Statistical analysis identifies initialization strategies for silicon photonic resonators, highlighting tradeoffs between speed and power efficiency.
Silicon photonic dense wavelength-division multiplexing (DWDM) links based on microresonators are a leading candidate for scaling optical interconnects in high-performance computing. However, fabrication-induced variations in resonant frequencies, free spectral ranges (FSRs), and thermal tuning efficiencies necessitate robust initialization strategies to align resonators with multi-wavelength laser sources. While many prior studies examine individual devices, comprehensive wafer-scale statistical analyses—and their implications for initialization algorithms remain scarce. Here, we report measurements from more than 4,600 fabricated resonators, including microdisks, tapered hybrid-bend (THB) racetracks, and microrings, fabricated on 300-mm wafers. We show that microrings exhibit the largest variation, while microdisks are significantly more uniform. Using these distributions, we evaluate initialization performance through Monte Carlo simulations of two widely used algorithms: barrel-shifting and global-search. Our results establish a clear tradeoff: barrel-shifting offers sub-millisecond initialization with modest power budgets for low-variation devices (e.g. microdisks), whereas global search, though approximately 300 times slower, achieves substantially lower power consumption for high-variation devices (e.g. microrings). This work provides the first wafer-scale statistical framework linking resonator variability to initialization outcomes, and offers practical design guidance for selecting initialization strategies that balance speed, power, and robustness.
No takes yet. Share an insight, caveat, or question.
Robinson et al. (2025) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: