Enhancing the sensitivity of surface-plasmon resonance sensors

Surface-plasmon-resonance (SPR) sensing devices have attracted tremendous interest in the past decade, both from a fundamental-physics perspective and as highly sensitive devices for optical detection of small biological or chemical entities in liquids.¹ The two main sensor types are based on extended and localized surface plasmons (SPs).² The former are considered more classical since they have been known longer. Extended SPs are longitudinal electromagnetic (EM) waves in a 2D electron gas on the surface of metals. Localized SPs, on the other hand, have become more mainstream only in the last two decades. They occur in metallic structures with dimensions less than half the wavelength of the exciting EM wave. The incident EM field must—in either case—be polarized in the plane of incidence. This is referred to as transverse-magnetic (TM) polarization.

A variety of nanophotonic structures³ can be used for SPR generation, including spherical nanoparticles and nanoshells, nanorods, nanopyramids, nanoholes in metallic films, subwavelength metallic grids, and nanorods as sculptured thin films,^4,5 among others. Figure 1(a) shows the most useful (‘Kretschmann’) configuration employing extended SPR. The EM field peaks at the metal-analyte interface and decays sharply within only a fraction of a wavelength, which is hence referred to as an ‘evanescent’ field. The EM-field distribution is the most important concept for SPR-sensing devices since the detection of molecules is done through their interaction with the evanescent field. The interaction may be described mathematically as the overlap integral between the evanescent field and the intrinsic molecular field. The effect of the molecules on the measured optical signal is proportional to the overlap volume. Hence, to increase the sensitivity of SPR sensors one needs to increase the overlap. This can be achieved by increasing the field's peak height and its extent from the surface (the evanescence region). The simplest and most direct observable is the reflectivity, expressed as either a function of wavelength or the incidence angle where a sharp dip is observed at the SPR (corresponding to an absorption or extinction peak). The dip location is a strong function of the medium's refractive index near the interface, hence its potential use as a sensitive refractive-index sensor. Typical sensitivity values are 50–100°/RIU (refractive-index units) or 2000–5000nm/RIU. Increasing the EM-field intensity near the metal interface enhances other optical signals specific to the molecules being sensed, such as their Raman scattering, fluorescence, second-harmonic generation, and other signals that depend nonlinearly on the evanescent field.⁶ The sensor specificity becomes higher when the measurand is an information-rich spectrum expressed as a function of either angle or wavelength.⁷

Figure 1. Surface-plasmon-resonance (SPR) excitation on the basis of (a) the standard Kretschmann configuration and (b) with addition of a top nanodielectric layer.

Figure 2. Electric-field intensity distribution showing an enhancement of a factor of 9 through addition of a 10nm-thick silicon (Si) layer to the silver film. Operating wavelength λ=653nm, refractive indices: n_prism=1.77641, n_Si=3.881+0.0116i, n_air=1, n_water=1.33, n_Ag=0.13+3.99i, and thickness of the silver layer d_Ag=43nm. NGWSPR: Near-guided-wave SPR.

Several methods can be used to enhance extended-SPR sensor sensitivity, including application of long-range SPR,⁸ bimetallic layers,⁹ and gratings^10,11 superimposed on the metal film. We recently developed a novel method based on two main concepts,¹² i.e., evanescent-field enhancement and use of smart algorithms to find the SPR-dip location with high precision. First, coating the metal surface with a nanometric layer (10–20nm thick) of material characterized by a high real part of the dielectric constant enhances the EM field at the interface with the analyte: see Figure 1(b). This is called near-guided-wave SPR (NGWSPR) because the top-layer thickness is less than that required to support the TM-guided mode. Second, we proposed the use of both inverse-scattering algorithms—by which one models the reflectivity curve to find the best fit to the experimental data—and dip-neighborhood approximation to a parabolic curve to find its minimum with subpixel accuracy.

Figure 2 shows an example of the field distribution, comparing the effect of a 10nm-thick silicon (Si) layer added to a silver film to that in the absence of the Si layer. An enhancement factor of four is observed. If a transparent dielectric layer is used, the enhancement can reach a factor of 10 if the real part of the dielectric tensor equals 17. Figure 3 shows experimental results and best-fitting simulation curves demonstrating sensitivity enhancement by a factor of about two. The increased sensitivity can be used to monitor chemical reactions of various types by attaching molecules to the surface and supplying probe molecules to the analyte fluid. Attachment of biological or chemical probes to the Si surface can be achieved by depositing a thin gold layer (with a thickness of a few nanometers) without affecting the sensitivity.

Figure 3. Measured (data points and dashed lines) and calculated (continuous lines) reflectivity for the structure of Figure 2. TM: Transverse magnetic.

We are now searching for sensitivity enhancements in other SPR configurations, such as grating-coupled SPR or nanophotonic structures covered with nanodielectric layers characterized by high real parts of the dielectric constant. Implementation of sensitivity-enhanced sensors for biological and chemical-pollutant sensing in water will follow soon.

I acknowledge support from the Israeli Ministry of Science under the Tashtiot program and discussions with Amit Lahav, Mark Auslender, and Atef Shalabaney.