Introduction
The development of a global quantum internet hinges on the ability to generate, transmit, and process quantum information using photons at telecom wavelengths around 1550 nm (the C-band). At these wavelengths, existing fiber-optic infrastructure can be leveraged, and photons experience minimal loss during transmission. However, a major challenge has been the scalable fabrication of efficient single-photon sources operating in the C-band. This tutorial will discuss a groundbreaking approach to overcome this obstacle by enabling the high-throughput fabrication of quantum photonic devices based on epitaxial semiconductor quantum dots (QDs) emitting indistinguishable single photons in the telecom C-band.
Enhancing QD Emission for Optical Imaging
The first key innovation is the design of a planar sample structure that significantly enhances the out-of-plane emission from single QDs, enabling their localization via optical imaging. The structure consists of an InP slab containing a single layer of InAs QDs, positioned atop a SiO2 layer with an aluminum (Al) mirror bonded to a silicon (Si) carrier wafer (Figure 1a). This geometry increases the photon extraction efficiency from a single QD by more than 7 times compared to bulk InP samples, reaching an efficiency greater than 10% for a numerical aperture (NA) of 0.4.
Fig. 1 illustrates a structure featuring quantum dots (QDs) within optimized circular Bragg gratings (CBGs) for imaging. (a) Details the layer stack designed for the precise positioning of single InAs/InP QDs emitting in the C-band. (b) Shows the calculated CBG Purcell factor (dark red line) alongside photon extraction efficiency for numerical apertures (NA) of 0.65 (full squares) and 0.4 (empty circles), with an inset depicting the far-field emission pattern. (c) Presents a scanning electron microscope image of a CBG cavity etched in InP over SiO2.
To facilitate QD localization, the top InP layer is structured into a mesh of 50 μm x 50 μm fields separated by 10 μm gaps, with the field edges serving as alignment marks (AMs) for imaging. The epitaxially grown InAs/InP QDs exhibit a surface density of 3.1 x 10^8/cm^2, but only a fraction (around 4 x 10^5/cm^2) emit in the C-band due to the inhomogeneous broadening typical of self-assembled QDs.
Optimizing Circular Bragg Grating (CBG) Cavities
The authors optimized the design of circular Bragg grating (CBG) cavities to enhance the cavity figures of merit at 1550 nm, namely the collection efficiency at the first lens (η) and the Purcell factor (Fp). The simplified CBG geometry consists of a central mesa and only four external rings, reducing footprint and fabrication complexity while maintaining high η and Fp values (Figure 1c).
Calculations show that the optimized CBG design achieves a broadband η of nearly 62% and 82% at 1550 nm for NA of 0.4 and 0.65, respectively (Figure 1b). The Purcell factor reaches a maximum value of 18.1 with a quality factor of 110, mimicking the CBG cavity mode centered at 1550 nm.
Optical Imaging and QD Localization
To localize the C-band emitting QDs, the authors developed a near-infrared (NIR) imaging technique utilizing a wide-field bright microscope configuration (Figure 2a). The sample is mounted in an optical cryostat at 4.2 K and illuminated with a 660 nm continuous-wave laser. The QD microphotoluminescence (μPL) and scattered light from the field edges (AMs) are collected by an objective and projected onto a thermo-electrically cooled InGaAs camera with a 1550 nm bandpass filter.
Fig. 2 showcases microphotoluminescence imaging of quantum dots (QDs). (a) Describes the optical setup with a bandpass filter and numerical aperture. (b) Displays a microphotoluminescence map of a (50 × 50) μm² InP field with seven localized InAs QDs emitting in the C-band. (c) Features a scanning electron microscope image of circular Bragg gratings (CBGs) fabricated over localized and preselected QDs. (d, e) Provide histograms detailing the accuracy of QD localization (d) and cavity placement (e) with indicators for the 10th distribution percentiles. (f, g) Show microphotoluminescence map cross-sections with Gaussian fits for QD localization and alignment marks. (h) Zooms in on the QD signal, highlighting a Gaussian profile with a full width at half maximum of ~1.6 μm and a signal-to-noise ratio (SNR) of 12.
Figure 2b shows a representative image of a field, where QDs appear as individual bright spots with a full-width half-maximum (FWHM) of approximately 1.6 μm and Airy rings around them. The QD positions are determined by fitting Gaussian profiles to the QD emission and AM signals in vertical and horizontal cross-sections (Figures 2f, 2g, and 2h).
For the brightest 10% of QDs with a signal-to-noise ratio (SNR) greater than 15.5, the authors achieve a QD localization accuracy of 80 nm in two dimensions (2D) (Figure 2d). Considering the electron beam lithography (EBL) alignment accuracy of 40 nm, the overall accuracy of 2D CBG placement around pre-selected QDs is estimated to be 90 nm (Figure 2e).
After localizing suitable QDs, CBGs are fabricated around them using EBL with proximity error correction and optimized inductively coupled plasma-reactive ion etching (ICP-RIE) (Figure 2c).
Deterministic Process Yield and QD-CBG Coupling
The authors define the process yield as the ratio between the number of QD-CBG devices with QD emission spectra matching the CBG mode and the total number of CBGs investigated. They obtain an impressive yield of 30%, a significant improvement over the yield of approximately 0.7% that would be achieved with a random placement approach.
Some of the QDs exhibit broadened emission spectra due to the impact of surface states and point defects introduced during cavity fabrication. Temperature-dependent μPL studies confirm that these broadened emission lines follow the expected Varshni trend, ensuring their association with QD emission.
The coupling between the QDs and CBGs is evidenced by the observation of reduced emission decay times compared to reference QDs in the planar structure. For two exemplary devices (QD-CBG #1 and QD-CBG #2), the authors measure Purcell factors of 5.0 ± 0.4 and 3.8 ± 0.3, respectively, which are comparable to previous non-deterministic fabrication approaches but lower than expected due to imperfect spatial and spectral matching between the QDs and cavity modes.
Photon Extraction Efficiency and Single-Photon Purity
Using a superconducting nanowire single-photon detector (SNSPD) in a calibrated optical setup, the authors evaluate the photon extraction efficiency (η) of their devices. They obtain η values of (16.6 ± 2.7)% for QD-CBG #1 and (13.3 ± 2.2)% for QD-CBG #2 using an objective with NA = 0.4. These values are comparable to previous reports on probabilistically fabricated C-band QD-CBG devices.
The single-photon purity of the devices is assessed by analyzing the second-order autocorrelation function g^(2)(τ). For QD-CBG #2, the authors measure a raw antibunching value of g^(2)(0)_raw = (3.2 ± 0.6) x 10^-3 under quasi-resonant excitation, setting a new record for QD-based single-photon sources in the C-band and outperforming most previous reports.
Photon Indistinguishability and Coherence
In a remarkable achievement, the authors demonstrate triggered two-photon interference (TPI) experiments for InP-based cavity-coupled QDs emitting in the telecom C-band, a crucial requirement for quantum information processing applications. For QD-CBG #2, they measure a TPI visibility of V = (19.3 ± 2.6)% and a post-selected value of V_PS = 99.8^+0.2_-2.6% at zero time delay (Figure 3c). This result represents the highest reported photon indistinguishability for QD-based single-photon sources in the C-band.
Fig. 3 presents quantum optical experiments on device #2. Panels (a) and (b) illustrate the second-order autocorrelation function g(2)(τ) for triggered photons, with (a) showing results from above-band excitation and (b) from LO-phonon-assisted quasi-resonant excitation. Panel (c) displays a Hong-Ou-Mandel histogram for cross- and co-polarized photons, demonstrating their indistinguishability with a two-photon interference visibility of V=(19.3±2.6)%V=(19.3±2.6)%.
The photon coherence time (T2) is extracted from the width of the central dip in the Hong-Ou-Mandel histogram, yielding T2 = (103 ± 13) ps for the data shown in Figure 3c. Additionally, using an all-fiber-based Michelson interferometer, the authors measure coherence times of up to (62 ± 3) ps for QD-CBG #2 under weak continuous-wave above-band excitation.
These coherence times compare favorably with previous reports for InAs/InP QDs, indicating that the microcavity integration does not degrade the optical coherence of the emitted photons.
Discussion and Future Outlook
The work presented in this tutorial article represents a significant breakthrough in the high-throughput fabrication of efficient quantum photonic devices operating in the telecom C-band. By combining a novel sample design, advanced optical imaging techniques, and deterministic nanofabrication processes, the authors have demonstrated the ability to integrate pre-selected, indistinguishable single-photon emitters into optimized CBG cavities with record-high performance metrics.
However, there is still room for further improvement to push the boundaries of photon indistinguishability and coherence, which are crucial for quantum information processing applications. Several avenues for future research are proposed:
Stabilizing the QD environment: The observed photon decoherence is attributed to the coupling of QDs to their semiconductor environment via charge and spin noise. Integrating the QDs into p-i-n junctions or implementing electrical charge stabilization via gates could significantly increase photon coherence and indistinguishability.
Tuning QD emission: To address the challenge of inhomogeneous broadening in the QD ensemble, techniques such as strain tuning or the quantum-confined Stark effect could be employed to fine-tune the QD emission energy and match it to the cavity mode more precisely. This would require a different cavity design to accommodate the tuning mechanism.
Coherent optical pumping: Implementing coherent optical pumping schemes, such as two-photon resonant excitation, while avoiding excess charge carriers, could further improve the photon coherence time and indistinguishability by minimizing dephasing mechanisms.
Improving imaging and positioning accuracy: Enhancements to the optical imaging setup, such as using higher-NA objectives and increasing the overall microscope magnification, could further improve the QD localization accuracy and, consequently, the device yield and performance.
Material optimization: Careful optimization of the epitaxial growth processes and dry etching techniques could minimize the introduction of defects and non-radiative recombination centers, thereby improving the overall device efficiency and yield.
The InP material system used in this work also appears advantageous compared to GaAs-based devices, as it avoids issues such as threading dislocations and dangling bonds that can lead to non-radiative recombination and efficiency degradation.
This tutorial has presented a groundbreaking approach for the high-throughput fabrication of quantum photonic devices emitting indistinguishable single photons in the telecom C-band. The combination of innovative sample design, advanced imaging techniques, and deterministic nanofabrication processes has enabled the realization of highly efficient and coherent single-photon sources, paving the way for their implementation in quantum communication networks and distributed quantum computing protocols. With further improvements in yield, coherence properties, and photon indistinguishability, this technology holds the potential to become a key building block for the future quantum internet.
Reference
[1] P. Holewa, D. A. Vajner, E. Zięba-Ostój, et al., "High-throughput quantum photonic devices emitting indistinguishable photons in the telecom C-band," Nature Communications, vol. 15, no. 3358, Apr. 2024. Online. Available: https://doi.org/10.1038/s41467-024-47551-7.
Comments