Resonance fluorescence revival in a voltage-controlled semiconductor quantum dot

International audienceWe demonstrate systematic resonance fluorescence recovery with near-unity emission efficiency in single quantum dots embedded in a charge-tunable device in a wave-guiding geometry. The quantum dot charge state is controlled by a gate voltage, through carrier tunneling from a close-lying Fermi sea, stabilizing the resonantly photocreated electron-hole pair. The electric field cancels out the charging/discharging mechanisms from nearby traps toward the quantum dots, responsible for the usually observed inhibition of the resonant fluorescence. Fourier transform spectroscopy as a function of the applied voltage shows a strong increase of the coherence time though not reaching the radiative limit. These charge controlled quantum dots can act as quasi-perfect deterministic single-photon emitters, with one laser pulse converted into one emitted single photon. Semiconductor quantum dots (QDs) are commonly considered as artificial atoms due to their discrete electronic structure and are very efficient sources of single and indistinguishable photons [1-3]. Their potential for applications is high as they can be easily integrated into nanophotonic devices [4, 5], defining building blocks for quantum information processing in the solid-state [6, 7]. During the past decade, a lot of effort has been devoted to minimize dephasing processes due to the coupling of QDs to their surrounding solid-state matrix. Indeed, coupling to phonons [8-10] as well as time jitter [11] reduces the degree of indistinguishability and charge or spin noise [12-14] lead to inhomogeneous broadening of the emission line. Therefore, strictly resonant excitation of the QD s-shell has appeared as an essential ingredient [15, 16] to preserve the coherence properties of the emitted photons. Charge noise is detrimental as it strongly limits or even suppresses the QD resonance fluorescence (RF) [17-19]. The RF quench has been attributed to the structure residual doping and defects which create a fluctuating electrostatic environment. This can lead to a Coulomb blockade effect preventing the photocreation of an electron-hole pair in the QD [18, 20]. To circumvent this difficulty and recover the RF, an additional very low power non-resonant laser can be used. Although this technique has been succesful [21-23], the exact physical process of this non-resonant pump has not been sufficiently addressed [23]. Moreover, the lack of control of this non-resonant pump prevents the realization of a fully on-demand single-photon source with a high degree of coherence for any probed QD. Here, we show how a revival of the RF can be achieved by using a suitably designed voltage-controlled device that stabilizes the resonantly photocreated electron-hole pair in the dot. The resonant excitation is realized in an in-plane waveguide geometry [8], while the single pho-tons are collected from the top. This geometry yields an almost complete suppression of the laser scattered light on the RF detection side. By controlling the QD electric field environment by a gate voltage, the charg-ing/discharging mechanisms from the nearby trap states to the QD are disabled. The resonantly photocreated electron-hole pairs give rise to a very intense RF line and an increase of the coherence time. However the radiative limit is not reached suggesting that charge and/or spin noise are still present in the structure leading to residual inhomogeneous broadening. Still, the gate control allows to convert one laser pulse into one emitted photon and optimization of the collection efficiency remains the last step to achieve for using such a device in quantum technologies applications. A low density InAs/GaAs self-assembled QD layer, grown by molecular beam epitaxy, was embedded at the center of a p-in doped GaAs/AlAs microcavity on a n-doped GaAs (001) substrate. The Bragg mirrors consist , for the n-doped bottom side, of 24 pairs silicon-doped at 2.10 18 cm −3 and 1.10 18 cm −3 closer to the cavity , while the p-doped Bragg mirror was carbon-doped at 2.10 18 cm −3 , except for the last two pairs at 2.10 19 cm −3 to improve the contact ohmicity. The cavity and the λ/4 layers were designed to obtain a cavity mode centered at 920 nm. The quality factor is only a few hundreds and does not induce any significant Purcell effect. Deep (ap-proximatively 1.5 µm) ridges were etched by inductively coupled plasma etching realizing one-dimensional waveg-uides with 0.8 to 1.2 µm width. Standard ohmic contacts were deposited and annealed on the back side. Top contacts were realized by resist planarization and deposi-tion of Ti/Au stripes perpendicular to ridges. The resist was then etched away between the stripes. A schematic view of the experimental geometry, a characteristic I-V curve and the simplified band structure are depicted in Fig. 1 (a), (b) and (c) respectively. In the following, U will denote the external applied bias and V BI the built-in voltage. Then, the QD potential is (V BI − U) for reverse bias. To investigate the resonant and non-resonant fluores-cence, the sample and the microscope objectives are mounted inside a He closed cycle temperature-variabl