Drift-free characterization of electro-optic tuning efficiency in lithium niobate photonic nanocavities

Technical Summary: Drift-free Characterization of Electro-optic Tuning Efficiency in Lithium Niobate Photonic Nanocavities

Problem Statement
Lithium niobate (LN) photonic crystal nanobeam cavities (PCNBC) are premier platforms for integrated electro-optics due to their deep sub-wavelength mode confinement and ultralow power consumption. However, accurately characterizing their electro-optic (EO) tuning efficiency ($\eta$) is fundamentally hindered by DC drift. This phenomenon, caused by charge redistribution, surface screening, or buffer layer relaxation under sustained electric fields, induces time-dependent spectral instabilities. In high-Q nanocavities, where resonance linewidths are extremely narrow (picometer scale), even minor drift causes severe tracking errors. Consequently, conventional quasi-static DC voltage scanning methods yield highly unreliable, non-reproducible, and inaccurate data for determining critical performance metrics like the $\pi$-phase shift voltage.

Methodology
The authors propose a drift-free, dynamic measurement methodology that exploits the vast separation of timescales between the instantaneous electronic Pockels effect (femtosecond response) and slow DC drift relaxation processes (milliseconds to seconds).

• Dynamic Drive: Instead of static voltage steps, the system employs high-frequency triangular-wave voltage sweeps (10 kHz – 1 MHz).
• Measurement Principle: A continuous-wave (CW) laser is fixed at the zero-bias resonance wavelength. As the triangular voltage sweeps, the cavity resonance shifts linearly in wavelength. The transmitted power traces a Lorentzian temporal envelope on a photodetector.
• Extraction Logic: The tuning efficiency $\eta$ is derived directly from the temporal width of the detected electrical signal ($\Delta t_{FWHM}$) and the known voltage sweep rate ($dV/dt$), using the relation $\eta = \Delta\lambda_{opt} / \Delta V_{FWHM}$.
• Key Advantage: Because the modulation frequency far exceeds the drift relaxation rate, the quasi-static drift term is effectively decoupled from the instantaneous Pockels response. This eliminates the need for absolute wavelength tracking and renders the measurement immune to drift-induced artifacts.

Device Design and Fabrication
To validate this method, the authors fabricated 35 distinct devices on a polymer-loaded Lithium Niobate on Insulator (LNOI) platform.

• Structure: The devices utilize a y-cut LN film (300 nm) with a patterned polymer photonic crystal layer (400 nm) to confine the optical mode.
• Electrodes: Butterfly-shaped coplanar electrodes were designed to maximize the overlap between the in-plane electric field and the optical mode.
• Parameter Space: The study systematically varied the electrode modulation length ($l$) from 10 to 200 $\mu$m and the lateral electrode-to-waveguide gap ($d$) from 3.1 to 3.9 $\mu$m.
• Simulation: Comprehensive 3D Finite-Element Method (FEM) simulations were performed to provide theoretical benchmarks for $\eta$ and the quality factor ($Q$) across the parameter space.

Key Results

1. Validation of Drift Immunity: A direct comparison on a single device ($d=3.1$ $\mu$m, $l=120$ $\mu$m) demonstrated the failure of conventional methods versus the success of the proposed technique.
   • Conventional DC Scanning: Extracted $\eta$ fluctuated irregularly with wait times, yielding a high coefficient of variation (CV) of 16.13%.
   • Triangular-Wave Sweep: Extracted $\eta$ remained stable across frequencies (10 kHz to 1 MHz), yielding a highly reproducible average of 4.39 pm/V with a CV of only 1.1%.
2. Scaling Behavior: The experimental data across 35 devices confirmed theoretical predictions:
   • $\eta$ increases monotonically as the electrode gap $d$ decreases.
   • $\eta$ increases with modulation length $l$ until saturating at approximately 120 $\mu$m, indicating the full coverage of the fundamental cavity mode volume.
3. Simulation Agreement: The experimentally extracted tuning efficiencies showed excellent quantitative agreement with 3D-FEM simulations. A minor systematic discrepancy (simulations ~0.2 higher than experiment) was attributed to idealized simulation assumptions (e.g., perfect sidewalls) versus real-world fabrication tolerances and interfacial losses.

Significance and Claims
The paper claims to establish a rigorous standard for the characterization and optimization of resonant cavity electro-optics. By providing a robust, drift-free technique, the authors enable:

• Reliable Quantification: Accurate extraction of intrinsic EO parameters without the confounding effects of DC drift.
• Design Optimization: A validated framework for optimizing electrode geometries in thin-film LN devices.
• Scalability: A generalizable methodology applicable to other high-Q cavity-based modulators where DC drift limits measurement accuracy.

The authors conclude that this approach accelerates the development of high-performance thin-film lithium niobate photonic integrated circuits by ensuring that performance metrics are derived from reproducible, drift-compensated data rather than unstable quasi-static measurements.