Sei sulla pagina 1di 8

914 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING. VOL. 38. NO. IO.

OCTOBER 1991

Development of Medical Pressure and Temperature


Sensors Employing Optical Spectrum Modulation
Roger A. Wolthuis, Senior Member, IEEE, Gordon L. Mitchell, Senior Member, IEEE, Elric Saaski,
James C. Hartl, and Martin A. Afromowitz, Member, lEEE

Abstract-Fiber optic Fabry-Perot sensors have been devel- fibers or fiber bundles as light guides, dictating the use of
oped whose optical reflectance varies with optical cavity depth relatively large (7-8 French) catheter housings.
(pressure) or with change in a material’s refractive index (tem- Later workers promoted new diaphragm materials, new
perature). These sensors employ a unique combination of fea-
tures: they are interrogated by an LED; they are designed to side-looking designs or new signal analysis techniques
operate within a single reflectance cycle; and their returned [5]-[SI. Even so, commercialization by one company
light is analyzed by a dichroic ratio technique. The sensors use (Camino Labs, San Diego, CA) has been less than suc-
a step index glass fiber and are relatively insensitive to absolute cessful, and many of the original problems with intensity
light levels and fiber bending. They have an expanded linear
operating range and can be built for low cost disposable appli-
modulation based designs remain.
cations. Sensor performance meets or exceeds established med- A new fiber optic pressure sensor design emerged in the
ical requirements. late 1970’s and early 1980’s based on the Fabry-Perot
interferometer [9], [ 101. In this design, one of two par-
allel, optical reflecting surfaces is a pressure sensitive dia-
INTRODUCTION phragm; as this diaphragm moves in response to pressure,
the Fabry-Perot reflectance peaks change accordingly.
B IOMEDICAL RESEARCHERS and clinicians have
long sought the development of small, inexpensive
sensors for measuring pressure and temperature. These
While this design resolved some of the intensity modu-
lation sensor limitations, other limitations remained and
it still required complex signal analysis instrumentation.
sensors would provide for direct, accurate measurements Turning to biomedical temperature measurements,
of pressure and temperature at discrete intravascular sites, small and inexpensive thermocouple and thermistor de-
yet minimize sensor presence effects on the measurement vices have been available for many years. Since these de-
itself. Furthermore, these sensors would also fit the mod- vices are electrically activated and employ conducting
ern emphasis on use of low cost, disposable devices. materials, they are not appropriate for making measure-
Piezoresistive pressure sensors became available for ments in a number of medical applications, e.g., in high
biomedical applications in the late 1950’s, and their pri- RF fields associated with hyperthermia treatment. Fur-
mary use was and remains intravascular pressure mea- thermore, available optical temperature sensors based on
surements. Millar Instruments (Houston, TX) miniatur- the use of rare earth phosphors [ 111 or on gallium arsenide
ized these sensors and placed them in catheters to allow [ 121 are relatively bulky and/or expensive.
for direct, accurate, intravascular measurements. The re- A program was initiated for development of a new gen-
sulting product was quite fragile and very expensive, pre- eration of medical pressure and temperature sensors based
venting use in all but a few restricted applications. Cur- on the Fabry-Perot interferometer. Specific program goals
rently, most intravascular pressure measurements are included:
made indirectly using a fluid-filled catheter and a larger,
externally-located pressure sensor. Hence, the desired small sensor size, ideally less than 0.1 mm3,
combination of small size, robustness and low cost has a single loo/ 140 pm optical fiber per sensor,
not been achieved with piezoresistive devices. ease of sensor fabrication/manufacture,
Fiber optic pressure sensors for biomedical applications low sensor cost,
have been under development since the mid-1960’s. Early sensor performance equal to or better than existing
workers focused on intensity modulation designs, but their American Association for Medical Instrumentation
sensors were quite large, difficult to manufacture, and (AAMI) specifications.
costly [ 11-[4]. Furthermore, these sensors used multiple The development, fabrication, and evaluation of these
sensors are described in this paper.
Manuscript received May 14. 1990; revised December 18, 1990.
R. A . Wolthuis, G . L. Mitchell, E. Saaski, and J . C. Hart1 are with SENSORDEVELOPMENT
MetriCor Inc.. Woodinville, WA 98072. The development program began with a thorough re-
M . A. Afromowitz is with the Department of Electrical Engineering,
University of Washington, Seattle, WA 98195. view of known optical approaches for pressure sensing.
IEEE Log Number 9102477. Emphasis was given to identifying an optical approach

0018-9294/91/1000-0974$01.OO 0 1991 IEEE


WOLTHIIIS cr <I/.: DEVELOPMENT OF MEDICAL PRESSURE A N D T E M P E R A T U R E SENSORS 915

that Pressure Sensitive

a) was relatively independent of absolute light levels,


i.e., not intensity modulated,
b) could operate using a single optical fiber, and
c) had performance equal to or greater than AAMI
specifications.
The Fabry-Perot interferometer [9] seemed to meet these
requirements, and subsequent development work focused Fig. I . A Fabry-Perot interferometer pressure sensor is illustrated. The
on expanding this optical approach in novel ways. dashed lines show light arriving at and reflecting from the two partially
reflecting surfaces. Pressure diaphragm thickness and optical cavity depth
are not drawn to scale.
Pressure Sensor
In a Fabry-Perot interferometer, two parallel, partially "1
reflecting surfaces are spaced less than a coherence length
apart, thereby forming an optical reflecting cavity. If one
of the partial reflecting surfaces is a pressure-sensitive
diaphragm, changes in external pressure will alter optical
cavity depth (Fig. 1). A change in optical cavity depth
will cause a change in optical cavity reflectance. For a
given wavelength, cavity depth and cavity reflectance are
0 1 2 3
related by an oscillating curve having multiple maxima
Cavity Dbpth. mluona
and minima (Fig. 2).
Fig. 2. In a Fabry-Perot interferometer, cavity depth and cavity reflec-
The literture describes a number of methods for making tance (at 850 nm) are related by a repeating curve having multiple maxima
Fabry-Perot pressure sensors, and these are summarized and minima. Vertical dashed lines define the area expanded in Fig. 3 .
in Table I. While each these methods have been imple-
mented in laboratory settings, each has one or more lim-
itations that have stood in the way of active commercial- reflectance of all LED wavelengths was summed and plot-
ization. For example, there is no absolute pressure ted as a function of optical cavity depth (i.e., the photo-
reference inherent in the Incremental method design; an current generated by the reflected light of an LED-Micro-
outside reference must be used each time the system is shift pressure sensor is plotted versus absolute pressure).
turned on. Intensity methods generally have a limited dy- As the data indicate, it is possible to operate between the
namic range and are sensitive to light level changes in- maxima and minima, but the sensor is linear over a very
duced by optical fibers and/or connectors. White light short pressure range.
methods suffer from limited source lifetimes and neces- The third feature of the LED-Microshift method is its
sary readout complexity. Finally, the LED-deep cavity use of a dichroic ratio technique for returned light signal
method is difficult to manufacture in quantity-reproduc- analysis. In this dichroic ratio technique,light returned
ing exact cavity depth (30-100 pm) over thousands of from the sensor is split into short and long wavelength
sensors is a major stumbling block. components by a dichroic mirror; the corresponding de-
In the present program, a new Fabry-Perot method was tected photocurrents can be used to form a ratiometric sig-
developed and named the LED-Microshift method [ 191. nal (Fig. 5 ) . One benefit of this technique is that the di-
This new method is characterized by three distinct fea- chroic ratio is relatively immune to changes in LED output
tures. First, as the name suggests, this method uses an intensity or to the efficiency of light-handling components
LED for interrogating the optical reflecting cavity. Since within the optical system; changes in light level affect both
the LED has an emission bandwidth of about 60 nm, the components of the ratio signal, leaving its absolute value
reflectance of each emission wavelength varies uniquely essentially unchanged. Another more important benefit of
with optical cavity depth. This is illustrated in Fig. 3 this technique is its ability to increase the linear operating
which shows the relationship between cavity depth and region of the pressure sensor by a factor of 3 to 4. Recall
reflectance for just three LED emission wavelengths. In that the light returned by the LED-Microshift pressure
effect, the spectral character of LED light reflected by the sensor (i.e., measured photocurrent) varies with absolute
sensor changes with cavity depth. pressure (Fig. 4). If, however, returned light is split into
The second feature of the LED-Microshift method has short and long wavelength components, the ratio of long
to do with its operating domain. The word "microshift" wavelength intensity to total intensity (short plus long
calls attention to the fact that this method operates over wavelength) yields an altered pressure response curve with
changes in optical cavity depth that occur between two a significantly extended linear range (Fig. 6). This ex-
reflectance peaks, e.g., on the reflectance changes occur- tended linear range allows use of the sensor without user
ring between a maximum and a minimum within a single correction across a wide range of ambient barometric
reflectance cycle. This is illustrated in Fig. 4 where the pressure settings.
976 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 38, NO. IO. OCTOBER 1991

TABLE 1
SUMMAR
OFYPUBLISHED
FABRY-PEROT
PRESSURE
SENSORS

Light Source ReHectance System


Method Source Width Max to Max Readout Reference

Incremental Laser < I nm > 10 nm Fringe count 13


Intensity Laser < I nm > 100 nm Intensity 14, 15
White Light Lamp > 100 nm - 10 nm Spectrometer 16
LED-Deep Cavity LED -60 nm < I O nm Spectrometer 17, I8

600 7 &- 0.9

1.8 1.9 2.0 2.1


200! , , I I , , ~! 0.4
0 500 1000 1500 2000
Cavity Depth, mlcronr
Absolute Pressure, mmHg
Fig. 3 . The relationship between interferometer cavity depth and cavity
reflectance is shown for LED emission wavelengths at 820. 850, and 880 Fig. 6 . Total sensor reHectance (measured photocurrent) and the output
nm. Note that reHectance is wavelength dependent at each cavity depth. from dichroic ratio signal analysis (dichroic ratio) are plotted with respect
to optical cavity depth (absolute pressure), as measured over part of a re-
Hectance cycle with a MetriCor Model 1400 instrument.

The output intensity, I, of the LED can be modeled as


I = 1,[1 + cos { T ( X - x,)/2w,}l/2, xc
2WL < < Xc 2WL
- x +
where X = wavelength, wL = LED spectral width, Xc =
LED center wavelength, I , = center wavelength inten-
200 I
sity. This modeled LED spectral shape matches measured
0 500 1000 1500 2000 spectra closely enough to permit an understanding of sen-
Absolute Pressure, mmHg sor performance.
Fig. 4. The relationship between total sensor reHectance (measured pho- Sensor reflectance is of the form (20)
tocurrent) and optical cavity depth (absolute pressure) is shown for an LED-
microshift pressure sensor, as measured over part of a reHectance cycle
with a MetriCor Model 1400 instrument. - cos 6

L
-F +
L
1 -cos6

- -1 +-1 - K cos 6
Fused Fiber
Optic Coupler To
K K2
I
Sensor
Optical
where we have taken the first two terms in an expansion,
considering cos 6 to be small, and where F = 4 R ‘ / ( 1 -
R‘)’, R‘ = ( r l r 2 ) ’ ” , r l . r2 are surface intensity reflec-
I tances, 6 = 47rd/X, d = surface spacing, and K = 1 +
Receiving 2/F.
Photodiodes We are interested in R in the neighborhood of a reflec-
tion peak, say the one that occurs at Xcf. At this wave-
Fig. 5 . Optical system design showing implementation o f the dichroic ra- length, &, = 2a(m + 1/2), where m is a particular in-
tio signal analysis technique. teger. For wavelengths h near this peak, that is 1 h - Act 1
<< Xcf it is readily shown that
COS 6 1 -COS A
The dichroic ratio signal analysis technique provides an
extended linear operating range because the LED spec- where A = ?r(X - Xcf)/2w, and w, = Xc,/4(m + 1/2).
trum is nearly symmetrical and closely matches the spec- Therefore,
tral reflectance peak-width of the sensor. This can be il- 1 1-K
R Z - - - cos A .
lustrated as follows: K K’
WOLTHUIS (’I U / . : DEVELOPMENT OF MEDICAL PRESSURE A N D TEMPERATURE SENSORS 911

Hence, the ratio of intensity of long-wave to long- plus Partially Reflecting


surfaces
short-wave reflected light takes the form
Silicon
Xr + 2 u /
[ IRdX
Optical Fiber

Fig. 7 . An LED-Microshift temperature sensor is shown. Some structures


are nor drawn to scale.
The result of this calculation is
1
Ratio = -
2
+ -2
T
1
2K
(1
-
-

(1
K ) sin A’
- K ) COS A’ 1 pressure sensor’s ability to measure pressure because a)
the two-optical cavities have significantly different optical
where A’ = n(Xc - Xc,)/2w and where we have let w = depths (1.43 pm with index 1.0 versus 4.03 pm with in-
W L = w,,. dex 3.6), and b) only the etched optical cavity changes
This expression for the ratio appears to be quite linear depth with change in pressure while the silicon “cavity”
for la’ I cc 7r/2, and applies to LED’s which have mod- does not.
erately symmetrical spectra illuminating Fabry-Perot sen- Subsequently, the design of an LED-Microshift tem-
sors in which the reflectance peak width of the Fabry- perature sensor based on silicon was pursued; the same
Perot is comparable to the LED emission peak width. three features used with the pressure sensor (i.e., use of
In practice, the LED-Microshift sensor has character- an LED source, operating within a single reflectance
istics that deviate from the theoretical model given above. cycle, and use of the dichroic ratio signal analysis tech-
For example, LED emission characteristics may vary from nique) were applied to the temperature sensor as well.
unit to unit; minor differences in LED peak wavelength Functionally, the outside surfaces of a thin layer of silicon
are tuned by controlling LED operating temperature, and defines an optical reflecting cavity (Fig. 7). As tempera-
spectral width variations are accommodated by gain ad- ture changes, the refractive index of silicon changes; this
justments. Furthermore, the diaphragm is not strictly index change is substantial when observed with light near
planar, therefore the width of the Fabry-Perot resonance the band gap of silicon. For example, with an 850 nm
is increased from theoretical values. However, we use LED source, the relative refractive index change with
only the center 25 % of the diaphragm area, and therefore temperature is several times larger than other changes oc-
the nonplanar effects are minimized. curring within the silicon, including the expansion of sil-
The LED-Microshift sensor shares the same design icon itself. As the refractive index of silicon changes, the
flexibility found in all diaphragm-based sensors, e.g., a spectral reflectance of each LED wavelength also changes
wide range of measurement applications can be accom- (see Fig. 3). These reflectance changes occur within a
modated by simple changes in one or more critical design sensor reflectance cycle and are of the same order of mag-
variables: Fabry-Perot cavity spacing, d ; cavity radius, nitude as seen for the pressure sensor.
a ; and/or pressure diaphragm thickness, r. Thus, the pres- Two competing effects must be considered in selecting
sure, P , which will cause a diaphragm center deflection the correct sensor silicon thickness: low sensitivity results
of X/4 is given by when the silicon is too thin, while high signal loss and
P = 6 4 ( D / a 4 )X / 4 restricted temperature range result when the silicon is too
thick. In our experience, single crystal silicon having a
where D is diaphragm stiffness defined by thickness of 0.8 pm provides a more than adequate linear
D = &/12(1 - d) range for medical temperture measurements.
and E is Young’s modulus for the diaphragm, r is dia- It is important to note that, because the pressure and
phragm thickness, and U is Poisson’s ratio for the dia- temperature sensor reflectance spectra are similar, both
phragm. Importantly, this design flexibility is only an as- sensors can use the same dichroic ratio signal analysis in-
set if the underlying variables can be controlled precisely strumentation (see Fig. 5 ) . This instrument commonality
in a manufacturing environment and within a production has significant user and commercial implications; most fi-
run. For the LED-Microshift sensor, photolighographic ber optic sensors to date have required separate, custom
micromatchining techniques provide for < 1 % sensor-to- instrumentation for each sensed parameter.
sensor variance within a production run, and the resulting
product yields are more than adequate for assuring low SENSORFABRICATION
cost sensors. As noted in the introduction, three of the five sensor
development program goals were related to sensor fabri-
Temperature Sensor cation: small sensor size, ease of sensor fabrication and
During pressure sensor development, it was suggested low sensor cost. To achieve these goals, sensor design
that the silicon diaphragm itself could act as an optical was weighted toward incorporating semiconductor pro-
reflecting cavity [21]. This observation did not affect the cess and micromachining techniques wherever possible.
978 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING. VOL. 3R. NO. IO. OCTOBER 1991

Optical Canty
with Ti02
Reflector pressure
Sensor

Gener Ring

Pressure Sensitive
Silimn Diaphragm

f
I t
I I -- Fiber-Capillary
Pedestal

Glass Optical I I
Fiber
I I
Fig. 8 . The detail of an LED-Microshift pressure sensor is shown. Some
structures are not drawn to scale.

Pressure Sensor
The completed LED-Microshift pressure sensor is a
small cube of glass with approximate dimensions of 300 Fig. 9. Detail of the fiber tip assembly used for mounting an LED-Micro-
x 300 x 275 pm. One face of this glass cube contains a shift pressure sensor is shown.
shallow cavity (1.4-1.7 pm deep x 200 pm diameter)
which is covered by a thin, pressure sensitive silicon dia- “Mallory” bonding technique [22]; the resulting bond is
phragm (Fig. 8). (While the sensor illustrated and de- permanent and hermetic. The undoped silicon is then
scribed in this report is designed for absolute pressure etched away with ethylene diamine pyrocatechol; etching
measurements, the same optical approch can and has been slows dramatically when the etchant reaches the layer of
used for the design of differential pressure sensors.) doped silicon, leaving the cavities covered with a thin sil-
Sensor fabrication centers around the masking and pro- icon diaphragm. Finally, the completed wafer assembly
cessing of a 275 pm X 75 mm O.D. polished substrate is diced into individual pressure sensors using a diamond-
made from Corning 7740 (Pyrex) glass. Following the ap- bladed dicing saw. The sensors are now ready for mount-
plication of photoresist to this glass substrate, the first ing on the tip of optical fibers.
masking step transfers an outline of sensor optical cavities The mounting of an LED-Microshift pressure sensor on
(200 pm diameter circles) to the substrate. These cavities the tip of an optical fiber is detailed in Fig. 9. Because
are then etched with HF to the desired depth; target depth the sensor is larger than the optical fiber cross section, the
is determined by the reflectance cycle one chooses to op- fiber cross section is first enlarged by attaching a short
erate in, and is typically in the range 1.4-1.7 pm. Next, piece of 500 pm O.D. glass capillary tubing to the fiber
a quarter-wave thick film of titanium dioxide is deposited end. The capillary tubing-fiber assembly is then polished
on the bottom of the etched cavities using conventional in preparation for sensor mounting. The completed as-
vacuum deposition techniques. This film increases reflec- sembly fits easily in a 1 mm O.D. ( 3 French) catheter
tance of the 7740/Ti02 /air surface to approximately that housing.
of the air/Si02 /Si surface on the opposite (diaphragm)
side of the optical cavity. The photoresist is removed and Temperature Sensor
the glass wafer is cleaned. The completed optical temperature sensor has approx-
Following the application of new photoresist, the next imate dimensions of 125 x 125 X 400 pm. As illustrated
masking step defines a donut-shaped iron getter ring near in Fig. 7 , the active sensor element is a thin layer of sil-
the outer circumference within each previously etched icon sandwiched between two pieces of glass.
cavity. This getter ring will later scavenge gases from the Sensor fabrication begins with anodic bonding of a 75
sealed optical cavity, thereby providing an absolute pres- mm diam X 200 pm thick polished glass wafer (Corning
sure sensor having a low temperature coefficient. The 7740) to a 75 mm diameter silicon wafer; the silicon wafer
photoresist is then removed. has been pre-processed by diffusing boron into its surface
The next fabrication step involves covering the pre- to a depth of 0 . 8 pm. Once again, anodic bonding [22] is
pared cavities with a silicon wafer. The silicon wafer is used to join the parts. The undoped silicon is then etched
preprocessed by diffusing an etch stop (boron) into the away with ethylene diamine pyrocatechol, leaving a thin
silicon to a depth consistent with the desired diaphragm layer of doped silicon (approximately 0.8 pm) attached to
thickness, nominally - 4 pm. The doped silicon surface the glass substrate. A second polished glass wafer is then
is anodically bonded in vacuum to the cavity-containing anodically bonded to the etched silicon surface, forming
glass wafer. Anodic bonding is accomplished using the a sandwich as shown in Fig. 7. The completed sandwich-
__

WOLTHUIS et U/.: DEVELOPMENT OF MEDICAL PRESSURE A N D TEMPERATURE SENSORS 979

wafer is diced into individual sensors with a diamond-


bladed dicing saw, and the sensors are mounted directly
onto the ends of 100/140 pm multimode glass fibers.
Each fiber-sensor assembly fits easily into a 2 French
catheter housing.

SENSOREVALUATION A N D CALIBRATION

Pressure Sensor Evaluation _.I. ,


600 700 800 900 1000 1100
Response of the LED-Microshift pressure sensor is Absolute Pressure, mmHg
shown in Fig. 10; the response to pressure is plotted as a Fig. IO. The response of an LED-Microshift pressure sensor is shown.
function of dichroic ratio, the signal analysis technique
used here and decribed earlier. The tradeoff between lin-
ear operating range and sensitivity is quite straightfor-
ward: increased linear range is achieved by operating over
a shorter portion of the reflectance cycle, e.g., by increas-
ing diaphragm thickness, while greater sensitivity is
achieved by operating over a longer portion of the reflec-
tance cycle, e.g., by decreasing diaphragm thickness. A
typical compromise provides 2 % of reading linearity and
0.06% rafio change/mmHg sensitivity; this sensitivity in 0.54 I I I I i
turn provides resolution (1 mmHg) and accuracy ( k l 600 700 800 900 1000 1100
mmHg) consistent with the needs of most medical appli- Absolute Pressure, mmHg
cations. Fig. 1 1 . The response of an LED-Microshift pressure sensor to four re-
Fig. 1 1 illustrates a lack of hysteresis in the LED-Mi- peated calibration runs is shown.
croshift sensor design. This absence of hysteresis is due
to the use of a) single crystal silicon in the pressure sen-
ICY I

sitive diaphragm and b) anodic bonding of silicon to the


glass substrate. Sensor designs which use organic dia-
phragm and/or adhesive materials are likely to have sub-
stantial hysteresis.
The frequency response of the LED-Microshift sensor
is flat up to 1000 Hz as measured with a piston-excited
I
pressure cell. In theory, frequency response should be flat .00 .25 .50 .75 1.00 i.25

from dc to 500 kHz because the diaphragm resonance is


greater than 1 MHz.
Finally, Fig. 12 is a composite showing recorded pres-
sure waveforms obtained with a Millar Micro-tip catheter
and an LED-Microshift sensor. As expected, the two sen-
sors provide identical waveforms, indicating the inter-
I
changeability of sensor technologies with respect to these .00 25 50 75 I bo 5 25

medical measurements. Tlme Isecl


One aspect of the LED-Microshit pressure sensor de- Fig. 12. The response of an LED-Microshift pressure sensor (upper trac-
sign deserves further comment, namely, its ability to sup- ing) is compared with that from a Millar Micro-tip catheter (lower tracing)
port both absolute and differential pressure sensor config- for measurements taken in the canine left ventricle.
urations. In the absolute pressure sensor configuration,
sensitivity and resolution are reduced to provide for an
extended linear operting range (nominally 1000 mmHg); sensor profile, and provision for optical cavity venting to
this allows use of the sensor at widely varying barometric the atmosphere.
pressures. Even so, the sensitivity and resolution are more
than adequate for accurate measurements of left heart and Temperature Sensor Evaluation
systemic arterial pressures. In the differential pressure The LED-Microshift temperature sensor response to
sensor configuration (not illustrated), the operating span temperature is shown in Fig. 13; as before, sensor re-
can be reduced because changes in barometric pressure sponse is plotted as a function of dichroic ratio. Over the
are automatically accommodated; hence, this configura- restricted range needed for medical applications, 1 % of
tion provides the sensitivity and resolution needed for, span linearity and 0.1 % ratio change/ "C sensitivity are
say, right heart or intracranial pressure measurements. readily achieved and provide a measurement resolution of
The cost for this enhanced performance is a slightly larger 0.2"C (0.02"C with sample averging). While sensor res-
980 IEEE TRANSACZTIONS ON BIOMEDICAL ENGINEERING. VOL. 38, NO. IO, OCTOBER 1991

With regard to the pressure sensor, recall that the an-


1 9 odic bonding is accomplished in vacuum, which, together
with the getter ring assure that there is little or no gas
inside the cavity. Thus, temperature changes do not cause
substantial calibration shifts due to cavity pressure
changes. Minor expansion of the Pyrex cavity material
results in a temperature coefficient of pressure of approx-
\3 imately 200 ppm of full scale/”C.
0.64 , I I I I I I I

0 10 20 30 40 50 60 70
Temperature, “C CONCLUSIONS
Fig. 13. The response of an LED-Microshift temperature sensor is shown From the start, the goals of this sensor development
for the range IO-60°C. program were very broad, ranging from creative sensor
design to robust sensor performance to incorporation into
commercially available medical products. The design
loor /---- concepts are truly unique and have been protected through
a series of U . S . and foreign patent applications; four of
0 t i these U.S. applications have already yielded patent
awards. Sensor performance goals have been largely met
or exceeded as illustrated by the data presented in this
paper. Finally, recent market introduction of the first of
several Baxter Edwards catheter products incorporating
the LED-Microshift sensor described above signals for-
mal incorporation of this sensor technology into commer-
I I I
0 100 200 300 400 500
cially available medical products. The time from begining
Time, msec of sensor design to product market introduction-seven
Fig. 14. The response time of an LED-Microshift temperature sensor is years.
shown, the sensor was moved quickly from ice water to boiling water for
this illustration.
ACKNOWLEDGMENT
The authors are indebted to Baxter Edwards Critical
olution is inadequate for thermal dilution curve measure- Care Laboratories for their generous support of this pro-
ments, total sensor span is nearly an order of magnitude gram. The authors express appreciation to D. Lawrence,
larger than that needed for medical applications. Metricor, for contributing expertise in semiconductor
Finally, sensor response to a step change in tempera- process and micromachining techniques; to D. Koop, J.
ture is shown in Fig. 14; data were obtained by moving a Barger, and G . Canny for laboratory support; and to Dr.
temperature sensor from ice water to boiling water. As K. Campberl and staff, Washington State University Car-
indicated, the sensor reaches 90% of its final value in ap- diovascular Research Laboratory, for taking and provid-
proximately 20 ms. The response time for this sensor will ing the in vivo data used in Figure 12.
vary with the application, and is generally rapid because
of the small sensor size. REFERENCES
[ I ] F. Clark, E. Schmidt, and R. DeLaCroix, “Fiber optic blood pres-
Calibration sure catheter with frequency response from dc into the audio range.”
Proc. Nut. Elec. C o n t . 1965, pp. 213-216.
Recall that the pressure and temperature sensors (Figs. [2] M. Polanyi. “Medical applications of fiberoptics,” Cunj Med. Elect.
B i d . E n g . , Tokyo. 1965. pp. 598-599.
1 and 7) consist of Pyrex glass and silicon, bound together (31 A. Ramirez, W. Hood, R . Wagner, and W . Abelmann, “A fiberoptic
anodically. The Pyrex and silicon have essentially the catheter for the measurement of intravascular pressures and sounds.”
same thermal expansion coefficient. Furthermore, no ad- Circulation, vols. 37, 38, supp. 6 p. 160. 1968.
141 S. Morikawa. Fiberoptic catheter-tip pressure transducer,” Jap. J.
hesives or polymers are used in these optically-sensitive “

Med. Electron.. vol. IO. pp. 36-39, 1972.


elements which could creep with time, or expand through 151 R. Pahler and A. Roberts, “Design of a fiber optic pressure trans-
absorption of moisture. Long-term stability and shelf-life ducer.” Trans. ASME, pp. 274-280. Feb. 1977.
161 H. Matsumoto and M. Saegusa, “The development of a fiber optic
are inherent in this design. catheter tip pressure transducer.” Jup. J . Med. Eng. Techno/. , vol.
Sensor-to-sensor variations can be minimized through 2, pp. 239-242. 1978.
state-of-the-art fabrication techniques, yielding sensors [7J W. Spillman, “Multimode fiber-optic pressure sensor based on the
photoelastic effect.” Opt. L e t t . , vol. 7. pp. 388-390, Aug. 1982.
with essentially the same performance characteristics. [8] I . Giles, D. Uttam. S . McNeil. and B . Culshaw, “Self compensating
This means that individual sensor calibration is not re- technique for remote fiber optic intensity modulated transducers.”
quired. Long-term measurement stability is determined SPIE, vol. 522, pp. 233-239. 1985.
191 E. Cox and B . Jones, “Fiber optic color sensors based on Fabry-
more by instrumentation performance than sensor consid- Perot interferometry,” presented at First Int. Conf. Opt. Fiber Sen-
erations. sors, London, 1983. pp. 122-126.
WOLTHUIS (,/ (I/ DEVELOPMENT OF MEDICAL PRESSURE AND TEMPERATURE SENSOKS Y8 I

1101 K . Belsley. D. Huber. and J. Goodman, "All-passive fiber-optic Gordon L. Mitchell (S'63-M'74-SM'77) re-
pressure sensor." l i i Tedi.. vol. 33, pp. 39-42, Dec. 1986. ceived the B.S. degree in 1964 and the Ph.D. de-
1111 K. Wickersheini, "Optical fibers in medicine 11.'' SPIE. vol. 713. gree in 1974 from the University of Washington
pp, 150-157. Sept. 1988. Department of Electrical Engineering. University
1121 D. Christensen, "A fiberoptic temperature sensor using wavelength- of Washington, Seattle.
dependent detection," SPIE. vol. 838, pp. 252-256. 1987. He has worked for MetriCor since 1983 and is
[ 131 T. Yoshino. K. Kurosawa. K . Itoh. and T. Ose, "Fiber-optic Fabry- presently the Manager. Advanced Prqjects. He
Perot interferometer and its sensor applications.'' lEEE Trcim. Mi- was with the Honeywell Systems and Reserch
wr\tuiv Tliror? Tech., vol. MTT-30. pp. 1612-1621. Oct. 1982. Center in Minneapolis for six years where he
[ 141 A . Kersey, D. Jackson. and M. Corke. "A simple tiber Fabry-Perot started a tiber optic sensor group. He has also held
sensor," Opr. Comiiiuii.. vol. 45. pp. 71-74, 1983. a research faculty position at the University of
[ I S ) C . Lee and H. Taylor. "Interferometric optical tiber sensors using Washington, assisted the startup of several companies, and Rew for the US
internal mirrors," Elrcfroii. Lrrr.. vol. 24. pp. 193-194. Feb. 1988. Navy. When not engaged i n the business o f tiber optic sensors he is active
1161 W. Quick. K . James, and J . Coker, "Fiber optics sensing tech- as a volunteer in OSA and IEEE activities.
niques." presented at First I n t . Conf. Fiber Opt. Sensors, London.
1983.
I 17) G . Beheim. K . Fritsch. and R . Poorinan. "Fiber-linked interf'ero-
metric pressure sensor.'' Rri.. Sc,i. / m / r m i . . vol. 5 8 , pp. 1655- 1669,
Sept. 1987.
[ 181 E. Cox and B . Jones." Fiber optic color sensors based on Fabry-
Perot interferometly." presented at First Int. Opt. Fiber Sensor Symp. Elric Saaski, photograph and biography not available at the time of pub-
London, Apr. 1983. I ication.
I191 G . Mitchell. E. Saaski, J . Hanl, R. Wolthuis. and M . Afroniowitz.
"A family of tiber optic sensors using cavity resonator inicroshifts."
presented at Fourth Int. Conf. Optical Fiber Sensors. Tokyo. Oct.
1986.
1201 M. Born and E. Wolf. Pritic.ip/c.\ of 0pric.s. Oxford: Perganion
Press, 1975.
12 I ] D. Christensen. "Tcniperature nieasurenient using optical ctalons." James C. Hart1 received the B.S. degree in 1976 and the M.S. degree in
presented at Opt. Soc. Anier. Ann. Meet.. Houston. TX. Oct. 15- 1979 from the Department of Chemical Engineering, University of Wash-
18, 1974. ington, Seattle.
1-72] G . Wallis. "Direct-current polarization during field- He has worked for MetriCor since 1981 and is presently Vice President
metal sealing."J. Airier. Ccr. Soc... vol. 53. pp. 563-567. Oct. 1970. of Research and Development. Previously he has worked in development
positions with Sigma Research and Battelle Pacific Northwest Labortories
both of Richland Washington.
Roger A. Wolthuis (M'73-SM'8 I ) completcd
undergraduate studies at [he Univer\ity of Michi
gan. Ann Arbor. and received the M S and Ph D
degree\ i n cardio\a\cular phy\iology troni Mich
igan State University. Ea\t Lan\ing. i n 1965 and
1968. respectively.
His professional interests have been a blend o f
applied cardiovascular research and medical prod- Martin A . Afromowitz (S'63-M'69-M'81) re-
uct development. He has held both technical and ceived the B . S . . M.S.. and Ph.D. degrees in elec-
management positions at the NASA Manned trical engineering from Columbia University. NY.
Spacecraft Center Cardiovascular Research Lab- in 1965. 1966, and 1969. respectively.
oratory, Houston, TX (1968-1974). the Department o f Medicine. U S A F Hc is currently Professor of Electrical Engi-
School of Aerospace Medicine. San Antonio, TX (1974-1079). Meironic neering and Ad,junct Protcssor of Bioengineering
Inc.. Minneapolis. MN (1979-1981). Squibb Medical Systems Division. at the University of Washington, Seattle, WA. His
Bellevue. WA (198 I - 1982). and MetriCor Inc.. Woodinvillc. WA (1983- research interests include bioinstrumentation, fi-
1990). He recently founded and is currently president of MetriLab Inc.. ber optic sensors. and microfdbrication.
Woodinville.WA.

Potrebbero piacerti anche