High-resolution surface plasmon resonance sensor
based on linewidth-optimized nanohole array
Kevin A. Tetz, Lin Pang, and Yeshaiahu Fainman
Department of Electrical and Computer Engineering, University of California, San Diego, California 92093
Received December 15, 2005; revised February 8, 2006; accepted February 11, 2006; posted March 1, 2006 (Doc. ID 66681)
A high spectral resolution, 2D nanohole-array-based surface plasmon resonance sensor that operates at nor-
mal or near normal incidence—facilitating high spatial resolution imaging—is presented. The angular and
spectral transmittance of the structure is modified from a Fano type to a pure Lorentzian line shape with a
parallel and orthogonal polarizer–analyzer pair. This change leads to a linewidth narrowing that maximizes
the sensor resolution, which we show to be of O?10−5? refractive index units (RIU). We estimate the potential
of this system of O?10−6? RIU under optimal conditions. © 2006 Optical Society of America
OCIS codes: 130.6010, 230.3990, 240.6680, 260.3910.
A wide variety of surface plasmon resonance (SPR)
sensors have been demonstrated. The most common
configurations employ the Kretschmann geometry or
a shallow grating coupler and monitor the resonance
shift (reflection minima) as a function of angle, wave-
length, or simply as differential intensity.1More re-
cent approaches have included phase sensitive varia-
tions, demonstrated in both interferometric2as well
as ellipsometric configurations.3One major drawback
with the conventional coupling methods is the diffi-
culty in incorporating the sensor elements in high
NA aperture imaging systems to increase spatial
resolution and the corresponding number of resolv-
able spots. Operating on a prism or with a first-or
higher-order diffraction mode places severe con-
straints on the depth of focus in the imaging system
needed for large arrays of assays. Massive parallel-
ism, and hence high throughput, is of primary impor-
tance in many potential SPR sensor applications, but
they are severely limited by most of the current de-
Recently, Ebbesen et al.6demonstrated enhanced
transmission through subwavelength hole arrays,
which is generally attributed to the resonant excita-
tion of surface plasmon polariton (SPP) waves. Such
structures may be exploited for sensor applications
due to their potential for significantly decreasing the
interrogation volumes while operating at normal or
near-normal illumination. This leads to a high pack-
ing density, minimal analyte volumes, and a large
number of parallel channels while facilitating high-
resolution imaging and wide field of view, supporting
a large space–bandwidth product (SBP). These ad-
vantages may make such devices preferable in a
number of applications despite the fact that the ulti-
mate spectral resolution is lower than the prism-
based equivalent because of SPR broadening due to
both radiative and material damping. Several au-
thors have suggested and demonstrated the use
applications,7,8and there are numerous numerical
and experimental studies on their spectral proper-
ties. In this Letter we demonstrate a SPR sensor
based on a metal film, perforated by a nanohole ar-
ray, and specifically show polarization properties that
facilitate narrowing the transmission linewidth (and
hence maximize resolution) while operating in a re-
gime that facilitates high SBP imaging.
Samples for our experiments are fabricated by de-
positing gold films of ?200 nm on glass substrate fol-
lowed by spin coating and patterning by holographic
lithography to achieve large usable areas ??1 cm2?.
Multiple exposures of a chemically amplified nega-
tive resist (SU-8) yield a 2D array of nanoholes, and
the exposure time and postexposure baking step al-
low fine control of the hole diameter9??200 nm?. To
facilitate large SBP imaging, we choose the period a
of the array to be close to the wavelength ? of the ex-
citation field ?a/??1? with the fabricated value of a
=1.4 ?m. The developed SU-8 is used as a mask for
etching nanoholes into the gold film by using induc-
tively coupled plasma (ICP)/reactive ion etching
(RIE) dry etching,and
(PDMS) mold with a microfluidic delivery channel
1 cm?2 mm?100 ?m is then bonded to the sub-
strate by an oxygen plasma. Measurements are car-
ried out using a simple setup shown in Fig. 1, where
1570 nm,6 dBm? of ?1 cm in diameter is used to ex-
cite a SPP field in the 2D nanohole array. The sample
is inserted between a polarizer–analyzer pair, and
the transmitted light is used to simultaneously im-
age an area of ?200 ?m?200 ?m of the sample onto
an InGaAs camera for alignment as well as a photo-
diode for transmission measurements.10Angular in-
terrogation is achieved using a mechanical rotation
stage rotating the sample in the y–z plane. We refer
to two polarization states in our measurements. (i)
Parallel polarizer–analyzer (PP): polarizer and ana-
lyzer axes are parallel and oriented at +?/4 with re-
spect to the [1,0] direction of the nanohole array (see
Fig. 1) yielding equal electric field amplitudes in the
x and y directions. (ii) Orthogonal polarizer–analyzer
(OP): the polarizer (analyzer) axis is oriented at +?/4
?−?/4? with respect to the [1,0] direction. Resonant
transmittance through the 2D nanohole array de-
pends on the interrogation angle and wavelength of
OPTICS LETTERS / Vol. 31, No. 10 / May 15, 2006
0146-9592/06/101528-3/$15.00© 2006 Optical Society of America
radiation and has a Fano-type11–13line shape for PP
and a Lorentzian shape for OP (see Fig. 1). There
have been a number of studies that have investigated
and explained the effects of the various geometric pa-
rameters on the shape of the resonant transmission
(e.g., hole size, metal film thickness, and optical prop-
erties of the metal), and we note that the critical fea-
ture (assume a relatively “thick” film) is the hole di-
ameter, which increases the scattering rate and
hence broadens the resonance linewidth.14This reso-
nant transmission mechanism, as elucidated by Bar-
et al.,15involves coupling to an SPP mode, evanescent
transmission through the below-cutoff waveguide
hole, and scattering of radiation again from the hole
array to produce propagating free-space modes. The
surface wave is excited by a projection of the incident
direction,16and the reradiated field is again projected
onto the analyzer. This effect has been explored pre-
viously with 1D gratings17and utilized in imaging
SPPs excited on 2D SPP grating couplers.
Normalized transmittance spectra for both wave-
length and angular interrogation are shown in Fig. 2
in the vicinity of ?−1,0? type SPP modes with an air
overlayer. We observe a characteristic Fano shape for
PP (dotted curves) and a pure Lorentzian shape for
OP (solid curves); with OP the background contribu-
tion is suppressed, leaving only the resonance compo-
nent of the transmission. The absolute transmittance
is low, −23 dB (0.50%) for PP, due to the small size of
the diameter of the holes (thus yielding relatively
narrow lines), and drops to approximately −29 dB
(0.13%) for OP due to additional polarization projec-
tion onto the analyzer. The extinction ratio of
?15–20 dB, limited by the linewidth as well as depo-
larization due to surface roughness in the etched
holes, is shown in the inset of Fig. 2(b). Under wave-
length interrogation the background level does not
drop to the same deep minimum levels within the
tuning range of our laser. The measured FWHMs for
wavelength interrogation [Fig. 2(a)] are 1.28 meV
?2.47 nm? and −2.86 meV ?5.53 nm? for OP and PP,
respectively, and the PP transmission peak is red-
shifted from that in OP by 0.40 meV ?0.77 nm?. Simi-
larly, the measured FWHMs for angular interroga-
tion [Fig. 2(b)] are 0.0012 ak?/2? (0.092°) and 0.011
(0.87°) for OP and PP, respectively, and the corre-
sponding peak shift is 0.0005 (0.04°).
through the 2D nanohole array for sensor applica-
tions by introducing an index-calibrated solution
through the microfluidic channel to create a con-
trolled gold–fluid interface. We repeat our experi-
ments on angular and wavelength interrogation ex-
citing the ?+1,0? type SPP modes and vary the
refractive index of the overlayer fluid (varying con-
centrations of Na2CrO4in H2O). Since the resolving
power and interrogation range are both higher in an-
gular interrogation, we focus our following study on
it. Figure 3 shows experimental results on the posi-
tion of the resonant transmission peak through angu-
tion of (a) energy (wavelength) and (b) parallel wave vector
(angle). In each case the dotted curves correspond to the PP
and the solid curves correspond to the OP polarization
states (as illustrated in Fig. 1 and described in the text).
The transmission in each case has been normalized to the
maximum to clearly illustrate the respective line-shape
functions. Inset in (b) are the same data plotted in a loga-
rithmic scale to show the ?15–20 dB background level re-
duction for the Lorentzian versus Fano-type resonances.
(Color online) Normalized transmission as a func-
nanohole-array-based SPR sensor. The input and output
polarization states of a tunable laser are controlled, provid-
ing variable spectral or angular Fano-type profiles. A mi-
crofludic channel is used to transport the analyte fluid to
the surface of the sensing area and can be used to control
the refractive index on the metal–dielectric interface to
tune the SPP resonance frequency. Also shown is a scan-
ning electron microscopy image of a representative sample.
(Color online) Conceptual diagram of the 2D
May 15, 2006 / Vol. 31, No. 10 / OPTICS LETTERS
lar interrogation versus the index of refraction of the Download full-text
fluid on the interface. Due to the strong absorption of
water in this wavelength range, the linewidths for
wavelength and angular interrogation broaden to
values of 4.32 meV ?8.31 nm? and 0.0064 ak?/2?
(0.52°), respectively, with OP. At shorter wavelengths
the damping due to water is reduced, but the metal
losses are larger. Also, at shorter wavelengths there
is a greater mode overlap of the resonant field with
the reaction of interest as the extent of the mode into
the dielectric is reduced. Error bars in the horizontal
direction are from uncertainty in the solution index
of refraction as well as variations in temperature.
Peak positions are determined by both the method of
moments (centroid position) and by fitting Lorentz-
ian functions, and the error bounds for these methods
in the presence of noise are shown as the various
shaded regions. This procedure corresponds to esti-
mated sensing limits of 5?10−6and 1?10−5refrac-
tive index units (RIU) for OP and PP, respectively.
The darkest region corresponds to the observed error
1.7?10−3(standard deviation) due to lack of full op-
timization in the feedback controls and therefore lim-
ited our direct measurement limit to ?1.5?10−5. We
estimate the limits for a nonabsorbing overlayer
(with a gaseous analyte, for example) with OP and an
optimized rotation stage (mechanical limits of ?10−4
in angle) to be of the order 1?10−6, which is shown
with the lightest shading.
While the peak position is typically determined
more precisely, it is useful to introduce the metric
??,??S?,?/??,?, which is a measure of the resolving
power that facilitates comparisons of different sen-
sors and interrogation methods.18Here S is the sen-
sitivity (i.e., derivative of the resonance position with
respect to the index of refraction), ? is the FWHM,
and the subscripts ? and ? refer to wavelength and
angular interrogations, respectively. We experimen-
S??1022±8 nm RIU−1
?78.4±0.6 deg RIU−1
?850 RIU−1and ???410 RIU−1with an air over-
?150 RIU−1and ???120 RIU−1with water broad-
ened transmission. These values compare favorably
?83?48? RIU−1for prism (grating) -based sensors, re-
spectively, at 850 nm.
We have demonstrated a high-resolution SPR sen-
sor based on transmission through nanohole arrays.
The transmission line-shape function was shown to
vary with the input and output polarization states,
showing a Fano-type dependence with a pure reso-
nant Lorentzian of minimal width when these two
states are orthogonal. In these structures, the SPP
propagation length may be reduced (from approxi-
mately several tens of micrometers in this case) to in-
crease the spatial resolution and limit the cross talk
between channels. This leads to a design trade-off
where the spectral or angular resolution (resolving
power) may be sacrificed for smaller interrogation
volumes depending on the particular application. In
addition, one can break the in-plane symmetry and
use, for example, elliptical16or chiral shaped holes to
have polarization dependence even at normal inci-
dence. These results will aid in the development of
future nanohole SPR sensors.
We thank the DARPA Optofluidics Center, NSF,
and AFOSR for support. K. A. Tetz’s e-mail address is
1. J. Homola, S. S. Yee, and G. Gauglitz, Sens. Actuators
B 54, 3 (1999).
2. S. Y. Wu, H. P. Ho, W. C. Law, C. L. Lin, and S. K.
Kong, Opt. Lett. 29, 2378 (2004).
3. I. R. Hopper and J. R. Sambles, Appl. Phys. Lett. 85,
4. J. Dostalek, J. Homola, and M. Miler, Sens. Actuators
B 107, 154 (2005).
5. Y. D. Su, S. J. Chen, and T. L. Yeh, Opt. Lett. 30, 1488
6. T. W. Ebbesen, H. J. Lezec, H. F. Ghaemi, T. Thio, and
P. A. Wolff, Nature 391, 667 (1998).
7. A. G. Brolo, R. Gordon, B. Leathem, and K. L.
Kavanagh, Langmuir 20, 4813 (2004).
8. T. Rindzevicius, Y. Alaverdyan, A. Dahlin, F. Hook, D.
S. Sutherland, and M. Kall, Nano Lett. 5, 2335 (2005).
9. L. Pang, W. Nakagawa, and Y. Fainman, Appl. Opt. 42,
10. K. A. Tetz, R. Rokitski, M. Nezhad, and Y. Fainman,
Appl. Phys. Lett. 86, 111110 (2005).
11. V. Lomakin and E. Michielssen, Phys. Rev. B 71,
12. C. Genet, M. P. van Exter, and J. P. Woerdman, Opt.
Commun. 225, 331 (2003).
13. M. Sarrazin, J. P. Vigneron, and J. M. Vigoureux, Phys.
Rev. B 67, 085415 (2003).
14. R. Muller, V. Malyarchuk, and C. Lienau, Phys. Rev. B
68, 205415 (2003).
15. W. L. Barnes, W. A. Murray, J. Dintinger, E. Devaux,
and T. W. Ebbesen, Phys. Rev. Lett. 92, 107401 (2004).
16. J. Elliott, I. I. Smolyaninov, N. I. Zheludev, and A. V.
Zayats, Opt. Lett. 29, 1414 (2004).
17. S. J. Elston, G. P. Bryan-Brown, and J. R. Sambles,
Phys. Rev. B 44, 6393 (1991).
18. J. Homola, I. Koudela, and S. S. Yee, Sens. Actuators B
54, 16 (1999).
whilethese values are reduced to
Fig. 3. (Color online) Resonance peak position shift versus
refractive index change (i.e., salt concentration in water) in
the fluidic overlayer. Black line, linear fit to the data.
Shaded regions, approximate peak position (absolute re-
fractive index) errors in the fitting procedure for the OP
and PP conditions for both air and water broadened line-
widths as well as estimated theoretical resolution limits.
OPTICS LETTERS / Vol. 31, No. 10 / May 15, 2006