68
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 38, NO. I , JANUARY 1991
Multisite Microprobes for Neural Recordings Norman A . Blum, Member, IEEE, Bliss G . Carkhuff, Harry K. Charles, Jr., Senior Member, IEEE, Richard L. Edwards, and Richard A. Meyer, Member, lEEE
Abstract-Multisite, passive microprobes have been developed to allow simultaneous recording of action potential activity from multiple neurons at different locations in the brain. The microprobes were fabricated using standard integrated circuit techniques. The probe is a planar structure that consists of gold electrodes sandwiched between two polyimide dielectric layers and bonded to a molybdenum structural support. Windows in the top dielectric layer expose the electrode sites and bonding pads. In two distinct versions of the probe four or six recordings sites, respectively, of approximately 25 pm2 are arranged on a dagger-shaped structure which can penetrate the pia. The bonding pads and interconnect wires at the probe head are entirely encapsulated in a tubular fixture that is packed with silicone RTV and sealed with epoxy to protect the interconnections from contact with body fluids. The site impedances at 1 kHz are typically between 2 and 4 MQ. Probe lifetimes for continuous immersion in physiological saline solution, as measured by impedance, have exceeded 750 h. The failure mechanism is believed to be due to moisture and ion absorption in the top dielectric layer. In acute neurophysiological experiments using the four site probes, action potential activity was recorded from physiologically identified neurons in the dorsal column nuclei of anesthetized rat.
I. INTRODUCTION
F
OR more than 30 years, fine-tipped microelectrodes have been inserted into the brain to record extracellular action potentials generated by individual neurons [ 11. This method of observing single neurons in succession has contributed to many significant advances in understanding the function of the brain. The applications of microelectronic techniques to the development of multielement microelectrodes makes it possible to record simultaneously from many different neurons. Multielement microelectrodes have many advantages over the conventional single site electrodes, including: 1) For single site recordings, functional relationships between cells must be inferred from the population data obtained from successive experiments. Simultaneous recordings from many neurons allow functional relationships to be assessed directly. 2) For single site recordings, a population of data is obtained by repeating the stimulus and/ or behavior paradigm while recording from different cells. This is laborious both for the experimenter and for the animals. Multielectrode recordings can dramatically increase the number of neurons per experiment and thus reduce the number of experiments and, perhaps more importantly, the number of animals necessary to collect data. 3) In many experiments, the evoked potential linked to the stimulus results in a noise signal which makes single unit recording difficult. Differential recording from neighboring pairs of the multielement electrodes allows for rejection of the noise signal common to the two electrodes. 4) Manuscript received October 16, 1989; revised May 4, 1990. This work was supported under Public Health Service Grant NS07226. A preliminary version of this paper was presented at the Fourteenth Northeast Bioengineering Conference [ 121. The authors are with the Applied Physics Laboratory, The Johns Hopkins University, Laurel, MD 20723. IEEE Log Number 9040624.
The fixed spacing of electrodes facilitates studies of topographic organization and functional relationships. 5) Prosthetic devices controlled by the CNS will require muiltiple recording and stimulating sites. Two major approaches to the development of multielement electrodes have been taken. The first, historically, was the fabrication of bundles of separate, single-wire microprobes that had either a fixed three-dimensional configuration [ 2 ] or a fixed twodimensional configuration with adjustable depth [ 3 ] . The capability of these arrays to record simultaneously from more than one neuron has been demonstrated. The second approach has utilized the application of thin film microelectronics technology for fabrication of multielectrode microprobes. Potential technological advantages of the use of thin film technology to fabricate microprobes with a linear array of electrode sites are: 1) a high degree of reproductibility once the design and processing sequence are developed, 2 ) a precise knowledge of the spatial distribution of electrode sites, 3 ) a high “packaging density” of electrode sites for a given implanted volume, 4) the possibility of incorporating the interface circuitry directly on the microprobe base, and 5 ) the distribution of electrodes in a specifiable geometric pattem. In addition to the Hopkins team, there are a number of other groups presently developing planar multielectrode microprobes. The Vienna probe [4]is perhaps the most advanced being offered on a commercial basis (Otto Sensors Corp., Cleveland, OH) with several variations for different aspects of neurophysiological research. It is made on a rigid glass substrate with a nominal thickness of 100 pm and uses low pressure CVD silicon nitride for insulation. The Michigan probe [5] is made on a silicon substrate which can be thinned in the implanted region to 15 pm and can have an almost arbitrary shape. Several probe designs have been fabricated, including ones with on-board, integrated electronics. A combination of SiO, and Si,N, are used as passivation or insulation layers. This probe appears to be the most technologically advanced sensor because of its integrated electronics and the sophisticated semiconductor processing necessary for its manufacture. The MIT probe [6] is made on molybdenum (18 pm thick) with up to 24 recording sites. The molybdenum is strong, but flexible, and is not subject to brittle fractures as are glass and silicon. Various dielectrics have been used, including photoresist and silicon nitride. The Hopkins probe [7] described herein contains the strong, flexible molybdenum substrate with an organic dielectric (polyimide). The probe is approximately 19 pm thick and currently contains either four or six recording sites. The molybdenum-polyimide structure is easy to fabricate using standard microelectric technology and does not require advanced semiconductor processing techniques. The flexibility and strength of both materials make a mechanically forgiving structure that should increase the ruggedness of the probes in various applications.
0018-9294/91/0100-0068$01.OO
0 1991 IEEE
-
’I
69
BLUM et a l . : MICROPROBES FOR NEURAL RECORDINGS
11. FOUR-SITEPROBE DESIGNAND FABRICATION The four-site microprobe, which has been used for in vivo experiments, is constructed as a sandwich of four principal layers: 1) a rigid supporting substrate of molybdenum, 2) a dielectric layer of polyimide, upon which 3) a conductor layer of gold is patterned, covered by 4) a top dielectric layer of polyimide with open areas to expose the site electrodes along the shank, and with bonding pads at the head. A plan view of Fig. 1 shows the shape and dimensions of the four-site probe. In this probe, which was originally designed for superficial recordings in monkey cortex, the electrode sites are colinear and 500 pm apart. The fabrication sequence is depicted in cross-sectional view in Fig. 2. Four photolithographic masks are required to manufacture the probes, and there are four individual probes on each molybdenum substrate. The 99.99% pure molybdenum foil substrate is procured from Goodfellow Metals, Ltd., in England. The foils are in the form of 1 in x 1-in squares, rolled to a thickness of 15 pm. They are high quality, free of pinholes, and have a smooth, rolled finish. The foils are lightly polished chemically in an acid etch ( H3P04: HNO, : CH3COOH : H 2 0 16: 1 : 1 :2 ) to further improve the surface. The polymide is formed from the precursor material DuPont Pyralin, Type PI2555. This material is in a liquid form that allows the precursor (polyamic acid solution) to be spin-coated onto the substrate. The thin chrome layer provides an adhesion surface for the polyimide, which does not adhere well to gold. After heat treatment, the polyimide layer is cured (imidized) and becomes a polyimide film approximately 1.O-1.5 pm thick depending on processing details. The gold electrodes and associated conductor structure are sputter deposited to a thickness of 0.5 pm from a pure gold target. Sputtered gold has good adhesion to the previously formed polyimide, is chemically stable, and is generally considered to be compatible with brain tissue [8], [9]. The probes are separated from the surrounding molybdenum substrate by wet chemical etching using the same acid etch that was used for polishing (see above). The initial layers of gold on the front and back of the substrate serve as etch masks. The probes, however, remain attached to the substrate by three narrow tabs that are cut just prior to mounting and assembly. Probes that are acceptable for mounting are identified by visual inspection under a microscope, and by electrical testing to assure electrical isolation of the recording sites from the substrate and from each other. The present yield indicates that about 50% of the fabricated probes pass these initial tests and can be mounted, having at least three sites suitable for neural recording; about 10%have four good sites. Failed sites (impedance out of acceptable range) are usually due to open tracks or connections within the mounting assembly (impedance too high), or to overetched or delaminated dielectric (impedance too low). In a few cases too high impedance is attributed to failure to completely clear the electrode windows, or too low impedance to pinholes in the polyimide layers. The good probes are separated from the substrate frames, and 0.008 in copper wires are manually attached by gold conductive epoxy to the bonding pads. The probes are then mounted by partial insertion into a tublar, polycarbonate fixture that can be handled easily, and which has an external plug for electrical connection to the recording apparatus. The region around the bonding pads, within the tube, is packed with silicone RTV and sealed with epoxy to ensure good insulation. Two thin (0.5 pm) gold flaps, about 75 pm long, that result from undercutting of the etching masks, are removed by care-
1-
I-
2 75 mm
\\II
I
Fig. 1. Pian view of the four-site probe.
0
MO substrate
rubrlrate
15um
1 0 ~n X 1 0
tn. X
0.0006 an 115rml
c Sputrered AuICr layer substrate +Sputtered
1 p m Au-
@ away
/
Mask step No 1 Pattern
t @
Au laver
AuiCr on top Au on bottom
t
1 to 1.5pm polyimide
Mask step No 2 Pattern
Polyimide msulatmn layer
substrate
1 to 1 5 p m polyimide
Pattern
Gold conductor
Wlndow !n polyimide Mask step No 4 Pattern
Polvimide top layer
Top polyimide @
Au electrode
AuICr
Bottom Dolvimide
Etched out probe
Fig. 2 . Schematic illustration of the major steps in the fabrication sequence of the four-site probe.
fully dipping the tip of the assembled probe in a gold etchant solution ( K I :I, : H,O - 4 : 1 : 4; by weight) while observing the operation under a microscope. Micrographs of the four site probe are shown in Figs. 3 and 4. ELECTRICALTESTING For in vitro testing, each assembled probe is mounted in a test fixture that is located inside an electromagnetically screened box. The probe is immersed beyond the uppermost electrode site in a 0.9%saline aqueous solution. When an alternating test current of 10 nA peak to peak is passed through one of the probe electrodes and the saline solution, a potential develops at the electrode-electrolyte interface. This potential is used to determine the probe site impedance. A computer controlled test system has been constructed to make impedance versus frequency measurements for each recording site on the probe. The system, which is shown sche-
1
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 38, NO. I , JANUARY 1991
70
(b) Fig. 3. Electron micrographs of (a) the entire shank and (b) the tip region of the four-site probe.
matically in Fig. 5, can make single frequency impedance measurements on any one of the recording sites, measure all recording sites at a single frequency in a scan mode, or measure all recording sites at multiple frequencies at preselected time intervals. A low distortion signal generator provides precision frequency and amplitude test signals to each of the stimulus/measurement units (SMU) in the system. The SMU electrical schematic diagram is shown in Fig. 6. Each SMU contains a voltage controlled current pump, a high impedance buffer and a filtered 5X amplifier. The current pumps have a transconductance of 2 p S providing a 10 nA peak to peak output current for a 5 mV peak to peak input signal. The high impedance buffer (1000 MQ) isolates each probe site from loading effects and provides input signal guarding to minimize cable capacitances and increase measurement bandwidth. The output of each SMU is multiplexed to a single amplifier that provides an additional 26 dB of gain and high frequency filtering. The final signal is measured using an ac rms voltmeter. Multiplexing control and system power are provided through the switch/control unit.
Fig. 4. Electron micrographs of (a) top surface view of entire four-site probe, (b) upper and lower dielectric layers near the probe tip, and (c) an electrode in a dielectric window.
The signal generator, voltmeter, switch/control unit, and +_ 15 V power supplies are connected to a desk-top computer via the IEEE 488 Bus. The computer software permits complete auto-
"I
71
BLUM et al.: MICROPROBES FOR NEURAL RECORDINGS
Signal return Return signal output bus
ground
0 9% by weight saline soluton
Final output signal level
Platinumwire reference electrcde
Ouler shield and signal
/Enclosure
wall
0 011 10
Switchiconlral
1
I
100
1000
11
Frequency I H r i
SMA
~
Fig. 7. Initial impedance values of different frequencies for each 25 X 25 pm site of a four-site probe; the 10 x 10 pm and 50 X 50 pm data are from [lo].
/Feedthrwgh I
Computer
I 1I
II
Double clad EP I - alass
Fig. 5 . Block diagram of test system for probe impedance measurements.
Signal Return
ti,-
outpu1
System Ground
Probe Site 2
StimuluslMeasurement Unit
The impedances at each site are recorded at frequencies between 100 Hz and 10 kHz in a l , 2, 5 sequence for a total time of at least 60 min in order to identify stabilized probe site impedances. From each production batch of four substrates ( 16 probes), one probe that has four good sites is selected and further tested by remaining in the saline solution until at least one site shows signs of failure, as denoted by rapidly decreasing impedance. This has proved to be indicative of the average performance of the other probes in the same batch that have site impedances in the acceptable range (2-4 M Q ) on initial (60 min) testing. Initial impedance versus frequency measurements for a probe with 25 x 25 pm recording sites are shown in Fig. 7 . These data compare well with those reported by Drake [lo] for 10 x 10 pm and 50 x 50 pm sites. Fig. 8 shows impedance measurements made on this probe during a 950 h immersion in a physiological saline bath. Impedances of three probe sites were stable for longer than 600 h, and declined to half their initial impedance values after about 750 h. Post failure electron micrograph studies of long-lived probes appear to rule out pin hole or delamination of the structure as explanations for the changes in impedance, which are most likely due to the effects of water absorption in the dielectric layers.
Fig. 6 . Schematic diagram of stimuludmeasurement unit.
IV. NEURALRECORDINGRESULTS matic control of test signals, SMU multiplexing and probe electrode selection, and data collection and analysis. The system can run unattended for long periods of time which is helpful when conducting life tests. Calibration of the system is done by placing precision 1% resistors in the test holder in place of the probe and correlating the measured potentials for each resistor with an impedance value. The system is calibrated from 100 kQ to 20 MQ with a sensitivity of 1 mV peak to peak per 1 kQ. Within the calibration range, measurement accuracy is 1-2%, repeatability is 12 % , and resolution is 1 kQ below 1 MQ, and 10 kQ above 1 MQ. System noise limits the minimum measurable impedance to 10 kQ.
7 -
For neurophysiological recordings, a low-noise preamplifier with a guarded input shield was constructed for each of the sites on the probe. In addition, injection of a 1 kHz constant current signal ( 1 nA p-p) provided a means of monitoring probe impedance in vivo. Total system noise was less than 5 pV rms, referred to the input, over a 100 Hz-10 kHz bandwidth. The spinal cords in the vicinity of the dorsal column nuclei (DCN) of chloralose anesthetized rats were used for our initial evaluation of the neural recording properties of these probes for three reasons: 1) A technique to stabilize the spinal cord for acute recordings free of motion artifacts had already been established [ 111. 2) The DCN are superficial structures which are reachable with the four-site probe design. 3) DCN cells can be activated by electrical stimulation of the dorsal columns and
I
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 38, NO. I. JANUARY 1991 0 20 so oad
-
01 0
100 200 300 400 500 600 700 800 900 Elapsed time (hi1
Fig. 8. Impedance at 1 .O kHz during a 950 h immersion in physiological saline solution, for each site of a four-site probe.
>-SITE 1
IMPEDANCE 1 9 Mn
DEPTH
Tip detail
1113 p
Fig. 10. Plan views of the six-site probe
U
3
21MO
112@
most often has been reinserted into the spinal cord eight times for a total duration of 20 h, and still was able to record neural activity. In a few cases, cells that were presumably located in nucleus gracilis, were identified based on their superficial location, their response to electrical stimulation of the dorsal column, and the presence of a cutaneous receptive field on the ipsilateral leg responsive to gentle touching of the skin.
V . SIX-SITEPROBEDESIGNAND FABRICATION EXPANDED SWEEP SITE
ni
L
as,",
Fig. 9 . Simultaneous neural recordings from three sites of a four-site probe. These spinal cord recordings were made in the vicinity of the dorsal column nuclei of chloralose anesthetized rat. Impedance (at 1 kHz) and approximate depth of recording sites are indicated; the bottom figure shows ten superimposed sweeps of a discriminated action potential from site 1 .
physiologically identified by the presence of a cutaneous receptive field. Simultaneous recordings from three sites of different neurons are shown in Fig. 9. Action potential signals greater than 100 pV peak-to-peak were often recorded. In addition, the background noise was usually less than 20 p V . We have been able to hold single cells for several hours without any significant change in the action potential shape. The probe we have used
-T
Recently, a version of the microprobe has been developed that has six electrode sites. While conceptually the same as the four-site probe, the new design differs from its predecessor in both configuration and details of construction. These changes reflect design modifications in the shank size and probe spacing that were required for acute neurophysiolgical experiments in rat and cat, as well as to implement significant improvements in the probe fabrication process. Fig. 10 shovCfS a plan view of the six-site microprobe. Compared with the previous design, the new probe has a longer shaft, a narrower maximum width, and more closely spaced recording sites in order to allow neural recordings from deeper structures in the brain. Several significant improvements in fabrication technique have been employed for the six-site microprobe design. The major processing steps are indicated in Fig. 11. Four photolithographic masks are required to manufacture the probes, and there are 12 probes on each 1 in x 1 in x 15 pm molybdenum substrate. All processing is performed on a single side of the substrate, eliminating the need for the front-to-back pattern alignment that was required in four-site versions of the probe.
1-
11
13
BLUM et al.: MICROPROBES FOR NEURAL RECORDINGS
0
22pm
Mask step 1
polyimide
I
O51unAu
CriAu
Pattern Bonom dielectric layer
I
Wandow in polyimide 1 -
I
I
Mask step 3 Pattern TOP dielectric layer
Au electrode
1
Top polyimide
@
Bottom polyimide
Mask step 4 Pattern Probe etch
CriAu All
Fig. 11. Schematic illustration of major steps of fabrication sequence of the six-site probe.
The upper and lower dielectric layers are formed from Selectilux HTR3- 100 light sensitive polyimide precursor, manufactured by EM Industries, Inc. This material can be directly patterned by photolithography, which simplifies the processing required to form the layers. Conventional photoresist, rather than a layer of gold, is used as the etch mask during the final molybdenum etch that separates the probes from the substrate, thus preventing the formation of the metal flaps described earlier. Different molybdenum etchants are also employed in the fabrication process. Etching the molybdenum foil in a one-toone solution of hyrochloric acid and 30% hydrogen peroxide produces a much smoother and more uniform initial substrate surface than using the acid etch previously described. A solution containing H,SO,, HNO, and H 2 0 ( 1 : 1 : 5 ) also greatly reduces the time required to etch through the substrate and free the probes. Micrographs of the six-site probe are shown in Fig. 12. The mounting fixture for the six-site microprobe has been modified in order to provide improved visibility during insertion of the probe into deep neural structures. The bonding pad end of the probe is attached to a ceramic support which is inserted into a stainless steel tube that is 1.O in long and 0.064 in in diameter. The 0.005 in wires which are attached to the bonding pads are passed through the tube to external connecting pins.
VI. CONCLUSIONS Four-site microprobes have been constructed on a molybdenum substrate and were used successfully to measure single unit neural activity in several acute preparations. New, six-site microprobes using a somewhat improved fabrication technology are currently undergoing electrical testing and characterization. The probes are relatively easy to construct using standard fine geometry, microcircuit techniques. Failure of the four-site
(b) Fig. 12. Electron micrographs of (a) the lower portion of the six-site probe shank and (b) the lowest two electrode sites.
probes after 500-600 h in a saline bath appears to be due to water permeation of the polyimide insulating layer. Significant improvement in the longevity of the four-site probes was achieved by increasing the thickness of the polyimide layers from 1 to 1.3 pm, and it is anticipated the 2.2 pm thick layers of the six-site probes will result in further increases in the probe’s immersion lifetime. Problems still to be solved relate to insulation and encapsulation of the bonding pads and connecting leads in an environment that is hostile to the fragile structures that must be used for increased longevity of brain implants. ACKNOWLEDGMENT The authors are pleased to acknowledge the contributions of M. B. Bender, W. C. Denny, and K. J. Mach for probe fabrication and assembly; E. W. Akeyson for valuable discussions of neurophysiological testing requirements and for the in vivo testing of the probes; and R. J . Johns, J. T. Massey, and V. B. Mountcastle for their encouragement and technical advice. REFERENCES [I] J. E. Rose and V. B. Mountcastle, “Activation of single neurons in the tactile thalamic region of the cat in response to a transient peripheral stimulus,” Bull. The Johns Hopkins Hospital, vol. 94, pp. 238-282, 1954. [2] J . Kruger and M. Bach, “Simultaneous recording of 30 microelectrodes in monkey visual cortex,” Exp. Brain Res., vol. 41, pp. 191-194, 1981. [3] H. J. Reitbock and G . Werner, “Multielectrode recording system for the study of spatio-temporal activity patterns of neurons in the central nervous system,” Experienria, vol. 39, p. 339, 1983. [4] 0. J . Prohaska er a l . , “Thin-film multiple electrode probes; Possibilities and limitations,” IEEE Trans. Biomed. Eng., vol. BME-33, pp. 223-229, 1986.
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 38, NO. I , JANUARY 1991
14
151 K. L. Drake, K. D. Wise, J. Farraye, D. J. Anderson and S . L. BeMent, “Performance of planar multisite microprobes in recording extracellular single-unit intracortical activity,” IEEE Trans. Biomed. Eng., vol. 35, pp. 719-732, Sept. 1988. [6] H. Eichenbaum and M. Kuperstein, “Extracellular neural recording with multichannel microelectrodes,” J . Electrophysiol. Tech., vol. 13, pp. 189-209, 1986. [7] H. K. Charles, Jr., J. T. Massey, and V. B. Mountcastle, “Polyimides as insulating layers for implantable electrodes,” in Polyimides, Vol. 2 , K. L. Mittal, Ed. New York: Plenum, 1984, pp. 1139-1 159. [8] R. L. White and T. J. Gross, “An evaluation of the resistance to electrolysis of metals for use in biostimulation microprobes,” IEEE Trans. Biomed. Eng., vol. BME-21, pp. 487-490, 1974. [9] S. S. Stensaas and L. J. Stensaas, “Histopathological evaluation of materials implanted in the cerebral cortex,” Acta Neuropathologica, vol. 41, pp. 145-155, 1978. [lo] K. L. Drake, “A planar multisite microprobe for recording simultaneous single-unit cortical activity,” Tech. Rep. 179, Solid State Electron. Lab., Dep. Elect. Eng. Comput. Sci., Univ. Michigan, Ann Arbor, MI, Oct. 1987, p. 27. 1111 E. W. Akeyson, M. M. Kneupfer, and L. P. Schramm, “Involvement of nucleus raphe magnus in the supraspinal control of splanchnic input to the thoracic spinal cord of rat,” Soc. Neurosci. Abst., vol. 12, p. 1057, 1986. [12] N. A. Blum and H. K. Charles, Jr., “Multi-microelectrode probe for neurophysiological experiments,” in Proc. 14th Northeast Bioeng. Conf., J. R. Lacourse, Ed., IEEE Pub. No. 88CHZ666-6, 1988.
Norman A. Blum (M’81) received the A.B. degree in 1954 from Harvard College, Cambridge, MA, and the M S . and Ph.D. degrees in physics from Brandeis University, Waltham, MA, in 1959 and 1964, respectively He did postdoctoral research on magnetism and Mossbauer spectroscopy at the MIT Frances Bitter National Magnet Laboratory, where he was also a member of the technical staff from 1960 to 1966 and a Visiting Scientist until 1970 After graduating College, he served as a Commissioned Officer in the U S Navy for three years and then worked at the Avco Research and Advanced Development Division in Massachusetts while attending Graduate School until he joined the Bitter Laboratory From 1966 until 1970 he was a Staff Scientist at the NASA Electronics Research Center in Cambridge, Massachusetts. In 1970 he joined the staff of The Johns Hopkins University, Applied Physics Laboratory, first in the Milton S Eisenhower Research Center where he was a Senior Member of the Professional Staff until 1984, and subsequently in the Microelectronics Group where he now is a member of the Principal Professional Staff and Supervisor of the Substrate Processing Section. He has performed research in optical and magnetic properties of matter, thin film physics, and photovoltaics, he has authored over 50 publications in these fields Dr. Blum is a member of APS, AVS, and ISHM.
Bliss G. Carkhuff received the Associate degree in electronics technology from the Ryder Technical Institute, Allentown, PA, in 1974, and has taken additional course work at the Milwaukee School of Engineering and at the Johns Hopkins University, Baltimore, MD. He is an Engineering Assistant at The Johns Hopkins University, Applied Physics Laboratory in the Microelectronics Group. He designs and implements computer-aided test systems to provide electrical test support in the- areas of hybrid and microelectronics manufacturing, biomedical implants, neuroscience research, space radiation effects on semiconductor devices, and studies of advanced ceramics.
-
-
7
Harry K. Charles, Jr., (S’66-M’73-SM’83) received the B.S.E.E. degree from Drexel University, Philadelphia, PA, and the Ph.D. degree in electrical engineering from The Johns Hopkins University, Baltimore, MD, in 1967 and 1972, respectively. In 1973, he joined the Microelectronics Group at the Applied Physics Laboratory (APL) of The Johns Hopkins University where he held various engineering and managerial positions including &our, SuDervisor. Since Januarv 1989, he has been leading thcEngingering’ and Fabrication Branch which is responsible for all electronic and mechanical design and fabrication at APL. He has been engaged in electronic research and advanced microcircuit packaging for 17 years. His current research interests include the modeling and experimental analysis of high speed interconnection and packaging structures for VLSI/VHSIC circuit applications. He has published over 100 papers in these and other fields. Dr. Charles is a member of the American Physical Society, ASM International, the American Ceramic Society, and the International Society for Hybrid Microelectronics (ISHM). In 1987, he was selected for an ISHM Technical Achievement Award. In 1989, he received Maryland’s Distinguished Young Engineer Award. He also received the John Hopkins University Whiting School of Engineering award for Excellence in Teaching for contributions to the Continuing Professional Programs.
Richard L. Edwards received the B.S. degree in physics from the University of Maryland, College Park, in 1983, and has taken several specialized short courses in various subjects related to microlectronics processing. He was employed by the Solarex Corporation from 1975 until 1981 where he worked on the development and manufacturing of silicon solar cells. He continued his work in photovoltaics research on joining the Department of Solid State Physics at COMSAT Laboratories in 1982 and later worked in the Microelectonics Division of COMSAT developing fabrication technology for gallium arsenide monolithic microwave integrated circuits. In 1986 he joined the Microelectronics Group at The John Hopkins University, Applied Physics Laboratory, where he is a Process Engineer working on microelectronics process development, particularly as applied to microwave and superconductor circuit fabrication, and to advancing the technology of sensors and biomedical devices.
Richard A. Meyer (S’66-M’69) received the B.S. degree in electrical engineering from Valparaiso University, Valparaiso, IN, and the M.S. degree in applied physics from The Johns Hopkins University, Baltimore, MD. He is a Principal Staff Engineer at The Johns Hopkins University, Applied Physics Laboratory, and Associate Professor of both Neurosurgery and Biomedical Engineering at The Johns Hopkins University School of Medicine. He is the Program Director of an interdisciplinary research program applying advanced technology to problems in the neurosciences. His research interests are in the psychophysical and neurophysiological mechanisms of pain. Mr. Meyer is a member of Tau Beta Pi, the International Association for the Study of Pain, the American Pain Society, and the Society for Neuroscience.
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 38, NO. I , JANUARY 1991
Multisite Microprobes for Neural Recordings Norman A . Blum, Member, IEEE, Bliss G . Carkhuff, Harry K. Charles, Jr., Senior Member, IEEE, Richard L. Edwards, and Richard A. Meyer, Member, lEEE
Abstract-Multisite, passive microprobes have been developed to allow simultaneous recording of action potential activity from multiple neurons at different locations in the brain. The microprobes were fabricated using standard integrated circuit techniques. The probe is a planar structure that consists of gold electrodes sandwiched between two polyimide dielectric layers and bonded to a molybdenum structural support. Windows in the top dielectric layer expose the electrode sites and bonding pads. In two distinct versions of the probe four or six recordings sites, respectively, of approximately 25 pm2 are arranged on a dagger-shaped structure which can penetrate the pia. The bonding pads and interconnect wires at the probe head are entirely encapsulated in a tubular fixture that is packed with silicone RTV and sealed with epoxy to protect the interconnections from contact with body fluids. The site impedances at 1 kHz are typically between 2 and 4 MQ. Probe lifetimes for continuous immersion in physiological saline solution, as measured by impedance, have exceeded 750 h. The failure mechanism is believed to be due to moisture and ion absorption in the top dielectric layer. In acute neurophysiological experiments using the four site probes, action potential activity was recorded from physiologically identified neurons in the dorsal column nuclei of anesthetized rat.
I. INTRODUCTION
F
OR more than 30 years, fine-tipped microelectrodes have been inserted into the brain to record extracellular action potentials generated by individual neurons [ 11. This method of observing single neurons in succession has contributed to many significant advances in understanding the function of the brain. The applications of microelectronic techniques to the development of multielement microelectrodes makes it possible to record simultaneously from many different neurons. Multielement microelectrodes have many advantages over the conventional single site electrodes, including: 1) For single site recordings, functional relationships between cells must be inferred from the population data obtained from successive experiments. Simultaneous recordings from many neurons allow functional relationships to be assessed directly. 2) For single site recordings, a population of data is obtained by repeating the stimulus and/ or behavior paradigm while recording from different cells. This is laborious both for the experimenter and for the animals. Multielectrode recordings can dramatically increase the number of neurons per experiment and thus reduce the number of experiments and, perhaps more importantly, the number of animals necessary to collect data. 3) In many experiments, the evoked potential linked to the stimulus results in a noise signal which makes single unit recording difficult. Differential recording from neighboring pairs of the multielement electrodes allows for rejection of the noise signal common to the two electrodes. 4) Manuscript received October 16, 1989; revised May 4, 1990. This work was supported under Public Health Service Grant NS07226. A preliminary version of this paper was presented at the Fourteenth Northeast Bioengineering Conference [ 121. The authors are with the Applied Physics Laboratory, The Johns Hopkins University, Laurel, MD 20723. IEEE Log Number 9040624.
The fixed spacing of electrodes facilitates studies of topographic organization and functional relationships. 5) Prosthetic devices controlled by the CNS will require muiltiple recording and stimulating sites. Two major approaches to the development of multielement electrodes have been taken. The first, historically, was the fabrication of bundles of separate, single-wire microprobes that had either a fixed three-dimensional configuration [ 2 ] or a fixed twodimensional configuration with adjustable depth [ 3 ] . The capability of these arrays to record simultaneously from more than one neuron has been demonstrated. The second approach has utilized the application of thin film microelectronics technology for fabrication of multielectrode microprobes. Potential technological advantages of the use of thin film technology to fabricate microprobes with a linear array of electrode sites are: 1) a high degree of reproductibility once the design and processing sequence are developed, 2 ) a precise knowledge of the spatial distribution of electrode sites, 3 ) a high “packaging density” of electrode sites for a given implanted volume, 4) the possibility of incorporating the interface circuitry directly on the microprobe base, and 5 ) the distribution of electrodes in a specifiable geometric pattem. In addition to the Hopkins team, there are a number of other groups presently developing planar multielectrode microprobes. The Vienna probe [4]is perhaps the most advanced being offered on a commercial basis (Otto Sensors Corp., Cleveland, OH) with several variations for different aspects of neurophysiological research. It is made on a rigid glass substrate with a nominal thickness of 100 pm and uses low pressure CVD silicon nitride for insulation. The Michigan probe [5] is made on a silicon substrate which can be thinned in the implanted region to 15 pm and can have an almost arbitrary shape. Several probe designs have been fabricated, including ones with on-board, integrated electronics. A combination of SiO, and Si,N, are used as passivation or insulation layers. This probe appears to be the most technologically advanced sensor because of its integrated electronics and the sophisticated semiconductor processing necessary for its manufacture. The MIT probe [6] is made on molybdenum (18 pm thick) with up to 24 recording sites. The molybdenum is strong, but flexible, and is not subject to brittle fractures as are glass and silicon. Various dielectrics have been used, including photoresist and silicon nitride. The Hopkins probe [7] described herein contains the strong, flexible molybdenum substrate with an organic dielectric (polyimide). The probe is approximately 19 pm thick and currently contains either four or six recording sites. The molybdenum-polyimide structure is easy to fabricate using standard microelectric technology and does not require advanced semiconductor processing techniques. The flexibility and strength of both materials make a mechanically forgiving structure that should increase the ruggedness of the probes in various applications.
0018-9294/91/0100-0068$01.OO
0 1991 IEEE
-
’I
69
BLUM et a l . : MICROPROBES FOR NEURAL RECORDINGS
11. FOUR-SITEPROBE DESIGNAND FABRICATION The four-site microprobe, which has been used for in vivo experiments, is constructed as a sandwich of four principal layers: 1) a rigid supporting substrate of molybdenum, 2) a dielectric layer of polyimide, upon which 3) a conductor layer of gold is patterned, covered by 4) a top dielectric layer of polyimide with open areas to expose the site electrodes along the shank, and with bonding pads at the head. A plan view of Fig. 1 shows the shape and dimensions of the four-site probe. In this probe, which was originally designed for superficial recordings in monkey cortex, the electrode sites are colinear and 500 pm apart. The fabrication sequence is depicted in cross-sectional view in Fig. 2. Four photolithographic masks are required to manufacture the probes, and there are four individual probes on each molybdenum substrate. The 99.99% pure molybdenum foil substrate is procured from Goodfellow Metals, Ltd., in England. The foils are in the form of 1 in x 1-in squares, rolled to a thickness of 15 pm. They are high quality, free of pinholes, and have a smooth, rolled finish. The foils are lightly polished chemically in an acid etch ( H3P04: HNO, : CH3COOH : H 2 0 16: 1 : 1 :2 ) to further improve the surface. The polymide is formed from the precursor material DuPont Pyralin, Type PI2555. This material is in a liquid form that allows the precursor (polyamic acid solution) to be spin-coated onto the substrate. The thin chrome layer provides an adhesion surface for the polyimide, which does not adhere well to gold. After heat treatment, the polyimide layer is cured (imidized) and becomes a polyimide film approximately 1.O-1.5 pm thick depending on processing details. The gold electrodes and associated conductor structure are sputter deposited to a thickness of 0.5 pm from a pure gold target. Sputtered gold has good adhesion to the previously formed polyimide, is chemically stable, and is generally considered to be compatible with brain tissue [8], [9]. The probes are separated from the surrounding molybdenum substrate by wet chemical etching using the same acid etch that was used for polishing (see above). The initial layers of gold on the front and back of the substrate serve as etch masks. The probes, however, remain attached to the substrate by three narrow tabs that are cut just prior to mounting and assembly. Probes that are acceptable for mounting are identified by visual inspection under a microscope, and by electrical testing to assure electrical isolation of the recording sites from the substrate and from each other. The present yield indicates that about 50% of the fabricated probes pass these initial tests and can be mounted, having at least three sites suitable for neural recording; about 10%have four good sites. Failed sites (impedance out of acceptable range) are usually due to open tracks or connections within the mounting assembly (impedance too high), or to overetched or delaminated dielectric (impedance too low). In a few cases too high impedance is attributed to failure to completely clear the electrode windows, or too low impedance to pinholes in the polyimide layers. The good probes are separated from the substrate frames, and 0.008 in copper wires are manually attached by gold conductive epoxy to the bonding pads. The probes are then mounted by partial insertion into a tublar, polycarbonate fixture that can be handled easily, and which has an external plug for electrical connection to the recording apparatus. The region around the bonding pads, within the tube, is packed with silicone RTV and sealed with epoxy to ensure good insulation. Two thin (0.5 pm) gold flaps, about 75 pm long, that result from undercutting of the etching masks, are removed by care-
1-
I-
2 75 mm
\\II
I
Fig. 1. Pian view of the four-site probe.
0
MO substrate
rubrlrate
15um
1 0 ~n X 1 0
tn. X
0.0006 an 115rml
c Sputrered AuICr layer substrate +Sputtered
1 p m Au-
@ away
/
Mask step No 1 Pattern
t @
Au laver
AuiCr on top Au on bottom
t
1 to 1.5pm polyimide
Mask step No 2 Pattern
Polyimide msulatmn layer
substrate
1 to 1 5 p m polyimide
Pattern
Gold conductor
Wlndow !n polyimide Mask step No 4 Pattern
Polvimide top layer
Top polyimide @
Au electrode
AuICr
Bottom Dolvimide
Etched out probe
Fig. 2 . Schematic illustration of the major steps in the fabrication sequence of the four-site probe.
fully dipping the tip of the assembled probe in a gold etchant solution ( K I :I, : H,O - 4 : 1 : 4; by weight) while observing the operation under a microscope. Micrographs of the four site probe are shown in Figs. 3 and 4. ELECTRICALTESTING For in vitro testing, each assembled probe is mounted in a test fixture that is located inside an electromagnetically screened box. The probe is immersed beyond the uppermost electrode site in a 0.9%saline aqueous solution. When an alternating test current of 10 nA peak to peak is passed through one of the probe electrodes and the saline solution, a potential develops at the electrode-electrolyte interface. This potential is used to determine the probe site impedance. A computer controlled test system has been constructed to make impedance versus frequency measurements for each recording site on the probe. The system, which is shown sche-
1
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 38, NO. I , JANUARY 1991
70
(b) Fig. 3. Electron micrographs of (a) the entire shank and (b) the tip region of the four-site probe.
matically in Fig. 5, can make single frequency impedance measurements on any one of the recording sites, measure all recording sites at a single frequency in a scan mode, or measure all recording sites at multiple frequencies at preselected time intervals. A low distortion signal generator provides precision frequency and amplitude test signals to each of the stimulus/measurement units (SMU) in the system. The SMU electrical schematic diagram is shown in Fig. 6. Each SMU contains a voltage controlled current pump, a high impedance buffer and a filtered 5X amplifier. The current pumps have a transconductance of 2 p S providing a 10 nA peak to peak output current for a 5 mV peak to peak input signal. The high impedance buffer (1000 MQ) isolates each probe site from loading effects and provides input signal guarding to minimize cable capacitances and increase measurement bandwidth. The output of each SMU is multiplexed to a single amplifier that provides an additional 26 dB of gain and high frequency filtering. The final signal is measured using an ac rms voltmeter. Multiplexing control and system power are provided through the switch/control unit.
Fig. 4. Electron micrographs of (a) top surface view of entire four-site probe, (b) upper and lower dielectric layers near the probe tip, and (c) an electrode in a dielectric window.
The signal generator, voltmeter, switch/control unit, and +_ 15 V power supplies are connected to a desk-top computer via the IEEE 488 Bus. The computer software permits complete auto-
"I
71
BLUM et al.: MICROPROBES FOR NEURAL RECORDINGS
Signal return Return signal output bus
ground
0 9% by weight saline soluton
Final output signal level
Platinumwire reference electrcde
Ouler shield and signal
/Enclosure
wall
0 011 10
Switchiconlral
1
I
100
1000
11
Frequency I H r i
SMA
~
Fig. 7. Initial impedance values of different frequencies for each 25 X 25 pm site of a four-site probe; the 10 x 10 pm and 50 X 50 pm data are from [lo].
/Feedthrwgh I
Computer
I 1I
II
Double clad EP I - alass
Fig. 5 . Block diagram of test system for probe impedance measurements.
Signal Return
ti,-
outpu1
System Ground
Probe Site 2
StimuluslMeasurement Unit
The impedances at each site are recorded at frequencies between 100 Hz and 10 kHz in a l , 2, 5 sequence for a total time of at least 60 min in order to identify stabilized probe site impedances. From each production batch of four substrates ( 16 probes), one probe that has four good sites is selected and further tested by remaining in the saline solution until at least one site shows signs of failure, as denoted by rapidly decreasing impedance. This has proved to be indicative of the average performance of the other probes in the same batch that have site impedances in the acceptable range (2-4 M Q ) on initial (60 min) testing. Initial impedance versus frequency measurements for a probe with 25 x 25 pm recording sites are shown in Fig. 7 . These data compare well with those reported by Drake [lo] for 10 x 10 pm and 50 x 50 pm sites. Fig. 8 shows impedance measurements made on this probe during a 950 h immersion in a physiological saline bath. Impedances of three probe sites were stable for longer than 600 h, and declined to half their initial impedance values after about 750 h. Post failure electron micrograph studies of long-lived probes appear to rule out pin hole or delamination of the structure as explanations for the changes in impedance, which are most likely due to the effects of water absorption in the dielectric layers.
Fig. 6 . Schematic diagram of stimuludmeasurement unit.
IV. NEURALRECORDINGRESULTS matic control of test signals, SMU multiplexing and probe electrode selection, and data collection and analysis. The system can run unattended for long periods of time which is helpful when conducting life tests. Calibration of the system is done by placing precision 1% resistors in the test holder in place of the probe and correlating the measured potentials for each resistor with an impedance value. The system is calibrated from 100 kQ to 20 MQ with a sensitivity of 1 mV peak to peak per 1 kQ. Within the calibration range, measurement accuracy is 1-2%, repeatability is 12 % , and resolution is 1 kQ below 1 MQ, and 10 kQ above 1 MQ. System noise limits the minimum measurable impedance to 10 kQ.
7 -
For neurophysiological recordings, a low-noise preamplifier with a guarded input shield was constructed for each of the sites on the probe. In addition, injection of a 1 kHz constant current signal ( 1 nA p-p) provided a means of monitoring probe impedance in vivo. Total system noise was less than 5 pV rms, referred to the input, over a 100 Hz-10 kHz bandwidth. The spinal cords in the vicinity of the dorsal column nuclei (DCN) of chloralose anesthetized rats were used for our initial evaluation of the neural recording properties of these probes for three reasons: 1) A technique to stabilize the spinal cord for acute recordings free of motion artifacts had already been established [ 111. 2) The DCN are superficial structures which are reachable with the four-site probe design. 3) DCN cells can be activated by electrical stimulation of the dorsal columns and
I
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 38, NO. I. JANUARY 1991 0 20 so oad
-
01 0
100 200 300 400 500 600 700 800 900 Elapsed time (hi1
Fig. 8. Impedance at 1 .O kHz during a 950 h immersion in physiological saline solution, for each site of a four-site probe.
>-SITE 1
IMPEDANCE 1 9 Mn
DEPTH
Tip detail
1113 p
Fig. 10. Plan views of the six-site probe
U
3
21MO
112@
most often has been reinserted into the spinal cord eight times for a total duration of 20 h, and still was able to record neural activity. In a few cases, cells that were presumably located in nucleus gracilis, were identified based on their superficial location, their response to electrical stimulation of the dorsal column, and the presence of a cutaneous receptive field on the ipsilateral leg responsive to gentle touching of the skin.
V . SIX-SITEPROBEDESIGNAND FABRICATION EXPANDED SWEEP SITE
ni
L
as,",
Fig. 9 . Simultaneous neural recordings from three sites of a four-site probe. These spinal cord recordings were made in the vicinity of the dorsal column nuclei of chloralose anesthetized rat. Impedance (at 1 kHz) and approximate depth of recording sites are indicated; the bottom figure shows ten superimposed sweeps of a discriminated action potential from site 1 .
physiologically identified by the presence of a cutaneous receptive field. Simultaneous recordings from three sites of different neurons are shown in Fig. 9. Action potential signals greater than 100 pV peak-to-peak were often recorded. In addition, the background noise was usually less than 20 p V . We have been able to hold single cells for several hours without any significant change in the action potential shape. The probe we have used
-T
Recently, a version of the microprobe has been developed that has six electrode sites. While conceptually the same as the four-site probe, the new design differs from its predecessor in both configuration and details of construction. These changes reflect design modifications in the shank size and probe spacing that were required for acute neurophysiolgical experiments in rat and cat, as well as to implement significant improvements in the probe fabrication process. Fig. 10 shovCfS a plan view of the six-site microprobe. Compared with the previous design, the new probe has a longer shaft, a narrower maximum width, and more closely spaced recording sites in order to allow neural recordings from deeper structures in the brain. Several significant improvements in fabrication technique have been employed for the six-site microprobe design. The major processing steps are indicated in Fig. 11. Four photolithographic masks are required to manufacture the probes, and there are 12 probes on each 1 in x 1 in x 15 pm molybdenum substrate. All processing is performed on a single side of the substrate, eliminating the need for the front-to-back pattern alignment that was required in four-site versions of the probe.
1-
11
13
BLUM et al.: MICROPROBES FOR NEURAL RECORDINGS
0
22pm
Mask step 1
polyimide
I
O51unAu
CriAu
Pattern Bonom dielectric layer
I
Wandow in polyimide 1 -
I
I
Mask step 3 Pattern TOP dielectric layer
Au electrode
1
Top polyimide
@
Bottom polyimide
Mask step 4 Pattern Probe etch
CriAu All
Fig. 11. Schematic illustration of major steps of fabrication sequence of the six-site probe.
The upper and lower dielectric layers are formed from Selectilux HTR3- 100 light sensitive polyimide precursor, manufactured by EM Industries, Inc. This material can be directly patterned by photolithography, which simplifies the processing required to form the layers. Conventional photoresist, rather than a layer of gold, is used as the etch mask during the final molybdenum etch that separates the probes from the substrate, thus preventing the formation of the metal flaps described earlier. Different molybdenum etchants are also employed in the fabrication process. Etching the molybdenum foil in a one-toone solution of hyrochloric acid and 30% hydrogen peroxide produces a much smoother and more uniform initial substrate surface than using the acid etch previously described. A solution containing H,SO,, HNO, and H 2 0 ( 1 : 1 : 5 ) also greatly reduces the time required to etch through the substrate and free the probes. Micrographs of the six-site probe are shown in Fig. 12. The mounting fixture for the six-site microprobe has been modified in order to provide improved visibility during insertion of the probe into deep neural structures. The bonding pad end of the probe is attached to a ceramic support which is inserted into a stainless steel tube that is 1.O in long and 0.064 in in diameter. The 0.005 in wires which are attached to the bonding pads are passed through the tube to external connecting pins.
VI. CONCLUSIONS Four-site microprobes have been constructed on a molybdenum substrate and were used successfully to measure single unit neural activity in several acute preparations. New, six-site microprobes using a somewhat improved fabrication technology are currently undergoing electrical testing and characterization. The probes are relatively easy to construct using standard fine geometry, microcircuit techniques. Failure of the four-site
(b) Fig. 12. Electron micrographs of (a) the lower portion of the six-site probe shank and (b) the lowest two electrode sites.
probes after 500-600 h in a saline bath appears to be due to water permeation of the polyimide insulating layer. Significant improvement in the longevity of the four-site probes was achieved by increasing the thickness of the polyimide layers from 1 to 1.3 pm, and it is anticipated the 2.2 pm thick layers of the six-site probes will result in further increases in the probe’s immersion lifetime. Problems still to be solved relate to insulation and encapsulation of the bonding pads and connecting leads in an environment that is hostile to the fragile structures that must be used for increased longevity of brain implants. ACKNOWLEDGMENT The authors are pleased to acknowledge the contributions of M. B. Bender, W. C. Denny, and K. J. Mach for probe fabrication and assembly; E. W. Akeyson for valuable discussions of neurophysiological testing requirements and for the in vivo testing of the probes; and R. J . Johns, J. T. Massey, and V. B. Mountcastle for their encouragement and technical advice. REFERENCES [I] J. E. Rose and V. B. Mountcastle, “Activation of single neurons in the tactile thalamic region of the cat in response to a transient peripheral stimulus,” Bull. The Johns Hopkins Hospital, vol. 94, pp. 238-282, 1954. [2] J . Kruger and M. Bach, “Simultaneous recording of 30 microelectrodes in monkey visual cortex,” Exp. Brain Res., vol. 41, pp. 191-194, 1981. [3] H. J. Reitbock and G . Werner, “Multielectrode recording system for the study of spatio-temporal activity patterns of neurons in the central nervous system,” Experienria, vol. 39, p. 339, 1983. [4] 0. J . Prohaska er a l . , “Thin-film multiple electrode probes; Possibilities and limitations,” IEEE Trans. Biomed. Eng., vol. BME-33, pp. 223-229, 1986.
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 38, NO. I , JANUARY 1991
14
151 K. L. Drake, K. D. Wise, J. Farraye, D. J. Anderson and S . L. BeMent, “Performance of planar multisite microprobes in recording extracellular single-unit intracortical activity,” IEEE Trans. Biomed. Eng., vol. 35, pp. 719-732, Sept. 1988. [6] H. Eichenbaum and M. Kuperstein, “Extracellular neural recording with multichannel microelectrodes,” J . Electrophysiol. Tech., vol. 13, pp. 189-209, 1986. [7] H. K. Charles, Jr., J. T. Massey, and V. B. Mountcastle, “Polyimides as insulating layers for implantable electrodes,” in Polyimides, Vol. 2 , K. L. Mittal, Ed. New York: Plenum, 1984, pp. 1139-1 159. [8] R. L. White and T. J. Gross, “An evaluation of the resistance to electrolysis of metals for use in biostimulation microprobes,” IEEE Trans. Biomed. Eng., vol. BME-21, pp. 487-490, 1974. [9] S. S. Stensaas and L. J. Stensaas, “Histopathological evaluation of materials implanted in the cerebral cortex,” Acta Neuropathologica, vol. 41, pp. 145-155, 1978. [lo] K. L. Drake, “A planar multisite microprobe for recording simultaneous single-unit cortical activity,” Tech. Rep. 179, Solid State Electron. Lab., Dep. Elect. Eng. Comput. Sci., Univ. Michigan, Ann Arbor, MI, Oct. 1987, p. 27. 1111 E. W. Akeyson, M. M. Kneupfer, and L. P. Schramm, “Involvement of nucleus raphe magnus in the supraspinal control of splanchnic input to the thoracic spinal cord of rat,” Soc. Neurosci. Abst., vol. 12, p. 1057, 1986. [12] N. A. Blum and H. K. Charles, Jr., “Multi-microelectrode probe for neurophysiological experiments,” in Proc. 14th Northeast Bioeng. Conf., J. R. Lacourse, Ed., IEEE Pub. No. 88CHZ666-6, 1988.
Norman A. Blum (M’81) received the A.B. degree in 1954 from Harvard College, Cambridge, MA, and the M S . and Ph.D. degrees in physics from Brandeis University, Waltham, MA, in 1959 and 1964, respectively He did postdoctoral research on magnetism and Mossbauer spectroscopy at the MIT Frances Bitter National Magnet Laboratory, where he was also a member of the technical staff from 1960 to 1966 and a Visiting Scientist until 1970 After graduating College, he served as a Commissioned Officer in the U S Navy for three years and then worked at the Avco Research and Advanced Development Division in Massachusetts while attending Graduate School until he joined the Bitter Laboratory From 1966 until 1970 he was a Staff Scientist at the NASA Electronics Research Center in Cambridge, Massachusetts. In 1970 he joined the staff of The Johns Hopkins University, Applied Physics Laboratory, first in the Milton S Eisenhower Research Center where he was a Senior Member of the Professional Staff until 1984, and subsequently in the Microelectronics Group where he now is a member of the Principal Professional Staff and Supervisor of the Substrate Processing Section. He has performed research in optical and magnetic properties of matter, thin film physics, and photovoltaics, he has authored over 50 publications in these fields Dr. Blum is a member of APS, AVS, and ISHM.
Bliss G. Carkhuff received the Associate degree in electronics technology from the Ryder Technical Institute, Allentown, PA, in 1974, and has taken additional course work at the Milwaukee School of Engineering and at the Johns Hopkins University, Baltimore, MD. He is an Engineering Assistant at The Johns Hopkins University, Applied Physics Laboratory in the Microelectronics Group. He designs and implements computer-aided test systems to provide electrical test support in the- areas of hybrid and microelectronics manufacturing, biomedical implants, neuroscience research, space radiation effects on semiconductor devices, and studies of advanced ceramics.
-
-
7
Harry K. Charles, Jr., (S’66-M’73-SM’83) received the B.S.E.E. degree from Drexel University, Philadelphia, PA, and the Ph.D. degree in electrical engineering from The Johns Hopkins University, Baltimore, MD, in 1967 and 1972, respectively. In 1973, he joined the Microelectronics Group at the Applied Physics Laboratory (APL) of The Johns Hopkins University where he held various engineering and managerial positions including &our, SuDervisor. Since Januarv 1989, he has been leading thcEngingering’ and Fabrication Branch which is responsible for all electronic and mechanical design and fabrication at APL. He has been engaged in electronic research and advanced microcircuit packaging for 17 years. His current research interests include the modeling and experimental analysis of high speed interconnection and packaging structures for VLSI/VHSIC circuit applications. He has published over 100 papers in these and other fields. Dr. Charles is a member of the American Physical Society, ASM International, the American Ceramic Society, and the International Society for Hybrid Microelectronics (ISHM). In 1987, he was selected for an ISHM Technical Achievement Award. In 1989, he received Maryland’s Distinguished Young Engineer Award. He also received the John Hopkins University Whiting School of Engineering award for Excellence in Teaching for contributions to the Continuing Professional Programs.
Richard L. Edwards received the B.S. degree in physics from the University of Maryland, College Park, in 1983, and has taken several specialized short courses in various subjects related to microlectronics processing. He was employed by the Solarex Corporation from 1975 until 1981 where he worked on the development and manufacturing of silicon solar cells. He continued his work in photovoltaics research on joining the Department of Solid State Physics at COMSAT Laboratories in 1982 and later worked in the Microelectonics Division of COMSAT developing fabrication technology for gallium arsenide monolithic microwave integrated circuits. In 1986 he joined the Microelectronics Group at The John Hopkins University, Applied Physics Laboratory, where he is a Process Engineer working on microelectronics process development, particularly as applied to microwave and superconductor circuit fabrication, and to advancing the technology of sensors and biomedical devices.
Richard A. Meyer (S’66-M’69) received the B.S. degree in electrical engineering from Valparaiso University, Valparaiso, IN, and the M.S. degree in applied physics from The Johns Hopkins University, Baltimore, MD. He is a Principal Staff Engineer at The Johns Hopkins University, Applied Physics Laboratory, and Associate Professor of both Neurosurgery and Biomedical Engineering at The Johns Hopkins University School of Medicine. He is the Program Director of an interdisciplinary research program applying advanced technology to problems in the neurosciences. His research interests are in the psychophysical and neurophysiological mechanisms of pain. Mr. Meyer is a member of Tau Beta Pi, the International Association for the Study of Pain, the American Pain Society, and the Society for Neuroscience.
