Photonic Dissipation Control for Kerr Soliton Generation in Strongly Raman-Active Media

Microcavity solitons enable miniaturized coherent frequency comb sources. However, the formation of microcavity solitons can be disrupted by stimulated Raman scattering, particularly in the emerging crystalline microcomb materials with high Raman gain. Here, we propose and implement dissipation control—tailoring the energy dissipation of selected cavity modes—to purposely raise or lower the threshold of Raman lasing in a strongly Raman-active lithium niobate microring resonator and realize on-demand soliton mode locking or Raman lasing. Numerical simulations are carried out to confirm our analyses and agree well with experiment results. Our work demonstrates an effective approach to address strong stimulated Raman scattering for microcavity soliton generation.

Figure 1

Bubbles: Soliton-forming modes in the optical microresonator (blue-colored region), numbered with respect to the pump mode. Red arrowed lines: Kerr nonlinear coupling with a rate of $g_k$. Blue arrowed lines: First SRS from the pump mode to the Stokes mode $R$ with a rate of $g_R$ and subsequent anti-SRS at mode $-R$. [5]. Gray coils: Cavity intrinsic dissipation rate (${\kappa }_{i,\mu }$). Green coils: Cavity-waveguide coupling rate (${\kappa }_{e,\mu }$).

Figure 2

(a) Threshold ratio ${ϵ}_{R,\mathrm{th}}/{ϵ}_{S,\mathrm{th}}$ as the Raman mode detuning ${\delta }_R/D_1$ and external coupling ratio ${\kappa }_{e,R}/{\kappa }_i$ are varied. The dashed curve separates the ${ϵ}_{R,\mathrm{th}}<{ϵ}_{S,\mathrm{th}}$ (I) and ${ϵ}_{R,\mathrm{th}}>{ϵ}_{S,\mathrm{th}}$ (II) regimes. (b)–(d) Simulated intracavity spectra for Raman lasing and soliton combs at ${\delta }_R/D_1=0$ and ${\kappa }_{e,R}/{\kappa }_i=1$, 1.5, and 2 marked by asterisks in (a), respectively. Insets show the temporal profiles. For (a)–(d), ${\kappa }_i/(2\pi )={\kappa }_e/(2\pi )=235\mathrm{MHz}$, $D_2/{\kappa }_i=8.6\times {10}^{-2}$. The soliton self-frequency shifts in (c),(d) are 0.9 THz.

Figure 3

(a) Schematic of a self-interferenced microring. Purple (pink) shaded line: The inner (outer) arm of the interferometer, with its length denoted as $L_{i(o)}$. (b) False-color SEM of the U-ring. (c) Schematic illustration of modulated ${\kappa }_{e,\mu }$ (brown line) to suppress Raman lasing. Cyan peaks: soliton-forming modes. Red arrow marks the pump mode. Blue shaded profile: the Raman gain. (d) Cyan line: measured transverse-electric transmission of (b). Dashed brown line: predicted ${\kappa }_e$ curve from Eq. (3). Cyan dots: extracted ${\kappa }_{e,\mu }$ from the measured transmission by fitting the linewidths of the resonances.

Figure 4

(a) External coupling rate ${\kappa }_{e,\mu }$, marked alternately with cyan or green dots, are extracted from the corresponding cyan or green dotted curve tracing the measured comb line powers in panel (b). The gray line in the background is the predicted continuous ${\kappa }_e$ curve. The deviation of the extracted external couplings from the theoretically calculated may be ascribed to local mode crossings that affect the extraction of ${\kappa }_{e,\mu }$. (b) The measured and normalized soliton comb output spectrum with an FSR of $\sim 445.7\mathrm{GHz}$ under on-chip pump power of 200 mW. Cyan or green dotted curve: Spectral envelope. Inset: Comb power under red-to-blue pump scan ($0.25\mathrm{GHz}/\mathrm{s}$). (c) The measured and normalized first-order Raman lasing spectrum. Inset: Stokes light power trace under the same scan for the inset of (b). (d) Cyan or green dots: ${\kappa }_{e,\mu }$ for the simulation, calculated by Eq. (3) with ${\kappa }_0/(2\pi )=240\mathrm{MHz}$ and $c/(L_o{n}_o-L_i{n}_i)=900\mathrm{GHz}$. Gray line is the same to (a). (e),(f) The simulated and normalized soliton comb and Raman lasing spectra on the output waveguide. Cyan or green dotted curve: Spectral envelope imposed by ${\kappa }_{e,\mu }$ (cyan or green dots) in (d). Insets: Intracavity temporal profiles. The time interval between the soliton and the feedback pulse is determined by the path difference $(L_o{n}_o-L_i{n}_i)/c=1.1\mathrm{ps}$.