62.6 GHz ScAlN Solidly Mounted Acoustic Resonators

Technical Summary: 62.6 GHz ScAlN Solidly Mounted Acoustic Resonators

Problem Statement
As wireless systems advance toward 5G/6G and millimeter-wave (mmWave) bands (above 30 GHz), scaling acoustic filters to higher frequencies presents substantial challenges. While Surface Acoustic Wave (SAW) devices face fabrication complexity and loss mechanisms when Interdigitated Transducer (IDT) dimensions shrink, Bulk Acoustic Wave (BAW) devices typically require sub-100-nm films to achieve fundamental modes in the mmWave regime, leading to degraded material quality and elevated loss. Although exploiting higher-order thickness-extensional (TE) modes offers a route to mmWave performance without extreme thinning, existing Solidly Mounted Resonators (SMRs) generally operate below tens of gigahertz. Maintaining both high quality factor ($Q$) and electromechanical coupling ($k_t^2$) at higher frequencies is difficult due to shrinking acoustic wavelengths, lateral-field excitation limitations, and incomplete mode confinement. Furthermore, recent high-frequency demonstrations using all-metal Bragg reflectors rely on complex fabrication flows and suffer from significant feedthrough issues.

Methodology
This work demonstrates a 63 GHz SMR utilizing a Sc$_{0.3}$Al$_{0.7}$N piezoelectric layer and a buried platinum (Pt) bottom electrode, positioned above an all-dielectric acoustic Bragg reflector. The design strategy involves:

• Stack Design: A finite-element analysis (FEA) guided the stack to confine and efficiently excite the third-order TE bulk acoustic wave (BAW) mode using thickness-field excitation. The structure consists of a 67.6 nm Sc$_{0.3}$Al$_{0.7}$N active layer sandwiched between a 40 nm buried Pt bottom electrode and a patterned Pt top electrode.
• Bragg Reflector: To minimize capacitive feedthrough and improve structural robustness compared to suspended FBARs, an all-dielectric reflector composed of 8.5 pairs of alternating SiO$_2$ (28.2 nm) and Ta$_2$O$_5$ (24.3 nm) layers was employed. These materials were selected for their high acoustic impedance contrast (12.4 vs. 33.3 Mkgm$^{-2}$s$^{-1}$), providing a fractional stopband bandwidth of approximately 60%.
• Fabrication: The devices were fabricated on high-resistivity silicon (HR-Si) substrates. The process involved sputtering the Bragg reflector, depositing the buried Pt electrode, and growing the ScAlN layer. The active region was defined via UV photolithography and ion milling, followed by low-temperature PECVD SiO$_2$ backfill for isolation. A top Pt electrode was patterned via electron-beam evaporation, and contact pads were formed with 300 nm Al.
• Characterization: Devices were characterized using GSG probes connected to a 67 GHz vector network analyzer (VNA). Two-port S-parameters were converted to Y-parameters, and a modified Butterworth–Van Dyke (BVD) circuit model was applied to extract parameters, accounting for routing parasitics ($L_s$, $R_s$).

Key Contributions

• Record Frequency: The fabrication of an SMR operating at 62.6 GHz, which represents the highest operating frequency reported for an SMR to date.
• All-Dielectric Reflector: The implementation of a SiO$_2$/Ta$_2$O$_5$ Bragg reflector to reduce capacitive feedthrough, a critical factor for high-capacitance-density mmWave SMRs, avoiding the complexity and feedthrough issues associated with all-metal reflectors.
• Thickness-Field Excitation: The successful utilization of a buried bottom electrode to leverage the $e_{33}$ coefficient for efficient thickness-field excitation of the third-order TE mode, rather than relying on lateral-field excitation.
• Design Validation: Experimental verification of the third-order TE mode confinement and the demonstration of a practical pathway to compact mmWave acoustic components.

Results

• Resonance Frequency: The fabricated device exhibits a clear series resonance at 62.6 GHz and an anti-resonance at 63.8 GHz. A second-order TE mode was observed at 40.4 GHz, and a first-order TE mode at 11.7 GHz.
• Performance Metrics: The device achieved an extracted piezoelectric coupling coefficient ($k_t^2$) of 0.8% for the third-order TE mode. The phase-derived Bode quality factor ($Q$) peaked at 50.91 at 63.75 GHz, resulting in a high $f \cdot Q$ product of $3.13 \times 10^{12}$.
• Discrepancies: The measured resonance frequency (62.6 GHz) was higher than the simulated target (approx. 49–50 GHz). Post-fabrication FEA matching suggests this shift is due to a thickness mismatch between the top and bottom Pt electrodes (estimated top Pt thickness of 28 nm vs. designed 40 nm) and variations in thin-film material properties.
• Loss Mechanisms: The relatively low $Q$ compared to lower-frequency SMRs is attributed to residual leakage from reflector bandgap mis-centering, intrinsic losses in the Pt/ScAlN/Pt cavity, and electrical loading from routing series resistance ($R_s = 52 \, \Omega$).

Significance
The authors state that these results establish a pathway toward mmWave SMR devices for filters and resonators in next-generation RF front ends. By demonstrating that solidly mounted architectures can operate effectively above 60 GHz using thickness-field excitation and all-dielectric reflectors, the work validates a design methodology that balances structural robustness with high-frequency operation. While the current $k_t^2$ and $Q$ are modest compared to lower-frequency counterparts, the achievement of the highest SMR frequency to date proves the viability of this platform for future compact, low-loss mmWave acoustic components. The paper concludes that future work will focus on reducing high-frequency losses, optimizing stack/electrode configurations to improve $k_t^2$, and integrating these resonators into filter prototypes.