Indistinguishable photons from deterministically integrated single quantum dots in heterogeneous GaAs/Si 3 N 4 quantum photonic circuits
Peter Schnauber, Anshuman Singh, Johannes Schall, Suk In Park, Jin Dong Song, Sven Rodt, Kartik Srinivasan, Stephan Reitzenstein, Marcelo Davanco
IIndistinguishable photons from deterministically integrated singlequantum dots in heterogeneous GaAs/Si N quantum photoniccircuits Peter Schnauber, Johannes Schall, Sven Rodt, and Stephan Reitzenstein
Institute of Solid State Physics, Technische Universit¨at Berlin, Berlin, Germany
Anshuman Singh
National Institute of Standards and Technology, Gaithersburg, MD, USA andMaryland NanoCenter, University of Maryland, College Park, USA
Suk In Park and Jin Dong Song
Center for Opto-Electronic Convergence Systems,Korea Institute of Science and Technology, Seoul, South Korea
Kartik Srinivasan
National Institute of Standards and Technology, Gaithersburg, MD, USA andJoint Quantum Institute, NIST/University of Maryland, College Park, USA
Marcelo Davanco ∗ National Institute of Standards and Technology, Gaithersburg, MD, USA a r X i v : . [ phy s i c s . a pp - ph ] M a y bstract Silicon photonics enables scaling of quantum photonic systems by allowing the creation of ex-tensive, low-loss, reconfigurable networks linking various functional on-chip elements. Inclusion ofsingle quantum emitters onto photonic circuits, acting as on-demand sources of indistinguishablephotons or single-photon nonlinearities, may enable large-scale chip-based quantum photonic cir-cuits and networks. Towards this, we use low-temperature in situ electron-beam lithography to de-terministically produce hybrid GaAs/Si N photonic devices containing single InAs quantum dotsprecisely located inside nanophotonic structures, which act as efficient, Si N waveguide-coupledon-chip, on-demand single-photon sources. The precise positioning afforded by our scalable fab-rication method furthermore allows observation of post-selected indistinguishable photons. Thisindicates a promising path towards significant scaling of chip-based quantum photonics, enabledby large fluxes of indistinguishable single-photons produced on-demand, directly on-chip. ∗ [email protected] . INTRODUCTION In the development of advanced photonic quantum information systems, exemplified byvarious devised schemes for quantum simulation [1] and communication [2], the ability toproduce, manipulate and detect multiple identical photons in multiple spatial modes is anecessity. Integrated photonics has a great potential to fulfill such tasks, by allowing thecreation of compact, complex, chip-scale photonic circuits that can implement phase-stable,reconfigurable, and integratable interferometric networks for linear optical operation at thesingle-photon level[3, 4].Silicon-based photonic integrated circuits are most promising for large system scaling, asfoundry services offer the fabrication of user-designed, high quality integrated circuits com-prising thousands of elements on shared wafer projects [5]. Importantly, photonic losses inon-chip waveguides and related linear elements - e.g., beam splitters and combiners, phasedelay paths and linear filters - can be reduced sufficiently through design and process control,to enable significant scaling of integrated quantum photonic systems. Adding to a favor-able set of characteristics, the introduction of solid-state quantum emitters [6] into silicon-based integrated quantum photonic circuits may yield unprecedented system scalability andfunctionality. Quantum emitters can e.g. act as high-rate, on-demand sources of indistin-guishable single photons [7–9], providing the large on-chip photon fluxes necessary for linearoptical quantum systems such as boson sampling simulators [10, 11]. Emitters with opticallyaddressable spins may furthermore act as stationary qubits in photonic networks, and, alongsimilar lines, single-photon nonlinearities in single-emitter quantum cavity-electrodynamicsystems [12, 13] may allow networks of deterministic quantum logic gates to be implemented.In terms of silicon-compatible quantum light emitters, color centers in SiC have beenshown to display promising optical and spin properties in a silicon-based material that isamenable to photonic integration [14, 15]. Equally attractive emitters have not yet been iden-tified in silicon or Si N . As a result, efforts to incorporate quantum emitters into photoniccircuit platforms based on such materials have relied on hybrid integration with guest/hostmaterial systems that provide the desired optical properties. For instance, nitrogen-vacancy(NV) centers in diamond [16], epitaxially grown InAs quantum dots (QDs) in GaAs [17], andInAsP QDs in InP [18] have been integrated with Si N waveguides. In addition, InAs QDsin InP [19] on silicon-on-insulator, InAs QDs on GaAs-on-insulator [20], as well as carbon3anotubes [21] and 2D materials [22] on silicon have also been shown. To date, however,Stranski-Krastanov (SK) self-assembled QDs have generally demonstrated superior opticalcoherence [7, 8, 23], commonly evidenced by high degrees of two-photon interference, whichis a pre-requisite to enable photon-photon interactions, e.g. in quantum gates. Thus, SKQDs currently offer the most favorable prospects for integrated quantum photonics.One important drawback of the SK growth mode is the QDs’ random spatial distributionacross the growth surface. This imposes considerable challenges for maximizing light-matterinteractions through nanophotonic geometries [24], which must be leveraged to create anefficient optical interface between the QD and the photonic circuit [17]. In such geometries,QDs must be positioned with high precision within the nanophotonic geometry, to maximizecoupling to specific spatial modes, and at the same time the QD must be sufficiently far awayfrom etched surfaces, to minimize effects detrimental to the QD coherence [25]. A number ofmethods have been developed for precisely locating individual SK QDs on a wafer surface,allowing subsequent fabrication of nanophotonic devices precisely located around selecteddots [26–30]. However, no hybrid devices have so far been demonstrated through suchtechniques [17–20].Here, we employ cryogenic cathodoluminescence (CL) spectroscopy and in situ electronbeam lithography (EBL) [31, 32] to deterministically create hybrid integrated quantumphotonic devices containing precisely positioned, pre-selected, individual InAs SK quantumdots. Our devices are based on a heterogeneous photonic integrated circuit platform, whereGaAs devices containing positioned QDs are produced on top of Si N waveguides [17]. Wedemonstrate triggered emission of single photons from a single QD in a hybrid nanowaveg-uide, coupled directly into a Si N waveguide. In addition, we report the observation oftwo-photon interference, which indicates generation of post-selected indistinguishable pho-tons from a single device. This is achieved through precise positioning of the single emitter atmaximum distance from etched surfaces through our deterministic approach. Single-photonindistinguishability is essential for quantum photonic systems based on linear optical oper-ations, and yet has never been reported in hybrid QD-silicon platforms. Our unprecedentedresults indicate good prospects for generation of on-demand indistinguishable photons in ascalable hybrid silicon photonic platform. 4 ) b)c) d) e) N Si WG SiO m o d e t r a n s f o r m e r p h o t o n c o l l e c � o n G a A s W G t a p e r d e v i c e preselected QD g)f) le � overGaAs N Si WG SiO GaAs marker 5µmGaAs WG taper device in-situ EBL
GaAs WG taper device 2µm N Si N Si WG SiO Si substrateSiO N Si GaAs QDsSiO
FIG. 1. a) Schematic layout of the wafer-bonded sample stack. b) Schematic GaAs-Si N devicedesign: The preselected InAs QD is hosted in a GaAs nanowaveguide that collects the QD’semission. The emission is then coupled into the Si N -SiO WG using mode transformers. c) - e):Visualization of key sample fabrication steps: c) in situ
EBL of a GaAs nanowaveguide patternand markers aligned to a QD which was preselected using low temperature CL spectroscopy. d)GaAs nanowaveguide and markers on Si N after etching the GaAs patterns and removing excessGaAs. e) fully-fabricated GaAs-Si N -SiO WG device. f) false-color optical micrograph of fully-fabricated device QD 1. g) false-color SEM image of device QD 1, showing the GaAs WG taper(yellow) and the Si N WG (pink).
II. DETERMINISTIC SAMPLE FABRICATION
Efficiently interfacing individual QDs embedded in a III-V host with Si N photonicwaveguides has two requirements. First, the III-V host must be carefully shaped to supportspatial modes into which emission from the individual QD can be efficiently funneled, andthese modes must be simultaneously and efficiently coupled to Si N WG modes [17]. Sec-ond, the QD must be located with high precision within the III-V host for optimal couplingto the desired spatial mode [24]. In the hybrid devices of ref. [18], individual InAsP QDs weregrown with high spatial precision within InP nanowires. Because such nanowires are gener-ated through self-assembled growth, geometrical control of the QD-hosting InP is limited,which results in less efficient QD-waveguide interfaces - for instance, limiting the ability to5reate small mode-volume cavity modes for Purcell enhancement [24]. Other groups [19, 20]have relied on lithography and etching to produce high-resolution, geometrically complexnanophotonic hosting geometries for embedded, randomly positioned SK QDs. No attempthas been made to position individual QDs precisely within the hosting geometries, however.In addition, in both demonstrations [19, 20], QD-containing III-V devices were producedseparately from the silicon photonic chip, then transferred onto the latter via pick-and-placeprocesses, which offer limited scalability. The heterogeneous integration technique usedin ref. [17] and this work, which starts from the wafer bonded stack in Fig. 1 a), allowsfor the creation of InAs SK QD-containing, complex GaAs nanophotonic devices directlyintegrated with Si N waveguides. Here, this technique is combined with the cathodolu-minescence spectroscopy and in situ electron-beam lithography of refs. [31, 32], which hasbeen shown to provide QD positioning accuracies of 34 nm [33]. To realize tapered GaAsnanowaveguides for mode transformers, it is crucial to achieve device features sizes in the50 nm to 100 nm range. Using the high patterning resolution of the in situ EBL along withproximity-correction grey-scale writing [34], feature sizes down to 50 nm [33] can be reliablyachieved. The combination of heterogeneous integration [17] with in situ
EBL [32] thereforeoffers a deterministic, high resolution and scalable, purely top-down fabrication scheme. Acomparison to other deterministic manufacturing approaches can be found in ref. [32].The in situ
EBL technique has been used to produce a variety of photonic devices withdeterministically positioned QDs, all on semiconducting (GaAs) [31, 32, 34, 35] or conducting(gold) substrates [36, 37]. For samples containing insulating layers like Si N and SiO , as inthis work, charging poses a major challenge. When the electron beam irradiates an insulatingsample, the induced charge is not drained to the scanning electron microscope (SEM) groundand charges accumulate. This can already be a problem in standard EBL with positive toneresist doses on the order of 50 µ C cm , which require conductive polymers or thin metal films to bedeposited onto the EBL resist. It becomes more severe for the in situ EBL which operates atelectron doses in the 5000-50000 µ Ccm range. Small amounts of charging lead to electron beamdeviations, and the fast build-up of large charge numbers leads to unstable beam jumps,which inhibit CL mapping or EBL patterning. High acceleration voltages reduce the amountof charge deposited in thin insulating layers [38], but likewise the number of electron-hole-pairs created in the GaAs layer decreases and QD excitation becomes inefficient. The presentwork demonstrates that a sufficient balance can be achieved, allowing for high resolution QD6ositioning and pattern definition on heterogeneous substrates with thin insulating layers.We fabricated a sample containing hybrid on-chip single-photon sources depicted schemat-ically in Fig. 1 b), with varying geometrical parameters. Such sources are composed by astraight GaAs nanowaveguide section (labeled ”photon collection”) which hosts the prese-lected SK InAs QD and captures its emission into guided modes that are strongly confinedin the GaAs ridge. Such GaAs-confined modes are subsequently converted into Si N modesby adiabatic mode transformers implemented at the two ends of the photon capture section.As discussed in ref. [17], such a geometry may offer QD-Si N coupling efficiencies in excessof 90 %, through a combination of high photon capture probabilities and modal transformerefficiencies. In our samples, the photon collection regions consisted of 5 µ m long straightGaAs ridges of widths ¿400 nm, on top of a wide Si N slab region of the same length, asseen in Figs. 1 e) to f). Over the mode transformer sections, the GaAs ridge was taperedfrom the central width down to 100 nm at the tip, whereas the underlying Si N waveguidemaintained a width of ≈
650 nm. The substrate cladding for the entire device consisted ofthermal SiO . A 100 nm thick spacer of SiO is also featured between the GaAs and Si N throughout the sample.Fabrication started with a low-temperature plasma-bonded wafer stack consisting of aSi substrate, 3 µ m thermal SiO , 250 nm low-pressure chemical vapor deposition (LPCVD)Si N , 100 nm plasma-enhanced chemical vapor deposition (PECVD) SiO and 190 nm ofGaAs containing SK InAs QDs at its center [17], as shown in Fig. 1 a). The plasma-bondedGaAs forms a uniform layer that spans over areas of several tens of square millimeters.GaAs nanowaveguides were deterministically patterned at the position of single preselectedInAs QDs through in situ EBL [31, 32] as follows: The sample was coated with a dual-toneEBL resist which exhibits high contrast and high resolution at cryogenic temperatures [39],mounted onto a custom-made liquid helium flow cryostat inside an SEM, and cooled to 7 K.In the chamber, the sample was excited by an electron beam, and emitted light was collectedthrough an NA = 0.8 elliptical mirror, then dispersed in a grating spectrometer. Throughthis process, spatially resolved CL spectrum maps over regions of hundreds square micronswith 500 nm steps were taken, while the electron dose remained well below the negative-toneonset-dose. Applying an acceleration voltage of 20 kV and a beam current of 0.5 nA, sampleand resist charging was minimized while still operating at a high QD excitation. Comparingthe CL imaging of sample regions with bonded GaAs to those without, we find that less7harge accumulates in regions with GaAs, indicating beneficial charge carrier diffusion inthe GaAs. On-the-fly spectral analysis of the CL maps yielded positions and spectra ofsuitable, individual QDs within a few minutes. Immediately after localization, proximity-corrected grey-scale in situ
EBL [34], as illustrated in Fig. 1 c), was performed at 7 K todefine 400 nm to 800 nm wide and altogether 45 µ m long symmetrical GaAs waveguide taperpatterns (see Supplementary Material) aligned to the identified QDs with an uncertainty ofabout 55 nm. The uncertainty is slightly higher than the previously achieved 34 nm [33] dueto a high QD density in the sample, which reduces the dynamic range of the 2D Gaussianfits used for QD localization if other emitters are spectrally and spatially within the fittingrange. In addition, four L-shaped marker patterns, also aligned to the QD positions, werewritten outside the CL mapping area. The sample was then brought to room temperature,developed, and the resist pattern was transferred into the GaAs with an inductively coupledplasma reactive ion etch. Unpatterned GaAs was subsequently removed in a nitric acid/cericammonium nitrate aqueous solution, using a resist mask to protect the etched device andalignment mark areas, resulting in the intermediate sample layout shown in Fig. 1 d). TheGaAs alignment marks (visible in the micrograph in Fig. 1 f)) were included for alignedEBL to be performed in a commercial 100 kV system. The Si N waveguide patterns weredefined as in ref. [17] and transferred via reactive-ion etching into the Si N (Fig. 1 e)), andthe sample was cleaved to allow endfire coupling to optical fibers inside of a cryostat.Figure 1 f) shows a false-color optical micrograph of a finalized device. The positionedQD is located at the center of the 5 µ m long, straight portion of the GaAs nanowaveguide.Figure 1 g) shows a false-color SEM image of the same device, in which an unintended,vertical displacement of ≈
60 nm between the fabricated GaAs and Si N waveguides isapparent. We note the GaAs marker dimensions and positions relative to the QD weremanually calibrated in the in situ EBL system, which likely led to write field distortion andscaling errors. As a result, the markers featured imperfections that disallowed nanometerprecision automatic alignment in the 100 kV EBL system. Vertical displacements were sys-tematically observed in all devices, and are likely due to the manual alignment procedureused for the fabrication of the Si N layer, based on visual information from SEM scans andinterferometric stage position readout. Incorporating the CL mapping system into state-of-the-art EBL equipment promises nanometer alignment fidelity throughout the whole processin the future. 8otably, some of the GaAs WG tapers that were fabricated with in situ EBL showeda bending at their left-hand side as shown in Fig. 2 b) and c), while their right-hand sideand the overall WG position remained unaffected. This bending stems from minor chargingwhich occurs in parts of the sample and is explained in more detail in the SupplementaryMaterial.
III. POST-FABRICATION DEVICE CHARACTERIZATION
The successful fabrication of heterogeneous waveguide devices and the integration of pre-selected QDs was checked through conventional microscopy, scanning electron microscopy aswell as micro-photoluminescence ( µ PL) and CL spectroscopy on the fully fabricated sample.For µ PL, the sample was mounted inside a closed-cycle cryostat and cooled down to 7 K.An off-resonant continuous wave (CW) laser at 821 nm was focussed through an NA=0.28microscope objective onto the designated QD position inside the GaAs WG. QD emissionwas collected from the Si N WG endface through a lensed single mode fiber. CL mapsand spectra are taken at a temperature of 7 K, an electron beam current of 4 nA and anacceleration voltage of 20 kV.Figure 2 a) - c) show microscope images of three example WG devices QD 1, QD 2and QD 3 with a GaAs nanowaveguide width of ≈
620 nm that have been successfullypositioned on preselected QDs. CL intensity maps of each QD taken during fabricationare visible in Fig. 2 d) - f) and after fabrication in Fig. 2 g) - i). The maps show theCL intensity integrated over those spectral regions that were used to determine the QDpositions during fabrication, marked by green dashed lines in Fig. 2 j) - l). The pre- andpost-fabrication CL maps for QD 1, QD 2 and QD 3 (Figs. 2 d) - i)) display spatiallymatching, localized high intensity spots, marked by red pixels within the pink guide tothe eye, that indicate successful, deterministic waveguide placement around the preselectedindividual QDs. Figure 2 j) - l) show CL spectra with an exposure time of 50 ms during(blue) and after (grey) fabrication from representative pixels at the positioned QD location,as indicated by each intensity maximum inside the pink guide to the eye in Fig. 2 d) - i). µ PL spectra were obtained with an exposure of 1 s after fabrication, but before the post-fabrication CL mapping, and are displayed in red. These spectra, which display broad CL,as well as sharp µ PL emission from individual QDs within the same spectral ranges, further9 ) e)h) f)i)d)
QD 1 QD 2 QD 3 a) c)b) y ( µ m ) y ( µ m ) I n t e n s i t y ( a . u . ) I n t e n s i t y ( a . u . ) x (µm)
12 1060 30 5060 10 30 50 y ( µ m ) y ( µ m ) I n t e n s i t y ( a . u . ) I n t e n s i t y ( a . u . ) x (µm)
12 1060 30 5060 10 30 50 j) k) l) x (µm) y ( µ m ) y ( µ m ) I n t e n s i t y ( a . u . ) I n t e n s i t y ( a . u . )
50 30 5050 10 30 50 102
900 910 920
Wavelength (nm) I n t e n s i t y ( a . u . ) I n t e n s i t y ( a . u . )
900 910 920
Wavelength (nm)
Wavelength (nm) I n t e n s i t y ( a . u . ) CL Fab.CL Post-Char.µPL Post-Char.
FIG. 2. a) - c): False-color optical micrographs showing devices QD 1-3, with GaAs colored inyellow for better contrast. The left-hand side of the GaAs WGs in b) and c) is bent downwardsdue to charging. d) - f): CL maps taken during in situ
EBL to locate QDs 1-3. The pink circlesmark the QD emission patterns that were used for QD localization. g) - i): CL maps taken onthe fully-fabricated devices. Here, pink circles mark the GaAs WG center. The CL intensity in allmaps d) - i) is integrated over those spectral regions, that were used to localize the QDs duringthe in situ
EBL. These spectral regions are highlighted in j) - l) by green dashed lines. j - l): CLspectra taken during (after) fabrication in blue (grey) along with µ PL spectra after fabrication inred. µ PL spectra are due to differences in QD excitation conditions, where the various QDexcitonic complexes are populated with different efficiencies. Also, in CL, the large injectedcharge density necessary for sufficient luminescence to be produced leads to spectrally broadQD lines. While QD 3 produced spectrally aligned pre-fabricated CL and µ PL emission, thepost-fabrication CL emission displays considerably less intensity within the same spectralrange. We believe that the repeated thermal cycles to which the sample was subjected,between the µ PL characterization and subsequent post-fabrication CL mapping, causedsuch degradation. Nonetheless, the pre-and post-fabrication CL maps still display the samelocalized intensity spots at the selected QD location. We note that all µ PL measurementswere obtained by pumping the center region of the QD devices. Moving the pumping spotto other locations caused the emission spectra to vary considerably, as expected.
IV. SINGLE-PHOTON EMISSION PROPERTIES
The finalized sample was tested inside of a closed-cycle cryostat at 7 K. The sample wasexcited from the top using CW or pulsed tunable lasers through an NA=0.28 microscopeobjective, and the emission was captured from the Si N WGs at the sample facet usingaligned lensed single mode fibers inside the cryostat[17]. Figure 3 a) shows a µ PL spectrumfrom the fabricated device housing QD 3, illuminated by a CW free-space laser beam tuned to ≈
904 nm, exciting the p-shell of the positioned QD 3 and giving rise to a narrow emission lineat ≈ . µ W. The PL intensity was obtained as the peak areaof a Gaussian model fit to each spectrum of the power series. Interestingly, a red-shift of theemission line is observed for increasing pump powers, see Fig. 3 c). This shift is likely due toa local increase in temperature in the GaAs WG, which lies on top of thermally insulatingSi N -SiO . To investigate this hypothesis, we assume a linear temperature increase withexcitation power due to linear absorption in the GaAs. With this model, we are able tofaithfully fit the power-dependent spectral position of the QD line with a Bose-Einsteinphonon law that describes the temperature dependence of the semiconductor bandgap [40],confirming a temperature-related effect. Moving towards strictly resonant excitation inthe future, sample heating can be neglected, as the necessary pump powers are orders of11agnitude lower than in p-shell excitation. More details and a comparison to temperatureseries measurements are given in the Supplementary Material.We next measured the lifetime of the ≈ . ≈
904 nm. The excitation laser was suppressedwith a ≈
500 pm free spectral range fiber-coupled grating filter with a transmission of ≈
60 %in addition to an edge pass filter. The filtered PL was detected on a superconductingnanowire single-photon detector (SNSPD) with a timing resolution of ≈
90 ps. The naturallogarithm of the data is plotted in the inset of Fig. 3 a) and shows a double exponentialdecay. By fitting two linear curves to the natural logarithm of the data, we extract twodecay constants of τ r = (1 . ± .
04) ns and τ r,2 = (3 . ± .
29) ns (uncertainties are standarderrors). The slower decay hints at a recapture process often seen in QDs [41]. To evaluatethe single-photon emission purity of QD 3, pulsed excitation close to saturation was used, asindicated by the red dot in Fig. 3 b). The collected PL was split in a 50/50 fiber beam splitterand then detected by two SNSPDs (overall timing resolution ≈
130 ps) in a Hanbury-Brownand Twiss (HBT) type configuration. Detection coincidences with time delay τ were trackedwith a 64 ps bin size. The normalized autocorrelation curve g (2) ( τ ) is depicted in Fig. 3 d).The data was fitted with a two-sided exponential decay function convolved with a Gaussianthat represented the experimental timing resolution, using a Poissonian statistics maximumlikelihood estimator [42] (see Supplementary Material). Without any corrections, we obtaina conservative estimate of g (2) (0) = 0 . ± .
04 (uncertainty marks the 95 % confidenceinterval), clearly showing that the positioned QD 3 emits triggered single photons into theSi N WG.Next, we estimate the emission efficiency of our hybrid single photon source from theQD into the lensed fiber. During the HBT measurement, a combined photon stream of ≈
50 kHz is measured on the detectors. Taking into account the grating filter and fibertransmission as well as the detector efficiency, we estimate a setup efficiency from collectionfiber to detector of η Setup ≈ .
09. Assuming 100 % quantum efficiency, the QD-to-fiberefficiency is η Source ≈
50 kHz / (76 MHz · η Setup ) ≈ . N guides, and12 )c) I n t e n s i t y E n e r g y ( e V ) Power (µW) -40 -20 0 20 40
Time delay (ns) N o r m . C o i n c i d e n c e s d) a) I n t e n s i t y ( c o un t s / s )
916 920918 02000400060008000
Wavelength (nm) L o g . i n t . ( a . u . ) Time (ns)
FitLife � me FIG. 3. a) PL spectrum of QD 3 in CW p-shell excitation. Inset: Natural logarithm of the excitedstate lifetime trace with two linear fits returning τ r = 1 .
39 ns and τ r,2 = 3 .
15 ns. b) PL intensity ofthe 916.3 nm line over excitation power, in 1 dB steps. c) Peak energy over excitation power (blue)and fit to data (red), showing a redshift with increasing excitation power. d) Pulsed excitationautocorrelation curve (blue) with fit to data (red) yielding g (2) (0) = 0 . coupling between the Si N WG and collection lensed fiber (details in the SupplementaryMaterial). Modelling the QD as linear dipole, we obtain η Source,1 ≈ η Source,2 ≈ η Source achievable withthe WG examined here. We note that the simulated QD to GaAs WG coupling efficiency is ≈
42 % for one propagation direction, and the overall coupling efficiency can potentially besignificantly increased, by improving the fiber-to-Si N waveguide coupling, the adiabaticmode transformer design, and introduction of a high reflectivity mirror on the back port ofthe GaAs waveguide [17]. V. EMISSION OF INDISTINGUISHABLE PHOTONS IN POST-SELECTION
While the transfer of single-photon emission from QDs into silicon photonic circuits hasbeen shown before using various sample preparation methods[17–20], the degree of indis-tinguishability of the emitted photons has never been evaluated in such hybrid systems. Infact, the close presence of dissimilar material interfaces to the QD introduces defect-richregions that can reduce the QD coherence through electronic interaction with surface ordefect states [25], inhibiting emission of indistinguishable photons. A high degree of single-photon indistinguishability, however, is necessary for quantum photonic systems based on13inear optical operations, and can serve as a baseline criterion for evaluating the quality ofthe fabrication process, regarding preservation of the QD coherence. In the following, weevaluate the coherence of photons emitted by the fully-fabricated device QD 3, pumped byCW laser light at ≈
904 nm, close to saturation with an excitation power of 123 µ W, markedby the red dot in Fig. 3 b).As a first indicator of photon coherence, we measured the linewidth of the ≈ . ≈
200 and free-spectral range of 40 GHz, and detecting the filteredsignal with an SNSPD. The recorded spectrum is shown in Fig. 4 a), where it is apparentthat QD 3 has a linewidth of ≈ V = (2 . ± .
19) GHz full-width at half-maximum(FWHM). A Lorentzian component of ∆Γ L = (1 . ± .
27) GHz FWHM suggests homo-geneous broadening beyond the Fourier limit of ≈ . G = (1 . ± .
26) GHz FWHM suggestsinhomogeneous linewidth broadening due to spectral diffusion [43]. All uncertainties markthe 95 % confidence interval. In our hybrid device both homogeneous and inhomogeneousbroadening values are comparable to those observed under p-shell excitation in refs. [42, 43]and resonant excitation in ref. [44] in purely GaAs-AlGaAs-InAs-based samples. From theLorentz linewidth we can extract an upper bound for the photon coherence time, yielding τ c, FPI = 1 /π ∆Γ L = (0 . ± .
03) ns [45].Next, we measured the two-photon interference (TPI) contrast of subsequently emittedphotons of the same ≈ . δτ ≈
10 ns arm imbalance. A variable half-wave plate in the long arm was used toalign photons parallel or orthogonal to each other at the HOM beamsplitter and coincidenceswere measured with SNSPDs. The full HOM setup is detailed the Supplementary Material.14 ) N o r m . c o i n c i d e n c e s Time delay (ns) N o r m . c o i n c i d e n c e s Time delay (ns) -45 0 45-15 15 b) parallel -45 0 45-15 15 orthogonal orthogonal parallel a) C o un t s Rela � ve frequency (GHz) -4 -2 0 2 4 R e s i du a l s -4010100 FPIVoigt fi t FIG. 4. a) top: FPI spectrum of QD 3 (blue) with a fitted Voigt profile (red). Bottom: Voigtfit residuals. b) TPI coincidence curve (blue) and fitted model (red) for parallel (orthogonal)configuration is shown in the left (right) hand side panel. c) Same as b), including error bars andmagnified around zero time delay for clarity.
The raw HOM autocorrelation traces for parallel and orthogonal photons are depicted inblue in Fig. 4 b) and magnified around τ = 0 in Fig. 4 c). The error bars in Fig. 4 c) arethe 1 / √ N Poissonian uncertainty for each time bin with N counts. At τ = 0, the trace forparallel-polarized photons is clearly below 0.5 and below the orthogonal trace, marking theemission of indistinguishable photons. The bunching around τ = 0 hints blinking due tocoupling of the QD to a dark state [44, 46].To extract an estimate for the coherence time τ c, HOM and the two-photon interferencevisibility V we follow Ref. [47] and model the parallel and orthogonal coincidence traces withthe functions g (2)HOM , (cid:107) ( τ ) and g (2)HOM , ⊥ ( τ ), respectively, where we include a HBT autocorrelationcurve g (2) ( τ ) with two separate two-sided exponentials to describe the bunching around zerotime delay (see Supplementary Material for details on all functions). Fitting g (2)HOM , ⊥ ( τ ) tothe orthogonal case of Fig. 4 b), we can extract g (2) ( τ ). We use this to fit g (2)HOM , (cid:107) ( τ ) to theparallel case of Fig. 4 b), with τ c, HOM and V as the only free parameters. Both fits are plottedalong the coincidence data in Fig. 4 b) and c). We obtain τ c, HOM = (0 . ± .
12) ns and V = 0 . +0 . − . , where the uncertainty states the 95 % confidence interval. τ c, HOM indicatesa post-selection time window where indistinguishable photons are available. Since we arepumping close to saturation, our estimate for the coherence time τ c, HOM represents a lowerbound of what can be achieved in our system [48]. Within the uncertainty range, τ c, HOM liesbelow the upper bound τ c, FPI deduced from the FPI measurement.Both the FPI spectrum analysis and the observation of two-photon interference indicate15hat our heterogeneous photonic integration platform can produce waveguide-coupled single-photon sources emitting light with a reasonable level of coherence. Both experiments wereperformed in quasi-resonant, and not strictly resonant excitation (a QD signal-to-pump lasernoise ratio of about 1:2 was estimated for resonant excitation, which prevented observationof resonance fluorescence - see Supplementary Material for details). Since the QD was notexcited resonantly - with which the highest optical coherence level can be achieved [23, 49] - aclear-cut evaluation of the adversity imposed by our fabrication process upon QD coherenceis not possible. In particular, QD 3 was not evaluated pre-fabrication, or even pre-waferbonding, so its starting optical properties are unknown. Nonetheless, the QD linewidths andcoherence times reported here are comparable with those observed from QDs in purely GaAs-based devices [42–44] with which high degrees of two-photon interference were demonstrated,indicating good prospects for our technique.
VI. DISCUSSION
Precise alignment of the EBL patterns with respect to the QDs is essential to avoidexcessive proximity to etched sidewalls, which may lead to degradation of quantum efficiencyand, especially, coherence [25]. Our observation of a 2.2 GHz linewidth from a positionedQD emission line, and subsequent demonstration of two-photon interference, indicate thatthe required precision can be met in our platform, and suggests that our fabrication methodhas minimal adverse effects on QD coherence. In order to increase the source efficiency η Source while preserving such high levels of photon coherence, photonic designs that avoidGaAs etched surfaces closer than 300 nm to the QD [50] while improving the emitter-WG-coupling are required. Creating sophisticated cavity-based devices [51] for such a goal canbe envisioned with our deterministic proximity-corrected grey-scale EBL process.Even though CL excitation is restricted to conditions that avoid charging, which may beproblematic for positioning precision and pattern resolution, we have been able to producestructures with widths ≤
620 nm with high precision around single, pre-selected QDs. We areable to consistently achieve 100 nm thin waveguide taper tips through proximity-correctedgrey-scale lithography in the in situ
EBL step. In our samples, some level of charge drainingis achieved through the GaAs top layer and Si substrate, so that, at a 20 kV electron-beamacceleration voltage, charging due to the Si N -SiO2 layers is avoided, allowing sufficiently16lear CL signals for QD positioning. Charging can nonetheless be further reduced throughvarious measures: lower QD densities in the GaAs would require lower beam currents forefficient excitation and precise localization; a thinner bottom SiO layer would absorb lessenergy without reducing optical confinement; electrically conductive polymers deposited onthe resist would reduce charging in the latter and connect the GaAs to the silicon and theground via the sample edge.We note that multi-emitter experiments can be envisioned with our platform as well, withQDs that produce indistinguishable photons at identical wavelengths. While in situ EBLallows pre-selection of QDs with closely spectrally matching emission lines, with a ¡1 nmaccuracy, fine spectral tuning mechanisms are likely necessary. Local electrical control overthe emission wavelength in GaAs p-i-n-doped layers through the quantum Stark effect is apromising approach, that is fully compatible with both the in situ
EBL and heterogeneoussample stacks. This could also allow control of the QD charge environment, which hasproven helpful in increasing the coherence of emitted photons [7].
VII. CONCLUSION
By applying in situ
EBL to a heterogeneous GaAs / Si N bonded wafer, we demonstratethe ability to deterministically produce GaAs nanophotonic devices with preselected andprecisely located InAs QDs, which can be efficiently accessed by Si N waveguides in an on-chip network. Our unprecedented demonstration of both triggered single-photon emissionand post-selected indistinguishable photons produced by a single QD in a hybrid photoniccircuit indicate a clear path towards highly scalable, chip-based quantum photonics. Thiscan enable experiments such as Shor’s algorithm [52] to be performed on-chip with triggeredphotons at rates substantially higher than currently available. FUNDING
A.S. acknowledges support by the Cooperative Research Agreement between the Uni-versity of Maryland and NIST-CNST. The authors acknowledge funding from the GermanResearch Foundation through CRC 787 ’Semiconductor Nanophotonics: Materials, Models,Devices’. 17
ISCLOSURES
The authors declare that there are no conflicts of interest related to this article. [1] C. Sparrow, E. Mart´ın-L´opez, N. Maraviglia, A. Neville, C. Harrold, J. Carolan, Y. N.Joglekar, T. Hashimoto, N. Matsuda, J. L. O’Brien, D. P. Tew, and A. Laing, Nature ,660 (2018).[2] A. Orieux and E. Diamanti, Journal of Optics , 083002 (2016).[3] J. Carolan, C. Harrold, C. Sparrow, E. Martin-Lopez, N. J. Russell, J. W. Silverstone, P. J.Shadbolt, N. Matsuda, M. Oguma, M. Itoh, G. D. Marshall, M. G. Thompson, J. C. F.Matthews, T. Hashimoto, J. L. O’Brien, and A. Laing, Science , 711 (2015).[4] N. C. Harris, G. R. Steinbrecher, M. Prabhu, Y. Lahini, J. Mower, D. Bunandar, C. Chen,F. N. C. Wong, T. Baehr-Jones, M. Hochberg, S. Lloyd, and D. Englund, Nature Photonics , 447 (2017).[5] M. Hochberg and T. Baehr-Jones, Nature Photonics , 492 (2010).[6] I. Aharonovich, D. Englund, and M. Toth, Nature Photonics , 631 (2016).[7] N. Somaschi, V. Giesz, L. D. Santis, J. C. Loredo, M. P. Almeida, G. Hornecker, S. L.Portalupi, T. Grange, C. Ant´on, J. Demory, C. G´omez, I. Sagnes, N. D. Lanzillotti-Kimura,A. Lema´ıtre, A. Auffeves, A. G. White, L. Lanco, and P. Senellart, Nature Photonics , 340(2016).[8] X. Ding, Y. He, Z.-C. Duan, N. Gregersen, M.-C. Chen, S. Unsleber, S. Maier, C. Schneider,M. Kamp, S. H¨ofling, C.-Y. Lu, and J.-W. Pan, Physical Review Letters , 020401 (2016).[9] J. Liu, R. Su, Y. Wei, B. Yao, S. F. C. da Silva, Y. Yu, J. Iles-Smith, K. Srinivasan, A. Rastelli,J. Li, and X. Wang, Nature Nanotechnology (2019), 10.1038/s41565-019-0435-9.[10] H. Wang, Y. He, Y.-H. Li, Z.-E. Su, B. Li, H.-L. Huang, X. Ding, M.-C. Chen, C. Liu, J. Qin,J.-P. Li, Y.-M. He, C. Schneider, M. Kamp, C.-Z. Peng, S. H¨ofling, C.-Y. Lu, and J.-W. Pan,Nature Photonics , 361 (2017).[11] J. C. Loredo, M. A. Broome, P. Hilaire, O. Gazzano, I. Sagnes, A. Lemaitre, M. P. Almeida,P. Senellart, and A. G. White, Phys. Rev. Lett. , 130503 (2017).
12] A. Javadi, I. S¨ollner, M. Arcari, S. L. Hansen, L. Midolo, S. Mahmoodian, G. Kirsanske,T. Pregnolato, E. H. Lee, J. D. Song, S. Stobbe, and P. Lodahl, Nature Communications ,8655 (2015).[13] S. Sun, H. Kim, Z. Luo, G. S. Solomon, and E. Waks, Science , 57 (2018),http://science.sciencemag.org/content/361/6397/57.full.pdf.[14] S. Castelletto, B. Johnson, V. Iv´ady, N. Stavrias, T. Umeda, A. Gali, and T. Ohshima, NatureMaterials , 151 (2014).[15] A. L. Falk, B. B. Buckley, G. Calusine, W. F. Koehl, V. V. Dobrovitski, A. Politi, C. A.Zorman, P. X.-L. Feng, and D. D. Awschalom, Nature Communications , 1819 (2013).[16] S. L. Mouradian, T. Schr¨oder, C. B. Poitras, L. Li, J. Goldstein, E. H. Chen, M. Walsh,J. Cardenas, M. L. Markham, D. J. Twitchen, M. Lipson, and D. Englund, Physical ReviewX , 031009 (2015).[17] M. Davanco, J. Liu, L. Sapienza, C.-Z. Zhang, J. V. M. Cardoso, V. Verma, R. Mirin, S. W.Nam, L. Liu, and K. Srinivasan, Nature Communications , 889 (2017).[18] I. E. Zadeh, A. W. Elshaari, K. D. J¨ons, A. Fognini, D. Dalacu, P. J. Poole, M. E. Reimer,and V. Zwiller, Nano Letters , 2289 (2016).[19] J.-H. Kim, S. Aghaeimeibodi, C. J. K. Richardson, R. P. Leavitt, D. Englund, and E. Waks,Nano Letters , 7394 (2017), http://dx.doi.org/10.1021/acs.nanolett.7b03220.[20] R. Katsumi, Y. Ota, M. Kakuda, S. Iwamoto, and Y. Arakawa, Optica , 691 (2018).[21] S. Khasminskaya, F. Pyatkov, B. S. Flavel, W. H. Pernice, and R. Krupke, Advanced Materials , 3465 (2014).[22] Z. Sun, A. Martinez, and F. Wang, Nature Photonics , 227 (2016).[23] A. V. Kuhlmann, J. H. Prechtel, J. Houel, A. Ludwig, D. Reuter, A. D. Wieck, and R. J.Warburton, Nature Communications , 8204 (2015).[24] P. Lodahl, S. Mahmoodian, and S. Stobbe, Reviews of Modern Physics , 347 (2015).[25] J. Liu, K. Konthasinghe, M. Davan¸co, J. Lawall, V. Anant, V. Verma, R. Mirin, S. W. Nam,J. D. Song, B. Ma, et al. , Physical Review Applied , 064019 (2018).[26] A. Dousse, L. Lanco, J. Suffczy´nski, E. Semenova, A. Miard, A. Lemaˆıtre, I. Sagnes, C. Roblin,J. Bloch, and P. Senellart, Physical Review Letters , 267404 (2008).[27] R. J. Coles, D. M. Price, J. E. Dixon, B. Royall, E. Clarke, P. Kok, M. S. Skolnick, A. M.Fox, and M. N. Makhonin, Nature Communications , 11183 (2016).
28] S. Unsleber, Y.-M. He, S. Gerhardt, S. Maier, C.-Y. Lu, J.-W. Pan, N. Gregersen, M. Kamp,C. Schneider, and S. H¨ofling, Optics Express , 8539 (2016).[29] M. Sartison, S. L. Portalupi, T. Gissibl, M. Jetter, H. Giessen, and P. Michler, ScientificReports , 39916 (2017).[30] L. Sapienza, M. Davan¸co, A. Badolato, and K. Srinivasan, Nature Communications , 7833(2015).[31] M. Gschrey, A. Thoma, P. Schnauber, M. Seifried, R. Schmidt, B. Wohlfeil, L. Kr¨uger, J. H.Schulze, T. Heindel, S. Burger, F. Schmidt, A. Strittmatter, S. Rodt, and S. Reitzenstein,Nature Communications , 7662 (2015).[32] P. Schnauber, J. Schall, S. Bounouar, T. H¨ohne, S.-I. Park, G.-H. Ryu, T. Heindel,S. Burger, J.-D. Song, S. Rodt, and S. Reitzenstein, Nano Letters , 2336 (2018),https://doi.org/10.1021/acs.nanolett.7b05218.[33] M. Gschrey, R. Schmidt, J.-H. Schulze, A. Strittmatter, S. Rodt, and S. Reitzenstein, Journalof Vacuum Science & Technology B , 021603 (2015).[34] P. Mrowinski, P. Schnauber, P. Gutsche, A. Kaganskiy, J. Schall, S. Burger, S. Rodt, andS. Reitzenstein, arXiv preprint (2019).[35] S. Fischbach, A. Schlehahn, A. Thoma, N. Srocka, T. Gissibl, S. Ristok, S. Thiele, A. Kagan-skiy, A. Strittmatter, T. Heindel, S. Rodt, A. Herkommer, H. Giessen, and S. Reitzenstein,ACS Photonics , 1327 (2017).[36] S. Fischbach, A. Kaganskiy, E. B. Y. Tauscher, F. Gericke, A. Thoma, R. Schmidt,A. Strittmatter, T. Heindel, S. Rodt, and S. Reitzenstein, Applied Physics Letters ,011106 (2017), http://dx.doi.org/10.1063/1.4991389.[37] S. Fischbach, M. v. Helversen, M. Schmidt, A. Kaganskiy, R. Schmidt, A. Schliwa, T. Heindel,S. Rodt, and S. Reitzenstein, arXiv preprint (2018).[38] B. Yacobi and D. Holt, Journal of Applied Physics , R1 (1986).[39] A. Kaganskiy, T. Heuser, R. Schmidt, S. Rodt, and S. Reitzenstein, Journal of Vacuum Science& Technology B, Nanotechnology and Microelectronics: Materials, Processing, Measurement,and Phenomena , 061603 (2016).[40] G. Ortner, D. R. Yakovlev, M. Bayer, S. Rudin, T. L. Reinecke, S. Fafard, Z. Wasilewski, andA. Forchel, Physical Review B , 201301 (2004).
41] E. Peter, S. Laurent, J. Bloch, J. Hours, S. Varoutsis, I. Robert-Philip, A. Beveratos,A. Lemaˆıtre, A. Cavanna, G. Patriarche, P. Senellart, and D. Martrou, Applied PhysicsLetters , 223118 (2007).[42] G. Kirsanske, H. Thyrrestrup, R. S. Daveau, C. L. Dreeßen, T. Pregnolato, L. Midolo, P. Tigh-ineanu, A. Javadi, S. Stobbe, R. Schott, A. Ludwig, A. D. Wieck, S. I. Park, J. D. Song, A. V.Kuhlmann, I. S¨ollner, M. C. L¨obl, R. J. Warburton, and P. Lodahl, Physical Review B ,165306 (2017).[43] A. Thoma, P. Schnauber, M. Gschrey, M. Seifried, J. Wolters, J.-H. Schulze, A. Strittmatter,S. Rodt, A. Carmele, A. Knorr, T. Heindel, and S. Reitzenstein, Physical Review Letters , 033601 (2016).[44] C. Nawrath, F. Olbrich, M. Paul, S. Portalupi, M. Jetter, and P. Michler, arXiv preprint (2019).[45] M. A. M. Versteegh, M. E. Reimer, K. D. J¨ons, D. Dalacu, P. J. Poole, A. Gulinatti, A. Giudice,and V. Zwiller, Nature Communications , 5298 (2014).[46] M. Davanco, C. S. Hellberg, S. Ates, A. Badolato, and K. Srinivasan, Physical Review B (2014), 10.1103/physrevb.89.161303.[47] R. B. Patel, A. J. Bennett, K. Cooper, P. Atkinson, C. A. Nicoll, D. A. Ritchie, and A. J.Shields, Physical Review Letters , 207405 (2008).[48] P. Michler, A. Imamo˘glu, A. Kiraz, C. Becher, M. Masona, P. Carson, G. Strouse, S. Buratto,W. Schoenfeld, and P. Petroff, Physica Status Solidi (b) , 399 (2002).[49] M. Reindl, J. H. Weber, D. Huber, C. Schimpf, S. F. C. da Silva, S. L. Portalupi, R. Trotta,P. Michler, and A. Rastelli, arXiv preprint (2019).[50] H. Thyrrestrup, G. Kirsanske, H. Le Jeannic, T. Pregnolato, L. Zhai, L. Raahauge,L. Midolo, N. Rotenberg, A. Javadi, R. Schott, A. D. Wieck, A. Ludwig, M. C.L¨obl, I. S¨ollner, R. J. Warburton, and P. Lodahl, Nano Letters , 1801 (2018),https://doi.org/10.1021/acs.nanolett.7b05016.[51] S. Hepp, S. Bauer, F. Hornung, M. Schwartz, S. L. Portalupi, M. Jetter, and P. Michler,Optics Express , 30614 (2018).[52] A. Politi, J. C. F. Matthews, and J. L. O’Brien, Science
53] M. Davan¸co, M. T. Rakher, D. Schuh, A. Badolato, and K. Srinivasan, Applied PhysicsLetters , 041102 (2011).[54] A. Young, A. Thijssen, D. Beggs, P. Androvitsaneas, L. Kuipers, J. Rarity, S. Hughes, andR. Oulton, Physical Review Letters , 153901 (2015).[55] S. Abrarov and B. Quine, Journal of Mathematics Research , p163 (2015).[56] M. Schwartz, U. Rengstl, T. Herzog, M. Paul, J. Kettler, S. L. Portalupi, M. Jetter, andP. Michler, Optics Express , 3089 (2016).[57] M. Schwartz, E. Schmidt, U. Rengstl, F. Hornung, S. Hepp, S. L. Portalupi, K. llin, M. Jetter,M. Siegel, and P. Michler, Nano Letters , 6892 (2018). UPPLEMENTARY MATERIAL: INDISTINGUISHABLE PHOTONS FROMDETERMINISTICALLY INTEGRATED SINGLE QUANTUM DOTS IN HETERO-GENEOUS GAAS/SI N QUANTUM PHOTONIC CIRCUITSI. CHARGING EFFECTS DURING FABRICATION
As can be seen from Fig. 2 b) and c) in the main text, some WGs show a downward-bent left-hand part. The WG bending amplitude correlates with the duration of the in-situ EBL process (stronger bending at later times during the 6 h in-situ EBL run). Italso correlates with the size of the GaAs layer area around the device (stronger bendingfor devices closer to GaAs layer borders). Both effects hint that during the in-situ EBLrun, charge continuously accumulates and at some point cannot be dissipated sufficientlyanymore. This leads to a charge build-up in the mapping process that deviates the beam inthe beginning of the patterning step. This effect can be reproduced in the post-fabricationmapping: Because no planar GaAs exists around the fabricated devices anymore, thereis insufficient charge dissipation and the beam is deviated in the beginning of the post-fabrication mapping process, as can be seen in the simultaneously acquired SEM image inFig. S1 a). Fig. S1 b) depicts an optical micrograph of the same WG device, showing thatit is not bent. The beam shift is analogous to the deviations for devices fabricated late inthe in-situ EBL run. In the post-fabrication maps Fig. 2 g) - i) in the main text, beamdeviations at the GaAs devices were avoided by starting the map far enough away for thebeam to stabilize before scanning the GaAs devices. As discussed in the main text, lowerQD densities requiring lower beam currents, as well as a thinner SiO layer and conductivepolymers on top of the EBL resist are expected to significantly reduce charging in the future.23 a) b) FIG. S1. a) SEM image of device QD 1 taken during post-fabrication CL mapping. In the firstfew seconds of the mapping process, the beam is deviated by a charge build-up, before stabilizingfor the rest of the map. b) False-color optical microscope image of the same device proving thatdevice QD 1 consists of straight WGs.
II. GAAS WAVEGUIDE TAPER PATTERN
The GaAs nanowaveguides, which host the InAs QD, are patterned using proximity-corrected grey-scale EBL. This is particularly important as the electron dose per pixel neededfor cross-linking the resist increases by a factor of more than 4 at the taper tip, as comparedto the taper center. Fig. S2 shows the target and the proximity-corrected pattern used towrite the nanowaveguide that hosts QD 1-3. A close-up of the grey-scale pattern taper tipis also shown. The electron dose is encoded linearly in the 256 grey-scale steps. QD target pa � ernproximity-corrected grey scale pa � ern 20µm5µm 0255 FIG. S2. top: Target GaAs nanowaveguide pattern used to integrate QD 1-3. btm: Proximity-corrected grey-scale EBL pattern for in-situ EBL to manufacture the target pattern along with aclose-up of the grey-scale taper-tip for better visualization of the grey scales. II. POWER-DEPENDENT EMISSION RED-SHIFT
To investigate the red-shift of the ≈ . T = T + η P in temperature with excitation power P , startingfrom a base temperature of T = 7 K, and fit this dependence with a Bose-Einstein phononlaw [40, 53] to the power-dependent peak positions E QD ( P ), see Fig. 3 c) in the main text. E QD ( P ) = E QD (0) − S E Ph coth (cid:18) E Ph T + η P ) k B (cid:19) (1)Here, S = 0 . ± . E Ph = (1 . ± .
06) meV is thephonon energy and η = (6700 ± ≈ . T = T + η P law. Thismeans that a more elaborate model is needed to fully describe the thermal behaviour ofour hybrid system. During the power series experiment, the excitation wavelength remainedunchanged. 25 avelength (nm) I n t e n s i t y ( a . u . ) a) Temperature (K) E n e r g y ( e V ) b) FIG. S3. a) Temperature series taken at temperatures of 7 K, 9 K, 14 K, 15 K and 20 K. We addedan arbitrary intensity offset in the spectra for better visualization. b) Peak energies from a) againstcold-finger temperature.
IV. HANBURY BROWN AND TWISS INTERFEROMETER EVALUATION
The HBT data shown in Fig 3 d) in the main text is modeled with a two-sided exponentialdecay function, taking into account only the faster decay rate τ r for a conservative estimateof g (2) (0): g (2) ( τ ) = A exp( −| τ | /τ r ) + A Side (cid:88) k (cid:54) =0 exp( −| τ − τ k | /τ r ) (2)Here, A is the central peak area, A Side the side peak area, τ r the radiative lifetime and τ k the side peak position. This model is convolved with the detector time response of129 ps to obtain g (2)Conv ( τ ) for the fit. Since the coincidence data around τ = 0 is close to0, we fit our model using a logarithmic Poissonian noise distribution[42] ln(Poiss( µ, K )) = − µ + K ln( µ ) − ln( K !) and a maximum likelihood routine that minimizes − (cid:88) i ln (cid:20) Poiss (cid:18) g (2)Conv ( τ i ) , N i (cid:19)(cid:21) . (3)Here, i enumerates the time bins, τ i is the time delay in bin i and N i is the number ofcoincidences in bin i . The peak areas A and A Side are the only free parameters in the fit,26ielding g (2) (0) = A /A Side = 0 . ± .
04, where the uncertainty gives the propagated 95 %confidence interval. 27 . UPPER BOUND ESTIMATE FOR THE QD-WAVEGUIDE AND QD-LENSEDOPTICAL FIBER COUPLING EFFICIENCY
To estimate the maximum expected coupling efficiency between QD 3 and the fundamen-tal TE Si N waveguide mode, we used Finite Difference Time Domain (FDTD) simulationsof electric dipoles emitting in a hybrid waveguide/mode transformer that approximated thegeometry observed in SEM. In this model, the GaAs, SiO spacer and Si N layers had athickness of 190 nm, 100 nm and 250 nm, respectively. The straight section of the GaAswaveguide had a width of 620 nm and length of 5 µ m, and the mode transformer was ta-pered down to a width of 100 nm over a length of 20 µ m. The GaAs nanowaveguide and theunderlying, 685 nm width Si N waveguide were horizontally misaligned from each other by44 nm. The FDTD simulations consisted of exciting the geometry with an electric dipolesource located at the geometrical center of the GaAs nanowaveguide, and calculating thesteady-state fields at a 916 nm wavelength, at the edges of the computational domain.The simulation incorporated perfectly-matched layers to emulate open domains. Thetotal emitted power of the dipole was obtained by integrating the steady-state Poyntingvector over all of the computational domain boundaries. The power carried by the var-ious guided modes supported by the GaAs and Si N waveguides were obtained throughoverlap integrals with the steady-state field at the waveguides’ cross-section. Simulationswere performed for horizontally oriented dipole moments, either transversal ( x -oriented) orlongitudinal ( z -oriented) to the GaAs nanowaveguide. Because we believe the ≈ . x - and z -componentswith a 90 ◦ phase between them)[54]. However, because a transversal dipole tends to cou-ple more efficiently to the fundamental TE-like GaAs mode, it yields a more conservativeupper-bound estimate for the QD-Si N waveguide coupling.The 620 nm wide GaAs waveguide supports 7 guided modes at 916 nm, as shown inFig. S4. The dipole emission is divided among such guided modes, as well as unguided,radiation or substrate modes. Figure S4 also shows the coupling ratios ( β x,y,z ) for eachguided mode, for the transverse ( x ), longitudinal ( z ), and rotating dipole ( c ) cases. Thehighest coupling ratio achievable in such multimode waveguides is of about 25 %, for thehorizontal dipole into the fundamental GaAs mode, which is TE-like, and has a major x N waveguide TE mode, however, the conversion efficiency is highest( ≈
69 %) for the fundamental one. Figure S5 shows the fundamental TE-like and TM-likeSi N waveguide modes, after the mode transformer. The coupling efficiencies from the QDto the these two modes, for the tree dipole configurations, is also displayed. A maximumefficiency of ≈
13 % is achieved for the transverse dipole, whereas for the rotating dipole theefficiency is of ≈ N waveguide and the lensedoptical fiber used in our experiments. The manufacturer specification for the lensed fiberwas such that it should produce at ≈ µ m spot size at its focus, at 980 nm, and so inour simulation, a 2 µ m spot-size, horizontally polarized Gaussian beam was launched at thegeometrical center of a 685 nm wide and 250 nm tall Si N waveguide facet. We note that, inour fabricated devices, the Si N waveguides unintendedly made a ≈ ◦ angle with respectto the cleaved facet plane. This imperfection was included in our model. An overlap integralwas then used to obtain, from the steady-state waveguide field, a coupling ratio of ≈
23 %into the fundamental TE-like Si N mode.Overall, we expect the QD-lensed fiber coupling efficiency to be of at most 0 . × . ≈ ≈ x ( m m) y ( m m ) x ( m m) x ( m m) x ( m m) x ( m m) x ( m m) x ( m m) 00.20.40.60.81|E| b = 0.25 x b = 0 z b = 0.05 c b = 0 x b = 0.04 z b = 0.03 c b = 0 x b = 0 z b = 0 c b = 0 x b = 0 z b = 0 c b = 0.17 x b = 0 z b = 0.06 c b = 0 x b = 0 z b = 0 c b = 0 x b = 0 z b = 0 c FIG. S4. Squared electric field profiles for bound modes of the hybrid nanowaveguide hosting QD 3.Modes 1, 2 and 5 have TE-like character, with major transversal ( x ) electric field component.Modes 3 and 4 have a TM-like character, with a major x magnetic field component. The β x,z,c factors are the coupling between transversal ( x ), longitudinal ( z ) or circular ( c ) dipoles to thecorresponding modes. − − − − |E| x ( � m)x ( � m) y ( � m ) � = 0.13 x � = 2×10 -3 z � = 0.09 c � = 0.07 x � = 4×10 -3 z � = 0.01 c TE-like TM-like
FIG. S5. Squared electric field profiles for bound modes of the Si N waveguide. Modes 1 and 2are TE-like and TM-like, respectively. The β x,z,c factors are the coupling between transversal ( x ),longitudinal ( z ) or circular ( c ) dipoles to the corresponding modes. VI. FABRY-PEROT INTERFEROMETER EVALUATION
For the evaluation of the Fabry-Perot-Interferometer (FPI) data, we fit the data with aVoigt [55] (Fig. 4 a) in the main text), Lorentzian (Fig. S6 a)) and Gaussian (Fig. S6 b))line functions. The Voigt fit gives an R of 0.9884, slightly superior to the Lorentzian andGaussian ( R = 0 . R = 0 . b)a) C o un t s ( a . u . ) R e s i du a l s FPILorentz fi t050-50 Rela � ve frequency (GHz) -4 -2 0 2 4 Rela � ve frequency (GHz) -4 -2 0 2 4 FPIGaussian fi t C o un t s ( a . u . ) R e s i du a l s FIG. S6. a) FPI data (blue) and Lorentzian fit (red) with residuals. b) FPI data (blue) andGaussian fit (red) with residuals II. HONG-OU-MANDEL INTERFEROMETER
In the Hong-Ou-Mandel (HOM) type experiment, the fiber-collected PL was first passedthrough a ≈
500 pm free spectral range fiber-coupled grating filter followed by a long wave-length pass filter, and then guided through quarter- and half-wave plates and a polarizationbeam splitter (PBS) cube. This was done to bring the polarization of the PL signal intoa linear state from an elliptical state that resulted in large part due to scrambling in thenon-polarization-maintaining SM collection fiber, though likely also from the inherent polar-ization of the collected QD emission. The linearly-polarized light was then passed througha half-wave plate and coupled into a polarization-maintaining (PM) fiber PBS. The half-wave plate was aligned to the slow-axis of the fiber PBS, such that the photon stream wasmaximized at one output port while being suppressed at the other. Throughout the experi-ment, we could monitor the suppressed polarization output of the fiber PBS at an SNSPD,to verify the long term stability of our polarization filtering system, which could change ifthe stress on the conventional SM fibers in the setup was accidentally altered. After thefiber-coupled PBS, the QD signal was guided into an unbalanced PM fiber-coupled Mach-Zehnder Interferometer (MZI), with a δτ ≈
10 ns arm imbalance. A variable half-wave plateinserted in the long interferometer arm allowed the (linear) polarization of photons travel-ling through either arm to be, at the second beamsplitter, parallel or orthogonal to eachother. We matched the intensity of the photon streams in the two MZI arms by controllablyloosening one fiber connection in the short arm, and measured coincidences on the two MZIoutputs using the same SNSPDs as in the HBT measurement.The raw HOM autocorrelation traces for parallel and orthogonal photons are depictedin blue in Fig. 4 b) and magnified around τ = 0 in Fig. 4 c) in the main text. We followRef. [47] to model the two traces with the functions g (2)HOM , (cid:107) ( τ ) and g (2)HOM , ⊥ ( τ ), where g (2) ( τ )accounts for the bunching around zero time delay: g (2) ( τ ) = 1 − A exp( −| τ | / τ ) + ( A −
1) exp( −| τ | / τ ) (4) g (2)HOM , ⊥ ( τ ) = 4( T + R ) R T g (2) ( τ ) (cid:124) (cid:123)(cid:122) (cid:125) G ( τ ) + 4 R T [ T g (2) ( τ − δτ ) + R g (2) ( τ + δτ )] (cid:124) (cid:123)(cid:122) (cid:125) G ( τ ) (5) g (2)HOM , (cid:107) ( τ ) = G ( τ ) + G ( τ )[1 − V exp( − | τ | / τ c, HOM ) ] (6)Here, A , τ and τ describe the bunching around τ = 0 in the CW autocorrelation curve g (2) ( τ ). R = 0 .
50 and T = 0 .
50 as well as R = 0 .
54 and T = 0 .
46 are the reflection and32ransmission coefficients of the first and second beamsplitter in the Mach-Zehnder interfer-ometer. From the fit of g (2)HOM , ⊥ ( τ ), we obtain A = 1 . ± . τ = (0 . ± .
02) ns and τ = (14 . ± .
32) ns, where the uncertainties are standard errors. The results of g (2)HOM , (cid:107) ( τ )are given in the main text. 33 III. RESONANT AND PHONON-MEDIATED EXCITATION
The coherent excitation of QDs through resonance fluorescence is an important steptowards the emission of Fourier-limited photons [23]. In resonance fluorescence, the QDsignal needs to be separated from the excitation laser. Waveguide architectures naturallyoffer spatial separation of pump and detection position, for free-space excitation orthogonalto the wafer. Furthermore, waveguides can act as polarization filters [56] allowing on-chipsuppression of pulsed excitation lasers up to a signal-to-noise ratio (SNR) of 40:1 [57].When tuning the excitation laser wavelength close to the QD emission, one can excite theQD through a longitudinal acoustic (LA) phonon-mediated process. Assuming that the LA-phonon mediated and strictly resonant excitation are approximately equally efficient, weuse this scheme to probe the QD emission to laser SNR that we can achieve in our system.Applying off-chip polarization filtering as in the HOM experiment and an excitation NA of0.28, we measure an SNR of about 1:2 as seen in Fig. S7 a). We check the origin of theexcess laser signal in our system with two methods. Firstly, we use an excitation NA of 0.1and repeat the phonon-mediated pumping, losing one order of magnitude in SNR. Secondly,we launch resonant laser light into the Si N waveguide at the sample facet, and measurelaser light that is scattered into an NA=0.28 towards the top with a conventional CCDcamera, see Fig. S7 b). Most of the light is scattered at the GaAs WG taper tips and inthe center of the GaAs WG, where the QD is located. Due to time reversal symmetry, weconclude that our current GaAs-Si N -SiO structure unfortunately scatters large amountsof the top excitation laser into the Si N WG, detrimental to laser suppression. Withimproved waveguide designs, that avoid vertical interfaces near the QD, and a tighter beamfocusing reducing the necessary pump powers, higher laser suppression enabling resonancefluorescence from preselected QDs on silicon chips may be possible in the future.34 n t e n s i t y ( a . u . ) Wavelength (nm)
916 920
LaserQD signal