Sensorsand Acruofors 8, 12 (1993)213-220 213

A fiber-optic chemical sensor based on surface plasmon resonance

R. C. Jorgenson and S. S. Yee

Sensor TechnologyResearch Group, Department of Electrical Engineering,Universityof Washington,Seattle, WA 98195 (USA)

(Received September 23, 1992;in revised form December 15, 1992;accepted f&ember 20, 1992)

A fiber-optic chemical sensor is presented which utilizes surface plasmon resonance excitation. The sensing element
of the fiber has been fabricated by removing a section of the fiber cladding and symmetricaBy depositing a thin layer
of highly refleoting metal onto the fiber core. A white-hgbt source is used to introduce a range of wavelengths into
the fiber optic. Changes in the sensed parameters (e.g., bulk refractive index, film thickness and f&a refractive index)
are determined by measuring the transmitted spectral-intensitydistribution, Experimentalresults of the sensitivity
and the dynamic range in the measurementof the refractiveindices of aqueous solutions are in agreementwith the
theoreticalmodel of the sensor.

1. introduction IncidentLight Reflected

Thl Polarized Light
Many optical sensing systems have been developed
utilizing the sensitivity of surface plasmon resonance
(SPR) to the refractive indices of bulk and thin-tihn
dielectrics and to the thicknesses of thin films. These
etal
sensing systems in conjunction with the appropriate
chemically sensitive layers have allowed researchers to
develop a variety of SPR-based chemical sensors, in- BalkDielectric
cluding immunoassay [l-3], gas [ 1,4], and liquid sen-
sors [5]. One such chemical-sensing system has been Fig. I. The Kretschmann configuration.
commercialized by Pharmacia (Uppsala, Sweden).
However, all these systems utilize bulk-optic configura-
tions that are limited by the use of a coupling prism.
The resulting systems are relatively large and expensive
and are also inapplicable for remote sensing applica-
tions.
Traditionally, SPR is measured using the Kretsch-
mann configuration, illustrated in Fig. 1, with a prism
and a thin highly reflecting metal iayer (e.g., silver or
gold) deposited upon the prism base [6]. The SPR
reflection spectrum (reflected light intensity versus angle
of incidence with respect to the normal of the metal/ Fig. 2. TheoreticaFSPR retkctioa spect+ for a bulk dielectric
dielectric interface) is measured by coupling transverse of water, a wavelength~of620nm, a 550A thick silver film, and a
magnetically (TM) polarized monochromatic light into fused silica prism. ’ .,!
the prism and measuring the reflected light intensity of
the ray exiting the prism versus the angle of incidence. A novel fiber-optic. SPR sensing configuration with-
An example of the resultant theoretical spectrum is out the required light-coupling prism is presented. This
illustrated in Fig. 2. The angle at which the minimum sensing con&uratiun allows for a small sensing element
reflection intensity occurs, t?,,, is the resonance angle at and sample volme, simplified optical design, potential
which coupling of energy occurs between the incident use for disposable fiber-optic sensors and the capability
light and the surface plasmon waves. for remote sensing.

0925-4C'J5/93/$6.00 @ 1993- ElsevierSequoia.All rights reserved

214

The sensing element is a segment of fiber in which the permittivities of the metal, a,,,, and the dielectric, ad.
cladding has been removed and a 550 8, thick silver film The propagation constant is expressed as
has been symmetrically deposited on the fiber core, via l/2
electron-beam evaporation. The length of the cladding (2)
removed was calculated to optimize the average number
of reflections for ah the propagating rays of light in the At the resonance angle 0,, the propagation constant of
fiber. Unlike traditional SPR measurements, which em- the /ncident beam parallel to the prism base (i.e., along
ploy a discrete excitation wavelength while modulating the z-direction, konp sin,&,) is equal to the real part of
the angle of incidence, the SPR fiber-optic sensor sys- the SPR propagation constant. This equality condition
tem uses a white-light source, resulting in a large range is expressed by
of excitation wavelengths. The range of incident angles
ki = k,,np sin t?,, (3)
is limited to only those angles that propagate in the
optical fiber. The sensed parameters can then be deter- where np is the refractive index of the prism.
mined from the measured resonance spectrum in the Equations (2) and (3) provide for the theoretical
transmitted spectral intensity distribution. transduction mechanism for SPR as a sensor. Specifi-
cally, an increase in refractive index of the sensed
dielectric will cause a shift in the resonance spectrum
2. Theory toward larger SPR coupling angles. By measuring the
SPR resonance parameters (coupling angle, Q,, full
2.1. Background width at half maximum, A@,and the reflected light
SPR is a resonance phenomenon in which surface intensity at the resonance coupling angle, Z&,) one can
plasmon waves (SPWs) are excited at a metal/dielectric determine the complex refractive index of the dielectric
interface. The electric field, E, of the SPW propagating using the calculated SPR propagation constant, k, [8].
in the z-direction (Fig. I) is expressed as [7] The SPR reflection spectrum in Fig. 2 is for a discrete
wavelength. For other excitation wavelengths the SPR
E,(x, 2, 1) = Et(x) exp(iwl-ik,z) (1) propagation constant will change accordingly, since the
where o = angular frequency and prism refractive index, bulk dielectric refractive index
and the metal permittivity are all a function of wave-
k,=k:+ik:
length [9]. A theoretical three-dimensional SPR reflec-
is the SPR propagation constant along the z-direction. tion spectrum for a range of wavelengths is illustrated
The SPR propagation constant, k,, is dependent in Fig. 3 with the value of one minus the normalized
upon the free-space wave number, k,, and the complex reflected intensity versus the angle of incidence and

Fig. 3. Theoretical three-dimensional SPR reflection spectra for a bulk dielectric of water, a 550 A thick silver film, and a fused silica
prism. The unlabelled axis is equal to 1-normalized reflection intensity.
 215

Surface Plasmon Wave Silver Film

Sensed Medium
I Fig. Illustration of the SPR fiber-optic sensing element.
500 600 700 800 900 loo0

Wavelength (mn)
2.2. Principle of the SPRJiber-optic sensor
Fig. 4. The set of curves labeled ‘Water’ is the contour plot of the
three-dimensional SPR spectra illustrated in Fig. 3. The set of The SPR sensor has eliminated the need for the
curves labeled ‘Sucrose Sol’n’ is the contour plot of the theoretical coupling prism by depositing the metal directly onto the
spectra assuming a 37.1% concentration of sucrose. core of the fiber as illustrated in Fig. 5. The SPR
sensing method of fixed angle of incidence and modu-
lated wavelength was selected since the wavelength in-
wavelength. Note that the discrete peaks in Fig. 3 tensity distribution may be preserved in a fiber optic,
correspond to the use of discrete data points in order to whereas the angular intensity distribution of light will
create the graph. However, in theory the three-dimen- be indistinguishable due to mode mixing as a result of
sional surface is actually a continuous peaked surface. the inherent bending of the multimode fiber optic in
The spectrum was calculated using a matrix method to practical sensing applications. Furthermore, since the
determine the Fresnel reflection coefficients of a multi- sensor is fabricated on a multimode fiber there is not a
layered structure [lo]. The calculations were made as- fixed angle of incidence, but rather a range of incident
suming a fused silica prism, a 550 8, thick silver layer, a angles that are allowed to propagate in the fiber.
bulk dielectric of water, and TM-polarized light. The The multimode fiber optic has a numerical aperture
refractive-index values used in our SPR spectra calcula- of 0.36. The fiber supports internal propagating angles
tions for the silver layer [ 111, fused silica prism [ 121, of light from 90.0 to 78.5”, with respect to the normal
and water [13] were obtained from the literature. The of the core/cladding interface. Within this incident an-
change in the SPR coupling angle as a function of gular region the predicted SPR coupling wavelengths
wavelength shown in Fig. 3 is almost exclusively due to are between 560 and 620 nm for a bulk dielectric of
the increased magnitude of the silver complex refractive water, as shown in Fig. 4. Thus the light intensity at
index over the wavelength range 400-1000 nm. This these wavelengths is expected to be attenuated due to
large change in the silver refractive index (i.e., one the excitation of surface plasmon waves at the silver
order of magnitude) is compared to the relatively small film/bulk dielectric interface. An optical system re-
change in both silica and water refractive indices, 0.019 quired for this fiber-optic sensor must have the capabil-
and 0.028 index of refraction units, respectively. ity for coupling a wide range of optical wavelengths
Figure 4 is a contour plot of two theoretical three- into the optical fiber and for measuring the transmitted
dimensional SPR reflection spectra for water and a spectral intensity distribution (intensity versus wave-
sucrose solution (37.1 wt.%) as the bulk dielectric me- length) at the output of the fiber.
dia. This plot illustrates the resonance coupling angle
dependency upon wavelength, and the three-dimen- 2.3. Modeling of thefiber-optic signal
sional SPR spectra dependence upon the bulk refractive Three considerations must be addressed in order to
index. It is clear from Fig. 4 that a very sharp reso- model the theoretical SPR fiber-optic sensor transmit-
nance would be observed at the longer-wavelength re- ted spectral intensity distribution. These considerations
gion for the fused silica prism and silver metal layer are: (1) the three-dimensional SPR reflection spectra;
configuration. Thus, it explains why many researchers (2) the number of reflections each propagating mode
involved in SPR sensing have employed the method of undergoes; (3) the density of propagating modes in the
fixed wavelength and modulated angle of incidence at fiber-optic sensor.
long wavelengths. These contour plots illustrate that it Figure 6 illustrates the theoretical SPR spectra of
is also possible to measure a resonance spectrum by the reflected intensity for one reflection versus wavelength for
method of fixing the angle of incidence and modulating a number of angles propagating inside the fiber-optic
the excitation wavelength, which is the fundamental sensor (90,87,84,81, and 78” with respect to the normal
basis for the development of the SPR fiber-optic sensor. of the core/metal interface). However, the number of
216

90
1

0.15.

80 BZ 84 86 88 90

Angleof Incidence w/r to the Normal of the Core/Metal Interface

620 640
Fig. 8. Mode density as a function of incident angle, calculated
Wavelength (nm) assuming a Gaussian distribution of propagating angles of inci-
dence within the fiber-optic SPR sensor.
Fig. 6. Theoretical SPR spectra of reflected light intensity vs.
wavelength for a number of angles propagating inside the fiber-
optic sensor, i.e., 90, 87, 84, 81, and 78”. Thesespectraassumea
core material of fused silica, bulk dielectric of water, a 550A
silver film and TM polarization of the light.

B 0.8

fiB 0.6

%
cc 0.4
Fig. 9. Theoretical SPR fiber-optic spectra for a bulk refractive
G index of water and a 37.1% sucrose solution, assuming random
g 0.2
polarized light.

0
540 560 580 600 620 640 five of the propagating angles. The spectrum with the
Wavelength (nm) lowest-order mode, 90” angle of propagation, travels
Fig. 7. Theoretical SPR spectra taking into account the number
parallel to the meridonial axis of the fiber and does not
of reflections each angle encounters. reflect off the interface. Therefore, the expected spec-
trum is that of the 90” spectrum in Fig. 6 raised to the
power of zero, which yields a constant spectrum corre-
reflections in the fiber-sensor area is a function of the sponding to no surface plasmon waves excited by the
mode propagation angle, 0, as well as of the diameter of 90” propagating mode. Correspondingly, the smallest
the fiber, d, and the length of the sensing area, L. This propagating angle in the fiber (78”) has an effective
relationship is given by spectrum that is greatly broadened since it has 5.31
L reflections within the sensing length of the sensor.
Number of reflections = - (4) The SPR fiber-optic signal represents an accumulated
d tan(Q)
spectrum for the entire range of propagating angles,
The calculated numbers of reflections for the angles since one cannot measure the SPR spectrum for a
propagating in the fiber are 0.00, 1.31, 2.63, 3.96, and specific mode. However, the signal is not an equally
5.31 for incident angles of 90, 87, 84, 81, and 78”, weighted average of all the angles propagating in the
respectively. The number of reflections was calculated fiber; it must be weighted with the non-constant density-
assuming a 400 pm fiber diameter and a sensing area distribution function of propagating angles in the fiber
length of 10 mm. (Fig. 8). The propagation angle density-distribution
To determine the effective SPR spectra (taking into function is assumed to be Gaussian, spanning the range
account multiple reflections), the spectrum for a single of the allowed propagating angles in the fiber. Figure 9
reflection (Fig. 6) is raised to the power of the number represents the theoretical SPR fiber-optic signal ob-
of reflections the specific propagating angle undergoes tained by weight averaging the angular spectra in Fig. 7
with the sensor interface. Figure 7 illustrates these with the density distribution of Fig. 8 for the bulk
effective SPR spectra of intensity versus wavelength for chemical water sample and a 37.1% concentration of
 217

sucrose. An approximate 200 run shift in the SPR cou- evaporator system in an ~r~gement whereby the flux
pling wavelength is observed; it corresponds to a of the evaporated metal was perpendicular to the axis
change of 0.06 index of refraction units of the chemical of the fiber. The fibers were then rotated during silver
sample. deposition, resulting in a 550 A silver film deposited
symmetrically about the fiber core. The deposition pro-
cess was monitored using a quartz-crystal detector.
Six sample solutions of high-fructo~ corn syrup di-
luted with deionized water were prepared, resulting in
The block diagram of the experimental set-up is various real refractive indices. The real refractive in-
shown in Fig. 10. The output of a tungsten-halogen dices of these sample solutions (1.333, 1.351, 1.364,
lamp is focused into a fiber optic. An SPR sensing 1.381, 1.393, and 1.404) were independently measured
surface has been fabricated in the middle section of the using an Abbe refractometer (Milton Roy Tabletop
length of fiber. A mode scrambler is used to populate Ref~ctometer 3L) at the 589 nm ~velength. The same
all the modes of the fiber optic. The SPR sensing experimental procedure was used for measuring the
surface is enclosed by a 3 ml flow cell constructed using transmitted fiber-optic SPR spectra in all three sensors.
a syringe with two syringe stoppers and inlet and outlet First, the transmitted spectral intensity distribution was
ports. The output of the fiber-optic sensor is connected measured while air was in the flow cell. Secondly, the
to a fiber-optic spectrograph (American Holographic, transmitted spectral intensity distribution was measured
Littleton, MA) via an SMA connector. The flat field for each of the six prepared sucrose solutions. 15 ml of
grating is an American Holographic model ##6.33, each sample solution was introduced via a syringe at
which disperses a range of wavelengths from 400 to the input port of the flow cell.
900 nm with a linear dispersion value of 20 run/mm.
The detector inside the spectrograph is a 1024-element
CCD linear array detector. The theoretical wavelength 4. Experimentalresults and discussion
resolution dete~ned by the linear dispersion value of
the grating and the 25.4 pm width of the CCD elements The air spectrum was used as a reference to normal-
is 0.5 nm. A data-acquisition board is used with an ize all the SPR fiber-optic sensor spectra taken of the
IBM-compatible computer for automated data acquisi- sample solutions, since there was no surface plasmon
tion. resonance excitation in this wavelength range for the
The fiber optic chosen for the experiment was a bulk refractive index of air, 1.0. This calibration pro-
silica~pol~er fiber (3M, Mi~~polis, MN, type FP- vided an effective means to measure the system transfer
400 LJHT). The ~0~~~760 pm (core, cladding and function, attributed by the light-source spectral output,
buffer diameters, respectively) fiber had a numerical the photodiode array spectral sensitivity, and the fiber
aperture of 0.36. The buffer and cladding layers were spectral absorbance. Figure 11 shows the normalized
removed using a technique employed by Weber and transmitted light intensity as a function of wavelength
Schultz [ 141.A hobby torch was used to burn away the measured by the 1Omm exposed core sensor for the
layers, and the resultant surface was wiped with Dyna- fructose solutions with refractive indices of 1.351, 1.393,
solve 100 (Dynaloy Inc.). and 1.404. The r~on~~ wavelength shift for increas-
Three fiber-optic sensors were fabricated by removing
6, 10, and 18 mm in length of the cladding/buffer layers.
The sensors were then mounted in an electron-beam

Wavelength(MI)
Fig. 11. SPR spectra collcctcd using the SPR fiber sensor with
1Omm of fiber core exposed. The spectra were cokctad from
Fig. 10. Illustration of the SPR fiber-optic experimentalapparatus. fructose samples with refractive indices of 1.351,1.393,and 1.404.
218

0.6
l

500 600 700 800 9c

Wavelength (nm)

Fig. 12. SPR spectra collected from all three fabricated sensors
j&//y jg
with the same fructose concentration with a refractive index of
1.351. 1.33 1.34 1.35 1.36 1.37 1.38 1.39 1.4 41
Real Refractive Index

Fig. 13. Experimental and theoretical SPR fiber-optic coupling

ing bulk refractive indices was consistent with the theo- wavelength response as a function of the refractive index of the
retical results illustrated in Fig. 9. However, the reso- chemical sample.
nance spectra appear to be broader than expected. This
may be partly attributed to the fact that the input fiber refractive index is 2.5 x 10V4at a wavelength of 500 nm
to the fiber-optic spectrograph is large, 400 pm, and and 7.5 x 10e5 at a wavelength of 900 nm, assuming the
thus the full spectrograph resolution is not optimized. optimal wavelength resolution of the spectrograph is
Narrower resonance spectra may be achieved by reduc- 0.5 nm.
ing the diameter of the input fiber to the spectrograph The experimental results presented verify the sensing
or by employing a narrow slit. mechanism of the surface-plasmon-resonance-based
Figure 12 is a plot of the SPR fiber-optic sensor fiber-optic sensor. Currently, we are directing our effort
spectra measured by the 6, 10, and 18 mm exposed core toward the optimization of the fiber sensor, taking into
sensors for a single fructose solution. The transmitted consideration type of metal, length of sensing area,
spectral-intensity distribution depends upon the length fiber-optic core material, numerical aperture, and fiber
of SPR sensing area. As shown in Fig. 12, the spectrum diameter, as well as a new sensing head for possible
measured by a longer sensing area exhibits a deeper disposable sensing applications.
resonance for the reasons discussed in Section 2.3. It
should be noted that due to the cylindrical geometry of
the fiber, both TE- and TM-polarized light with respect 5. Improvedsensor and system design
to the core/metal interface are allowed to propagate in
the multimode fiber optic. The expected optimal trans- An improved sensor and system configuration has
mitted light intensity at resonance is 0.50 rather than been developed that has many advantages for practical
0.0, since SPR can only be excited with TM-polarized chemical-sensing applications, including: (1) elimina-
light. tion of flow-ceil limitations; (2) the possibility of dis-
The theoretical and experimental SPR coupling posable fiber-sensor probes; (3) amenable for use in
wavelength versus the refractive index of the fructose remote sensing applications,’ The improved sensor and
sample solutions are plotted in Fig. 13 for all three SPR system design, illustrated in Fig. 14, utilizes a termi-
fiber-optic sensors. The response of all three sensors is nated sensor. This terminated reflection-based SPR
in good agreement with the calculated values. In order fiber-optic sensor is based on the same principles as the
to plot the Abbe refractometer measured refractive in-line transmission-based sensor. However, the termi-
indices of the sample solutions in Fig. 13, the values nated sensor design utilizes a microfabricated mirror at
were corrected for wavelength using the refractometer the end of the probe to reflect the light back through
dispersion measurement table [ 151.The theoretical sen- the fiber. In this configuration the light travels through
sitivity of the fiber-optic SPR sensor to refractive index the sensing area twice, thus the sensing length must be
can be calculated from the SPR wavelength response one half the length of the sensing area of the in-line
curve. Because the response is non-linear, the sensitivity transmission-based sensor.
will be a function of wavelength, with better sensitivity In the improved sensor configuration light is coupled
at the longer wavelengths. The theoretical sensitivity to into a single branch of a 50:50 two-way fiber-optic
 219

fiber-optic SPR sensing probes (capable of remote and

in situ monitoring) will likely be developed toward
applications in process analytical chemistry, environ-
mental monitoring, and biochemical and biomedical
P sensing.

Acknowledgements

The authors wish to thank Frank Lawrence for his

assistance in the sensor fabrication, fructose solution
preparation,anddataacquisition. WealsothankDr Lloyd
Burgess, Dr Martin Afromowitz and Kevin Kuhn for their
helpful discussions, Dr Robert Darling for use of his
electron-beam evaporation system, and Kyle Johnston for
his assistance in fabricating the ‘reflected’sensor configu-
2Mx)A Silver Mirror ration. This work is supported by the Center for Process
Sensor
Analytical Chemistry (CPAC) an NSF/Industry Cooper-
Fig. 14. Illustration of the SPR fiber-optic sensor and sensor ative Research Center at the University of Washington
system.
and an NSF grant, # EID-9212314.

splitter from a tungsten-halogen lamp. 50% of the References

coupled light is then transferred to the probe branch of
the splitter. An SMA connector is used to connect the 1 B. Liedberg, C. Nylander and 1. LundstrGm, Surface plasmon
splitter to the disposable SPR fiber-optic sensor probe. resonance for gas detection and biosensing, Sensors and Actu-
The light is transmitted down the probe to the sensor ators, 4 (1983) 299-304.
2 P. B. Daniels, J. K. Deacon, M. J. Eddowes and D. Pedley,
area and reflected back up the probe by the use of a Surface plasmon resonance applied to immunosensing,Sensors
microfabricated mirror. The returned light is split again and Actuators, 15 ( 1988) I1 17.
and connected to a fiber-optic spectrograph, which is 3 R. C. Jorgenson, S. S. Yee, K. K. Chittur and L. W. Burgess,
used to measure the spectral intensity of the signal light. In-situ characterization of adsorbed protein films using surface
A reference signal can also be measured by use of the plasmon resonance, IEEE/_&g. Medicine Biology Sot. Proc.,
Vol. 12, 1990, pp. 440-442.
remaining splitter arm.
4 J. Gent, P. Lambeck, H. Kreuwel, G. Gerritsma, E. Sudholter,
The results obtained from the reflected sensor D. Reinhoudt and J. Popma, Optimization of the chemooptical
configuration are consistent with the data and theory of surface plasmon resonance based sensor, Appl. Opt., 29 (1990)
the in-line transmission-based SPR fiber-optic sensor 2843-2849.
presented in this paper. 5 K. Matsubaru, S. Kawata and S. Minami, Optical chemical
sensor based on surface plasmon measurement, Appl. Opt., 27
(1988) 1160-1163.
6 E. Kretschmann and H. Raether, Radiative decay of non-
6. Conclusions radiative surface plasmons excited by light, 2. Naturforsch., Teil
A, 23 (1968) 2135-2136.
7 J. Gent, P. Lambeck, H. Kreuwel, G. Gerritsma, E. Sudholter,
A fiber-optic SPR sensor has been presented that
D. Reinhoudt and J. Popma, Chromoionophores in optical ion
utilizes the sensitivity of surface plasmon resonance sensors, Sensors and Actuators, 17 (1989) 297-305.
combined with fiber-optic technology. The sensor signal 8 E. Fontana, R. H. Pantell and M. Moslehi, Characterization of
and response were theoretically modelled and the opti- dielectriccoated, metal mirrors using surface plasmon spec-
mal sensitivity was found to be between 4.5 x 10m4and troscopy, Appl. Opt., 27 (1988) 3334-3339.
9 R. Jorgenson, Surface plasmon resonance as an optical probe
7.5 x lo-’ index of refraction units. T’his is comparable
of the metal/dielectric interface, Master’s Thesis, University of
to the sensitivity of traditional SPR bulk-optic systems. Washington, 1991.
The sensing principle of the device was based upon IO C. Jung, Surface plasmon resonance, Master’s Thesis, Univer-
fixed angle of incidence and modulated wavelength sity of Washington, 1991.
measurements, instead of the more traditional fixed 11 G. Hass and L. Hadley, Optical properties ofmetals, in D. Gray
(ed.), American Instifute of Physics Handbook, McGraw-Hill,
wavelength and modulated angle of incidence measure-
New York, 1972, pp. 149-151.
ments used in bulk-optic SPR systems. 12 M. Querry, D. Wieliczka and D. Segelstein, Water (H,O), in
The measured responses of the sensing probes were E. Palik (ed.), Handbook of Optical Constants of Solids II,
in good agreement with the theoretical model. The Academic Press, Boston, 1991,pp. 1059-1077.
220

13 I. H. Malitson, Interspecimen comparison of the refractive Battelle Memorial Institute (1987, 1988) investigating
index of fused silica. J. Oat. Sot. Am.. 55(1%5) 1205-1216. bulk-optic surface plasmon resonance for the measure-
14 A. Weber and .I. S. &h&z, Fiber-optic fludrimefry in biosen-
ment of in situ protein adsorption. His main research
sors: comparison between evanescent wave generation and
distal-face generation of fluorescent light, Biosensors Bioelec-
interests are in chemical and biomedical optical sensors.
won., 7 (1992) 193-197.
15 Abbe Refractometer Operator’s Manual and Dispersion Table, Sinclair S. Yee received B.S., M.S., and Ph.D. degrees
Milton Roy Company, Analytical Products Division, in electrical engineering from the University of Califor-
Rochester, New York, 1986.
nia, Berkeley, in 1959, 1961, and 1965, respectively. In
1964, he joined the Lawrence Livermore Laboratory as
Biographies a research engineer working on semiconductor devices.
Since 1966 he has been on the faculty of the Department
Ralph C. Jorgenson received B.S. and MS. degrees in of Electrical Engineering at the University of Washing-
electrical engineering from the University of Washing- ton, where in 1974 he became a full professor. In
ton in 1988 and 1991, respectively. He is currently a 1972-1974, he was an NIH Special Research Fellow,
Ph.D. candidate in the Department of Electrical En- and since 1979, he has been an IEEE Fellow. His main
gineering at the University of Washington. He has research interests are concerned with silicon-based and
interned at International Biomedics Inc. (1986), re- GaAs quantum-well optical modulator devices and sur-
searching an in vivo fiber-optic blood-gas sensor and at face plasmon resonance, as well as microsensor arrays.