Abstract

Cavity solitons are self-sustaining ultrastable pulses looping in an externally driven cavity [1,2]. They exist in an equilibrium state between the nonlinear phase accumulation from the Kerr Effect counterbalancing the chromatic dispersion and cavity detuning, while the energy lost in the cavity is supplied through the driving laser. They offer an attractive way of generating optical frequency combs by simply extracting a fraction of the cavity power each roundtrip. The extracted signal, taking the form of a pulses train with a repetition rate equal to the cavity roundtrip, acts naturally as a frequency comb source. Several limitations still prevent their use for real-life applications. Both the comb bandwidth and the soliton peak power are limited by the intracavity losses and are typically lower than what can be obtained with mode-locked lasers. Minimum possible losses are desirable which contrasts with the desired output for which the highest possible tapping ratio is desired. This is also particularly critical for fiber resonators in which the loss budget is usually tight. This means that typically only a small fraction of the useful signal can be extracted. To overcome this limitation, active cavity solitons (ACS) were recently introduced [3]. They are hosted in an active cavity, for which the gain medium, pumped under the lasing threshold, compensates a major part of the cavity loss, including large input/output coupling ratios. This near-all loss compensation allows very low effective losses, even with the extraction of a large part of the cavity signal. This design however suffers from gain saturation as the gain decreases with increasing intracavity power, meaning that the effective low loss is only maintained at low average power. This intrinsically limits the number of solitons the cavity can sustain and prevents the spontaneous generation of solitons through a detuning sweep [4, 5]. The energetic efficiency is also low as only a small fraction of the pump power is absorbed in the gain medium.

© 2023 IEEE