1. Introduction

The development of quantum communication from toy model demonstrations to commercial applications and reliable quantum networks demands the fabrication and characterization of integrated and packaged light sources with non-classical emission characteristics. Most approaches in quantum communication still rely on the use of attenuated laser pulses or heralded photons from spontaneous parametric down-conversion [1, 2, 3]. But a true deterministic single-photon source is inevitable since security in long distance communication is crucially limited by Poisson statistics and is also a requisite for realizations of all-optical quantum computation [4]. Photoluminescence from single semiconductor quantum dots [5] has become an important player in the field meeting the demanded deterministic emission of single-photon states [6, 7] and even opening the gates to coherent mapping of information onto stationary solid-state quantum bit systems for information processing [8] due to its narrow linewidth.

The standard procedure for single-photon generation from single quantum dots is selective optical excitation via a microscope objective with high numerical aperture [9]. However, electrical excitation is required to allow for the fabrication of integrated compact devices and to avoid large-scale external pump sources. Early schemes utilized a simultaneous Coulomb-blockade for electrons and holes in a semiconductor triple quantum well nanostructure [10]. Other realizations embedded single self-organized quantum dots in pin-diode structures [11, 12, 13, 14, 15]. Their emission can be enhanced and directed into distinctive modes by the growth of Bragg reflectors [16]. But the desired ultimate control in integrated ready-to-go single-photon sources for quantum communication includes the deterministic injection of a single electron and a single hole into a quantum dot since it avoids the need for spectral or spatial filtering of the emission. Recently, it was shown [17] that it is possible to selectively pump a single QD in a pin-diode and to obtain a pure emission spectrum with only a single exciton line. Here, we report for the first time on non-classical photon statistics from such a device.

2. Sample fabrication

A promising approach for single-photon generation based on electrical pumping uses a micron-size aluminum oxide aperture to restrict the current flow to a single dot [13, 14]. Thus, electrical excitation of more than one dot was significantly suppressed, and non-classical statistics of the electroluminescence was measured. But that work could not finally prove the injection of a single electron and a single hole into the pin-junction that generates sub-Poissonian statistics without external filtering. The structure of our device has been described elsewhere [17, 18] and will be sketched only shortly (Fig. 1). It is grown on semi-insulating (100) epi-ready GaAs substrates using a Riber-32P MBE system. The light emitting diode (LED) consists of an un-doped GaAs layer with InAs quantum dots of low density inserted, a 60 nm thick aperture layer of high aluminum content AlGaAs, and p- and n-type GaAs electrical contact layers. The low quantum dot density of 10⁸ cm^-2 was obtained in the Stranski-Krastanow mode by deposition of 1.8 ML of InAs. Cylindrical mesas were processed by inductively coupled plasma reactive ion etching, and selective oxidation of the high aluminum content AlGaAs layers led to submicron-size oxide current apertures. Subsequent Si ₃N₄ deposition allowed Au/Pt/Ti and Au/Au-Ge/Ni metallization to form p- and n-contacts, respectively.

 

Fig. 1. Schematic cross section of the device structure.

Download Full Size | PDF

3. Measurements

In order to probe the electrical excitation of only a single quantum dot, we measured the electroluminescence spectrum at an injection current of 870 pA and a bias voltage of 1.65V (Fig. 2(a)). It reveals just one single line, and emission of the wetting layer is also completely suppressed. For experiments related to entangled pair generation using biexciton-exciton decays [20], emission from an uncharged dot is requisite. Via high resolution spectroscopy, we could determine a fine structure splitting of 55 µeV due to electron-hole exchange interaction. The existence of a splitting proves the electroluminescence to originate from an exciton rather than a trion state and will allow further experiments towards entanglement generation with this device. Still, other dots may be excited and their electroluminescence blocked by a shadow mask. But, the micro-photoluminescence of a few dots (Fig. 2(b)) at 10 K exhibited a set of several discrete lines. This proves that the oxide aperture above the dot under study is transparent for near infrared light, and the absence of light emission other than from the exciton decay in the electroluminescence spectrum clearly demonstrates the pumping of indeed only a single quantum dot.

 

Fig. 2. (a) Electroluminscence spectrum at a current of 870 pA and 1.65 V bias voltage (b) Micro-photoluminescence with a laser spot size of 2 µm.

Download Full Size | PDF

In Fig. 3, six electroluminescence spectra are depicted to further characterize the emission of our device at increasing injection current. At around 1 nA, corresponding to an injection of 5 electrons/ns, the exciton intensity saturates. With the typical exciton lifetime of 1 ns [19, 21, 22] and an internal quantum efficiency of about unity [5, 23, 24], this gives a remarkable injection efficiency of about 20% since the capture time for carriers is in the picosecond regime. This is two orders of magnitude better than in previously reported structures [11, 12, 13, 14]. At higher currents, two additional lines appear. They both originate from the same dot since they obey the same spectral jitter. The high-energy peak is assigned to the biexciton decay due to its super-linear dependence on the injection current. The third line is unpolarized and does not show any splitting; thus, it can be attributed to the trion. As already mentioned, the emission from uncharged dots paves the way for on-demand generation of entangled photon pairs from our device which was not possible in earlier realizations [14].

Given these extremely clean emission features and excellent carrier control, we further exploited the use of our device as a single-photon source. Non-classical photon statistics is characterized by the second-order intensity correlation function via a Hanbury-Brown and Twiss setup (HBT) [25] consisting of a 50:50 beam splitter and two avalanche photo diodes (APDs). In contrast to all previous HBT setups related to quantum dot experiments, no spatial or spectral filtering of the exciton line was needed (except a 10 nm FWHM band pass filter centered at 953 nm in front of one APD to avoid cross-talk between the detectors).

 

Fig. 3. Electroluminescence spectra at different current injection. For clarity an offset is added to each spectrum.

Download Full Size | PDF

 

Fig. 4. Autocorrelation measurement under continuous wave current injection (0.9 nA, 1.65 V) at 10 K. No spectral filtering was used to isolate a single transition in a single quantum dot. The dashed line shows a fit function as described in the text.

Download Full Size | PDF

The result of our autocorrelation measurement is depicted in Fig. 4. In order to interpret our experimental data, we use the rate model described in [26]. While an ideal single-photon source shows an antibunching dip of the correlation function down to zero, the limited time resolution of our HBT setup must be taken into account. Thus, the ideal function g ⁽²⁾(Δt)=1−exp(−Δt/τ) was convoluted with a Gaussian (FWHM: 800 ps) and yields a dip depth of around 65 %. The time resolution of 800 ps was independently determined by an autocorrelation measurement of ultra-short laser pulses. The convoluted fit shown in Fig. 4 agrees well with our experimental data and proves that our quantum dot LED represents indeed an ideal electrically pumped single-photon source.

4. Conclusion

In summary, the measurements characterize our quantum dot LEDs with submicron-size aperture as highly efficient light sources that allows the controlled injection of a single electron and a single hole into a single quantum dot. Photons from uncharged dots can be extracted in well-defined polarization modes. Photon statistics exhibits strong antibunching. Together with its high efficiency and unmatched spectral purity, our structure is predestined for the generation of single-photon states as required for numerous applications in quantum communication. Tuning the fine structure splitting to zero [27] would allow to implement the method of cascaded decay of a biexciton proposed in [20] to generate polarization entangled photon pairs from an electrically pumped device. Similar as in the case of recently demonstrated polarization entangled photon pair generation from optically excited quantum dots [28, 29], a quantum state tomography by a linear combination of cross correlation measurements using 16 different polarization combinations could be performed with our setup. A source of entangled photon pairs on demand should allow the usage for various experiments in quantum information processing where resources of single photons and entangled photon pairs are needed. Although the purity of state generation by parametric down-conversion has not been obtained yet, first results based on single quantum dots [28, 29] are very promising. As an add-on to optically excited on-demand sources, electrical excitation provides the potential for future easy to use as well as highly integrated real devices.