments, these tests require precisely tailored nonmaximally entangled states. One of them is equivalent to the two-setting and two-outcome Bell test requiring a minimum detection efficiency. The reported experiments are still affected by the locality and detection loopholes, but are free of the post-selection loophole of previous energy-time and time-bin Bell tests.
We discuss the problem of finding the most favorable conditions for closing the detection loophole in a test of local realism with a Bell inequality. For a generic nonmaximally entangled two-qubit state and two incompatible bases to be adopted for alternative measurements of two observables $\mathrm{a\ifmmode \hat{}\else \^{}\fi{}}$ and $\mathrm{b\ifmmode \hat{}\else \^{}\fi{}}$ on each party, we apply Hardy's proof of nonlocality without inequality and derive an Eberhard-like inequality. For an infinity of nonmaximally entangled states we find that it is possible to refute local realism by requiring perfect detection efficiency for only one of the two observables, say $\mathrm{b\ifmmode \hat{}\else \^{}\fi{}}$, to be measured on each party: The test is free from the detection loophole for any value of the detection efficiency corresponding to the other observable $\mathrm{a\ifmmode \hat{}\else \^{}\fi{}}$. The maximum tolerable noise in such a loophole-free observable-asymmetric test is also evaluated.
The certification of quantum nonlocality, which has immense significance in designing device-independent technologies, confronts severe experimental challenges. Detection loopholes, originating from the unavailability of perfect detectors, are one of the major issues among them. In the present study we focus on the minimum detection efficiency (MDE) required to detect various forms of genuine nonlocality, originating from the type of causal constraints imposed on the involved parties. In this context, we demonstrate that the MDE needed to manifest the recently suggested
T2-type nonlocality deviates significantly from perfection. Additionally, we have computed the MDE necessary to manifest Svetlichny's nonlocality, with the state-independent approach markedly reducing the previously established bound. Finally, considering the inevitable existence of noise we demonstrate the robustness of the imperfect detectors to certify
T2-type nonlocality.