arXiv:2610.20005
Hyperentanglement in intermediate phase-space planes: benchmarking a photon-pair source for quantum key distribution
Murilo H. Magiotto, Millena B. Novaes, Wesley B. Cardoso, Ardiley T. Avelar, 1 Paulo H. Souto Ribeiro, Stephen P. Walborn, Guilherme L. Zanin, Rafael M. Gomes
Abstract
We characterize a source of photon pairs simultaneously entangled in polarization and in the transverse spatial degree of freedom, with the spatial measurements performed in intermediate planes of the fractional Fourier transform. The source consists of two crossed type-I crystals in a non-collinear geometry, such that the transverse variables carry information about the crystal of origin. Detection in the crystal plane reveals this information completely and destroys the polarization entanglement, so that a scheme using the near and far fields as conjugate spatial bases cannot support a parallel polarization channel. In intermediate planes both spatial bases combine position and momentum, such that the which-crystal information is partially erased, and the polarization entanglement survives in all four combinations of spatial bases. Applying Bell and Einstein-Podolsky-Rosen criteria to the same photon pairs certifies both degrees of freedom simultaneously. We evaluate quantum bit error rates and asymptotic secret key rates in each degree of freedom, and find that the source delivers two simultaneously usable channels per photon pair.
AI Summary
The paper asks whether a non-collinear, crossed-type-I SPDC source can provide simultaneous polarization and transverse-spatial entanglement when two conjugate spatial bases are needed. Transverse displacement between emissions from the two crystals carries which-crystal information: near-field detection suppresses polarization coherence. The proposed readout instead uses fractional-Fourier-transform planes near $\pi/4$ and $3\pi/4$, partially erasing this information in both bases. Experiments with two 1-mm BBO crystals characterize polarization correlations, conditional spatial distributions, and joint error statistics. Polarization CHSH values are $S=2.57$–$2.68$ across all four spatial-basis combinations, with concurrence lower bounds of $0.81$–$0.91$; near-field measurements give $S=1.45\pm0.03$. Opposite spatial bases yield a conditional-variance product $S^{CV}=0.071\pm0.009$, below the Reid bound $1/4$. Joint errors are consistent with independent signal errors plus a shared random background of $(0.74\pm0.13)\%$. For a two-cell spatial encoding, measured errors give asymptotic benchmarks of $0.558\pm0.013$ polarization and $0.837\pm0.011$ spatial secret bits per sifted pair, totaling approximately $1.40$ when both bases match. A continuous-variable entropy bound gives $0.47$ bits per detected pair against a quantum-memory adversary. This is source characterization, not demonstrated QKD: measurements use scanning single detectors over a laboratory path, no key is distributed, and no finite-key analysis is performed. Spatial variances are sampled at only two conditioning positions. The two-cell analysis idealizes discarded events, while basis-dependent optics and slit widths introduce efficiency mismatches requiring security analysis. Practical operation would require multiplexed detection.
Central claim may not hold: The reported spatial steering certification and continuous-variable key bound use conditional widths at only two Bob outcomes, without establishing the outcome-averaged inference variances or conditional entropies required by those bounds. The 1.40-bit combined key benchmark assumes a complementary spatial-qubit encoding not established by the two-cell measurements and does not account for the acknowledged detection-efficiency mismatch. Compatibility of measured error patterns with independent signal errors and uniform background does not establish factorization of the polarization–spatial quantum state. Equation (Kint) inverts the kappa-dependent prefactor relative to Equation (Kgauss), although the quoted 1.5-bit estimate uses the correct expression.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.