Non-classical states of light, such as entangled photon pairs and number states, are essential for fundamental tests of quantum mechanics and optical quantum technologies. The most widespread technique for creating these quantum resources is spontaneous parametric down-conversion of laser light into photon pairs1. Conservation of energy and momentum in this process, known as phase-matching, gives rise to strong correlations that are used to produce two-photon entanglement in various degrees of freedom2, 3, 4, 5, 6, 7, 8, 9. It has been a longstanding goal in quantum optics to realize a source that can produce analogous correlations in photon triplets, but of the many approaches considered, none has been technically feasible10, 11, 12, 13, 14, 15, 16, 17. Here we report the observation of photon triplets generated by cascaded down-conversion. Each triplet originates from a single pump photon, and therefore quantum correlations will extend over all three photons18 in a way not achievable with independently created photon pairs19. Our photon-triplet source will allow experimental interrogation of novel quantum correlations20, the generation of tripartite entanglement12, 21 without post-selection and the generation of heralded entangled photon pairs suitable for linear optical quantum computing22. Two of the triplet photons have a wavelength matched for optimal transmission in optical fibres, suitable for three-party quantum communication23. Furthermore, our results open interesting regimes of non-linear optics, as we observe spontaneous down-conversion pumped by single photons, an interaction also highly relevant to optical quantum computing.
Figures at a glance
Figure 1: Schematic of photon-triplet generation and experimental set-up. 
a, A down-conversion source (SPDC 1) produces a pair of photons in spatial modes 0 and 1, where the photon in mode 0 creates another photon pair in the second source (SPDC 2) in modes 2 and 3, generating a photon triplet. b, The primary source, pumped by a 405-nm laser, produces photon pairs at 775 nm and 848 nm. The 848-nm photon is directly detected by a silicon avalanche photodiode (D1), and the 775-nm photon serves as input to the secondary source, creating a photon pair at 1,510 nm and 1,590 nm that is detected by two InGaAs avalanche photodiodes (D2 and D3). A detection event at D3 represents a measured photon triplet. BS, beam splitter; F0, F1, band-pass filters; FP, long-pass filter; G, gate; TAC, time acquisition card; PC, computer.
Figure 2: Triple-coincidence histograms. 
a, Measured triple coincidences obtained in 20 h. Each bin corresponds to a 0.8-ns time interval between events at D3 and D1 (ΔτD3–D1). The sharp peak indicates a strong temporal correlation between all three detection events, as expected of the C-SPDC process. b, Triple-coincidence histograms with varying delays of τ = 0 and ±0.5 ns between D2 and D3, resulting in a decrease of the coincidence peak. The absolute rate reduction for τ = 0 results from a different setting on the InGaAs detectors for this measurement series. Error bars, 1 s.d.
Figure 3: Phase-matching and triple-coincidence dependence on crystal temperatures. 
a, Central wavelengths of the pair of photons produced by the secondary source as a function of the PPLN temperature for input wavelengths of 775.4 nm (circles) and 776.0 nm (squares). The dashed lines show the theoretical phase-matching curves with the poling period as the only fit parameter. Triple coincidences were measured for different settings of the PPLN temperature and the input pump photon wavelength. The PPLN temperature was 60 °C for setting A and 50 °C for settings B and C; the input photon wavelength was 776.0 nm for settings A and B and 775.4 nm for setting C. b, Measured triple coincidence histograms over 20 h for each measurement setting. For A and C, the PPLN temperatures lie on the respective phase-matching curves and a triple-coincidence peak is observed. For B, the temperature is outside the 776.0-nm phase-matching curve and no peak is present. Wavelength changes in the input photons, needed for the measurements shown in Fig. 3b, were achieved by altering the temperature of the PPKTP crystal (43.6 °C for settings A and B, 40.8 °C for C). Error bars, 1 s.d.
right
