Single Self-Assembled $\mathrm{InAs}/\mathrm{GaAs}$ Quantum Dots in Photonic Nanostructures: The Role of Nanofabrication

Single self-assembled $\mathrm{InAs}/\mathrm{GaAs}$ quantum dots are a promising solid-state quantum technology, with which vacuum Rabi splitting, single-photon-level nonlinearities, and bright, pure, and indistinguishable single-photon generation have been demonstrated. For such achievements, nanofabrication is used to create structures in which the quantum dot preferentially interacts with strongly confined optical modes. An open question is the extent to which such nanofabrication may also have an adverse influence, through the creation of traps and surface states that could induce blinking, spectral diffusion, and dephasing. Here, we use photoluminescence imaging to locate the positions of single $\mathrm{InAs}/\mathrm{GaAs}$ quantum dots with respect to alignment marks with $<5\mathrm{nm}$ uncertainty, allowing us to measure their behavior before and after fabrication. We track the quantum-dot emission linewidth and photon statistics as a function of the distance from an etched surface and find that the linewidth is significantly broadened (up to several gigahertz) for etched surfaces within a couple hundred nanometers of the quantum dot. However, we do not observe an appreciable reduction of the quantum-dot radiative efficiency due to blinking. We also show that atomic-layer deposition can stabilize spectral diffusion of the quantum-dot emission and partially recover its linewidth.

Figure 1

Experimental setup. Photoluminescence (PL) images are generated by exciting the sample simultaneously with 630- and 730-nm LEDs, with PL from the QDs and reflected 730-nm light from the alignment marks separated from unwanted light through 800-nm band notch filters (NFs) and a 700-nm long-pass filter (LPF), and sent into an electron-multiplying charge-coupled device (EMCCD) camera. PL spectra from specific QDs within the image are obtained by pumping them with a fiber-coupled Ti:sapphire laser whose wavelength is tuned to the wetting layer (spot diameter $<5\mu \mathrm{m}$), collecting the emission into a single-mode fiber, and sending it into a grating spectrometer. Time-correlated single-photon-counting (TCSPC) measurements are performed by sending the collected emission into a fiber-coupled tunable grating filter, with the filtered signal then going into either one SNSPD for lifetime measurements or into a $50:50$ fiber coupler and two SNSPDs for intensity autocorrelation measurements. Alternately, the filtered signal can be sent into a scanning Fabry-Perot (SFP) cavity for high-resolution spectroscopy, where the output of the SFP is again coupled into a SNSPD.

Figure 2

QDs in circular Bragg grating cavities. The optical properties of QDs before and after the creation of circular Bragg gratings are measured. (a),(b) Photoluminescence images of the QDs (a) before and (b) after device fabrication. The alignment mark separation is $50\mu \mathrm{m}$. (c) Photoluminescence spectrum, (d) photoluminescence decay, (e) emission linewidth, and (f)–(h) intensity autocorrelation recorded for one of the QDs before (black lines and circles) and after (red lines and circles) fabrication, under wetting layer excitation. The data in (d)–(h) are taken for the brightest emission line in the spectrum from (c) (near 911 nm). The photoluminescence decay data in (d) are fit to a monoexponentially decaying function, and a lifetime ${\tau }_{\mathrm{bef}}=1.5\pm 0.2\mathrm{ns}$ (${\tau }_{\text{after}}=0.6\pm 0.1\mathrm{ns}$) is extracted for the QD state before (after) fabrication, where the uncertainty is a one-standard-deviation value from the fit. The emission linewidth data (circles) in (e), measured by the SFP cavity, are fit to a Gaussian function (solid line) to determine the full width at half maximum listed on the plot, with the uncertainty being a one-standard-deviation value from the fit. The intensity autocorrelation measurement in (f) is recorded over a duration long enough to enable the evaluation of $g^{(2)}(\tau )$ out to a time delay as long as 1 s, and the time axis is given in a logarithmic scale. The intensity autocorrelation measurements in (g) and (h) are presented over a narrower range of time delays, to focus on the antibunching dip and potential presence of shorter timescale blinking. For both the before- and after-fabrication data in (g), the quantum efficiency (QE) is extracted from a three-level system fit to the data, as described in the main text, and the uncertainty value is a one-standard-deviation value from the fit. The quoted $g^{(2)}(0)$ values are determined by additionally deconvolving the SNSPD or TCSPC timing response from the fit. An enlarged view of the raw data, fit (no deconvolution), and fit including deconvolution is shown in (h), for the QD state after device fabrication. The before-fabrication data in (g) are vertically shifted up by 1.0 units for clarity. (i) Energy level diagram of the model used in blinking studies. $X_B$, $X_D$, and $G$, are the optically bright, dark, and ground states of the exciton, respectively. The QD is pumped from the ground state $G$ with a rate of $r_{\mathrm{up}}$, the bright state $X_B$ decays to the ground state with a spontaneous decay rate $r_{\mathrm{down}}$.

Figure 3

Behavior of QDs in etched nanopillars. Behavior of QDs before (black lines and circles) and after (red lines and circles) fabrication of nanopillars of varying diameter, under wetting layer excitation. (a) Scanning electron microscope (SEM) image, (d) intensity autocorrelation, and (g) emission linewidth for a QD in the smallest nanopillar fabricated, with a nominal diameter of 100 nm. (b) SEM image, (e) intensity autocorrelation, and (h) emission linewidth for a QD in a nanopillar with a nominal diameter of 300 nm. (c) SEM image, (g) intensity autocorrelation, and (i) emission linewidth for a QD in a nanopillar with a nominal diameter of 600 nm. The before-fabrication data in (d)–(f) are vertically shifted up by 1.2 units for clarity. The QE values in (d)–(f) are extracted from a three-level system fit to the data (solid lines), as described in the main text, and the uncertainty value is a one-standard-deviation value from the fit. The emission linewidth data in (g)–(i) are fit to Gaussians to determine the full width at half maximum listed on the plots, with the uncertainties being one-standard-deviation values from the fits. The scale bar displayed in (a) represents 1000 nm and is applicable to the SEM images in (b) and (c) as well.

Figure 4

Linewidth of QDs as a function of the proximity to etched surfaces. Compilation of QD emission linewidths under wetting layer excitation as a function of the maximum distance between the QD and an etched surface. For the circularly symmetric structures studied in this work, this is given by the radius of the central region of the circular Bragg grating (Fig. 2) and the radius of the nanopillar 80 nm below the top surface (Fig. 3). The emission linewidth is measured by the SFP method, and the error bars are given by the one-standard-deviation uncertainties from nonlinear least-squares fits of the data to Gaussian functions.

Figure 5

Stabilizing QD emission through atomic-layer deposition. Photoluminescence spectrum, recorded as a function of the time in 1-s acquisition intervals, for a QD (a) before and (b) after ALD of an ${\mathrm{Al}}_2{\mathrm{O}}_3$ capping layer. In each case, the emission intensity is normalized to the maximum value within the spectrum (scale bar shown to the right). (c) and (d) are the histograms of the photon counts as a function of the wavelength in (a) and (b), respectively, with a spectral bin size of 0.05 nm. (e)–(g) Emission linewidth for QDs in nanopillars before (red lines and circles) and after (blue lines and circles) ALD. The nanopillar diameters are (e) 100 nm, (f) 300 nm, and (g) 600 nm. The emission linewidth is measured by the SFP method, and the uncertainties are given by the one-standard-deviation uncertainties from nonlinear least-squares fits of the data to Gaussian functions.