Uniform arrays of three shapes ($gauss$, $hat$, and $peak$) of GaAs microlenses (MLs) by wet-etching are demonstrated, ∼200 nm spatial isolation of epitaxial single QDs embedded ($\lambda$: 890–990 nm) and broadband ($\Delta \lambda \sim 80$ nm) enhancement of their quantum light extraction are obtained, which is also suitable for telecom-band epitaxial QDs. Combined with the bottom distributed Bragg reflector, the $hat$-shaped ML forms a cavity and achieves the best enhancement: extraction efficiency of 26%, Purcell factor of 2 and single-photon count rate of $7\times {10}^6$ counts per second at the first lens; while the $gauss$-shaped ML shows a broader band (e.g., longer $\lambda$) enhancement. In the MLs, single QDs with featured exciton emissions are observed, whose time correlations prove single-photon emission with multi-photon probability $g^{\left(2\right)}\left(0\right)=0.02$; some QDs show both biexciton $XX$ and exciton X emissions and exhibit a perfect cascade feature. This work could pave a step towards a scalable array of QD single-photon sources and the application of QD photon-pair emission for entanglement experiments.

We proposed a precise calibration process of Al _0.9Ga_0.1As/GaAs DBR micropillar cavity to match the single InAs/GaAs quantum dot (QD) exciton emission and achieve cavity mode resonance and a great enhancement of QD photoluminescence (PL) intensity. Light-matter interaction of single QD in DBR micropillar cavity (Q ∼ 3800) under weak coupling regime was investigated by temperature-tuned PL spectra; a pronounced enhancement (14.6-fold) of QD exciton emission was observed on resonance. The second-order autocorrelation measurement shows g⁽²⁾(0)=0.070, and the estimated net count rate before the first objective lens reaches 1.6×10⁷ counts/s under continuous wave excitation, indicating highly pure single-photon emission at high count rates.

We present a broadband enhanced single photon source device of Gaussian-shaped GaAs microlens (ML) arrays coupled with a gold bottom mirror hybrid structure. The extrapolated light extraction efficiency of (60.6 ± 1.5)% at maximum (for collection numerical aperture = 0.45 and based on the an actually measured raw detection efficiency of ∼0.775%) and a broad spectral band of 890–960 nm were obtained. The ML-Au structure with embedded epitaxial quantum dots exhibits a high single-photon purity of (96.6 ± 0.7)% [g(2)(0) = 0.034 ± 0.003] under pulsed excitation with 20% saturation. Wet-etched microlenses with smooth surfaces achieve simultaneous enhancement of both XX and X exciton emissions, which is prerequisite for promising entangled photon pair generation. This work proposes the bright potential of integrated hybrid structures to enable further photon pair entanglement applications.

Deep neural networks have been successfully applied to forward predicting optical response and inverse designing topological structure of nanophotonic devices. However, the existing deep learning based methods need sufficient simulated data to train the model effectively. For those devices with complex structures that containing many design variables, obtaining enough training data through numerical simulations will become extremely time-consuming. In order to reduce the requirement of large amounts of training data, we present a new deep learning approach based on the Capsule Network in this paper. By employing the proposed model, we have designed and verified a series of silicon-based wavelength demultiplexer with more than one thousand design variables. The numerical simulations validate that the trained model can both effectively predict the optical response with a fixed topological structure, and inverse design the approximate topological structure for a needed given optical response. Comparison with the classical convolutional neural networks show that our model can obtain nearly the same performance when using only 60% of the training data.