--- title: Stokes Microcombs in Silicon Nitride url: https://www.emergentmind.com/topics/stokes-microcombs-in-silicon-nitride type: topic --- # Stokes Microcombs in Silicon Nitride Stokes microcombs in silicon nitride (Si₃N₄) microresonators comprise optically generated frequency combs based on stimulated Raman scattering (SRS), observed in on-chip devices with normal group velocity dispersion (GVD). These microcombs are characterized by the generation of Stokes lines, forming a broadband frequency comb shifted from the pump frequency by the characteristic Raman frequency of the medium. Improvements in nanofabrication have resulted in ultra-high quality-factor (Q) values up to $10^7$, enabling the manifestation of pronounced nonlinear effects, including the interplay between Kerr and Raman nonlinearities essential for Stokes microcomb formation in Si₃N₄ photonic platforms [2511.05767]. ## 1. Silicon Nitride Microresonator Platform Si₃N₄ microresonators employ a waveguide geometry of 300 nm × 2 500 nm (Si₃N₄ core, SiO₂ cladding), engineered to support high-Q optical modes. Key platform parameters are: | Parameter | Critically Coupled Ring | Undercoupled Ring | |----------------------------|-----------------------------------|------------------------------| | Intrinsic Q ($Q_0$) | $\sim$10–20 × 10⁶ | $\sim$10–20 × 10⁶ | | Loaded Q | 3.7–5 × 10⁶ (η=0.5) | $\sim$10 ×10^6$ (η≈0.25) | | FSR ($D_1/2\pi$) | 202 GHz | 184 GHz | | GVD ($D_2/2\pi$) | –39.4 MHz | –32.6 MHz | The normal GVD regime ($\beta_2>0$ in this sign convention, $D_2<0$) is central to Stokes microcomb generation. The observed stimulated Raman shift is $\Omega_R/2\pi \approx 9$ THz, with a bandwidth of approximately 5 THz. ## 2. Mean-Field Model: Lugiato–Lefever Equation with Raman Term The dynamics of Stokes microcombs are governed by a mean-field model extending the Lugiato–Lefever equation (LLE) to include Raman interactions. The intracavity envelope $E(\tau,t)$, where $\tau$ represents the fast time and $t$ the slow time, is described by: \[ \frac{\partial E}{\partial t} =\left[-\alpha - i\delta_0 + i\frac{\beta_2}{2}\frac{\partial^2}{\partial \tau^2} + i\gamma|E|^2\right]E + F + i\,\Gamma_R\,E\;\otimes\;h_R(\tau)\,. \] Here: - $\alpha = \kappa/2$: cavity half-linewidth, - $\delta_0 = \omega_p-\omega_0$: pump–resonance detuning, - $\beta_2$: GVD coefficient (normal in these experiments), - $\gamma = n_2\omega_0/(c A_\text{eff})$: Kerr nonlinear coefficient, - $F = \sqrt{\kappa_c P_\text{in}/(\hbar\omega_0)}$: pump amplitude, - $\Gamma_R$: Raman index perturbation parameter, - $h_R(\tau)$: Raman response function, typically $H(\tau)\,(\tau_1^{-2}\,\tau e^{-\tau/\tau_2})$ with $\tau_1 \approx 12.2$ fs, $\tau_2 \approx 32$ fs. This formalism captures the critical interplay between instantaneous Kerr and delayed Raman nonlinearities, enabling analysis of both Kerr- and Raman-dominated comb generation and transition phenomena such as platicon formation. ## 3. Experimental Pump Schemes and Threshold Phenomena Two principal experimental pump configurations are deployed: 1. **Tunable external-cavity laser (ECL) with isolator and EDFA:** Lensed-fiber coupled to chip, $P_\text{in}$ controllable from 30–90 mW. Pump wavelength is red-detuned and swept across resonance. Stokes comb onset occurs at $P_\text{in} \approx 5$–9 mW (first Stokes at 1601–1663 nm for pump 1520–1570 nm), with cascaded Stokes combs forming for $P_\text{in} \approx 14$–60 mW and spanning $>100$ nm. 2. **Self-injection-locked (SIL) distributed-feedback (DFB) diode laser:** $\lambda \approx 1546$ nm, butt-coupled to chip (no isolator), with on-chip $P_\text{in} \approx 6$–36 mW. The SIL configuration stabilizes laser frequency and detuning via back-reflection phase. Kerr or Raman comb states are controllable by varying the locking phase (i.e., laser–chip separation). These methods provide in situ tunability of the nonlinear regime, enabling deterministic access to either predominantly Kerr- or Raman-driven microcomb states. ## 4. Comb Spectra, Temporal Regimes, and Platicon Generation The microresonator supports distinctive comb regimes contingent on pump parameters and detuning: - **Predominantly Raman comb:** Achieved via ECL scan (e.g., pump $\lambda_p=1564$ nm, $P_\text{in}=60$ mW). The first Stokes line appears at 1640 nm (9 THz shift). Comb FSR is $\sim$202 GHz; span exceeds 100 nm. Stokes line power is $-5$ dB to $-8$ dB relative to pump. A weak Kerr platicon remains at the pump, $>30$ dB below the pump level. - **Predominantly Kerr comb:** Achieved under appropriate SIL phase. FSR is again 202 GHz; the spectrum exhibits a characteristic dark-pulse (platicon) envelope with a spectral dip at pump wavelength, spanning $\sim$10 nm centered at 1546 nm, with no measurable Stokes content above 1600 nm. - **Temporal profiles:** Both Raman- and Kerr-platicons reconstructed numerically exhibit pulse durations $\tau_\text{pulse} \approx 3$ ps, each manifesting as intensity dips against a continuous-wave (cw) background. Platicon formation at the Stokes frequency is thus experimentally and numerically corroborated, with synchronization between pump and Stokes platicons inferred from both spectral and temporal data. ## 5. Numerical Simulations and Stability of Stokes Microcombs Numerical analysis employs coupled-mode equations for forward and backward, pump and Stokes fields (see Eqs. (1)–(4) in the original work). Key normalized parameters include: - Dimensionless pump strength: $f=3.5\ldots12$ - Normalized dispersion: $D_2/\kappa \approx -0.2\ldots -0.1$ - Raman fraction: $f_r \approx 0.18$ - Shock time: $\tau_r \approx 0.3$ - Normalized Raman gain: $G_r \approx 0.38$ - CW/CCW coupling: $\beta \approx 6$ With periodic fast-time boundary conditions, the simulations reveal: - Stokes sidebands appear with increasing detuning ($\delta$), - Platicon steps at both pump and Stokes frequencies synchronize for normalized $\delta \approx 20$, - Spectral shapes and spans ($\Delta\lambda \sim 2$ nm in normalized units) closely track experiments after rescaling, - Stable platicon states persist over detuning intervals $\Delta\delta\approx 5$; disabling Raman gain eliminates platicon seeding—establishing SRS as critical for comb initialization and stability. ## 6. Regime Switching: Kerr–Raman Control via Injection-Locking Phase Regime switching between Kerr- and Raman-dominant comb operation is achieved by sub-micrometer tuning of the SIL laser–chip separation, influencing the SIL phase $\phi_\text{lock}$ and consequently the effective pump detuning $\delta_\text{eff}$. The operational regimes are: - $\phi_\text{lock} = \phi_\text{Kerr}$: $\delta_\text{eff}$ selects the Kerr four-wave mixing regime, producing platicon at the pump with no Stokes comb. - $\phi_\text{lock} = \phi_\text{Raman}$: $\delta_\text{eff}$ favors SRS threshold, yielding a strong Stokes comb and suppressing Kerr comb by pump power depletion. Switching does not require external modulators, instead relying on precise mechanical adjustment of the chip position, yielding all-electronic control in a monolithic platform. ## 7. Applications and Implications Stokes microcombs in Si₃N₄ microresonators extend the functionality of integrated photonic frequency combs in several domains: - On-chip Raman lasers and comb sources at 1.6–1.7 µm, relevant for gas spectroscopy (e.g., methane absorption features), - Broadband wavelength-division multiplexing (WDM) beyond the C-band, with $>100$ nm comb span, - Dual-comb spectroscopy by simultaneous generation of pump and Stokes combs with locked repetition rates, - Integrated microwave photonics leveraging 200 GHz comb spacing for RF-generation and optical clocks, - Electrically reconfigurable, CMOS-compatible frequency comb sources without additional moving parts. *This suggests* that the demonstrated method for regime control and platicon synchronization in Si₃N₄ microresonators establishes a foundational platform for advanced nonlinear photonic devices and new approaches to spectroscopic sensing, microwave photonics, and frequency synthesis. Source: https://www.emergentmind.com/topics/stokes-microcombs-in-silicon-nitride