A 3D-printed piezoelectric metamaterial uses topology-driven charge transport to locate underwater sound sources from a single compact transducer, replacing bulky hydrophone arrays.
(Nanowerk Spotlight) Piezoelectricity, the ability of certain materials to generate voltage under mechanical stress, has been understood since the Curie brothers described it in 1880. The effect arises from the lack of inversion symmetry in a crystal lattice: when stress deforms the structure, charge displaces along axes dictated by the atomic arrangement. This mechanism underpins a vast range of technologies, from medical ultrasound probes to sonar systems to vibration energy harvesters.
Engineers have developed many ways to enhance piezoelectric performance, including domain engineering, compositional tuning, and composite structures, all of which can increase the magnitude of specific coupling coefficients. But the symmetry structure of the piezoelectric tensor, which defines which directional couplings exist and how they relate to one another, is fixed by the material’s crystal class. Coefficient values can be optimized, but the architecture of the response cannot be rewritten.
This constraint has shaped acoustic transducer design for over a century. Each piezoelectric element produces one fixed directional sensitivity pattern determined by its tensor symmetry. A cylindrical hydrophone, for instance, is locked into an omnidirectional response perpendicular to its axis. To determine where a sound originates, engineers must arrange many transducers into spatially distributed arrays and compute phase or arrival-time differences between them.
The arrays this demands can be vast. At 20 kHz in water, reaching 1° angular resolution requires an aperture of roughly 4 m and nearly 100 hydrophones. Towed sonar systems exceed 100 m in length, with sensors spaced 5 to 10 m apart, requiring precise synchronization, complex wiring, and computationally intensive real-time processing.
A study published in Advanced Materials (“Intelligent Acousto‐Electrical Metamaterials (IAM) for Sound Source Detection”) reports a form of acoustic-electric coupling that escapes this constraint. The researchers created what they call intelligent acousto-electrical metamaterials: three-dimensional micro-architected piezoelectric lattices whose topology, rather than crystal symmetry alone, governs how charge flows in response to sound. Acoustic waves passing through these lattices excite multiple vibration modes simultaneously, and the resulting electrical output changes with the frequency, direction, and structural geometry of the material.
Acoustic-vibration-electric coupling in architected metamaterials. a-i) Conventional hydrophone towed array used for DoA, which provides multiple signals and uses phase computation for localization. This design is subjected to crosstalk noise between individual hydrophones and requires large spacing to perform optimally. ii) Proposed metamaterial concept, where one material is capable of displaying diverse voltage response for multiple source detection. b) Comparison between the electromechanical response and charge flow of a piezoelectric single-crystal and a metamaterial unit-cell. c) Visualization of the impact on the wavelength on the response of an auxetic and honeycomb lattice. In hydrostatic condition, the charge flow follows the ligaments. At higher frequencies, the charge flow varies with the pressure sign along the strut, which depends on the wavelength and direction of arrival of the wave. d) Visualization of the impact on the wavelength on the response of two distinct piezoelectric micro-lattice. For ϕ1 the stress varies mostly along the x3 axis when it oscillates along the x1 axis for ϕ2. (Image: Reproduced from DOI:10.1002/adma.202513205, CC BY) (click on image to enlarge)
At low frequencies, where the wavelength far exceeds the unit cell size, all structural elements called ligaments experience nearly uniform pressure and generate coherent charge. At higher frequencies, different parts of the structure encounter different phases of the pressure wave simultaneously, producing partial charge cancellation in some regions and redistribution elsewhere. Adjusting ligament angle and thickness allows the researchers to amplify, suppress, or reverse the voltage response for specific directions and frequencies.
Charge pathways can be selectively reconfigured by changing the lattice topology. This means the piezoelectric behavior is encoded in geometry rather than inherited exclusively from the crystal. By assembling different lattice types side by side within a single cylinder, the researchers created transducers with complex directional sensitivity profiles that no single crystal element could produce.
The team developed an inverse design framework that starts from a desired beam pattern. The method segments the target profile into angular sections, assigns an appropriate lattice architecture to each using a precomputed sensitivity map, and assembles them into one cylindrical device. Simulations confirmed close agreement with target profiles, and the designs remained stable even when piezoelectric coefficients varied by up to 20%.
A validation test demonstrated the concept without machine learning. The researchers placed a 10 mm diameter architected hydrophone in a water tank with two 300 kHz sources separated by 120°. During a full rotation, the device recorded two distinct voltage peaks matching the source directions. A commercial omnidirectional hydrophone tested alongside it produced only a flat response, unable to distinguish the two sources.
Microscale additive manufacturing made the structures possible. The team used a high-resolution light-based 3D printing system to fabricate PZT ceramic lattices, which they then sintered into dense piezoelectric structures. They applied silver epoxy electrodes to each architectural section and then polarized the material to activate its piezoelectric response.
The cylinder is not uniform. It is divided into distinct architectural sections, each built from a different lattice topology with its own strut angles and densities. Each section carries its own electrode pair, so the device produces a set of distinct magnitude and phase signals rather than a single output. This internal diversity is what gives the system enough information for machine learning to work with.
To convert these signals into direction estimates, the researchers paired the metamaterial with a multilayer perceptron neural network. The model compares magnitude and phase readings from all sections against a training dataset spanning multiple angles and frequencies. It represents angles as sine-cosine pairs to avoid discontinuities at the 0°/360° boundary.
The finished transducer measured 3 cm in diameter and operated across 20 to 100 kHz, achieving an R² score of 0.95 when tested against experimentally acquired training data. Underwater tests showed the device tracking source movements with localization error within ±10°. It distinguished two simultaneous sources emitting at different frequencies through Fourier-based signal separation, and a real-time test with a toy fish moving unpredictably confirmed that position estimates updated within hundreds of milliseconds.
Acoustic crosstalk between transducer sections produced a counterintuitive benefit. Conventional arrays treat mechanical coupling between adjacent elements as unwanted noise and suppress it through physical separation. In the metamaterial transducer, connected sections with strong coupling generated richer signal patterns that the neural network exploited for improved accuracy. Coupled configurations consistently outperformed physically separated ones.
The device maintained high prediction accuracy even in noisy conditions, with reliable performance for signal-to-noise ratios down to 20 dB. The 3 cm transducer achieves directional sound sensing that conventionally requires apertures of 4 m or more for comparable angular resolution, a reduction of more than two orders of magnitude in size. Packaging imperfections currently limit its signal-to-noise ratio below that of bulk piezoelectric transducers, but the architecture compensates through geometry-driven tunability and far fewer active channels.
Because the approach programs directional response into material structure rather than relying on array geometry and phase computation, it could extend beyond underwater acoustics to in-air sensing for autonomous vehicles or robotic swarms.
For authors and communications departmentsclick to open
Lay summary
Prefilled posts
Nanowerk Newsletter
Get our Nanotechnology Spotlight updates to your inbox!
Thank you!
You have successfully joined our subscriber list.
Become a Spotlight guest author! Join our large and growing group of guest contributors. Have you just published a scientific paper or have other exciting developments to share with the nanotechnology community? Here is how to publish on nanowerk.com.
