Sie sind auf Seite 1von 9

IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 3, NO.

5, OCTOBER 2009 277

Studies of MEMS Acoustic Sensors as


Implantable Microphones for Totally
Implantable Hearing-Aid Systems
Wen H. Ko, Life Fellow, IEEE, Rui Zhang, Ping Huang, Jun Guo, Xuesong Ye, Darrin J. Young, and
Cliff A. Megerian

Abstract—There is a need for high-quality implantable micro- tations, such as ear canal irritations, distortion, and occasional
phones for existing semiimplantable middle-ear hearing systems ringing, and social problems with being perceived as handi-
and cochlear prosthesis to make them totally implantable, thus capped, have deprived many patients of clear hearing. Partially
overcoming discomfort, inconvenience, and social stigma. This
paper summarizes and compares the results of an in-vitro study implantable cochlear and middle ear hearing-aid systems can
on three design approaches and the feasibility of using micro- enable those with severe hearing loss to gain improved hearing
electromechanical system acoustic sensors as implantable micro- and speech function. However, the external microphone and
phones to convert the umbo vibration directly into a high-quality electronics create concerns about reliability, inconvenience, and
sound signal. The requirements of sensors were selected including social stigma. It is, therefore, highly desirable to develop totally
the ability to withstand large body shocks or sudden changes
of air pressure. Umbo vibration characteristics were extracted implantable hearing-aid systems with implantable microphones
from literature and laboratory measurement data. A piezoelectric [2], [3]. Several approaches of implantable microphones coupled
vibration source was built and calibrated to simulate the umbo to middle ear bones have been reported. Piezoelectric material
vibration. Two laboratory models of the acoustic sensor were can be used to sense the malleus vibration [4]. However, the
studied. The model-A device, using electrets-microphone as the material is stiff and difficult to maintain precise contact with
sensor, was designed and tested in the laboratory and on temporal
bones. The results verify that the laboratory measurement is umbo. An optic-fiber approach is complex; it consumes sizable
consistent with the temporal bone characterization and achieves a power and may temporarily lose signals [5]. These approaches
near flat frequency response with a minimum detectable signal of all suffer from performance degradation when large shocks or
a 65-dB sound-pressure-level (SPL) at 1 kHz. The model-B sensor sudden changes of air pressure occur. A magnetic sensor attached
was then designed to increase the sensitivity and provide an easy
on malleus head was tried with encouraging results [6]. However,
mounting on umbo. The model-B device can detect 40-dB SPL
sound in the 1–2 kHz region, with 100-Hz channel bandwidth. the loading effect is large, and magnetic material is MRI incom-
The results of model-A and model-B displacement sensors and patible. The subcutaneous microphone is being evaluated [7],
the acceleration sensor are summarized and compared. A prelim- [8]; the noise from chewing and body movement as well as the
inary design of the implantable displacement sensor for totally tissue growth and sensitivity degradation with time are concerns.
implantable hearing-aid systems is also presented.
Our research team has studied three microelectromechanical-
Index Terms—Displacement sensor, implantable microphone, system (MEMS) acoustic sensors based on accelerometer and
implantable sensor, microelectromechanical-system (MEMS) displacement sensors [9]–[11]. The design and results of micro
acoustic sensors, totally implantable hearing systems, vibration
sensor.
accelerometer approaches were reported [9]. The sensitivity at
low frequency, mass loading, weight, and the size are challenges.
This paper presents: 1) introduction and implant microphone re-
I. INTRODUCTION quirements; 2) laboratory simulation unit; 3) design and evalua-
tion of Model-A displacement sensor; 4) design and evaluation of
T is estimated one out of ten Americans suffers from conduc-
I tive as well as sensorineural hearing loss [1]. Conventional
hearing aids can offer moderate rehabilitation, inherent limi-
Model-B displacement sensors; 5)comparison of three acoustic
sensors and future work; and 6) the conclusion.
The concept of using a displacement sensor with an inter-
face circuit as an implantable microphone in totally implantable
Manuscript received September 29, 2008; revised February 10, 2009 and May cochlear hearing-aid systems is shown in Fig. 1.
19, 2009; accepted August 17, 2009. Current version published September 25,
2009. This work was supported part by Grant 5R21-DC-006850 of the National The displacement sensor is located inside the middle ear
Institute of Health, USA. This paper was recommended by Associate Editor cavity coupled to or mounted on the umbo. The sensor output
Eugenio Culurciello. can be used as a microphone output for various implantable
W. H. Ko, R. Zhang, P. Huang, J. Guo, X. Ye, and D. J. Young are with
the Department of Electrical Engineering and Computer Science, Case School hearing-aid systems, including the totally implantable cochlear
of Engineering, Case Western Reserve University, Cleveland, OH 10900 prosthesis system or totally implantable middle ear hearing
USA (e-mail: whk@cwru.edu; rxz33@cwru.edu; Ping.Huang@case.edu; systems, where the stimulator output may be applied to the
Jun.Guo@cwru.edu; yexs@mail.bme.zju.edu.cn; darrin.young@utah.edu).
C. A. Megerian is with the Otolaryngology Department, Medical School, stapes or the oval round window.
Case Western Reserve University, Cleveland, OH 10900 USA (e-mail: Cliff.
Megerian@UHhospital.org). A. Requirements of the Implantable Acoustic Sensor
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. The requirements of the implantable acoustic sensor were
Digital Object Identifier 10.1109/TBCAS.2009.2032267 set to include 1) and 2), for conventional microphones, and 3),
1932-4545/$26.00 © 2009 IEEE
278 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 3, NO. 5, OCTOBER 2009

Fig. 2. Umbo displacement characteristics and driving voltage needed for the
Fig. 1. Totally implantable cochlear hearing-aid system block diagram and lo- PZT simulator.
cation of the implantable displacement sensor.

4), and 5) for implantable microphones coupled to middle ear


bones. They are:
1) nearly flat frequency response from 250 to 8 kHz;
2) input sound range from 40 to 100 dB SPL;
3) small size and low power consumption;
4) less than 20 mg mass and force loadings on umbo to limit
the loading effect on the umbo vibration; [12];
5) withstand large low-frequency displacements of umbo
while sudden air pressure changes or large body shocks Fig. 3. Principle of model—A displacement sensor.
occur. Under these situations, the umbo displacement
may be up to 10–100 m, which is times that of
the umbo amplitude responding to a small input sound input sound levels. The simulation unit was calibrated with
stimulation. Therefore, especially designed sensor and a Laser–Doppler-Vibrometer to exhibit adequate linearity
coupling techniques need to be developed to accommo- between the driving voltage and vibration amplitude, flat fre-
date this large input range. quency response, and large loading capability. The PZT driving
From the requirement 3), the microelectromechanical-system voltages needed to simulate umbo vibrations at 80-dB and
(MEMS) capacitive sensors are selected. Both the accelerom- 100-dB SPL input sound levels, as functions of frequency, are
eter and displacement sensor approaches are pursued. The re- given in Fig. 2 as dotted lines, using the right-side axis scale.
quirements 1)–5) represent significant new engineering chal- For inputs lower than 60-dB SPL the driving voltages are in the
lenges that call for feasibility and prototype design studies be- microvolt range, making the measurement inaccurate.
fore device implementation. Laboratory model-A displacement sensors were measured
with the simulation unit. Characterization on several temporal
II. LABORATORY SIMULATION UNIT OF UMBO VIBRATION bones was performed to verify that the measurement results
A laboratory simulation unit for umbo vibration charac- with the simulation unit were consistent with the temporal bone
teristics measurement was assembled for the development results, and the simulation unit can generate vibrations which
of acoustic sensors as implantable microphones. The umbo accurately emulate umbo vibrations with specific input sound
vibration characteristics with an input sound between 60 dB to levels. This unit is used to evaluate the design and problems of
100 dB SPL were measured and studied from temporal bones acoustic sensors.
characterizations [9], [13]–[17]. The umbo displacement (vi-
bration) amplitude at a given frequency is linearly related to the III. MODEL-A DISPLACEMENT SENSOR
input sound pressure level in dB SPL. The umbo displacement This Model-A sensor is aimed to study the feasibility and
amplitudes versus frequency characteristics with 80-dB and to develop the coupler that can satisfy the requirements 4) and
100-dB SPL input sound levels are shown as solid curves in 5). The principle of the Model-A displacement sensor with a
Fig. 2. coupling spring is shown in Fig. 3. The sensor is mounted on
A piezoelectric vibration source (PZT) [#P-810.10 from the temporal bone structure near the umbo and coupled to umbo
PI Co.] driven by an oscillator with accurately controlled through a spring. The sensor output is fed to an interface circuit
output frequency and voltage was assembled as the umbo to convert it into an electrical signal and further amplified as a
simulation unit to emulate umbo vibrations under various microphone output. The key component is the coupling spring.
KO et al.: STUDIES OF MEMS ACOUSTIC SENSORS 279

Fig. 4. Measurement setup for the spring transmission coefficient.

Fig. 5. Model-A characteristics measured with the simulation unit.


A. Coupling Spring
In order to satisfy the requirement 4) and 5), a specially
B. Results Measured With the Simulation Unit
designed spring with proper weight and stiffness is used to
couple the sensor to umbo. The spring has to absorb large The Model-A sensor used a commercial electrets capacitive
low-frequency displacements, in the order of 100 m, without microphone with a built-in capacitance-to-voltage converter
overloading umbo, while transmitting the sound signal, with circuit [WM-65A103,with 50-dB sensitivity and 55-dB
nanometer amplitude, to the sensor efficiently. When the spring signal-to-noise ratio (SNR)] as the displacement sensor in
is compressed or extended up to 100 m, the loading to the umbo Fig. 3. The sensor was coupled through the spring to the PZT
should be less than 20 mg which is equivalent to 196 N force. vibration source replacing the umbo in Fig. 3. The interface
This would define a range of mass and stiffness of the spring. At circuit is a low-noise amplifier which exhibits a gain of 1000,
the same time, the vibration signal that needs to be transmitted flat frequency response from 250 Hz to 8000 Hz, and an input
to the sensor with small attenuation; that is, the transmission equivalent noise floor of 2.6 nV Hz. The frequency response
coefficient, needs to be close to 100%. Thus, the specific density of each Model-A sensor was measured several times with
and sound transmission velocity of the spring material need to the input amplitudes in the range from 60 dB to 100 dB SPL
be considered. A 100- m diameter coil spring, made of tungsten at 1 kHz. The results were recorded with a dynamic signal
wire, weighs 15 mg (the commercial sensor weighs about 5 mg) analyzer and an oscilloscope. The outputs at all frequencies are
and has a self-resonant frequency of 249 Hz and a stiffness of linearly proportional to the input SPL level. Fig. 5 shows the
10 N/m will be used for Model-A devices [10]. typical frequency characteristics and noise floor of the Mode-A
In order to design the spring properly, the vibration transmis- sensor at 100-dB SPL input.
sion coefficient of the spring and the three modes of sound trans-
mission were studied. The three modes are: 1) the acoustic cou- C. Results From Temporal Bone Experiments
pling through the air; 2) the vibration movement of the spring; The Model-A sensors were measured on temporal bones.
and 3) the sound-wave propagation through the spring wire. Fig. 6(a) and (b) shows the setup. The calibrated 60-dB and
Fig. 4 shows the experimental setup. 95-dB SPL sound signals were applied to the ear canal through
The spring is connected between the PZT vibration source the ER-2 Instrument. The Model-A sensor was coupled to the
and a stretched aluminum foil modeling the sensor diaphragm. umbo through the coupling spring. The amplifier output was
A Laser-Doppler–Vibrometer is used to measure the vibration measured and recorded with a dynamic signal analyzer. The
amplitude of the diaphragm over the frequency range. The three measured sensor characteristics and noise floor of the Model-A
modes of transmission were measured by different arrange- sensor are shown in Fig. 7. The minimum detectable SPL at
ments of the springs. Four springs were made from the same 1 kHz is 38 dB for 1-Hz channel bandwidth, and 58 dB for
batch of tungsten wire with the same coil diameter and length the 100-Hz channel bandwidth. This is limited by the sensor
but with different stiffness. Measured results at 800 Hz were sensitivity and the amplifier noise.
recorded. The transmission coefficient increased slowly from By comparing the results measured using the simulation unit
68% to 78% when spring stiffness increased from 5.1 N/m and measured on temporal bones, it is verified that the sim-
to 205 N/m. For the spring with 5.1-N/m stiffness, the total ulation unit can generate vibrations which accurately emulate
transmission coefficient at 800 Hz is 67%. The Model-A sensor umbo vibration with specific input sound levels [11].
used a spring with 10-N/m stiffness and 67% transmission Although, the Model-A sensor showed that it is feasible to
coefficient. The transmission coefficient can be increased to satisfy all of the requirements by using the designed coupling
75% if the end of the spring is enlarged to match the mechanical spring, the sensor unit needs to be mounted on a nearby bony
impedances of the spring and the diaphragm better [11], [17]. structure. The alignment of the spring to the umbo to have
This is less than 100% desired and other means of raising the proper initial spring compression, to maintain contact with
transmission coefficient would be studied. umbo under all conditions, will be delicate. The engineering
280 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 3, NO. 5, OCTOBER 2009

Fig. 8. Structure and principle of the Module-B sensor.

IV. MODEL-B DISPLACEMENT SENSOR


The Model-B sensor incorporates a novel approach to elimi-
nate the coupling spring by mounting the sensor unit (sensor, in-
terface I.C., and package) on the umbo; and an innovative design
so that the microsensor responds differently to large low-fre-
quency shocks than to small high-frequency sounds. The de-
sign is based on the reasoning that large body shocks or large
air pressure changes, when transmitted to the ear, are large in-
puts with much lower frequency than the voice or sound signals.
The sensor unit has two parts: one responds to low-frequency
shocks, and the other responds to the sound frequency band. The
Model-B sensor can be mounted on umbo directly and can sat-
isfy all requirements listed in Section I.
The high-performance silicon chip capacitive strain sensors
with 0.1 micro-strain sensitivity at 10-kHz bandwidth designed
Fig. 6. Measurement setups for temporal bone experiments. (a) Block diagram. previously in our laboratory [19] were modified as a high sen-
(b) Photograph of the experiment setup. sitivity MEMS capacitive displacement sensor. The schematic
structure of the Model-B sensor is shown in Fig. 8. The capac-
itive sensor diaphragm and the mounting stub have a mass .
The diaphragm is connected to a set of springs S, with negli-
gible mass and a total stiffness . The other end of the spring is
connected to a base with a mass , and . The sensor
unit has a total mass of less than 20 mg. The ca-
pacitance between the diaphragm and the base represents
the displacement sensor. For low-frequency (e.g., 200-Hz) vi-
brations, the sensor mass and would move together with
the input displacement, resulting in a small sensor output. For
large amplitude low-frequency shocks, the whole sensor unit
would move together with the umbo. The sensor may
be saturated but not damaged. The total loading on the umbo
would be limited to the 20-mg mass. When the input is in the
sound frequency (500 Hz–8 kHz) range, the base would be
nearly standstill, and the diaphragm would respond to the
input, converting vibrations into capacitance changes. The in-
terface amplifier circuit then converts the capacitance changes
into electrical signals, resulting in the acoustic sound voltage
output.

Fig. 7. Results of the Model-A sensor measured on temporal bones. A. Analysis of the Model-B Sensor Principle
As shown in Fig. 8, the Model-B sensor receives the driving
displacement from umbo through the mounting stub to the di-
design of the mounting structure and the implant procedure aphragm . The base is coupled to through a set of
would be difficult. Other approaches to resolve these problems springs with a combined spring constant . The capacitance
were studied. The Model-B sensor was developed. C between the and base is the acoustic sensor. The initial
KO et al.: STUDIES OF MEMS ACOUSTIC SENSORS 281

positions , and included the effect of earth gravitational


force are assumed to be constant.
When the input displacement is applied, assume
that the displacement of is ; the length change of the spring
is ; the total damping is ; and , , and are functions of
frequency and time. The system response can be described by
the following force equations:

(1)

(2)

(3)

The solution of the Laplace Transform of (3) for is

(4)

The system characteristics are determined by two factors: 1) the


critical frequency and 2) system damping ratio
where

(5)

For all hearing applications, the phase variation is not impor- Fig. 9. Response of the sensor to input displacement. (a) “x=A .” “y=A ”
tant. The Model-B sensor is a heavily damped device. The and responses to input A versus frequency F. (b) C response to input A versus
amplitudes responding to the input depend on the value of frequency F.
and are plotted in Fig. 9(a). Considering the best damped re-
sponse, where is equal to , as approaches zero,
approaches one, and approaches zero. The and B. Model-B Displacement Sensor
responses for (solid line) and
The Model-B sensor module structure and the attachment
(dotted line) cases, are indicated in Fig. 9(a). The capacitance
method to the umbo when implanted, as well as the sensor struc-
response to (and ) is shown in Fig. 9(b).
ture, are schematically illustrated in Fig. 10.
The sensor output is the capacitor variation. The capacitance
The design of silicon chip Model-B sensor is outlined. The
of a parallel plate capacitor, neglecting the fringe effect, is
Model-B displacement sensor has a desired mass of
20 mg, and 200 Hz. According to (6), the spring
(6) constant should be 31.2 N/m. There are four folded-cantilever
beams supporting the diaphragm, acting as the springs in the
where , , , and denote the capacitance, electrode area, system, the value of each beam should be 7.8 N/m. For a given
dielectric constant, and separation between the electrodes, re- spring constant, the dimension of the cantilever-beams springs
spectively. is the initial capacitance when . The can be calculated from
amplitude of capacitance change due to input and the
change of spring length is
(8)

(7)
where , , and are the thickness, length, and width of the
spring; and E is the Young’s modulus. The calculated silicon
where is the initial length of the spring. The length change cantilever beam dimension, with 200-Hz critical frequency
is a function of spring constant and . Assuming the umbo and 20- m thickness, should be 80 m in width and 1454 m
has large driving capability, the input displacement would in length. However, due to compromises on silicon chip size
be equal to umbo vibration, and the sensor would truly be a and layout constraints from the shared fabrication processes
displacement sensor. The frequency on the silicon-on-insulator (SOI) wafer, the fabricated springs
can be designed to be less than 200 Hz. As frequency increases are shorter than that designed. The length of springs in the
beyond the sound frequency band, the force loading of “m” on prototype sensor is 850 m. For a spring with 20 mg,
umbo increases with . The increased loading effect would m, m, and m, the calculated
decrease the responses and at high frequencies. Beyond is 445 Hz. The prototype Model-B sensor weighs about 25 mg
a frequency , the sensor sensitivity would drop off. and the SOI wafer used also has 2- m silicon dioxide on the top
282 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 3, NO. 5, OCTOBER 2009

Fig. 11. Test setup of the prototype Model-B sensor.

of the amplifier represents the output of an implantable micro-


phone. The sensor unit is an integrated module; it includes the
sensor, the mounting, and the interface circuit. The sensitivity
of the sensor module is limited by the total noise of the module.
The capacitive sensor itself has very low noise and does not
consume any real average power. In order to satisfy the system
sensitivity, a low-noise interface circuit is needed. The circuit
noise is usually specified by the input-referred noise power spec-
tral density. For the low-noise interface circuit that is needed,
the interface electronics designed for the high-performance ca-
pacitive strain sensor by this group were used [19]. The inter-
face circuit has a resolution, in capacitance change, of 250 aF
over a bandwidth of dc to 10 kHz, and an input-referred voltage
Fig. 10. Model-B structure and attachment methode on umbo. (a) Overall ar- noise power spectral density of 5 Hz. The architecture
rangement. (b). Model-B sensor.
of the interface electronics is shown in Fig. 12. The prototype
Model-B sensor and a fixed capacitor are used as the differen-
tial input capacitances in the “MEMS sensor” block, in Fig. 12.
of the 20- m device layer. If the difference in Young’s modulus They are driven by a 1-MHz clock signal with 3-V amplitude
of silicon and silicon dioxide is neglected, the recalculated and are interfaced by a differential charge amplifier, which con-
is 459 Hz. The measured on the prototype Model-B sensor verts the sensor capacitance change to an output voltage. The
is about 480 Hz. This 4.4% increase in may contribute to high clock frequency is chosen to modulate the sensor infor-
the normal dimensional variation in fabrication processes. mation away from the low-frequency noise, such as the
The Model-B sensor was fabricated on a 2 2-mm silicon noise of the amplifier, a critical means to achieve high sensi-
chip. A printed-circuit board (PCB) substrate is used to simu- tivity. An input common-mode feedback (ICMFB) circuit and
late the of the interface circuit; the total weight of the sensor an output common-mode feedback (OCMFB) circuit are incor-
unit is 25 mg. The sensors were measured by using the simu- porated with the charge amplifier to minimize its common-mode
lation unit and experimental setup described in Sections II and shift caused by the driving clock; hence, suppressing any offset
III. The PZT vibration amplitude is set at a constant sound pres- signal due to the parasitic capacitance mismatch and drift over
sure level, corresponding to 97-dB SPL sound input at 1 kHz. time. The charge amplifier output is then mixed by the same
The sensor capacitance output is fed through an interface cir- clock signal and low-pass filtered to obtain an output voltage
cuit to convert capacitance into voltage and then measured by , which represents the desired sensor information. The in-
a signal spectrum analyzer. The measurement setup is illus- terface circuit is fabricated by MOSIS using 1.5- m technology,
trated in Fig. 11. The sensor is mounted on a printed-circuit and it consumes 1.5 mA at 3 V [20]. The power consumption can
board (PCB) substrate and connected to the interface circuit also be reduced by choosing smaller feature size MOSIS technology
mounted on the substrate. The connections were made through in future designs.
1-mil diameter flexible gold wires to minimize the strain inter-
ference to the sensor diaphragm and the spring. The sensor unit D. Results of the Model-B Sensor
is attached on the solid beam coupler by a thin layer of commer- The Model-B sensor was tested with a simulation unit at the
cial hard adhesive. set amplitude equivalent to a sound input of 97-dB SPL at 1
kHz. The measured voltage outputs as well as the system noise
C. Interface Circuit
as functions of frequency are shown in Fig. 13. This proto-
The sensor output is the capacitance change . It needs an type sensor has a critical frequency around 480 Hz, which is
interface circuit functioning as a capacitance to a voltage con- higher than the desired value of 200 Hz for reasons, explained in
verter to convert into voltage and an amplifier. The output Section IV-B. The displacement of the substrate of another
KO et al.: STUDIES OF MEMS ACOUSTIC SENSORS 283

Fig. 12. Capacitive sensor interface electronics architecture.

In order to show that can be designed according to (6),


the device mass of the same Model-B sensor was changed from
25 mg to 125 mg by adding a weight on the substrate. The mea-
sured results on the 125-mg sensor showed that the cutoff
frequency was shifted from 480 Hz to less than 200 Hz. Both of
these two Model-B sensors with different masses have a near flat
frequency response from 800 to 8 kHz with a variation of less
than 6 dB. As shown in Fig. 14, for the Model-B sensor with
30-mg mass, the signal-to-noise voltage ratio (signal analyzer
bandwidth is 50 Hz) is 74 dB, at 1 kHz and 97-dB SPL input.
This indicates that an SNR of 71 dB can be achieved if the in-
terested signal bandwidth is 100 Hz. Therefore, a minimum de-
tectable input sound level (with 6-dB SNR) of 32-dB SPL can
be achieved around 1 kHz with 100-Hz channel bandwidth.
V. THREE APPROACHES OF MEMS ACOUSTIC SENSORS AND
FUTURE WORK
As mentioned in the Introduction, our team had studied three
Fig. 13. Test result of the prototype sensor with 30-mg mass. approaches for the microacoustic sensor for implantable micro-
phones aimed to meet the same requirements for implantable
hearing systems. They are the accelerometer described in pre-
vious works [9], [12], the spring coupled Model-A displacement
sensor, and the direct mounting Model-B displacement sensor
described in this paper. All three approaches can be designed to
satisfy the requirements set in Section I, and their power con-
sumptions are about 4.5 mW. The power is nearly all consumed
by the interface circuit; the capacitive sensors do not consume
any real average power except for the small series resistance of
the capacitive sensor. With advanced IC processes, the power
consumption can be reduced. Each of the three sensors and their
interface circuits can be integrated on a millimeter-size silicon
chip that weighs 1–2 mg. Both the accelerometer sensor and the
Model-B sensor are to be attached on the umbo adhesives. The
minimum detectable sound pressure levels versus frequency of
these three approaches are shown in Fig. 15, where all hearing
systems are assumed to have the channel bandwidth of 100 Hz
and an SNR of 6 dB. The performance of all approaches can be
improved with next generation of designs. However, the shapes
of these curves are the characteristics of each approach. The
Fig. 14. Measured displacement x of the mass M. lower sensitivity of the accelerometer approach in the low-fre-
quency branch may be compensated by a special interface cir-
cuit; the Model-A approach needs higher sensitivity and a deli-
sensor unit with the same mass under the same conditions was cate structure to attach the sensor unit to the bone. The Model-B
measured and is shown in Fig. 14, where the response near the sensor has high sensitivity and near flat frequency response from
critical frequency and the decrease in amplitude beyond 800 Hz up to 10 kHz. Model-B is a promising approach for fur-
are clearly illustrated as predicted. ther investigation.
284 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 3, NO. 5, OCTOBER 2009

systems was studied and established. The capacitive accelerom-


eter and displacement sensors approaches may all meet the
requirements of implantable microphones and can be used
to convert the umbo vibration into electrical sound signals
with high fidelity when redesigned with improved technology
and the information gained from this study. The laboratory
simulation unit was assembled and confirmed that it can ac-
curately emulate the umbo vibrations responding to sound
inputs. The Model-A sensor was evaluated on the simulation
unit and temporal bones. The results indicated that 20- to
30-dB higher sensitivity or lower noise is needed to meet the
40-dB operation requirements; and the accurate mounting of
the sensor package in the middle ear cavity is difficult. The
Model-B sensor was designed to overcome these problems.
A novel design was used to separate the large low-frequency
movements from the nanometer amplitude sound frequency
Fig. 15. Minimum detectable signal levels of three acoustic sensors studied.
vibrations. The Model-B sensor can be mounted directly on
, and can respond differently to large shocks than regular
sound signals. This Model-B sensor with a low noise interface
circuit can detect 40-dB SPL sound in a 1- to 2-kHz region
with 100-Hz channel bandwidth. A comparison of achieved
performance of three acoustic sensor approaches studied by the
team is summarized. A design example of the Model-B sensor
as an implantable microphone is given for future studies.
REFERENCES
[1] Better Hearing Institute, Prevalence of hearing loss & demographics.
[Online]. Available: http://www.betterhearing.org/research/fac-
toids.cfm
[2] W. Ko, A. Maniglia, and R. Zhang, “A preliminary son the implantable
middle ear hearing aid,” in Proc. IEEE 9th Annu. Conf. Eng. Med. Biol.,
Fig. 16. Cross-sectional view of the preliminary design of the sensor. 1987, p. 1890.
[3] A. Maniglia, W. Ko, and M. Rosenbaum, “A contactless electromag-
netic implantable middle ear device of the ossicular stimulating type,”
Ear Nose Throat J., vol. 73, no. 2, p. 78, 1994.
In order to develop a practical implantable microphone for [4] H. Zenner, M. Maassen, R. Lehner, J. Baumann, and H. Leysieffer, “An
implant evaluation, further studies are needed. They would in- implantable hearing aid for inner ear hearing loss: Short-term implan-
clude: 1) package of the device to have long implant lifetime; tation of microphone and transducer,” Otolaryng. Head Neck Surgery,
vol. 45, no. 10, pp. 872–880, Oct. 1997.
2) the lead connection from the sensor/microphone to the main [5] A. Vujanic, R. Pavelka, N. Adamovic, C. Kment, S. Mitic, W. Brenner,
body of the implant system for power input and signal output; and G. Popovic, “Development of a totally implantable hearing aid,” in
3) the proper adhesive materials, such as the bone cement [21]; Proc. 23rd Int. Conf. Microelectronics, Yugoslavia, May 2002, vol. 1,
NIŠ.
and d) the system reliability in the implanted environment. With [6] A. Maniglia, H. Abbass, T. Azar, M. Kane, P. Amantia, Garverick, W.
the information gained from this study, the requirements of a Ko, W. Frenz, and T. Falk, “The middle ear bioelectronic microphone
practical device can be specified; serious effort may start to de- for a totally implantable cochlear hearing device for profound and total
hearing loss,” Amer J. Otol., vol. 20, pp. 602–611, 1999.
velop an implantable acoustic sensor unit for totally implantable [7] D. Chen, D. Backous, M. Arriaga, R. Garvin, D. Kobylek, T. Littman,
middle ear or cochlear hearing systems. As an example, a pre- S. Walgre, and L. David, “A totally implantable middle ear device for
liminary design of the Model-B sensor is proposed based on the sensorineural hearing loss,” Otolaryng. Head Neck Surgery, vol. 131,
no. 6, pp. 904–916, 2004.
results presented in this paper and is shown in Fig. 16. The basic [8] A. Jenkins, J. Atkins, H. Drew, M. Hoffer, B. Balough, J. Arigo, G.
principle is the separation of responses to low-frequency shock Alexiades, and W. Garvis, “U.S. Phase I preliminary results of use of
and to high-frequency sound signals. The additions are: 1) the the Otologics MET fully-implantable ossicular stimulator,” Otolaryng.
Head Neck Surgery, vol. 137, pp. 206–212, 2007.
sensor module is sealed in the silicon chamber by a top plate [9] D. Zurcher, J. Young, M. Semaan, C. Megerian, and W. Ko, “MEMS
that also serves as the spring, 2) the sensor unit is covered with middle ear acoustic sensor for fully implantable cochlear prosthesis,”
a biocompatible silicone coating, 3) the coupling stub used to in Proc. 20th IEEE Int. Conf. MEMS, Japan, 2007, pp. 11–14.
[10] P. Huang, J. Guo, C. A. Megerian, D. Young, and W. Ko, “A laboratory
attach the sensor to umbo is integrated with the sensor during study on a capacitive displacement sensor as an implant microphone in
fabrication, and 4) stops are added to limit displacements of the totally implant cochlear hearing aid systems,” in Proc. Int. Conf. IEEE
capacitive sensor in both directions. EMBS, Lyon, France, Aug. 2007, pp. 5691–5694.
[11] W. Ko, P. Huang, J. Guo, R. Zhang, D. Young1, and C. Megerian,
“MEMS acoustic sensors for totally implantable hearing aid systems,”
in Proc. IEEE Int. Symp. Circuits and Systems, Seattle, WA, May 2008,
VI. CONCLUSION pp. 1812–1817.
[12] S. Nishihara, H. Aritomo, and R. L. Goode, “Effects of changes in mass
The feasibility of designing displacement sensors as im- on middle ear function,” Otolaryng. Head Neck Surgery, vol. 109, pp.
plantable microphones to be used in totally implantable hearing 899–910, 1993.
KO et al.: STUDIES OF MEMS ACOUSTIC SENSORS 285

[13] M. Zucher, “Development of a MEMS middle ear acoustic sensor Jun Guo received the B.S. degree in electrical engineering from Tsinghua Uni-
for a fully implantable cochlear prosthesis,” M.Sc. dissertation, Case versity, Beijing, China, in 1993, and the M.S. degree in electrical engineering
Western Reserve Univ., Cleveland, OH, 2006. and computer science from Case Western Reserve University, Cleveland, OH,
[14] U. B. Willi, “The dynamic behavior of the incudo malleolar joint and in 2000, where he received the Ph.D. degree in 2007 .
its role during the transmission of sound,” Ph.D. dissertation, Dept. His current research is oriented toward the design, fabrication, and testing of
Radiol. Magn. Resonance, Univ. Zürich, Zurich, Switzerland, 2003. microelectromechanical-system sensor and nanoelectromechanical devices for
[15] R. L. Goode, G. Ball, S. Nishihara, and K. Nakamura, “Laser Doppler a variety of applications.
Vibrometer (LDV)—A new clinical tool for the otologist,” Amer. J.
Otol., vol. 17, no. 6, pp. 813–822, Nov. 1996.
[16] R. Z. Gan, M. W. Wood, and K. J. Dormer, “Human middle ear transfer
function measured by double laser interferometry system,” Otol. Neu- Xuesong Ye received the Ph.D. degree in biomedical
rotol., vol. 25, no. 4, pp. 423–435, Jul. 2004. engineering from Zhejiang University, Hang Zhou,
[17] S. Nishihara, H. Aritomo, and R. L. Goode, “Effects of changes in mass China, in 1997.
on middle ear function,” Otolaryng. Head-Neck Surgery, vol. 109, pp. He was an Assistant Professor, Associate Pro-
899–910, 1993. fessor, and Professor with the Department of
[18] P. Huang, “A laboratory study of a capacitive displacement sensor as an Biomedical Engineering at Zhejiang University
implant microphone,” M.Sc. dissertation, Case Western Reserve Univ., since 1997. From 2006 to 2008, he was a Senior Re-
Cleveland, OH, 2007. search Associate with the Department of Electrical
[19] W. H. Ko, D. Young, and J. Guo et al., “A high-performance MEMS Engineering and Computer Science, Case Western
capacitive strain sensing system,” Sens. Actuators A: Phys., vol. 133-2, Reserve University, Cleveland, OH. His research is
pp. 272–277, 2007. focused on neural-network computing, neural chip
[20] M. Suster, N. Chaimanonart, J. Guo, W. H. Ko, and D. Young, “Re- fabrication, and neural signal detecting and processing for brain computer
mote-powered high-performance strain sensing microsystem,” in Proc. interface based on microelectromechanical-system sensor and complementary
IEEE Techn. Dig., Int. Conf. MEMS, Miami, FL, 2005, pp. 255–258. metal–oxide semiconductor integrated-circuit design.
[21] A. J. Maniglia, N. Nakabayashi, M. M. Paparella, and J. W. Werning,
“A new adhesive bonding material for the cementation of implantable
devices,” Otol. Surgery. Amer. J. Otol, vol. 18, pp. 322–327, 1997.
Darrin J. Young received the B.S. degree (Hons.),
M.S., and Ph.D. degrees in electrical engineering and
computer science from the Department of Electrical
Engineering and Computer Sciences at the Univer-
Wen H. Ko (LF’90) received the B.S. degree in elec- sity of California, Berkeley, in 1991, 1993, and 1999,
trical engineering from Amoy (Xiamen) University respectively.
of China, Xiamen, China, in l946, and the M.S. and His doctoral dissertation focused on micro
Ph.D. degrees in electrical engineering from Case electromechanical devices design and fabrication
Institute of Technology, Cleveland, OH, in l956 and technologies for radio-frequency analog signal
l959, respectively. processing. From 1991 to 1993, he was with
He has been an Assistant, an Associate, and a Full Hewlett-Packard Laboratories, Palo Alto, CA,
Professor of Electrical Engineering and Biomedical designing a shared memory system for a digital signal processor-based
Engineering at Case Western Reserve University multiprocessor architecture. Between 1997 and 1998, he was with Lawrence
(CWRU), Cleveland, in l959, l962, and l967, respec- Livermore National Laboratory, working on the design and fabrication of
tively. He became a Professor Emeritus in Electrical 3-D radio-frequency microelectromechanical-systems (MEMS) inductors for
Engineering at CWRU in 1993. wireless communications.
Dr. Ko is a fellow of the American Institute of Medical and Biological Engi- Dr. Young joined the Department of Electrical Engineering and Computer
neering. He is active in research on microelectromechanical systems, biomed- Science, Case School of Engineering, Case Western Reserve University, Cleve-
ical implants, as well as micropackaging. land, in 1999, where he is currently an Associate Professor. His research in-
terests include MEMS and nanoelectromechanical devices design, fabrication,
and integrated analog circuits design for sensing, communication, biomedical
implant, and general industrial applications.
Rui Zhang received the B.Sc. degree in mechanical
and electrical engineering from Xiamen University of
China, Xiamen, China, in 2003 and is currently pur-
suing the Ph.D. degree in electrical engineering and Cliff A. Megerian received the M.D. degree from the
computer science from Case Western Reserve Uni- University of Michigan Medical School, Ann Arbor,
versity, Cleveland, OH. in 1988
His current research includes the design and fab- He served his residency at Otolaryngology and
rication of microelectromechanical-systems sensors Head and Neck at University Hospital, Cleveland,
and actuators, and complementary metal–oxide semi- OH. He was the first Clinical Fellow in otology
conductor interface circuit design. and neurotology at the Massachusetts Eye and Ear
Infirmary and Harvard Medical School. Currently,
he is a Professor and Vice Chairman of otolaryn-
gology—head and neck surgery, and Professor
of neurological surgery with the Otolaryngology
Ping Huang received the M.S. degree in electrical engineering from Case Department, Medical School, Case Western University, Cleveland. He is also
Western Reserve University, Cleveland, OH, in 2007, in the field of microelec- the Medical Director of the Adult and Pediatric Cochlear Implant Program and
tromechanical-systems (MEMS) technology and sensor design. Co-Director of the Skull Base Surgery Program. His clinical interests include
His interest is in MEMS devices and micromachining technology. otology, neurotology, cochlear implantation, and skull-base surgery.

Das könnte Ihnen auch gefallen