1. Introduction

Multiple fields bear witness to a broad range of plasmonic nanostructure application prospects, those of which include dynamic and static structural colors [1–2], highly integrated biosensing detection [3–4], high-efficiency solar cells [5–7], and other high-performance optoelectronic devices [8–10]. Their excellent optical performance is attributed to their strong field confinement and enhancement on incident light field at nanoscale through the plasmon resonance excitation [11]. Particularly, the localized near field of plasmonic nanostructures is extremely sensitive to small changes of surface surroundings, which makes them very promising for chemical and biological sensing detection [12–13]. Currently, a wide variety of biosensors have been developed depending on plasmon resonance in nanostructures, for example, ultrasensitive detection or even disease biomarkers detection at the single-molecule level for early-stage diagnostics, and the miniaturized and integrated designs for convenient point-of-care detection, as well as efficient, time-saving multiplexed and high throughput assays [14–18]. Despite numerous researches have been endeavored, plasmonic nanostructure sensors still face dramatic challenges in high sensitivity, miniaturization, multi-parameter, and high throughput detection.

Recently, the coupling and hybridization among different plasmonic modes in nanostructures have drawn prodigious attention in the design of high-performance biosensors. It is an effective strategy to generate new plasmon resonances and adjust their linewidths and amplitudes [19–20]. In order to achieve plasmon resonances with sharp linewidths and high absorptions, it is critical to control the interaction among different plasmonic modes in precision by adjusting the structural dimensions appropriately [21–22]. Therefore, plasmonic nanostructure hybrid systems open up a new route to realize high-performance optical biosensing. However, previous extensive researches mainly concentrated on an individual hybrid mode, there are few comparative studies on the sensing performance of multiple coupled modes in an individual plasmonic nanostructure, which has potential prospects in multi-parameter detection and various application scenarios [19,23–30].

In this paper, we focus on an asymmetric step-shaped three-layer metal-insulator-metal (MIM) plasmonic metamaterial, which features three perfect absorption peaks by coupling different plasmonic modes in the near-infrared region. Due to distinct resonant coupling mechanisms, these absorption peaks show different near-field enhancements and spatial localizations, resulting in a range of sensing capabilities. The absorption peak based on SLR mode at the shortwave range, originating from the structural symmetry broken of the proposed step-shaped groove array, exhibits an ultra-narrow linewidth of 2 nm, corresponding to a high optical quality factor, and a high bulk sensitivity of 1605 nm/RIU. In comparison with the above SLR mode, these other two absorption peaks possess superior surface sensitivities, which has great advantages for small-size biomolecular detection in practical applications. It is worth noting that the middle absorption peak disappears in aqueous solutions. The complementary sensing performance among these resonant absorption peaks will promote the development of versatile sensing platforms for various biomolecules.

2. Model construction

Figure 1 illustrates schematics of the proposed asymmetric step-shaped plasmonic metamaterial and its characteristic spectra. As shown in Fig. 1(a), our plasmonic metamaterial consists of a 50 nm thick silica spacer (H) inserted in between an asymmetric step-shaped slit-groove array and an opaque gold membrane of 150 nm with all incident light reflected. The cross-sectional view shows three structural unit cells with period P. Each unit cell of the uppermost slit-groove array is composed of a nano-slit (width d) in immediate contact with a groove (width w₂ and depth t₂) in a gold film (thickness t₁). The coexistence of periodic nano- slits and nano-grooves in the gold film leads to the emergence of periodic nano stripes (width w₁) in the proposed structure. In this paper, optical properties are characterized by utilizing two-dimension finite difference-time domain (FDTD) method, which has advantages in calculating complex metallic nanostructures [31]. For all simulations, the incident light illuminates the proposed structure from the negative z-direction with wavelength from 1400 nm to 3000 nm, and only transverse magnetic (TM) polarization (magnetic field parallel to y direction) is considered. The periodic boundary condition is adopted in x-direction and the perfectly matched layers condition in z-direction. The permittivity of silica spacer is 2.13 and that of gold is described by Drude-Lorentz model [32]. All calculations with extremely high accuracy and well convergence are performed with a finer mesh of 0.5 × 0.5 nm². Furthermore, the proposed plasmonic metamaterial can be experimentally realized using the combination techniques of vacuum coating and focus ion beam.

 

Fig. 1. Schematics and geometrical parameters of the asymmetric step-shaped plasmonic metamaterial and its representative optical spectral characteristics. (a) Three-dimensional schematics and cross section of three unit cells of the proposed plasmonic metamaterial with geometric dimensions specified. (b) Reflection (red solid line) and absorption (black solid line) spectra for the proposed plasmonic metamaterial with P = 1600 nm, slit width d = 100 nm, the thickness of middle dielectric spacer H = 50 nm, groove width w₂ = 750 nm, groove depth t₂ = 30 nm, the width of top gold stripe w₁ = 750 nm, and the thickness of gold film t₁ = 390 nm. As a direct comparison, reflection spectrum (blue dashed line) for symmetric case is also shown in top panel. Inset shows a magnified view of SLR dip A with a narrow linewidth of 2 nm.

Download Full Size | PDF

3. Results and discussion

As illustrated by the black solid line in Fig. 1(b), there are three remarkable perfect absorption peaks (marked as peaks A, B, and C, respectively) exhibited in the absorption spectrum of the asymmetric structure in air environment, which is equivalent to three resonant reflection dips due to the opaque gold film at the bottom of the structure (red solid line in the middle panel). To facilitate the explanation of the physical origins, reflection spectra and corresponding reflection dips are considered in the following. It is clearly observed that three near-zero reflection dips show different linewidths and wavelength positions. At the short wavelength range, an Fano-type asymmetric narrowband near-zero reflection dip appears at the wavelength of 1607 nm with a linewidth of 2 nm. The magnified inset of dip A further confirms this point. Moreover, dip A has a high dependence on the structural period, which is shown in the following text. The above-mentioned two important features of dip A are consistent with SLR mode in the previously reported literatures [23,33–35]. In addition, the other two perfect absorption peaks in the long-wavelength range possess relatively broad linewidths, 48 nm for peak B and 172 nm for peak C, which is more one quarter of 100 times than that of peak A. As a direct comparison, the reflection spectrum of the symmetric case without nano-groove is also depicted by the blue dashed line in the top panel of Fig. 1(b). Evidently, the narrowband refection dip A at the short wavelength range vanishes, leaving only a broad adjacent reflection peak due to Rayleigh anomaly. And the linewidths and wavelength positions of the other two reflection dips remain nearly unchanged for the symmetric case. The comparison results of symmetric and asymmetric structures corroborate that reflection dip A originates from the existence of the top groove in the top step-shaped slit-groove array.

In order to distinctly elucidate the generation mechanism of the SLR mode (dip A) induced by structural asymmetry, the spatial electric field and corresponding surface charge distributions for two different wavelengths are plotted in Fig. 2, respectively. The white solid lines indicate the outlines of the structure. Here, the SLR mode with an asymmetric narrowband spectral feature, characterized as a special type of Fano resonance with the structural period considered, is attributed to the coupling of two hybridized plasmon modes (the super-radiative bright mode and the sub-radiative dark mode) due to the structural asymmetry of the top surface [34–36]. Figure 2(a) shows the electric field distribution of reflection dip A at the wavelength of 1607 nm, where the strong and enhanced electric field is mainly localized around the top surface of the structure. This near-field distribution corresponds to the sub-radiative dark mode, which can be further confirmed by the surface charge distribution at the top surface of the structure in Fig. 2(b). Obviously, the dipolar modes at the top surfaces of the gold stripe and groove have exactly opposite directions. The out-of-phase dipolar interaction induced by structural asymmetry reduces the dipole momentum and minimizes the radiative damping of the proposed structure, which is the representative feature of the sub-radiative dark mode. Furthermore, the spatial electric field distribution of the flat and broad reflection peak at the wavelength of 1630 nm adjacent to dip A is also plotted in Fig. 2(c). In comparison to the case of dip A, the electric field magnitude at the wavelength of 1630 nm is significantly reduced and its localization is also weakened. This near-field distribution corresponds to the radiative bright mode of the Fano resonance mentioned above. The surface charge distribution in Fig. 2(d) at the top surface of the proposed structure exhibits only one big dipolar mode formed by the whole top surface, which possesses high and extremely broad reflection in the studied wavelength range. Notably, the SLR mode in our work is mainly determined by the sub-radiative dark mode. Due to the ultra-narrow spectral feature of dip, the proposed asymmetric step-shaped nanostructure is very suitable for biosensing applications.

 

Fig. 2. Spatial distributions of magnitude of electric field and corresponding surface charge distributions and influence of structural asymmetry on resonant reflection dip A. (a) Electric field and (b) surface charge distribution at reflection dip A with wavelength of 1607 nm corresponding to sub-radiative dark mode of SLR mode. (c) Electric field and (d) surface charge distribution at reflection peak close to reflection dip A with wavelength of 1630 nm corresponding to radiative bright mode of SLR mode. Due to charge distribution in top surface is much larger than other surfaces, only the charge distribution of the small region around top surface is displayed in (b) and (d) to highlight the charge distribution in top surface. (e) Reflection spectra for different lateral shift d_x of 0 nm, 75 nm, 150 nm, and 375 nm. Inset shows the charge distribution of d_x = 0 nm at the wavelength of 1607 nm. And the linewidth of dip A for different lateral shift d_x is also given in (e).

Download Full Size | PDF

For further illustrating the influence of structural asymmetry on the spectra and the generation mechanism of the SLR mode, Fig. 2(e) shows the reflection spectra with the upper metal stripe localized in different positions. Evidently, the structural asymmetry is a crucial parameter for the generation of reflection dip A. Here, structural asymmetry is characterized by the lateral shift d_x between the center points of the uppermost thin metal stripe and the thicker one below. It can be clearly observed that the dark mode in the symmetric case with the lateral shift d_x= 0 nm is not radiatively coupled and has no radiative losses, leading to the disappearance of dip A. In this case, the charge distribution at the wavelength of 1607 nm corresponds to the super-radiative dipolar mode. Breaking the symmetry by moving the top gold strip (changing the lateral shift d_x) horizontally allows us to controllably introduce radiative coupling to the dark mode. The dark mode with small radiative losses then leads to the generation of the ultra-narrow SLR mode, which further confirms that the structural asymmetry is a necessary condition for the generation of dip A. Furthermore, with the increase of the lateral shift d_x, the linewidth and depth of the SLR mode increases [36–37].

The physical mechanisms of dips B and C are evidently different from dip A. As shown in Figs. 3(a) and 3(b), the energy of incoming light is effectively collected and funneled into both the nano-slit and the middle silica spacer, resulting in extremely strong electric field enhancement and confinement inside the nano-slit as well as the middle silica spacer. However, the electromagnetic field above the top surface of the proposed structure is very weak. Generally, the resonant modes excited in the nano-slit and the nano-spacer are regarded as cavity mode and gap plasmon mode, respectively. Due to the coexistence of cavity mode and gap plasmon mode, the reflection dips B and C are generated through the coupling and hybridization of these two modes. Notably, the remarkable near-field distribution difference between dips B and C is the order of gap plasmon mode in the silica spacer. For dip B in Fig. 3(a), the gap plasmon mode in the spacer layer corresponds to third order as the number of nodes formed in the silica layer in each unit cell is three, which can be further verified by its surface charge distribution in Fig. 3(c). Besides, the charge distribution builds some current loops between the bottom surface of the gold stripe and the top surface of the opaque gold substrate, which emerges as magnetic dipole resonance with high absorption. Similarly, the gap plasmon mode of dip C corresponds to the fundamental mode of the resonance because there is only one node on the bottom surface of metal stripe, as shown in Figs. 3(b) and 3(d). Notably, dips B and C have wider linewidths in comparison to dip A, which contributes greatly to the extremely strong confinement of cavity and gap plasmon modes, thus increasing the inherent dissipation loss of metal. Although the gap plasmon modes in the spacer layer for dips B and C cannot contact with the external environment directly, the strong confined and enhanced cavity mode near field is still sufficiently for high sensitivity biosensing.

 

Fig. 3. Spatial electromagnetic field distributions and corresponding surface charge distributions. (a) Electric field and (c) surface charge distributions at dip B with wavelength of 1757 nm. (b) Electric field and (d) surface charge distributions at dip C with the wavelength of 2772 nm. Solid lines in (a) and (b) indicate outlines of structure respectively.

Download Full Size | PDF

Next, we investigate the influence of nano-silt width d on three resonant reflection dips A, B, and C. As shown in Fig. 4, when the nano-slit width d increases from 100 nm to 350 nm, three resonant dips are always available in the reflection spectra. However, these dips have distinct tendencies of evolution. The wavelength positions of dip A almost remain unchanged while the intensities remain remarkably very low and gradually approach to zero with the stepwise increase of slit width d from 100 nm to 350 nm, as shown in Figs. 4(a) and 4(b). Meanwhile, its linewidths also become narrower progressively, even reaching 0.3 nm when the slit width d = 350 nm. The weak dependence of dip A on slit width d makes us confirm that the generation of dip A is merely associated with the near-field distribution at the top surface of the structure and is an advantage for the uncontrollable deviation in the manufacturing process. Likewise, as shown in Fig. 4(c), the resonant wavelengths and linewidths of dip B have little change, which is primarily attributed to the relatively weaker cavity mode in the nano-slit compared with the gap plasmon mode in the dielectric spacer. This phenomenon illustrates that the gap plasmon mode in the spacer layer plays a dominant role in the excitation of dip B, which can be clearly observed from the near-filed distribution of Fig. 3(a). The near-field distribution for dip C in Fig. 4(d) is just the opposite: the electric field intensity in the nano-slit is stronger than the counterpart in the spacer layer. Therefore, the wavelength positions and intensities of dip C are extremely sensitive to the change of slit width d. As the slit width d increases, the linewidths of dip C become wider and the reflection intensities become higher.

 

Fig. 4. Influence of slit width d on three-resonant reflection dips. (a) Reflection spectra with nano-slit width of d = 100 nm, 150 nm, 200 nm, 250 nm, 300 nm, and 350 nm. Linewidths and intensities of (b) dip A, (c) dip B, and (d) dip C for various slit width d.

Download Full Size | PDF

The above-mentioned results show that the three-resonant reflection dips of the proposed nanostructure exhibit almost identical near-field enhancement intensities. Nonetheless, the near-field localization extent of these resonant modes is completely different, resulting in distinctive linewidths of these resonant modes: dip A possesses weak near-field localization, namely, large decay length, and narrow line width, which is promising for the detection of large-sized biomolecules, such as cells, viruses, and the like. And yet, the other two dips, particularly for dip C, possesses strong near-field localization, that is, small decay length, and relatively wide linewidth, which especially allows for biological small molecules like glucose, heparin, etc. Therefore, due to their distinct decay lengths, the multi-resonant modes of the proposed plasmonic nanostructure are probably applied in various application scenarios for biosensing detection.

Besides the dimensional effect of nano-silt width d, we also investigate the influence of other structural parameters on these three resonant reflection dips. As shown in Fig. 5(a), when the thickness of gold film t₁ has stepwise change from 360 nm to 420 nm, three resonant dips have relatively small changes. More specifically, wavelength positions and linewidths of dip A has small changes. The wavelength positions of both dips B and C have redshifts, and the redshifts of dip C are more obvious. Figure 5(b) shows that the small change of groove depth t₂ has little influence on three resonant reflection dips in the range from 25 nm to 50 nm, which provides an excellent tolerance for preparation of the designed structure. The dependence of reflection spectra on structural period P is demonstrated in Fig. 5(c). Three resonant reflection dips have obvious redshifts with the increase of structural period P while spectral line-shapes and linewidths keep almost unchanged. Remarkably, dip C exhibits more evident redshifts in comparison to dips A and B. Similar to groove depth t₂, the width of top gold stripe w₁ has a minor impact on these three resonant reflection dips as well.

 

Fig. 5. Influence of other structural parameters on three-resonant reflection dips. (a) Thickness of gold film t₁, (b) Groove depth t₂, (c) Structural period P, and (d) Top gold stripe width w₁.

Download Full Size | PDF

To elaborately evaluate and compare the sensing performance of multi-resonant modes in our asymmetric step-shaped plasmonic metamaterial, bulk refractive index (RI) sensitivity and surface sensitivity are universally utilized as two crucial criteria. Here, bulk RI sensitivity is defined as the ratio of the wavelength shift of reflection dip to the refractive index change of ambient bulk RI, which is determined by the slope of linear fitting. Surface sensitivity is expressed as the amount of wavelength shift of reflection dip with a fixed nanometric dielectric layer thickness (equivalent to a biomolecule layer). We first focus on the bulk RI sensitivity, as shown in Fig. 6. When the environmental RI increases from 1 to 1.5, the resonant modes A and C have significant redshifts with their lineshapes and depths remaining almost unchanged. However, mode B gradually disappears and then reappears to the left of dip A, which is different from dips A and C. Therefore, in the following, the investigation on bulk RI sensitivity is divided into two parts for ambient RI range: one is air environment with RI of 1, the other is the aqueous solution with RI range from 1.33-1.38. Specifically, with the increases of environmental RI from 1.00 to 1.05, as shown in Fig. 6(b), there are three resonant modes used for sensing applications. The depths of dips A and C always keep a fixed value and that of dip B has only a slight decrease. The dependence of extracted resonance wavelengths related to these three reflection dips on the environmental RI is shown in Fig. 6(d), which exhibits good linearity in the whole studied RI range. The bulk sensitivity is approximately obtained by calculating the slope of the linear-fitting curve, which of the three modes are 1605 nm/RIU, 650 nm/RIU, and 1000 nm/RIU, respectively. However, most sensing applications are carried out in aqueous solution, it is very important to calculate the bulk RI sensitivity in aqueous environment. Figure 6(c) shows the case of RI from 1.33 to 1.38. In this case, dip B vanishes and dips A and C remain their lineshapes and depths unchanged. Figure 5(e) shows the dependence of extracted resonance wavelengths related to dips A and C on the environmental RI and the bulk sensitivities of them are 1607 nm/RIU and 764 nm/RIU, respectively. Briefly, the bulk RI sensitivity for dip A is the highest among these resonant modes, which is mainly attributed to strong electric field enhancement and large near field decay length. A large decay length means dip A possesses relatively large detection spatial dimension, so it is sensitive to large-sized analyses. Furthermore, another crucial parameter, figure of merit (FOM), defined as the ratio of RI sensitivity to linewidth, is also used for evaluating the sensing performance. Due to the ultra-narrow line width of dip A, its FOM is achieved up to 803 RIU^-1 in both air and aqueous environment. The high FOM will make distinguishing tiny shift of resonant wavelength induced by ultra-low concentration of analyst effortlessly. Compared to dip A, dip C demonstrates modest bulk RI sensitivity of 1000 nm/RIU in air and 764 nm/RIU in water. However, due to its wide line width, the FOM of dip C is reduced to 6 RIU^-1 in air and 4 RIU^-1 in aqueous. The low FOM of dip C is an unfavorable factor for low concentration detection. For a direct comparison, the bulk RI sensitivities, linewidths of resonant modes, and FOMs of some previously reported plasmonic nanostructure sensors [19,23–25,38–40] are summarized in Table 1. Obviously, the bulk RI sensitivity and FOM of dip A in our proposed metamaterial sensor outperform most previous studies [19,38–40] and are comparable to the works in Refs. [23–25]. Furthermore, the more comprehensive analysis for the sensing performance of our nanostructure in two situations including aqueous solution and air environment is investigated. As for the reflection dip B of the proposed metamaterial sensor, both bulk sensitivity and FOM are lower than that of modes A and C due to its weak electric field enhancement in the nano-slit for the air environment. However, its bulk RI sensitivity is also higher than some previously reported works [19,39–40]. As mentioned above, reflection dip B only occurs in air environment and disappears in aqueous solution environment, which will limit the applications of dip B. Fortunately, dip B may be applied in the field of surface-enhanced Raman scattering (SERS) for biomolecular detection when the excitation wavelength of laser is consistent with resonant dip B because SERS detection is generally performed in air environment.

 

Fig. 6. Bulk sensing performance of the proposed plasmonic metamaterial. Reflection spectra with environmental (a) RI from 1 to 1.5, (b) RI from 1.00 to 1.05, and (c) RI from 1.33 to 1.38. (d) Dependence of extracted wavelength position of three reflection dips on environmental RI from 1.00 to 1.05, and (e) RI from 1.33 to 1.38.

Download Full Size | PDF

Table 1. Comparison of the sensing performances of plasmonic nanostructure sensors

View Table

The above-mentioned bulk RI sensitivities only represent the response of resonant modes on the whole ambient environment. More specifically, RI changes from the structure surface to infinity. Therefore, in order to comprehensively evaluate the sensing performance of the proposed plasmonic metamaterial, surface sensitivity, generally reflecting the electric field distribution on the surface, is considered. We simply evaluate it by calculating the wavelength shifts of resonant dips with various layers of high RI (1.52) thin dielectric films modeling bio-molecular films such as proteins, DNAs, and glucose on the structure surface at the nanoscale. Figures 7(c) and 7(d) demonstrate the dependence of reflection spectra on the thickness of the dielectric layer in aqueous solution with environment RI of 1.33. The wavelength shifts of the two resonant reflection dips A and C induced by respective changes in the thickness of dielectric layer from 0 nm to 900 nm and from 0 nm to 180 nm are shown in Figs. 7(e) and 7(f). Evidently, when the thickness of the dielectric layer increases to 900 nm, the wavelength position of dip A still has an evident redshift, which is a direct proof for weak near-field localization and large decay length of dip A. Whereas, the wavelength shift of dip C reaches to a plateau with a 90 nm-thick dielectric layer in Fig. 7(f). This illustrates dip C is more sensitive to the surface RI changes than dip A. Hence, the sensing performances between dips A and C are exactly complementary for both bulk RI and surface sensitivities. Specifically, dip A possesses high bulk RI sensitivity, and yet lower surface sensitivity while dip C is just opposite. This phenomenon further consolidates the drawn conclusion: the near field of dip C possesses strong localization (small decay length), and yet, that of dip A has relatively weak localization and longer decay length. Therefore, the proposed plasmonic metamaterial own different near-field sensitivity extent and distinct bulk RI sensitivities and FOMs, which makes it very promising for various application scenarios.

 

Fig. 7. Surface sensing performance of the proposed plasmonic metamaterial. Schematic of high refractive index dielectric layer covering the top surface of structure with (a) film thickness smaller than 50 nm and (b) film thickness larger than 50 nm. (c) Reflection spectra with different thickness of dielectric layers from 0 nm to 900 nm in aqueous solution with an interval of 180 nm. (d) Reflection spectra with different numbers of dielectric layers from 0 nm to 90 nm in water solution with an interval of 16 nm. Wavelength shifts of (e) dip A and (f) dip C on the thickness of dielectric layer.

Download Full Size | PDF

4. Conclusions

In this paper, we have theoretically demonstrated an asymmetric step-shaped plasmonic metamaterial exhibiting three near perfect absorption peaks, which can be utilized for various application scenarios. These three absorption peaks correspond to three different plasmonic hybridized modes with distinguished near-field distributions. The SLR mode at the short wavelength range with both high bulk refractive index sensitivity and figure of merit possesses larger near-field spatial dimension, which enables the detection of large-sized biomolecules. However, the absorption peak at the long wavelength range, stemming from the coupling between slit-cavity mode and gap plasmon resonance, shows high surface sensitivity and modest bulk RI sensitivity, which especially allows detecting biological small molecules. Therefore, the complementary near-field distributions between different resonant modes in the proposed nanostructure can be used to detect a broad range of biomolecules with various sizes, which is very promising for developing high performance versatile sensing platform for various application scenarios.