HTML
-
Biosensing is one of the most sought-after applications in the field of nanophotonics because the enhanced light–matter interactions via sub-wavelength confinement and amplification of optical near-fields give access to unprecedented real-time, label-free and highly sensitive optical signal transductions1-10. For example, nanoengineered plasmonic devices have revolutionized the performance of traditional Raman11, 12, fluorescent13 and infrared14 spectroscopy techniques enabling extraction of essential data on complex molecular structures, as well as single-molecule detection15. In particular, refractometric plasmonic sensors, such as those based on surface plasmon resonance (SPR)16 and localized-SPR (LSPR)17, 18 effects, have opened up new horizons in label-free optical biosensing, overcoming limitations of cumbersome and time-consuming labeled approaches, such as enzyme-linked immunosorbent assay and fluorescence detection. Moreover, the small-volume, near-field resonant plasmonic modes act as perfect probes to monitor minute local refractive index changes caused by molecular-binding events on the sensor surface. Most recently, facilitated by convenient far-field spectral intensity-measuring techniques, versatile on-chip plasmonic biosensors enabling clinically relevant detection limits of various biomaterials, such as bacteria19, viruses20, exosomes21 and proteins22, have been reported. Despite the recent progress, several challenges still exist for nanoplasmonic sensor technologies, including the need for simplified and robust optical readout systems and mass-scalable nanostructured chips with low manufacturing costs.
From a more applied perspective, on-site biosensing without dedicated personnel and infrastructure is essential for the biomedical community23-25. In particular, simple, robust and affordable bio-detection technologies that can rapidly and quantitatively detect multiple analytes within small sample volumes would not only benefit medical diagnostics but also other fields, such as food safety and environmental surveillance26. Several attempts have been made to engineer point-of-care (POC) biosensors that leverage nanoplasmonics27-30. Most of these devices rely on amplitude (intensity) interrogation in which a narrow-band illumination source, such as a light-emitting diode (LED), is tuned to the plasmonic resonance, and the intensity variations due to the spectral resonance shifts that are induced by local refractive index changes are quantified on an imaging sensor, for example, a cellphone camera31, among other techniques32-34. Although this mode of detection can achieve spectral monitoring without the need for expensive and bulky spectrophotometers, its detection capabilities are limited by the plasmonic resonance mode properties, such as sensitivity to refractive index and quality factor (Q-factor, inversely proportional to the bandwidth of the mode) as well as the level of environmental and instrumental noise. Moreover, they do not usually allow multiplexed and high-throughput detection as the operational sensing area is limited either by the optical beam, when microscopes are used for read-out, or the size of the nanoplasmonic sensor, which is often small when fabricated using time-consuming and sophisticated techniques.
An alternative optical signal transduction mechanism for sensing is phase interrogation, which measures phase retardations of electromagnetic wavefronts. Using phase rather than intensity in optical detection has improved microscopy of thin and transparent objects35, 36 and elucidated the crystal structures of atomically thin layers in the field of X-ray diffraction37, among many other applications38. Phase-based detection was introduced into SPR technology by Kabashin et al39 to exploit phase shifts varying sharply at plasmonic resonances, first reported by Abeles et al40 Conventional light sources suffer from high-intensity (amplitude) noise, which in turn reduces the detection sensitivity of intensity-based plasmonic sensors. Instead, phase detection allows for lower noise and versatile signal-processing possibilities, such as spatial and temporal mapping due to the inherent relative measurement of phase with respect to a reference beam41. Moreover, in the traditional intensity-based setting, sensing is done at wavelengths that correspond to the side slopes of resonant modes rather than at center wavelengths where the intensity varies minimally as the resonance shifts. Despite that, the side slopes of the intensity spectrum change moderately when compared to the abrupt change that the phase spectrum exhibits at the center of the modes. Therefore, the phase response allows for superior refractive index sensing42.
Within the context of phase interrogation, nanostructured plasmonic surfaces are still mostly unexplored and their phase response investigation is limited15, 43, 44. Recently, Kravets et al15 used an ellipsometric spectrometer to probe the sharp phase transitions in plasmonic metamaterials for biosensing. These conventional ellipsometer-like optical readers utilize orthogonally polarized components of the same beam as reference and signal in an oblique reflectance setup. Alternatively, Junesh et al43 proposed to use an interferometric substrate with an Al mirror layer in order to access the phase information associated with the resonance modes of short-range ordered nanohole arrays in transmission and extracted the phase data from oscillating spectral interference patterns. The common drawback of these techniques is that they cannot provide spatial information, which is essential for high-throughput multiplexed biosensing. Importantly, despite their advantages, these traditional spectrometer-based optical setups are complex and bulky, thus limiting the use of phase interrogation, especially for POC applications.
Differential interference contrast microscope, which uses sheared beams, is a phase-reading technique that allows two-dimensional phase mapping on a limited field of view (FOV), but its precision relies on critical alignment of lens and Wollaston–Nomarski angle-shearing prism45. Similarly, interferometric scattering microscopy is a label-free platform with high phase sensitivity that suffers from limited FOV due to the use of high numerical aperture objectives46, 47. Recently, a novel phase-interrogating large FOV interferometric microscope was introduced, offering ultrasensitive axial topographic resolution in a lens-free and compact form48. Phase detection was achieved by first shearing the incoming light into signal and reference optical beams, and then interferometrically combining them in a collinear common-path geometry. The corresponding beams are quasi-overlapped and thus subjected to similar environmental instabilities, leading to reduced noise. Measurements on transparent substrates showed the potential of the device, but the demonstrated sensitivity was not sufficient for biosensing applications.
In this paper, we present a new plasmonic phase-sensitive detection platform for measuring protein biomarker microarrays with high throughput. It consists of a specifically designed metal nanostructured chip that enhances phase effects of near-field light–matter interactions, which are then detected at the far field using a large FOV lens-free differential interference contrast microarray (LIM) imager. The plasmonic interaction increases the phase sensitivity of the previously demonstrated48 LIM setup by more than one order of magnitude, which is crucial for efficient biomarker detection.
Figure 1a depicts the large-area plasmonic microarray plate, sandwiched between two polarizer (P), P1 and P2, and Savart plate (SP), SP1 and SP2, pairs. The output from a collimated LED source is split by SP1 into two orthogonally polarized beams sheared with respect to each other (shear distance is 25 μm) that traverse the microarray plate. The two sheared beams are recombined by SP2 and interfered by P2, which is orthogonal to P1. This optical design allows for the projection of minute topographical changes, enhanced by the plasmonic interaction, onto a CMOS image sensor (~30 mm2). Consequently, any phase difference between the two beams are mapped over the entire sensor chip. In Figure 1a, we show a CMOS image of the experimentally measured optical path difference (OPD) map from a plasmonic chip patterned with arrays of 10 nm thin silica (SiO2) layer. In order to emphasize the flexibility of our method in recognizing microspots with different diameters and interspot distances, we designed this pattern where the diameter of the spots decrease from center to edge (350–50 μm), as well as the spacings in between them (min 25 μm). The corresponding OPD contrast from the microarrayed silica spots is clearly observed (Supplementary Fig. 4S). The OPD maps are computed using multiple interferograms collected at different SP1 tilt positions controlled by a stepper motor and applying a phase-shifting interferometry method.
Fig. 1
Large FOV interferometric microarray imager (LIM) and experimental setup. (a) Collinear optical light-path configuration of the LIM setup. Collimated LED light beam is polarized (P1) and then sheared by a SP (SP1). This generates quasi-spatially overlapped and orthogonally polarized light beams that traverse the plasmonic microarray plate and are subsequently recombined using a second SP (SP2) and interfered by a second polarizer (P2). The interferogram is finally imaged by the CMOS sensor. The image shown on the schematic CMOS sensor is a real measured interferogram of a 10 nm-thin silica (SiO2) pattern on a plasmonic chip. (b) Photograph of a wafer comprising 1 cm × 1 cm plasmonic Au-NHA chips fabricated using high-throughput, wafer-scale nanofabrication tools. (c, d) Artificially colored scanning electron microscopy images of 10 nm thin silica microarrays on uniformly patterned plasmonic Au-NHAs. (e) Photograph of a plasmonic chip with 200 pl volume protein droplet microarrays formed using low-volume liquid dispensing tool. (f) Disposable capillarity-based microfluidic platform assembled on the plasmonic microarray plate. (g) Portable interferometric microarray imager operated through an interface running on an ordinary personal computer.For the first time, we exploit the sharp phase transitions at the resonances of plasmonic gold nanohole arrays (Au-NHAs)49, instead of the most commonly used amplitude-sensitive approach and demonstrate high-throughput biosensing using the LIM imager. The microarray plate, consisting of uniformly nanostructured Au-NHAs over the entire sensor surface, is fabricated on robust glass wafers using high-throughput and low-cost deep-ultraviolet lithography. Such scalable and robust manufacturing techniques are necessary for disposable and affordable biosensors. Figure 1b shows a 4″ wafer comprising 50 Au-NHA chips (1 cm × 1 cm). To characterize the proposed plasmonic phase-sensitive detection platform, we first probed it using dielectric thin films of amorphous silica (Figure 1c and 1d), and showed that atomically thin (angstrom-level) topographical changes in a microarray setting with potentially millions of protein biomarker spots can be detected. Subsequently, to validate the platform as a biosensor, we created microarrays of protein solution droplets on the plasmonic chips via low-volume liquid dispensers (Figure 1e). The bioassays were carried out using a capillarity-based disposable microfluidic platform, which enables on-chip, pumpless fluidic manipulation (Figure 1f). We experimentally extracted the OPD sensitivity of our phase-sensitive reader as 9000 nm per refractive index unit (RIU) and the minimum detectable refractive index change as 5.7 × 10−4 RIU, which is an order of magnitude higher than the intensity-based imaging system values reported by Cetin et al28. Moreover, the LIM imager is highly integrated (portable) and built with low-cost off-the-shelf consumer electronic and optic products, such as CMOS image sensors and LEDs (Figure 1g), and it can also be operated by a standard personal computer through a simple user interface. The readout time for a single measurement is 30 s. All of these features are essential for practical and rapid POC biosensing.
-
The plasmonic Au-NHAs retain the unique extraordinary transmission (EOT) phenomenon induced by the interplay of two resonance coupling mechanisms in a classical asymmetric Fano-type spectral line profile22, 50. Perpendicularly incident light generates in-plane surface plasmon polariton sub-radiant modes satisfying the Bragg's coupling condition through wavevector matching of the grating's momentum. The sharp dark mode strongly couples to the subwavelength holes in the Au film and generates a bright radiant mode that scatters light into the free space. These destructive and constructive near-field interactions are strongly dispersive and manifest themselves in the far-field intensity spectrum as multiple dips and peaks that constitute different modes associated with either the supporting substrate or the top medium. It is imperative for plasmonic biosensing that the sensing mode be spectrally isolated from other modes as the refractive index of the monitored medium changes51, 52. A well-designed mode allows sensing with different background media, such as air (nair=1) or water (nwater=1.33), and ensures high sensitivity over a wide dynamic range. In this work, we use the primary (−1, 0) EOT sensing mode of the Au-NHA sensor (hole diameter=200 nm and period=600 nm) and suppress the other substrate modes by increasing the thickness of the Ti adhesion layer (10 nm) between the glass substrate and the nanostructured Au film. Figure 2b shows the numerically computed intensity dispersion curve of the primary (−1, 0) EOT sensing mode, which is spectrally well isolated. The slope of this continuous EOT mode allows us to calculate the conventional bulk refractive index sensitivity in the visible-to-near-infrared range as 615 nm per RIU ($S_{\mathrm{bulk}}=\frac{\mathrm{d} \lambda_{\mathrm{EOT}}}{\mathrm{d} n}$, where λEOT is the transmission peak wavelength and n is the refractive index of the top medium), which is in line with the experimental findings reported in the literature22, 51. In comparison to the intensity, the phase response of the plasmonic resonances has not been widely explored thus far. The incident light that couples to the plasmonic surface experiences temporal retardation at the resonance modes, resulting in sharp phase transitions in the far field53, which can be exploited for sensing the refractive index variations on sensor surface. Figure 2c shows that phase dispersion curve exhibits the same bulk sensitivity calculated on the intensity dispersion curve of Figure 2b.
Fig. 2
Phase interrogation principle of LIM on plasmonic substrates. (a) Orthogonally polarized and symmetrically sheared, that is, partially overlapped, beams (red and blue columns) are both intensity and phase modulated upon traversing the Au-NHA sensor, due to the plasmonic mode coupling. The plasmonic phase and intensity modulation show spatial difference on the microarray spots (ON) with respect to the bare plasmonic surface (OFF). When the light beams are recombined (that is, the shear is removed), they create fringe patterns indicated by blue and red regions on the CMOS sensor due to phase differences induced by the distinct ON and OFF plasmonic phase modulation. (b, c) Numerically computed transmission intensity and phase dispersion of the EOT mode plotted as a function of the refractive index (RI) of the top media. The redshift of the mode associated with the RI increase can be observed both in the transmission intensity and phase plots (see contrast in the color maps). The bulk sensitivity (Sbulk) of the Au-NHAs (period 600 nm and diameter=200 nm) is calculated as 615 nm per RIU. The LED source, with peak wavelength λpeak=656 nm, is also indicated on the plot. (d) Representative intensity and phase modulation spectra associated with the ON and OFF regions with a RI difference (Δn) of 0.01. The phase derivative (φder) corresponding to the EOT peak, which is a significant parameter in phase interrogation, is marked on the plot. (e) Phase contrast (Δφ) and the corresponding OPD between the ON and OFF regions as a function of Δn are calculated at the EOT resonance wavelength (λEOT) of the bare sensor. The refractometric LIM sensitivity can be numerically calculated from the slope of the curve.$\left(\mathrm{OPD}=\frac{|\Delta \phi|}{2 \pi / \lambda}\right)$ In order to access the phase information from the plasmonic Au-NHA sensor, we use the LIM imager introduced in Terborg et al48. This imager offers precision and stability over a large FOV by interfering sheared common-path collinear beams. The orthogonally polarized and sheared beams, shown in Figure 2a as red and blue columns, traverse the plasmonic sensor where they are both intensity and phase modulated. As the Au-NHA plasmonic resonances are polarization-independent28, the modulation functions vary only spatially depending on whether the material to be detected is present or absent at a given region on the sensor surface (In Figure 2a, the pink region represents a microarray element to be detected). Specifically, the beam that traverse the microarray element is referred to the signal and is modulated by the ON functions; whereas, the beam that traverse the bare sensor is referred to the reference and is modulated by the OFF functions, as shown in Figure 2d. Note that the presence of material on the sensor surface induces a redshift in the spectral positions of both the intensity and phase modulation functions. When the beams reach the second SP and polarizer, they are recombined and interfered, and the topography of the microarray pattern on the sensor surface is projected as interference contrast fringes on the CMOS sensor. In Figure 2a, the green pixels of the CMOS sensor correspond to regions where the sheared beams go through identical optical paths, and therefore yield zero intensity after interference. Whereas the blue and red pixels correspond to the regions where the beams are modulated by different phase transmission functions (φON and φOFF), resulting in a net intensity after interference. In contrast to transparent substrates (that is, no plasmonic structure) where the phase difference depends only on the thickness of the material and its refractive index contrast with the medium, in our system the phase difference is amplified by the plasmonic phase function, which results in a sensitivity increase.
Unlike the previous intensity-based plasmonic biosensors, which rely only on the spectral position of the EOT intensity peak, our methodology uses the OPD between the signal and reference beams (OPD=Δφ/k, where k=λEOT/2π is the wavevector and Δφ=|φON−φOFF|). In our interferometric reader, this phase contrast value (Δφ) is extracted from multiple images recorded at different phase bias using the computational phase-shifting interferometry technique48. Figure 2e shows the numerically computed phase contrast Δφ (left axis) and OPD (right axis) as functions of the refractive index difference (Δn=nmaterial−nmedium). The sensitivity of our system can also be analytically stated as follows:
$$ S_{\mathrm{Pl}-\mathrm{OPD}}=\mathrm{d}(\mathrm{OPD}) / \mathrm{d} n=S_{\mathrm{bulk}} \times \phi_{\mathrm{der}} \times \lambda_{\mathrm{EOT}} / 2 \pi, $$ where φder=dφ/dλ=10 [deg nm] at the EOT peak (indicated in Figure 2d). The OPD sensitivity can, therefore, be numerically estimated as 1.1 × 104 nm per RIU over a dynamic range of 0.025 RIU, which is sufficient for biosensing applications. In our system, this dynamic range is mainly defined by the spectral shape of the plasmonic mode, which correlates to the derivative of the phase function (φder), as well as the spectrum of the illumination source. Details on the plasmonic phase sensitivity (SPl-OPD) derivation are presented in the Supplementary Information.