New technologies for air-coupled ultrasonic transducers

Mate Gaal and Daniel Kotschate

Bundesanstalt für Materialforschung und -prüfung (BAM), Berlin, Germany,
[email protected]

Abstract

Air-coupled ultrasonic testing (ACUT) has experienced rapid growth within the last
years. It is especially well suited to inspection of lightweight structures consisting of
composite materials and adhesive joints. Uniform coupling and easy maintenance are its
advantages compared to contact technique. However, the impedance mismatch between
the transducer and air poses a major challenge to the development of ACUT
transducers. Commercially available air-coupled transducers consist of a
piezocomposite material and matching layers. Their fabrication is difficult in handling
and their signal-to-noise ratio sometimes not sufficient for various testing requirements.
However, there are several innovative approaches using other materials and other
physical principles to transmit and receive an ultrasonic pulse.
We present a review of the latest advances in research on air-coupled transducers for
non-destructive testing, including previously unpublished results. We recognize two
major directions as most promising: ferroelectrets and thermoacoustic transducers.
Ferroelectrets are charged cellular polymers exhibiting piezoelectric properties. Their
small acoustic impedance is matched to air better than matching layers applied in
conventional air-coupled transducers. Applying bias voltage to a ferroelectret receiver is
the latest development in this field, which increased the received signal by 12 to 15 dB.
Thermoacoustic transducers use heat to initiate an ultrasonic wave, acting as
transmitters. The working principle is known from nature as thunder and lightning:
thermal energy of an electrically heated material, which can also be air, is converted
into acoustic energy. Some thermoacoustic transmitters consist of a conductive layer
with a thickness in the nanometer range deposited on a solid substrate. Another
possibility is to use an electric spark. For the first time, measurements of the sound field
of an electric spark up to 500 kHz were performed. Thermoacoustic transducers enable
excitation of extremely broadband pulses while producing high pressure levels, which
opens new possibilities for advanced signal processing.

1. Introduction
Since its beginnings in 1970s, air-coupled ultrasonic testing (ACUT) has experienced
rapid growth as a non-destructive method to detect defects (1). Typical applications
include inspection of lightweight structures consisting of composite materials like fibre-
reinforced polymers or sandwich structures and metal adhesive joints. Compared to
contact technique, it offers uniform coupling conditions and easier maintenance, since
there is no running water. Some materials with sensitive surfaces, which would be
damaged by a fluid couplant, are inspected with ACUT. However, the impedance

Creative Commons CC-BY-NC licence https://creativecommons.org/licenses/by-nc/4.0/

mismatch between any solid and air poses a major challenge to this technique. The
signals are much smaller than with the contact technique, so that the sensitivity of air-
coupled transducers is the main issue in their development. For this reason,
commercially available air-coupled transducers consist of a piezocomposite material
and matching layers matched to air. Their fabrication is difficult and their signal-to-
noise ratio sometimes not sufficient for testing requirements of thicker probes or probes
with higher damping. However, there are several innovative approaches using other
materials and other physical principles to transmit and receive an ultrasonic pulse.

This contribution is a review of the latest advances in research on air-coupled

transducers for non-destructive testing, including some previously unpublished results.
Two major directions are recognised as most promising: transducers based on
ferroelectrets and thermoacoustic transducers.

2. Ferroelectret transducers
2.1 The functioning principles

Ferroelectrets are charged cellular polymers exhibiting piezoelectric and pyroelectric

properties (2)(3)(4). Typically, their cellular structure is strongly anisotropic, so that
they are extremely soft in one direction, which is also the direction of their polarisation.
Although they were named after ferroelectrics (or piezoelectric materials), the
underlying mechanism of their piezoelectric properties is different. In ferroelectrets, the
stress in the 3-direction reduces the dipole moments, so that the piezoelectric coefficient
d33 is positive.

Most part of research on ferroelectrets was performed on cellular polypropylene. The

manufacturing of this material begins with melting and foamization, followed by
extrusion, cooling to the crystallization temperature, heating to the orientation
temperature and finally biaxial orientation, which is responsible for the cellular structure
(3). After this last step, the cellular polypropylene film is subjected to corona charging,
so that ions are trapped at the cell walls, resulting in remanent polarization (5). This
polarization leads to piezoelectric properties of ferroelectrets. They were named after
ferroelectrics and electrets, because they unite their properties.

Cellular polypropylene has a thickness between 50 and 100 µm, an acoustic impedance
about 0.03×106 Pa·s/m at frequencies around 250 kHz and a piezoelectric constant d33
around 200 pC/N. Its Young modulus c33 is strongly frequency-dependent, so that d33
and the acoustic impedance are frequency dependent too (6)(7). A study of their
viscoelastic properties revealed that the Young modulus of cellular polypropylene with
the product name HS06-20-BR by company EMFIT varies from 0.04 to 0.8 MPa in the
frequency range from 0.3 Hz to 300 kHz, which was well modelled by Cole-Cole
relation describing viscoelastic materials (7).

The thickness change of a ferroelectret transmitter is described by the relation

𝜀 𝜀𝑉0
∆ℎ𝑡 = 2𝑐 𝑉𝑡2 + 𝑐 𝑉𝑡 , (1)
33 ℎ 33 ℎ

2
where ε is the permittivity of the ferroelectret, c33 its Young modulus in thickness
direction, h its thickness, V0 its internal voltage caused by polarization and Vt the
excitation voltage (2)(3). The first of the two terms describes the thickness change due
to electrostatic force (the functioning principle of a capacitive loudspeaker) and the
second term the piezoelectric properties. However, it should be considered that the
elasticity is frequency-dependent, so that the thickness change depends on the time
behaviour of the excitation signal, which is not considered in Equation (1). When the
same ferroelectret transducer is used as a receiver, the generated voltage can be
described by the equation
∆ℎ
∆𝑉 = 𝑉0 ℎ , (2)

where V0 is the internal voltage and Δh the thickness change of the receiver. This
equation also describes capacitive microphones, with V0 standing for external bias
voltage.

2.2 Transducers based on ferroelectrets

Typically ferroelectret films are glued on an electrode, while the other electrode is
deposited at the other surface (8)(9). The electrode deposition is performed using
electron beam evaporation, evaporating layers of aluminium, gold or titanium with a
thickness between 10 and 500 nm. The extraordinary flexibility of thin ferroelectret
films enables cylindrical and even spherical focusing, which combined with high-
voltage excitation creates sound pressure levels above 140 dB (10). When excited with
square pulses of 1800 V, ferroelectret transducers with a diameter of 19 mm and a
resonance frequency between 200 and 300 kHz produce much higher signals than the
commercial transducers, leading to an increase of the signal-to-noise ratio by 20 dB.
These transducers were successfully applied to inspection of adhesive joints and CFRP
(carbon-fiber-reinforces plastic) plates, detecting inserts of 1 mm size within joints and
plates with a total thickness of about 4 mm. Applying additional bias voltage increases
the signal additionally by 12 to 15 dB, which is described later in this section. Stacking
several layers together decreases the resonance frequency with some loss of sensitivity
(9)(11)(12).

Since the ferroelectret film is glued onto one of the electrodes, various transducer
geometries can be easily created by structuring that electrode, while the electrode
deposited at the other side of the ferroelectret may be continuously connected. Not only
phased array probes can be easily produced this way (13), but also a twin transducer
with the transmitter and the receiver placed at the same piece of ferroelectret film. Two
concentric electrodes were placed on the same circuit board, which was spherically
focussed with curvature radius of 50 mm. As shown in Figure 1(a), the inner electrode
was a circle with a varied diameter and the outer electrode a ring around that circle with
an outer diameter of 27 mm. A ferroelectret film HS06-20-BR by company EMFIT was
glued onto the circuit board and deposited with 100 nm aluminium using electron beam
evaporation, applying the technology earlier developed for focusing ferroelectret
transducers (10). This is why the structure of the electrodes and the division into the
transmitting and the receiving part is not visible (see Figure 1(b)). The sound field of
this transducer was evaluated by using the reflection from a circular rod, as illustrated in

3
Figure 2(a). This rod with a 6 mm diameter was moved in the sound field of the
transducer and the reflected signal was recorded and evaluated as a C-Scan, as seen in
Figure 2(b). The sound field resembles a sound field of a circular focussing transducer.
The 6dB focal size depends on the size of the receiver, which is the circular electrode in
the middle. For the receiver diameter of 8, 10 and 12 mm the following focal sizes were
measured: 6.5, 3.1 and 3.7 respectively.
27

8
Type A

Type B
10

Type C
12

Transmitter
Receiver

(a) (b)
Figure 1. (a) The layout of the electrodes and (b) a photography of the ferroelectret twin probe. The
electric units for the transmitter and for the receiver are placed in separate housings.

(a) (b)

Figure 2. The sound field of the twin probe Type B recorded as a C-Scan of the reflection from a
circular rod with a diameter 6 mm. The transducer was positioned at x = – 9 mm, y = 0 while the
rod was moved. (a) The experimental setup and (b) the resulting C-Scan.

If ferroelectrets are understood as capacitive microphones with an internal bias voltage

V0, then the next natural thought is to apply additional bias voltage to increase their
sensitivity (8). Ferroelectret receivers with no additional bias voltage were constructed
and built consisting of two modules: one containing the deposited ferroelectret and the
other containing a pre-amplifier (10). An additional module was constructed that
provided bias voltage VDC from -2000 to +2000 V (8). When this bias voltage module
was connected and +2000 V applied, the received voltage increased by 12 to 15 dB,
while the noise increased by only 1 dB. This was a result of an experiment with a
through transmission of air, where the received peak-to-peak signal was evaluated
against varying bias voltage as shown in Figure 3(a).

A physical model explaining this result can be obtained by expanding Equation (2), so
that the bias voltage consists of ferroelectret internal voltage and additional DC voltage:

4
 Vdc = 0
1000

800
Receiver Signal (mV)

V0
Peak-to-Peak

600

400 Vdc = 2 kV

200

0
-2000 -1000 0 1000 2000
Vdc (V)

(a) (b)
Figure 3. (a) The influence of additional bias voltage on the sensitivity of the ferroelectret receiver.
V0 is the internal voltage of the ferroelectret receiver and Vdc the additional bias voltage. (b) C-Scan
of spruce timber with a thickness 35 mm with and without additional bias voltage, with a colour
scale in dB.

∆ℎ
∆𝑉 = (𝑉0 + 𝑉𝐷𝐶 ) . (3)
ℎ

The intercept gives an estimate of the internal voltage V0 of the evaluated single-layer
transducer, which is around 500 V. Inspection of spruce timber in Figure 3(b) gives an
example how the application of additional bias voltage increases the signal-to-noise
ratio.

3. Thermoacoustic transducers
We all know the natural phenomenon called thunder and lightning well from our daily
life. The electrical discharge of the charged cloud particles causes a rapid change in the
temperature gradient, which generates the rumble of thunder. Due to the short and
strong excitation, the pressure of the fluid is increased and propagates as shock wave
front, travelling kilometres until it vanishes. Various observations of thermoacoustically
induced acoustic oscillations have been reported over time. Byron Higgins wrote the
first scientific article about this phenomenon when he described the generation of sound
during glass blowing process (14). Plasma speakers make use of thermoacoustic effect
to reproduce audible sound with limited power.

One of the types of thermoacoustic transducers consist of a deposited conductive layer

based on fused silica. Since the conductive layer acts as a regular conductor, electrical
energy is dissipated and converted to heat by Joule's heating. This heat is responsible for
an increase of the pressure, which is how an acoustic wave is initiated. The heat is
proportional to the electrical power, so that a sine-wave electric excitation leads to an
acoustical signal with the double frequency. Such transducers were already applied in

5
non-destructive testing. A thermoacoustic transmitter was combined with a ferroelectret
receiver to inspect a 4 mm thick CFRP test piece (15).

Another possibility is to use gas discharges (electric spark or arc). As described, the
most familiar natural phenomena are lightning and thunder. Gas discharges excite a
wideband acoustic signal. For this paper, measurements of this excitation on electric
spark discharges up to 500 kHz were performed. Two significant effects affect the
acoustic excitation: a) the electrodynamic and b) the thermoacoustic effect. The
electrodynamic influence relies on the production of charged particles and the body
force caused by the electric field acting on these. We can neglect the electrodynamic
influence compared with the thermoacoustic one, because the recombination processes
between ions and other chemical compounds in air (e.g. carbon monoxide CO and
atmospheric methane CH4) dominate. Since the applied electrical field accelerates the
charged particles, they interact with neutral particles inside the volume. Elastic and
inelastic collisions between charged (electrons, ions) particles and neutral molecules of
the gas mixture cause the temperature increase. Depending on the kind of discharge, the
thermal change can be continuous (arc) or limited in time (spark).

We investigated the acoustic emission of gas discharges using a prototype as shown in

Figure 4, with 1 mm distance between the electrodes, applying the sound particle
velocimetry method described in (15). For the pulse-shaped emission initiated with
7.5 kV pulses, frequency components in a broad frequency range were observed. Figure
5 illustrates the characteristics of these acoustic emissions. The recorded signal is not
perfectly broadband, but it consists of several resonance frequencies with different
amplitudes. Due to the nature of the gas discharges, the acoustic response is strongly
dependent on the environmental settings. However, we could reproduce similar
characteristics over a period of one year without any modifications on the experimental
setup, using the same prototypes. Several types of gas discharges were tested and an
electric spark could produce a maximum sound pressure level of about 137 dB (Figure
5), which is comparable to commercial transducers.

The electric spark discharge generates pressure variation, which tends to be non-linear.
Figure 5 shows the sound pressure level over distance emitted by an electric spark
discharge. The non-linear behaviour dominates for distances below 400 mm and
consists of two dominating effects, which both were observed experimentally (16): a)
the wave-steering and b) the acoustic saturation. Wave-steering describes the change of
the signal shape depending on the distance from the source, whereas higher harmonics
in the acoustic spectrum appear. The acoustic saturation is caused by the streaming
processes of the fluid. For low-intensity acoustic excitations, the influence of streaming
processes is very weak and can be neglected. For high-intensity acoustics, they consume
energy and must be considered. The distance independence of the sound pressure level
of spherical waves apparently contradicts the conservation of energy, but it can be fully
explained by the convective terms and the presence of wave-steering. The depicted
effect is called acoustic saturation and was observed by various authors (16). In non-
linear regime, the energy is contained in higher harmonics.

6
 Figure 4. Schematic representation of the initiation of an electric spark.

Figure 5. (a) Measured acoustic excitation of a gas discharge and (b) the corresponding Fourier
transformed. The acoustic signal was recorded at 440 mm distance from the source.

Figure 6. Sound pressure level over distance of an electric spark discharge.

7
4. Conclusions
Cellular polypropylene is a ferroelectret material offering new possibilities for the
construction of air-coupled transducers. It can be easily formed, bended and structured,
so that various apertures are possible. A recent innovation is a twin probe with a
transmitter and a receiver placed on the same ferroelectret film. For common
frequencies for air-coupled ultrasonic testing, it offers sensitivity 20 dB higher than
transducers based on piezoceramics. If additional bias voltage is applied to the receiver,
the sensitivity increases further by 12 to 15 dB. This makes it the most sensitive
ultrasonic transmission system for non-destructive testing available. Ferroelectret
transducers open new possibilities for the inspection of components with high signal
losses. Beyond non-destructive testing, other applications are possible, where highly
sensitive microphones in ultrasonic range are required.

The newest generation of thermoacoustic transmitters offer extremely high bandwidth

and sound pressure levels comparable to conventional air-coupled transducers.
Combined with other types of receivers, they are being considered for ultrasonic
inspection. With their broad bandwidth they potentially offer more information about
the inspected object, which would broaden the application field of air-coupled ultrasonic
testing.

Acknowledgements

We would like to express our gratitude to: Florian Schadow for the preparation of the
twin probes and for the sound field measurement; Uwe Beck and Matthias Weise for the
electron beam evaporation in all our projects; to Joachim Döring, Viktor Bovtun and
Jürgen Bartusch for years of fruitful cooperation. Our work was funded by programs
ZIM, MNPQ-Transfer and VIP of the German government.

References

1. D. E. Chimenti, “Review of air-coupled ultrasonic materials characterization,”

Ultrasonics, vol. 54, no. 7, pp. 1804–1816, 2014.
2. G. M. Sessler and J. Hillenbrand, “Electromechanical response of cellular electret
films,” Appl. Phys. Lett., vol. 75, no. 21, pp. 3405–3407, 1999.
3. M. Paajanen, J. Lekkala, and K. Kirjavainen, “ElectroMechanical Film (EMFi) -
a new multipurpose electret material,” Sensors Actuators, A Phys., vol. 84, no. 1,
pp. 95–102, 2000.
4. S. Bauer, R. Gerhard-multhaupt, G. M. Sessler, S. Bauer, R. Gerhard-multhaupt,
and G. M. Sessler, “Ferroelectrets : Soft Electroactive Foams for Transducers,”
Phys. Today 57, vol. 2, no. 2004, pp. 37–43, 2004.
5. M. Wegener et al., “Controlled inflation of voids in cellular polymer
ferroelectrets: Optimizing electromechanical transducer properties,” Appl. Phys.
Lett., vol. 84, no. 3, pp. 392–394, 2004.
6. J. Hillenbrand and G. M. Sessler, “DC-biased ferroelectrets with large
piezoelectric d33 -coefficients,” J. Appl. Phys., vol. 103, no. 7, 2008.
7. M. Gaal, V. Bovtun, W. Stark, A. Erhard, Y. Yakymenko, and M. Kreutzbruck,

8
 “Viscoelastic properties of cellular polypropylene ferroelectrets,” J. Appl. Phys.,
vol. 119, no. 12, 2016.
8. M. Gaal, R. Caldeira, J. Bartusch, and M. Kupnik, “Air-Coupled Ultrasonic
Ferroelectret Receiver with Additional DC Voltage,” Proc. Eurosensors 2017,
vol. 1, no. 4, p. 362, 2017.
9. J. L. Ealo, J. C. Prieto, and F. Seco, “Dynamic response estimation of multilayer
ferroelectret-based transducers using lumped-element electromechanical
models,” Proc. - IEEE Ultrason. Symp., 2009.
10. M. Gaal, J. Bartusch, E. Dohse, F. Schadow, and E. Köppe, “Focusing of
ferroelectret air-coupled ultrasound transducers,” in AIP Conference
Proceedings, 2016, vol. 1706.
11. A. Streicher, “Luftultraschall- Sender-Empfänger-System für einen künstlichen
Fledermauskopf,” Universität Erlangen-Nürnberg, 2008.
12. M. Wegener, S. Bergweiler, W. Wirges, A. Pucher, E. Tuncer, and R. Gerhard-
Multhaupt, “Piezoelectric two-layer stacks of cellular polypropylene
ferroelectrets: Transducer response at audio and ultrasound frequencies,” IEEE
Trans. Ultrason. Ferroelectr. Freq. Control, vol. 52, no. 9, pp. 1601–1605, 2005.
13. J. F. Pazos-Ospina, J. L. Ealo, and J. Camacho, “New dual-focalization
ferroelectret-based array for air-coupled ultrasonic inspection of textiles,” NDT E
Int., vol. 74, pp. 50–57, 2015.
14. K. T. J. Feldman, “Review of the Literature on Rijke Thermoacoustic
Phenomena,” J. Sound Vib., vol. 7, no. 1, pp. 83–89, 1968.
15. M. Daschewski, M. Kreutzbruck, J. Prager, E. Dohse, M. Gaal, and A. Harrer,
“Resonanzfreie Messung und Anregung von Ultraschall,” Tech. Mess., vol. 82,
no. 3, pp. 156–166, 2015.
16. M. F. Hamilton and D. T. Blackstock, Nonlinear acoustics. Academic Press,
1998.