Ultra-broadband, Low-loss Wavelength Combiners and Filters: Novel Designs and Experiments in Thin-film Lithium Niobate

Abstract

Thin-film lithium niobate (TFLN) has emerged as a leading platform for large-scale programmable photonic circuits for quantum and classical applications. As circuits scale in complexity, low-loss routing of broadband pump and signal fields becomes essential. Here, we present closed-form analytical models and experimentally demonstrate compact, fast-quasi-adiabatic driving-optimized wavelength combiners and filters operating at the fundamental harmonic (FH, 1550 nm) and second-harmonic (SH, 775 nm) wavelengths. Our designs achieve ultra-low loss below 0.06 dB across a 90 nm bandwidth at FH, while maintaining extinction ratios exceeding 25 dB. At SH, the loss remains below 0.12 dB over a 45 nm bandwidth with extinction ratios greater than 19 dB. Devices fabricated on a 300-nm TFLN platform exhibit added loss below 0.1 dB across 1550 - 1600 nm, with minimum values of 0.04 dB around 1580 nm and 0.021 dB at 775 nm. Combined with recent advances in on-chip quantum state generation, low-loss interferometers, and detection, these results enable high-fidelity quantum photonic circuits on the TFLN platform.

Figures (6)

• Figure 1: FAQUAD-optimized adiabatic directional coupler: design methodology and simulated performance. (a) Top-view geometry of the coupler showing the engineered longitudinal variation. Region I: Constant-gap section enabling controlled adiabatic supermode evolution. Region II: Cubic bend section designed to preserve adiabaticity during waveguide separation. Region III: Euler bends for low-loss routing and curvature-optimized transitions. (b) Spatial evolution of the waveguide top-width difference and inter-waveguide top gap. (c) Prescribed FAQUAD coupling angle $\chi(z)$ ensuring controlled adiabatic evolution between supermodes. (d) Mode profiles of the symmetric and antisymmetric supermodes at the device's midpoint ($g_m = 800$ nm) and at the end of FAQUAD evolution ($g_c = 1200$ nm) where coupling is negligible. (f) Simulated nominal performance at FH and SH mode propagation, showing the high coupling for FH and high extinction for SH, as desired.
• Figure 2: Simulated performance and fabrication-error tolerance. (a) Extinction Ratio at FH for our design [top] and others of the same length. (b) Extinction Ratio at SH. (c) Solid: Total loss at FH and SH. Dashed: Power lost from both outputs. (d) Total loss at FH over fabrication parameter sweep.
• Figure 3: Experimental characterization. (a) Top: Measurement configuration used for FH characterization. Bottom: Representative resonance dips measured around the FH wavelength (1530–1600 nm) for the control resonator (black) and the resonator containing the DUT (red). The reduced resonance depth and increased linewidth in the DUT resonator indicate additional loss. (b) Top: Optical microscope image of the fabricated devices. Bottom: Extracted excess loss relative to the control resonator as a function of wavelength. Red points correspond to measured values, while the dashed curve shows the smoothed fit with the shaded region indicating the 95% confidence interval.
• Figure 4: Characterization at the SH wavelength. (a) Top: Optical micrograph of the DUT where the FAQUAD-optimized coupler is used. Bottom: Micrograph image of the circuit in resonance when coupled from the DUT port. (b) Representative transmission resonance measured near $\lambda = 775$ nm. The red trace shows the measured transmission spectrum, while the gray curve corresponds to the fit obtained using the resonator transmission model.
• Figure 5: (a) Deviation of the implemented coupling angle $\chi$ from the target model (in degrees). (b) Relative error of the coupling magnitude, $\Gamma$. (c) Simulated adiabaticity relative to the design value in the FAQUAD sections of the coupler.