Open Access
Add to BibSonomyAbstract
Specific detection of proteins is demonstrated using planar photonic crystal waveguides. Using immobilized biotin as probe, streptavidin was captured, causing the waveguide mode cut-off to red-shift. The device was shown to detect a 2.5 nm streptavidin film with a 0.86 nm cut-off red-shift. An improved photonic crystal waveguide sensor design is also described and shown to have a 40% improved bulk refractive index response.
©2008 Optical Society of America
1. Introduction
Effective identification of biological molecules remains critical for rapid disease detection, drug discovery, and many fundamental problems in molecular biology. In particular, the detection of an expanding number of biomarkers associated with pharmacological responses [1] and disease states such Alzheimer’s [2] and cancer [3] continues to generate considerable interest. Current technologies such as DNA microarray, mass spectrometry, and nuclear magnetic resonance spectroscopy are expensive and too cumbersome for many applications. The development of low cost, sensitive transducers for specific biomarker detection has therefore been of considerable interest in recent years.
Optical biological detection techniques typically employ radioactive or fluorescent labels. Use of such labels is labor intensive and thus considerable effort has been invested in the development of label-free approaches to protein analysis. For instance, refractive index (RI) sensing techniques detect an analyte by a local refractive index shift. Surface plasmon resonance devices and optical waveguides are common examples of such platforms. These approaches, however, require relatively large sensing areas, limiting their potential towards integration in lab-on-a-chip systems.
Photonic crystals (PC) are periodic structures that do not support electromagnetic propagation within certain frequency ranges. This photonic bandgap is a function of the refractive index modulation which defines the crystal. While most RI sensing approaches rely on the interaction between a weak evanescent wave and the analyte, photonic crystals allow strong light confinement to the sensing region, potentially even primarily within the analyte itself. Photonic crystal fibers (PCF) have been demonstrated for sensing antibodies [4] and DNA [5]. These fiber sensors are, however, difficult to implement in a compact and automated manner. White light reflectometry with bulk photonic crystal structures has also been reported for detection of proteins [6] and antibodies [7].
A common planar photonic crystal design consists of a hexagonal array of cylindrical air holes in a thin silicon film. Waveguides or cavity structures can then be created by introducing defects which support field localizations in the photonic bandgap. The design flexibility of planar photonic crystal structures enables greater light confinement, increased analyte interaction and decreased volume of interaction. These structures can be fabricated using existing microelectronic fabrication processes, allowing integration with microelectronics, microfluidics or other photonic devices.
Planar PC resonant cavities have already shown promise as thin film sensors [8,9] and single particle sensors [10]. Planar photonic crystal waveguides (PCW) have also been demonstrated for refractive index measurements [11], and analysis of liquid crystal orientation [12]. A preliminary PCW biosensor has recently been demonstrated with non-specific detection of bovine serum albumin [13]. While not possessing as high potential for ultimate sensitivity relative to the PC cavity biosensor, the fabrication requirements of the photonic crystal waveguide sensor are simpler than those of cavities, while still providing high field localization.
A simple photonic crystal waveguide (the ‘W1’ waveguide) can be created by removing a single row of air holes from the lattice. The dispersion characteristics of this waveguide are such that it is effectively single-moded at low frequencies, with a cut-off falling inside the band-gap (Fig. 1). This mode cut-off causes an abrupt drop of the output power in the transmission spectrum. Adding a thin surface film modifies the local refractive index, changing the effective refractive index of the slab and the refractive index contrast between the ‘hole’ and ‘slab’ regions. The device can thereby be used as a sensor by monitoring the cut-off wavelength shift resulting from the attachment of the target on the sensor surface.
Fig. 1. Band diagram for a two-dimensional hexagonal lattice silicon W1 photonic crystal waveguide projected in the propagation direction, simulated by the plane-wave expansion technique. Two waveguide modes are visible. As indicated by the lightly-shaded region, at lower frequency the waveguide is effectively single-moded before reaching a cut-off.
The deployment of a practical sensing device will require analysis of protein mixtures, where many potential targets are intermingled. To this end, we present the specific biological detection of a protein using planar silicon photonic crystal waveguides. The detection of streptavidin is employed for such proof of concept, with biotin as probing agent. We also introduce an improved waveguide sensing design showing an increased sensitivity compared to the standard PCW design.
2. Experimental details
Waveguides were designed such that the cut-off of the lowest order photonic bandgap guided mode fell near λo=1570 nm. The lattice of the photonic crystals was therefore set to Λ=406 nm with a hole radius of r=0.38Λ. The photonic crystal waveguide was patterned with a length of 24Λ. Coupling to and from the PCW is accomplished with traditional strip waveguides (Fig. 2). Each strip waveguide tapers from 2 µm at the chip edge to 450 nm at the photonic crystal waveguide interface, in order to minimize the mode mismatch between itself and the photonic crystal waveguide.
Fig. 2. A single sensing element composed of a 24Λ long W1 photonic crystal waveguide coupled to strip waveguides which have been tapered to 450 nm at the interface.
2.1 Fabrication
The silicon-on-insulator (SOI) wafer used here had a 340 nm silicon device layer over a 1 µm buffer oxide layer. Fabrication of the devices started by cleaning a ten millimeter square SOI chip in Piranha solution (3:1 H2SO4:H2O2) to remove any residual organics. Microchem PMMA 950K A2 was then spun onto the chip at 500 rpm for 5 s, at 4000 rpm for 30 s, and then baked on a hotplate at 185°C for 5 min to remove any residual solvent, creating an approximately 90 nm thick resist layer. The PMMA was then exposed using a Raith 150 electron beam lithography system with an accelerating voltage of 2 kV, an aperture of 20 µm, and an electron dose of 24 µC/cm2. Using low energy electrons reduced scattering and proximity effects, eliminating the need for proximity effect correction by other means.
The resist was then developed with a Microchem 1:3 MIBK:IPA solution for 30 s followed by a 15 s IPA rinse. Exposed areas of the SOI wafer were then anisotropically dry-etched in a Stanford Technical Systems ICP-RIE at a pressure of 20 mTorr, with a gas flow of 80 sccm C4F8 and 110 sccm SF6. The plasma was energized with 2.5 kW of ICP power and 20 W of forward RF power. Following the silicon etch, the remaining PMMA was stripped with acetone in an ultrasonic bath, the chip was cleaved to expose waveguide end facets, then re-cleaned with Piranha solution to remove any surface contaminants. In the interest of mechanical durability, the oxide under-layer was retained.
2.2 Measurement technique
The relevant spectral properties of the photonic crystal waveguides were measured by transmission spectroscopy. Figure 3 illustrates the custom coupling apparatus used to observe the transmission modes of the photonic crystal waveguide. A Santec TSL-210V tunable (1510–1630 nm) continuous-wave laser was employed as the light source. The source was fixed to be dominantly TE polarized with a Thorlabs looped-fiber polarization controller. Since the optical fiber was not polarization maintaining, the polarization was set with a polarizing beam cube at the tapered fiber output, which was then removed prior to waveguide coupling.
The silicon strip waveguides which lead to the photonic crystal waveguide sensing element were excited by butt coupling an input fiber with their cleaved edges. With a core diameter of 9 µm, a standard SMF-28 fiber was poorly matched to the silicon strip waveguide, which was only 0.34 µm thick and 2 µm wide. The beam was therefore focused by antireflection coated tapered and lensed fiber to increase the coupling efficiency (Fig. 4).
Fig. 4. Tapered & lensed fiber coupling to 2 µm strip waveguide at the chip edge (left). The strip waveguide continues to the PCW, narrowing to a terminal width of 450 nm at the interface (right).
The fibers were then aligned with a pair of piezoelectrically controlled 3-axis translation stages and a 20X long working distance objective. Coarse alignment was achieved visually with the objective and a camera. Fine tuning was performed by maximizing transmitted power. In the same manner as light was coupled into the strip waveguide, output light was collected with a tapered/lensed fiber and transmitted to a Newport InGaAs photodetector. The spectral response was measured by recording the transmitted power as the laser output wavelength was scanned, both automated with a custom LabVIEW controller.
2.3 Bulk refractive index sensing
Photonic crystal waveguide sensors of the type discussed here are most sensitive to changes in the refractive index near their surface. The response tapers off as the distance from the surface increases and the associated field decreases. As such, the sensor’s marginal response will decrease as the film thickness increases. The response to an effectively-infinite layer therefore sets an upper limit on the refractive index sensitivity. This limit of sensitivity was measured using a homogenous covering layer of de-ionized water (n=1.33).
Figure 5 shows the simulated transmission spectra (TE polarization) of a W1 silicon photonic crystal waveguide computed with the 2D FDTD effective-index method. The structure was simulated using a 10 nm step size, a 0.024 fs time step, and Mur’s absorbing boundary conditions. The response was normalized to the input broadband pulse. Spectra are shown for an air clad photonic crystal waveguide with air filled holes and for a water clad waveguide with water filled holes. The wavelength of the mode cut-off red-shifts by 31 nm as the refractive index increases from n=1 to n=1.33. The effect of the cladding layer is accounted for in a 2D simulation by modifying the high-index material’s effective index. Substitution of air for water in the photonic crystal holes was found to have a significantly greater effect than the substitution of the cladding layer.
2.4 Biofunctionalization
Specific biosensing relies on an immobilized biological receptor which binds selectively to the target analyte. In the case of refractive index biosensors, this binding event increases the local refractive index (Fig. 6). With biotin acting as a molecular probe, streptavidin was chosen as the target analyte to demonstrate the specific detection of protein. This molecular system is well documented, and possesses a high binding affinity and high specificity. Biotin was bound to the sensor surface with a linking molecule, 3-mercaptopropyltrimethoxysilane (MPTMS). If present, one of streptavidin’s four receptor sites will then bind to the immobilized biotin.
Fig. 5. Transmission response of a 2D silicon photonic crystal waveguide for air (solid) and water (dashed) infiltrated holes, simulated by FDTD. There is a significant red-shift associated with the infiltration of a higher refractive index material.
Biofunctionalization begins by immersing a freshly piranha-cleaned silicon waveguide in a 20:1 toluene to MPTMS solution at 100°C, under Ar atmosphere, then rinsing with toluene, ethanol, acetone and drying with nitrogen. Second, the silanized samples were immersed for 1 hour in a 1 mg/mL concentration of biotin in a 1:1 dimethyl sulfoxide (DMSO)/phosphate buffered saline (PBS) and then rinsed with DMSO/water, ethanol, acetone and dried with nitrogen. Lastly, a 10 mM solution of streptavidin in a phosphate buffer solution with 1% Triton X surfactant was diluted to 10 µM. Using this solution, the biotinylated samples were exposed to the streptavidin solution for 1 hour, then rinsed with a PBS/Triton X solution, ethanol, and acetone and dried with nitrogen.
Fig. 6. The surface of a silicon PCW sensor. A bioreceptor, biotin, is bound to the silicon surface with an MPTMS layer. If present streptavidin proteins will bind to the biotin immobilized on the sensor surface.
A negative control experiment was also conducted to confirm that the red-shift was due to specific binding of the streptavidin, rather than non-specific surface adsorption. Streptavidin has four binding sites to which the biotin can attach. A solution of streptavidin whose four pockets were pre-saturated with biotin was flowed over the sensor. Binding between the surface biotin layer and the saturated streptavidin was not expected under these conditions, since all the binding sites of the streptavidin were already occupied.
3. Results and discussion
A de-ionized water (n=1.33) layer was first dispensed onto the surface of the waveguide in order to test the response to a change in the cover layer refractive index. The resulting mode cut-off shift due to a 0.33 refractive index unit (RIU) change of the cover layer refractive was 29 nm. This corresponds to a bulk response of 88 nm/RIU, similar to the response of 31 nm predicted in Fig. 5, and to previously reported values [13].
The group velocity of the waveguide mode increases dramatically with wavelength, falling to zero at the mode cut-off. The increased loss associated with low group velocity in a lossy cover material makes it difficult to distinguish the mode cut-off wavelength. For this reason, an alternate reference wavelength (point A in Fig. 7) was used for the assessment. It is believed that this wavelength corresponds to the onset of coupling with TM-like photonic crystal slab modes [14,15]. Fabry-Perot oscillations due to the formation of a standing wave between the cleaved waveguide edge and the photonic crystal waveguide have been filtered from the results presented here.
Fig. 7. Observed mode cut-off under ambient air condition (dashed) and under a homogenous de-ionized water layer (solid). The mode cut-off red-shifts in the presence of a high refractive index cover layer. Point A is chosen as a reference.
Specific label-free protein detection was accomplished by measuring the spectral transmission profile near the mode cut-off, recorded following biotin immobilization and following streptavidin capture (Fig. 8). As expected, the mode cut-off red-shifts after the binding of streptavidin. A streptavidin protein layer was successfully observed as a red-shift of 0.86 nm. The streptavidin film thickness was measured to be 2.5 nm using ellipsometry with an assumed refractive index of 1.4 on a separate but identically processed silicon (100) chip.
A negative control experiment using biotinylated streptavidin confirmed that this shift was due to the specific binding of streptavidin, rather than non-specific binding of streptavidin or another species. Indeed, only a very slight 0.1 nm blue-shift was observed, likely a result of a thin residue being removed in the process.
Fig. 8. Photonic crystal waveguide mode cut-off following biotin attachment (dashed) and following streptavidin capture (solid). The mode cut-off red-shifts by 0.86 nm.
6. Improved waveguide design
Photonic crystal waveguide sensors are most sensitive near the silicon surfaces of the waveguide, where the electromagnetic field is the most intense. In the photonic crystal plane, this means that the region near the line defect is more sensitive than outlying regions. It also means that the central sidewalls are more sensitive than the upper silicon film surface. The response of the sensor is therefore dominated by sidewall deposition [10].
To improve the sensitivity, we have also investigated a modified donor-defect waveguide (Fig. 9) where rather than eliminate a row of holes forming a W1 photonic crystal waveguide, a line of holes are shrunk to a fraction of their original size. The response of this waveguide is similar to that of the W1 waveguide, except that the modes are shifted to higher frequencies inside the bandgap. This structure significantly increases the amount of surface area available for sensing in the central ‘high-field’ regions, thus significantly increasing its sensitivity.
Fig. 9. Modified photonic crystal waveguide sensor for increased sensitivity. In this case, a line defect is created by shrinking a row of holes rather than removing them.
This structure was fabricated with a lattice period of 446 nm and hole radii 0.37Λ for the exterior holes and 0.22Λ for the central holes. As before, the modified photonic crystal waveguide architecture was immersed in de-ionized water to determine the bulk response to an effectively infinite homogenous film. A 41 nm shift was observed corresponding to a sensitivity of 120 nm/RIU, a 40% improvement relative to the W1 sensor (Fig. 10). Work is currently underway to demonstrate this enhanced sensitivity towards the detection of proteins.
Fig. 10. Response of a modified photonic crystal waveguide as in seen in Fig. 9. Observed mode cut-off under ambient air condition (dashed) and under a homogenous de-ionized water layer (solid). As before, the mode cut-off red-shifts in the presence of a high refractive index cover layer.
6. Conclusions and future work
We have presented a photonic crystal waveguide biosensor capable of specific biological detection. Using a surface immobilized biotin film, streptavidin protein was captured on the sensor surface causing the waveguide mode cut-off to observably red-shift. Specific binding of a 2.5 nm streptavidin film induced a 0.86 nm cut-off red-shift. The bulk sensitivity of this sensor was 88 nm/RIU. We have also presented an improved sensing architecture featuring increased surface area in the high field region. This device was demonstrated to have a 40% increased bulk response.
The waveguides reported here were significantly longer than required to extinguish unguided modes. Incomplete polarization isolation was thus the primary factor limiting the cut-off depth. It should therefore be possible to further reduce the size of the photonic crystal without affecting the resolvability of the mode cut-off.
It should also be possible to improve the sensor by exploiting the ‘slow-light’ dispersion. Mach-Zehnder interferometry (MZI) is a relatively common sensing approach to making highly sensitive phase-based measurements [16]. By placing the photonic crystal waveguide in the sensing arm, the output signal becomes dependent on the cumulative phase difference between the two paths. Near the cut-off, the group velocity of the photonic crystal waveguide travelling mode decreases dramatically, falling to zero at the mode cut-off. This wave compression can be exploited to create compact MZIs, where the required interaction length to produce a fixed phase offset is reduced by a factor of c/vg [17]. Integrated silicon photonic crystal modulators have already been reduced in length by two orders of magnitude versus a strip waveguide using this approach [18]. One would expect that a similar gain would be possible for photonic crystal waveguide sensors, enhancing their ability to detect films with very small index contrast.
Acknowledgments
This work has been supported by the U.S. Air Force Office of Scientific Research through its SBIR/STTR program. Steven Buswell also acknowledges additional support from the Natural Sciences and Engineering Research Council of Canada (NSERC). Device fabrication was performed at the University of Alberta Nanofab. Electron microscopy facilities were provided by the National Research Council of Canada National Institute of Nanotechnology (NRC-NINT). The authors wish to acknowledge Ashok Prabhu for his assistance in performing photonic crystal plane-wave simulations.
References and links
1. J. S. Ross, D. P. Schenkein, O. Kashala, G. P. Linette, J. Stec, W. F. Symmans, L. Pusztai, and G. Hortobagyi, “Pharmacogenomics,” Adv. in Anat. Pathol.11, 211–220 (2004).
2. J. H. Growdon, “Biomarkers of alzheimer disease,” Arch. Neurol. 56, 281–283 (1999). [CrossRef] [PubMed]
3. P. R. Srinivas, M. Verma, Y. M. Zhao, and S. Srivastava, “Proteomics for cancer biomarker discovery,” Clin. Chem. 48, 1160–1169 (2002). [PubMed]
4. J. B. Jensen, P. E. Hoiby, G. Emiliyanov, O. Bang, L. Pedersen, and A. Bjarklev, “Selective detection of antibodies in microstructured polymer optical fibers,” Opt. Express 13, 5883–5889 (2005). [CrossRef] [PubMed]
5. L. Rindorf, J. B. Jensen, M. Dufva, L. Pedersen, P. Hoiby, and O. Bang, “Photonic crystal fiber long-period gratings for biochemical sensing,” Opt. Express 14, 8224–8231 (2006). [CrossRef] [PubMed]
6. L. Stefano, L. Rotiroti, I. Rendina, L. Moretti, V. Scognamiglio, M. Rossi, and S. D’Auria, “Porous silicon-based optical microsensor for the detection of l-glutamine,” Biosens. Bioelectron. 21, 1664–1667 (2006). [CrossRef]
7. O. Meskini, A. Abdelghani, A. Tlili, R. Mgaieth, N. Jaffrezic-Renault, and C. Martelet, “Porous silicon as functionalized material for immunosensor application,” Talanta 71, 1430–1433 (2007). [CrossRef] [PubMed]
8. M. Lee and P. M. Fauchet, “Two-dimensional silicon photonic crystal based biosensing platform for protein detection,” Opt. Express 15, 4530–4535 (2007). [CrossRef] [PubMed]
9. L. W. Mirkarimi, S. Zlatanovic, S. Sigalas, M. A. Bynum, K. Robotti, E. Chow, and A. Grot, “Toward single molecule detection with photonic crystal microcavity biosensors,” Digest of the LEOS Summer Topical Meetings (IEEE2006) pp. 29–30. [CrossRef]
10. M. Lee and P. Fauchet, “Two-dimensional Si photonic crystal microcavity for single particle detection,” Proceedings of the 4th IEEE International Conference Group IV Photonics (IEEE2007) pp. 1–3. [PubMed]
11. J. Topolancik, P. Bhattacharya, J. Sabarinathan, and P. C. Yu, “Fluid detection with photonic crystal-based multichannel waveguides,” Appl. Phys. Lett. 82, 1143–1145 (2003). [CrossRef]
12. R. Ferrini, J. Martz, L. Zuppiroli, B. Wild, V. Zabelin, L. Dunbar, R. Houdre, M. Mulot, and S. Anand, “Planar photonic crystals infiltrated with liquid crystals: optical characterization of molecule orientation,” Opt. Lett. 31, 1238–1240 (2006). [CrossRef] [PubMed]
13. N. Skivesen, A. Ttu, M. Kristensen, J. Kjems, L. Frandsen, and P. Borel, “Photonic crystal waveguide biosensor,” Opt. Express 15, 3169–3176 (2007). [CrossRef] [PubMed]
14. Y. Tanaka, T. Asano, Y. Akahane, B. Song, and S. Noda, “Theoretical investigation of a two-dimensional photonic crystal slab with truncated cone air holes,” Appl. Phys. Lett. 82, 1661–1663 (2003). [CrossRef]
15. Y. Tanaka, T. Asano, R. Hatsuta, and S. Noda, “Analysis of a line-defect waveguide on a silicon-on-insulator two-dimensional photonic crystal slab,” J. Lightwave Technol. 22, 2787–2792 (2004). [CrossRef]
16. R. G. Heideman and P. V. Lambeck, “Remote opto-chemical sensing with extreme sensitivity: design, fabrication and performance of a pigtailed integrated optical phase-modulated Mach-Zehnder interferometer system,” Sens. Actuators. B 61, 100–127 (1999).
17. M. Soljacic, S. G. Johnson, S. H. Fan, M. Ibanescu, E. Ippen, and J. D. Joannopoulos, “Photonic crystal enhancement of nonlinear phase sensitivity,” J. Opt. Soc. Am. 19, 2052–2059 (2002). [CrossRef]
18. L. L. Gu, W. Jiang, X. N. Chen, L. Wang, and R. T. Chen, “High speed silicon photonic crystal waveguide modulator for low voltage operation,” Appl. Phys. Lett. 90, 071105 (2007). [CrossRef]
