High-coherence parallelization in integrated photonics

Abstract

Coherent optics has profoundly impacted diverse applications ranging from communications, LiDAR to quantum computations. However, developing coherent systems in integrated photonics comes at great expense in hardware integration and energy efficiency. Here we demonstrate a high-coherence parallelization strategy for advanced integrated coherent systems at minimal cost. By using a self-injection locked microcomb to injection lock distributed feedback lasers, we achieve a record high on-chip gain of 60 dB with no degradation in coherence. This strategy enables highly coherent channels with linewidths down to 10 Hz and power over 20 dBm. The overall electrical-to-optical efficiency reaches 19%, comparable to that of advanced semiconductor lasers. This method supports a silicon photonic communication link with an unprecedented data rate beyond 60 Tbit/s and reduces phase-related DSP consumption by 99.99999% compared to traditional III-V laser pump schemes. This work paves the way for realizing scalable, high-performance coherent integrated photonic systems, potentially benefiting numerous applications.

Fig. 1. High-coherence parallelization based systems.

a Using a self-injection locked DFB laser, we pump a Si₃N₄ microresonator to generate the microcomb, which is then employed to drive the DFB laser array (iii). We can obtain a highly coherent parallel light source that combines the advantages of high power and high WPE of the DFB laser array (i) and narrow the linewidth of the microcomb (ii). b Conceptual illustration of the high-coherence parallelization system. This light source can find wide applications in integrated photonic scenarios and exhibits significant potential across various fields, such as communications, LiDAR, quantum information and neural networks.

Fig. 2. Fundamental characteristics of the self-injection locked microcomb and injection locking amplification.

a Photograph of the integrated DFB laser and a Si₃N₄ microresonator. b The link used for fundamental characteristics measurement. DFB distributed feedback laser, DEMUX demultiplexer, Circ circulator, MUX multiplexer, OSA optical spectrum analyzer, PM power meter, AOM acousto-optic modulator, PD photodetector, OSC oscilloscope. c Photograph of the slave DFB laser. d Spectrum of the microcomb pumped by the self-injection locked DFB laser. e Spectra of the comb line at 1550.62 nm before and after injection locking amplification, show that a gain of more than 60 dB can be achieved. f Combined spectra of each comb line after being filtered and injection locking amplified. g WPE of different cases. The purple triangles and blue circles represent the WPE of each comb line before and after injection locking amplification. The green triangle represents the overall WPE of the original microcomb. The dashed line represents the overall WPE of all comb lines after amplification including the power supplied to the source DFB laser. h Linewidths of different cases. The purple triangles and blue circles represent the intrinsic linewidths of the comb lines before and after injection locking amplification, respectively. Both of them have linewidths below 200 Hz, while the pump DFB laser has a linewidth over 20 kHz. i Single-sideband frequency noise of different amplification methods. Injection locking amplification has no significant noise-floor increment.

Fig. 3. Parallel coherent silicon photonic communications.

a Architecture of the coherent communication system utilizing our power-efficient, high-coherence parallel source. Photographs of the transmitter and receiver chips are shown at the top. Tx transmitter, Rx receiver, PRBC polarization rotator and beam combiner, PRBS polarization rotator and beam splitter, MPD monitor photodetector, VOA variable optical attenuator, Hyb 90^∘ hybrid, FA fiber array, TIA trans-impedance amplifier. b Combined spectra of modulated carriers for all 34 channels. c BERs across 34 wavelengths and 6 cores. d Recovered constellation diagrams across 34 channels and 6 cores. X Pol X polarization, Y Pol Y polarization.

Fig. 4. DSP reduction induced by high-coherence parallelization.

a Brief architecture and DSP modules in our strategy. Pre-proc pre-processing, EQ-MIMO equalization multiple-input-multiple-output, FOE frequency offset estimation, CPE carrier phase estimation. b (i) Using standalone DFB lasers as the carrier and the LO, both the frequency offset and the phase difference change rapidly, requiring frequent CPE and FOE in DSP. b (ii) DSP reduction results in the degradation of constellation diagrams when using standalone DFB lasers. c (i) In our scheme, the frequency offset is 0, and the phase difference is stable. Therefore, we can significantly reduce the CPE ratio and omit FOE. c (ii) Recovered constellation diagrams are well maintained with different ratios of CPE and FOE when using our strategy.