A mode-locked random laser generating transform-limited optical pulses

The study addresses whether a random laser—whose mode spacing is inherently ill-defined due to stochastic feedback from randomly distributed scattering centers—can be mode-locked to generate transform-limited pulses. Conventional mode-locked lasers achieve transform-limited pulses via precise phase locking of uniformly spaced cavity modes. Random lasers, however, lack well-defined spacing, complicating phase locking. The authors propose a simplified one-dimensional random fiber laser in which gain and scattering are spatially separated: extremely weak Rayleigh backscattering from a short, arbitrarily selected section of standard single-mode telecom fiber provides a faint, spectrally selective distributed mirror, while time-gated amplification synchronizes feedback from only the addressed section. This approach aims to impose narrow spectral filtering that fixes the lasing wavelength and ensures cavity mode phase locking, enabling transform-limited pulse generation across tunable durations and repetition rates, with potential impact in spectroscopy, telecommunications, and distributed sensing.

The paper situates the work within several domains: (1) ultrafast lasers and mode locking in conventional cavities, which have produced transform-limited pulses since the 1960s; (2) random lasers based on multiple scattering and stochastic feedback, including demonstrations of coherent feedback and reports on mode dynamics and attempts at mode locking in random media (dyes, powders, terahertz devices); (3) random fiber lasers using distributed Raman or Brillouin feedback over kilometer scales and hybrid architectures with lumped gain (SOA/EDFA). Prior studies have shown quasi mode-locking and picosecond pulse generation in random fiber lasers, but not Fourier-transform-limited operation. The authors note that the Rayleigh backscattering spectral “fingerprint” of fiber sections is well known from phase-sensitive OTDR and that weak, long random gratings behave like faint, long FBGs with narrow peaks. Building on these insights, the present work aims to realize true synchronous mode locking to the Fourier limit in a random laser using only centimeter–meter scale backscattering sections and extreme net gain to overcome cavity loss.

Experimental architecture: A sigma-configuration fiber laser cavity comprising a fiber loop and an open-ended patch. The loop contains a gated semiconductor optical amplifier (SOA) and 125 m of delay fiber connected via a 50/50 fused coupler. The open-ended branch includes a 27 m erbium-doped fiber (EDF) amplifier without isolators (providing bidirectional CW gain) and a passive single-mode telecom fiber spool serving as the distributed random mirror via Rayleigh backscattering. In some experiments, an additional commercial EDFA and an adjustable spectral filter are inserted in the loop to increase gain and suppress ASE. Gain gating and addressing: The SOA is driven by nanosecond current pulses (repetition rate 0.714–1.220 MHz) to open a gain window synchronized with the round-trip time such that only backscattered light originating within a selected short section (3 cm–2 m) of the passive fiber experiences periodic amplification. The length of the addressed section is set by the electrical pulse duration (e.g., ~1 m for ~10 ns pulses). Light scattered outside the addressed section does not overlap the gain window and decays. Cavity design considerations: The 125 m intra-cavity delay ensures that doubling the shortest possible cavity length exceeds any point in the addressed fiber, uniquely mapping repetition rate to a single cavity length. Extremely low backscattering (≈ −72 dB/m mean) necessitates very high net round-trip gain (on the order of 100 dB); coherent reflectivity spikes ∼10 dB above average Rayleigh levels are required to reach threshold. Dispersion and nonlinear effects are negligible given powers and cavity lengths. Measurements: Temporal profiles are recorded with a 26 GHz amplified photodiode and a 3.5 GHz oscilloscope. High-resolution optical spectra are measured on an OSA with 20 MHz frequency resolution (0.16 pm). Mode coherence is characterized via self-homodyne detection using a Mach–Zehnder interferometer with ~100 km path imbalance; polarization is adjusted to overlap pulses and obtain mode-beat spectra around the repetition tone and harmonics. Additional single-shot synchronized temporal and spectral acquisitions are performed to directly compare pulses with their Fourier transforms. Simulations: Rayleigh backscattering spectra are modeled by summing contributions from random-index segments (e.g., d = 5 mm) to obtain frequency-dependent reflectivity and peak width statistics for section lengths 0.5–2 m over 10–100 GHz spans. Materials: Experiments predominantly use dispersion-shifted telecom fiber for the backscattering section; standard SMF-28 is also used when adding extra amplification. Pulse durations explored span 0.34–20 ns; repetition rates 0.714–1.22 MHz. Environmental conditions are laboratory ambient (no temperature stabilization) with noted acoustic background influencing mode linewidth.

• Transform-limited random laser pulses: The mode-locked random fiber laser produces pulses that are exactly or nearly transform-limited across durations from 0.34 ns to 20 ns and cavity lengths 170–285 m. The pulse duration is set by the SOA gate, which simultaneously selects the effective distributed mirror length and thus the spectral bandwidth, maintaining the time-bandwidth limit.
• Spectral bandwidth versus duration: For 5, 10, and 20 ns electrical drive pulses, measured optical FWHM bandwidths are ~227 MHz, 125 MHz, and 65 MHz, respectively, closely matching sinc^2 Fourier transforms of the temporal profiles and inversely proportional to pulse duration.
• Sub-nanosecond operation: Single-shot measurements confirm a 340 ps Gaussian pulse whose spectrum matches the Gaussian Fourier transform without width adjustment; time-bandwidth products are ~0.89 for square-like pulses and ~0.44 for Gaussian-like pulses as expected.
• Mode coherence and linewidth: Self-homodyne interferometry shows all optical modes share the same linewidth (~4.5 kHz), with a bell-shaped beat around the repetition tone and harmonics, indicating phase-locked modes. The linewidth is consistent with ambient acoustic noise in the ~200 m fiber cavity.
• Laser threshold behavior: Clear random-laser thresholds are observed. Longer addressed sections (longer pulses) provide higher feedback, reducing EDFA pump power at threshold and increasing slope efficiency. Modeling and measurements indicate that coherent reflectivity spikes about 10 dB above the average Rayleigh level are necessary to reach lasing.
• Tunability: Repetition rate is adjustable from ~0.714 to 1.22 MHz; pulse duration from ~0.34 to 20 ns; temperature tuning of the lasing wavelength is feasible due to the Rayleigh spectral fingerprint of the addressed section.
• Application in distributed temperature sensing: By addressing a ~9.0 cm section (900 ps pulses) immersed in water and scanning repetition rate along the fiber, the system measures wavelength (frequency) shifts with high SNR, demonstrating ~9.0 cm spatial resolution, frequency sensitivity ≈1.19 GHz/K, and precision of 0.0033 K. Direct frequency readout avoids post-processing and is robust to power fluctuations.
• Extreme gain regime: Lasing with centimeter-scale Rayleigh feedback requires net gains on the order of 100 dB, among the highest reported, yet is achievable with the hybrid SOA/EDFA architecture.

The results demonstrate that a short, weakly reflecting Rayleigh-scattering section of standard telecom fiber acts as a narrow, colored distributed mirror whose dominant spectral peak defines the lasing wavelength and bandwidth. By synchronizing gain to the time-of-flight from that addressed section, only its backscattered field is periodically amplified, enabling synchronous mode locking of many cavity modes to the phase of the modulation waveform. This mechanism yields pulses at or near the Fourier limit across a wide range of durations and repetition rates, despite the intrinsically random mode spacing of the cavity. The findings validate the hypothesis that separating gain from scattering and using time-gated amplification can impose phase coherence in a random laser to the transform limit. The approach provides high spectral efficiency and stability suitable for precision sensing: the sharp, transform-limited line allows highly sensitive detection of thermally induced refractive-index changes in the addressed section, enabling cm-scale distributed temperature sensing with mK precision. Beyond sensing, the ability to generate transform-limited nanosecond-to-sub-nanosecond pulses from a random laser can impact spectroscopy and telecommunications, offering narrow linewidths and electronically controllable parameters without conventional high-reflectivity mirrors.

The study introduces a mode-locked random fiber laser that leverages the Rayleigh backscattering ‘fingerprint’ of short telecom fiber sections, together with time-gated amplification, to generate transform-limited pulses with tunable duration and repetition rate. Key contributions include experimental verification of phase-locked cavity modes with kHz-level linewidths, confirmation of Fourier-limited temporal-spectral pairs from 340 ps to 20 ns, and a practical demonstration of distributed temperature sensing with 9.0 cm spatial resolution and 0.0033 K precision. The work shows that centimeter-scale Rayleigh feedback is sufficient for laser action when combined with extreme net gain, opening paths to simpler, robust, and highly sensitive sensing architectures. Future directions include pushing into the picosecond regime (electronics-limited here), enhancing Rayleigh feedback via UV exposure or engineered scattering fibers to relax gain requirements, extending the concept to other scattering media, and exploring free-space implementations for lasing-LIDAR and studies of quantum-limited pulse initiation.

• Extreme gain requirement: Achieving threshold with centimeter-scale Rayleigh feedback requires net round-trip gains on the order of 100 dB; an additional EDFA and spectral filtering were sometimes necessary.
• Electronics-limited pulse duration: Minimum pulse durations below ~0.3 ns were not reached due to available drive electronics, constraining exploration of shorter regimes.
• Environmental noise: The observed ~4.5 kHz mode linewidth is limited by acoustic perturbations of the ~200 m fiber cavity; environmental isolation or stabilization could further improve coherence.
• Random reflectivity dependence: Lasing relies on coherent reflectivity spikes (~10 dB above average) in the addressed section; performance depends on the stochastic Rayleigh spectrum of the selected fiber segment.
• Temperature sensing calibration: Absolute temperature measurements require calibration of wavelength versus temperature for each addressed section; the absolute mapping is not known a priori.
• No active temperature stabilization of the cavity: Although robust in practice, lack of stabilization can influence mode phases, potentially impacting long-term stability without affecting the narrow lasing line of the addressed mirror.