High-Q and Compact Fabry-Perot Microresonators on Thin-Film Lithium Niobate

TL;DR

This work targets high-performance, compact photonic resonators on thin-film lithium niobate to overcome limitations of WGM structures caused by material anisotropy. By implementing tapered photonic crystal mirrors around a straight, suspended LNOI waveguide, the authors realize a 100 μm Fabry-Perot cavity with a measured loaded Q of $5.7\times10^{5}$ at $1531.8$ nm and a 4.8 nm free spectral range, aided by an optimized taper that yields near-unity reflectivity within the mirror band. Integration of on-chip electrodes enables piezo-optomechanical modulation, revealing thickness modes up to about $18.7$ GHz and confirming strong RF-to-optical coupling. The results demonstrate a viable, high-finesse FP platform on LNOI that broadens design options for electro-optic and optomechanical devices, with potential improvements via Bloch-mode engineering and triply-resonant cavity configurations, ultimately enabling higher cooperativity $C$ through enhanced mode overlap $\Gamma$ and finesse $\mathcal{F}$, $C \propto \Gamma^{2} \mathcal{F}$.

Abstract

Thin-film lithium niobate (TFLN) has played a pivotal role in the advancement of integrated photonics, by supporting a diverse range of applications including nonlinear optics, electro-optics, and piezo-optomechanics. The effective realization and enhancement of these interactions rely heavily on the implementation of high quality photonic microresonators. The pursuit of novel resonator architectures with optimized properties thus represents a central research area in TFLN photonics. In this work, we design and fabricate TFLN Fabry-Perot microresonators, by placing a straight section of waveguide between a pair of tapered photonic crystal mirrors. The resonator features a high quality factor of 600k at 1530 nm and a compact length of 100 um. The functionality of the device is further demonstrated by integrating on-chip electrodes for high-frequency piezo-optomechanical modulation. Our device can serve as an appealing candidate for developing high-performance photonic components on the TFLN platform.

Figures (4)

• Figure 1: (a) Schematic drawing of the photonic crystal unit cell, together with the direction of lithium niobate crystal axis. It consists of a suspended ridge waveguide with air holes etched through at the center. The lattice constant of the unit cell is $a=498$ nm and the waveguide width is $w$=1.2 µm. All other dimensions can be found in the text. The optical mode studied here is the fundamental TE mode with electric field components mainly along the x-axis. (b) Simulated reflection spectrum of the photonic crystal, formed by cascading 30 unit cells. The reflectivity is clearly suboptimal with poor performance at shorter wavelength. (c) The reflectivity is significantly improved by linearly tapering down the lattice constant and hole size at the cavity side. The number of taper cells $n$ is varied to show the improvement. No apparent change in reflection is seen when $n$>8, we thus choose $n$=10 for the FP resonator device.
• Figure 2: (a) SEM image of the device. The device consists of a straight waveguide placed in between two photonic crystal mirrors. The waveguide is suspended and supported by two pads at each ends. A grating coupler is placed at one side after a short section of bus waveguide. The number of unit cells is reduced for the mirror on the coupling side, allowing access to the resonance via reflection measurement. The red, blue, and orange boxes correspond to the zoom-in view of components in (b)-(d). (b) Zoom-in view of the photonic crystal mirror. A taper structure is implemented on the cavity side. (c) Zoom-in view of the waveguide. (c) Zoom-in view of the grating coupler. It has an apodized structure and exhibits single-pass coupling efficiency of 35 % at center wavelength 1530 nm.
• Figure 3: (a) Measurement results for the FP cavity without the implementation of tapered mirrors. The bottom plot is the reflection spectrum; the top plot shows the fitted linewidth together with the lower limit calculated from mirror reflectivity simulation. The resonances exhibit low Q factor and are only seen at longer wavelength, which aligns with the prediction from simulation. (b) Measurement results for the device with tapered mirrors. The resonator Q is significantly improved compared to the case in (a). The trend of resonance linewidth agrees qualitatively with the simulated lower limit. The dashed orange box indicates the highest-Q resonance plotted in (c). (c) The resonance at 1531.8 nm, with a loaded Q of 5.7$\times$10$^5$ and 8.3 dB extinction ratio. The orange line is a Lorentzian fitting.
• Figure 4: (a) Experimental setup for measuring piezo-optomechanical modulation. The vector network analyzer (VNA) applies RF output to the on-chip gold (Au) electrodes via an RF probe. It then excites the mechanical modes supported by the suspended waveguide via piezoelectric effect. By tuning the laser into optical resonance, the light will experience optomechanical modulation, which is measured via the beating signal after the high-speed photodetector. (b) The piezo-optomechanical modulation spectrum measured up to 30 GHz. The prominent peaks are the first-order and third-order thickness modes at 7.3 GHz and 18.7 GHz, respectively, which are most strongly coupled to the RF excitation. Their normalized displacement in the waveguide cross section is shown and are mostly along the crystal x-axis.