Near-Transform-Limited Single Photons from an Efficient Solid-State Quantum Emitter

By pulsed $s$-shell resonant excitation of a single quantum dot-micropillar system, we generate long streams of 1000 near-transform-limited single photons with high mutual indistinguishability. The Hong-Ou-Mandel interference of two photons is measured as a function of their emission time separation varying from 13 ns to $14.7\mu \mathrm{s}$, where the visibility slightly drops from 95.9(2)% to a plateau of 92.1(5)% through a slow dephasing process occurring at a time scale of $0.7\mu \mathrm{s}$. A temporal and spectral analysis reveals the pulsed resonance fluorescence single photons are close to the transform limit, which are readily useful for multiphoton entanglement and interferometry experiments.

Figure 1

Generation and characterization of single photons from a QD embedded in a micropillar. (a) An illustration of the experimental setup. The QD is grown by molecular beam epitaxy and sandwiched between 25.5 (15) lower (upper) distributed Bragg reflectors stacks. Pillars with a diameter of $2\mu \mathrm{m}$ are defined via electron beam lithography [30]. The quality factor of the micropillar cavity is measured to be 5349 (see the Supplemental Material, Fig. S1 [31]). The emitted RF single photons are sent to a Fabry-Perot cavity for high-resolution spectral analysis, or to an unbalanced Mach-Zehnder interferometer for $g^1(\tau )$, $g^2(\tau )$, and HOM measurements. (b) Temperature-dependent 2D intensity plot of the QD photoluminescence and the cavity mode. The excitation laser is at 780 nm with a power of $\sim 0.3\mathrm{nW}$. (c) Detected single photon counts as a function of the pump field strength. (d) Intensity-correlation histogram of the RF photons. The measured second-order correlation at zero delay is $g^2(0)=0.007(1)$.

Figure 2

Single photon indistinguishability measured from HOM experiments at various photon emission time separations $\mathrm{\Delta }$. (a) $\mathrm{\Delta }=13\mathrm{ns}$. The input two photons are $\pi$-pulse excited and prepared in cross (black) and parallel (red) polarizations. The data accumulation time was 5 min. The data points presented are raw data without background subtraction. (b) $\mathrm{\Delta }=14.7\mu \mathrm{s}$, while keeping all the other physical conditions the same as in (a). (c) The extracted photon indistinguishability as a function of emission time separation. The small window of a few percent difference in the indistinguishability is precisely measured with a small error bar, which is possible due to the high photon count rate (that eliminates the shot noise) from the efficient QD-micropillar device [11]. The same data are shown in the inset that plots the $x$ axis in log scale and highlights the number of consecutively emitted single photons for each data point. The red curve is a fit using the model derived in Ref. [27] assuming non-Markovian noise.

Figure 3

Characterization of the single-photon source. (a) Time-resolved RF counts at QD-cavity resonance (4.3 K) and at far detuning (32 K). (b) Measurement of the coherence time of the single photons using a Mach-Zehnder interferometer, showing the fringe contrast versus path-length difference. The fringe contrast was calculated from registered single photon counts of one of the interferometer outputs while scanning one of the arm lengths over about 10 wavelengths. (c) A high-resolution RF spectrum when excited by a $\pi$ pulse, obtained using a home-built Fabry-Pérot scanning cavity with a finesse of 170, a linewidth of 220 MHz (full width at half maximum), a free spectral range of 37.4 GHz, and a total transmission rate of 61%. The blue line was a fit using a Lorentzian profile. The red line was a fit using a Voigt profile with the residual shown in the bottom panel. The Fourier transform of this spectrum after deconvolution with the 220-MHz Fabry-Pérot instrumental resolution gives a coherence time of 291(8) ps, in good agreement with (b).