High-speed measurement-device-independent quantum key distribution with integrated silicon photonics

source.

• 1.25 GHz polarization-encoding MDI-QKD is experimentally demonstrated with silicon photonic chip transmitters.The system uses two chip transmitters and an untrusted relay for Bell-state measurements.Source: source · pp. 3, 5
• The experiment reports higher key rates than previous MDI-QKD experiments and supports low-cost, scalable networks with untrusted relays.The authors describe the system as a step toward wafer-scale manufactured MDI-QKD systems and quantum networks.Source: source · pp. 3, 7
• Each transmitter integrates intensity modulation, polarization modulation, and variable optical attenuation for QKD state preparation.The integrated components support decoy intensities, four BB84 polarization states, and attenuation to single-photon levels.Source: source · p. 4
• The packaged chip provides stable polarization encoding and decoy-state modulation, while QBER remains low over several hours.The device has a 4.8 × 3 mm^2 chip footprint and is assembled in a compact packaged system designed for low-cost production.Source: source · p. 5

• 1.25 GHz pulses are generated by individually driven master and slave gain-switched lasers.The lasers use 500-ps and 200-ps square-wave pulses, with a 1-ps electrical delay for temporal overlap.Source: Methods · p. 8
• Seeding photons produce low-jitter, phase-randomized pulses approximately 100 ps wide.The generated pulses pass through a 10-GHz filter to reduce frequency chirp.Source: Methods · p. 8
• The two-photon interference test attenuates each detector count rate to approximately 3.5 MHz and records data for 100 s.The coincidence time window is 600 ps.Source: Methods · p. 8
• 48.4% visibility is obtained in the two-photon interference experiment.Source: Methods · p. 8

• The chip integrates an intensity modulator, polarization modulator, and variable optical attenuator.The intensity modulator uses a Mach-Zehnder interferometer, while the VOA attenuates pulses to single-photon levels.Source: Si transmitter chip · pp. 8–9
• Four intensity choices, {µ, ν, ω, 0}, are randomly generated for decoy-state operation.The integrated intensity modulator has approximately 30 dB static and 20 dB dynamic extinction ratios; an external LiNbO3 modulator improves vacuum extinction.Source: Si transmitter chip · p. 8
• The polarization modulator prepares the four required states by controlling arm amplitudes and imposing phase shifts θ ∈ {0, π/2, π, 3π/2}.The polarization rotator combiner converts and recombines the two polarization components.Source: Si transmitter chip · p. 9
• The prepared states are characterized in Charlie using polarization controllers and polarization beam splitters aligned to rectilinear and diagonal bases.Source: Si transmitter chip · p. 9
• An average polarization extinction ratio of approximately 23 dB is achieved, sufficient for low-error MDI-QKD operation.Source: Si transmitter chip · p. 9

• Charlie electrically distributes the synchronization clock with tunable 1-ps delays to compensate temporal drifts among Alice, Bob, and Charlie.Source: Detection · p. 9
• A 50-ohm shunt resistor prevents detector latching at the system’s GHz repetition rate without sacrificing detection efficiency.The system requires SNSPDs to tolerate peak counting rates above 5 MHz.Source: Detection · p. 10
• The high-speed time tagger supports a maximum data transfer rate of 65 MHz.Source: Detection · p. 10
• A 600-ps coincidence window provides a trade-off between detection efficiency and X-basis error rate.Source: Detection · p. 10

• Four-intensity decoy-state MDI-QKD uses three decoy intensities in the X basis and one signal intensity in the Z basis.Alice and Bob randomly prepare weak coherent pulses in the Z or X basis.Source: SUPPLEMENTARY MATERIALS · p. 20
• The Z basis generates secret key, while the X basis supports decoy-state analysis.Source: SUPPLEMENTARY MATERIALS · p. 20
• Alice and Bob use the same six optimized parameters [s, µ, ν, P_s, P_µ, P_ν] for intensities and probabilities.The setup performs a full optimization of these parameters.Source: SUPPLEMENTARY MATERIALS · p. 20
• Joint constraints combine common observables and can produce a higher key rate than independent constraints.Source: SUPPLEMENTARY MATERIALS · p. 20
• The secret key is extracted using a formula based on observed gains and QBER together with bounded single-photon yield and phase-error terms.The error-correction efficiency is set to 1.16.Source: SUPPLEMENTARY MATERIALS · p. 20
• 48.4% visibility is measured in the Hong-Ou-Mandel interference experiment.The photon count rate is approximately 3.5 MHz per detector, with 100 s of data collection and a 600-ps coincidence window.Source: SUPPLEMENTARY MATERIALS · pp. 20–21

Intensity modulation

• 19.8 dB dynamic extinction ratio is obtained at a 1.25-GHz repetition rate.The extinction ratio decreases from 21.5 dB at 625 MHz to 19.8 dB at 1.25 GHz.Source: Si chip transmitter · p. 21
• 1.25 GHz operation supports random intensity modulation for one signal state and three decoy states.Four RF voltage levels generate the intensities s, µ, ν, and ω.Source: Si chip transmitter · p. 21
• ∼110 dB maximum attenuation is achieved using three cascaded PIN-diode variable optical attenuators.A single attenuator provides 38.0 dB, while differential biasing across the cascade raises the maximum attenuation.Source: Si chip transmitter · pp. 21–22

Polarization modulation

• Four polarization states in conjugate bases are prepared with ∼26 dB extinction ratio and ∼0.993 degree of polarization.The states are characterized using a polarimeter after applying appropriate RF signals to the polarization modulator.Source: Si chip transmitter · p. 22
• ∼23 dB average polarization extinction ratio is obtained after aligning the measurement device to rectilinear and diagonal bases.The alignment uses EPC adjustments and RF voltages between 0 and 7.5 V.Source: Si chip transmitter · p. 22

Motivation

• A shared polarization reference is required among Alice, Bob, and Charlie for the experiment.The alignment procedure addresses polarization drift in the quantum channel.Source: Polarization alignment · p. 22

Alignment procedure

• The automatic alignment method rapidly calibrates the polarization reference through three sequential adjustment steps.The procedure is represented by a dedicated alignment-system schematic and flowchart.Source: Polarization alignment · p. 22
• Bob and Charlie first establish a common reference by adjusting EPC-1 and EPC-2.This is the first step of the flowchart-based procedure.Source: Polarization alignment · p. 23
• Alice then aligns her bases by adjusting EPC-A.This follows the second step of the alignment flowchart.Source: Polarization alignment · p. 23
• Charlie finally aligns one PBS to the Z-basis by adjusting EPC-2.This completes the three-step polarization alignment procedure.Source: Polarization alignment · p. 23

Board architecture

• The FPGA board integrates memory, serial/parallel conversion, delay, thermoelectric-control, synchronization, and analog-output modules.All modules except analog output are implemented on a Xilinx Kintex-7 FPGA.Source: Electronic control board · p. 23

High-speed control

• 12.5 Gbps data transmission is achieved by converting low-speed serial data into high-speed parallel streams.The conversion supports simultaneous transmission across parallel channels.Source: Electronic control board · p. 23
• 1 ps delay resolution enables temporal overlap of laser pulses and accurate intensity and polarization modulation.The delay module independently adjusts the timing of each channel.Source: Electronic control board · p. 24

Synchronization

• Four delayed pulse sequences can be synchronized to internal or external clocks, allowing all experimental stations to operate synchronously.The synchronization module generates up to four precisely delayed output sequences.Source: Electronic control board · p. 24

Laser synchronization

• A 48.4% Hong-Ou-Mandel interference visibility is measured for the gain-switch lasers.The result is associated with the laser-seeding characterization used in the experiment.Source: Figure 5 · p. 25