A frequency comb is one of the most orderly objects in physics: a laser whose spectrum is a ladder of perfectly equidistant frequencies, precise enough to serve as the optical ruler that won John Hall and Theodor Hansch a share of the 2005 Nobel Prize in Physics. Most combs demand a near-perfect resonator or a carefully mode-locked laser. A team at the University of Ottawa has now built one from the opposite ingredient. Their device relies on Rayleigh scattering, the faint reflection of light off random density fluctuations inside glass, which engineers usually spend entire careers trying to suppress. In their design, that disorder does not fight the comb. It is the comb’s second mirror.
The work, published August 4 in Nature Communications by Da-Peng Zhou, Gerard Tatel, Yuan Wang, Paul S. Westbrook, Liang Chen and Xiaoyi Bao, is the first demonstration of a random Kerr optical frequency comb, which the authors describe in the paper as the spectral-synthesized analog of a random laser. Random lasers, proposed by V. S. Letokhov in 1966, replace the two mirrors of a conventional laser with multiple scattering in a disordered medium. The Ottawa device does something similar in the frequency domain: it takes the concept of feedback through disorder and applies it to comb generation.
A cavity with one mirror missing
Conventional Kerr combs work by trapping a continuous-wave laser inside a tiny ring resonator, where the Kerr nonlinearity converts one frequency into many through four-wave mixing. The resonator must be almost lossless, with a quality factor above one million, so that light circulates long enough for the cascade of new frequencies to build up. The Ottawa setup abandons that requirement almost entirely.
The cavity here is half-open. At one end sits a fiber Bragg grating, a narrowband reflector etched into the fiber. At the other end there is no mirror at all, only a 1 kilometer length of fiber whose Rayleigh scattering has been enhanced about 13 dB over standard single-mode fiber by ultraviolet exposure. A pulsed pump laser at 1550 nanometers sends 10 to 60 nanosecond pulses into the fiber at a 150 kHz rate, with a peak power of about 5 watts.
The mechanism is subtle. As a pulse propagates, modulation instability, the same nonlinear process that fragments intense light in fiber, spontaneously generates faint sidebands at new frequencies. A small fraction of that light scatters backward off the fiber’s random density fluctuations. The fiber Bragg grating filters the scattered light and reflects it back down the fiber, where it meets the next incoming pump pulse. The encounter imprints a very weak periodic modulation onto that pulse, roughly 45 dB below the pump power. That whisper of a signal is enough. Modulation instability amplifies and compresses the weak modulation as it travels, reshaping the sine wave into a train of picosecond pulses. A second fiber stage, a 5.25 kilometer dispersion-decreasing fiber, compresses the pulses further to about 390 femtoseconds, broadening the spectrum to roughly 100 nanometers.
The result is a comb with lines spaced about 1 nanometer apart, a spacing the team could tune over roughly 0.3 nanometers by adjusting the pump wavelength within the modulation instability gain band. The soliton repetition rate was tunable from 120 to 160 GHz, and the output could be gated on and off at the pump rate.
What earns a device the name comb
The name itself became a scientific argument. In peer review, one referee challenged the term comb on the grounds that a comb traditionally implies a phase-locked set of coherent lines, and argued the output should be described instead as a cascaded four-wave mixing source or a soliton pulse train generator. The objection was not pedantry: if the spectral lines were not locked in phase with one another, the device would be little more than a collection of discrete wavelengths, useless as a ruler.
The authors answered with a beat-note experiment. They filtered out individual comb lines, mixed each one with an external continuous-wave laser of 5 kHz linewidth, and recorded the beat signal over 200 microseconds, a window containing more than 20 output pulses. All 20 pulses produced the same beat tone, and the behavior was consistent across comb lines, demonstrating fixed phase relationships between the lines and pulse-to-pulse coherence. The measurement cannot resolve detailed frequency noise, the authors note, but it establishes the constant phase relation that the word comb requires. They kept the title.
Honest limits
The device has real constraints, and the authors state them plainly. The output is a gated train of pulses, not a continuous stream, because a pulsed pump is used, and the gate rate is limited by the length of the Rayleigh-enhanced fiber: only one pump pulse may occupy the fiber at a time, or the scattered feedback from different locations interferes destructively. Long-term stability is dominated by the environment. Both the grating and the Rayleigh spectrum shift by roughly 1 GHz per kelvin of temperature change, and over a one-hour measurement the comb lines drifted with a standard deviation of about 360 MHz while individual line power fluctuated by about 8.5 percent. The team insulated the fiber and grating in a styrofoam box to keep these effects manageable, and they estimate the resulting timing jitter of the soliton train at about 21 femtoseconds. Spectral bandwidth is currently capped near 100 nanometers by the power that can be delivered into the compressor stage, and conversion efficiency is low.
What the device loses in stability it gains in simplicity. It needs no feedback control loops, no external modulators and no additional pump sources, and it is built entirely from standard fiber components. The authors suggest the comb could serve applications that tolerate its gated output, including frequency synthesis, microwave photonics and high-repetition-rate pulse generation, and they point to distributed fiber-optic sensing, where streams of soliton pulses are used as probes to locate disturbances along a fiber, as a particularly natural fit. With better dispersion engineering, they argue, pulses below 100 femtoseconds and correspondingly wider spectra should be reachable.
The deeper point is conceptual. For fifty years, random lasers showed that order can emerge from disorder when enough amplification is present. The Ottawa comb shows the same principle operating on the spectrum of a precision instrument: a measurement tool whose defining feature is that it built its cavity out of the glass’s own imperfections.
Source: Zhou, D.-P., Tatel, G., Wang, Y., Westbrook, P. S., Chen, L. & Bao, X. “All-fiber random Kerr optical frequency combs.” Nature Communications (2026). DOI: 10.1038/s41467-026-76335-4.