A bulk acoustic resonator with vertical electrodes for wideband filters

Radiofrequency (RF) front ends for current and next generation (5G and 6G) wireless communication demand acoustic filters that combine wide bandwidth, high power capability, and thermal stability. Existing surface and bulk acoustic wave (SAW and BAW) technologies face inherent trade-offs between electromechanical coupling, lithographic tunability, and robustness. Here we introduce the bulk acoustic resonator with vertical electrodes (VBAR), a device that combines the advantages of suspended and solidly mounted resonators. VBARs use lithium niobate (LiNbO3) ridges with sidewall electrodes to excite a shear-horizontal bulk acoustic resonance, providing frequency control through lithography in a configuration that is mechanically anchored to the substrate. Fabricated VBARs exhibit electromechanical coupling coefficients exceeding 30% in the 2-4 GHz range, enabling ladder filters with fractional bandwidths of nearly 20%. While further optimization is necessary to minimize losses, the VBAR concept offers an alternative route toward wideband and robust RF filters for next-generation wireless systems.

💡 Research Summary

The paper introduces a novel bulk acoustic resonator architecture called the Vertical‑Electrode Bulk Acoustic Resonator (VBAR) that aims to meet the stringent requirements of next‑generation wireless front‑ends, namely wide fractional bandwidth (FBW), high power handling, and temperature stability. Conventional surface acoustic wave (SAW) and bulk acoustic wave (BAW) technologies each excel in certain aspects—SAW offers lithographic frequency tuning, while BAW provides high electromechanical coupling (kₜ²) and robust mounting—but they cannot simultaneously deliver all three performance metrics. VBAR bridges this gap by combining a lithographically defined resonant dimension with a mechanically anchored, solidly mounted structure.

The device consists of lithium niobate (LiNbO₃) ridges fabricated on a silicon substrate. Narrow silicon pedestals act as anchors, isolating the acoustic cavity from the bulk substrate while providing a thermal path for heat dissipation. Aluminum sidewall electrodes are deposited on the ridge faces, creating an in‑plane electric field that excites a shear‑horizontal bulk acoustic wave (SH‑BAW) mode. Because the acoustic cavity is bounded laterally by the ridge sidewalls, the resonant frequency is primarily set by the ridge width (w_ridge), which can be precisely controlled by standard photolithography. This contrasts with traditional BAW resonators where the film thickness determines frequency.

Finite‑element simulations demonstrate that varying w_ridge from 350 nm to 600 nm tunes the resonance from 2.5 GHz to 4 GHz. The electromechanical coupling coefficient kₜ² reaches 7 %–40 % for a Y‑cut 36° LiNbO₃ ridge without any compensating layer. Adding a SiO₂ overlay reduces kₜ² proportionally to its thickness but dramatically improves the temperature coefficient of frequency (TCF) from roughly –70 ppm/K to –30 ppm/K, owing to the positive thermal expansion of SiO₂ compensating the negative coefficients of LiNbO₃, Al, and Si. The simulations also reveal that the acoustic mode is well confined within the ridge, and that a narrow silicon anchor suppresses leakage into the substrate without requiring high‑velocity substrates such as SiC or diamond.

Fabrication proceeds through a series of lithography and etching steps. First, the ridge pattern is defined by deep reactive‑ion etching, yielding near‑vertical sidewalls (85°–90°). Aluminum is then conformally deposited, and a self‑aligned lift‑off removes planar metal, leaving only sidewall electrodes that connect to external pads. Finally, a controlled undercut of the silicon anchor is performed using XeF₂ or SF₆ plasma, achieving sub‑micron precision so that the anchor width remains less than half of w_ridge, thereby minimizing acoustic loss while preserving mechanical support.

Electrical characterization of fabricated VBARs shows clear resonance–anti‑resonance pairs with a spacing of about 500 MHz, confirming the intended SH‑BAW mode. Measured Q‑factors (Bode Q) range from 100 to 150, decreasing for narrower ridges and higher frequencies, indicating that intrinsic ridge losses dominate over substrate leakage. The presence of SiO₂ layers introduces a spurious mode near the anti‑resonance, highlighting the need for careful stack design to avoid mode coupling. Despite the modest Q, the devices exhibit electromechanical coupling exceeding 30 % across the 2–4 GHz band, enabling ladder‑type filters with fractional bandwidths approaching 20 %, a substantial improvement over conventional BAW filters (typically ≤15 %).

Thermal and power tests reveal that VBARs can sustain continuous RF power up to 30 dBm with only minor non‑linear distortion, thanks to the solid anchoring that provides an efficient heat‑spreading path. The SiO₂ compensation layer effectively reduces temperature sensitivity, making the resonators suitable for environments with significant temperature swings.

In summary, VBAR offers a compelling alternative to existing SAW and BAW technologies by delivering high kₜ², lithographic frequency control, and robust mechanical anchoring in a single platform. While current Q‑values are lower than those of state‑of‑the‑art SAW resonators, the demonstrated bandwidth, power handling, and temperature stability address key challenges for 5G/6G front‑ends. Future work should focus on Q‑enhancement through loss‑reduction strategies, suppression of spurious modes, and scaling to higher frequencies (>4 GHz) via finer lithography and optimized electrode designs. If these issues are resolved, VBAR could become a cornerstone technology for next‑generation wideband RF filters.