Direct Epitaxial Growth and Deterministic Device Integration of high-quality Telecom O-Band InGaAs Quantum Dots on Silicon Substrate

TL;DR

This work demonstrates direct epitaxial growth of telecom O-band InGaAs/GaAs quantum dots on silicon via a GaP buffer and strain-reducing layer, paired with deterministic in-situ electron-beam lithography to place individual QDs into circular Bragg grating cavities. The QD-CBG devices achieve high photon extraction efficiency around the 40% level and exhibit exceptional single-photon purity at 4 K, with robust performance up to 77 K, highlighting practical cryogenic operation. The combination of material engineering and precise nanofabrication provides a scalable path toward silicon-based quantum light sources and large-scale, chip-integrated quantum photonics. The strong agreement between experiment and FEM simulations underlines the reliability of the design and positioning approach for future quantum information systems on silicon.

Abstract

Semiconductor quantum dots (QDs) are key building blocks for photonic quantum technologies, enabling practical sources of non-classical light. A central challenge for scalable integration is the direct epitaxial growth of high-quality emitters on industry-compatible silicon platforms. Furthermore, for long-distance fiber-based quantum communication, emission in the telecom O- or C-band is essential. Here, we demonstrate the direct growth of high-quality InGaAs/GaAs QDs emitting in the telecom O-band using a strain-reducing layer approach on silicon. Deterministic integration of individual QDs into circular Bragg grating resonators is achieved via in-situ electron-beam lithography. The resulting devices exhibit strong out-coupling enhancement, with photon extraction efficiencies up to $(40 \pm 2)\%$, in excellent agreement with numerical simulations. These results highlight the high material quality of both the epitaxial platform and the photonic nanostructure, as well as the precise lateral positioning of the emitter within 20~nm of the resonator center. At cryogenic temperature (4~K) and low excitation power ($0.027\times P_\text{sat}$), the devices show excellent single-photon purity, exceeding 99\%. Operation at elevated temperatures of 40~K and 77~K, compatible with compact Stirling cryo-coolers and liquid-nitrogen cooling, reveals robust performance, with single-photon purity maintained at $(88.4 \pm 0.6)\%$ at 77~K. These results demonstrate a practical and scalable route toward silicon-based quantum light sources and provide a promising path for cost-effective fabrication and seamless integration of quantum photonics with classical electronics, representing an important step toward large-scale, chip-based quantum information systems.

Figures (4)

• Figure 1: (a) Scanning electron microscopy (SEM) image of the investigated QD-CBG structure fabricated on a GaP/Si template. A CL intensity map spanning 1326–1345 nm is overlaid at the center of the central mesa. The inset shows a magnified view, where the white circle with cross indicates the geometric center of the mesa, and the green circle with cross marks the fitted position of the integrated quantum emitter. The emitter is offset from the mesa center by only (20 $\pm$ 2) nm. (b) CL spectrum of the QD-CBG indicating the spectral range of the relevant QD. Due to a rather high QD density of $\sim (1-5 \times 10^9$) cm$^{-2}$ multiple QDs with distinct emission energies were incorporated into the central mesa region of the structure.
• Figure 2: (a) Waterfall plot of µ PL spectra from the QD-CBG device presented in Fig. \ref{['Fig:CL']}. The measurements were performed under pulsed excitation at a wavelength of 1240 nm, with increasing excitation powers of 1, 3, 9, 34 µ W, at a temperature of 4 K. Four distinct emission lines are visible and are attributed to the neutral exciton (X, black), charged exciton (X$^\pm$, red), biexciton (XX, blue), and charged biexciton (XX$^\pm$, yellow), where the line assignment was performed using excitation and polarization dependent µ PL measurements presented in Fig. S5 panels (a) and (b). Insets show Gaussian fit of the X$^\pm$ line, yielding a linewidth of $(90 \pm 8)$ µ eV. (b) µ PL spectra of the same QD at an excitation power of 20 µ W, measured at two orthogonal polarization angles: 0° (black) and 90° (red). The insets show emission of the X and XX emission lines, from which an FSS of $(34.2 \pm 0.4)$ µ eV is extracted. The degree of linear polarization is plotted in blue, showing the characteristic polarization dependence of the neutral excitonic transitions. (c) Raw count rate of the X$^\pm$ emission as a function of excitation power, as measured with a SNSPD. A maximum count rate of $(0.72~\pm~0.03)$ MHz was observed at saturation, and was used to determine the PEE of the SPS.
• Figure 3: (a) Waterfall plot of µ PL spectra from the investigated structure recorded at excitation powers of 2.7, 7.1, 14.5, 30, 60, and 100 µ W. These correspond to the powers at which the second-order photon autocorrelation function $g^{(2)}(\tau)$ shown in panel (b) was measured. All measurements were performed under pulsed excitation at a wavelength of 1240 nm and at a temperature of 4 K. (b) Second-order photon autocorrelation function $g^{(2)}(\tau)$ measured for the charged exciton line at various excitation powers. The corresponding $g^{(2)}(0)$ values are indicated in the legend. The data demonstrate clean single-photon emission behavior, with a $g^{(2)}(0)$ value of $(0.007 \pm 0.001)$ at weak excitation and $(0.172 \pm 0.004)\%$ at saturation power.
• Figure 4: (a) Waterfall plot showing the µ PL emission from the investigated QD-CBG measured at four different temperatures of 4, 20, 40, 77 K under pulsed excitation. (b) The FWHM (black data points) and emission wavelength (red data points) of the XX$^{\pm}$ as a function of temperature. (c) Measured second-order photon correlation function $g^{(2)}(\tau)$ at 4, 20, 40, 77 K, demonstrating pure single-photon emission up to 77 K.