IMPROVED EQUIVALENT CIRCUIT AND FINITE ELEMENT METHOD

MODELING OF CAPACITIVE MICROMACHINED ULTRASONIC

TRANSDUCERS
Goksen G. Yaralioglu, Mohammed H. Badi, A. Sanli Ergun, and Butrus T. Khuri-Yakub,
E. L. Ginzton Laboratory, Stanford University, Stanford, CA, 94305
Abstract Equivalent circuit model has been widely used
to predict the bandwidth of capacitive micromachined
ultrasonic transducers (CMUTs). According to this model,
the lower cutoff of the bandwidth is determined by the time
constant of the parallel RC where R is dictated by the
radiation and C is determined by the electrical capacitance
of the transducer. The higher cutoff, on the other hand, is
determined by the membranes anti-resonance. In the
mechanical part of the model, the radiation impedance is
simply added to the membrane impedance assuming that the
membrane impedance does not change when it operates in
the immersion medium. Therefore, the mass loading effect of
the medium is neglected. Our finite element method
calculations showed that the mass loading on the membrane
impedance drastically lowers the membrane anti-resonance
frequency degrading the bandwidth. In this paper, we
present results of equivalent circuit modeling combined with
finite element analysis. We constructed a 3D finite element
model for one element of a 1D array. The element has 7
hexagonal membranes in the width dimension and it is
assumed that the membranes are replicated in the length
dimension infinitely by using symmetry boundary conditions.
By combining membrane impedance with equivalent circuit
model, we found that the center frequency of operation is 11
MHz and the bandwidth is 12.5 MHz close to the collapse
voltage. We also investigated the effect of the DC bias on the
center frequency. Decreasing the bias voltage increased the
center frequency without affecting the bandwidth assuming
the source impedance is zero.

I. INTRODUCTION
The analogy between mechanical structures and
electrical circuits is widely used for the analysis of
mechanical systems. Using the analogy, forces are
replaced by voltage sources and velocities are replaced
by electrical currents. Then, an equivalent circuit of
the system is constructed. This method becomes an
even more powerful tool for the analysis of
electromechanical systems where some parts of the
system are already in electrical domain. For example,
equivalent circuit analysis was successfully employed
for piezoelectric transducers for their design and
optimization [1]. Similarly, Hunt used the same
approach for the modeling of electrostatic transducers
[2]. Recently, the equivalent circuit model is employed

0-7803-7922-5/03/$17.00 (c) 2003 IEEE

for the characterization of capacitive micromachined

ultrasonic transducers (CMUTs) [3,4]. Fig. 1 shows
the equivalent circuit for a CMUT transducer. In the
electrical part, Co is the clamped capacitance of the
device. Spring softening capacitance and the
mechanical membrane impedance constitute the
mechanical part. The two parts are coupled together
through an electromechanical transformer. For a
parallel plate capacitor, electrical capacitance,
electrical field and conversion ratio are given by

C0 =

oS
V
, E em =
, n = C 0 E em
gx
gx

(1)

where S is the area, g is the initial gap, x is the

membrane displacement and V is the bias voltage.
When the transducer is operated in vacuum, the
mechanical port of the circuit is short-circuited. For
immersion devices, the mechanical port is simply
terminated by the radiation impedance.
In this paper, we will test the validity of the
equivalent circuit model by using finite element
method (FEM). First, the results of input impedance
calculations of the transducer in vacuum will be
compared using the two methods to find accurate
values for the circuit elements. This analysis will
include the membrane deflection. Then, the bandwidth
of the device will be evaluated for immersion
operation.

c soft

+
V

(1 : n )

C0

2
C em
C0

cm

Zm

lm

v
F

Figure 1. Electrical equivalent circuit of a CMUT.

2003 IEEE ULTRASONICS SYMPOSIUM-469

We used ANSYS 5.7 for the FEM calculations. In

the following subsections, 2D and 3D models will be
developed for vacuum and immersion devices. For
immersion devices, the bandwidth of the transducer
will be evaluated using both circuit model and FEM.
A. Input impedance in vacuum
Fig. 2 depicts the finite element model that was used
in the calculations. The model was assumed to be
axisymmetric around the line passing through the
membrane center. The membrane was meshed by
plane elements and the gap was meshed by
electrostatic parallel plate capacitors (Ci) as shown in
the figure. The clamped capacitance and turns ratio
are then given by
C 0 = C i , n = C i Ei
(2)

where Ci and Ei are the capacitance and electric field

in the small capacitor elements, respectively. By using
this model first, we calculated the input impedance of
a membrane that moves parallel to the substrate. We
simply coupled the movement of the nodes on the
electrode together in the vertical direction. This
structure resulted in a collapse voltage of 199.6 V.
Assuming 199 V bias voltage, the Co and n found to
be 13.84 fF and 1.3x10-5 nt/V, respectively. The
electrical input impedance is given by

Z in =

VAC
I cap

(3)

Fig. 3 shows the input impedance calculated using

FEM and equivalent circuit for a parallel plate
membrane. Results obtained using the two methods
agree very well.
1 m thick silicon membrane
membrane radius = 20 m
electrode size = 10 m

J
.
Ci , Ei

Figure 2. Finite
calculations

g = 0.3 m

element

model

for

vacuum

Imag. Part (ohms)

II. FINITE ELEMENT METHOD CALCULATIONS

Frequency (MHz)

Figure 3. Input impedance for a parallel plate

capacitor.
After testing the FEM model on a parallel plate
capacitor, we released the coupling over the electrode
region. For a membrane motion, the device
capacitance can be divided into two parts: a parallel
plate capacitance and parasitic capacitance [5]. This
evaluation can be carried out through the
electromechanical coupling coefficient calculation. For
more accurate circuit modeling this parasitic
capacitance should be included. Our calculations
showed that for the geometry depicted in Fig. 2, the
parasitic capacitance is 8%. Another difficulty arises
due to the calculation of turns ratio in (2). The
electrical forces are only applied to the electrode
region. However, the membrane impedance is
calculated as the ratio of the uniform applied force all
over the membrane to the average membrane velocity.
To solve this discrepancy, one needs to define an
effective n* through

n* = n

Zm
Zp

(4)

where Zm is the membrane impedance and Zp is the

partial impedance calculated assuming a uniform force
over the electrode. We calculated Zm and Zp using
FEM and for the geometry in Fig.2, the effective n* is
n * = 2.3n
(5)
If the above improvements are incorporated in the
circuit model one can obtain a very good match
between the circuit model and FEM calculations as
shown in Fig.4.

2003 IEEE ULTRASONICS SYMPOSIUM-470

Imag. Part (ohms)

Frequency (M H z )

Figure 4. Input impedance using improved circuit

model and FEM.

the FEM model. Then, we used the same model to

calculate the radiation impedance for a pressure
release baffle, which is more realistic boundary
condition for devices operating in relatively high
impedance medium such as water. According to
equivalent circuit model, the membrane impedance
equal to sum of this radiation impedance and the
membrane impedance in vacuum. Figs. 5 and 6 show
the calculation result. We also calculated the
membrane impedance in water by applying a uniform
force over the membrane surface in water and
calculating the resulting average velocity. This result
is also depicted in Figs. 5 and 6.

B. Immersion FEM calculation (single cell)

For immersion devices, the mechanical port of the
circuit model is terminated by the radiation impedance
of the medium. Therefore, the impedance seen by the
membrane is given by the summation of the
mechanical impedance of the membrane in vacuum
and the radiation impedance. In this subsection, we
will test the validity of this approach for a single
membrane by calculating its mechanical impedance in
the immersion medium.
Fig. 5 shows the FEM model of a single membrane
immersed in water.

Figure 6. Real part of the membrane impedance

calculated using equivalent circuit approach and FEM.

absorbing
boundary

membrane

pres=0

Figure 7. Imaginary part of the membrane impedance

calculated using equivalent circuit approach and FEM.

Figure 5. FEM mesh for a single membrane.

Using the above model, first we calculated the
radiation impedance of a circular area assuming a
rigid baffle boundary condition next to the radiating
region. For this configuration, radiation impedance
can be calculated analytically. Our results matched
well with the analytical solution proving the validity of

From above figures it is clear that membrane

impedance in water is not simply equal to the addition
of the two impedances. Due to the water loading antiresonance of the membrane shifts to a lower
frequency.

2003 IEEE ULTRASONICS SYMPOSIUM-471

C. Immersion FEM calculation (many cells)

In a CMUT transducer, many of the membranes
operate in parallel and resulting radiation impedance is
real. For this case we constructed a 3D FEM model.
By using the similar approach used in the previous
subsection, our calculations showed that addition of
the membrane impedance and the radiation impedance
did not provide an accurate estimate of the membrane
impedance in water. In addition, we also incorporated
electrostatic elements into the 3D model and compared
transmit pressure to that of calculated using equivalent
circuit model.

width dimension and it is infinitely long in the length

dimension. Fig. 10 shows the output pressure. The
bandwidth of the device is 12.5 MHz in transmit.

Figure 10. Transmit pressure of a 1D array.

III. CONCLUSION

Figure 8. Transmit pressure calculated on the

transducers surface.
Above figure compares calculated transmit
pressures for a transducer. The transmit bandwidth is
determined by the anti-resonance which is not
predicted correctly using the equivalent circuit model.
Another interesting result we obtained using the 3D
model is that half electrode coverage results in
optimum transmit bandwidth as depicted in Fig. 9.

Figure 6. Transmit pressure. The pressure is

calculated away from the transducer surface.
D. FEM of a 1D element
We have also created an FEM model for a 1D array
element. The element has 7 hexagonal cells in the

The transmit bandwidth of a CMUT transducer is

determined by the anti-resonance of the membrane.
Equivalent circuit model cannot predict the antiresonance correctly for immersion operation. In
immersion the anti-resonance frequency depends on
the electrode coverage on the membrane. Half
electrode coverage results in maximum bandwidth
since it shifts the anti-resonance to a higher frequency.
REFERENCES
[1] W.P. Mason, Electromechanical Transducers
and Wave Filters, D. Van Nostrand Company
Inc., London, 1948.
[2] F.V. Hunt, Electroacoustics; the analysis of
transduction, and its historical background,
Cambridge: Harvard University Press, 2nd ed.
1982.
[3] I. Ladabaum, Xuecheng Jin, H. T. Soh, A. Atalar,
and B. T. Khuri-Yakub, Surface micromachined
capacitive ultrasonic transducers, IEEE Trans.
Ultrason., Ferroelect., Freq. Contr., vol. 45, pp.
678-690, May 1998.
[4] Butrus T. Khuri-Yakub et al., Silicon
Micromachined Ultrasonic Transducers, Jpn. J.
Appl. Phys. 39, 2883-2887 (May 2000).
[5] Yaralioglu et al., Calculation and measurement
of electromechanical coupling coefficient of
CMUT transducers, IEEE Trans. Ultrason.,
Ferroelect., Freq. Contr., vol. 50, pp. 449-456,
April 2003.

2003 IEEE ULTRASONICS SYMPOSIUM-472