This text will be replaced

Right-click to download chirp sound from the nanospeakers. (0.33 MB MPEG4) The classical music tracks also rendered by the speakers have not been reproduced here for copyright reasons.

“We now have a lab of macroscale phenomena for nanostructures,” says Brown, a researcher at the University of Colorado in the US. He adds, “This work on thermoacoustic speakers is a step from passive structures towards actively controlled nanodevices.”

The devices they create consist of tungsten films around 25nm thick and 17μm x 2μm in area that produce readily audible sounds and even tunes. Operated in parallel in large arrays, the devices produce sound pressure levels of 90-100dB at frequencies above 10kHz and 9.71W electrical input power. While at around 100dB per mW commercial headphones are about 3500 times more efficient, as Brown points out the tungsten nanospeakers benefit from having planar structures with no moving parts, and it is possible to control the phase with arrays of the devices, which could allow directional sound production.

Potential applications include pressure transduction in microfluidics, non-destructive testing and medical ultrasonics. In addition the ALD themoacoustic tungsten nanobridges have already proved profitable as a testbed for thermoacoustic modelling. The propagation of sound from a nanostructure can be described in spherical waves – which fit assumptions of a substantial distance from the source – or planar waves as they would be experienced at close encounters. However, the relevant distances depend on the size of the source and the wavelength, which raised questions for modelling nanosources.

“It wasn’t clear how to reconcile nanoscale models with macroscale experiments,” says Brown. This latest study reports an approach to accomplish this, and confirms spherical wave propagation of the sound, while providing insights of the limits for the different regimes.

How is it made?

Thermoacoustic systems use pulsed heating from an a.c. current to generate pressure waves that propagate as sound. The absence of voice coils and magnetic materials prone to resonance effects makes thermoacoustics an attractive approach for generating sound from a nanoscale source, and this has not been missed by other groups. However, previous work has focused on systems based on novel material systems such as graphene and carbon nanotubes, which require high processing temperatures above 600 °C.

Some prior work also created metal nanobridges but required removal of underlying silicon in order to suspend the bridges, making it difficult to integrate the devices in this prior work with underlying circuitry or other structures. In contrast, the ALD fabrication processes the Colorado researchers adopt are performed at 130 °C and use a sacrificial polymer layer that can be removed to suspend the nanobridges without damaging underlying microstructures, creating the potential for integration with CMOS and roll-to-roll fabrication.

By exposing surfaces to cycles of self-limiting chemical reactions due to exposure to gaseous reagents, ALD achieves impressive structural and chemical control. Although more commonly used for ceramics, the researchers had the expertise within the team – in particular with Victor M Bright who leads a research group that has been bringing ALD into devices for almost 15 years – to apply the approach to metals such as tungsten.

How is it modelled?

Brown and his team were motivated by how their devices might fit into models of thermoacoustic devices, but as Brown is prompt to point out, the maths can get very complicated. “I’m interested in making a bridge with physics, and the emphasis is on getting behind the maths and translating from fundamental physics to an engineering approach.”

The researchers monitored the efficiency of their device as they varied a number of parameters to test the models. For this it was important to use a lumped element approach to test geometric parameters. Previous work has used continuum models that assume infinite-length beams and this ignores the effect of thermal conduction within finite beams.

From their experiments – largely performed by Nathan C Moore, who is still an undergraduate at the University of Colorado – the researchers noted that the efficiency increased with both power and frequency. Importantly they noted a cubic proportionality with frequency, which suggests spherical wave propagation; planar propagation would lead to a directly proportional relationship.

Brown is keen to look into how their models might apply to thermal-piezoelectric resonators and other microdevices. He is also interested in the performance of their speakers at frequencies beyond a MHz, at which point they would expect some of their assumptions about the relative transport properties of the thermal and pressure waves to break down. However, the next steps will be looking into how to progress their technology for integration into flexible substrates and CMOS.

Full details are reported in Nanotechnology.