Direct measurement of heralded single-photon statistics from a parametric down-conversion source

We present a full triple-coincidence analysis of photon-pair states generated by spontaneous parametric down-conversion. By increasing the coherence time of the source with the help of an intracavity setup, our measurements are not spoiled by detection time jitter. Signal-idler, but also thermal signal-signal, correlations are clearly resolved in this regime. Via introduction of an artificial coincidence window, we discuss in detail the transition to the previously studied cases where typically no single-arm correlation is observed. We investigate the heralded antibunching characteristics to show that in our system further studies of continuously generated photon states, possibly higher-photon-number entangled states, can be performed with respect to their (non)applicability in quantum information tasks.

Figure 1

Experimental setup for heralded single-photon generation consisting of a frequency doubled pump laser (not shown) and a double-resonant optical parametric oscillator (DRO) built from a PPKTP crystal inside a optical cavity with added KTP compensating crystal. The signal and idler photons are spectrally filtered and fed to a Hanbury Brown–Twiss setup and to a single avalanche photodiode (APD), respectively. All APD clicks are recorded by a time-tag recording electronic system.

Figure 2

(a) Signal-idler correlation function ${\overline{P}}_{\mathit{si}}(\tau )$ for an idler photon at $t_i=0$ and signal photons at $t_1=\tau$. The shape reflects the OPO cavity linewidth of 7.2 MHz. (b) Measured signal-signal autocorrelation function ${\overline{P}}_{\mathit{ss}}(\tau )$. The reduced peak height of the measured data is attributed to photon bunching at the beam splitter in the HBT setup. Both measurements were binned with the same bin size ${\tau }_b=3$ ns $\ll \Delta t$.

Figure 3

Measured (simulated) triple-coincidence rate ${\overline{P}}_{\mathit{ssi}}(t_1,t_2,t_i)$ [$P_{\mathit{ssi}}(t_1,t_2,t_i)$] for an idler photon at $t_i=0$ and two signal photons at $t_1-t_i$ and $t_2-t_i$. The ridges of the plots are normalized without binning to 1, which represents the theoretic result in the case of no background noise [25]. The measurement in (a)–(c) as well as the simulation in (d)–(f) is binned with bin sizes of 5, 10, and 40 ns, respectively.

Figure 4

Calculated second-order signal coherence function conditioned on the detection of an idler photon. The black dots represent measurements; the solid red line was simulated according to Eq. (3). The dotted red line represents the same simulation but without the background noise.