On-Chip Surface-Based Detection with Nanohole
Angela De Leebeeck,†L. K. Swaroop Kumar,‡Victoria de Lange,‡David Sinton,*,†
Reuven Gordon,*,‡and Alexandre G. Brolo*,§
Department of Mechanical Engineering, Department of Electrical and Computer Engineering, and Department of Chemistry,
University of Victoria, Victoria, BC, Canada
A microfluidic device with integrated surface plasmon
resonance (SPR) chemical and biological sensors based
on arrays of nanoholes in gold films is demonstrated.
Widespread use of SPR for surface analysis in laboratories
has not translated to microfluidic analytical chip plat-
forms, in part due to challenges associated with scaling
down the optics and the surface area required for common
reflection mode operation. The resonant enhancement of
light transmission through subwavelength apertures in a
metallic film suggests the use of nanohole arrays as
miniaturized SPR-based sensing elements. The device
presented here takes advantage of the unique properties
of nanohole arrays: surface-based sensitivity; transmis-
sion mode operation; a relatively small footprint; and
repeatability. Proof-of-concept measurements performed
on-chip indicated a response to small changes in refractive
index at the array surfaces. A sensitivity of 333 nm per
refractive index unit was demonstrated with the integrated
device. The device was also applied to detect spatial
microfluidic concentration gradients and to monitor a
biochemical affinity process involving the biotin-strepta-
vidin system. Results indicate the efficacy of nanohole
arrays as surface plasmon-based sensing elements in a
microfluidic platform, adding unique surface-sensitive
diagnostic capabilities to the existing suite of microfluidic-
based analytical tools.
Surface plasmons (SPs) are surface-bound electromagnetic
waves formed at the interface between a metal and a dielectric.
Although SPs from a smooth metallic surface cannot be directly
excited by light, the surface plasmon resonance (SPR) condition
can be achieved by either prism or grating coupling.1,2Due to
their interfacial nature, SPs are very sensitive to the near-surface
dielectric constant (index of refraction) and are well-suited to the
detection of surface binding events. This characteristic has been
exploited in a wide variety of SPR-based sensors, particularly
biosensors,3,4for which commercial devices are available.5The
most common methodology involves the Kretschmann configu-
ration,1where a prism is used for the light-SP coupling at the
surface of a thin gold film (∼50 nm). The probe light undergoes
total internal reflection on the inner surface of the prism. At the
SPR angle, an evanescent light field travels through the gold thin
film and excites SPs at the metal-dielectric interface, reducing
the intensity of the reflected light. The SPR angle is sensitive to
several factors,6including the wavelength of light, the thickness
of the gold film, the physical and optical properties of the prism,
and the index of refraction of the medium near the metallic
interface (which is typically <200 nm6). The latter sensitivity,
combined with signal averaging over a relatively large area,7is
commonly exploited in these sensor applications. Device-level
miniaturization of SPR-based sensing and integration with micro-
fluidics has been limited,7,8in part due to the following: the utility
of averaging over a large surface; the challenges associated with
scaling down the prism and optical infrastructure needed for
reflection mode operation; and difficulty in incorporating high
numerical aperture optics to achieve high spatial resolution.
Recently, SPR reflectance imaging was applied to monitor con-
centration during the replacement of water with ethanol in a
straight microfluidic channel7and to visualize the mixing in a flow
cell.8In those studies, the prism surface comprised one wall of
the microchannel, and the concentration changes were determined
from SPR reflectance imaging.
SPR may also be tailored via nanometer-scale structures,9and
the ability to pattern such structures in metallic films presents
many opportunities. Ebbesen et al.10reported enhanced transmis-
sion of light through arrays of subwavelength holes in optically
thick metallic films, at normal incidence. The enhanced transmis-
sion is attributed to initial scattering of the incident light into SPs
that penetrate the nanoholes and are again scattered on the other
side of the film.9The extent of SP generation depends on the
combination of incident light wavelength, hole geometry/periodic-
* To whom correspondence should be addressed. E-mail: email@example.com.
Phone: 250-721-8623. Fax: 250-721-6051. E-mail: firstname.lastname@example.org. Phone: 250-
472-5179. Fax: 250-721-6052. E-mail: email@example.com. Phone: 250-721-7167.
†Department of Mechanical Engineering.
‡Department of Electrical and Computer Engineering.
§Department of Chemistry.
(1) Nice, E. C.; Catimel, B. BioEssays 1999, 21, 339-352.
(2) Homola, J. Anal. Bioanal. Chem. 2003, 377, 528-539.
(3) Homola, J.; Yee, S. S.; Gauglitz, G. Sens. Actuators, B 1999, 54, 3-15.
(4) Cooper, M. A. Anal. Bioanal. Chem. 2003, 377, 834-842.
(5) Mullett, W. M.; Lai, E. P. C.; Yeung, J. M. Methods 2000, 22, 77-91.
(6) Jung, L. S.; Campbell, C. T.; Chinowsky, T. M.; Mar, M. N.; Yee, S. S.
Langmuir 1998, 14, 5636-5648.
(7) Kim, I. T.; Kihm, K. D. Exp. Fluids 2006, 41, 905-916.
(8) Iwasaki, Y.; Tobita, T.; Kurihara, K.; Horiuchi, T.; Suzuki, K.; Niwa, O. Meas.
Sci. Technol. 2006, 17, 3184-3188.
(9) Barnes, W. L.; Dereux, A.; Ebbesen, T. W. Nature 2003, 424, 824-830.
(10) Ebbesen, T. W.; Lezec, H. J.; Ghaemi, H. F.; Thio, T.; Wolff, P. A. Nature
1998, 391, 667-669.
Anal. Chem. 2007, 79, 4094-4100
Analytical Chemistry, Vol. 79, No. 11, June 1, 2007
10.1021/ac070001a CCC: $37.00 © 2007 American Chemical Society
Published on Web 04/21/2007
ity and material/medium dielectric constants. For wavelengths
that satisfy SPR conditions for a given system, incident light
transmission can be higher than that expected based on area of
the nanoholes. This finding is in stark contrast to classical theory11
that would predict a much lower transmission through subwave-
length apertures. The result is a wavelength- and medium-sensitive
subwavelength focusing mechanism that has potential applications
in several areas including optical filters,12imaging,13nanolithog-
raphy,14,15photonic circuits,16and others. The central role of SPs
in the enhanced transmission has motivated the application of
nanohole arrays to surface-based chemical and biological detec-
tion. The unique optical properties of nanohole arrays and their
application in this area have recently been investigated by our
In the context of on-chip chemical and biological analysis,
nanohole arrays have several unique advantages: (1) in contrast
to reflective mode SPR, transmission mode operation at normal
incidence simplifies alignment, facilitates the use of high numerical
aperture optics,29and permits eventual device-level miniaturization
and integration of supporting optics; (2) the footprint of a nanohole
array is small relative to that typically required in reflective mode
SPR,28enabling miniaturization and integration into microfluidic
architectures, and higher spatial resolution; (3) in contrast to other
local surface plasmon strategies based on colloidal nanoparticles,32
or roughened surfaces, nanohole arrays can be fabricated with
high reproducibility; (4) the high sensitivity of the optical response
to hole shape,22,33periodicity, and lattice versus basis orientation23
provides a large variety of handles with which to tailor array/
sensor response; and (5) the unique geometry differentiating each
array element is fixed within the structure and is generally more
robust than an adsorbed probe. A perceived disadvantage of
grating-based SPR systems is reduced sensitivity compared to the
Kretschmann configuration; however, significantly improved sen-
sitivities have recently been achieved for arrays of nanoholes.28,29
Motivated by the above advantages, we have demonstrated the
application of nanohole arrays to detect the binding of organic
and biological molecules.19In that case, a shift of the SPR peak
in the transmission spectrum from normal incident white light
was detected following the adsorption of a mercaptoundecanoic
acid monolayer and also subsequent to absorption of a bovine
serum albumin. The sensitivity was found to be 400 nm per
refractive index.19Lui et al. presented an affinity biosensor platform
utilizing enhanced fluorescence transduction and demonstrated
that enhanced fluorescence output may be obtained in both
periodic and randomly distributed nanocavities.26The polarization
properties of nanohole arrays have been exploited to achieve
significantly higher sensitivity at near-normal incidence.29In work
to date, the confinement offered by a simple microfluidic channel
has already proved useful in the delivery of solutions to an
array.26,29The focus, however, has been on the characteristic
optical phenomena from an isolated nanohole array and the unique
sensing opportunities provided. The incorporation of arrays of
nanohole arrays within a microfluidic framework for spatial and
temporal detection on-chip is a natural progression of this
Here we demonstrate nanohole arrays as discrete, 20 µm ×
20 µm, sensing elements capable of both chemical sensing and
biomolecule adsorption monitoring for spatial/temporal measure-
ments in a microfluidic chip platform. The integrated device is
manufactured using established soft lithographic microfabrication
and focused ion beam nanofabrication. Sensitivity and proof-of-
concept tests are performed using a microscope and fiber-optic
spectroscopy. The sensor is applied to detect changes in refractive
index in sucrose solutions of varying concentration, to measure
cross-stream concentration gradients, and to monitor surface
Nanohole Array Fabrication. Figure 1 shows the arrange-
ment of nanoholes and integration into the microfluidic chip. The
nanohole arrays were fabricated by focused ion beam milling on
a 100-nm optically thick commercially coated gold film on a 25.4
mm × 25.4 mm glass substrate. The arrays were fabricated and
imaged using a FEI 235 dual-beam gallium ion beam and field
emission scanning electron microscope. The ion beam was set to
30 keV with a milling rate of 1.6 nm/µs, and the beam current
was 300 nA. The nanoholes were ∼150 nm in diameter and were
made to form two rows of six arrays each. One row contained
arrays of periodicity (center-to-center spacing between holes) 350,
450, 550, 650, 750, and 850 nm, and the second row contained 6
arrays of 650 nm periodicity. The individual nanohole arrays were
∼20 µm × 20 µm each, and the footprint of the entire set of arrays
was less than 1 mm × 100 µm. The sensor array was positioned
off center such that the fluidic connections could be maximally
separated from the microscope objective during operation.
Microfluidic Chip Fabrication and Integration. The mi-
crofluidic channel structures were produced using established soft
(11) Bethe, H. A. Phys. Rev. 1944, 66, 163-182.
(12) DiMaio, J. R.; Ballato, J. Opt. Express 2006, 14, 2380-2384.
(13) Smolyaninov, I. I.; Elliott, J.; Zayats, A. V.; Davis, C. C. Phys. Rev. Lett. 2005,
(14) Srituravanich, W.; Durant, S.; Lee, H.; Sun, C.; Zhang, X. J. Vac. Sci. Technol.
B 2005, 23, 2636-2639.
(15) Srituravanich, W.; Fang, N.; Sun, C.; Luo, Q.; Zhang, X. Nano Lett. 2004,
(16) Yin, L. L.; Vlasko-Vlasov, V. K.; Pearson, J.; Hiller, J. M.; Hua, J.; Welp, U.;
Brown, D. E.; Kimball, C. W. Nano Lett. 2005, 5, 1399-1402.
(17) Gordon, R.; Kumar, L. K. S.; Brolo, A. G. IEEE Trans. Nanotechnol. 2006,
(18) Brolo, A. G.; Arctander, E.; Gordon, R.; Leathem, B.; Kavanagh, K. L. Nano
Lett. 2004, 4, 2015-2018.
(19) Brolo, A. G.; Gordon, R.; Leathem, B.; Kavanagh, K. L. Langmuir 2004,
(20) Brolo, A. G.; Kwok, S. C.; Moffitt, M. G.; Gordon, R.; Riordon, J.; Kavanagh,
K. L. J. Am. Chem. Soc. 2005, 127, 14936-14941.
(21) Brolo, A. G.; Kwok, S. C.; Cooper, M. D.; Moffitt, M. G.; Wang, C. W.;
Gordon, R.; Riordon, J.; Kavanagh, K. L. J. Phys. Chem. B 2006, 110, 8307-
(22) Gordon, R.; Brolo, A. G.; McKinnon, A.; Rajora, A.; Leathem, B.; Kavanagh,
K. L. Phys. Rev. Lett. 2004, 92, 037401.
(23) Gordon, R.; Hughes, M.; Leathem, B.; Kavanagh, K. L.; Brolo, A. G. Nano
Lett. 2005, 5, 1243-1246.
(24) Gordon, R.; Brolo, A. G. Opt. Express 2005, 13, 1933-1938.
(25) Liu, Y.; Blair, S. Opt. Lett. 2003, 28, 507-509.
(26) Liu, Y.; Bishop, J.; Willians, L.; Blair, S.; Herron, J. Nanotechnology 2004,
(27) Willians, S. M.; Stafford, A. D.; Rodriguez, K. R.; Rogers, T. M.; Coe, J. V.
J. Phys. Chem. B 2003, 107, 11871-11879.
(28) Stark, P. R. H.; Halleck, A. E.; Larson, D. N. Methods 2005, 37, 37-47.
(29) Tetz, K. A.; Pang, L.; Fainman, Y. Opt. Lett. 2006, 31, 1528-1530.
(30) Dintinger, J.; Klein, S.; Ebbesen, T. W. Adv. Mater. 2006, 18, 1267-1270.
(31) Rindzevicius, T.; Alaverdyan, Y.; Dahlin, A.; Hook, F.; Sutherland, D. S.;
Kall, M. Nano Lett. 2005, 5, 2335-2339.
(32) Sun, Y. G.; Xia, Y. N. Analyst 2003, 128, 686-691.
(33) Koerkamp, K. J. K.; Enoch, S.; Segerink, F. B.; Hulst, N. F. v.; Kuipers, L.
Phys. Rev. Lett. 2004, 92, 183901.
Analytical Chemistry, Vol. 79, No. 11, June 1, 2007
lithography techniques.34,35Briefly, photomasks of each channel
structure were created and printed on transparencies using high-
resolution printing. The photoresist SU-8 50 was spin-coated on a
silicon wafer ramping to 500 rpm in 5 s, then dwelling for 8 s,
then ramping to 2000 rpm in 5 s, and spinning for 25 s, resulting
in a uniform thickness of 50 µm. The coated silicon wafers were
prebaked at 65 °C for 6 min and then at 95 °C for 20 min to harden
the photoresist. The photomask was positioned over the silicon
wafer, and it was exposed under collimated UV light for 45 s. This
was followed by a hard bake, first at 65 °C for 1 min and then at
95 °C for 5 min to promote polymerization of the photoresist. The
exposed silicon wafer was then immersed in SU-8 developer until
all the unexposed photoresist was removed, creating the master
templates. The microchannel structures were created by curing
4 mm of PDMS on the masters. This PDMS thickness enabled
direct tubing connection via friction fit and ensured channel
stiffness under suction. The PDMS was then degassed for 45 min
in vacuum and cured at 95 °C for 2 h. The PDMS channel
structure was cut from the master, holes for tubing connections
were punched, and the PDMS surface was exposed to oxygen
plasma for 20 s. The oxygen plasma treatment rendered the PDMS
surface hydrophilic and facilitated filling of the device. Irreversible
bonding via plasma treatment, as with PDMS-PDMS or PDMS-
glass interfaces,35is not possible with PDMS-gold. Instead, a
conformal bond, reversible seal was employed and the seal was
mechanically ensured via an acrylic top plate and clamping
mechanism. The top plate was machined to allow tubing connec-
tions to the PDMS layer and direct optical access to the area of
the chip containing the sensor array. This layered configuration
enabled chip disassembly, cleaning of the arrays,19and switching
of the microfluidic layers as required. In this work, the same
multiarray gold-on-glass substrate was employed with over 10
microfluidic structures. The size of the chip assembly was ∼7.5
× 25.4 × 25.4 mm3.
Chemicals. Solutions of known and controllable refractive
indices were required to establish the sensitivity of the nanohole
array sensors and to demonstrate local chemical detection within
the microfluidic framework. Aqueous sucrose solutions of refrac-
tive index varying from 1.332 (pure water) to 1.359 (concentrated
sucrose solution) were employed. The refractive index of each
solution was determined with a refractometer. For measurements
across on-chip-generated concentration gradients, a sucrose solu-
tion with a refractive index of 1.359 was used in conjunction with
pure water. The sucrose concentration profile across the sensor
array was varied with flow rate. Napthol blue black, mixed at 1%
by weight in distilled water, was employed to demonstrate the
generation of a cross-stream concentration gradient across the
sensor array. Chemicals employed in the microfabrication are
discussed in the microfabrication section above, and details of the
chemicals employed in the on-chip assembly of the cysteamine-
biotin-streptavidin system are provided with the results. Volume
flow rate of solutions to/from inlets/outlets was ensured via a
syringe pump. When possible, the chip was operated in vacuum
mode (applying a suction pressure to the outlet) rather than
positive pressure at an inlet. Operation in vacuum mode promotes
sealing of the channel structure, which is an advantage when a
reversible seal is employed.
Optical Measurements. Figure 2 shows a schematic of the
setup for the optical measurements. A broadband white light from
a halogen light source was provided to the nanohole arrays at
normal incidence through a 50× long working distance micro-
scope objective (Leica Microsystems, Wetzlar, Germany). This
objective ensured that only one array (20 µm × 20 µm) received
the incident light, and the 8.1-mm working distance provided
ample clearance for the assembled chip. Transmission spectra
through the nanohole arrays were obtained with a fiber-optic cable
connected to a spectrometer. The transmitted light intensity was
digitally recorded, in counts, versus wavelength in the range λ )
RESULTS AND DISCUSSION
Figure 1 shows the architecture of the chip at relevant
centimeter-, micrometer-, and nanometer-length scales. Two
configurations of the microfluidic layer, shown at the top of Figure
1, were developed to deliver solutions to the nanohole array
sensors. The most basic configuration is a single-inlet, single-outlet
channel with an expanded section to facilitate alignment with the
sensor array (top right in Figure 1). This arrangement ensures
all arrays experience similar conditions and was used to compare
differences between arrays based on periodicity and to acquire
redundant measurements on similar arrays sampling the same
solution. The second microfluidic layer configuration (shown top
(34) Duffy, D. C.; McDonald, J. C.; Schueller, O. J. A.; Whitesides, G. M. Anal.
Chem. 1998, 70, 4974-4984.
(35) McDonald, J. C.; Duffy, D. C.; Anderson, J. R.; Chiu, D. T.; Wu, H. K.;
Schueller, O. J. A.; Whitesides, G. M. Electrophoresis 2000, 21, 27-40.
Figure 1. Schematic and images illustrating the architecture of the
microfluidic chip with embedded nanohole arrays at relevant centi-
meter-, micrometer-, and nanometer-length scales. Shown are two
microfluidic configurations, the sensing element pattern with period-
icities indicated, and sample SEM images of the nanoholes (scale
bar size indicated.)
Analytical Chemistry, Vol. 79, No. 11, June 1, 2007
left in Figure 1) has two inlets, one outlet, and an expanded section
to deliver flow rate-controlled cross-stream concentration variations
during colaminar flow. The sensor array was composed of two
rows of nanohole arrays, offset from the center of the substrate
to facilitate tubing (world-to-chip interface) without interfering with
the collinear optical access. One row contained 650-nm periodicity
arrays, and the second row contained arrays differing in periodicity
from 350 to 850 nm. Periodicities in this range were selected, as
they result in wavelength-specific, SP-enhanced transmission of
light in the visible spectrum, to which the spectrometer is
sensitive. Portions of the 350- and 850-nm periodicity arrays and
a representative nanohole, imaged through scanning electron
microscopy, are shown in the lower portion of Figure 1.
As a preliminary test, transmission spectra obtained directly
from the nanohole array in air, without a PDMS channel, were
compared with those obtained from the assembled chip without
any solution. Wavelength- and periodicity-specific enhanced
transmission of incident light was evident in both cases, and the
measured wavelengths of peak transmission were similar to the
limit of detector resolution. All subsequent measurements were
obtained with the assembled chip under dynamic solution flow.
Five sucrose solutions with known refractive indices varying from
1.332 (pure water) to 1.359 were employed to quantify the
sensitivity of the device. A transmission spectrum from each array
was obtained on chip for the pure water case, and all sucrose
solutions. Following the sucrose solutions, pure water was
reintroduced into the chip and a transmission spectrum was
Figure 3a shows spectra collected for the pure water case, two
sugar solutions spanning the range, and the final pure water rinse
for the 450-nm periodicity array. The general form of the
transmission spectrum is a product of several factors. The
periodicity determines the wavelength(s), upon which surface
plasmon resonances are excited leading to enhanced transmis-
sion.9,10Several resonance conditions can be satisfied for a given
nanohole array periodicity, and multiple peaks are generally
observed from a broadband input. Other influencing factors
include the power spectrum of the light input and the sensitivity
of the detector as a function of wavelength. From the perspective
of chemical and biological sensing, the transmission spectrum
should exhibit SP-induced definable features (high gradients), and
the spectral shift in those features should be a predictable function
of near-surface refractive index attributable to concentration or
surface binding. As shown in Figure 3a, the spectrum is red-shifted
in response to solutions with increased refractive index. The red-
shift shown for the 450-nm periodicity array in Figure 3a was
Figure 2. Schematic of the transmission mode optical configuration
and the fluidic connections employed with the assembled device
Figure 3. Results of sensitivity test employing sucrose solutions of
known refractive indices. Transmission spectra were obtained through
the 450-nm periodicity array for the cases of pure water, sucrose
solutions with increasing refractive index, and again for water. (a)
Sample transmission spectra plotted together. (b) The relative red-
shift in peak wavelength for three SPR peaks (at wavelengths 606,
644, and 674 nm) exhibited by the array as a function of refractive
index. The two sets of data for the pure water case (n ) 1.332)
correspond to the first and last measurements. Estimated error for
all measurements is indicated for the 606-nm wavelength case.
Analytical Chemistry, Vol. 79, No. 11, June 1, 2007
representative of the response of the other arrays in agreement
with established theory of light transmission through nano-
holes.10,36In the case of the 850- and 750-nm periodicity arrays,
however, the intensity of the main resonance peak occurred at
wavelengths beyond the efficient range of the detector. Therefore,
the spectral shifts from the 750- and 850-nm periodicity arrays,
although similar to those observed in the 350-650-nm periodicity
arrays, could not be quantified with the same accuracy. The red-
shift in the three peak wavelengths shown in Figure 3a were
recorded as a function of solution refractive index with the results
plotted in Figure 3b. Error bars representative for all measure-
ments are indicated on the 606-nm wavelength peak case, and
the two sets of data for the pure water case correspond to the
first and last measurements. The plot indicates that increasing
the index of refraction over the total range, 0-2%, resulted in a
significant, linear increase in relative peak wavelength from 0 to
10 nm. Also, the first and last pure water measurements collapse
well near n ) 1.332 as expected. The demonstrated sensitivity is
333 nm/refractive index unit (RIU) in very good agreement with
previous measurements with similar arrays and experimental
Figure 4a shows a microscope image of a cross-stream
concentration gradient in the Y-channel configuration. The set of
six, 650-nm periodicity arrays were employed to spatially resolve
a cross-stream concentration gradient in a microfluidic flow.
Several applications of microfluidics to concentration gradient
generation have been motivated by the importance of concentra-
tion gradients to biological processes37and interest in monitoring
molecular interactions.38,39Microfluidic gradient generation was
facilitated here by colaminar microfluidic streaming from the
Y-channel intersection geometry shown in Figure 1. Due to the
laminar nature of the flow, mixing was limited to diffusion alone,
and the width of the mixing region of the two adjacent liquid
streams could be controlled by varying the flow rate. In the case
of ideal, pluglike, electroosmotic flow, the mixing process is one-
dimensional. Cross-stream velocity gradients in pressure-driven
flows, however, result in more rapid mixing near the microchannel
walls than at the midplane.40Nanohole array sensors sample
exclusively in the near-wall region where the mixing width can
be expected to scale as (DhCy/U)1/3, where D is the diffusion
coefficient, U is the average flow speed in the channel, hCis the
channel height, and y is the distance traveled along the channel.40
Characteristic near-wall mixing widths were estimated based on
the Y-channel configuration shown in Figure 1, with a diffusion
coefficient of 5 × 10-10m2/s, channel height of 50 µm, channel
width of 200 µm, and mixing distance of 7 mm (between the
intersection and the expanded region containing the nanohole
arrays). Flow rates of 0.001 and 0.050 mL/min resulted in mixing
width length scales corresponding to 24 and 7% of the channel
width, respectively. A colored dye solution was employed to
demonstrate the gradient generation across the expanded region
at 0.001 mL/min. The dye was introduced at one inlet while the
other inlet provided pure water. The generated concentration
gradient is effectively spread across the expanded region contain-
ing the arrays and reformed at the outlet. Due to the increased
channel width in this region, negligible cross-stream mixing is
expected despite the relatively long residence time. Similar
geometries have been employed to increase axial diffusion at the
expense of cross-stream mixing.41
Marker-free concentration detection was achieved here using
the six 650-nm periodicity arrays labeled in the image in Figure
4a. The dye solution was replaced with a sucrose solution, and
the resonance peak wavelengths were recorded at each cross-
stream 650-nm periodicity array under different flow rates. The
distribution of the sensor arrays shown in Figure 4a indicates that
array 6 would be expected to sample the pure water stream, and
given identical flow rates of each solution, the mean concentration
would be expected along the centerline between arrays 2 and 3.
(36) Krishnan, A.; Thio, T.; Kima, T. J.; Lezec, H. J.; Ebbesen, T. W.; Wolff, P.
A.; Pendry, J.; Martin-Moreno, L.; Garcia-Vidal, F. J. Opt. Commun. 2001,
(37) Dertinger, S. K. W.; Chiu, D. T.; Jeon, N. L.; Whitesides, G. M. Anal. Chem.
2001, 73, 1240-1246.
(38) Hatch, A.; Kamholz, A. E.; Hawkins, K. R.; Munson, M. S.; Schilling, E. A.;
Weigl, B. H.; Yager, P. Nat. Biotechnol. 2001, 19, 461-465.
(39) Kamholz, A. E.; Weigl, B. H.; Finlayson, B. A.; Yager, P. Anal. Chem. 1999,
(40) Ismagilov, R. F.; Stroock, A. D.; Kenis, P. J. A.; Whitesides, G.; Stone, H. A.
Appl. Phys. Lett. 2000, 76, 2376-2378.
(41) Coleman, J. T.; McKechnie, J.; Sinton, D. Lab Chip 2006, 6, 1033-1039.
Figure 4. Application of spatially separated (similar periodicity)
nanohole arrays to resolve a cross-stream concentration gradient in
a microfluidic flow. (a) A 5× magnification image of colored dye
(instead of sucrose solution) to demonstrate gradient generation
across the arrays at a flow rate of 0.001 mL/min. The array periodicity
is 650 nm. (b) A plot of resonance wavelength for each of the six
arrays for the cases of uniform water solution (diamonds and dotted
line), uniform sucrose solution (triangles and dot-dash line), and
gradient generation at 0.001 (blue squares and line) and 0.05 mL/
min (red line and crosses). Array numbers correspond to those labeled
Analytical Chemistry, Vol. 79, No. 11, June 1, 2007
Figure 4b compares the resonance wavelength recorded at each
cross-stream array when the channel contained pure water, pure
sucrose solution, a gradual gradient (0.001 mL/min), and a
sharper gradient (0.05 mL/min). The resulting resonance wave-
length profiles reflect expected cross-stream concentration be-
havior in each case. Specifically, the resonance wavelength was
consistent across the channel for the pure water case and showed
a uniform ∼10-nm red-shift in response to the sucrose solution.
At both focusing flow rates, array 6 reported very little deviation
in resonance wavelength from the pure water case. The expected
shapes of both the sharp (0.05 mL/min) and more gradual (0.001
mL/min) gradient are reflected in progressive red-shifts in arrays
5 through 1. The monotonic increase in red-shift in the case of
the gradual gradient is in qualitative agreement with prediction
based on flow rate although it is difficult to estimate the width of
the mixing zone other than that it is less than the channel width.
In the case of the sharper gradient, the near-wall mixing region
is found to be confined at the centerline between arrays 2 and 3,
indicating a mixing length on the order of 10% of the channel
width, in agreement with the scaling analysis above.
In analogy with traditional bioanalytical applications of SPR,5,42,43
it is expected that on-chip SPR-based sensors will find most utility
where the surface-specific nature of SPs can be exploited; in the
monitoring of adsorption processes. The surface-specific nature
of nanohole array-based sensing was exploited here to monitor
surface binding events of biological molecules in a flow-through
microfluidic format. The cysteamine-biotin-streptavidin model
system was chosen because it is readily available, the biotin-
streptavidin bond is generally quite strong, and the biochemistry
is well established.44A schematic of the final protein binding is
shown in Figure 5a, and it is a product of a three-part process.
First, a solution of cysteamine with a molar mass of 65 g/mol
was prepared (0.002 g of cysteamine in 5 mL of water). The arrays
of nanoholes were rinsed with acetone and methanol, placed in
an ultrasonic bath in methanol for 3 min, plasma cleaned for 15
min, and immersed in the cysteamine solution for 72 h to assemble
a monolayer on the gold surface. The gold chip was removed from
the cysteamine solution after the monolayer formation and cleaned
with ethanol and with distilled water. The microfluidic device was
then assembled, the PDMS channel structure and cover plate were
aligned and clamped over the gold-on-glass substrate, and the
channels were filled with pH 7 phosphate buffer solution (PBS).
The transmission spectrum from each 650-nm periodicity array
was obtained. In the second step, a solution of biotin linker (EZ-
Link NHS-LC-LC-Biotin) was prepared by dissolving 12 mg of
biotin into 2 mL of dimethyl sulfoxide (DMSO). This biotin
solution was introduced to the cysteamine monolayer via the
microfluidic chip for 45 min at a rate of 0.02 mL/min. The chip
was flushed with PBS buffer solution, and the transmission
spectrum from each 650-nm periodicity array was obtained.
The third and final step in the protein binding experiment
involved the streptavidin protein solution (∼1 mg of streptavidin
dissolved in 4 mL of PBS buffer). The nanohole arrays were
exposed to the streptavidin solution for 15 min at a flow rate of
0.02 mL/min. The transmission spectra were then obtained with
the streptavidin-PBS solution in the channel and following a flush
with PBS buffer. The SPR peaks from each step in the flow-
through assembly process were measured. The wavelength shifts
(∆λ) were calculated using the SPR peak of cysteamine as
reference and plotted in Figure 5b. The total surface group
assembly process corresponded to resonance peak red-shifts of
∼3.5-4 nm, indicating an increased surface-average refractive
index as expected from protein adsorption.45Results were analyzed
from each of the six arrays in three separate runs. The error bars
in Figure 5b correspond to the standard deviation calculated for
a particular run. The scatter in the biotin data was typical in the
assembly runs conducted. This may be due to the use of the
organic solvent (DMSO), which may remain as trace contamina-
tion during the rinsing or introduce some impurities into the
system by interacting with the PDMS. A decreased response from
arrays close to the lateral wall of the microchannel (for instance,
see arrays 5 and 6 in Figure 3) was observed in one of the runs,
and this was attributed to partial removal/damage of the cysteam-
ine monolayer during alignment. The shifts measured with
streptavidin in solution decreased by 1 nm after flushing with the
PBS solution. The change in SPR peak following the final rinse
indicates both the formation of adsorbed proteins and the surface-
specific nature of the SPR-based sensors. The magnitude of the
(42) Frutos, A. G.; Corn, R. M. Anal. Chem. 1998, 70, 449a-455a.
(43) Hanken, D. G.; Jordan, C. E.; Frey, B. L.; Corn, R. M. Electroanal. Chem.
1998, 20, 141-225.
(44) Neuert, G.; Kufer, S.; Benoit, M.; Gaub, H. E. Rev. Sci. Instrum. 2005, 76,
(45) Jung, L. S.; Nelson, K. E.; Campbell, C. T.; Stayton, P. S.; Yee, S. S.; Perez-
Luna, V.; Lopez, G. P. Sens. Actuators, B: Chem. 1999, 54, 137-144.
Figure 5. Application of on-chip nanohole arrays to detect surface
binding in the assembly process of a cysteamine monolayer-biotin
linker-streptavidin protein system. (a) Schematic of the streptavidin
protein binding strategy. The four biotin binding sites present in
streptavidin are indicated. (b) SPR shift measured relative to cys-
teamine after the addition of biotin, streptavidin, and PBS rinsing.
Analytical Chemistry, Vol. 79, No. 11, June 1, 2007
shifts reported here are in general agreement with earlier Download full-text
experimental work by our group,19but not as high as reported
The sensitivity of the nanohole integrated device in resonance
units (RU), where 1 RU is 1 pg of biomaterial/mm2of the sensor
surface, was estimated using the methodology described else-
where.6The slope of the refractive index calibration (Figure 3),
the wavelength shift due to the protein adsorption (Figure 5), an
estimated 200-nm decay length for the SP field,36,46and a specific
volume of 0.71 cm3/g for the protein47were used in the calculation.
A sensitivity of ∼6500 RU was obtained from our data. Typical
sensitivity values from commercial SPR devices are ∼70 000 RU.
It is important to emphasize that the sensitivity is not a funda-
mental limitation for the array of nanoholes, and it can match the
observed from typical commercial SPR systems. The sensitivity
reported here could be easily improved by using a laser source
combined to a better detection system.28The geometric param-
eters of the array can also be tailored to increase sensitivity. For
instance, the periodicity could be designed to allow the match
between the SPR resonance and the laser source,28and substrates
with optimized hole diameter29and shape48would lead to sharper
resonances and provide polarization-sensitive response. Moreover,
the amount of protein in the microchannel during incubation (17.5
µg) and the sample volume of our microfluidic sensing chamber
(0.05 µL) are comparable to commercial devices (Biacore X, for
instance). However, as can be seen in Figure 1, our device
supports 12 sensing elements (arrays) within this volume.
The benefits of nanohole arrays for on-chip detection, as
demonstrated in this paper, are balanced by some limitations. Due
to the inherent surface-specific nature of surface plasmons, the
detection volume is limited to the near-surface region. The
response to refractive index changes is also not selective. As in a
typical SPR instrument, the specificity of the sensor is given by
the immobilization of appropriated targets. Also, in contrast to
electrochemical-based detection methods, optical access and
associated optical infrastructure are required for nanohole-based
detection. Sensitivity of nanohole arrays has previously been
demonstrated to be similar to commercial SPR systems;28,29
however, the additional optical processing and infrastructure
required present some additional cost as compared to the
relatively simple setup employed here. Last, fabrication of the
nanohole arrays via focused ion beam milling, as employed here,
is relatively expensive. Effective nanohole array fabrication has,
however, also been demonstrated using photolithography.29,49
The results presented here effectively demonstrate nanohole
arrays as SPR-based sensing elements in a microfluidic platform.
Discrete nanohole arrays serve as effective on-chip sensing
elements and fit well with the existing microarray concept. An
alternative architecture would be a continuously perforated metal-
lic film as a substrate for whole-chip monitoring. Integrated with
small bandwidth sources, such as laser diodes, and CCD detection,
imaged intensity could be interpreted directly in terms of surface
properties, in real time. Reports on the utilization of arrays of
nanoholes fabricated by photolithographic methods are available
in the literature,29,49indicating that large-area, cost-effective nano-
hole substrates can in principle be mass produced for widespread
SUMMARY AND CONCLUSION
A microfluidic device with an integrated array of nanohole
arrays serving as SPR-based chemical and biological sensors was
developed and demonstrated. Measurements indicate that arrays
of nanoholes can be integrated and used as effective SPR detectors
in an on-chip format. The integrated device was successfully
applied to detect changes in refractive index with a sensitivity of
333 nm/RIU, which is comparable to previously reported values.
A set of six, 650-nm periodicity arrays were employed to spatially
resolve a cross-stream concentration gradient in a microfluidic
flow. The unique surface sensitivity of nanohole arrays was
exploited to detect surface binding in the assembly process of a
cysteamine monolayer-biotin linker-streptavidin protein system.
In gradient-based measurements, cross-stream concentrations
determined at different flow rates agreed well with a dye-based
experiment and a scaling analysis. In the protein binding event
monitoring experiments, average shifts in resonance wavelength
of 3.5-4.0 nm were observed. These shifts are in general
agreement with earlier experimental work. The results presented
here effectively demonstrate the application of nanohole arrays
as surface plasmon-based sensing elements in a microfluidic
platform. Nanohole array substrates as a platform for chip-based
analysis, add unique surface-sensitive diagnostic capabilities to the
existing suite of microfluidic-based analytical tools.
The authors are grateful for the financial support of the Natural
Sciences and Engineering Research Council (NSERC) of Canada,
through discovery research grants, and a postgraduate scholarship
to A.D. This work was also supported by equipment grants from
the Canada Foundation for Innovation (CFI). A fellowship from
the University of Victoria to A.D. is also gratefully acknowledged.
Received for review January 1, 2007. Accepted March 19,
(46) Chang, S. H.; Gray, S. K.; Schatz, G. C. Opt. Express 2005, 13, 3150-3165.
(47) Pahler, A.; Hendrickson, W. A.; Kolks, M. A.; Argarana, C. E.; Cantor, C. R.
J. Biol. Chem. 1987, 262, 13933-13937.
(48) Jung, Y. S.; Sun, Z.; Wuenschell, J.; Kim, H. K.; Kaur, P.; Wang, L.; Waldeck,
D. Appl. Phys. Lett. 2006, 88, 243105.
(49) Henzie, J.; Barton, J. E.; Stender, C. L.; Odom, T. W. Acc. Chem. Res. 2006,
Analytical Chemistry, Vol. 79, No. 11, June 1, 2007