Quantum dot single photon source on SiN integrated with coupled crossover waveguides

Hybrid integration of InAs/GaAs quantum dot (QD) single-photon sources (SPSs) is a promising approach for introducing quantum light into SiN photonic integrated circuits. However, the large refractive-index mismatch between GaAs and SiN poses a challenge for efficient optical coupling. Here, we propose and experimentally demonstrate hybrid integration of an InAs/GaAs QD-SPS on SiN using a coupled crossover waveguide structure. A photonic crystal nanocavity is employed for coupling QD emission into a GaAs waveguide, which efficiently transfers photons to a SiN waveguide at the crossover section. We observed Purcell-enhanced single-photon emission, on-chip propagation, and outcoupling through a SiN grating coupler.

💡 Research Summary

This paper presents a hybrid integration scheme that couples an InAs/GaAs quantum‑dot (QD) single‑photon source (SPS) to a silicon‑nitride (SiN) photonic integrated circuit (PIC) using a coupled crossover waveguide architecture. The authors first motivate the need for high‑performance SPSs in SiN platforms, noting SiN’s ultra‑low loss, broadband transparency, and CMOS compatibility, but also highlighting the difficulty of integrating efficient emitters due to the large refractive‑index mismatch (nGaAs ≈ 3.4, nSiN ≈ 2.0). Conventional approaches rely on long adiabatic tapers or evanescent coupling with reduced confinement, both of which suffer from alignment sensitivity and limited coupling efficiency.

To overcome these limitations, the authors propose a three‑stage design: (1) a two‑dimensional photonic‑crystal (2D‑PhC) nanocavity that funnels QD emission into a single cavity mode with a high β‑factor; (2) a PhC waveguide (W1) and a tapered PhC waveguide that connect the cavity to a GaAs wire waveguide, achieving a coupling efficiency h₂ ≈ 92.7 %; and (3) a coupled crossover waveguide where the GaAs wire waveguide evanescently couples to an underlying SiN waveguide at a small crossing angle (≈ 8°) and a sub‑200 nm gap. Full‑wave 3D FDTD simulations predict a near‑unity transfer efficiency h₃ ≈ 99.4 % for the crossover, with weak dependence on angle and wavelength, thus relaxing fabrication tolerances. The overall theoretical efficiency η = β·h₁·h₂·h₃ is estimated at 91.8 %, limited primarily by h₂.

Experimentally, the authors fabricate each component separately. The QD wafer is grown by molecular‑beam epitaxy, patterned into the 2D‑PhC nanocavities, and released as a suspended GaAs membrane. A 300 nm glass thin film is prepared to restore vertical symmetry. Commercial SiN waveguides with grating couplers are obtained from a foundry and planarized with spin‑on glass. Transfer printing is used to place the GaAs membrane onto the SiN chip, aligning the GaAs wire waveguide with the SiN waveguide at a rotation of ≈ 9° (adjusted from the design value to compensate for fabrication deviations). A glass cap is then placed over the assembly, and air bubbles are removed by gentle scrubbing.

Low‑temperature micro‑photoluminescence (PL) measurements at 55 K under 633 nm CW excitation reveal a cavity resonance at 917.9 nm and a nearby QD line at 918.3 nm. Spectra collected from the GaAs and SiN grating couplers are nearly identical, confirming that only cavity‑coupled emission reaches the waveguides. The measured cavity Q‑factor (≈ 2,070) is lower than the design value, leading to an experimental h₁ of ≈ 42.6 % (the reduction is attributed to a lower intrinsic Q of ≈ 3,600). By accounting for the simulated out‑coupler efficiencies (≈ 4.6 % each), the crossover transfer efficiency h₃ is experimentally estimated at ≈ 43 %, indicating that fabrication imperfections (e.g., waveguide bending, dimensional deviations) degrade the ideal performance.

Time‑resolved PL under pulsed excitation shows a QD lifetime of 0.2 ns when resonant with the cavity, three times faster than bulk QDs (≈ 0.6 ns), corresponding to a Purcell factor of 3.3. Using the measured Purcell enhancement and the photonic band‑gap suppression factor, the authors infer a β‑factor of ≈ 97 %. Combining the experimentally obtained β, h₁, and h₃ with the simulated h₂ yields an overall on‑chip SPS efficiency of ≈ 16.5 %.

Second‑order correlation measurements through the SiN grating coupler at 40 K under CW excitation give g^(2)(0) ≈ 0.55. After correcting for background cavity emission (≈ 12 % of the detected signal), the corrected value is g^(2)(0) ≈ 0.34, confirming non‑classical single‑photon emission albeit with limited purity due to residual cavity leakage.

The study demonstrates that coupled crossover waveguides can efficiently bridge materials with large index contrast, providing a compact (≈ 5 µm interaction length) and tolerant coupling mechanism that does not require long adiabatic tapers. When combined with high‑β PhC nanocavities, this approach enables Purcell‑enhanced, on‑chip single‑photon generation in SiN platforms. The current experimental efficiencies are limited by fabrication‑induced Q‑factor degradation and sub‑optimal waveguide tapering; however, the authors argue that further optimization of the PhC taper, improved lithographic control, and higher intrinsic cavity Q could raise the total efficiency well above 50 %. This work thus offers a practical pathway toward scalable quantum photonic circuits that integrate deterministic solid‑state SPSs with low‑loss SiN waveguides.