IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. BME-28, NO.

3, MARCH 1981 301

mean rate, duration, and total number of spikes for each vigi- Microthermocouple for Soft Tissue Temperature
lance state. These data serve as a base for statistical analysis of Determination
global rates for all cells in a given state.
For intervals with stimulation (STINT), verification of stim- ERIC J. GUILBEAU AND BRUCE I. MAYALL
ulus timing is performed to determine if any stimuli synchro-
nizing pulses are absent or if synchronization artifacts have Abstract-Platinum-tellurium microthermocouples with a sensing junc-
been introduced during playback. This test is especially im- tion of approximately 1.0 ,um were constructed by vapor depositing
portant for stimulus trains, since deletion of sync pulses or tellurium on the exposed conical tip of a glass insulated platinum wire.
addition of synchronizing artifacts can upset determination of The thermoelectric power, linearity, and hysteresis of the microthermo-
the serial position of a given stimulus in a train. The routine couples were determined by comparing the output of the microthermo-
calculates the time intervals between stimuli and, based on a couple with that of an iron-constantan macrothermocouple. The micro-
previous keyboard entry of the number of stimuli per train, thermocouple response was linear and exhibited no hysteresis over the
constructs a timing matrix. Examination of the matrix by the temperature range 27-100°C. Thermoelectric power was typically
CC reveals any discrepancies in the timing pattern and sub- 300 ,uV/°C. Typical impedance was several hundred MQ. The response
routines are provided for removal by the operator of spurious time was less than 50 ms.
"stimuli,"1 or addition of missing stimuli indicators.
-DATA FILE STRUCTURE
Each file contains data for one particular project or anatom- INTRODUCTION
ical recording area. The first component of the file provides Recent emphasis in physiology has been directed toward
an index of cell numbers, times, and data locations to permit obtaining a better understanding of the function of organs by
automatic retrieval of a data segment for a particular cell and the measurement of important physiological parameters at the
time. In addition a frequency matrix gives for each cell the cellular level. This has required the development of transducers
pooled weighted mean rates for all epochs in each available that can be introduced into the environment of a single cell
vigilance state. This matrix is used for analysis of global (for without altering the activity and function of the cell. Suitable
all cells) and single-cell rate changes with respect to changes in trainsducers must be small in size relative to the cell, have a
vigilance state. pointed configuration to allow for easy penetration of micro-
The base data for each epoch are stored in consecutive file scopic structures, and have a response time rapid enough to
components, the first of which contains identification and follow the transient behavior of the signal to be measured.
timing information to permit retrieval of the data proper. The Those devices with conical tip configurations having tip
interval and sequential mean rate data are stored consecutively diameters of 50 ,im or less and a sensing surface of several
in time order in the following components of the file with a Pim2 to less than 1 gim2 are called microtransducers, or more
maximum of 4000 values per component. Each component simply, microelectrodes.
also contains the start and stop times of the data segment as Sensors capable of measuring resting membrane potential
well as the identification component location. For epochs and the action potential of single cells are widely used. Micro-
containing interval data, the sequential mean rate is computed electrodes have also been developed and used to measure
for the entire epoch and stored for future retrievals on the first oxygen tension, potassium ion activity, sodium ion activity,
occasion that it is requested. chloride ion activity, and pH [3], [7], [8], [O1], [11]. Cain
CONCLUSIONS and Welch [21 fabricated a thin-film temperature sensor for
biological measurement with a probe tip as small as 10 Jim.
The system described has achieved the aim of permitting They used a quartz substrate with a flat end surface to support
automated transfer and storage of large volumes of neural thin metallic films insulated from each other by a polymer
time-interval data with a minimum of operator intervention. coating except at the tip of the probe where the thin films
The data file organization permits rapid access to any com- overlapped to form a thermoelectric junction. When nickel
onent by specifying the start and stop times of the required and copper were used to form the thermoelectric junction,
segment and the cell number. This feature can be used for they obtained a thermoelectric EMF of 21 ,uV/ C which was
searching large cell groups for certain features or for determin- linearly dependent on temperature over the range normally
ing global characteristics of the group. Since in general, most encountered in biological measurements. Although these
data are stored in the form of interspike intervals, no informa- devices exhibited excellent characteristics, they have not been
tion is lost and later, newer techniques can be used to analyze widely used to measure temperature in biological media by
old data. In addition, since the main data analysis phase is other investigators because the polymer coating used to insulate
effected by the central computer, with all its resources includ- the thin films was Parylene C® and can only be deposited
ing interactive programming features, the researcher has the using a special process developed by Union Carbide. Although
potential to accomplish virtually any data analysis or display Parylene C has been demonstrated to be an excellent insulator
task. With simplified data transfer and storage facilities, the for microelectrode applications [6], the use of Parylene and the
researcher can retain accessibility to -his data. This "hands-on" process for its deposition is regulated by patents on materials
approach is especially valuable to research students who must the deposition process. The object of this investigation
have at least a minimum of exposure to data handling and and was to develop a microthermocouple with 1 jim tip diameter
computer techniques as part of their training. suitable for the measurement of temperature in soft tissue.
ACKNOWLEDGMENT
The author wishes to thank Dr. M. Steriade for his encourage- Manuscript received February 4, 1980; revised August 18, 1980, and
ment in this project and P. Giguere for his invaluable technical October 23, 1980.
assistance. E. J. Guilbeau is with the Department of Chemical and Bioengineering,
REFERENCES Arizona State University, Tempe, AZ 85281.
B. I. Mayall was with the Department of Chemical and Bioengineering,
[1] W. Simon, S. Blumenthal, J. H. Young, and C. Elsbeck, "The scien- Arizona State University, Tempe, AZ 85281. He is now with American
tific laboratory instrument to computer coupler," IEEE Trans. Edwards Laboratories, Santa Ana, CA 92711.
Biomed. Eng., vol. BME-27, pp. 55-57, Jan. 1980. ®Registered service mark of the Union Carbide Corporation.
0018-9294/81/0300-0301$00.75 i 1981 IEEE
302 IEEEeTRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. BME-28, NO. 3, MARCH 1981
METHODS
The microthermocouples were constructed by vapor depos-
iting tellurium on the exposed conical tip of a glass insulated
platinum wire. Thin platinum wires 25 ,um in diameter and
approximately 1.5 cm long were joined to a larger 14 gauge
tinned copper wire using conductive silver paint and were
coated with glass by the method described by Ballintijn [ 11 as
modified by Silver [101. The platinum was straightened by
gentle heating in a small flame, washed, and degreased. It was
electropolished by dipping the extreme tip just below the sur-
face of a solution of sodium nitrite prepared by diluting a sat-
urated solution with an equal volume of water and by passing a
3 kHz alternating current through the solution between -the
platinum and a carbon rod. Using this technique, a long taper
which ended in a very sharp point (1 pm diameter or less) was
imparted to the extreme end of the wire. After polishing, the
wire was again washed and degreased. The sharpened wire was
then inserted into a glass tube (Kimble Glass Company type
R-6) redrawn to an inside diameter of 1.5 mm ± 0.05 mm
and a wall thickness of 0.25 mm (Garner Glass Company, Clare-
mont, CA). This tube had previously been drawn into a
pipette with a long straight tip approximately 100 pm in
diameter. The platinum wire was pushed down into the glass
tube until the copper wire seated against the taper at the point
where the pipette tip began to form. In this position the
platinum wire extended into the lumen of the 100 pm pipette
tip. The copper wire was fixed to the glass with adhesive, and
the glass tube was placed in a micromanipulator with the point
vertically downward, and a small weight (200 mg) was at-
tached to the end of the pipette tip. The glass pipette tip was
moved by the manipulator into a small heating loop and was
observed through a dissecting microscope. By gradually
increasing the current through the heating loop the glass
pipette tip surrounding the sharpened platinum wire was
heated until it began to melt. As it melted, it was drawn down
by the weight and thinned in the process. As the melted glass
flowed down the platinum wire, the heating loop was lowered
at the same rate to keep abreast of the melting front so that
the relative position of the melted glass boundary and heating
loop did not change. This resulted in a very thin glass coating
on the platinum wire (less than 1.0 pum thick and 0.5-1.0 cm
long). The process was continued until the glass coating
reached the taper on the extreme tip of the platinum wire.
Then the conical point of the platinum wire was coated by
gradual adjustment of the heat. The glass coating could be
extended to over three quarters of the distance down the
conical point past which it suddenly separated and in so
doing coated all but the very tip of the platinum, leav-
ing an exposed cone of platinum, the base of which was
approximately 1 pum in diameter. This technique is essentially
the same as described by Silver (10) in the construction of
oxygen microelectrodes, and it has been used successfully by
many investigators to produce glass insulated platinum wires
with exposed platinum tips of 1 pm or less. The relative posi-
tion of the glass insulated platinum tip, the tinned copper
wire, and the unmelted glass pipette barrel are the same as in
the Silver oxygen microelectrode [ 101 .
These glass insulated platinum wires (oxygen microelec-
trodes) served as the basic substrate for fabrication of the
microthermocouples. They were again washed and degreased,
allowed to air dry, and mounted vertically in a vapor deposition
chamber where a thin film of tellurium was evenly deposited
over the entire outer surface of the assembly using high-vacuum
metal evaporation techniques. The thermocouple junction
consisted of the 1 pum platinum-tellurium junction at the very
tip of the probe. It was important to control carefully the
evaporation process to prevent the tellurium charge from being
blown from the evaporation boat [9]. Experience showed
that this problem could be minimized by keeping the tellurium
THERMOELECTRIC JUNCTION
PLATINUM
GOLD GLASS
SHIEW_ 'INbULAI
tsmClli ATIIUN \
----
T4\\lELLURIUM
Fig. 1. Schematic diagram of the platinum-tellurium microthermo-
couple.
powder under high vacuum for at least one day before attempt-
ing evaporation. The appearance of the films varied from a
dull black to a lustrous grey color and seemed to depend on
the deposition rate. Others have shown that the structure of
thin films of tellurium depends upon the deposition rate [4].
Variations in the appearance of the tellurium film did not
affect the performance of the completed microthermocouple;
those with a lustrous gray color exhibited essentially the same
thermoelectric power as those with a dull black appearance.
Microthermocouples in this configuration were found to per-
form well in air; however, when the tips were immersed in
aqueous solutions or tissue, the thermocouple junction was
destroyed and the thin film of tellurium separated from the
substrate in noticeable flakes. To protect the tellurium film'
and the thermoc-ouple junction, the microthermocouples were
completely coated with a thin film of negative photo-resist
(Eastman Kodak KMER), which was polymerized by exposure
to ultraviolet light. This coating provided mechanical protec-
tion of the fragile tellurium film and allowed the microther-
mocouple to be operated in aqueous solutions and tissue.
With the tellurium-platinum junction protected in this way, it
was possible to coat the entire assembly with a final thin film
of gold via vacuum vapor deposition that served to shield the
electrode from electromagnetic interference and provided a
nontoxic, biocompatible surface. A cross-sectional schematic
diagram of the completed electrode is shown in Fig. 1.
PERFORMANCE CHARACTERISTICS
The thermoelectric power, linearity, and hysteresis of the
microthermocouples were determined by comparing the output
of the microthermocouple with that of a large iron-constantan
thermocouple with well-documented performance character-
istics. Fig. 2 shows the change in the thermoelectric EMF
of the microthermocouple as it was slowly heated and sub-
sequently cooled in an aluminum heating block. The circled
points were recorded during heating, the squares represent
values measured during cooling. The reference junction was
maintained at 270C. The microthermocouple exhibits: 1)
a linear response over the range 27-100 C; 2) essentially no
hystersis; 3) a thermoelectric power of 302 pV/°C; and 4) a
thermoelectric EMF of zero at the reference junction tempera-
ture. Fig. 3 shows the time response of the microthermocouple.
This response was obtained by rapidly dnrving the thermo-
electric junction into a bath of heated water. Total micro-
thermocouple response was consistently measured to occur in
less than 50 ims. The impedance of the electrodes was typically
several hundred MQ2. They exhibited an equivalent noise level
of less than 0.25 C.
To establish that the thermocouple junction was restricted
to a 1 pum area of the microthermocouple tip, scanning elec-
tron micrographs of the platinum-tellurium microthermocouple
were made. In Fig. 4 the tellurium-coated, glass-insulated,
electropolished platinum wire is magnified 610 times. For
reference purposes, the diameter of the wire is approximately
25 pm. At this magnification the terminal edge of the glass
insulation is not clearly visible, but appears simply as an irregu-
larity at the most extreme surface of the conical point. The
tellurium coated glass appears smooth with irregularly spaced
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. BME-28, NO. 3, MARCH 1981 303
I II )
So -a
TELLURIUM- PLATINUM
CALIBRATION

15-
OHEATING
EMF. mV OCOOLING
(re,ternce * 27 C)
10-

5
THERMOELECTRIC
POWER - 302 $V/ C
Fig. 3. Time response of the platinum-tellurium microthemocouple
O0 10 20 30 40 50 60 70 80 90 100
when rapidly immersed in a heated water bath.
TEMPERATURE, *C
Fig. 2. Temperature response of the platinum-tellurium microthermo-
couple when gradually heated and cooled in an aluminum block.

Fig. 4. Scanning electron micrograph of the microthermocouple tip,

magnification = 610X (power = 20 keV).

craters randomly present in the tellurium film. Fig. 5 shows
the very tip of the microthermocouple at a magnification of
14 400 X. The irregular edge of the glass insulator is clearly
visible. At this magnification, X-ray emission analysis showed
that: 1) platinum was present at all points along the axis of
the microthermocouple; 2) sodium was present only above the
irregular edge; and 3) tellurium was present at all points on the
microthermocouple. Since sodium was present only as a constit-
uent of the glass insulator, this established that the irregular
edge visible in these micrographs was the edge of the glass insu-
lator and that tellurium was indeed in contact with platinum
at the most extreme tip to form the thermoelectric junction.
At first we suspected that the "craters" which are randomly
distributed about the surface were holes in the glass insulation.
If so, each would act as a thermoelectric junction and the
temperature measurement would not be confined to a 1 ,um
area of the tip. Additional micrographs of glass insulated plati-
num wires always demonstrated a smooth crater-free glass
coating. When these "craters" were observed at even higher
magnifications and subjected to X-ray emission spectrum anal-
ysis, it was demonstrated that sodium as well as platinum and
tellurium were present. It was concluded that the "craters"
were simply indentations in the tellurium film or tellurium
"flakes," either of which would most likely result from the
rapid sublimation characteristic of tellurium evaporation
which almost always caused some of the charge to be forc-
ibly blown from the crucible. These micrographs show the
microthermocouple prior to application of the photoresist
304 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. BME-28, NO. 3, MARCH 1981

Fig. 5. Scanning electron micrograph of the microthermocouple tip,

magnification = 14 400x (power = 20 keV).

insulator and gold shield so that the glass insulator is clearly between 30 and 100IC. All values for thermoelectric power
visible. The application of these additional layers does not add are somewhat lower than those reported in the literature [5 ].
appreciably to the tip diameter. The strength characteristics Elevating the temperature of the tellurium film toward the
of the electrodes are essentially equivalent to those of the annealing temperature stabilized the thermoelectric power of
Silver oxygen microelectrodes [10]. They are adequate for the microthermocouple. The microthermocouples were al-
penetrating soft tissue such as brain. No attempts to use the ways used in pairs (one as the measuring junction, the other
thermocouples in other tissue types (i.e., skeletal muscle, maintained at the reference temperature). While other junc-
cardiac muscle, kidney, or others) have been made. tions of dissimilar metals were unavoidable in order to make
electrical contact, the effects of these uncontrolled junctions
DISCUSSION were minimized by keeping them as close together as feasible.
The slight overshoot of the response shown in Fig. 3 is prob- Thus, any ambient temperature variation would be sensed
ably not a part of the true microthermocouple response. equally and the effects would cancel. This technique was very
Although it might be a result of series microthermocouple effective and the microthermocouples could easily be cali-
junctions along the tip, we feel it was most probably due brated to within ± 0.250C.
to thermal stratification of the liquid into which the micro- These microthermocouples have proven to provide a stable
electrode was driven or an artifact associated with the elec- method for the measurement of temperature in microenvi-
tronics of the measurement system. The gradual increase in ronments. We have used them successfully for initial studies in
temperature exhibited in Fig. 3 prior to the sudden increase the brain of an anesthetized rabbit. Their fabrication requires
was due to the boundary layer above the surface of the bath. relatively simple metal evaporation equipment and expertise
The long response time was probably due to the thermal mass in the fabrication of glass insulated platinum wires (oxygen
of the platinum wire. microelectrodes).
In addition to the platinum-tellurium microthermocouples
described in the preceding paragraphs, microthermocouples REFERENCES
were also fabricated by vapor depositing tellurium on
glass-insulated 70 percent platinum-30 percent iridium wires.
These microthermocouples were more difficult to electropolish [1] C. M. Ballintijn, "Fine tipped metal microelectrodes with glass
to a fine conical point but were more resistant to deformation insulation," Experientia, vol. 17, pp. 523-526, 1961.
when inserted into semisoft media and therefore less fragile. [2] C. P. Cain and A. J. Welch, "Thin-film temperature sensor for
biological measurements," IEEE 7rans. Biomed. Eng., vol. BME-
Their thermoelectric power (248 ,V/OC) was slightly lower 21, no. 5, pp. 421-423, 1974.
than the pure platinum-tellurium probes (302 ,V/0C). Their [31 N. W. Carter, "The production and testing of double-barreled
time response was essentially the same as that shown in Fig. 3, pH glass microelectrodes for measurement of intratubular pH,"
less than 50 ms, and they also exhibited linear characteristics Yale J. Biol. Med., vol. 45, pp. 349-355, 1972.
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. BME-28, NO. 3, MARCH 1981
[4] D. M. Chizhikou and V. P. Schastlivyi, Tellurides. London,
[6]
[7]
[81
England: Collet's, 1970.
[51 P. A. Kinzie, Thermocouple Temperature Measurement. New
York: Wiley, 1973.
G. E. Lobe, M. J. Bak, M. Saloman, and E. M. Schmidt, "Parylene
as a chronically stable, reproducible microelectrode insulator,"
IEEE 7rans. Biomed. Eng., vol. 24, no. 2, pp. 121-128, 1977.
T. 0. Neild and R. C. Thomas, "New design for a chloride sensi-
tive microelectrode," J. Physiol., vol. 231, p. 7P, 1973.
J. O'Doherty, J. F. Garcia-Diaz, and W. McD. Armstrong, So-
dium-selective liquid ion-exchange microelectrodes for intra-
cellular measurements," Science, vol. 203, pp. 1349-1351, 1979.
[9] L. 0. Olson, C. S. Smith, and E. C. Crittenden, Jr., "Techniques
for evaporation of metals," J. Appl. Phys., vol. 16, pp. 425-434,
1945.
[101 I. A. Silver, "Some observations on the cerebral cortex with an
ultra-micro membrane-covered oxygen electrode," Med. Elec.
Biol. Eng., vol. 3, pp. 377-387, 1965.
[11] J. L. Walker, Jr., "Ion specific liquid ion exchanger microelec-
trodes," Anal. Chem., vol. 43, pp. 89A-92A, 1971.
Dielectric Behavior of Selected Animal Tissues
in Vitro at Frequencies from 2 to 4 GHz
M. M. BRADY, S. A. SYMONS, AND S. S. STUCHLY
Abstract-The permittivities of bovine muscle, artery, kidney, and
liver are measured at 37°C in vitro at 2.0, 3.0, and 4.0 GHz, using a
coaxial-line reflection technique and an automatic network analyzer.
Uncertainties, evaluated by measurements on organic alcohols whose
permittivities are well known, are consistently less than ± 10 percent in
both e' and e". The data compare well with those published for similar
animal tissues.
INTRODUCTION
The dielectric properties of biological materials have long
been studied at radio frequencies (RF's) and microwave fre-
quencies (MF's), starting with the work of Cook [ 1], including
extensive studies by Schwan and his associates [21, and contri-
butions by many other workers. The data available up to 1980
are tabulated in [3 1.
In this work, the permittivities of bovine muscle, artery, kid-
ney, and liver are measured at 376C in vitro at 2.0, 3.0, and
4.0 GHz, using a coaxial-line reflection technique, with elec-
tronic data processed by an automatic network analyzer.
Whenever possible, the data are compared to those previously
published [ 3 1.
EXPERIMENTAL TECHNIQUES
For many years, coaxial-line techniques have been used to
measure the dielectric properties of materials [4]. Numerous
coaxial-line configurations have been proposed and used to ex-
tract information on dielectric properties of a sample by re-
flection techniques. A survey of these methods has recently
been published [ 5 ].
The method of this work comprises computation of permit-
tivity from admittance data, referred to the open end of a co-
Manuscript received August 25, 1980. This work was supported by
the Ministry of Labor, Province of Ontario, Canada, under Applied Re-
search Award 095/R.
The authors are with the Department of Electrical Engineering, Uni-
versity of Ottawa, Ottawa, Ont., Canada KlN 6N5.
0018-9294/81/0300-0305$00.75 © 1981 IEEE
c r AG
C2
.-
Z.k IIZ
(b)
z -O
(a)
305
Fig. 1. (a) Open-ended coaxial line. (b) Its lumped-element equiva-
lent circuit.
axial line in contact with the dielectric material being measured.
At (MF's) the coaxial lines used as sensors are small, for exam-
ple a line 3.6 mm (0.141 in) in diameter was used for the mea-
surements at 2-4 GHz reported here. This means that the
method may be used to measure small samples, and that the
samples need not be cut, altered, or otherwise prepared to suit
the measurement scheme. Admittance data are processed using
an automatic network analyzer, and the system may be cali-
brated and uncertainty may be evaluated by measuring the
permittivities of liquids whose permittivities are well known at
the frequencies in question.
THEORETICAL RELATIONSHIPS
The admittance of an open-ended coaxial line may, at the
plane of the open end, be represented as a lumped-element
circuit comprising a capacitance and a conductance [61. The
equivalent circuit is shown in Fig. 1. The general relationship
for the normalized admittance at the plane of the open circuit
is
y
= jcoCl Zo + CDzo C2(w e^) + Zo G(w, t) (1)
Yo
where Z0 is the characteristic impedance of the coaxial line,
Yo = l/Z0, X is the angular frequency, e is the permittivity of
the medium outside the line, C1 is the fringing capacitance
within the line, C2 (co, e) is the fringing capacitance outside the
line, a function of co and c, and G(co, e) is the radiation con-
ductance, also a function of co and e. In general, C1 may also
be a function of co. For coaxial lines whose dimensions are
small compared to a wavelength, C1 and C2 are frequency in-
dependent, while G is proportional to the fourth power of
frequency (6].
An open-ended coaxial line immersed in a lossy medium
radiates into that medium. Hence the open-ended line may be
regarded as a functional antenna. For antennas in lossy media,
it has been shown [ 71 that
Y(CO, g) = \fiY(\f@CO, eo). (2)
Hence (1) may be rewritten
-=2CoClZo + jCe,C2Zo + 5/2GZ. (3)
Yo
Equation (3) is of the form
=K +K3e +K3E (4)
Yo
where factors K1, K2, and K3 are, in general, complex. When
these factors are known, measurements of normalized admit-
tance may be used to compute unknown permittivities using
(4). There are two approaches to evaluating the three factors.