Measuring the quantum nature of light with a single source and a single detector

An elementary experiment in optics consists of a light source and a detector. Yet, if the source generates nonclassical correlations such an experiment is capable of unambiguously demonstrating the quantum nature of light. We realized such an experiment with a defect center in diamond and a superconducting detector. Previous experiments relied on more complex setups, such as the Hanbury Brown and Twiss configuration, where a beam splitter directs light to two photodetectors, creating the false impression that the beam splitter is a fundamentally required element. As an additional benefit, our results provide a simplification of the widely used photon-correlation techniques.

Figure 1

Cartoon of a coherent state (a) and a single-photon Fock state (b). Spatial-temporal modes (indicated by lines) of a specific coherence time are populated by discrete excitations. For a coherent state this image corresponds to a snapshot, since the number of excitations per mode $n$ can only be predicted with a certain probability $P\left(n\right)$. In a single-photon Fock state there is exactly one excitation per mode. A detection event projects the mode to the vacuum state. (c) Schematics of a standard Hanbury Brown and Twiss setup, where the light is split on a beam splitter allowing intensity correlation measurements within time intervals shorter than the individual detector dead times. (d) Direct statistical analysis of light by detecting single-photon arrival times with a single detector.

Figure 2

Experimental setup. (a) Scheme of the fiber-coupled detector chip. (b) A single N-V center in a diamond nanocrystal is excited with a 532 nm cw laser. Its emission is collected via a confocal optical microscope, coupled to a single mode (sm) optical fibre, and detected in two configurations: (1) a single fiber-coupled SSPD and (2) a standard free beam Hanbury Brown–Twiss (HBT) setup containing a beam splitter (BS) and two avalanche photodiodes (APDs). Correlations are analyzed by a fast 1 GHz oscilloscope or by a time interval counter. (c) Typical voltage trace with two detection events recorded with the oscilloscope. Only double events with a time difference of between 5 ns and 200 ns were collected.

Figure 3

Correlation measurements with classical light. (a) Measured $g^{\left(2\right)}$ function of an attenuated laser beam sent to a single commercial APD and analyzed via the fast oscilloscope by evaluating traces when a second detection event occurs within a time window of 5 ns to 200 ns. The dashed line indicates the APD dead time. The bunching observed between correlation times of 30 ns and 50 ns is due to afterpulsing. For a rate of 300 000 s${}^{-1}$ and an afterpulsing probability of 0.5$\%$ one out of eleven events is an afterpulsing event and contributes to the area above the solid line [$g^{\left(2\right)}\left(\tau \right)=1$]. (b) Same measurement, but with a SSPD with a dead time shorter than 5 ns. Since only time differences larger than 5 ns were recorded no dead time effect was resolved. The absence of any correlation indicates a Poissonian photon number distribution. The solid line is a linear fit. (c) $g^{\left(3\right)}$ function from an attenuated laser beam measured with a single SSPD obtained by analyzing oscilloscope traces containing 3 or more detection events; ${\tau }_1$ and ${\tau }_2$ are the time intervals between detection events. The 4 ns dead time of the detector yields the expected flat $g^{\left(3\right)}$ function for times longer than 5 ns. Bin sizes are 1 ns for the $g^{\left(2\right)}$ measurements and 5 ns for the $g^{\left(3\right)}$ measurement.

Figure 4

Antibunching measured with a single detector. Measurement of the $g^{\left(2\right)}$ function in the single-detector configuration 1 of Fig. 2b [blue (dark grey) dots] and the standard HBT configuration, configuration 2 of Fig. 2b [red (light grey) dots], respectively. The black line is a fit to a three-level rate equation model. An additional bunching is observed due to occasional population of a metastable singlet state.

Figure 5

Two detection events separated by 2.7 ns measured on a specially designed SSPD with reduced dead time.