Access Official Technical PDF

Share Summary:LinkedInEmail

ScAlN SAW resonators offer a route to stronger electromechanical coupling in high-frequency acoustic devices, while retaining silicon as the substrate. In reported devices, single-crystal scandium aluminum nitride (ScAlN) and aluminum nitride (AlN) layers grown by molecular beam epitaxy (MBE) achieved effective coupling coefficients (kt²) of up to 7.8% and 5.0%, respectively. The highest reported quality-factor–coupling product, Q×kt², was 5.4 for a ScAlN resonator on silicon.

The results highlight two design considerations for engineers evaluating surface acoustic wave (SAW) technology: the piezoelectric layer’s crystal quality can influence coupling, and acoustic reflectors can improve energy confinement and quality factor. Both matter when balancing resonator performance for filtering and other high-frequency applications.

Why ScAlN and crystal quality matter

SAW resonators use a piezoelectric material to convert electrical signals into acoustic waves at a surface and back again. AlN is of interest for high-frequency devices because of its high surface phase velocity compared with traditional materials such as lithium niobate. It also has favorable temperature stability relative to lithium niobate, although its electromechanical coupling is a recognized limitation.

Adding scandium to AlN increases its piezoelectric response, making ScAlN a candidate for devices that need stronger electromechanical coupling. The performance of either material, however, depends in part on the quality of the piezoelectric layer. The study’s central approach was to use single-crystal layers rather than the polycrystalline films commonly produced by reactive magnetron sputtering.

Sputtering is attractive for its relative simplicity, cost, and growth rate, but its films can have lower crystal quality than epitaxial layers. This distinction becomes more important as devices are scaled for higher frequencies. The researchers therefore grew AlN and ScAlN on silicon by MBE, using nitrogen plasma and evaporated metal sources. The layers were 400 nm thick; the ScAlN contained 11.8% scandium.

ScAlN SAW resonators on silicon: device design and results

The fabricated devices used interdigitated transducers (IDTs), which convert between electrical and acoustic energy. The layouts varied by the number of IDT sets—one-port or two-port—and by whether reflector structures were included to help confine acoustic waves. Most devices had a 0.4 µm pitch, with additional reflector-equipped devices built at a 0.3 µm pitch.

For devices with a 0.4 µm pitch, the one-port ScAlN resonator without reflectors reached a measured kt² of 7.8% at a center frequency of 3.718 GHz. The corresponding AlN device reached 5.0% at 3.992 GHz. These results illustrate the stronger coupling measured in ScAlN, while also showing that the device configuration affects the final value: reflector-equipped devices had lower measured kt² than their counterparts without reflectors.

The strongest reported Q×kt² result came from a two-port ScAlN device with reflectors. At a center frequency of 3.648 GHz, it achieved kt² of 3.7% and an unloaded Q of 146, giving a figure of merit of 5.4. This product combines coupling and quality factor, providing a way to consider both properties together rather than assessing coupling alone.

How reflectors affect resonator performance

Reflectors are intended to reduce acoustic energy leakage by confining the wave near the active region. Finite element analysis showed greater displacement and less lateral and vertical acoustic leakage for a resonator with reflectors than for one without them. Measurements were consistent with the expected quality-factor benefit: devices with reflectors exhibited more than four times the Q of devices without reflectors.

That improvement came with a lower measured coupling coefficient in the tested layouts. For example, the two-port ScAlN device with reflectors had kt² of 3.7%, compared with 7.3% for the two-port device without reflectors. This contrast is a practical reminder that resonator design involves trade-offs. A configuration that improves acoustic confinement and Q may not maximize coupling, so comparisons should consider the intended figure of merit and application requirements.

Measurement and modeling considerations

The devices were characterized from 1 to 10 GHz using S- and Y-parameter measurements. Resonant frequencies and effective coupling were extracted from the measured response, while the unloaded quality factor was obtained by fitting the admittance data to a modified Butterworth–Van Dyke model.

Finite element analysis also helped identify a modeling issue: using the commonly cited low-frequency dielectric constant for AlN produced a simulated coupling value substantially below measurement. Accounting for the lower dielectric constant at high frequency brought the modeled resonant frequency into agreement with the measured value. The work also notes that simulated coupling and measured effective coupling are not identical quantities, and that layer stress and the extraction method can contribute to differences.

With a smaller 0.3 µm pitch, reflector-equipped devices reached a center frequency of 4.112 GHz and kt² of 7.4%; a higher-order mode was observed at 10.05 GHz. These results indicate operation beyond the principal 3.6–4.0 GHz range reported for the main device set, without implying that every configuration will deliver the same combination of frequency, coupling, and Q.

Practical implications for acoustic device development

For engineers and device developers, the findings make a case for evaluating the piezoelectric layer and resonator geometry together. MBE-grown single-crystal AlN and ScAlN on silicon demonstrated coupling and Q×kt² performance that improved on the cited prior work using sputtered ScAlN on silicon. Silicon also provides access to established micromachining processes, an integration consideration identified by the researchers.

The results do not remove the need to optimize each design for its target frequency and performance priorities. Instead, they show why crystal quality, scandium incorporation, reflector design, and high-frequency material properties should all be considered during that process. Single-crystal ScAlN SAW resonators on silicon provide a promising combination of coupling and acoustic confinement, with a reported Q×kt² of 5.4 offering a useful benchmark for further device development.

Access Official Technical PDF

About The Author