INVITED PAPER Microelectrodes, Microelectronics,and ......INVITED PAPER Microelectrodes,...

INV ITEDP A P E R

Microelectrodes,Microelectronics, andImplantable NeuralMicrosystemsProgress in development of tiny electrodes, cables, circuitry, signal processors

and wireless interfaces promises to advance understanding of the

human nervous system and its disorders.

ByKensall D. Wise, Life Fellow IEEE, Amir M. Sodagar, Member IEEE, Ying Yao, Member IEEE,

Mayurachat Ning Gulari, Gayatri E. Perlin, and Khalil Najafi, Fellow IEEE

ABSTRACT | Lithographically definedmicroelectrode arrays now

permit high-density recording and stimulation in the brain and

are facilitating new insights into the organization and function of

the central nervous system. They will soon allow more detailed

mapping of neural structures than has ever before been possible,

and capabilities for highly localized drug-delivery are being

added for treating disorders such as severe epilepsy. For chronic

neuroscience and neuroprosthesis applications, the arrays are

being used in implantable microsystems that provide embedded

signal processing and wireless data transmission to the outside

world. A 64-channel microsystem amplifies the neural signals by

60 dB with a user-programmable bandwidth and an input-

referred noise level of 8 �Vrms before processing the signals

digitally. The channels can be scanned at a rate of 62.5 kS/s, and

signals above a user-specified biphasic threshold are transmitted

wirelessly to the external world at 2 Mbps. Individual channels

can also be digitized and viewed externally at high resolution to

examine spike waveforms. The microsystem dissipates 14.14 mW

from 1.8 V and measures 1.4 � 1.55 cm2.

KEYWORDS | BioMEMS; implant; microelectrodes; microstimu-

lation; microsystems; neural probes; neural recording

I . INTRODUCTION

Of all the parts that make up the human body, the nervous

system is by far the least understood and its disorders are the

most difficult to treat. Research on the use of electrical

stimulation to restore movement to paralyzed limbs goes

back at least 250 years to the work of Franklin [1] and

others, but it was midway through the last century thatphysiologists began to use electrical recording to try to

understand the nervous system at the cellular level [2].

Electrolytically sharpened metal wire electrodes [3] were

used for extracellular recording, while glass micropipettes

[4] were used to probe individual cells. Significant progress

was made in understanding the behavior of single neurons

and some sensory areas of the brain, but it became

increasingly clear that understanding the organization andsignal-processing techniques used at the system level would

require simultaneous recording and stimulation from many

sites having known spatial locations in tissueVsomething

that existing technology did not then permit.

From the 1950s onward, there was also growing

interest in developing electronic interfaces to the brain

for use in prosthetic devices for treating various disorders.

Work on visual prostheses began with Brindley’s earlyefforts [5], but by the late 1960s, a program to develop a

cortical prosthesis for the blind was also under way under

Dobelle at the University of Utah [6], [7]. Initial work on

cochlear prostheses for the deaf began about the same time

[8]–[10]. Most early work involved experiments using

wire-based stimulating electrode arrays, but it was quickly

recognized that better technology was needed, not only for

Manuscript received June 18, 2007; revised January, 30 2008. This work was

supported by the Engineering Research Centers Program, National Science

Foundation, under Award EEC-9986866.

The authors are with the Engineering Research Center for Wireless Integrated

MicroSystems, Department of Electrical Engineering and Computer Science,

The University of Michigan, Ann Arbor, MI 48109-2122 USA.

Digital Object Identifier: 10.1109/JPROC.2008.922564

1184 Proceedings of the IEEE | Vol. 96, No. 7, July 2008 0018-9219/$25.00 �2008 IEEE

Authorized licensed use limited to: National University of Singapore. Downloaded on October 6, 2008 at 6:15 from IEEE Xplore. Restrictions apply.

forming the arrays themselves but also for the implantedmicrosystems that would be required for any practical

device. In addition to chronically implanted electrodes,

reliable leads, signal-processing circuitry, telemetry inter-

faces, and power sources would be needed. The required

technology did not exist at that time, but it was coming.

Microelectronics had taken a significant step forward

in 1959 with the invention of the lithographically based

planar process for integrated circuit fabrication, and by themid-1960s, the use of this process for fabricating

extracellular microelectrode arrays for neurophysiology

was under way [11]–[13]. This work drew on the selective

silicon etching technology being developed at Bell

Telephone Laboratories [14] and established the basis for

subsequent work on micromachined integrated sensors,

microelectromechanical systems (MEMS), and the micro-

systems of today. After 40 years, most of the technologyneeded for the realization of chronically implantable

neuroelectronic microsystems is now in place [15].

Advanced electrode design and wireless implantable

microsystems are hot topics today [16], with entire

conferences and journals devoted to them. Optimism is

being fueled by the spectacular successes achieved by

cochlear implants for the deaf and by deep brain

stimulation (DBS) in the subthalamic nucleus forParkinson’s Disease. Over 125 000 cochlear microsystems

have now been implanted worldwide, allowing profoundly

deaf patients to function normally in a hearing world, and

the use of DBS for the suppression of Parkinson’s tremor

has been remarkably successful, with very few side effects

when the electrodes are positioned correctly. Improved

prostheses for deafness and Parkinson’s Disease are beingdeveloped [17]–[19], along with new devices aimed at

treating blindness [20], [21], paralysis [22], [23], epilepsy

[24], and other disorders. Extracellular microelectrodes,

which provide the critical interface between the nervous

system and microelectronics, are discussed first below,

followed by a look at the present state-of-the-art in

implantable microsystems, including the interconnect

cables, site-interface circuitry, embedded signal processor,and wireless link to the external world.

II . MICROELECTRODE TECHNOLOGY

Extracellular microelectrodes record the voltages produced

by ionic current flow around neurons as their cell

membranes depolarize as the result of inputs received

from other cells. These neural Bspike[ potentials representthe electrical half of an electrochemical system, with

amplitudes as high as several hundred microvolts and a

frequency content extending up to perhaps 10 kHz. Almost

40 years ago, authors questioned whether this ionic current

flow was effectively isotropic or whether it was constrained

to flow only in narrow clefts between the surrounding

nonneuronal cells [3], making the observation of a cell

discharge (Bsingle unit[) largely a matter of luck. Thisspeculation was partly a result of the observation that in a

single pass through tissue, an electrode typically Bsees[ onlya few neurons, far fewer than the number known to exist

anatomically. However, based on many years of recording

with multisite probes, there is no evidence that electrode-

cell coupling is anisotropic. Cortical neurons can typically

Fig. 1. Basic structure of a micromachined multielectrode probe for recording or stimulation in the central nervous system.

Wise et al. : Microelectrodes, Microelectronics, and Implantable Neural Microsystems

Vol. 96, No. 7, July 2008 | Proceedings of the IEEE 1185


be seen electrically above background noise as far as 100 �maway, and in areas such as the cochlear nucleus, where cells

are smaller, recording fields are 40–50 �m in extent. Thus,

in order to record simultaneously from essentially all

neurons within a single block of tissue, sites on 80–200 �mcenters are required. This is possible today using the

lithographic techniques developed for integrated circuits.

The use of integrated circuit technology to create dense

arrays of thin-film electrodes for single-unit recording inthe nervous system began at Stanford University in 1966.

The resulting neural probes [11]–[13] had the general

structure shown in Fig. 1, which is typical of virtually all

lithographically defined electrode arrays reported since

that time. The probe structure consists of a selectively

etched (Bmicromachined[) substrate, with conducting

leads insulated above and below by inorganic dielectrics

and recording/stimulating sites formed by an area ofexposed metal. In 1970, it was difficult to fabricate such

devices reproducibly, but as MEMS technology developed

in the 1970s and 1980s, diffused boron etch-stops, reactive

ion etching (RIE), and silicon-on-insulator wafer technol-

ogy allowed probes to be realized reproducibly with high

yield using single-sided processing of wafers having normal

thickness [25].

As electrode technology developed, a wide variety ofmaterials were explored for use as the probe substrate

[26]–[30], including silicon, metals, glass, sapphire, and

polymers. While some of the resulting devices recorded

from neurons acutely, shaping the substrate, the strength

of that substrate [31], and the lack of suitable encapsulat-

ing dielectrics for chronic use were all important

problems. Most present approaches to probe formation

use silicon substrates [32]–[35]. Silicon can be shaped witha precision that probably exceeds that of any other material

[36] and taps directly into a plethora of processing

knowledge and equipment developed for the microelec-

tronics industry. It is compatible with the use of high-

quality silicon dioxide and silicon nitride dielectrics along

with some polymers, as well as with the formation of on-

chip circuitry [37]. It is also compatible with the

realization of probe structures that are very small (a fewmicrometers or even submicrometers in width).

The probe substrate must be realizable in reasonable

volumes with high yield, and that implies the ability to

define it lithographically in a batch process. Since silicon

wafers are typically 500–800 �m thick, it is important that

the probe shanks be formed using some form of etch stop,

either as a diffused-boron layer [25] or a buried-oxide layer

[34], [35]. Either approach can be effective, althoughboron etch-stops permit the use of multiple substrate

thicknesses and produce a smooth surface that is probably

helpful in minimizing tissue damage (Fig. 2). The boron

etch-stop also avoids any possibility of silicon substrate

dissolution in-vivo. It should be noted that not all silicon

electrode arrays are formed lithographically in a planar

process. The two-dimensional silicon depth arrays devel-

oped at the University of Utah [38], [39] are based on glass

reflow, sawing, and etching. They have produced some ofthe longest chronic implants reported to date.

Using silicon as the substrate and thermally grown

silicon dioxide (fused quartz) and low-pressure chemical-

vapor-deposited (LPCVD) silicon nitride as the insulating

dielectrics, conductors from the sites to bonding pads,

cables, or circuitry at the rear of the probe can be formed

using refractory metals, metal silicides, or polysilicon. In

spite of its relatively high resistance (10 �/square),polysilicon is adequate for most recording electrodes. For

stimulating probes, however, lower resistance materials

are needed to reduce drive voltages and the resulting

Fig. 2. Scanning electron microscope views of the tips of neural

probes fabricated using a boron-etch stop for the silicon substrate.

(Top) Side view of the substrate only. (Middle) Perspective view of a

probe tip. (Bottom) The lateral boron diffusion has been removed

using an RIE field etch.


1186 Proceedings of the IEEE | Vol. 96, No. 7, July 2008


voltage-stress on the encapsulating dielectrics. Typical cell

thresholds for stimulation are 10–20 �A, but stimulating

currents can reach 100 �A or more. Since the conducting

substrate below the leads and the extracellular fluid above

them both act as ground planes, interchannel crosstalk is

virtually negligible on such probes, and the small area of

the lithographically defined leads minimizes shunt capac-

itance so that signal attenuation is negligible as well. Siteimpedances are dominated by the series capacitance of the

metal-electrolyte double layer. Recording sites are usually

formed using gold or platinum, although anodically formed

iridium oxide [40], [41] is increasingly used. It produces

significantly lower recording impedances than other

materials and is essential for stimulating sites, permitting

more than 20 times the charge delivery to tissue than

platinum or gold at the same voltage [15], [42].Sites are thought to average the potential field seen in

tissue over their area. For recording from small cells, site

areas of 100 �m2 have often been used, although larger

areas of 300–400 �m2 have been more typical recently to

reduce thermal electrode noise, which is inversely related

to area. For stimulating sites, 1000 �m2 is suitable for

current levels of about �70 �A at supply voltages of �5 V,

limited by the spreading resistances in tissue near thesite [43].

Fig. 3 shows several neural probes on the back of a U.S.

penny. More than 7500 such devices have been supplied to

the neuroscience community by the University of

Michigan and have changed research directions in systems

neurophysiology. One of the additional benefits of using

silicon as the substrate material is that circuitry for site

Fig. 3. Several different probe designs shown on the back of a U.S. penny.

Fig. 4. A 64-site probe on a U.S. postage stamp. The sites are spaced

100 �m in depth with shanks on 200 �m centers. The probe is

configured for mounting in a 3-D array.




selection, signal amplification, stimulus current genera-tion, command decoding, and self-test can be integrated in

the probe substrate itself. Fig. 4 shows a 64-site eight-

channel recording probe on a U.S. postage stamp [44],

[45], while Fig. 5 shows a 64-site four-channel stimulating

array [46]. Several such probes can be microassembled

[47] to form three-dimensional (3-D) electrode arrays, as

in the 256-site four-probe 16-channel array also shown in

Fig. 5. This array has sites on 400 �m centers in threedimensions. Three-dimensional arrays of recording and

stimulating probes are poised to launch a revolution in our

understanding of neural systems, allowing computer-

controlled mapping of the connections between differentareas of brain in a few minutes that would be impossible

using discrete wire electrodes.

Lithographically defined probes can also support more

localized studies at the cellular level. By omitting the

boron diffusion in selected areas of the substrate, back-

looking sites can be realized as well as top-looking sites and

combinations of the two. Fig. 6 shows a diagram and

photograph of an 18-�m-wide probe containing top, back,and double-sided sites separated in depth by 40 �m. The

5-�m-diameter back-looking sites are separated from the

conducting substrate by less than 2 �m and yet record

Fig. 5. A 64-site eight-channel stimulating probe with sites on 400 �m centers and with CMOS electronics for stimulus generation,

recording, and self-test. Below, four such probes are mounted in a micromachined silicon platform to form a 256-site 32-channel array.




well, with the substrate appearing as an insulator to the

extracellular current [48]. Fig. 7 shows recordings made

with a 50-�m-wide probe having top-, double-, and back-

sided sites on 20 �m centers as it advances in guinea pig

inferior colliculus. Cells can clearly be handed off from one

site to the next, but the substrate here is screening some

cells from some sites. The double-sided sites average the

potentials seen on their two surfaces, but as shank width

Fig. 6. Diagram and backside view of an 18-�m-wide probe shank having top-, double-, and back-looking sites. The sites are

separated in depth by 40 �m.

Fig. 7. Recordings made with a 50-�m-wide neural probe having top-, double-, and back-sided sites on 20 �m centers as it advances

in depth through guinea pig inferior colliculus.




decreases, screening becomes less pronounced. For explor-

ing questions in neuroscience using acute extracellular

recording and stimulation, the necessary electrode array

technology is here now and is sufficient to allow rapid

progress in our understanding of neural structures. The

challenges are to provide high-density two-dimensional(2-D) and 3-D electrode arrays commercially and to develop

ways of asking the right questions of the recorded data using

appropriate algorithms embedded in smart multichannel

data acquisition systems.

For studies of plasticity and learning and for practical

neural prostheses based on recording, viable interfaces

with tissue for months, years, or even decades will be

required. Long-term (years) stimulation and shorter term(weeks) recording is possible now, but single-unit

recording for years will require further progress at the

probe–tissue interface. Probes implanted chronically are

encapsulated by a layer of glia a few micrometers thick.

Stimulating currents can penetrate this layer, but it can

shield small recording sites from extracellular potentials

that may be present. A variety of approaches are being

explored to increase recording life in-vivo from weeks ormonths to years [49]. These include coating the sites with

materials that prevent protein adsorption or attract

neuronal growth along with forming projections on the

sites that extend beyond the glial sheath.

Tresco [50] has also found that 100–150 �m-wide probe

shanks produce a 30–40% cell loss within 100 �m of the

shank (throughout the recording field of a typical site) and

elevated levels of astrocytes considerably beyond this. Onepossible explanation for these far-field effects is mechanical

motion of the probe in tissue, which is exacerbated by any

tethering of the probe to the skull. Designing probes with

shank widths that approach cellular dimensions and with

back-ends that fold over flat on the cortical surface [51] to

allow the dura to be replaced over the implant is one

approach to reducing such motion. Ultimately, scaling probe

size to smaller dimensions will be limited by strength rather

than technology, especially if penetrating the pia arachnoid

is required. Fig. 8 shows a probe whose shank consists of a10-�m-wide lattice of supporting silicon but which is

otherwise open, allowing cellular processes to regrow

through the shank. Histological studies are now under

way to determine the efficacy of such structures. The probes

still have sufficient strength for cortical use.

Fig. 8. Back side of a multisite lattice recording probe on a human hair, with an inset of the tip of the probe.

Fig. 9. Diagram of a probe containing devices for both electrical

and chemical recording and stimulation.




Fig. 10. View of a 50-�m-wide probe shank containing a microchannel and an integrated thermal flowmeter. Insets show two such

probes on a U.S. penny and the cross-section of a channel before etch-back to form the channel wall [57].

Fig. 11. Neural spike counts in the presence of ten 100 ms injections of AMPA (an excitatory drug) delivered over a 10-s period [57].




Considerable progress has been made recently in

addressing the chemical aspects of the electrochemical

neural system, with the goal of realizing probes that, in

addition to stimulating/recording electrodes, have a

complete set of microfluidic components: drug delivery

channels, flowmeters, valves, and chemical sensors. Fig. 9

shows the diagram of such a probe. Probes having fluidicdrug-delivery channels [52], shutters [53], and flowmeters

[54], [55] have been demonstrated, and integrated

microvalves [56] are being developed. Fig. 10 shows

probes having drug-delivery channels and integrated

thermal flowmeters. Such probes can resolve flows below

200 pL/s, have thermal time constants of about 200 �s,and are vacuum-shielded to limit any temperature rise in

tissue to less than 1 �C. Fig. 11 shows 10 �M AMPA, anexcitatory neural agonist, being delivered in ten 100-ms-

long pulses over a 10-s interval [57]. The neural spike rate

increases after delivery, gradually returning to pre-AMPA

levels. The use of voltammetry to measure in-vivo chemical

concentrations (e.g., dopamine) has also been recently

demonstrated [58]. By combining electrical recording and

stimulation with chemical recording and stimulation, new

windows are being opened in neuropharmacology, with

new prospects for the treatment of disorders such as severe

epilepsy by providing therapies at point of need.

Fig. 12. A neural microsystem consisting of 2-D and 3-D arrays of cortically implanted electrodes with ribbon cables connecting

them to a subcutaneous electronics package containing circuitry for amplification, spike detection and encoding,

and the wireless transmission of power and bidirectional data.

Fig. 13. Block diagram of a wireless implantable microsystem for

cortical single-unit recording along with its external user interface.




III . FULLY IMPLANTABLE NEURALMICROSYSTEMS

Chronic recording and/or stimulation in neuroscience and

in neural prostheses requires more than electrodes.Electronic site selection, stimulus generation, signal

amplification, spike detection and encoding, and the

wireless transmission of power and bidirectional data are

also required, in addition to any microfluidic functions

needed. Fig. 12 shows one possible configuration for such a

system, where two- and three-dimensional electrode arrays

interface with the cortex and are connected to a

subcutaneous electronics package using ultraflexible rib-bon cables. Within this package, the neural signals are

amplified, interpreted, and wirelessly interfaced to the

outside world. The remote location of the electronics with

respect to the cortex here allows more flexibility in power

dissipation since, in cortex, any temperature rise must be

less than 2 �C to avoid damage. It also allows better

shielding of the front-end neural signals from the wireless

power link and permits satellite platforms for antennas orfluidic reservoirs when needed. The tissue displaced by the

electrode array must be small enough to avoid any

significant disruption of the physiological system, and its

vertical rise above the cortical surface should be less

than 1 mm to allow the array to remain free of the skull

and float in tissue. Based on this implant architecture, the

various components of the microsystem will now be

discussed.

A. Interconnect Cables and Site-Interface CircuitryThe recent integration of parylene films, deposited and

patterned at wafer level prior to probe release, into the

probe process [59] is important not only because it allows

the use of parylene over the upper surface of the probe butalso because it permits the fabrication of parylene ribbon

cables as an integral part of the probe structure. Such

cables are more robust than silicon alternatives [60] and

potentially solve the mechanical interconnect problems

between the electrodes and the electronics package.

Accessing a few leads from an acute electrode array is

relatively simple, but for more than 30–40 sites, the back-

ends of such probes become prohibitively large. Forchronic implants, the situation is even worse, and since the

first silicon probes were produced in the 1960s, the

integration of circuitry on the probe itself has been an

important goal. The primary reason for on-chip circuitry is

to reduce the number of external leads (through site

selection and/or multiplexing) and permit the use of more

sites, but other benefits are to reduce impedance levels,

increase signal amplitudes, and decrease sensitivity tocable leakage.

The design of on-chip amplifiers for neural probes has

progressed from open-loop designs [44], [61] to capaci-

tively coupled closed-loop designs [62], [63]. Reported

amplifiers provide closed-loop gains of 1000, equivalent

input noise levels of 8 �Vrms, an upper cutoff frequency of

10 kHz, and lower cutoffs below 100 Hz [62]. Program-

mable low-frequency cutoffs have recently been demon-strated [64] to include or reject slow-wave activity,

depending on the application. State-of-the-art amplifiers

[45] dissipate 75 �W from �1.5 V and fit in an area of

0.07 mm2 in 0.5 �m complementary metal–oxide–

semiconductor (CMOS). Recognizing that most sites will

not be near neurons of interest, a front-end site selector

allows the user to choose from a large number of sites on

Fig. 14. Block diagram of a simple multichannel neural recording system.

Fig. 15. Functional block diagram of a 32-channel digital spike

detector [76].




the probe (electronic site positioning). The external probe

interface then typically requires seven leads (three power,

data in, data out, clock, and strobe), independent of the

number of sites. For a stimulating probe, the desired

current amplitudes and site addresses are entered serially,

and the generated currents are steered to the sites through

Fig. 16. Functional block diagram of the signal processor [77].

Fig. 17. The formation of data packets by time-division multiplexing the spike detector outputs.




the site selector while providing recording access to anysite [46], [65]. Microsystems using both active (multi-

plexed and nonmultiplexed) and passive electrode arrays

have been reported, but as these systems evolve, more will

likely incorporate on-chip circuitry to reduce lead counts

and improve system reliability.

B. Neural Signal ProcessorsFig. 13 shows a block diagram of the complete

microsystem required for wireless cortical recording,

either for neuroscience or for neuroprosthesis applica-

tions. After amplification of the neural signals, here using16- or 64-channel amplifier chips housed in the

electronics package, the signals are fed to a neural signal

processor (NPU). In its simplest form, such a processor

could be realized by a simple time-domain analog

multiplexer as in Fig. 14. This approach keeps the system

complexity low as long as the number of recording

channels is small. For instance, the three-channel analog

recording system reported in [66] easily fits on a 2.2 �2.2 mm2 chip in 1.5 �m CMOS technology. As soon as the

number of channels grows, however, the amount of neural

Fig. 18. (a) Time-domain data compression in an eight-channel spike detector module by using a local memory and (b) forming the

outgoing data packets out of the neural activity retrieved from four such modules.




information to be wirelessly transmitted will increase tosuch an extent that it cannot be easily handled. Thus, the

signal processor in Fig. 13 must take a more active role in

processing and compressing the recorded data. Much of

the emphasis in future generations of implantable

microsystems will focus on improving both the quantity

and quality of neural recordings. Work on the former will

focus primarily on increasing the number of channels,

while quality will focus on preserving as many features ofthe neural signal as possible, starting from the spike width

above a user-supplied threshold and progressing to a high-

resolution digital record of the spike waveform. In this

evolution, tradeoffs among system capability, power, and

size will be important. Sophisticated digital signal-

processing (DSP) algorithms for spike detection and

classification [67]–[73] require too much power and size at

present, so reported spike detectors have used user-programmable thresholds to provide information on spike

occurrence. This is adequate for some applications [74], [75]

and significantly reduces the amount of transmitted data.

A simplified block diagram of a mixed-signal neural

processor [76] is shown in Fig. 15. It receives four time-

multiplexed analog signal channels, each containing the

neural signals recorded on eight sites. After 5-bit analog-to-

digital conversion, the resulting signal amplitudes areseparated into 32 individual channels. A special-purpose

digital spike processor then computes the signal averages

and standard deviations for the individual channels and

calculates appropriate biphasic thresholds based on a user-

specified number of standard deviations. The amplitude of

each channel is then compared to its threshold to detectneural spikes. If the channel is active (contains a spike), the

sample is tagged with the associated channel address and

put in a buffer to be transmitted wirelessly at 2.5 Mbps.

This NPU provides full waveform information on neural

spikes above threshold, ignores subthreshold noise, saves a

factor of about 12 in output bandwidth, and increases the

number of allowable channels from 25 to 312.

The 32-channel signal processor [77] shown in Fig. 16supports two operational modes.

• In Scan Mode, all neural channels are scanned for

the occurrence of neural spikes. The addresses of

the active channels, i.e., channels with above-

threshold neural activity, are sorted, packed, and

sent to the outside world through a reverse

(outgoing) telemetry link. The threshold polarity

(positive, negative, or biphasic) and its level are setby the user for each channel individually.

• In Monitor Mode, a neural channel is selected,

sampled at high resolution, and transmitted to the

outside world.

This processor is equipped with additional circuitry

that allows it to be used as a 32-channel module in a 64-

channel neural processing architecture. Intrachip modu-larity (implemented by using eight-channel spike detectormodules (SD-8) in parallel) helps achieve high channel

scan rates. The number of channels handled by each

module, the number of parallel modules, and the size of

the local memory on each module are parameters that

can be optimized according to the targeted recording

Fig. 19. Data packaging [80]: (a) Incoming data packet, (b) outgoing data packet in Scan Mode, and (c) outgoing data packet in Monitor Mode.




capacity and speed. Interchip modularity allows expansionof the system by simply configuring one processor as the

master and another one (or more) processor(s) as the

slave(s).

C. Data CompressionPerhaps the simplest scheme for sending the detected

neural activity to the outside world is to use time-division

multiplexing, as illustrated in Fig. 17. The binary outputsof all the spike detectors are periodically scanned and

sent to the external interface after data packet formation

[78]. However, since the extracellularly recorded action

potentials from cortical neurons have typical durations of

approximately 1 ms with firing rates from G 1 to 150

spikes per second [79], the output of each spike detector

contains useful information (neural activity) only a small

Fig. 20. Wireless transfer of data packets along with a synchronized clock [80]: (a) functional block diagram and

(b) actual operating waveforms.




portion of time. Hence, transmitting the outputs of all thespike detectors (independent of their contents) wastes

significant recording bandwidth. A more efficient ap-

proach is to transmit only spike activity information

from active channels [77]. Here, when a spike is

detected on a given channel, the address of that channel

is sent to the external host. As shown in Fig. 16, the

local memory in each SD-8 module temporarily stores

the active channel addresses. The memory space is sharedamong all eight channels. The data fusion core polls the

four SD-8 modules to fetch the active channel addresses,

completing the process of spike detection, channel-

address tagging, spike sorting, and spike queuing. The

addresses retrieved from the four parallel queues will then

be used to make an outgoing data packet, as illustrated in

Fig. 18.

D. Bidirectional Wireless InterfacesImplantable recording microsystems need to be

powered and programmed through a forward telemetry

link in order to perform long-term wireless recording via a

reverse wireless path, as shown in Fig. 13. To support

these requirements, a major building block in such

systems is a bidirectional wireless interface that contains

at least two parts:/ a forward telemetry front-end that retrieves an

incoming stream of data packets containing setup

commands and data and provides the microsystem

with regulated supply voltages derived from the

modulated inductively coupled radio-frequency

carrier.

/ a reverse telemetry back-end that prepares the out-

going stream of neural data packets, and transmitsthe recorded information to the external world.

E. Data Transfer and PackagingWireless digital data transfer between the implantable

microsystem and the external host is basically asynchro-

nous, so to simplify detection and minimize power, a

synchronized clock is embedded in the data stream. To

facilitate recognition of a data packet and help the receiverseparate it from the preceding one, a start pulse (usually a

predefined pattern of 0s and 1s) is sent ahead of each

packet, and parity bits are added to each packet to allow

error detection. Fig. 19 shows incoming and outgoing data

packets [80], designed to program the microsystem and

transmit the recorded neural information, respectively. In

the incoming data packet, B0 and B1 encode four

commands, and B2–B9 carry the channel address ofinterest for either spike detection setting or for Monitor

Mode operation, or to convey required settings to the

channel of interest for operation in Scan Mode. The

outgoing data packet format in Scan Mode reports the

detected neural activity on one channel per each SD-8

using four bits. Every 8 bits are accompanied by a parity

bit. In Monitor Mode, each data packet carries two

consecutive 8-bit amplitude samples, along with two paritybits. The chip address bit (CAB), seen in the outgoing data

packet in both modes, is used when two 32-channel neural

processors of the same type (shown in Fig. 16) are

connected in a master–slave configuration to realize a 64-

channel processor. The CAB represents the chip that has

Fig. 21. Wireless recording in Monitor Mode.




prepared the data packet (B1[ for the master chip and B0[for the slave).

Fig. 20 presents the functional details of the reverse

telemetry link. First, the data packets are Manchester

encoded in order to incorporate the system clock. Then,

the encoded bit stream is on–off key (OOK) modulated

and is wirelessly transmitted to the external system. On

the external receiver side, after amplification, the reverse

process consisting of OOK demodulation and Manchesterdecoding is performed in order to retrieve the received

data and a synchronized clock.

F. Wireless RecordingWireless implantable recording microsystems have

recently been demonstrated [78], [80]. Fig. 21 shows in-

vivo recordings obtained in Monitor Mode from the guinea

pig auditory cortex using one of these systems [80]. TheBoriginal signal[ here is the amplified neural signal before

analog-to-digital conversion on the microsystem side, and

the Breconstructed signal[ is the retrieved signal after

passing through the wireless link, external receiver, and

postprocessing software in the host computer (see Fig. 13).

The signals shown in Fig. 21 are a small slice of a 30-s

recording in which two separate units were identified

based on postrecording analysis. For the first unit, thesignal/noise ratio (SNR) decreased from 11.21 on the

system side to 8.77 after recovery on the external side;

the second unit experienced an SNR reduction from 4.56

to 3.35 in passing through the same path. This degradation

in the signal quality comes from analog/digital conversion

with 8-bit resolution. Wireless data transfer does not

contribute to SNR reduction because digital data

modulation/transfer is used. Averaged over 24 886 datapackets, the packet error rate was 0.33%, which is excellent

performance for wireless data transfer in this application.

For a 2 MHz clock, the channel scan rate for spike

detection in this system is 62.5 kS/s and the total system

power dissipation at 1.8 V is 14.4 mW. The implantable

version of the microsystem measures 1.4 � 1.55 cm2,fitting on a U.S. penny.

IV. CONCLUSIONS

The technology for creating neural probes capable of

acutely measuring the neuronal activity throughout a

volume of tissue is now in place. Using three-dimensional

arrays of stimulating and recording electrodes, detailedmapping of connections in the nervous system should soon

be possible. Such studies will give important insights into

the signal-processing techniques used in the nervous

system and into the disorders that disrupt them. For

chronic studies, further advances are needed in the

electrode–tissue interface, and these needs are driving

electrode size toward cellular dimensions and below.

Beyond the electrodes themselves, substantial progresshas been made in the cables, site-selection and amplifica-

tion circuitry, embedded neural signal processors, and

wireless interfaces needed for chronic investigations in

neuroscience and for neural prostheses. The first com-

pletely implantable neural microsystems are now emerging

and should stimulate substantial progress in both areas. The

coming decade should see some dramatic breakthroughs in

our understanding of the nervous system and in our abilityto treat its disorders. h

Acknowledgment

The authors would like to thank Dr. F. Terry Hambrecht

and Dr. W. J. Heetderks of the Neural Prosthesis Program,

National Institute of Neurological Disorders and Stroke, for

their vision and support of this work over many years, aswell as the many faculty, students, and staff at the

University of Michigan who contributed to this research

in so many important ways. The support and encourage-

ment provided by P. V. Anderson of Cupertino, CA, is also

gratefully acknowledged.

REFERENCES

[1] B. Franklin, BAn account of the effects ofelectricity in paralytic cases,[ Phil. Trans.Lond., vol. L, pp. 481–483, 1975.

[2] M. Verseano and K. Negishi, BNeuronalactivity in cortical and thalamic networks,[ J.Gen. Physiol., vol. 43, pp. 177–195, 1960.

[3] D. A. Robinson, BThe electrical properties ofmetal microelectrodes,[ Proc. IEEE, vol. 56,pp. 1065–1071, Jun. 1968.

[4] O. F. Schanne, M. Lavallee, R. Laprade, andS. Gagne, BElectrical properties of glassmicroelectrodes,[ Proc. IEEE, vol. 56,pp. 1072–1082, Jun. 1968.

[5] G. S. Brindley and W. S. Lewin, BThesensations produced by electrical stimulationof the visual cortex,[ J. Physiol. (Lond.),vol. 196, pp. 479–493, 1968.

[6] W. H. Dobelle, M. G. Mladejovsky, andJ. P. Girvin, BArtificial vision for the blind:Electrical stimulation of visual cortex offershope for a functional prosthesis,[ Science,vol. 183, pp. 440–444, Feb. 1974.

[7] BAn electronic link to the visual cortex may letblind Fsee_,[ Electronics, Dec. 20, 1974.

[8] F. B. Simmons, C. J. Mongeon, W. R. Lewis,and D. A. Huntington, BElectrical stimulationof acoustical nerve and inferior colliculus:Results in man,[ Arch. Otolaryngol., vol. 79,pp. 559–567, 1964.

[9] F. A. Spelman, BThe past, present, and futureof cochlear prostheses,[ IEEE Eng. Med. Biol.Mag., pp. 27–33, May 1999.

[10] W. F. House and K. I. Berliner, BCochlearimplants: From idea to clinical practice,[ inCochlear Implants: A Practical Guide,H. Cooper, Ed. San Diego, CA: Singular,1991, pp. 9–33.

[11] K. D. Wise, J. B. Angell, and A. Starr, BAnintegrated circuit approach to extracellularmicroelectrodes,[ in Dig. 8th Int. Conf. Eng.Med. Biol., Jun. 1969, p. 14.5.

[12] K. D. Wise and J. B. Angell, BAmicroprobe with integrated amplifiers forneurophysiology,[ in Dig. IEEE Int. Solid-StateCircuits Conf., Philadelphia, PA, Feb. 1971,pp. 100–101.

[13] K. D. Wise and J. B. Angell, BAlow-capacitance multielectrode probe forneurophysiology,[ IEEE Trans. Biomed. Eng.,vol. BE-22, pp. 212–219, May 1975.

[14] M. P. Lepselter, BBeam-lead technology,[Bell Syst. Tech. J., vol. 45, pp. 233–254,Feb. 1966.

[15] K. D. Wise, D. J. Anderson, J. F. Hetke,D. R. Kipke, and K. Najafi, BWirelessimplantable microsystems: Electronicinterface to the nervous system,[ Proc. IEEE(Special Issue on Biomedical Applications forMEMS and Microfluidics), pp. 76–97,Jan. 2004.

[16] R. Harrison and K. D. Wise, BImplantable andprosthetic devices: Life-changing circuits,[ inDig. IEEE Int. Solid-State Circuits Conf.,San Francisco, CA, Feb. 2007, pp. 376–377.

[17] J. Wang, M. Gulari, and K. D. Wise, BAnintegrated position-sensing system for aMEMS-based cochlear implant,[ in Dig. IEEEElectron Devices Meeting, Washington, DC,Dec. 2005, pp. 129–132.




[18] P. T. Bhatti, S. Lee, and K. D. Wise, BA 32-site4-channel cochlear electrode array,[ IEEE J.Solid-State Circuits, pp. 2965–2973,Dec. 2006.

[19] P. Limousin, P. Krack, P. Pollack,A. Benazzouz, C. Ardouin, D. Hoffmann, andA. Benabid, BElectrical stimulation of thesubthalamic nucleus in advanced Parkinson’sdisease,[ New Eng. J. Med., vol. 339,pp. 1105–1111, 1998.

[20] J. D. Weiland and M. S. Humayun, BAbiomimetic retinal stimulating array,[ IEEEEng. Med. Biol. Mag., vol. 24, pp. 14–21,Sep. 2006.

[21] E. Margalit, M. Maia, J. D. Weiland,R. J. Greenberg, G. Y. Fujii, G. Torres,D. V. Piyathaisere, T. M. O’Hearn, W. Liu,G. Lazzi, G. Dagnelie, D. A. Scribner,E. de Juan, Jr., and M. S. Humayun, BRetinalprosthesis for the blind,[ Surv. Ophthal.,vol. 47, pp. 335–356, 2002.

[22] D. M. Taylor, S. I. Tillery, and A. B. Schwartz,BDirect cortical control of 3-Dneuroprosthetic devices,[ Science, vol. 296,pp. 1829–1832, Jun. 2002.

[23] L. R. Hochberg, M. D. Serruya, G. M. Friehs,J. A. Mukand, M. Saleh, A. H. Caplan,A. Branner, D. Chen, R. D. Penn, andJ. P. Donoghue, BNeuronal ensemble controlof prosthetic devices by a human withtetraplegia,[ Nature, vol. 442, pp. 164–171,Jul. 13, 2006.

[24] D. J. Anschel, E. L. Ortega, A. C. Kraus, andR. S. Fisher, BFocally injected adenoseneprevents seizures in the rat,[ Exp. Neurology,vol. 190, pp. 544–457, 2004.

[25] K. Najafi, K. D. Wise, and T. Mochizuki,BA high-yield IC-compatible multichannelrecording array,[ IEEE Trans. Electron Devices,vol. ED-32, pp. 1206–1211, Jul. 1985.

[26] G. A. May, S. A. Shamma, and R. L. White,BA tantalum-on-sapphire microelectrodearray,[ IEEE Trans. Electron Devices,vol. ED-26, pp. 1932–1939, Dec. 1979.

[27] M. Kuperstein and D. A. Whittington,BA practical 24-channel microelectrode forneural recording in-vivo,[ IEEE Trans. Biomed.Eng., vol. BE-28, pp. 288–293, Mar. 1981.

[28] O. J. Prohaska, F. Olcaytug, P. Pfundner, andH. Dragaun, BThin-film multiple electrodeprobes: Possibilities and limitations,[ IEEETrans. Biomed. Eng., vol. BE-33, pp. 223–229,Feb. 1986.

[29] N. A. Blum, B. G. Carkhuff, H. K. Charles,R. L. Edwards, and R. A. Meyer, BMultisitemicroprobes for neural recordings,[ IEEETrans. Biomed. Eng., vol. 38, pp. 68–74,Jan. 1991.

[30] P. J. Rousche, D. S. Pellinen, D. P. Pivin, Jr.,J. C. Williams, R. J. Vetter, and D. R. Kipke,BFlexible polyimide-based intracorticalelectrode arrays with bioactive capability,[IEEE Trans. Biomed. Eng., vol. 48,pp. 361–371, Mar. 2001.

[31] K. Najafi and J. F. Hetke, BStrengthcharacterization of silicon microprobes inneurophysiological tissues,[ IEEE Trans.Biomed. Eng., vol. 37, pp. 474–481, May 1990.

[32] T. Kewley, M. D. Hills, D. A. Borkholder,I. E. Opris, N. I. Maluf, C. W. Storment,J. M. Bower, and G. T. A. Kovacs,BPlasma-etched neural probes,[ Sens.Actuators A, vol. 58, pp. 27–35, 1997.

[33] T. H. Yoon, E. J. Hwang, D. Y. Shin, S. I. Park,S. J. Oh, S. C. Jung, H. C. Shin, and S. J. Kim,BA micromachined silicon depth probe formultichannel neural recording,[ IEEE Trans.Biomed. Eng., vol. 47, pp. 1082–1087,Aug. 2000.

[34] P. Norlin, M. Kindlundh, A. Mouroux,K. Yoshida, and U. G. Hofmann, BA 32-siteneural recording probe fabricated by DRIE ofSOI substrates,[ J. Micromech. Microeng.,vol. 12, pp. 414–419, 2002.

[35] K. C. Cheung, K. Djupsund, Y. Dan, andL. P. Lee, BImplantable multichannelelectrode array based on SOI technology,[IEEE J. Microelectromech. Syst., vol. 12,pp. 179–184, Apr. 2003.

[36] K. E. Petersen, BSilicon as a mechanicalmaterial,[ Proc. IEEE, vol. 70, pp. 420–457,May 1982.

[37] K. Najafi and K. D. Wise, BAn implantablemultielectrode array with on-chip signalprocessing,[ IEEE J. Solid-State Circuits,vol. SC-21, pp. 1035–1045, Dec. 1986.

[38] E. M. Maynard, C. T. Nordhausen, andR. A. Normann, BThe Utah intracorticalelectrode array: A recording structure forpotential brain-computer interfaces,[Electroenceph. Clin. Neurophys., vol. 102,pp. 228–239, 1997.

[39] R. A. Normann, D. J. Warren, J. Ammermuller,E. Fernandez, and S. Guillory,BHigh-resolution spatio-temporal mapping ofvisual pathways using multi-electrode arrays,[Vision Res., vol. 41, pp. 1261–1275, 2001.

[40] J. D. Weiland and D. J. Anderson, BChronicneural stimulation with thin-film, iridiumoxide electrodes,[ IEEE Trans. Biomed. Eng.,vol. 47, pp. 911–918, 2000.

[41] X. Beebe and T. L. Rose, BCharge injectionlimits of activated iridium oxide electrodeswith 0.2 ms pulses in bicarbonate bufferedsaline,[ IEEE Trans. Biomed. Eng., vol. 35,pp. 494–495, Jun. 1988.

[42] S. J. Tanghe, K. Najafi, and K. D. Wise, BAplanar IrO multichannel stimulating electrodefor use in neural prostheses,[ Sens. Actuators,vol. B1, pp. 464–467, Jan. 1990.

[43] C. Kim and K. D. Wise, BA 64-site multi-shankCMOS low-profile neural stimulating probe,[IEEE J. Solid-State Circuits, vol. 31,pp. 1230–1238, Sep. 1996.

[44] J. Ji and K. D. Wise, BAn implantableCMOS circuit interface for multiplexedmicroelectrode recording arrays,[ IEEE J.Solid-State Circuits, vol. 27, pp. 433–443,Mar. 1992.

[45] G. E. Perlin, A. M. Sodagar, and K. D. Wise,BNeural recording front-end designs for fullyimplantable neuroscience applications andneural prosthetic microsystems,[ in Dig. IEEEInt. Conf. Eng. Med. Biol., New York,Sep. 2006, pp. 2982–2985.

[46] M. D. Gingerich, J. F. Hetke, D. J. Anderson,and K. D. Wise, BA 256-site 3D CMOSmicroelectrode array for multipointstimulation and recording in the centralnervous system,[ in Dig. Int. Conf. Solid-StateSens. Actuators (Transducers’01), Munich,Germany, Jun. 2001.

[47] Q. Bai and K. D. Wise, BA high-yieldmicroassembly structure forthree-dimensional microelectrode arrays,[IEEE Trans. Biomed. Eng., pp. 281–289,Mar. 2000.

[48] G. E. Perlin and K. D. Wise, BThe effect of thesubstrate on the extracellular neural activityrecorded with micromachined siliconmicroprobes,[ in Dig. IEEE Conf. Eng. Med.Biol., San Francisco, CA, Sep. 2004,pp. 2002–2005.

[49] X. Cui, V. A. Lee, Y. Raphael, J. A. Wiler,J. F. Hetke, D. J. Anderson, and D. C. Martin,BSurface modification of neural recordingelectrodes with conducting polymer/biomolecule blends,[ J. Biomed. Mater. Res.,vol. 56, pp. 261–272, 2001.

[50] R. Biran, D. C. Martin, and P. A. Tresco,BNeuronal cell loss accompanies the braintissue response to chronically implantedsilicon microelectrode arrays,[ Exp.Neurology, Jul. 2005.

[51] Y. Yao, M. N. Gulari, S. Ghimire, J. F. Hetke,and K. D. Wise, BA low-profile three-dimensional silicon/parylene stimulatingelectrode array for neural prosthesisapplications,[ in Dig. IEEE Conf. Eng. Med.Biol., Shanghai, China, Sep. 2005.

[52] J. Chen, K. D. Wise, J. F. Hetke, andS. C. Bledsoe, Jr., BA multichannel neuralprobe for selective chemical delivery at thecellular level,[ IEEE Trans. Biomed. Eng.,vol. 44, pp. 760–769, Aug. 1997.

[53] D. Papageorgiou, S. C. Bledsoe, M. Gulari,J. F. Hetke, D. J. Anderson, and K. D. Wise,BA shuttered probe with in-line flowmetersfor chronic in-vivo drug delivery,[ in Dig.IEEE Microelectromech. Syst. Conf., Interlaken,Jan. 2001, pp. 212–215.

[54] D. Papageorgiou, S. Shore, S. Bledsoe, andK. D. Wise, BA shuttered probe with in-lineflowmeters for chronic in-vivo drug delivery,[IEEE J. Microelectromech. Syst., vol. 15,pp. 1025–1033, Aug. 2006.

[55] Y. Li, K. Baek, M. Gulari, and K. D. Wise, BAdrug-delivery probe with an in-line flowmeterbased on trench refill and chemicalmechanical polishing techniques,[ in IEEESensors Conf., Atlanta, GA, Oct. 2007.

[56] K. Baek, Y. Li, M. N. Gulari, and K. D. Wise,BA pneumatically-actuated microvalve forspatially-selective chemical delivery,[ in Dig.North Amer. Conf. Solid-State Sens., Actuators,Microsyst., Hilton Head, SC, Jun. 2006,pp. 155–158.

[57] Y. Li, BAn integrated drug-delivery probe withan in-line flowmeter,[ Ph.D. dissertation,Univ. of Michigan, Ann Arbor, 2006.

[58] M. D. Johnson, R. K. Franklin, K. A. Scott,R. B. Brown, and D. R. Kipke, BNeural probesfor concurrent detection of neurochemicaland electrophysiological signals in vivo,[ inProc. IEEE Eng. Med. Biol. Conf., Shanghai,China, Sep. 2005, pp. 7325–7328.

[59] J. Wang, M. N. Gulari, and K. D. Wise, BAparylene-silicon cochlear electrode array withintegrated position sensors,[ in Dig. IEEE Int.Conf. Eng. Med. Biol., New York, Sep. 2006,pp. 3170–3173.

[60] J. F. Hetke, K. Najafi, and K. D. Wise,BFlexible miniature ribbon cables forlong-term connection to implantablesensors,[ Sens. Acuators, vol. A23,pp. 999–1002, Apr. 1990.

[61] Q. Bai and K. D. Wise, BSingle-unit recordingwith active microelectrode arrays,[ IEEETrans. Biomed. Eng., pp. 911–920, Aug. 2001.

[62] R. H. Olsson, III, M. N. Gulari, andK. D. Wise, BA fully-integrated bandpassamplifier for extracellular neural recording,[in Proc. 1st Int. IEEE EMBS Conf. Neural Eng.,Capri, Italy, Mar. 2003, pp. 165–168.

[63] R. R. Harrison and C. Charles, BA low-powerlow-noise CMOS amplifier for neuralrecording applications,[ IEEE J. Solid-StateCircuits, vol. 38, pp. 958–965, Jun. 2003.

[64] R. H. Olsson, III, D. Buhl, A. M. Sirota,G. Buzsaki, and K. D. Wise, BBand-tunableand multiplexed integrated circuits forsimultaneous recording and stimulation withmicroelectrode arrays,[ IEEE Trans. Biomed.Eng., vol. 52, pp. 1303–1311, Jul. 2005.

[65] Y. Yao, M. N. Gulari, J. F. Hetke, andK. D. Wise, BA low-profile three-dimensionalneural stimulating array with on-chip currentgeneration,[ in Dig. IEEE Conf. Eng. Med. Biol.,San Francisco, CA, Sep. 2004, pp. 1994–1997.




[66] P. Mohseni and K. Najafi, BWirelessmulti-channel biopotential recording using anintegrated FM telemetry circuit,[ IEEE Trans.Neural. Syst. Rehabil. Eng., vol. 13,pp. 263–271, Sep. 2005.

[67] C.-W. Ko, Y.-D. Lin, H.-W. Chung, andG.-J. Jan, BAn EEG spike detection algorithmusing artificial neural network withmulti-channel correlation,[ in Proc. IEEE Eng.Med. Biol. Conf., 1998, vol. 4, pp. 2070–2073.

[68] T. Kalayci and O. Ozdamar, BWaveletpreprocessing for automated neural networkdetection of EEG spikes,[ IEEE Eng. Med. Biol.Mag., vol. 14, pp. 160–166, Mar. 1995.

[69] Y. Oomori, F. Yano, and S. P. Ninomija, BAcomputer-assisted spike detecting systemusing normalized method for epilepsydiagnosis,[ in Proc. 21st Ann. Hawaii Int. Conf.Syst. Sci., 1988, vol. 4, pp. 51–54.

[70] L. S. Pon, M. Sun, and R. J. Sclabassi, BThebi-directional spike detection in EEG usingmathematical morphology and wavelettransform,[ in Proc. 6th Int. Conf. SignalProcess., 2002, vol. 2, pp. 1512–1515.

[71] S. Kim, J. McNames, and K. Burchiel, BActionpotential detection with automatic templatematching,[ in Proc. Int. Conf. Eng. Med. Biol.,San Francisco, CA, Sep. 2004, pp. 41–44.

[72] M. Adjouadi, D. Sanchez, M. Cabrerizo,M. Ayala, P. Jayakar, I. Yaylali, andA. Marreto, BInterictal spike detection usingthe walsh transform,[ IEEE Trans. Biomed.Eng., vol. 51, pp. 868–872, May 2004.

[73] Z. Nenadic and J. W. Burdick, BSpikedetection using the continuous wavelettransform,[ IEEE Trans. Biomed. Eng., vol. 52,pp. 74–87, Jan. 2005.

[74] A. B. Schwartz, BCortical neural prostheses,[Annu. Rev. Neurosci., vol. 27, pp. 487–507,2004.

[75] D. M. Taylor, S. I. H. Tillery, andA. B. Schwartz, BInformation conveyedthrough brain-control: Cursor versus robot,[IEEE Trans. Neural Syst. Rehabil. Eng., vol. 11,pp. 195–199, Jun. 2003.

[76] R. Olsson, III, and K. D. Wise, BAthree-dimensional neural recordingmicrosystem with implantable data

compression circuitry,[ IEEE J. Solid-StateCircuits, vol. 40, pp. 2796–2804, Dec. 2005.

[77] A. M. Sodagar, K. D. Wise, and K. Najafi,BA fully-integrated mixed-signal neuralprocessing module for implantablemulti-channel cortical recording,[ IEEETrans. Biomed. Eng., vol. 54, pp. 1075–1088,Jun. 2007.

[78] R. Harrison, P. T. Watkins, R. J. Kier,R. O. Lovejoy, D. J. Black, B. Greger, andF. Solzbacher, BA low-power integratedcircuit for a wireless 100-electrode neuralrecording system,[ IEEE J. Solid-State Circuits,vol. 42, pp. 123–133, Jan. 2007.

[79] M. A. L. Nicolelis, A. A. Ghazanfar,B. M. Faggin, S. Votaw, and L. M. O. Oliveira,BReconstructing the engram: Simultaneous,multisite, many single neuron recordings,[Neuron, vol. 18, pp. 529–537, Apr. 1997.

[80] A. M. Sodagar, G. E. Perlin, Y. Yao,K. D. Wise, and K. Najafi, BAn implantablemicrosystem for wireless multi-channelcortical recording,[ in Proc. Int. Conf.Solid-State Sens., Actuators, Microsyst.(Transducers’07), Lyon, France, Jun. 2007.

ABOUT THE AUTHORS

Kensall D. Wise (Life Fellow, IEEE) 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 and from 1972 to 1974, he

was a Member of Technical Staff with 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, Stanford, working

on the development of micromachined solid-state sensors. In 1974, he

joined the Department of Electrical Engineering and Computer Science,

University of Michigan, Ann Arbor, where he is now the J. Reid and Polly

Anderson Professor of Manufacturing Technology and Director of the

Engineering Research Center for Wireless Integrated MicroSystems. His

present research focuses on the development of integrated microsys-

tems for health care and environmental monitoring. In 2002, he was

named the William Gould Dow Distinguished University Professor at the

University of Michigan, where he also held the 2007 Henry Russel

Lectureship.

Dr. Wise is a member of the U.S. National Academy of Engineering. He

organized and was 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, IEEE-

EDS National Lecturer (1986), and Technical Program Chairman (1985)

and General Chairman (1997) of the IEEE International Conference on

Solid-State Sensors and Actuators. He 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.

Amir M. Sodagar (Member, IEEE) received the

B.S. degree from K. N. Toosi University of Tech-

nology (KNTU), Tehran, Iran, in 1992 and the M.S.

and Ph.D. degrees from Iran University of Science

and Technology (IUST), Tehran, in 1995 and 2000,

respectively, all in electrical engineering.

He was with S. Rajaee University as a Lecturer

from 1995 to 2000, with Iran Telecommunication

Research Center (ITRC) as a Design Engineer

during 1996–1997, with VLSI Circuits and Systems

Laboratory, University of Tehran, as a Research Engineer from 1997 to

1998, and with EMAD Semicon as a Senior Design Engineer from 1998 to

2000. He was a Guest Lecturer at several institutions during this time and

was a member of the Electrical and Electronics Engineering Technical

Committee of the International Kharazmi Youth Innovation Festival

during 1995–2000. In 2000, he joined the National Science Foundation

Engineering Research Center for Wireless Integrated MicroSystems

(WIMS), University of Michigan, Ann Arbor, as a Research Fellow, where

he worked on integrated microsystems for electric nerve stimulation. In

2002, he joined KNTU as an Assistant Professor. Since 2004, he has been

with WIMS as a Visiting Associate Research Scientist and subsequently as

the Technical Director for Biomedical Microsystems. His research

interests are in mixed-signal integrated circuits and biomedical implant-

able microsystems. He is the author of Analysis of Bipolar and CMOS

Amplifiers (Boca Raton, FL: CRC Press/Taylor & Francis Group, 2007).

Ying Yao (Member, IEEE) received the B.S.

degree in electrical engineering from Wuhan

University, China, in 1997 and the M.S. and Ph.D.

degrees in electrical engineering from the Univer-

sity of Michigan, Ann Arbor, in 2000 and 2005,

respectively.

Since 1999, she has been a Research Assistant

and then a Research Fellow with the Engineering

Research Center for Wireless Integrated Micro-

Systems, University of Michigan. Her research

interests focus on the development of implantable wireless integrated

microsystems for neural stimulating and recording with applications in

brain–machine interfaces, neuroscience, and neuroprostheses.




Mayurachat Ning Gulari received the B.S. degree

in chemical technology and the M.S. degree in

polymer science and engineering from Petroleum

and Petrochemical College, Chulalongkorn Uni-

versity, Bangkok, Thailand, in 1993 and 1995,

respectively.

She was a Planning Engineer with Star Petro-

leum Refining Company (a joint venture of

Chevron and Texaco), where she worked on

optimization of feedstock and products. She is

currently a Research Engineer in the Engineering Research Center for

Wireless Integrated MicroSystems, University of Michigan, Ann Arbor,

where her research activities involve designing and fabricating micro-

machined silicon probes for drug delivery, recording, and stimulation.

Gayatri E. Perlin received the B.S.E. and M.S.E.

degrees in electrical engineering from the Univer-

sity of Michigan, Ann Arbor, in 2001 and 2003,

respectively, where she is currently pursuing the

Ph.D. degree in electrical engineering.

Her thesis is focused on the development of a

fully implantable microsystem for neural prosthe-

ses. Her research interests include microfabrica-

tion, microelectromechanical devices, and

integrated circuits for biomedical and other

applications.

Khalil Najafi (Fellow, IEEE) received the B.S., M.S.,

and Ph.D. degrees from the Department of Elec-

trical Engineering and Computer Science, Univer-

sity of Michigan, Ann Arbor, in 1980, 1981, and

1986, respectively, all in electrical engineering.

He is the Schlumberger Professor of Engineer-

ing in the Electrical Engineering and Computer

Science Department, University of Michigan. He

was Director of the Solid-State Electronics Labo-

ratory from 1998 to 2005. He has been Director of

National Science Foundation (NSF)’s National Nanotechnology Infra-

structure Network (NNIN) since 2004 and Deputy Director of the NSF

Engineering Research Center on Wireless Integrated Microsystems

(WIMS) at the University of Michigan. His research interests include

micromachining technologies, micromachined sensors, actuators, and

microelectromechanical systems; analog integrated circuits; implantable

biomedical microsystems; micropackaging; and low-power wireless

sensing/actuating systems. He has been active in the field of solid-state

sensors and actuators for more than 20 years. He has been involved in

several conferences and workshops dealing with microsensors, actua-

tors, and microsystems, including the International Conference on Solid-

State Sensors and Actuators, the Hilton Head Solid-State Sensors and

Actuators Workshop, and the IEEE/ASME Micro Electromechanical

Systems Conference. He is an Associate Editor of the Journal of

Micromechanics and Microengineering and an Editor for Sensors and

Materials.

Dr. Najafi is a Fellow of AIBME. He received a National Science

Foundation Young Investigator Award from 1992 to 1997. He is an

Associate Editor of the IEEE JOURNAL OF MICROELECTROMECHANICAL

SYSTEMS. He was an Associate Editor of the IEEE JOURNAL OF SOLID-

STATE CIRCUITS from 2000 to 2004, the Editor for Solid-State Sensors of

the IEEE TRANSACTIONS ON ELECTRON DEVICES from 1996 to 2006, and an

Associate Editor of the IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING

from 1999 to 2000.




INVITED PAPER Microelectrodes, Microelectronics,and ......INVITED PAPER Microelectrodes,...

Documents

Transcript of INVITED PAPER Microelectrodes, Microelectronics,and ......INVITED PAPER Microelectrodes,...