Electret transducers for measuring acceleration and

structure-borne sound
J. Hillenbrand, S. Haberzettl, T. Motz, and G. M. Sessler
Darmstadt University of Technology, Institute for Telecommunications Technology,
Merckstrasse 25, D-64283 Darmstadt, Germany
email: j.hillenbrand@nt.tu-darmstadt.de

Abstract
The number of electret microphones produced every year significantly exceeds that of all other
microphone types. This is due to the very good acoustical properties and the simple and low-cost design of
these electret air-borne sensors. In contrast, most of the discrete structure-borne sound sensors (or
accelerometers) are based on the piezoelectric effect. In the present work, accelerometers utilizing the
electret principle were constructed, built and characterized. These electret accelerometers comprise a
metallic seismic mass, covered by an electret film, a ring of a soft cellular polymer, and a metallic
backplate which replace membrane, spacer, and back electrode, respectively, of the electret microphone.
An adjustable static pressure to the seismic mass is generated by one or two metal springs. The dynamic
characterization of the electret accelerometers was carried out in the frequency range from 10 Hz to 8 kHz
by using an electrodynamic shaker and an external charge amplifier. Sensors with seismic masses from 8
to 25.5 g were investigated. With these masses typical charge sensitivities from 10 to 40 pC/g, voltage
sensitivities from 600 to 2000 mV/g, and resonance frequencies from 3 to 1.5 kHz were measured. The
relation between sensitivity, seismic mass and resonance frequency was calculated by a simple analytical
model and compared with measured results. Good agreement of experimental and calculated data was
found. The present work shows that sensitive, lightweight, and inexpensive electret accelerometers can be
built for applications in various fields, such as active noise control and adaptronics.

Introduction

Accelerometers [1] are acoustical sensors [2], just as electret microphones [3], conventional condenser
microphones, or dynamic microphones. They belong to the group of vibration sensors, used for detecting
vibrations of solid objects and structure-borne sound. However, accelerometers are frequently used also in
various non-acoustic applications, for example in the airbag deployment system of cars, in portable
electronic devices, such as smartphones and PDAs, and in the controllers for video game consoles, where
they are employed to measure, often very low-frequency, accelerations. Accelerometers can be based on
several physical principles. Capacitive, piezoresistive, and piezoelectric accelerometers are the most
common types [4].
Recently, piezoelectric accelerometers based on polypropylene (PP) ferroelectrets were constructed and
investigated [5]. Ferroelectrets [6], sometimes also referred to as piezoelectrets [7,8], are piezoelectric
films of charged cellular or porous polymers and have comparable or larger d33-coefficients than lead
zirconate titanate (PZT). It was shown [5,9], that these piezoelectret-based accelerometers are sensitive,
lightweight, and inexpensive. In addition, they can be used at very low frequencies and their output signals
are rather insensitive to temperature variations, which is due to the low conductivity and the small
pyroelectric effect of the piezoelectret films, respectively. However, charges in piezoelectrets made of
cellular PP are not stable for temperatures above 60C and thus the sensitivity of PP based piezoelectret
accelerometers gradually decreases when subjected to such environmental conditions.

571

572

P ROCEEDINGS OF ISMA2010 INCLUDING USD2010

Two alternative routes exist to overcome this drawback. The first one is to replace the PP piezoelectrets by
more temperature-stable composite piezoelectrets made of the fluoropolymers PTFE and FEP [10].
Results obtained with this approach will be published elsewhere. The second possible route is described in
the present paper. Its basic idea is to employ in one sensor both a cellular PP and a solid FEP film. The
cellular film operates as a spring and the FEP film stores electret charges, while in a piezoelectret
accelerometer both functionalities have to be fulfilled by a single cellular film. Accelerometer designs
based on these considerations are very similar to standard electret microphone designs and thus
comparable temperature stabilities can be obtained. As is well-known, the sensitivity of electret
microphones is stable for many years even at elevated temperatures and high humidity. Such microphones
can thus be used in various applications and the same can be expected for electret accelerometers.

Experimental procedures

2.1

Electret accelerometers

For the studies presented in the following, a relatively large and heavy accelerometer design was used. It
allows easy and fast replacement of its components and also the adjustment of the static pressure acting on
the soft cellular ring. Smaller and lighter accelerometers, with housings protecting them from acoustic and
electrical noise, are planned and will be realized in the future.
The present accelerometer design and its main components are shown in Figure 1. Two spring/screwcombinations act on the two sides of a metal arbor, which in turn is in direct contact to the seismic mass.
This generates an adjustable static force to the seismic mass and the cellular polymer ring and anchors the
components in the accelerometer in such a way, that movement and vibration of the seismic mass in
measuring direction is not hindered but movements in other directions are restricted. The static force not
only anchors the components of the accelerometer but also influences its sensitivity and resonance
frequency. This is due to a dependence of Youngs modulus of the cellular polymer ring on static pressure.
Besides serving as a spring, the cellular ring with a thickness of about 40 m also adjusts the distance
between the Teflon FEP-film surface and the back electrode. The FEP-films are corona-charged to
voltages up to about 1000 V before assembling the accelerometer. The two output electrodes consist of the
metallic backplate and the onesided metallized FEP-film.

onesided metallized
FEP-electret film

seismic mass

cellular PP film ring

output

backplate
airgap
Figure 1: Design of an electret accelerometer. A static force is generated by two spring/screwcombinations allowing free movement and vibration of the seismic mass in z-direction. The
cylindrical seismic mass has a diameter of 11.5 mm and variable height. The outer cellular ring
diameter is 11.5 mm, the inner diameter 10 mm and the ring area 25 mm2. The output signal is
taken from the metallized side of the FEP-film and the metallic backplate.

A DAPTRONICS

2.2

573

Acceleration measurements

The electret accelerometers is mounted on an electrodynamic vibration exciter (B&K 4809). The exciter is
driven by a power amplifier (B&K 2713). The charge or voltage generated by the accelerometer is
amplified by a charge amplifier (B&K 2635) or an electrometer (Keithley 600B), respectively. An audio
analyzer (R&S UPD) is used to generate the signal for the power amplifier and to record and analyze the
output signal of either the charge amplifier or the electrometer.
In the majority of cases, sinusoidal excitation signals were used in the range from 10 Hz to 8 kHz and the
output signal of the charge amplifier or the electrometer was filtered with a digital bandpass before the
amplitude was calculated. In some experiments, white noise excitation was employed and the output of
the charge amplifier was analyzed with a Fast Fourier Transformation (FFT). As expected, both methods
yield very similar frequency responses for the tested accelerometers.
The entire measuring setup described above does not have a flat frequency response. This is basically due
to the vibration exciter. For a constant input voltage, the resulting acceleration significantly decreases
below 100 Hz and above a resonance of the shaker at about 10 kHz. Therefore, calibration measurements
of the setup were performed using commercial accelerometers (B&K 4332 and 4344). The smaller
weights of the commercial accelerometers, compared to the ferroelectret accelerometers, were
compensated by mounting additional masses on the shaker during the calibration measurements, which
ensures that in this case the same acceleration is generated by the shaker.
Usually, the frequency dependent acceleration generated by the shaker was accepted and finally
compensated numerically by the calibration curves. However, if linearity was tested or non-linear effects
were expected or observed, in particular non-linear behavior of Youngs modulus of the ferroelectret
films, the equalization possibility of the audioanalyzer was utilized: Measuring the frequency response of
the output signal in a first step and applying the inverse response to the input in further measurements
assures relatively frequency independent accelerations at the shaker.

Sensitivity measurements with charge output

An electret accelerometer can be operated in two modes: Either the generated charge or the generated
voltage is amplified and measured. Measurements with a charge amplifier (B&K 2635) are presented in
this Section, measurements with an electrometer (Keithley 600B), operated as a high impedance voltage
amplifier, are discussed in Section 4.

3.1

Variation of the seismic mass

Measured frequency responses of the charge output sensitivity of an accelerometer are shown in Figure 2.
The various measurements were performed with seismic masses from 8 to 25.5 g, the static pressure on
the cellular PP ring was kept constant at a value of 77 kPa. For all measurements, a relatively flat or
slightly decreasing response for lower frequencies and a resonance peak at higher frequencies was
observed. Furthermore, when the seismic mass is enlarged, the sensitivity of an accelerometer increases
and its resonance frequency decreases. These findings can be understood when the cellular ring, with
Youngs modulus Y, and the seismic mass ms is considered as a spring-mass-system. A brief derivation of
this model is given in the following.

574

P ROCEEDINGS OF ISMA2010 INCLUDING USD2010

Sensitivity [pC/g]

100

ps = 77 kPa

10
Seismic mass ms:
8,0 g
10,2 g
12,3 g
13,9 g

1
10

16,7 g
18,9 g
22,6 g
25,5 g

100

1000

Frequency [Hz]
Figure 2: Charge output sensitivity of an accelerometer with different seismic masses from 8 to
25.5 g. For all measurements the same static pressure of 77 kPa was applied to the cellular ring.
The metallic backplate, the airgap, the FEP-film and its metallization form a capacitor (see Figure 1).
Since the film is charged, a change of the airgap and thus a change of the capacity generates charges in the
two output electrodes. Taking into account Newtons second law, under short-circuit conditions a charge
Q is generated in the accelerometer with the seismic mass ms, when an external acceleration a is applied:

Q=

c
ms a .
Y A

(1)

The constant c depends on properties of the accelerometer, such as airgap distance, electret voltage,
electret area, Youngs modulus and area of the cellular ring. The charge sensitivity S=Q/a of an
accelerometer can then be written as

S=

Q
c
=
ms ,
a Y A

(2)

indicating that the sensitivity S is linearly dependent on its seismic mass and inversely proportional to
Youngs modulus Y and the area A of the cellular ring.
The resonance frequency f res = res / 2 of a spring-mass-system is given by res = D / m with the
stiffness D = YA / t , where Y is Youngs modulus, A the area, and t the thickness of the cellular ring.
Therefore, the resonance frequency of the accelerometer can be written as

f res =

1
2

Y A
.
ms t

(3)

A DAPTRONICS

575

In Figure 3, calculated values from eqs. (2) and (3) were fitted and compared with values obtained from
the frequency responses of Figure 2. Measured sensitivities (@100 Hz) and resonance frequencies are
shown as symbols and the two solid lines are the calculated values. Both, the linear and the inverse root
dependency are in good agreement with the measured data points.

Sensitivity [pC/g]

2.5
30

2.0
1.5

20

1.0
10
measured sensitivity
measured resonance frequency

10

15

20

25

0.5
0.0

Resonance frequency [kHz]

3.0

40

Seismic mass [g]

Figure 3: Sensitivity (@100 Hz) and resonance frequency taken from the various curves shown in
Figure 2 (symbols) and calculated from eqs. (2) and (3) (solid lines).

3.2

Variation of the static pressure

Applying a static pressure to the seismic mass ensures a stable bilateral fastening of the seismic mass in
the accelerometer without hindering the seismic mass to move and to vibrate in the direction perpendicular
to the film plane. The static pressure, which can be quite large compared to the acceleration forces,
influences sensitivity and resonance frequency of the electret accelerometers. While the dependence of the
two parameters on the seismic mass can be explained by a spring-mass-system (Section 3.1), their
dependence on the static pressure is more complicated and will be illustrated in the following.
In Figure 4, various frequency response measurements of the sensitivity for static pressures on the cellular
ring from 41 to 125 kPa are shown. For all measurements a seismic mass of 8 g was used.
As for the measurements with various seismic masses, nearly flat responses for low frequencies and
resonance peaks at high frequencies were observed. The sensitivities at 100 Hz and resonance frequencies
of Figure 4 are presented in Figure 5 (solid circles) as a function of the applied static pressure. It can be
seen, that the resonance frequencies increase and the sensitivities decrease with increasing pressure.

576

P ROCEEDINGS OF ISMA2010 INCLUDING USD2010

Sensitivity [pC/g]

100

10
Static pressure ps:
41 kPa
53 kPa
65 kPa
77 kPa

1
10

89 kPa
100 kPa
115 kPa
125 kPa

100

ms = 8 g

1000

Frequency [Hz]
Figure 4: Sensitivity of an accelerometer with a seismic mass of 8 g. The static pressure on the
cellular ring was varied from 41 to 125 kPa.
The behavior of sensitivity and resonance frequency as a function of the static pressure can be explained
with eqs. (2) and (3), if a pressure dependent Youngs modulus Y(ps) and area A(ps) are assumed:

S=

Q
c
=
ms ,
a Y ( ps ) A( ps )

(4)

Y ( ps ) A( ps )
.
ms t

(5)

f res =

1
2

It is further assumed, that Y(ps) is constant for low pressures and increases (or decreases) for higher
pressures and that A(ps) starts at a small value for zero pressure and approaches the total area of the
cellular ring for high pressures. The first assumption concerning Y(ps) is true for virtually any material,
however, the pressure needed to significantly change Y is very low for the very soft cellular polymers. The
second assumption becomes obvious, when the rough and non-flat cellular polymer ring in contact with
Teflon on one side and brass on the other side is imagined under increasing pressure: the two contact areas
gradually increase with increasing pressure. Taking eqs. (4) and (5) and the assumptions about Y(ps) and
A(ps), decreasing sensitivities and increasing resonance frequencies of Figure 5 are qualitatively correct.
A quantitative comparison can be made when the unknown quantity Y(ps) A(ps) is eliminated in eqs. (4)
and (5). One obtains
S=
and

c
2
4 f res
t
2

(6)

A DAPTRONICS

577

f res =

1
2

c
.
S t

(7)

20
18
16
14
12
10
8
6
4
2
0

3.0
2.5
2.0
1.5
1.0

measured sensitivity
calculated sensitivity
measured resonance frequency
calculated resonance frequency

40

60

80

100

0.5
120

0.0

Resonance frequency [kHz]

Sensitivity [pC/g]

With these two equations, sensitivities can be calculated from resonance frequencies and vice versa.
Measured values, directly taken from Figure 4 (solid circles, thick lines), and values, calculated by eq. (7),
are presented in Figure 5 (open circles, thin lines). Measured and calculated values are in very good
agreement.

Static pressure [kPa]

Figure 5: Measured sensitivities (@100 Hz) and resonance frequencies of the various frequency
responses presented in Figure 4 (solid circles) and calculated values (open circles).

Sensitivity measurements with voltage output

As for piezoelectric accelerometers, the output signals of electret accelerometers can be amplified by a
charge amplifier or by a voltage amplifier. The relation between the output voltage Vout and the output
charge Qout of an accelerometer with the capacity Cacc is given by Vout = Qout / Cacc . Since the capacity of a
PZT plate is inversely proportional to its thickness and directly proportional to its dielectric constant,
which is in the range of 1000, a large thickness is needed, to obtain high voltage output signals. This
results in large, heavy and expensive PZT accelerometers. In contrast, the dielectric constant of an
electret transducer is close to 1. This allows the construction of thin, light and inexpensive accelerometers
with high voltage output. Electret microphones are sensors with exactly these attributes.
Since for all sensors with voltage output, the actually measured voltage depends on the capacities of
sensor, cable and amplifier, open circuit voltages are given in the following.
In Figure 6, frequency responses of the open circuit sensitivity of an accelerometer with seismic masses
from 8 to 22.6 g are shown. The static pressure on the cellular PP ring was kept at a constant value of
77 kPa for all measurements. Again, a relatively flat or slightly decreasing response for lower frequencies
and a resonance peak at higher frequencies were observed. Increasing the seismic mass increases the
sensitivity and decreases the resonance frequency. The similarity of the voltage output and the charge
output measurements experimentally supports the above mentioned linear relation Vout = Qout / Cacc .

578

P ROCEEDINGS OF ISMA2010 INCLUDING USD2010

10000

Sensitivity [mV/g]

Cs = 21,5 pF
ps = 77 kPa

1000
ms = 8,0 g
ms = 10,2 g
ms = 12,3 g
ms = 13,9 g

100
10

ms = 16,7 g
ms = 18,9 g
ms = 22,6 g

100

1000

Frequency [Hz]
Figure 6: Voltage sensitivity of an accelerometer with different seismic masses from 8 to 22.6 g. For
all measurements a static pressure of 77 kPa was applied to the cellular ring.

1750

3.0

1500

2.5

1250

2.0

1000
1.5
750
1.0

500
250
0

measured sensitivity
measured resonance frequency

10

15

20

0.5
0.0
25

Resonance frequency [kHz]

Sensitivity [mV/g]

Furthermore, the above presented model of a spring-mass-system, all statements given above, and the eqs.
(1) to (7) are valid also for accelerometers with voltage output, if simply the constant c is adapted. This
can also be seen in Figure 7, where calculated values from eqs. (2) and (3) are compared with values
obtained from Figure 6 and good agreement is found.

Seismic mass [g]

Figure 7: Voltage sensitivity (@100 Hz) and resonance frequency taken from the various curves
shown in Figure 6 (circles) and calculated from eqs. (2) and (3) (lines).

A DAPTRONICS

579

Conclusions

Electret accelerometers are lightweight, inexpensive, and sensitive. The charge and the voltage sensitivity
as well as the resonance frequency can be adjusted by several design details. These are area, Youngs
modulus, and thickness of the cellular ring, size of the seismic mass, thickness and electret voltage of the
FEP film, and static pressure on the cellular ring. Charge sensitivities of more than 40 pC/g and voltage
sensitivities of up to about 1600 mV/g were measured for resonance frequencies between about 1.5 and
3 kHz. Electret accelerometers can be used at extremely low frequencies, since both electret and signal
charges are virtually not dissipating due to the very low conduction inside such sensors. Furthermore, the
stability of electret accelerometers at elevated temperatures is expected to be comparable to that of electret
microphones and thus high enough for many potential applications. Finally, the output signals of electret
accelerometers should be rather insensitive to temperature variations within the temperature range of
electret stability. For both temperature effects, direct experimental evidence and qualitative measurements
are still missing but planned for the near future.

Acknowledgements
The authors gratefully acknowledge financial support by the Hessische Ministerium fr Wissenschaft und
Kunst and by the Deutsche Forschungsgemeinschaft.

References
[1] G. Gautschi, Piezoelectric Sensorics, Springer, Berlin (2002).
[2] T. D. Rossing (Ed.), Springer handbook of acoustics, Springer, New York (2007).
[3] G. M. Sessler (Ed.), Electrets, 3rd Ed. Morgan Hill, CA: Laplacian Press, Vol. 1, (1999).
[4] S. P. Beeby, G. Ensel, M. Kraft, N. White, MEMS Mechanical Sensors, Artech House, London
(2004).
[5] J. Hillenbrand, M. Kodejska, Y. Garcin, H. v. Seggern, and G. M. Sessler, High-sensitivity
ferroelectret-film accelerometers, Trans. on Dielectr. and Electr. Ins., Vol. 17 (2010).
[6] S. Bauer, R. Gerhard-Multhaupt, and G. M. Sessler, Ferroelectrets: Soft Electroactive Foams for
Transducers, Physics Today, Vol. 57(2), pp. 37-43, (2004).
[7] J. Hillenbrand and G. M. Sessler, High-sensitivity piezoelectric microphones based on stacked
cellular polymer films, J. Acoust. Soc. Am., Vol. 116, pp. 3267-3270, (2004).
[8] S. Bauer, Piezo-, pyro- and ferroelectrets: soft transducer materials for electromechanical energy
conversion, Trans. on Dielectr. and Electr. Ins., Vol. 13(5), pp. 953-962, (2006).
[9] M. Kodejska, J. Hillenbrand, and G. M. Sessler, Ferroelectret-film accelerometers with high
sensitivities, in: International Conference on Acoustics, NAG/DAGA 09, Deutsche Gesellschaft fr
Akustik, pp. 144-145, (2009).
[10] X. Zhang, J. Hillenbrand, and G. M. Sessler, Ferroelectrets with improved thermal stability made
from fused fluorocarbon layers, J. Appl. Phys. Vol. 101, 054114, (2007).

580

P ROCEEDINGS OF ISMA2010 INCLUDING USD2010