High-fidelity entangled photon pairs from a quantum-dot-based single-photon source

Abstract

Entangled photon pairs are a ubiquitous resource in quantum technologies, used in quantum key distribution and quantum networking as well as fundamental tests of non-locality. For scalable quantum networks, pairs that are indistinguishable in all unentangled degrees of freedom are essential, as they enable high-fidelity entanglement swapping across network nodes. To date the most-studied sources of "swappable" entangled photon pairs have been based on spontaneous parametric down-conversion (SPDC) in non-linear crystals. However, the probabilistic nature and unavoidable trade-off between brightness and unwanted multi-photon emission limits their performance in lossy channels. Here, we demonstrate a high-fidelity source of "swappable" entangled photon pairs using a semiconductor quantum dot (QD) coupled to a tunable microcavity. By actively modulating the QD emission between orthogonal polarisation states, delaying one path in a low-loss Herriott cell, and recombining the two on a balanced beam splitter, we generate entangled photon pairs with a fidelity of $96.1\pm0.5$ %. We identify and mitigate fidelity-limiting factors, achieving a maximum fidelity of $98.1\pm0.5$ % through time-resolved post-selection. The scheme suppresses residual multi-photon events concentrated near the excitation pulse and has only a modest impact on the rate. Furthermore, the photons are mutually indistinguishable, enabling efficient entanglement swapping. Our results establish semiconductor QDs as a viable platform for quantum network-compatible swappable entangled photon pair generation, with feasible entanglement generation rates exceeding 0.5 Gpairs/s.

Figures (14)

• Figure S1: Experimental setup. (a) Schematic of a postselected entanglement source using a stream of indistinguishable photons. (b) Representation of our experimental setup. The polarisation of every other photon from the quantum dot (QD) is switched via the resonant electro-optic modulator (EOM) and the polarising beam splitter (PBS). After recombining at a non-polarising beam splitter (NPBS), the resulting entangled photon pairs are detected via coincidence counting using superconducting nanowire single photon detectors (SNSPDs), while the half-waveplate (HWP), quarter-waveplate (QWP), and PBS in each output arm of the beam splitter are used for quantum state tomography to reconstruct the density matrix.
• Figure S2: Indistinguishable entangled photon pairs. (a) Two-photon density matrix of the singlet fraction for the source running at repetition rate $R_{L}$, yielding a fidelity of $F=96.1\pm 0.5 \%$ for the state $\ket{\psi} = \ket{HV}+e^{i\phi}\ket{VH}$, where $\phi \approx -11$ degrees (the singlet fraction is obtained by applying single-qubit rotations that optimise the overlap with the target Bell state; this results in a density matrix that has only real components). (b) Hong-Ou-Mandel interference experiment for excitation pulse separation of 500 ps, showing visibility of $V_{HOM} = 97.6 \pm 1.5\%$. (c) Two-photon density matrix of the singlet fraction for excitation pulse separation of 500 ps, with a fidelity of $F = 95.2\pm 0.5 \%$.
• Figure S3: Dependence of fidelity on source properties. (a) Fidelity as a function of the temporal misalignment between the two interfering photons. At zero delay ($\Delta t = 0$ ps), the fidelity is $F = 95.8\pm 0.5 \%$. It decreases to $F = 69.1\pm 0.5 \%$ at $\Delta t = 60$ ps, corresponding to one radiative lifetime. (b) Fidelity versus $g^{(2)}(0)$. For $g^{(2)} = 1.3\%$, the fidelity is $96.4\pm 0.5 \%$, while for $g^{(2)} = 7.7\%$, $F$ decreases to $88.8\pm 0.5 \%$. In both panels, black lines represent theoretical fits (see Supplementary Information).
• Figure S4: Improving fidelity with filtering. (a) Entanglement fidelity as a function of etalon detuning for the 5-GHz bandwidth etalon. Zero detuning corresponds to the etalon being resonant with the single photons. The maximum fidelity is $98.3\%$ for $\delta = 0.8$ GHz. The asymmetry in the entanglement fidelity results from the asymmetric transmission of the etalon. (b) Main plot: two-photon entanglement fidelity versus switching time, $t_{\rm ON}$. In the data analysis, coincidences for $t<t_{\rm ON}$ are rejected and coincidences for $t \ge t_{\rm ON}$ are retained. $t=0$ represents the time of the maximum intensity of the laser pulse (see Supplementary Information). Inset: Example histogram of the timing between the data and trigger signals. The dark blue curve corresponds to the raw histogram, while the lighter blue curve shows the histogram after temporal filtering. Only counts within the shaded region are included in the fidelity calculation; counts in the unshaded region are excluded.
• Figure S5: Entanglement-swapping rates across platform. Entanglement-swapping (E.-Swap) rate comparison between a postselected quantum-dot source and various SPDC-based sources. Plot assumptions: SPDC extraction efficiency of 0.8; optimised SPDC pump power to achieve entanglement fidelity $F > 97\%$; QD extraction efficiency of 0.71; spatial multiplexing with insertion efficiency of 1 and switch efficiency of 0.97; detector efficiency of 0.9 with photon-number resolution (PNR).