The fifth generation of mobile communication systems (5G) has defined more than 100 frequency bands in which wireless signals are transmitted. The coexistence of all these frequency bands requires the use of many bandpass filters (near 100 in the latest mobile phones). These filters select only the relevant signals from the ambient spectrum, hence limiting the computational power – and thus the power consumption – needed to extract the information from radio signals. They also ensure that the emitted signals do not interfere with other communications or applications.

Discussions are in progress to prepare the sixth generation of mobile communication systems (6G). In particular, the World Radio Conference has proposed in September 2024 the use of the FR3 frequency range (7.125 – 24.25 GHz) to extend wireless network capabilities, in order to support the growth in data consumption driven by applications such as the Internet of Things or Augmented Reality. This frequency range offers a sweet spot between the sub-7 GHz FR1 range, which starts to become congested and cannot offer wide bandwidths, and the millimetre-wave FR2 range (above 24 GHz) in which signal attenuation and reflections by buildings restrict applications to relatively short range, line of sight, though high data rate applications.  

Filtering technologies used in the FR1 range are based on acoustic resonators, in which an extreme miniaturisation is obtained by replacing electromagnetic waves by acoustic waves propagating in solids. Therefore, filters based on resonators in the range of tens to hundreds of microns can be easily mass produced with techniques inherited from the semiconductors and MEMS industries. This is made possible by the use of piezoelectric materials which are capable of transducing electrical signals into mechanical vibrations, and conversely, in a fully passive way. Extending these technologies to higher frequencies requires even further miniaturization, as critical dimensions scale inversely with frequencies. This causes considerable challenges regarding maintaining suitable material properties and proper dimensional control to meet reasonable specifications in terms of centre frequency and bandwidth.

To explore these challenges, teams from VTT and CEA-Leti designed two Bulk Acoustic Wave Filters (BAW), based on two different approaches.

VTT designed its filter considering AlScN as the piezoelectric material. Reaching such high frequencies requires very thin piezoelectric layers (~110 nm), in order to keep reasonable electrode thicknesses (>150 nm) for low ohmic losses. To ensure that the bandwidth requirements can be met despite the piezoelectric layer remaining less than a third of the resonator stack, a high Sc content (Al_0.7Sc_0.3N) was considered. Also, to offer an interesting compromise between acoustic confinement in the piezoelectric layer and conductivity, bi-metal multilayer W/Al electrodes were considered.

On the other hand, CEA-Leti considers a single crystal LiNbO₃ as the piezoelectric material. To keep electrodes size reasonable (50 and 100/150 nm) and the piezoelectric layer thicknesses (200 nm), the resonators exploit the third resonance. The high piezoelectric properties of LiNbO₃ ensures that this mode of operation remains relevant to synthesize the intended band, even when considering light, but highly conductive electrodes such as Al. Exploiting an overtone resonance needs to take care of lower order resonances. Their respective influence was decreased by designing a specific acoustic Bragg mirror which preserves only the mode of interest.

These designs are a first step in the FAMES project. They define the goal to reach, now driving developments towards implementing in practice these filters with proper reproducibility and dimensional control to ensure that these devices meet their specifications.

Once fabricated, these filters will be assembled into a demonstrator, along with an RFIC module, to form an FR3 transceiver. These designs may be also adapted to cover other frequency bands than the ones initially targeted.

FAMES provided the funding to dedicate time and effort to perform this work.