Microcomb-driven silicon photonic systems

Abstract

Microcombs have sparked a surge of applications over the past decade, ranging from optical communications to metrology^1-4. Despite their diverse deployment, most microcomb-based systems rely on a large amount of bulky elements and equipment to fulfil their desired functions, which is complicated, expensive and power consuming. By contrast, foundry-based silicon photonics (SiPh) has had remarkable success in providing versatile functionality in a scalable and low-cost manner^5-7, but its available chip-based light sources lack the capacity for parallelization, which limits the scope of SiPh applications. Here we combine these two technologies by using a power-efficient and operationally simple aluminium-gallium-arsenide-on-insulator microcomb source to drive complementary metal-oxide-semiconductor SiPh engines. We present two important chip-scale photonic systems for optical data transmission and microwave photonics, respectively. A microcomb-based integrated photonic data link is demonstrated, based on a pulse-amplitude four-level modulation scheme with a two-terabit-per-second aggregate rate, and a highly reconfigurable microwave photonic filter with a high level of integration is constructed using a time-stretch approach. Such synergy of a microcomb and SiPh integrated components is an essential step towards the next generation of fully integrated photonic systems.

Fig. 1. Microcomb-based SiPh optoelectronic systems.

Conceptual drawings for several integrated optoelectronic systems (data transmission, microwave photonic signal processing, optical beam steering and photonic computing) realized by combining a microcomb source with silicon photonic chips. With III–V-on-silicon photonic integration, the chips are expected to contain all the essential functions (for example, laser-microcomb generation, passive and active optical components, and the electronics for supporting signal processing and system control).

Fig. 2. Comb generation and fundamental characteristics.

a, Optical image of the InP DFB laser chip and the AlGaAsOI microresonators for dark-pulse generation. b, Normalized comb power when tuning the pump frequency across the resonance at around 1,552 nm. With 10-mW pump power, a dark-pulse Kerr comb could be accessed in a large frequency window (tens of gigahertz). CW, continuous wave. c, d, Two-FSR dark-pulse spectra (top) and the ‘turnkey’ behaviours (bottom) pumped by a commercial external laser (c) or a DFB laser chip (d) with an equal on chip power of 10 mW. A pair of flat wings besides the pump is formed in both spectra, exhibiting typical profile of the coherent dark-pulse microcombs. Inset: comb intensity noise (resolution BW of 100 kHz). The intensity noise of the dark-pulse Kerr comb is at the same power level as the electrical spectrum analyser background. P, power; f, frequency; PZT, lead zirconate titanate. e, Long-term stability of a free-running comb. f–k, Optical images and main performance of several Si-based fundamental devices, including a depletion-mode Si MZM (f), a TiN microheater (g), a Si spiral waveguide delay line (h), a vertical epitaxial Ge PD (i), a microring filter (j) and a CMOS driver for MZMs (k). More details can be found in Methods.

Fig. 3. Transmission results.

a, Schematic of the microcomb-based data transmission set-up. The dark-pulse Kerr comb source is pumped by a continuous-wave laser, which can be generated by a commercial external cavity diode laser (ECDL, i) or a distributed feedback laser chip (ii). The generated comb is then sent into a SiPh T/R chip. iso, isolator; NF, notch filter; DEMUX, demultiplexer; MUX, multiplexer. Scale bar, 500 μm. b, A 20-line comb spectrum in the C band as the multiwavelength source before injection into the SiPh T/R chip. c, Typical eye diagrams of the chosen channel after modulation by SiPh modulators at different symbol rates (32 Gbaud, 40 Gbaud and 50 Gbaud). d, BER for each comb line. The blue squares and red circles indicate the ECL-pumped comb data transmission results at symbol rates of 32 Gbaud and 50 Gbaud, respectively. All channels are considered within the given HD-FEC (3.8 × 10⁻³) or SD-FEC (2 × 10⁻²) threshold (blue and orange dashed lines, respectively). The grey diamond markers show the performance when pumping the AlGaAs microresonator with a DFB chip. The wavelength-dependent BERs mainly result from the increased noise of the pre-amplifier at the edge of its operation band. The optimized receiving power for each channel is about 2–3 dBm. e, BER versus receiving power comparison between an on-chip Ge–Si PD and a commercial PD with the variation of the receiving power. The main limitation of the Ge–Si PD is the non-optimized frequency response (Methods).

Fig. 4. Reconfigurable MPF results.

a, Schematic of the setup to perform microcomb-based reconfigurable MPF. The time delays between comb lines are produced by on-chip spiral delay lines (setup 1) and dispersive propagation from a spool of SMF (setup 2). Scale bar, 200 μm. b, Optical spectra of Gaussian-apodization comb lines for BW programming (σ, Gaussian factor; blue, experiment (Exp); red, Gaussian fitting). c, RF filtering responses of the MPF with various passband BWs, based on the setup 1 (top) and setup 2 (bottom). The red dashed curves show the theoretical fitting results (Sim.) (Supplementary Note III). d, Proof-of-concept demonstration of RF filtering of a wideband RF signal. From top to bottom: RF spectra of original signal, signal after 1.1-GHz BW filter and signal after 0.9 GHz BW filter. e, f, Optical spectra (e) and corresponding RF responses (f) of the MPF with various FSRs, produced by modifying the comb line spacing and based on setup 2. ∆λ, wavelength distance between adjacent comb lines. g, Proof-of-concept demonstration on RF filtering of a complex dual-channel RF signal. From top to bottom: RF spectra of original signal, signal after 3.6-GHz FSR filter and signal after 7.2-GHz FSR filter.

Extended Data Fig. 1. Performance of building block devices.

a, Measured linewidth of the DFB laser. b, The measured 3dB bandwidth of the Si-Ge PD photodiode. c, Design, fabrication and measurement results of the 10:90 asymmetric MMI as on-chip monitor for silicon photonics engines. d, Optical image of the grating coupler (left) and its normalized transmission.

Extended Data Fig. 2. Turnkey AlGaAs dark-pulse microcomb generation.

a, Experimental setup. b, ECL and c, DFB laser chip driven comb spectra and the comb power variations along with the control signal in five consecutive switching tests.

Extended Data Fig. 3. Detailed information for data transmission.

a, Detailed experimental set-up for the odd/even test band for the comb-based silicon photonics data link. b, measured 100 Gbps PAM-4 eye diagrams from the sampling oscilloscope for each channel at receiving end.

Extended Data Fig. 4. Microcomb generation with different pump schemes.

a, Measured linewidth of the DFB laser. Comparison of the comb spectra pumped by b, a commercial external cavity laser and c, a DFB laser chip.

Extended Data Fig. 5. Dispersive delay-line MPF.

Experimental setup of the second dispersive delay-line scheme based TDL-MPF.

Extended Data Fig. 6. The response time measurement for the reconfiguration of RF filtering profiles.

a, Measured temporal response of the MRR under a square-wave electrical signal driving. b, MRR switched from minimum to maximum transmission, and c, maximum to minimum transmission. 90/10 rise/fall times are 15 μs and 53 μs, respectively.