Fig. 1. (a) Scanning electron micrograph of the AlN microresonator after dry etching. (b) Photograph of a quarter of a wafer fabricated with standard photolithography. (c) Microscope image of one fabricated device. (d) Transmission spectrum of the microresonator used for octave-spanning soliton generation. (e) Zoomed-in region of the two close resonances near 1550.6 nm, with fits to determine the Q values. (f) Simulated integrated dispersion Dint. The circles correspond to the experimental results of around 50 resonances. (g) Schematic of the pump transmission at high power. The transmissions of TE00 and TE10 modes are plotted separately in (i), where three pump positions are marked. (ii) is the direct combination of the two resonances without achieving a soliton regime. (iii) indicates the accessible soliton under appropriate pump parameters. (h) Principle of the passive compensation of the circulating power in the microresonator by the TE10 resonance to achieve soliton shown in the state (iii) of (g). The laser is only coupled into TE00 mode at λp1 (i). At λp2 (ii), most of the power is coupled into the TE00 mode, while both modes will redshift due to the thermal effect. (iii) shows the transition into a soliton state at λp3. The sudden reduction of the pump power during the soliton generation enables the blueshift of two resonances. The pump will move to the red-detuned side of the cavity resonance, while the TE10 mode can compensate for the intracavity power change, thus stabilizing the soliton state. In this state, the pump resonance splits into C-resonance (cavity resonance) and S-resonance (soliton resonance) [38].

Fig. 2. Experimental setup used for single DKS generation and characterization. FPC, fiber polarization controller; EDFA, erbium-doped fiber amplifier; OSA, optical spectrum analyzer; OSA1, 600–1700 nm; OSA2, 1200–2400 nm; AC, autocorrelator; PD, photodiode; OSC, oscilloscope; ESA, electrical spectrum analyzer; FBG, fiber Bragg grating.

Fig. 3. (a) Resonance transmissions at an on-chip power of 350 mW and a pump tuning speed of 1 nm/s. (b) Frequency comb evolution map. (c) Optical spectral snapshots of different comb states, corresponding to the dash lines in (b). The dashed lines show the fitted sech2 envelope. Insets are the microscope images of the green light emission from the microresonator. (d) Low-frequency intensity noise of the microcombs and the PD noise floor. (e) SHG-based autocorrelation measurement for the soliton 1 state. (f) A single pulse with sech2 fitting, where the trace FWHM needs to be multiplied by 0.648 to yield the real pulse width.

Fig. 4. Power difference between all and pump transmissions measured at on-chip powers with a laser tuning speed of 1 nm/s. The green polygon indicates the soliton existence regime.

Fig. 5. (a) Schematic of the laser wavelength tuning with step mode. (b) Soliton accessing possibility among 50 attempts, versus the λstart and λstop. (c) Record pump transmission traces when λstart/λstop are (i) 1550.620/1550.720 nm, (ii) 1550.630/1550.730 nm, and (iii) 1550.640/1550.740 nm, respectively. Green rectangles and red circles indicate the successful and failed attempts, respectively.