1. Introduction

Surface enhanced Raman scattering (SERS) is one of the most active field in spectroscopy due to its great potentials for a variety of applications such as single molecules detection and bio-sensing. Localized surface plasmon resonance (LSPR) excitation of metal nanoparticles is in the heart of the signal enhancement of this technique. A variety of geometries of colloidal metal nanoparticles such as disks [1], rings [2], crescent moons [3,4] and triangles [5] have been produced by chemical process usually, enabling their LSPRs to be tuned by altering their morphology [6], size, composition [7] and interparticle distances [8]. However, the very nature of present colloids means that the Raman signals are poor in reproducibility due to the lack of control in the size, shape and distribution of the particles. Other factors hindering wide applications of SERS are the limited substrate materials (Ag, Au and Cu, etc.) and limited excitation ranges due to the inherent nature of the LSPRs of nanoparticles for a specific metal, for example, intense SERS on Au and Cu nanoparticles were only observed with red light excitations.

The essence natures of metamaterial, highly reproducible and stable artificial structures with periodic lattice of identical elements exhibiting extraordinary electromagnetic properties from its structure other than directly from its composition, may provide new strategies to break the inherent limitations of conventional SERS substrates and bring it to more practical/industrial applications. Metamaterial have recently sparked considerable interests [9–11] and various designs of specific unit cells have been proposed to realize desired magnetic and electric plasmon resonances for achieving a multitude of counterintuitive physical effects and very appealing applications [12–18]. Most recently, Clark et al. [19] developed nanophotonic split-ring resonators as dichroics while Ding et al. [13] designed polarization-dependent metal-molecule-metal sandwich metamaterial for molecular spectroscopy. In this paper, we report a novel metamaterial design of a #-shaped gold wires array and explore its extraordinary features such as SPRs, local field hotspots and enhancements for SERS applications. Sensitivity of the SPRs to the dielectric environment and figure of merits (FOM) are also investigated. It is shown that the novel metamaterial exhibits excellent properties as SERS substrates such as tunable double pronounced SPRs from visible to near-infrared range, symmetrically distributed local field hotspots, high field enhancement factors, superior sensitivity and FOMs over most of the conventional ones.

2. Structure design and numerical tools

Figures 1(a) and 1(b) show a typical periodically arranged structure of the #-shaped metamaterial and a sketch of a unit cell with all relevant geometrical parameters. The parameters are chosen to match closely those of technologically feasible physical systems i.e. the period P=260 nm, gold thickness h=30 nm, arm/leg’s lengths a=b=140 nm, line widths c=d=40 nm, and separations e=f=10 nm. Within a laser focus (generally ~2 μm in diameter in the modern Raman systems), there should be at least 36 unit cells being excited.

 

Fig. 1 (a) Schematic of a #-shaped gold wires metamaterial; (b) a unit cell with geometric scales P=260 nm, a=b=140 nm, c=d=40 nm, e=f=10 nm, and the gold thickness h=30 nm

Download Full Size | PDF

The spectral response of the structure was calculated by three-dimensional finite element method with Microwave Studio (CST Inc). Its reliability and accuracy were proved by Ref [20]. To comply with the Raman experiments for SERS, we let the polarized electromagnetic wave propagate along the z axis, with the electric field in the y direction, and the magnetic field in the x direction. The metal is taken as gold following the Drude dispersion model to fit its realistic characteristic at visible frequencies [21,22]. The structure is surrounded by an effective homogeneous medium with a refractive index of n=1.05 for a good agreement between experimental and theoretical results because using an index of vacuum in simulating would slightly blue shift all resonance frequencies [23].

3. Results and discussion

3.1 Dual-band resonances

Figure 2(a) shows the extinction coefficients as a function of wavelength for the regular structure with P=260 nm, a=b=140 nm, c=d=40 nm, e=f=10 nm, and the gold thickness h=30 nm as designated in Fig. 1(b). A single metal cross shows primarily a single, well-pronounced resonance [24]. By using the #-shaped structure, we obtained two pronounced SPRs around respectively 753 nm and 605 nm along with a third order resonance at about 529 nm. Due to the symmetry of the structure which responses to x and y polarized light equivalently, both resonances can thus be excited simultaneously with either x or y polarized light.

 

Fig. 2 (a) Extinction spectrum in #-shaped gold wires arrays for either electric field polarizations. The first resonance (N=1, M1) is at 753 nm and the second resonance (N=2, E1) is at 605 nm. (b) Distributions of induced currents at the excitation of 753 nm (N=1, M1) and (c) at the excitation of 605 nm (N=2, E1).

Download Full Size | PDF

In order to get insight into the origins of the dual-band resonances and the correlation between SPRs and the structure geometry, we calculated the surface current distributions in the #-shaped gold wires array by excitation with photon energies corresponding to each resonance wavelength which are shown in Figs. 2(b) and 2(c), respectively. Figure 2(b) shows clearly that the first resonance originates from the concentrated surface currents flowing (or charge oscillation) around the left and right gaps between the two arms (see the two curved arrows) which are induced by excitation of the y-polarized photons with energy corresponding to 753 nm. Since the currents flow oppositely in the two arms which build up intense magnetic fields constructively in the gaps, the first resonance is therefore attributed to a magnetic surface plasmon resonance (M1 mode) in metal split-ring resonators [25, 26]. The arms serve as an inductance L and the gap between the two arms provides a capacitance C. This mode is thus characterized by a LC-resonance. However, the second resonance (E1) results from the electrical surface plasmon resonance with currents flow around the outer corners [from arms to legs or vice visa, see the four curved arrows in Fig. 2(c)]. Both pairs of dipoles on the left and right sides form a bonding state at each corresponding gap region. The in-phase oscillating of the two bonded dipole pairs contributes to second pronounced resonance in the symmetrically structured metamaterial. It is found that the third order resonance (E2) arises from a multipole interaction in the two arms or the two legs (not shown here).

3.2 Tunability of SPRs

Tuning the plasmon resonance is of great interest for various applications. Figure 3 shows tunability of the magnetic (M1) and electric (E1) mode wavelengths with altering the arm/leg’s length and gap’s width. Figure 3(a) reveals that the M1 mode can be much more effectively tuned than the E1 mode by changing the arm’s length. This can be attributed to the effective increase in inductance L and capacitance C of the split ring resonators by increasing the arm’s length since the M1 mode can be described by an LC-resonance ω_LC=(LC)^-1/2. Due to a simultaneous reduction in inductance L and capacitance C with increasing the gap’s width f, the M1 mode can also be tuned by altering the gaps width f [triangles in Fig. 3(b)]. However, this mode is not expected to be effectively tuned by altering the gap’s width e because of a cancel effect in capacitance C and inductance L. Since the E1 mode originates from the charge oscillating in the two bonded dipole pairs, the opening up of the gaps between the arms or legs has nearly equivalent effect in shortening the charge oscillating paths. It can thus be effectively tuned by altering the widths of either gaps [Fig. 3(b)]. By changing the arm’s length from 140 to 240 nm the M1 center wavelength can be tuned from ~750 to ~1160 nm while the E1 can be tuned from ~605 to ~630 nm. By changing the gap’s width from 10 to 30 nm, the M1 can be further tuned from 750 to 690 nm and E1 from 605 nm to less than 550 nm, respectively.

 

Fig. 3 Dependence of the SPRs on (a) the arms length a; and (b) the gap’s widths e (■,●) and f (▲,▼)

Download Full Size | PDF

3.3 Hotspots of localized field and Raman enhancement factor

Figure 4 shows hotspots distribution and the magnitude of the electric field (|E|) at the surface of the #-shaped gold wires in the x-y plane for both SPRs. It is found that the maximum electromagnetic fields (hotspots) are localized in the gaps for the M1 SPR (N=1, λ_M1=753 nm) as shown in Figs. 4(a) and 4(b) and in the gaps as well as the corners for the EPR (N=2, λ_E1=605 nm) as shown in Figs. 4(c) and 4(d). The intense field in the gaps for the MPR originates from the inductive magnetic dipoles moments, while the intense field in the gaps for EPR confirms the bonding state of both dipole pairs on the left and right sides of the structure as discussed above. The electromagnetic field enhancement determined from the amplitude ratio of the calculated local field E to the incident field E₀ reaches 35.9 for the M1 SPR and 39.9 for E1 SPR, respectively. Since the Raman enhancement factor is proportional approximately to the fourth power of the field amplitude enhancement [2], we achieved Raman enhancement factors of 1.7×10⁶ and 2.5×10⁶ at the magnetic SPR and electric SPR, respectively. Even with the relatively weak third order resonance, a Raman enhancement factor of 10⁵ can be obtained. These values are sufficiently high for SERS and sensing applications, considering that the Raman enhancement factors resulting from pure electromagnetic enhancement in silver/gold colloids are generally around 10³ ~ 10⁴ [27].

 

Fig. 4 The near field distribution and the corresponding field enhancement (|E|/|E₀|): (a) and (b) at the excitation of 753 nm (N=1, M1); (c) and (d) at the excitation of 605 nm (N=2, E1).

Download Full Size | PDF

In addition to the high enhancement factors, an important and interesting aspect of the metamaterial is its wavelength tunability, particularly, the higher order resonances to the shorter wavelength. Conventional SERS substrates with gold exhibit very poor or even no enhancement with green or blue light excitations due to its LSPR being at larger wavelength and aggregation of the nanoparticles pushes the LSPR to even longer wavelength [28]. With metamaterial electromagnetic response concept, considerable Raman enhancement factors can be achieved with gold at green light excitation via higher order resonances. This broadens greatly usable wavelength range of gold substrates and applications of SERS.

3.4 Sensing figure of merit

As a standard measure for assessing a nanoparticle’s sensing potential, the sensing figure of merit (FOM) is defined as the ratio of SPR sensitivity (eV/RIU) to full width at half maximum (FWHM) (eV) [28]. In order to assess the property of the #-shaped metamaterial as chemical sensors, we investigated the sensitivity of the SPRs to the refractive index of environment. Figures 5(a) and 5(b) show that both SPRs are sensitive to the dielectric environment and redshift linearly with the refractive index of the surrounding medium. The magnetic surface plasmon resonance is much more sensitive to the dielectric environment than the electric surface plasmon resonance. Similar effect was also observed in a metal-molecule-metal sandwich metamaterial [29]. Table 1 lists the peak position, refractive index sensitivity, FWHM and FOM of the metamaterial for three gap sizes. The FOMs of #-shaped gold wires metamaterial exceed most of those reported [4] for nanocrescents (~3), nanoshells (~ 1.7), triangular prisms (~ 1.7) and nanorice (~ 1).

 

Fig. 5 SPR sensitivity to the local dielectric environment: (a) the SPR extinction spectra for P=260 nm, a=b=140 nm, c=d=40 nm, e=f=10 nm upon a change in dielectric environment; (b) the SPR sensitivity to dielectric environment is equal to the value of the slope.

Download Full Size | PDF

Table 1. Peak Positions and Environmental Sensitivity Parameters for Different Gaps Structures

View Table

Finally, we would like to point out that the central hole in the unit cell has little influence on the extinction spectra and field enhancement although a #-shaped structure is designed here.

4. Conclusion

We have designed a #-shaped gold wires array and investigated its SPRs, frequency tunability, hotspots distribution and field enhancement. It is shown that the designed structure exhibits pronounced and polarization-independent dual-band SPRs which can be tuned to cover most of the Raman excitation lines from near-infrared to green regions. The field hotspots locate symmetrically at the opposite gaps for the magnetic suface plasmon resonance and in the gaps and corners for the electric surface plasmon resonance. Raman enhancement factors of ~10⁶ are achieved for both SPRs with superior FOMs and high dielectric sensitivity. It is also possible to break the usable wavelength range of conventional gold SERS substrates by higher order (second and third orders) excitations of the #-shaped metamaterials. These excellent properties together with the highly reproducibility and easy-fabricating of the simple planner structure make it attractive as substrates for SERS and sensing applications.