Adiabatic and Deterministic Routes to Soliton Combs in Non-Hermitian Kerr Cavities
We present a cardinal solution for the long-standing and fundamental problem associated with the adiabatic, reversible, and controlled excitation of both dark and bright solitons in Kerr micro-resonators with normal group velocity dispersion. Our findings stem from the inclusion of a localised non-Hermitian potential, which we use to drastically reshape the characteristic collapsed snaking structure associated with such solitons. Consequently, we demonstrate a novel snaking-free bifurcation landscape where solitons of all possible widths are continuously connected via the dynamic change of the cavity detuning, and hence dissipative localised states of unprecedentedly high pump-to-comb conversion efficiencies can be excited in an adiabatic, deterministic, and reversible fashion. Our fundamental discovery has practical implications of paramount importance for frequency comb generation in all-normal dispersion cavities, which are key to comb generation in most spectral regions away from the telecom bands.
💡 Research Summary
The authors address a long‑standing obstacle in normal‑group‑velocity‑dispersion (GVD) Kerr micro‑resonators: the deterministic, adiabatic excitation of both dark and bright dissipative solitons and their associated frequency combs. In the conventional setting, solitons in the normal‑GVD regime exhibit a “collapsed snaking” bifurcation structure. This structure consists of a cascade of saddle‑node (SN) bifurcations separated by intrinsically unstable branches, which makes it extremely difficult to select a particular soliton width or to switch between dark and bright states without resorting to stochastic perturbations, abrupt pump‑power jumps, or self‑injection locking schemes.
The key innovation of the paper is the introduction of a localized non‑Hermitian potential V(x) = m e^{−iϕ} cos(qx) Θ(3π/2q − |x|) into the driven‑damped nonlinear Schrödinger equation that models the intra‑cavity field. Physically, such a potential can be realized by integrating phase and amplitude electro‑optic modulators (EOMs) into the micro‑cavity, allowing independent electrical control of the depth m, spatial frequency q, and phase ϕ. The non‑Hermitian nature (complex refractive‑index modulation) provides a pinning mechanism for the oscillatory tails of the soliton fronts, thereby suppressing the collapse of the snaking diagram.
As m is increased, the unstable branches (U₁, U₂, …) disappear because the corresponding SN points coalesce into a series of cusp bifurcations C₁, C₂, …. Each cusp marks a qualitative change in the solution set: the unstable region is eliminated and the stable soliton branches of different widths become continuously connected along the same detuning (δ) axis. When the first cusp C₁ is removed (m ≳ 0.61 for the parameters studied), even the widest dark soliton can be accessed directly from the homogeneous continuous‑wave (CW) state by slowly red‑shifting the laser detuning. Further red‑shifts lead to an abrupt SNₙ transition where a single dark soliton splits into a pair of bright solitons. Conversely, a blue‑shift can abruptly nucleate a bright soliton from the CW background, after which a subsequent red‑shift allows smooth manipulation of its width. Thus, the entire family of solitons—dark, double‑dark, and bright—can be traversed in a reversible, adiabatic manner using only the detuning as a control knob.
The spatial frequency q controls the width of the potential and therefore the number of cusps that can be generated. Reducing q (e.g., q ≈ 0.03) widens the potential, creates many more cusps, and ultimately yields a fully unfolded bifurcation diagram in which all stable localized states (both dark and bright) lie on a single continuous branch emerging from the CW solution. In this limit the classic Maxwell point disappears, there are no unstable intervals, and the system becomes a true “turn‑key” platform for soliton comb generation.
Numerical simulations of the full time‑dependent equations confirm the theoretical predictions. Figure 3 shows a smooth transition from the CW state to a dark soliton family as δ is slowly increased, followed by a sudden jump to a double‑bright state at the SNₙ point. Figure 4 demonstrates that, with a very shallow q, the pump‑to‑comb conversion efficiency η (estimated from the intra‑cavity line contrast) can exceed 30 %, far surpassing typical efficiencies (<10 %) reported for normal‑GVD resonators that rely on dispersion engineering or injection locking.
Beyond the specific Kerr‑microresonator context, the work highlights a general strategy for eliminating collapsed snaking in a broad class of pattern‑forming systems (Swift–Hohenberg, Ginzburg‑Landau, Boussinesq, etc.) by employing localized non‑Hermitian perturbations. Experimentally, the required non‑Hermitian potential can be implemented with existing silicon‑nitride or lithium‑niobate platforms that already integrate high‑speed EOMs, making the approach readily translatable to current photonic foundries.
In summary, the paper delivers a comprehensive solution to the deterministic excitation of soliton frequency combs in all‑normal‑dispersion Kerr cavities. By reshaping the bifurcation landscape through a tailored non‑Hermitian potential, the authors achieve:
- Elimination of multi‑stability and unstable branches,
- Continuous, reversible control of soliton width and type via simple detuning sweeps,
- Substantially higher pump‑to‑comb conversion efficiencies,
- A practical, electrically tunable implementation compatible with integrated photonic technologies.
These advances open a clear path toward low‑power, high‑efficiency comb sources for telecommunications, LIDAR, spectroscopy, and other applications that demand robust, all‑normal‑dispersion operation.
Comments & Academic Discussion
Loading comments...
Leave a Comment