Fig. 1. TFLN microresonator with smooth sidewall. (a) Schematic illustration depicting the structure of a racetrack resonator. (b) Illustration of the transmission spectra demonstrating the characteristic features of a resonator. (c) Optical microscope image showing a racetrack resonator with 3 μm width and 500 μm length straight section. (d) SEM image (false-colored) offering an overview of the coupling region of a racetrack with 0.5 μm coupling gap and 3 μm width. (e) SEM image (false-colored) providing a detailed view of the coupling region of the same racetrack resonator. (f) AFM image capturing coupling region’s topography. (g) Lumerical eigenmode simulation representing the fundamental TE mode at the cross-section of 3 μm width ring racetrack resonator.

Fig. 2. Monolithic high-Q microresonators on TFLN. (a) Selected resonator spectrum spanning from wavelengths 1573.78 nm to 1574.06 nm. The corresponding racetrack features a width of 4.5 μm, length of 10 mm, coupling gap of 0.6 μm, and bending radius of 200 μm. Background modulation is attributed to the cavity formed by reflections between the two facets of the chip. (b) The highest-Q resonance features an intrinsic Q factor of 29 million at the wavelength of 1574 nm. (c)–(g) Resonances at wavelengths 1573.80 nm, 1573.85 nm, 1573.90 nm, 1573.95 nm, and 1574.04 nm, all belonging to the same high-Q mode family.

Fig. 3. Statistical analysis of intrinsic Q factor versus racetrack length and width. The bars depict the mean value of the top 50 intrinsic Q factors of a device, while horizontal marker lines are error bars, indicating the mean values plus and minus their standard deviations. (a) Average intrinsic Q of racetracks with a width of 3 μm and lengths of 1 mm, 2.5 mm, 5 mm, 7.5 mm, and 10 mm. (b) Average intrinsic Q of racetracks with a length of 10 mm and widths of 3.0 μm, 3.5 μm, 4.0 μm, 4.5 μm, and 5.0 μm.

Fig. 4. Resonance calibration with RF-modulated laser light. (a) Schematic of measurement setup incorporating a phase modulator capable of generating optical sidebands. (b) The original resonance without applying RF power exhibits a loaded Q of 16.07 million. (c) Same resonance with sidebands generated by activating the 100 MHz RF sources. (d) Calibration using the sideband positions to redefine the x-axis as frequency and refit the resonance. The loaded Q increases to 18.39 million, slightly higher than the original loaded Q.