A manufacturable platform for photonic quantum computing

Abstract

Although holding great promise for low noise, ease of operation and networking¹, useful photonic quantum computing has been precluded by the need for beyond-state-of-the-art components, manufactured by the millions^2-6. Here we introduce a manufacturable platform⁷ for quantum computing with photons. We benchmark a set of monolithically integrated silicon-photonics-based modules to generate, manipulate, network and detect heralded photonic qubits, demonstrating dual-rail photonic qubits with 99.98% ± 0.01% state preparation and measurement fidelity, Hong-Ou-Mandel (HOM) quantum interference between independent photon sources with 99.50% ± 0.25% visibility, two-qubit fusion with 99.22% ± 0.12% fidelity and a chip-to-chip qubit interconnect with 99.72% ± 0.04% fidelity, conditional on photon detection and not accounting for loss. We preview a selection of next-generation technologies: low-loss silicon nitride (SiN) waveguides and components to address loss, as well as fabrication-tolerant photon sources, high-efficiency photon-number-resolving detectors (PNRDs), low-loss chip-to-fibre coupling and barium titanate (BTO) electro-optic phase shifters for high-performance fast switching.

Fig. 1. Manufacturable integrated quantum photonic stack.

a,b, Schematics of key components and process modules. We highlight (on the right) further process steps included in our next-generation platform. c, A 300-mm wafer containing single-photon sources, superconducting single-photon detectors and quantum benchmarking circuits. d, A cryogenic assembly containing a photonic die, heat spreader, electronic PCB and 100-channel telecommunications fibre attach unit. e–j, Optical micrograph, scanning electron microscope or transmission electron microscopy images of: photon source (top-down) (e); optical waveguide (cross-section) (f); deep/shallow trench scattered light shield (cross-section) (g); single-photon detector (top-down) (h); thermal isolation trench (cross-section) (i); single-photon detector on waveguide (cross-section) (j). k, Custom cryostat used in benchmarking experiments with >10 W cooling power at 2.2 K. Scale bars, 20 μm (e,h), 1 μm (f,i), 10 μm (g), 40 nm (j). AMZI, asymmetric Mach–Zehnder interferometer.

Fig. 2. Key building blocks of the platform.

a, Schematics of photon source, filter network, interferometer and detector. b, Measured joint spectral intensity of an interferometrically coupled resonator photon source, indicating a spectral purity of 99.5% (Supplementary Information). c, Response of our pump filter network. We shade the pump, signal and herald frequency bands and show the measured herald (orange) and signal (blue) filter spectrum, characterized with on-chip SNSPDs. d, Measured response of a Mach–Zehnder interferometer to heralded single-photon illumination on a fully integrated platform. The extinction ratio at the transmission port is >50 dB. The asymmetry in the Mach–Zehnder interferometer response is an artefact of a non-constant step size, which is finer around one feature only. There is no marked variation in performance across a circuit or among different circuits. e, Measured on-chip detection efficiency as a function of detector bias current (I_B) normalized by the detector switching current (I_SW) and the detector count rate (blue) and dark count rate (orange) per second (inset) (Supplementary Information).

Fig. 3. Quantum benchmarking circuits.

These circuits are reconfigurable by means of thermal phase shifters indicated in red in the schematics. a–d, Schematics of: quantum state preparation and measurement (a); point-to-point qubit network (b); two-photon quantum (HOM) interference (c); two-qubit fusion measurement (d). e, SPAM fidelity of the reconstructed state with the target state for Pauli eigenstates. f, HOM interference. g, Measured Pauli transfer matrix of chip-to-chip qubit interconnect channel. h, Reconstructed two-qubit density matrix after fusion (grey bars indicate magnitude below the 0.01 threshold).

Fig. 4. Cascaded resonator source and PNRD.

a, Schematic of the source. b, Measured joint spectral intensity of a cascaded resonator source showing up to 99.35% purity, assuming flat spectral phase (Supplementary Information). c, Measured indistinguishability of two source copies as a function of the resonance wavelength offset (Supplementary Information). d, Top-down optical micrograph of a SiN-waveguide-coupled PNRD, in which single-photon detectors (SNSPDs) are crossing a waveguide and absorb light from the waveguide through evanescent coupling. Sets of SNSPDs are connected through on-chip resistors to comprise a unit cell. Identical unit cells are connected in series. e, On-chip detection efficiency for the PNRD shown in d as a function of normalized bias current, showing the average across six unique devices (Supplementary Information). e, Inset, distribution of single-shot detection efficiency for each of the unique devices biased at roughly 0.9I_SW at two input power levels. f, Left, persistent plot of the electrical photodetection signal (voltage traces) of a four-unit-cell PNRD. The traces were recorded using a cryogenic amplifier. The voltage traces show five distinct levels, corresponding to 0, 1, 2, 3 and 4 unit cells detecting photons simultaneously. Right, voltage trace histogram.

Fig. 5. Waveguide and component loss and BTO optical switch.

a–c, Loss of SiN-based components with mean (black line) and median (white line). a, SiN waveguide loss measurement, showing results across example wafers for both multimode (MM) and single-mode (SM) waveguides (Supplementary Information). b, SiN component loss for waveguide splitters and crossings (Supplementary Information). c, Chip-to-fibre loss. The fibre-to-chip coupling is measured in the low-loss regime using repeated transmission measurements on two exemplary devices designed for SMF-28 fibre and an exemplary device designed for UHNA fibre (Supplementary Information). d, Free-space electro-optic measurement of the effective Pockels coefficient of a BTO film grown by molecular-beam epitaxy, with hysteresis. e, Scanning electron microscope cross-section of a fully fabricated BTO-on-SiN phase shifter. Scale bar, 1,000 nm. f, Cutback-based propagation loss measurement of a BTO-on-SiN phase shifter (data points and guide line), with 95% confidence intervals provided (dashed lines). g, Measured optical transmission of a Mach–Zehnder interferometer with a L = 2-mm-long BTO phase shifter. A voltage was applied to one arm of the Mach–Zehnder interferometer, resulting in a V_πL = 0.62 V.cm in a non-push–pull configuration (Supplementary Information), in which V_π is the voltage required to change the phase by π radians. Wafer maps of these results can be found in the Supplementary Information.