All-optical supercontinuum switching

The study addresses the central challenge in all-optical switching: achieving high fan-out and efficient switching without relying on resonant enhancement, which often limits speed, increases dispersion, and complicates channel separation. Conventional Kerr-based schemes require strong pulses to switch weaker ones, undermining fan-out. Resonant cavities and slow-light approaches enhance nonlinearity but introduce issues like group delay, reduced speed, and limited separation. Integrated photonics (e.g., silicon nitride waveguides, microring combs) promises high data rates and strong nonlinearities, yet a key missing component is an all-optical transistor enabling time-division multiplexing beyond 100 Gbit/s with near-unity fan-out. The authors build on prior proposals exploiting group-velocity matching between a dispersive wave (DW, in normal dispersion) and a soliton (in anomalous dispersion), creating a long-lived temporal lock akin to resonance but without cavity-related drawbacks. Earlier work showed a weaker DW can induce substantial soliton frequency shifts, satisfying fan-out, but required challenging pulse parameters. The hypothesis here is that using higher-order solitons undergoing soliton fission near the zero-dispersion wavelength can greatly enhance switching efficiency by enabling control over supercontinuum (SC) generation via an optical event horizon interaction, leading to fast, conservative, multi-gate logic functionality with high contrast, fan-out, and partial cascadability.

The paper situates its contribution within several threads of prior work: (1) Optical switching and fan-out limitations; proposals leveraging resonant enhancement in cavities improve nonlinearity but suffer from speed and dispersion constraints and limited channel separation (e.g., Kerr-based devices and cavity-enhanced schemes). (2) Slow-light mediated χ(3) enhancement faces similar dispersion-induced drawbacks. (3) Integrated photonic platforms such as silicon nitride waveguides enable on-chip SC generation and frequency combs, supporting extremely high data rates and potentially revolutionizing communications with cavity solitons; however, a practical all-optical transistor for ultrafast TDM remains absent. (4) Optical event horizon analogies: reflections at moving refractive index fronts inspired by fluid dynamics have been explored experimentally and theoretically in fibers, leading to studies on diametric acceleration, quantum bouncing, rogue waves, and stimulated Hawking radiation. Early implementations exhibited limited efficiency. (5) Prior theoretical proposals by the authors demonstrated that group-velocity-matched DW–soliton interactions can control soliton dynamics and SC generation, with significant soliton frequency shifts driven by weaker DWs (satisfying fan-out), but required demanding pulse parameters. This work builds on those foundations, targeting efficient, deterministic switching within soliton fission-driven SC generation.

Experimental: An OPCPA source produced sub-7 fs pulses (>1.5 µJ, 200 kHz, 650–1150 nm). A folded double-pass 4f pulse shaper (knife-edges + folding mirrors) selected two spectral components to form a DW and a higher-order soliton with adjustable relative delay (mirror on translation stage, 0.2 fs nominal resolution) and tunable energy ratio via an attenuator. The two collinear pulses were coupled by a microscope objective (NA 0.40) into a 0.47 m photonic crystal fiber (NKT NL-PM-750) with two zero-dispersion wavelengths (ZDW1 ≈ 750 nm, ZDW2 ≈ 1250 nm). Output was collimated and analyzed by an optical spectrum analyzer (Ando AQ-6315A). Input wavelengths: DW at 680 nm and soliton at 800 nm, chosen symmetrically around ZDW1 to achieve nearly identical group velocities (≈2.02 × 10^8 m/s), with DW slightly slower. Input energies: soliton 1 nJ (soliton order N ≈ 25), DW 750 pJ (energy ratio ≈ 3/4). Pulse durations: soliton 200 fs, DW 1 ps. The relative timing between pulses was scanned to explore interaction regimes (leading, overlapping, trailing). Measurements used two spectrometers to cover spectral ranges for DW and soliton outputs. Experimental observables included spectral shifts and the presence/absence of SC components at designated output wavelengths (e.g., ~540, 680, 1100 nm). Numerical simulations: Pulse propagation modeled by the generalized nonlinear Schrödinger equation (GNLSE) for the complex field envelope and, for cross-check, the forward unidirectional propagation equation for the analytic signal; both gave consistent results. The GNLSE included higher-order dispersion (β2 to β11), self-phase modulation, self-steepening, and Raman response (instantaneous Kerr and delayed Raman with fr = 0.18, τ1 = 12.2 fs, τ2 = 32 fs). Fiber dispersion expansion around ωs = 2.325 rad fs^-1 (λ = 810 nm) with coefficients: β2 = −0.38 ps^2 km^-1, β3 = 4.97×10^-3 ps^3 km^-1, β4 = 2.68×10^-4 ps^4 km^-1, β5 = −1.25×10^-5 ps^5 km^-1, β6 = 7.7×10^-7 ps^6 km^-1, β7 = −5.83×10^-8 ps^7 km^-1, β8 = 3.0×10^-9 ps^8 km^-1, β9 = −3.74×10^-10 ps^9 km^-1, β10 = −7.45×10^-11 ps^10 km^-1, β11 = 1.696×10^-11 ps^11 km^-1; γ = 3.8 W^-1 km^-1; anomalous dispersion between ZDWs (λzDW1 ≈ 1252 nm, λzDW2 ≈ 743 nm). Initial conditions were sums of hyperbolic secant pulses: a soliton (τs ≈ tens of fs in some simplified cases; in experiment 200 fs higher-order soliton that undergoes fission) and a DW (τDW ≈ 70 fs in single-soliton simulations; 1 ps in experiments) with adjustable delay. Quantum noise seeding was included in selected simulations via spectral synthesis to assess robustness. Numerical integration used a split-step pseudospectral method. Simulations reproduced experimental spectra, delay-dependent switching maps, temporal/spectral evolutions, and XFROG traces, and explored sensitivity to slight dispersion variations and input chirp. Ensemble averaging over 100 disorder instances was used for comparison to experiments.

• Demonstrated an all-optical switching scheme that controls supercontinuum generation via interactions between a higher-order soliton fission cascade and a group-velocity-matched dispersive wave, exploiting an optical event horizon analogy. - Achieved three parallel logical AND gate functionalities with distinct output channels, with high switching contrast. Example channels: 540 nm (DW2 out) present when soliton and DW overlap (S & DW), suppression of the 1100 nm soliton output in that case, and a cascadable 680 nm output channel that remains unshifted when S & DW holds. - Spectral shifts observed: DW blue-shifted from 680 nm to ≈540 nm upon interaction; soliton red-shifted towards ≈1100 nm in the absence of interaction (limited by ZDW2 at 1250 nm), with shifts controlled by relative delay. Output and input channels are separated by at least ~100 nm. - Fan-out: With an experimental energy ratio DW/soliton ≈ 3/4, strong control is demonstrated; simulations show control persists down to ≈1/5 energy ratio, indicating the ability to switch larger soliton energies with weaker DWs and potentially switch multiple solitons in parallel. - Timing control: Delay scans reveal regimes of acceleration (leading-edge collision) or deceleration (trailing-edge collision) of ejected solitons, with corresponding blue/red shifts. Direct temporal overlap inhibits soliton fission and enables strong four-wave mixing, generating new components ω1 = 2ωDW − ωs and ω2 = 2ωs − ωDW. - Robustness and suppression of rogue events: Ensemble simulations and prior theory indicate that the presence of the DW suppresses soliton-soliton collisions and rogue soliton formation, enabling deterministic gate behavior particularly for the DW-related outputs. - Speed and practicality: Switching occurs on femtosecond time scales; with two ZDWs, critical biasing is avoided for the 1100 nm port, and the 680 nm port is cascadable as it remains unshifted under S & DW conditions. - Agreement: Simulations using measured/estimated fiber dispersion and Raman parameters qualitatively match experiments; slight quantitative differences in exact frequency shifts are attributed to uncertainties in dispersion and possible input chirp.

The findings validate the hypothesis that leveraging higher-order soliton fission near a zero-dispersion wavelength and interacting it with a group-velocity-matched DW provides efficient, conservative all-optical switching. By exploiting the optical event horizon condition, the interaction yields strong, delay-controlled frequency shifts and even suppression of the fission process under overlap, enabling logic operations without resonant enhancement and its associated penalties. The approach directly addresses the fan-out bottleneck by demonstrating that a weaker DW can control stronger soliton outputs and even multiple solitons simultaneously, thus meeting a key criterion for optical logic. The use of a fiber with two ZDWs confines soliton red-shifts and stabilizes the output channel near 1100 nm, effectively removing the need for critical biasing. The unshifted 680 nm output under AND conditions offers partial cascadability as per Miller’s criteria. The ability to perform three parallel AND operations with channels separated by >100 nm aids channel isolation. Numerical modeling corroborates the physical mechanisms (group-velocity locking, Raman-driven shifts, four-wave mixing) and supports the deterministic nature of the switching, including suppression of solitonic collisions and rogue events. Overall, the work demonstrates a practical pathway toward ultrafast optical logic elements capable of high-speed TDM, with potential for significant energy scaling in integrated platforms.

This work experimentally demonstrates an all-optical transistor-like switch based on controlled interactions between a dispersive wave and a higher-order soliton undergoing fission, enabling three concurrent AND gate functionalities with femtosecond switching, high contrast, fan-out, and partial cascadability without resonant enhancement. The scheme controls SC generation and soliton dynamics via group-velocity-matched interactions akin to an optical event horizon. The use of fibers with two ZDWs stabilizes output channels and avoids critical biasing. Simulations confirm the mechanisms and suggest scalability to multi-soliton control and improved fan-out (down to ~1/5 energy ratio). Future directions include dispersion-engineered fibers or integrated waveguides (e.g., silicon-based platforms) to drastically reduce pulse energy requirements by orders of magnitude, enabling THz-rate TDM with few-watt to sub-watt power budgets. Engineering ZDW2 (e.g., to ~1600 nm) together with erbium amplification and frequency conversion could render additional channels cascadable; further dispersion engineering could map soliton outputs back to DW inputs. Ultimately, the approach lays groundwork for practical ultrafast optical logic circuits, though substantial development remains to reach full circuit-level functionality.

• Current implementation requires relatively high pulse energies (nJ-level soliton, hundreds of pJ DW) in fiber; integrated platforms are proposed to reduce energy by orders of magnitude. - Only partial cascadability is presently demonstrated (680 nm channel); achieving full cascadability across all channels requires additional dispersion engineering, amplification, and frequency conversion stages. - Exact quantitative agreement between experiment and simulation is sensitive to precise fiber dispersion parameters and potential input chirp, indicating some uncertainty in device-level predictability without thorough characterization. - The soliton fission process is inherently susceptible to noise; while the presence of a DW suppresses rogue events and collisions, ensuring fully deterministic behavior across all functionalities (especially the soliton channel AND operation) may require careful control and stabilization. - The demonstrated logic functionality is an early-stage step toward practical circuits; substantial integration, scaling, and system-level design are needed for complex, large-scale optical logic.