Phonon Trapping Lateral Field Excited Suspended Bulk Acoustic Wave Resonators (XBARs)

TL;DR

FBAR-based devices traditionally rely on quasi-plane acoustic waves with transverse extent $λ_a$, which limits strong 3D confinement needed for efficient microwave–optical transduction. The authors propose and demonstrate phonon-trapping lateral-field excited suspended XBARs by sculpting the ScAlN piezoelectric layer into a spherical lens, achieving 3D confinement of shear-horizontal bulk overtone modes and a ~4× increase in the $fQ$ product relative to unlensed devices. By combining 2D/3D FEM with RF measurements, they show reduced lateral leakage and a trapped high-$Q$ mode, while also highlighting remaining challenges in achieving higher $k^2_{eff}$ and lower dissipation. The findings offer a pathway toward more efficient MW-OTs and improved XBAR-based RF devices, with suggested optimizations in material composition, electrode geometry, and potential integration with microwave cavities for enhanced coupling.

Abstract

Film bulk acoustic wave resonators (FBARs) underpin modern wireless communication by enabling compact, high-performance RF filters in modern smartphones. Traditionally, these FBAR devices work with quasi-plane waves of sound where the transverse extent of the acoustic field $\gg$ the acoustic wavelength ($λ_a$). On the other hand, strong modal confinement is needed for achieving the interaction strengths necessary for building efficient microwave to optical quantum photon transducers (MW-OT) around an FBAR opto-mechanical cavity platform. Here, we fabricate a small mode-volume phonon trapping lateral field excited FBAR resonator (XBAR) by shaping the piezoelectric layer into a spherical lens, show an improvement in modal confinement and quality factor ($\approx$ 4$\times$), and discuss the improvements needed for building efficient MW-OTs around this XBAR geometry.

Figures (5)

• Figure 1: (a) Microscope image of a representative XBAR device (flat-flat cavity without mode trapping) fabricated on a ScAlN-on SOI substrate. The layer thicknesses are noted in the inset. (b) RF reflection spectrum ($S_{11}$) of the device showing a series of successive overtone resonances of the cavity. A zoomed-in spectrum of one of the modes is shown in the inset showing an asymmetric (Fano) lineshape. (c) 2D FEM simulation of one of the XBAR resonances under lateral field excitation. The $\vec{x}$-component of the displacement is plotted clearly indicating the resonant mode has a shear horizontal (SH) character.
• Figure 2: Phonon trapping microcavity: By shaping the ScAlN layer into a spherical (lens-like) surface, one can prevent lateral leakage in the XBAR mode. (a) shows the $k$-space plot of the modal displacement of the XBAR cavity mode shown in Fig.\ref{['Unlensed_ScAlN_XBAR']}(c). Modes with non-zero $k_x$ which correspond to leakage through the sides appear as circular arcs as indicated. The same $k$-space plot for the lensed cavity mode (c) is plotted in (b) in which the modal leakage is minimized as shown by the absence of the non-zero $k_x$ modes. (d) 3D modal confinement is evident from a full 3D FEM simulation of the lensed cavity showing the field ($x-z$ cross-section) is mainly confined under the lens at resonance. The inset shows the 2D cut along the orthogonal ($y-z$) direction indicated by the dashed line.
• Figure 3: (a) Microscope image of a representative lensed XBAR membrane device showing the spherical lens embedded in the ScAlN piezoelectric film. The cross-section of the device is shown in the inset with a Gaussian mode trapped in the cavity shown by illustration. The topography of the lens as measured by a profilometer (Dektak) is also shown via overlay. (b) RF reflection ($S_{11}$) spectrum for a representative mode in a lensed device (red) with the same mode in a reference (unlensed) device (blue) shown for comparison ($f_1$). A higher frequency comparison ($\approx$4.3) is shown in the inset ($f_2$). The $S_{11}$ spectra are normalized to remove the background tilt, cf. Fig.\ref{['Unlensed_ScAlN_XBAR']}(b). (c) Zoom-in $S_{11}$ spectrum of the lensed device clearly showing a doublet with two modes with different $Q$. The lower $Q$ mode corresponds to the untrapped mode existing in the base ScAlN layer and the higher $Q$ mode is trapped in the lens region, as shown schematically in the inset. (d) Histogram of the measured $fQ$ product of the modes from lensed vs reference devices showing that lensed devices give an overall 4$\times$ improvement in $Q$.
• Figure 4: Microscope image of a 1D lensed XBAR device mode-trapping cavity fabricated on ScAlN using a reflow etching process, along with a Dektak scan showing a lens thickness of 250 nm and a radius of 32 µ m, placed between electrodes with a gap of 8. (b) RF reflection ($S_{11}$) comparison between bare, 1D-lens and 2D-lens devices. The shift in frequency between the bare and the 1D lens devices agrees well with the 2D FEM simulation (inset) showing the geometry works as intended. The 2D lens data has additional structure due to the presence of the background mode (cf. Fig.\ref{['Lensed_ScAlN_XBAR']}(c)).
• Figure 5: Schematic illustration of the lensed XBAR fabrication process. The process includes backside patterning and substrate etching using a SiO$_2$ hard mask, membrane release via deep reactive ion etching (ICP Bosch), and subsequent lens fabrication on a 375 nm-thick ScAlN layer using photoresist reflow and plasma etching. 1D and 2D lenses are formed with final thicknesses of $\sim$250 nm and radii of $\sim$32 µ m. This is followed by standard lithography steps for electrode fabrication, including gold evaporation and lift-off. S1805 and S1813: photoresists, LOR: lift-off resist, BOE: buffered oxide etch.