Heterogeneous integration for on-chip quantum photonic circuits with single quantum dot devices

Abstract

Single-quantum emitters are an important resource for photonic quantum technologies, constituting building blocks for single-photon sources, stationary qubits, and deterministic quantum gates. Robust implementation of such functions is achieved through systems that provide both strong light-matter interactions and a low-loss interface between emitters and optical fields. Existing platforms providing such functionality at the single-node level present steep scalability challenges. Here, we develop a heterogeneous photonic integration platform that provides such capabilities in a scalable on-chip implementation, allowing direct integration of GaAs waveguides and cavities containing self-assembled InAs/GaAs quantum dots-a mature class of solid-state quantum emitter-with low-loss Si₃N₄ waveguides. We demonstrate a highly efficient optical interface between Si₃N₄ waveguides and single-quantum dots in GaAs geometries, with performance approaching that of devices optimized for each material individually. This includes quantum dot radiative rate enhancement in microcavities, and a path for reaching the non-perturbative strong-coupling regime.Effective use of single emitters in quantum photonics requires coherent emission, strong light-matter coupling, low losses and scalable fabrication. Here, Davanco et al. stride toward this goal by hybrid on-chip integration of Si3N4 waveguides and GaAs nanophotonic geometries with InAs quantum dots.

Fig. 1

Principle of operation and device geometry. a Conceptual quantum photonic circuit composed of a passive waveguide network with a directly integrated GaAs nanophotonic device (exemplified by a nanowaveguide) containing a single quantum dot. A zoomed-in image of the GaAs device region (inside the dashed boundary box) shows details of the geometry and operation principle of the hybrid photonic integration platform. The light–matter interaction section of the device promotes efficient coupling between a confined electromagnetic field (in this case, a wave confined in a GaAs nanowaveguide) and a single-InAs QD embedded in the GaAs. Adiabatic mode transformers allow light from the QD in the light–matter interaction region to be efficiently transferred to a Si₃N₄ waveguide, and, conversely, also allow the QD to be accessed efficiently with resonant light guided by the Si₃N₄ waveguide. b, c Cross-sections of passive Si₃N₄ and active GaAs waveguides that form the core elements of the integration platform

Fig. 2

Nanophotonic design. a Left panel: electric field distribution for the fundamental TE GaAs supermode of the waveguide stack in Fig. 1c, with dimensions specified in the main text. Center panel: electric field distribution across the mode-transformer cross-section, for a GaAs mode launched at z = 0. At z ≈ 10 μm, the GaAs and Si₃N₄ guides are phase-matched, and power is efficiently transferred from the top GaAs to the bottom Si₃N₄ guide. Right panel: fundamental TE mode of the Si₃N₄ waveguide at the end of the mode transformer. b Coupling efficiency (β), as a function of GaAs width and emission wavelength, of photons emitted by a dipole located at x = 0 and 74 nm below the top surface, into the GaAs waveguide mode traveling in either the +z or −z direction. c Modal power conversion efficiency from the GaAs mode into the Si₃N₄ mode in (a) as a function of wavelength

Fig. 3

Device fabrication. a Top: schematic of the bonded wafer stack used in fabrication, consisting of a top III–V layer, containing InAs QDs, that is directly bonded on top of a Si/3 μm SiO₂/550 nm Si₃N₄ stack. Bottom: cross-sectional scanning electron microgaph (SEM) of bonded wafer stack. The ≈30 nm SiN_x layer was grown on the GaAs wafer surface prior to bonding. b Top: GaAs device lithography and etching steps. Bottom: optical micrograph of etched GaAs microring resonator and bus waveguide. Au aligment marks used for registered electron-beam lithography are visible. A wet etch protection resist mask (not depicted in the schematic—Supplementary Note 2) is also visible. c Top: Si₃N₄ waveguide lithography (aligned to the previously etched GaAs device) and etching steps. Bottom: optical micrograph of GaAs microring resonator and bus waveguide, and underlying Si₃N₄ waveguide. d False-color SEM of tip of mode-transformer geometry, common to both devices in (e, f). e False-color SEM of fabricated GaAs waveguide (yellow) on top of Si₃N₄ (red) waveguide. Blue regions are exposed SiO₂. Insets show details of the mode transformer end tip and the QD photon capture waveguide. f False-color SEM of GaAs microring and bus waveguide, and underlying Si₃N₄ waveguide. Inset shows details of the microring-bus waveguide evanescent coupling region

Fig. 4

Characterizing mode transformer efficiency with a photonic crystal reflector. a Schematic of a PhC reflector device. Top: top-view. Bottom: cross-section. Green arrows indicate pathways taken by the optical signal injected at the input port. R and T stand for PhC modal power transmission and reflection spectra, and T _dev, R _dev transmission and reflection spectra through the entire device, including lensed fibers. b False-color SEM of fabricated GaAs PhC reflector (yellow) on top of a Si₃N₄ (pink) waveguide, on top of exposed SiO₂ (blue). c FDTD-simulated TE modal transmission (T, purple) and reflection (R, yellow) spectra as a function of wavelength for the PhC (without mode transformers), for two different lattice constants a. d Experimental transmission spectra for various PhC reflectors with a = 250 nm (top) and a = 290 nm (bottom), normalized first to the transmission spectrum of a baseline Si₃N₄ waveguide (without GaAs sections), then to the mean transmission at wavelengths between 1250 and 1300 nm. Different colors indicate different devices. e Experimental transmission and reflection spectra for a PhC reflector with a = 290 nm, normalized to the transmission spectrum of a baseline Si₃N₄ waveguide. Gray areas have transmission <−15 dB in (c, d), <−20 dB in (e)

Fig. 5

Quantum dot-waveguide coupling. a Photoluminescence spectrum for a single QD inside a GaAs waveguide as in Fig. 3e, pumped with 1061 nm wavelength laser light. The PL is transferred to the bottom Si₃N₄ waveguide, and collected with a lensed optical fiber inside of a Liquid Helium flow cryostat (Supplementary Note 6). Sharp lines are exciton transitions from a single QD. Inset: Fit of PL peak at 1130.18 nm. b Second-order correlation as a function of time delay τ for the 1130.18 nm line. Circles mark experimental data, red line is a fit (Methods section and Supplementary Note 7). c Zoom-in of b near τ = 0. The blue curve and quoted g ⁽²⁾(0) are obtained from the red fit by deconvolving the detection time-response. Uncertainties for g ⁽²⁾(0) are 95% fit confidence intervals (two standard deviations). d Photoluminescence decay trace for the 1130.18 nm line. Gray dots are experimental data, the red line is a fit with a monoexponential function with lifetime τ _sp. The uncertainty is obtained from the fit and corresponds to a single standard deviation

Fig. 6

Quantum dot-cavity coupling. a Photoluminescence (PL) spectra as a function of wavelength from a QD ensemble pumped with laser light at 975 nm, emitting inside three different GaAs microring resonators. Peaks are whispering gallery modes (WGMs) with quality factor Q, which increases with the microring-bus waveguide gap width. b PL spectrum for a single QD coupled to a Q ≈ 1.1 × 10⁴ WGM. Inset: fit of cavity-coupled QD emission near 1126 nm. c, Left: second-order correlation g ⁽²⁾(τ) for the 1126 nm exciton line in b. Right: close-up near τ = 0. Circles are experimental data, red lines are a fit. The blue curve and quoted g ⁽²⁾(0) are obtained from the red fit by deconvolving the detection time-response. d PL spectrum for a single QD in a microring, coupled to a Q ≈ 6 × 10³ WGM. Circles: experimental data. Blue continuous line: fit. Dashed lines: fitting Lorentzians for the cavity and two excitons, X₁ and X₂. e g ⁽²⁾(τ) for X₁ in d. Inset: close-up near τ = 0. f, Left panel: PL spectra for varying spectral detuning Δ between X₁ and the cavity. Δ is obtained from fits as in d. All spectra are normalized to the intensity maximum at Δ ≈ −0.07 nm. The color scale indicates normalized intensity. Right panel: corresponding X₁ photoluminescence decay curves. Gray dots are experimental data, red (blue) lines are biexponential (monoexponential) decay fits. For biexponential fits, τ ₁ is the fast lifetime. g, Left panel: integrated intensity as a function of Δ for the filtered X₁ exciton contribution to the the PL spectra in f, obtained from Lorentzian fits as in d, normalized by the integrated intensity of the full fitted spectrum. Right panel: decay lifetimes for the fits in f, as a function of Δ. Open blue circles are the fast biexponential decay lifetimes, closed blue circles are the monoexponential decay lifetimes. Uncertainties for g ⁽²⁾(0), Δ and the X₁ magnitude are 95% fit confidence intervals (two standard deviations). Lifetime uncertainties are single standard deviations from the exponential decay fits