Optical excitation of a surface plasmon (SP) on a thin metallic surface is widely applied for sensitive biosensing.1 The standard approach employs attenuated total reflection (ATR) in a glass prism to excite an SPW on a thin gold film coated on the prism. ATR biosensors are known to be very sensitive to surface environmental changes. The change of surface refractive index unit (RIU) can be measured on the order of for a precise angular measurement ( degree) or for a 0.02-nm wavelength shift in the optical spectrum.2 However, the ATR setup is typically bulky and expensive and requires a large amount of sample solution. Due to its optical configuration, it is hard to be applied for high-throughput and chip-based detections, such as DNA and protein microarrays.3, 4
The studies on modern nano-plasmonics show that SPs can also be efficiently excited in metallic nanostructures.5, 6, 7 Moreover, the resonance of SPs in a periodic nanostructure causes an extraordinary transmission in certain wavelengths. The resonant condition is decided by the nanometer aperture, period of the structure, and surface condition. Since the resonant wavelength is changed under different surface conditions, the metallic nanostructures are expected to be good sensing elements for high-throughput detections. Recently, Brolo proved this concept by using an array of nanoholes in a 200-nm-thick gold film.8 A sensitivity of was achieved by measuring the resonant wavelength shift. Light is usually cut off in the nanoholes. The enhanced transmission is attributed to the surface plasmon resonance (SPR) on the top and bottom of the metallic film. The resonance occurs when the incident wavelength and period of the nanostructure satisfy the phase-matching condition. However, in a glass biochip, the refractive indices of the substrate and outside air or water environment have different phase-matching conditions. This results in a low transmission of light and a lower SPR sensitivity for such nanohole structures.
Recently, we found that a transverse-magnetic (TM) polarized wave can directly transmit through a nanoslit without cutoff.9 Different from nanoholes, the extraordinary transmission of a TM wave was explained by the SP resonance in the nanoslit. The SP can also propagate along the exit metallic surface and cause interference pattern when nanoslits are close enough.10, 11 If the period of the nanoslit array satisfies the phase-matching condition, constructive interference will result in an additional SPR peak in the transmission spectrum. Such SPR is only related to the outside surface condition and is decoupled from the substrate SPR mode. Thus, it is expected to have a higher sensitivity than nanohole structures. Besides, light is not cut off in the nanoslit. The transmission intensity is much stronger than that of the nanoholes. As a result, the nanoslit array is a better candidate than the nanohole array in the chip-based biosensor. In this paper, we verify this result by fabricating an SPR microarray consisting of periodic nanoslits. For a nanoslit array with a 50-nm-gap, 150-nm thickness, and 600-nm period on a glass substrate, the measured sensitivity is , about 1.7 times larger than that for the nanohole array. The measured transmission in the nanoslit array is about one order of magnitude larger than that in the nanohole array.
Fabrication of Nanoslit Array
Metallic nanoslits were fabricated by using electron beam lithography and a reactive ion etching method. We used a soda-lime glass as the substrate. Because gold has poor adhesion to the glass surface, a 5-nm-thick Ti film and a 150-nm-thick gold film were sequentially deposited on the glass sample by using an electron gun evaporator. After the deposition, a 350-nm-thick PMMA resist (MicroChem Corp., Newton, MA) was spin-coated. A field emission scanning electron microscope (LEO-1530) modified with a nano-pattern generating system (NPGS) was used to write nanoslits on the PMMA resist. The patterns were then transferred into film by using argon sputtering in a reactive ion etching machine (Oxford Instrument, plasmalab 80plus). The power of the radio-frequency (RF) electromagnetic wave in the reaction chamber was . The chamber pressure was , and the flow rate of Ar gas was . The PMMA resist was removed by rinsing the sample in acetone for a few hours. To make sure that no PMMA resist residue remained on the gold nanoslits, we further cleaned the sample by using ozone sputtering in the RIE chamber. The sample was then put in ultrapure water, placed in an ultrasound bath for , and purged dry by nitrogen. Figure 1 shows slit arrays on the glass substrate. The scanning electron microscope (SEM) images show that each array consists of 600-nm-period and 50-nm-gap nanoslits. The area of the nanoslit array is . The separation between adjacent arrays is . This dimension is comparable with the spot size and separation in a DNA microarray. In addition to the fabrication of slit arrays, an array of nanoholes with a 600-nm period and 100-nm-hole diameter was also made to compare their optical transmissions.
The transmission spectrum for a single array was tested by using a white light source ( , halogen lamp). Figure 2 shows the optical measurement setup. The light was spatially filtered by using a lens, an iris diaphragm, and a collimation lens. The incident polarization was controlled by a linear polarizer. The polarized light was then focused on a single array by using a objective lens. Note that to avoid the influence from other arrays, the beam size focused on the array needs to be smaller than . This was accomplished by adjusting the aperture size of the diaphragm. The transmission light was then collected by another objective lens and focused on a fiber cable. The transmission spectrum was taken by using a fiber-coupled linear CCD array spectrometer (BWTEK, BTC112E).
Figure 3a shows the transmission optical images of slit arrays in air and in water. It is seen that both colors are quite different due to different SPR conditions on the metallic surface. Figure 3b shows the measured transmission spectra of a nanoslit array in air and water under transverse-electric (TE) or TM polarization. For a TE polarized wave, the transmission is very weak because the optical wave is under a cutoff condition. The intensity drops quickly when the incident wavelength increases. For TM polarized wave in air, there are two transmission resonant peaks, at 630-nm and 730-nm wavelength, respectively. These two modes are attributed to two different resonant conditions in the nanoslit array: the SPR on the metallic surface and the resonant cavity mode in nanoslits.12 The SPR mode is relative to the period of slit and surface condition. It occurs when incident wavelength meets the phase-matching condition;5is the resonant wavelength, is the period , and and are the dielectric constants for the surface medium and the metal, respectively. In air, the SPR peak is at ( , for gold at the wavelength). Hence, the measured 630-nm peak wavelength came from the resonant SP on the metallic surface, and the 730-nm wavelength was the cavity mode in nanoslits. In water [ and for gold at the 800-nm wavelength], the estimated SPR peak was at . The measured 807-nm peak wavelength meets quite well with the prediction of the SPR mode. The figure also shows the measured results for a 600-nm-period nanohole array. It can be seen that an SPR peak also appears near the 600-nm wavelength in air. In water, there is an SPR peak near . But the transmission intensity is about one order of magnitude smaller than that in the nanoslit array.
In most SPR biosensors, an aqueous environment is preferred. Hence, the sensitivity at the 807-nm-wavelength SPR mode was tested. Figure 4a shows the transmission spectra in water and different methanol/water mixtures for TM polarized waves. There was an apparent redshift when the outside medium changed from water to a methanol-rich medium. Figure 4b shows the peak wavelengths against the refractive index of the outside medium. The slope of the fitting curve shows that the RIU sensitivity is . This value is times larger than the array of nanoholes, which is .8 The red shift and sensitivity can be predicted well by Eq. 1. In most cases, the absolute value of is much larger than . Hence, the change to the SPR wavelength can be approximated byis the change of the refractive index. The predicted sensitivity is for the 600-nm-period nanoslit array. This value is close to the measured result and is comparable with grating coupled-based SPR sensors.2 It is noticeable that Eq. 2 indicates that the SPR sensitivity is proportional to the period. An increase in the period can increase the sensitivity. However, if the period is larger than , the SPR wavelength in water will be in the infrared region. A sensitive and high-resolution IR spectrometer is necessary for the measurement. In our experiments, the period was , and the SPR wavelength in water was . The commercial CCD spectrometer is sensitive to this wavelength.
It is worthwhile to compare the nanoslit-based sensor with the grating coupled-based sensor. For a grating coupled device, the phase-matching condition is written as13is the incident angle and is the period of the grating. Hence, a precise control of the incident angle is necessary to excite the surface plasmon. In the nanoslit device, the SPR wavelength is insensitive to the incident angle. In addition, the spectrum in a grating coupled device is measured by the reflection light from the top surface. Scattering of the aqueous medium would cause substantial noise in the measurement. On the contrary, the spectrum in nanoslit device is taken in the transmission mode. Transmission light is dominant by the evanescent surface plasmonic waves. The scattering of the aqueous medium has a small effect on this device. The major drawback of the nanoslit-based sensor is the fabrication cost. Most nanostructures are made by using electron beam lithography or the focused ion beam method. For the grating coupling device, the grating is easily fabricated by using the two laser beam interference method.
The aforementioned sensitivity was obtained by fulfilling the fluidic channel with different refractive index media. For verifying the surface sensitivity, an antigen–antibody interaction was tested. The antigen was serum amyloid P component (SAP), a 25-kDa pentameric protein. It was first identified as the pentagonal constituent of in vivo pathological deposits14 and is thought to be an important contributor to the pathogenesis of a related group of diseases called the amyloidoses.15 We applied the nanoslit array to detect the anti-SAP and SAP interactions on the gold surface. In the experiment, pure water and buffer solution, , were first injected to the fluidic channel to clean the gold surface. The transmission spectrum of the nanoslit array in the buffer solution was taken. Then of anti-SAP in the buffer solution was injected into the channel. The sample was placed overnight in order to make anti-SAP bind well to the gold surface. A new buffer without anti-SAP was then injected two times to remove the unbound antibodies. The spectrum of anti-SAP on gold film in the buffer solution was taken. Next, SAP was injected into the channel. After several hours of displacement for the antigen–antibody interactions, the unbound SAP was washed away by twice filling the channel with new buffer. The transmission spectrum was then taken. Figure 5 shows the measured spectra for different surface conditions. The anti-SAP on a gold surface made an red shift of the SPR peak. The shift was increased to when SAP was bound to anti-SAP. The 25-kDa-sized SAP resulted in a 1.4-nm wavelength shift. It implies that if the resolution of a spectrometer is , the gold nanoslit array can detect -sized biomolecules attached to the surface.
In summary, we studied the surface sensitivity of a gold nanoslit array in an aqueous environment. The RIU test shows that the sensitivity is 1.7 times larger than that of an array of nanoholes. The anti-SAP and SAP interaction test verified good surface sensitivity for bio-affinity studies. Our measurements were taken in a small area of , and the acquisition time was shorter than . The high-throughput detection is thus feasible when thousands of arrays of nanoslits were fabricated on a glass chip.
This research is supported by the National Science Council, Taiwan (Grant no. NSC 95-2215-E-001-001) and the Thematic Project of Academia Sinica, Taiwan.