IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 41, NO.

12, DECEMBER 2006

2965

A 32-Site 4-Channel High-Density Electrode

Array for a Cochlear Prosthesis
Pamela T. Bhatti, Member, IEEE, and Kensall D. Wise, Fellow, IEEE

AbstractThis paper describes a 32-site 4-channel high-density intracochlear electrode array. Combining MEMS-based
processing technology with active circuitry, the thin-film device
realizes a stimulating site density three times that of commercial
systems, delivering 500 A biphasic current levels with 8-bit
resolution and less than 1% charge mismatch. The minimum
pulsewidth is 4 s, and the per-channel power dissipation is
2.5 mW from 2.5 V. The active circuit chip has a footprint of
2.4 mm 2.9 mm. Serving as the end-effector of a cochlear prosthesis, the active array also provides site-impedance measurement
and position-sensing functions.
Index TermsBioMEMS, cochlear implant, current stimulator,
electrode impedance, neural prostheses.

I. INTRODUCTION
INCE their introduction in the 1980s, cochlear implants
have been the most successful neural prostheses, with
nearly 100 000 profoundly deaf and severely hearing impaired
users worldwide [1]. By directing current through 1622 intracochlear wire electrodes, a cochlear implant exploits the
frequency-to-place mapping in the cochlea of the inner ear
to directly stimulate the auditory nerve electrically, thereby
replacing defective hair cells that in a normal individual would
transduce sound energy into nerve impulses (Fig. 1). Although
such implants have been remarkably effective, there remain
significant variations in speech perception among patients
exhibiting similar pre-implant etiologies, as well as difficulties
in understanding tonal languages and appreciating music [2]. A
potential solution to these problems is to develop electrode arrays having significantly increased numbers of stimulating sites
so that the arrays can more easily adapt to differing patterns
of neural survival and make use of multi-polar current shaping
to activate distinct nerve populations and enhance pitch perception [3]. However, merely scaling todays hand-assembled,
wire-bundle-based electrode arrays is precluded by the size of
the cochlea, which tapers from a diameter of about 1 mm to
about 200 m over its length. This paper presents a promising
alternativea MEMS-based microfabricated thin-film array
capable of supporting an electrode density three times that
of commercial systems while remaining consistent with the
cochlear space. The technology enables the integration of position sensors on the array to help optimize in vivo placement,
minimize insertion damage, and therefore preserve a patients

Manuscript received June 23, 2006; revised August 24, 2006. This work was
supported in part by the NSF-WIMS ERC under cooperative agreement EEC9986866 and by a gift from the late Ms. Polly V. Anderson of Cupertino, CA.
The authors are with the Engineering Research Center for Wireless Integrated
MicroSystems, The University of Michigan, Ann Arbor, MI 48109-2122 USA
(e-mail: pamelat@umich.edu; wise@umich.edu).
Digital Object Identifier 10.1109/JSSC.2006.884862

Fig. 1. A contemporary cochlear electrode array composed of bundled wire

electrodes and banded electrode sites. (Drawing courtesy of Cochlear Ltd.)

residual hearing [4]. On-board circuitry for stimulus generation

and position-sensing is incorporated, realizing an active 32-site
4-channel array designed for animal studies (guinea pig). The
array offers the same features and site densities needed for
a 128-site 16-channel full-length human array and is serving
as the front-end of a cochlear prosthesis being developed at
the University of Michigans Engineering Research Center for
Wireless Integrated MicroSystems (WIMS-ERC).
Section II provides an overview of cochlear prosthesis systems and the WIMS-ERC Cochlear Prosthesis in particular.
Section III summarizes the array fabrication process and circuit integration, and in Section IV the on-board circuitry and
experimental results are described. Finally, in Section V the
work is summarized and prospects for developing a 128-site
16-channel human array are examined.
II. SYSTEM OVERVIEW: THE WIMS-ERC
COCHLEAR PROSTHESIS
A contemporary cochlear prosthesis consists of an external
module (a microphone and speech processor) and an internal
portion consisting of an implanted receiver/stimulator and
electrode array (Fig. 2). Externally, sound is captured by a
microphone worn behind the ear and is directed to a speech
processor that filters and compresses the captured signals.
At frequencies where significant acoustic energy is present,
stimulation commands specifying electrode and stimulation
level (current amplitude and phase duration) are output by
the processor and sent to the internal receiver/stimulator
via an inductively coupled forward telemetry link that also
transmits power. The implanted stimulator then translates the
commands into current pulses that are presented to sites along

0018-9200/$20.00 2006 IEEE

2966

IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 41, NO. 12, DECEMBER 2006

Fig. 2. A cochlear prosthesis microsystem, composed of a speech processor worn behind the ear, an external coil for power/data transfer, implanted
electronics, and an electrode array placed in the cochlea. (Drawing courtesy of Cochlear Ltd.)

Fig. 3. The WIMS-ERC cochlear prosthesis.

the electrode array. Stimulating site impedances, intracochlear

voltages, and diagnostic information may also be transmitted
back through a reverse telemetry link.
The WIMS Cochlear Prosthesis functionally parallels
commercial devices, but adjusts the system partitioning to
explore moving the signal processing internally (Fig. 3).
Furthermore, this implementation combines the processing-intensive circuitrya low-power microcontroller (MCU) and
DSP corewith a wireless interface and an A/D converter
in a front-end electronics package. An application-specific
integrated circuit (ASIC) resides at the base of the array for
stimulus generation. Moving toward a system-on-a-chip allows
for studying the feasibility of a fully implantable cochlear prosthesis. Such systems have been explored by others primarily
through the implementation of low-power analog circuits [5],

[6], but this is the first known effort to develop a fully integrated
wireless microsystem that includes the electrode array.
The integrated DSP core implements the Continuous
Interleaved Sampling (CIS) speech-processing algorithm
using the lowest active power consumption reported to date
(1.79 mW) [7]. Designed for biomedical and environmental
monitoring applications, the 16-bit MCU contains a 32 kB
on-chip memory and two peripheral interfaces, while providing
significant programmability and flexibility for optimizing a
patients speech processing parameters. Additionally, stimulation commands for other speech-processing algorithms
may be configured, as well as current steering and high-stimulus-rate research profiles. Controlling the wireless interface
is a bidirectional telemetry module [8]. Based on an FSK
transmission scheme, data and clock are recovered from 8-bit

BHATTI AND WISE: A 32-SITE 4-CHANNEL HIGH-DENSITY ELECTRODE ARRAY FOR A COCHLEAR PROSTHESIS

2967

Fig. 4. Cross-section of the electrode array.

Fig. 5. The 32-site high density electrode array.

Fig. 6. The circuit chip implementing the cochlear array electronics.

data packets transmitted at a 46 MHz carrier frequency. The

inductively coupled link also transmits power that is recovered
and rectified to produce multiple regulated power supplies
[9]. The wireless interface and MCU-DSP unit are housed in
a hermetically sealed glass-on-silicon package using a Au-Si
eutectic wafer-level bonding process. Signals are transferred
via polysilicon feedthroughs. Accelerated testing has shown
mean-times-to-failure of several decades [10]. Recent work
incorporating silicon getters has reduced package leak rates to
1 mTorr/year levels [11].
The electrode array directly interfaces with neural tissue, resulting in higher required voltages (520 V) than for the signal
processing electronics. Segregation of the comparatively lowvoltage MCU and DSP (1.8 V) in the electronics package allows
future scaling of the processor, while retaining a mixed-signal
high-voltage option for driving the tissue. Such a strategy requires level-shifting and buffering at the ASIC at the base of the
electrode array, but also reduces the leakage and relaxes the voltages on the 8-lead polymeric cable connecting the electronics
package to the ASIC.

multi-level metal and on-chip monolithic circuitry will be

needed for 128-site arrays for human use.
The 11-mask passive array process is compatible with integrated on-chip BiCMOS circuitry [12]; however, to meet
the area restrictions in small animals, commercially fabricated
circuitry has been used here to take advantage of smaller
industrial feature sizes. Realized in 0.5- m AMI CMOS, the
2.4 mm 2.4 mm circuit chip (Fig. 6) is flipped and ultrasonically wedge-bonded to the cochlear array via 45 gold lead tabs
formed using the array process (Fig. 7). The hybrid assembly is
then coated with medical grade silicone.

III. ARRAY FABRICATION AND CIRCUIT INTEGRATION

The array uses a boron etch-stop to form a 2.5- m-thick
silicon substrate [12] under stress-compensated silicon dioxide
and silicon nitride dielectrics (Fig. 4). A parylene overcoat can
be defined at wafer level prior to an ethylene-diamine-pyrocatechol substrate release etch that separates the array from
the wafer [13]. Sized for a shallow-insertion into the guinea
pig cochlea, the 8-mm-long substrate tapers from a width of
500 m to 200 m and supports 32 180- m-diameter iridium
oxide (IrO) stimulating sites on 250- m centers (Fig. 5). In
contrast to platinum, routinely employed in stimulating arrays, activated iridium electrodes provide a ten-fold increase in
charge-delivery; this is important for dense electrodes as well as
for increasing system efficiency [14]. A 14- m-thick boron-diffused silicon area forms the base of the array, which remains
outside the cochlea to support circuitry for current generation,
site selection, and position sensing. While single-level metal is
sufficient to connect the circuitry to 32 sites, stress-controlled

IV. THE COCHLEAR ASIC

The array communicates with the microcontroller using a
5-wire serial peripheral interface (SPI) protocol [15]. Level
shifters translate the 03 V signaling used on the input lines
into the 2.5 V levels used by the cochlear ASIC. The ground
(GND) is referenced to animal ground and is implemented with
a large electrode placed in the tissue. Sixteen-bit command
words (Fig. 8) are serially clocked into a 16-bit shift register at
a maximum rate of 10 MHz on the data line (DIN). A strobe
signal (STB) captures each command in a 16-bit latch that
directs the word to an error detection block implementing an
even parity check. Valid words are flagged by driving the data
valid (DAV) line high, whereas an invalid command keeps the
DAV line low causing the SPI port to issue a timeout to the
MCU (Fig. 9). Data is output from the ASIC on the data-out
line (DT0) and is routed to an A/D converter block.
Fig. 10 shows a block diagram of the cochlear ASIC. The
array functions in one of two modes as determined by the mode
bit. The first mode configures the array for current stimulation,
while the second is used to implement functions such as position
measurement, system tests, and electrode recording. In stimulation mode, current is output on four parallel channels, each of
which is routed to one of eight sites. Each input data word is
used to select a stimulating site (bits 1014) and output the desired current level (bits 07) on that site. Bits 1011 select from
one of four stimulators, with each stimulator driving a dedicated
channel. Bits 1214 control an output switch matrix selecting

2968

IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 41, NO. 12, DECEMBER 2006

Fig. 7. Hybrid interconnects between the microcontroller, the cochlear ASIC, and the electrode array. The left inset is a scanning electron microscope view of a
gold rivet bond.

The electrode recording function bit designation is described in

Section IV-B.
A. Stimulation Mode

Fig. 8. Designations for the eight cable leads and the input data word structure
for the cochlear array.

Fig. 9. Error detection block output on the Data Valid line (DAV).

one of the eight sites for stimulation for the addressed channel.
In the second mode, extended-mode, bits 07 define the command while bits 1014 control operations within the command.

A single 8-bit DAC is shared over four stimulus channels

and converts the command current level into an analog voltage
ranging between 1 V. A sample-and-hold (S/H) circuit retains
the current level for each channel during applied pulses. Each
S/H output is buffered and applied to an analog current driver
as shown in Fig. 11. When in stimulation mode, STIMSEL high
enables the current driver, and the S/H voltage is compared to
VREF. If the S/H voltage is greater than VREF, the top half
of the circuit is active and converts the voltage difference to
a drive current ( ) determined by . The current is mirrored
to MN1 and MN2 and used to set the bias for the low-voltage
cascode stage that sources 5 . When the S/H voltage is less than
VREF, the bottom half of the circuit sets the bias point using an
analogous scheme and sinks current from the output. The stimulus output driver generates a biphasic stimulus current ranging
from 0 to 500 A with a charge mismatch less than 1%
(Fig. 12). The output current levels are consistent with in vivo
estimates. The monopolar and bipolar auditory thresholds (the
levels of current necessary to elicit an auditory percept) in
guinea pig are 50 A and 300 A, respectively [3]. For situations requiring output currents of more than 500 A, multiple
sites can be driven in parallel with reduced spatial resolution.
The output voltage compliance range of 2 V is sufficient to
A and 40 nC. Minimum
support biphasic stimulation at
pulse widths are 4 s, enabling the system to deliver pulses
faster than the stimulation rates required clinically. An output
impedance greater than 50 M is seen at the cascode current
mirror output stage, and the maximum per-channel power
dissipation is 2.5 mW. Although the shared-DAC approach
implemented here is vulnerable to stimulus droop due to S/H
node leakage, it saves die area compared with a per-channel
current-steering DAC design, and the output stimulus currents
are sufficiently stable for use in vivo.

BHATTI AND WISE: A 32-SITE 4-CHANNEL HIGH-DENSITY ELECTRODE ARRAY FOR A COCHLEAR PROSTHESIS

2969

Fig. 10. Block diagram of the cochlear array circuitry.

Fig. 11. Schematic diagram of a stimulus output current driver.

B. Extended Mode Recording Function

An on-chip recording amplifier and selection matrix allow
the user to independently record from any site.1 To assess site
impedance, the extended mode recording command is issued.
The site is selected using bits 1014, the amplifier is powered
on, and a switch directing the amplifier output to the DTO line
is closed. A user-specified current level is output to the selected
site, and the recording amplifier captures the voltage response.
In clinical systems, impedances are determined immediately
after implantation to confirm functionality, open circuits presenting high impedances and short circuits presenting low, as
well as during patient follow-up visits. In Fig. 13, a representative in vivo voltage response to an applied charge-balanced
anodic-first biphasic current pulse is illustrated, exhibiting two
Fig. 12. Charge-balanced cathodic-first 140 A cochlear stimulation pulses on
two parallel output channels (100 s/div, DIN and STB at 5 V/div).

1Implementation of the recording block increases the chip footprint by 5 mm

to 2.4 mm 2.9 mm.

2970

Fig. 13. In-vivo voltage response to a 125-A, 160-s biphasic pulse.

IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 41, NO. 12, DECEMBER 2006

package to digitize the voltage response. A simple and robust

two-stage operational amplifier configured as an inverting amplifier doubles the signal to match the A/Ds input voltage range.
The amplifier has a closed-loop 3 dB frequency of 1.7 MHz that
is more than adequate for measuring the electrode potentials.
Operating from 2.5-V dual supplies, an active-load nMOS differential pair at the input amplifies the electrode potential referenced to a common potential (Fig. 15). The common node
is a large-area electrode, independent from the return
electrode in monopolar stimulation and considered as the animal
ground. Laser trimming of resistive taps allows for tuning the
bias point of the differential pair to adjust for process and in vivo
temperature variations. The single-ended output is directed to a
pMOS common-source output stage, sized to provide maximum
swing and substantial signal drive for the track-and-hold 12-bit
A/D. The sampled signal may then be reconstructed with the fidelity to accurately determine the impedance components.
C. Performance Summary

Fig. 14. Model of the electrode site impedance.

main regions. The first step-like change is regarded as the access voltage, , in the system from the sourcing electrode to the
sinking or return electrode. Losses such as the lead resistance,
contact resistance, and tissue spreading resistance are lumped
into an access resistance, . The second (more slowly moving)
response, the polarization voltage , is the loss due to charging
of the site interface capacitor and charge transfer via Faradic reactions at the electrode-electrolyte interface. This is modeled as
, in parallel with
. At present, clinical
a leaky capacitor,
software simply assesses the impedance by measuring the maximum response at the end of a 75- A 25- s pulse, and divides by
. By resolving the voltage response into
the current,
its essential components,
, , and
(Fig. 14), and monitoring their respective changes over time, the tissue response
at the electrode-tissue interface and endocochlear environment
can be inferred [16]. During the initial phase of tissue response,
it is believed that the surface quality of the electrode is altered by
a layer of protein on the surface, increasing . It is readily dispersed once stimulation commences. A more long-term tissue
response, exemplified by changes in the , is the formation of
a fibrous tissue sheath encapsulating the array as the foreign object is walled-off from the body. Such changes can reshape
the applied field and may even account for some variations in
speech perception [17], as well as increasing the power consumption of the system.
Our implementation allows the measurement of at least three
,
, and
time points of the voltage response for estimating
. A buffer on the cochlear ASIC provides the signal drive
to allow a commercial A/D converter located in the electronics

The performance of the cochlear ASIC is described in

Table I. Although the 4-V compliance voltage is sufficient,
based on acute in vivo measurements, clinical systems range
from 920 V in compliance. The use of IrO sites may contribute
to the compliance difference; there is potentially less voltage
buildup at the interface per charge delivered compared with the
platinum electrodes used clinically. To increase compliance and
scale the output current levels of our system, the AMI 0.5- m
process offers a high-voltage (20-V) thick-gate oxide option
that could readily be implemented if needed chronically. Such
voltage levels would allow the system to remain in compliance
while meeting increased threshold and comfort stimulus levels
that can change over time.
The stimulator achieves a biphasic current mismatch less
than 1%. While seeming adequate, still better charge balance
must be achieved for neural stimulation. Any excursion past
electrochemically safe reactions, violating the water-window,
can result in the evolution of oxygen or hydrogen gas and/or
local pH shifts at the interface. Histopathological studies examining the impact of elevated DC levels reveal a highly correlated
loss of neural elements (spiral ganglion cells) observed at levels
ranging from 0.42.8 A DC (200- s monophasic current
pulses), further underscoring the importance of charge balance
[18]. The addition of DC-blocking capacitors, while effective,
is not possible here due to space limitations. An alternative is to
short all electrodes to ground between pulses. It has been shown
A ensures safe long-term
that an average DC level
electrical stimulation, even at high stimulus rates, and can be
achieved by interpulse shorting [19]. Maintaining a DC current
under 0.1 A would otherwise require a biphasic mismatch less
than 0.02% at 500 A. Therefore, shorting is the best option
and can easily be implemented in the system presented here.
V. DISCUSSION
This paper describes the first high-density cochlear electrode
array with on-board integrated electronics for neural stimulation and recording. Four 500- A biphasic current stimulators
are multiplexed to 32 electrode sites. A recording function for
in vivo electrode impedance measurements is also presented.

BHATTI AND WISE: A 32-SITE 4-CHANNEL HIGH-DENSITY ELECTRODE ARRAY FOR A COCHLEAR PROSTHESIS

2971

Fig. 15. Schematic of the recording amplifier.

TABLE I
COCHLEAR ASIC PERFORMANCE SUMMARY

The hybrid approach implemented here combines industry-standard circuit features with electrode fabrication and flexible cabling technologies.
The present array has been sized for the guinea pig animal model and serves as a template for a potential 128-site
16-channel human system. The cochlear ASIC presented is
directly portable to the human. The 16-bit command word
could readily specify site and channel addresses for site multiplexing on 128-sites and stimulation on 16 channels. Scaling
the cochlear ASIC to accommodate a human system would
result in an estimated chip footprint of 5 mm 5 mm, well
within the space available outside the human cochlea. The digital circuitry could also be compressed further using a smaller
feature-size technology, further reducing circuit area; however,

in designing circuitry for integrated prostheses, the compliance

voltage makes a high-voltage option attractive along with a
mixed-signal process technology.
The success of the human device will require long-term stability and protecting the electronics from the corrosive tissue environment. Signal transfer from the cochlear ASIC to 128 traces
on the electrode array necessitates well-insulated high-density
interconnects. Fabricating the device monolithically, where the
electronics are integrated into the array substrate, is probably a
better option. The cochlear circuit area could then be hermetically sealed within a glass capsule.
Considering the electrode array, silicon substrates have
proven to be robust and reliable in a number of cortical stimulating and recording applications, and exhibit a high degree

2972

of out-of-plane flexibility2 [20], [21]. The 2.5- m-thick substrate of the cochlear array may be coiled around a 500- m
cylinder; more than sufficient to follow the radius of curvature
exhibited by commercial devices in the basal and second turn,
1.8 mm and 0.75 mm, respectively [22]. However, although
silicon arrays can sustain some rotation and bending3 [23], the
anatomical requirements for an intracochlear electrode array are
challenging. The array must bend to match the scala tympanis
intra-cochlear tilt [24], [25]. In our in vivo insertion trials,
we found this in-plane bending to be especially pronounced
in small animals. In other applications requiring flexible substrates, polymers such as parylene-C or polyimide (Pyralin) are
attractive thin-film alternatives, exhibiting a Youngs Modulus
of 24 GPa and 7 GPa, respectively, compared with 200 GPa
for boron-doped silicon [26]. Polyimide-based electrode arrays have been explored for cortical and visual applications
[27], [28], and a parylene-silicon substrate cochlear array
for position-sensing has been developed [29]. An additional
benefit of these mixed-material approaches is the ability to integrate flexible cables with the process. Trade-offs in long-term
stability and electrical isolation compared with silicon-only
substrates must be investigated. Additionally, the development
of a stress-compensated multilevel metal process coupled
with high-quality inter-level dielectrics is essential to provide
interconnects for 128 sites.
Contemporary arrays are flexible and self-curling, thereby encouraging the sites to closely approach the modiolus. A patient
study comparing a straight banded array (Fig. 1) with a perimodiolar array reports significantly lowered threshold and comfort levels with the contoured device [30]. Such a perimodiolar
resting state requires a backing to the array and an insertion
tool coupled with the thin-film. We are collaborating with the
Epstein Laboratory at the University of California, San Francisco (UCSF) to develop a molded silicone carrier for the 32-site
silicon array. A full assembly has been demonstrated in vivo.
We are also working with Michigan Technological University
to develop a pneumatically actuated insertion tool as part of
an overall closed-loop feedback system [31]. A human implant
based on thin-film arrays coupled to a suitable backing structure
and introduced using an actuated tool should achieve deep insertion and therefore span a wider perceived frequency range;
at the same time it can closely couple the sites to the cochleas
central core (modiolus), achieving the highly focused stimulation that is capable only with high-density sites.
The work presented here is part of an on-going effort to
improve the electrical-neural interface to further enhance the
sound perception possible with a cochlear implant. Paralleling
the development of other neural prosthetic systems (visual and
cortical), we have taken a hybrid approach here. It is through
the sustained effort and collaboration among many individuals
in the neural prostheses community that we will ultimately be
2In-plane refers to the x-y plane, i.e., the plane of a wafer. Analogous to the
roll, pitch, and yaw angles of navigation dynamics; the out-of-plane bending
angle can be considered the pitch angle.
3Rotation describes when one end of the array remains fixed and the array is
twisted about an axis in the plane of the array (roll). Bending describes when
the free-end is rotated in-plane (yaw).

IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 41, NO. 12, DECEMBER 2006

able to offer significantly improved treatments for sensorineural

deafness.
ACKNOWLEDGMENT
The authors thank S. Lee and I. Bhatti for design support,
M. Gulari, J. Wang, and K. Beach for array fabrication, B. Casey
and J. Hetke for system assembly, S. Lee, R. Gordenker,
A. Sodagar, and C. Ellinger for assistance with testing, and
R. Anderson for technical illustration. The authors also thank
J. Patrick and C. Treaba at Cochlear, Ltd., for design review.
REFERENCES
[1] Cochlear Implants, National Institute on Deafness and Other Communication Disorders, National Institutes of Health, Bethesda, MD,
2005 [Online]. Available: http://www.nidcd.nih.gov/health/hearing/
coch.asp
[2] L. Xu, C. S. Thompson, and B. E. Pfingst, Relative contributions of
spectral and temporal cues for phoneme recognition, J. Acoustic Society of America, vol. 17, no. 5, pp. 32553267, May 2005.
[3] P. T. Bhatti, B. Y. Arcand, J. Wang, N. V. Butala, C. R. Friedrich, and
K. D. Wise, A high-density electrode array for a cochlear prosthesis,
in Proc. IEEE Int. Conf. Solid-State Sensors and Actuators (Transducers03), Boston, MA, Jun. 2003, pp. 17501753.
[4] J. Wang, M. Gulari, and K. D. Wise, An integrated position-sensing
system for a MEMS-based cochlear implant, in IEDM 2005 Dig.,
Washington, DC, Dec. 2005, pp. 129132.
[5] R. Sarpeshkar et al., An ultra-low-power programmable analog bionic
ear process, IEEE Trans. Biomed. Eng., vol. 52, no. 4, pp. 711727,
Apr. 2005.
[6] J. Georgiou and C. Toumazou, A 126- W cochlear chip for a totally
implantable system, IEEE J. Solid-State Circuits, vol. 40, no. 2, pp.
430443, Feb. 2005.
[7] E. D. Marsman, G. A. Carichner, S. Kubba, and M. S. McCorquodale, A DSP architecture for cochlear implants, in Proc. IEEE Int.
Symp. Circuits and Systems (ISCAS 2006), Kos, Greece, May 2006,
pp. 657660.
[8] A. M. Sodagar and K. D. Wise, Wireless interface chip for power and
bidirectional data telemetry dedicated to a cochlear implantable microsystem, presented at the 2006 IEEE Biomedical Circuits and Systems Conf. (BioCAS06), London, U.K., Nov.-Dec. 2006.
[9] A. M. Sodagar, K. Najafi, K. D. Wise, and M. Ghovanloo, A fullyintegrated CMOS power regulator for telemetry-powered implantable
biomedical microsystems, in Proc. IEEE CICC, Sep. 2006.
[10] T. J. Harpster, S. Nikles, N. Dokmeci, and K. Najafi, Long-term hermeticity and biological performance of anodically-bonded glass-silicon implantable packages, IEEE Trans. Device Mater. Reliabil., vol.
5, no. 3, pp. 458466, Sep. 2005.
[11] J. Mitchell, G. R. Lahiji, and K. Najafi, Long-term reliability, burn-in
and analysis of outgassing in Au-Si eutectic wafer-level vacuum packages, in Proc. Solid-State Sensors and Actuators and MicroSystems
Workshop, Hilton Head Island, SC, Jun. 2006, pp. 376379.
[12] K. D. Wise, D. J. Anderson, J. F. Hetke, D. R. Kipke, and K. Najafi, Wireless implantable microsystems: high-density electronic interfaces to the nervous systems, Proc. IEEE, vol. 92, no. 1, pp. 7697,
Jan. 2004.
[13] Y. Yao, M. N. Gulari, S. Ghimire, J. F. Hetke, and K. D. Wise, A low
profile three-dimensional silicon/parylene stimulating electrode array
for neural prosthesis applications, in Dig. IEEE Conf. Engineering in
Medicine and Biology, Shanghai, China, Sep. 2005.
[14] J. D. Weiland and D. J. Anderson, Chronic neural stimulation with
thin-film, iridium oxide electrodes, IEEE Trans. Biomed. Eng., vol.
47, no. 7, pp. 911918, Jul. 2000.
[15] J. Zhang, Z. Wang, and A. Mason, A universal micro-sensor interface chip with network communication bus and highly programmable
sensor readout, in Proc. IEEE 45th Midwest Symp. Circuits and Systems, Tulsa, OK, Aug. 2002, pp. 246249.
[16] M. Tykocinski, L. T. Cohen, and R. S. Cowan, Measurement and analysis of access resistance and polarization impedance in cochlear implant recipients, Otology and Neurology, vol. 26, no. 5, pp. 948956,
Sep. 2005.

BHATTI AND WISE: A 32-SITE 4-CHANNEL HIGH-DENSITY ELECTRODE ARRAY FOR A COCHLEAR PROSTHESIS

[17] T. Hanekom, Modeling encapsulation tissue around cochlear implant

electrodes, Medical and Biological Engineering and Computing, vol.
43, no. 1, pp. 4555, Feb. 2005.
[18] R. K. Shepherd et al., Chronic electrical stimulation of the auditory
nerve using non-charge-balanced stimuli, Acta Otolaryngol, vol. 119,
no. 6, pp. 674684, 1999.
[19] J. Xu et al., Chronic electrical stimulation of the auditory nerve at high
stimulus rates: a physiological and histopathological study, Hearing
Research, vol. 105, no. 1, pp. 129, Mar. 1997.
[20] Y. Yao, M. N. Gulari, J. F. Hetke, and K. D. Wise, A low-profile
three-dimensional neural stimulating array with on-chip current generation, in Dig. IEEE Conf. Engineering in Medicine and Biology, San
Francisco, CA, Sep. 2004, vol. 3, no. 4, pp. 19941997.
[21] R. H. Olsson, III and K. D. Wise, A three-dimensional neural
recording microsystem with implantable data compression circuitry,
IEEE J. Solid-State Circuits, vol. 40, no. 12, pp. 27962804, Dec.
2005.
[22] P. Wardrop et al., A temporal bone study of insertion trauma and intracochlear position of cochlear implant electrodes. I: comparison of
nucleus banded and nucleus contour electrodes, Hearing Research,
vol. 203, pp. 5467, 2005.
[23] T. Lisby et al., Mechanical characterization and design of flexible silicon microstructures, J. Microelectromechan. Syst., vol. 13, no. 3, pp.
452464, Jun. 2004.
[24] C. Todd and F. Naghdy, Virtual cochlear implant insertion for medical education, in Proc. 1st Joint Eurohaptics Conf. and Symp. on
Haptic Interfaces for Virtual Environment and Teleoperator Systems,
Pisa, Italy, Mar. 2005, pp. 505506.
[25] S. Hatsushika et al., Dimensions of the scala tympani in the human and
cat with reference to the cochlear implants, Annals Otology Rhinology
Laryngology, vol. 99, pp. 871876, 1990.
[26] K. Najafi and K. Suzuki, Measurement of fracture stress, Youngs
modulus and intrinsic stress of heavily boron-doped silicon microstructures, Thin Solid Films, no. 181, pp. 251258, 1989.
[27] P. J. Rousch et al., Flexible polyimide-based intracortical electrode
arrays with bioactive capability, IEEE Trans. Biomed. Eng., vol. 48,
no. 3, pp. 361371, Mar. 2001.
[28] M. Humayun et al., Visual perception in a blind subject with a chronic
microelectronic retinal prosthesis, Vision Res., vol. 43, no. 24, pp.
25732581, Nov. 2003.
[29] J. Wang, M. N. Gulari, and K. D. Wise, A parylene-silicon cochlear
electrode array with integrated position sensors, in Dig. IEEE Int.
Conf. Engineering in Medicine and Biology, New York, NY, Sep. 2006,
pp. 31703173.
[30] E. Saunders, Threshold, comfortable level and impedance changes as
a function of electrode-modiolar distance, Ear and Hearing, vol. 23,
no. 1 Suppl., pp. 28S40S, Feb. 2002.
[31] B. Y. Arcand, P. T. Bhatti, N. V. Butala, J. Wang, C. R. Friedrich,
and K. D. Wise, Active positioning device for a perimodiolar cochlear
electrode array, Microsyst. Technol., vol. 10, no. 67, pp. 478483,
Oct. 2004.
[32] P. T. Bhatti, S. Lee, and K. D. Wise, A 32-site 4-channel cochlear
electrode array, in IEEE Int. Solid-State Circuits Conf. Dig. Tech. Papers, San Francisco, CA, Feb. 2006, pp. 5051.

2973

Pamela T. Bhatti (S05M06) received the B.S. degree in biomedical engineering from the University
of California, Berkeley, in 1989, and the M.S.E.E.
degree from the University of Washington, Seattle,
in 1993. She received the Ph.D. degree in electrical
engineering from the University of Michigan, Ann
Arbor, in August 2006.
In January 2007, she will join the School of Electrical and Computer Engineering at the Georgia Institute of Technology as an Assistant Professor. From
1987 to 1990, she was a Member of the Technical
Staff at Alza Corporation, where she conducted clinical and experimental processing of controlled release drug delivery systems. From 1994 to 1995, she was
with Motorolas Semiconductor Products Sector, where she developed and participated in an engineering rotation program for the MOS Digital and Analog
Integrated Circuits Division, and later as an Applications Engineer with Motorola-LONWORKS. In 1996, she joined Microware Systems Corporation as
an Embedded Systems Software Engineer where she developed device drivers
enabling infrared data transfer. From 1997 to 2000, she served as a Clinical
Trials Research Associate with the Department of Radiology at the University
of Michigan where she investigated the discrimination and characterization of
breast cancer using 3-D Doppler ultrasound imaging. Her current research interests are in the general area of biomedical sensors and subsystems, bioMEMS,
and microelectronics. Specifically, she is interested in implantable cochlear electrode arrays, vestibular prostheses, and the integration of controlled-release drug
delivery technology with flexible neural recording/stimulating arrays.
Dr. Bhatti is a member of the IEEE, Society of Women Engineers, and Engineering in Medicine and Biology Society.

Kensall D. Wise (S61M69SM83F86) received the B.S.E.E. degree with highest distinction
from Purdue University, West Lafayette, IN, in
1963, and the M.S. and Ph.D. degrees in electrical
engineering from Stanford University, Stanford, CA,
in 1964 and 1969, respectively.
From 1963 to 1965 (on leave 19651969) and from
1972 to 1974, he was a Member of Technical Staff
at Bell Telephone Laboratories, where his work was
concerned with the exploratory development of integrated electronics for use in telephone communications. From 1965 to 1972, he was a Research Assistant and then a Research
Associate and Lecturer in the Department of Electrical Engineering at Stanford, working on the development of integrated circuit technology and its application to solid-state sensors. In 1974, he joined the Department of Electrical
Engineering and Computer Science at the University of Michigan, Ann Arbor,
where he is now the J. Reid and Polly Anderson Professor of Manufacturing
Technology and Director of the NSF Engineering Research Center for Wireless
Integrated MicroSystems. His present research interests focus on the development of integrated microsystems for health care and environmental monitoring.
Dr. Wise organized and served as the first chairman of the Technical Subcommittee on Solid-State Sensors of the IEEE Electron Devices Society (EDS).
He was General Chairman of the 1984 IEEE Solid-State Sensor Conference,
Technical Program Chairman of the IEEE International Conference on SolidState Sensors and Actuators (1985), and IEEE-EDS National Lecturer (1986).
He served as General Chairman of the 1997 IEEE International Conference
on Solid-State Sensors and Actuators. Dr. Wise received the Paul Rappaport
Award from the EDS (1990), a Distinguished Faculty Achievement Award from
the University of Michigan (1995), the Columbus Prize from the Christopher
Columbus Fellowship Foundation (1996), the SRC Aristotle Award (1997), and
the 1999 IEEE Solid-State Circuits Field Award. In 2002 he was named the
William Gould Dow Distinguished University Professor at the University of
Michigan. He is a Fellow of the IEEE and a member of the United States National Academy of Engineering.