Answers are grounded in this paper and cite their sources.
Waveguide crossing is the indispensable building block for large-scale photonic integrated circuits. However, current existing silicon waveguide crossing still suffers from excessive insertion loss, limited bandwidth, and large footprint, which severely restricts on-chip photonic circuit scalability. Here, we propose an ultra-compact waveguide crossing that achieves a 400-nm bandwidth (1250–1650 nm) with average insertion loss below 0.19 dB and crosstalk below −17 dB, using a novel, to the best of our knowledge, root-mean-square transmittance figure of merit (RMS-FOM) methodology. Meanwhile, the device maintains an ultra-compact footprint of 4 × 4 μm 2 . Comprehensive tolerance analysis reveals that the optimized devices sustain insertion losses below 0.4 dB at 1310 nm and 0.35 dB at 1550 nm across manufacturing deviations of up to ±100 nm, confirming the robustness and reliability of our device under realistic fabrication. Large-scale experimental validation demonstrates that the fabricated device exhibits low crosstalk of below −18.8 dB and −20 dB and insertion losses of 0.32 dB and 0.35 dB in two operating bands (1250–1350 nm and 1500–1600 nm). We believe the device establishes a new benchmark for silicon waveguide crossings by integrating optimized overall performance with an ultra-compact footprint, broadband, and fabrication tolerance, thereby enabling high-density on-chip optical interconnects in communication, data centers, and quantum technologies.
No takes yet. Share an insight, caveat, or question.
Zhou et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: