1. Introduction

Biosensing is one important application of plasmonics [1–4]. Plasmonic nanostructures strongly confine and enhance the incident electric field, thus inducing an efficient electromagnetic coupling with the surrounding material. The plasmon resonance frequencies depend sensitively on the refractive index of this surrounding medium, so that plasmonic resonators can be used to probe refractive index changes. Plasmonic resonators for surface enhanced spectroscopies and surface plasmon resonance (SPR) sensing devices make use of these exceptional optical properties.

The mid-infrared (mid-IR) is a spectral range of interest for biosensing as it comprises the characteristic molecular absorption lines that can be used to identify different substances [5]. The weak interaction between IR light and molecules, caused by the mismatch of their characteristic length scales, can be increased when the molecules are located in the vicinity of plasmonic resonators, notably in an enhanced electric field. This effect is referred to as surface enhanced infrared absorption (SEIRA). Furthermore, SPR sensing in the mid-IR has been proposed for its enhanced sensitivity compared to sensors operating in the visible spectrum [6]. This is because the molecular absorption features are accompanied by variations of the refractive index n [7], resulting therefore in larger plasmon resonance shifts.

Typically, gold nanostructures are used for plasmonic biosensing and surface enhanced spectroscopies [8–16]. They perform well in the visible and near-IR spectral range but they show limitations especially at higher wavelengths due to intrinsic losses. Research for tunable and adapted materials is an essential requirement to create biosensors for the mid-IR. Several propositions for alternative plasmonic materials have been made, for example group IV materials as Si [17, 18], Ge [19, 20] and graphene [21], transparent conductive oxides (TCOs) and III-V semiconductors [22], where notably InAs features excellent properties [23–26]. InSb has also been investigated for mid-IR plasmonics [27].

Here, we propose 1-D periodic grating structures consisting of highly doped InAsSb on GaSb substrates. These nanostructures are realizable by large area surface patterning methods as photolithography or interferential lithography. It has been demonstrated previously that they support localized surface plasmon resonances (LSPR) [28]. InAs_0.91Sb_0.09 is lattice matched to the substrate, guaranteeing a high crystalline quality. Besides, it can be doped up to 10²⁰ cm⁻³ and has a small effective electron mass $m_e^{*}$, both important parameters to tune the plasma frequency

In this article, we theoretically study the influence of the grating’s geometric parameters on the sensing properties and optimize the structure with regard to high electric field enhancement and the refractive index sensitivity evaluated for a thick, bulk like material layer. These objectives take into account the requirements for both SEIRA spectroscopy and SPR sensing and allow the combination of both techniques on one sensing device. We assess in simulations the tunability of the LSPR wavelength through variation of the ribbon width and thickness. We find a large range of resonant structures at a chosen working wavelength for sensing. Theoretical sensitivity values of around 900 nm RIU⁻¹ are obtained for SPR sensing of bulk material embedding the resonators. We also investigate the SPR shift and SEIRA effects for a thin adlayer on the plasmonic resonator grating using simple permittivity models. Moreover, we carry out a comparative study between the highly doped InAsSb and Au gratings and demonstrate experimentally, in agreement with the simulations, that the semiconductor based grating outperforms the gold grating in the investigated spectral range.

2. Numerical finite-difference time-domain (FDTD) simulations

2-D FDTD simulations have been performed with a commercial software package to calculate the optical response and the electric near field distribution of plasmonic resonator structures [29]. We assumed translational invariance of the structures in z-direction so that calculations could be restricted to the xy-plane.

A cross sectional view of the modeled, idealized structure is shown in Fig. 1(a). The grating is made of rectangular ribbons placed onto the substrate. The geometric parameters used for the optimization are indicated in the schematic. Ribbon thickness t and width w are variable while the pitch Λ is fixed at 500 nm. This pitch is compatible with interferential lithography surface patterning techniques and allows for a high densification of the plasmonic resonators.

 

Fig. 1 (a) Schematic illustration of the simulation set-up and the geometry of the grating. The ribbon thickness t and width w are variable. The pitch Λ is set to 500 nm. The modeled unit cell of the periodic structure is indicated by the dashed line surrounding the ribbon in the middle. The broad-band light source is placed above the plasmonic resonator structure. The incident waves, indicated by the wave vector $\overrightarrow{k}$, propagate in negative x-direction. The electric field vector $\overrightarrow{E}$ oscillates along the y-direction. (b) Schematic of the grating embedded in bulk material. This model was used to investigate the LSPR wavelength shift. (c) Schematic of the grating topped with a SAM applied to investigate the sensitivity on refractive index variations within the close proximity of the resonators.

Download Full Size | PDF

The material of the plasmonic grating resonators, highly doped InAs_0.91Sb_0.09, is modeled by a Drude-function

We investigate the reflectance of plasmonic gratings as well as the scattering properties of isolated ribbons as the constituting element of gratings. For both types of simulations, the initial field distribution is introduced by a broad-band source emitting normal to the structure as illustrated in Fig. 1(a) in a spectral range of 1–15 µm. Simulations are performed with incident light polarized parallel to the axis of the ribbon width, so that the LSPR can be excited. Accordingly, the electric field vector $\overrightarrow{E}$ oscillates along the y-direction. Because the ribbon is infinitely extended in the direction of its length axis along the z-direction, no LSPR can be observed for polarization along this axis. For periodic structures, a plane wave source is used. Only one unit cell is modeled and the periodicity is incorporated by Bloch conditions as boundaries in y-direction. Perfectly matched layer (PML) boundary conditions are used in the non-periodic x-direction. When investigating isolated ribbons, a Total-Field-Scattered-Field (TFSF) source is applied [33]. PML boundary conditions are used to delimit the calculation zone in both directions. The ribbon is placed at least one wavelength away from the PMLs.

Sub-gridding techniques with mesh cell sizes of 1 nm or 0.25 nm for detailed characterization of selected structures were applied around the resonator structures to have high spatial resolution in the areas of strongest refractive index variation and highest field enhancement. Convergence testing was performed by variation of the distance of the structure to the PMLs, the number of PML layers, the mesh accuracy, source and monitor’s positions, the source’s bandwidth, and the mesh size of the sub-gridding.

3. Results and discussion

3.1. Resonances in plasmonic semiconductor gratings

Calculated reflectance spectra of InAsSb plasmonic gratings of variable ribbon width w and constant pitch Λ = 500 nm are shown in Fig. 2(a), exemplary for 100 nm thick gratings. Two LSPR modes can be observed in each spectrum: one weak excitation at around 6 µm and one more pronounced, larger mode at higher wavelength that depends sensitively on the chosen geometry. The latter mode continuously shifts to higher wavelength as the ribbon width increases. We define the central peak position of this mode as the resonance wavelength λ_R of a structure. Higher reflectance intensity for this principal LSPR mode is observed with increasing width of the resonators, that is, with increasing ratio of the strongly reflecting highly doped semiconductor to the transparent substrate within one unit cell.

 

Fig. 2 (a) Reflectance spectra of 100 nm thick InAsSb gratings on GaSb substrate for several ribbon widths w. The pitch Λ is 500 nm. (b) Electric field profile associated to the mode at high wavelength. Note that the color map scale is reduced to 20 in order to determine how the electric field enhancement decreases with distance from the structure. (c) Electric field profile associated to the mode close to 6 µm. The field profiles have been calculated with a narrow mesh, revealing therefore high electric field values thanks to the fine resolution around the corners of the resonator. (d) Resonance wavelength λ_R as a function of the geometrical parameters (ribbon width w and thickness t) of the InAsSb grating. The chosen working wavelength of 10 µm is indicated by the dashed line. Several geometric configurations close to the dashed line provide resonance maxima near 10 µm. (e) Averaged values of the electric field E_avg at 10 µm (normalized to the incident field strength E₀), within an area of 50 nm × 50 nm centered at the ribbon’s corner towards the substrate, obtained for different geometric configurations of the InAsSb grating. The geometries corresponding to the resonant structures close to the dashed horizontal line in (d) are located on the black line. To establish the numerous data points, the mesh size was larger than for the exemplary field profiles shown in (b) and (c) due to computing power constraints.

Download Full Size | PDF

Plasmonic resonances are associated to an electric field enhancement which is one essential factor for sensing. The electric field profiles at the wavelengths of the two LSPR modes for one exemplary structure are shown in Figs. 2(b) and 2(c). The electric field E is normalized to the amplitude of the electric field E₀ of the source. The enhanced electric field associated to the principal LSPR mode is localized around the resonator’s corners towards the substrate whereas the second, smaller mode in the reflectance spectrum features an enhanced electric field mainly localized on the top corners. The maximum field enhancement for this exemplary resonator geometry corresponds to 155 and 64, for the strong and weak LSPR peaks, respectively. This highly enhanced electric field is decreases quickly, so that at 15 nm distance from the resonator its value has decreased to below 20. Strong electric field confinement is attractive for the detection of molecular absorption lines [34], especially for small molecules.

The resonance wavelength λ_R has been determined for a set of different geometrical configurations, the investigated range of ribbon thicknesses and widths (t from 50 to 200 nm, w from 50 to 400 nm) being chosen with regard to technological feasibility. Figure 2(d) displays the obtained wavelengths λ_R for several grating thicknesses. The resonance wavelength can be tuned to specific values within the mid-IR by variation of the ribbon width while keeping a constant ribbon thickness.

A working resonance wavelength of λ_R = 10 µm was chosen to optimize the sensing properties. This sets an example for sensors operating far into the mid-IR. Around this wavelength, a transparency window of the atmosphere facilitates spectroscopic measurements as the signals will not be superposed by H₂O or CO₂ absorption lines. Also, several aliphatic amines have vibrational modes in this spectral range [35], so that there is interest in operating biosensors around this wavelength. As indicated by the dashed horizontal line in Fig. 2(d), several geometrical configurations can be envisaged resulting in this peak wavelength. This set of geometrical configurations will be referred to as the resonant structures throughout this manuscript, keeping in mind that this is related to the chosen working wavelength.

Having investigated the far field reflectance spectra, we now analyse the influence of the geometry on the electric near field. It is known from literature that the near field maximum is red-shifted from the far field reflectance maximum, because the plasmons excited in resonators with finite size behave like driven, damped harmonic oscillators [36,37]. Structures having their far field resonance wavelength at 10 µm might even achieve higher field enhancement values at a slightly higher wavelength. However, far field quantities like the reflectance maximum are easier accessible in simulation and experiment. We will use the far field maximum at 10 µm to choose the resonant structures and assume a negligible detuning between near and far field quantities, especially as we investigate systems with plasmonic resonances of non-negligible full width at half maximum (FWHM) and therefore good overlap between their plasmonic and the vibrational modes of molecules. Figure 2(e) shows the averaged electric field strength E_avg at 10 µm normalized to the incident electric field strength E₀ as a function of the geometric parameters ribbon width w and thickness t. In order to minimize the influence of singular values due to the meshing used in the model, we use an averaged value of the electric field within an area of 50 nm × 50 nm centered at the ribbon’s corner towards the substrate. The resonant structures are indicated by the black line. It can be observed that the resonant structures, especially those of more than 100 nm ribbon thickness, generate high field enhancement at the investigated wavelength, compared to the far off-resonant structures, so that they comply with our objectives for the sensing device.

3.2. Sensitivity for SPR sensing

Besides the electric field enhancement, the sensitivity on refractive index changes caused by bulk material around the resonators is investigated. To model this, the plasmonic resonators were embedded in a dispersionless material of varying thickness d with refractive index n = 1.61, as depicted in the inset of Fig. 3(a). The enhanced electric field is evanescent in the free space around the resonators, thus probing the environment over a certain distance. Hence, the resonance wavelength shift induced by the material depends on its thickness [see Fig. 3(a)]. Yet, a saturation value can be obtained for a sufficiently thick layer that was found to be at least 1 µm. In Fig. 3(b), it is shown that the resonance shift is very sensitive on the material thickness for thin layers up to 200 nm. Then we find a zone with moderate thickness dependence and finally the saturation for material layers of around 1 µm and above, as indicated by the exponential decay fit function. The increasing thickness of the deposited layer leads then mainly to a decrease of the reflectance intensity and has only little influence on the resonance wavelength.

 

Fig. 3 (a) Reflectance spectra of an exemplary InAsSb grating (ribbon thickness 100 nm, ribbon width 260 nm) embedded in dispersionless material of varying thickness d, as shown in the inset. The material has a refractive index of n =1.61. (b) Resonance wavelength λ_R as a function of the layer thickness d of the embedding material for three structures of different geometry as indicated in the graph. The dashed lines correspond to exponential decay fits and serve as guide to the eye to highlight the saturation of the wavelength shift with increasing material thickness.

Download Full Size | PDF

Simulations of the resonant structures embedded in 1 µm of the dispersionless material were performed. Based on those simulations, the sensitivity S and figure of merit (FOM) were calculated according to the expressions given by Mayer et al. [4], where an approximately linear dependence of the resonance wavelength shift on the variation of the refractive index n is assumed:

As we underlined the dependence of the resonance wavelength shift on the material layer thickness, we will furthermore use a material thickness dependent sensitivity $S_d^{*}$:

Figure 4 presents the sensitivity $S_{1\mu m}^{*}\approx S$ and the FOM of the resonant structures. Sensitivity values of (900 ±20) nm RIU⁻¹ were obtained. The simultaneous variation of both geometric parameters, ribbon thickness t and the accordingly chosen ribbon width w, and the discretization of the values introduces the dispersion of the data points. A slightly decreasing trend is found for the FOM due to the increasing FWHM of the plasmonic mode [see inset of Fig. 4], that is related to the bigger resonator size and, correspondingly, to the increased damping. For SPR sensing, small FWHM are usually requested in order to easily determine the resonance wavelength and its shift. In contrast, the FOM is usually not taken into account for pure SEIRA applications, as large plasmonic resonances provide the advantage of presenting a spectral overlap with a wide range of absorption lines.

 

Fig. 4 Sensitivity S (black dots, left ordinate) and figure of merit FOM (red triangles, right ordinate) for the resonant structures. Values were obtained for plasmonic resonators embedded in 1 µm of dispersionless material with n = 1.61. Sensitivity values of (900 ± 20) nm RIU⁻¹ were reached for all structures. The FOM drops slightly which is mainly caused by the increasing FWHM of the LSPR with increasing resonator size as shown exemplary for the smallest and the largest investigated structure in the inset.

Download Full Size | PDF

The sensitivity of the here proposed structures exceeds those of metal-based systems for the visible spectral range: the theoretical sensitivity of typical gold grating coupler-based SPR sensors is 309 nm RIU⁻¹ at 630 nm operating wavelength, and 630 nm RIU⁻¹ at 850 nm [38]. Gao et al. [14] report on a sensitivity of around 480 nm RIU⁻¹ at 650 nm for a plasmonic interferometer sensor made of silver and Vincenti et al. [12] indicate a sensitivity of 650 nm RIU⁻¹ in the visible spectrum for 2-D periodic gold patch gratings on silicon substrate, with exceptionally narrow linewidths leading thus to high FOMs. Focusing on sensors for the mid-IR, Law et al. [25] demonstrated wavenumber shifts of up to 71 cm⁻¹ corresponding to a wavelength shift of 740 nm for InAs nano pillars on GaAs substrates coated with 50 nm polymethylmethacrylate (PMMA; refractive index n = 1.49). This indicates that the sensitivity is equal to or higher than $S_{50\text{nm}}^{*}=1510\text{nm}{\text{RIU}}^{-1}$. Nano pillars have a larger active surface for sensing than the here proposed, simpler ribbon geometry, and both studies consequently demonstrate the suitability of InAs and InAsSb, respectively, for the mid-IR.

3.3. Optimized plasmonic grating geometries

For the following detailed study on sensing behavior we have chosen a resonator geometry of 100 nm thickness, 260 nm ribbon width, and 500 nm pitch, that is, one structure guaranteeing a good trade-off between high electric field enhancement and the FWHM of the LSPR, envisaging both SEIRA and SPR sensing applications.

Figure 5(a) displays an overview reflectance spectrum of the chosen resonator geometry. The resonance wavelength λ_R has been precisely tuned to 10 µm. The associated electric field profiles to the LSPR modes have previously been shown [see Figs. 2(b) and 2(c)]. For this structure, the sensing properties for monolayers were investigated. The choice of the material model allows to study SPR sensing- and SEIRA effects separately. For both models, the layer thickness of the analyte material was 2 nm. For the dispersionless material with n = 1.61, applied to analyze the SPR sensing capabilities, a clear red shift of the resonance wavelength of (45 nm ±3) nm was observed as shown in Fig. 5(b), corresponding to a thickness dependent sensitivity $S_{2\text{nm}}^{*}=70\text{nm}{\text{RIU}}^{-1}$. This corroborates the excellent sensitivity obtained for sensing of thicker material layers.

 

Fig. 5 (a) Overview reflectance spectrum of one selected resonator geometry of 100 nm thickness and 260 nm ribbon width. (b) Reflectance spectra of the InAsSb grating (black dotted curve) and the InAsSb grating covered by a 2 nm SAM with constant refractive index n = 1.61 (red dashed curve). The vertical dashed lines indicate the maximum reflectance position of the SPR peak. The wavelength difference between the two vertical lines is Δλ = 45 nm as labeled. (c) Reflectance spectra of the InAsSb grating, an unstuctured InAsSb layer and a GaSb substrate covered by an absorbing 2 nm SAM. The material was modeled as a Lorentz oscillator material with an absorption line at 10 µm. In each case, the reflectance of the model with constant refractive index is shown as well, but the spectra overlap greatly apart from the range around 10 µm.

Download Full Size | PDF

In order to simulate the SEIRA effect, an artificial absorption line was then introduced via the Lorentz material model. The vibrational signal was investigated for the material on the optimized resonator grating, and as a comparison on an unstructured InAsSb layer and directly on the GaSb substrate without the highly doped semiconductor. In Fig. 5(c), the reflectance spectra for the different models are shown. When the absorption feature and the maximum of the plasmonic mode overlap, the absorption line induces a dip in the reflectance spectrum, similar to other resonator systems found in the literature [8, 39]. On the InAsSb layer and the GaSb substrate, the signal from the adlayer is extremely weak. The enhancement is consequently induced by the plasmonic resonators patterned into the InAsSb layer. As a reference, the reflectance spectra of a 2 nm thin adlayer with a constant refractive index are shown for each case. Only for the InAsSb grating there is a clearly visible difference between the layer with absorption feature and without. The spectra also serve to extract the vibrational signal by taking the rapport between the reflectance calculated with the Lorentz model and the one with constant refractive index. From this rapport it can be deduced that the absorption line results in a reflectance variation of 9.9% for the adlayer on the plasmonic grating, whereas the reflectance only changes of 0.3% on the GaSb substrate and of 0.19% on the InAsSb layer for the same strength of the vibration.

3.4. Comparison of semiconductor and gold gratings

When typical plasmonic materials like gold are used for mid-IR applications, it is necessary to adapt the resonances of the material to the desired spectral range by design considerations. Many structures feature for example nanometer sized gaps to confine the electric field and to shift the resonances to higher wavelengths [39–41]. Another possibility to reach high electric field strength for sufficient signal enhancement for SEIRA is to use carefully designed nanorod antenna arrays to excite collective plasmon resonances [42, 43]. In the following, the simple semiconductor gratings are compared to equal structures made of the state-of-art material in plasmonics.

Firstly, the scattering maximum of single gold resonators was tuned to 10 µm by adapting the ribbon width w in order to have the same working wavelength as for the InAsSb grating. We fixed the ribbon thickness t to 100 nm to have the same thickness for the semiconductor and the gold grating. We found a linear relationship between the gold ribbon width w and the scattering maximum as indicated in Fig. 6(a). The higher the refractive index of the substrate, the steeper the slope of the linear relationship, as it had been shown for micrometer-sized nanorod antennas on different substrates by Huck et al. [15].

 

Fig. 6 (a) Scattering maximum of isolated, 100 nm thick gold ribbons as a function of their width. A linear fit was performed (dotted lines). (b) Reflectance spectra of gold gratings on GaSb substrate with 1.6 µm wide ribbons and different pitches Λ. The different diffraction orders are highlighted in the graph: solid lines mark the diffraction orders from the gold-grating GaSb interface, dashed lines the diffraction orders from the air-gold grating interface, beginning with the first order at highest wavelength. For Λ = 2.8 µm, the first order diffraction from the lower interface has been tuned to be in resonance with the scattering maximum of the ribbons.

Download Full Size | PDF

On GaSb substrate, approximately 1.6 µm large gold ribbons provide a scattering maximum at 10 µm. Consequently, a mid-IR biosensor based on gold gratings will be at least four times larger in scale compared to the semiconductor grating for the same number of resonators and thus the same number of active zones where field enhancement occurs. Moreover, for plasmonic gratings made of gold, the pitch is necessarily in the order of magnitude of the wavelengths incident on the grating so that diffraction effects become important. Working under normal incidence in reflectance configuration, diffraction arises from the air-gold-grating interface as well as from the lower gold-grating-GaSb interface. Yet, as shown in the simulations in Fig. 6(b), no plasmonic excitations can be observed as diffraction dominates and might hide all plasmonic effects. For this reason, it is necessary to investigate and compare isolated resonators made of InAsSb or gold.

Figures 7(a) and 7(b) show the scattering spectra of an isolated InAsSb ribbon and an isolated gold ribbon, respectively, with same dimensions as for the gratings with and without a dispersiveless 2 nm thin SAM on top of the ribbon. While a red shift of (34 nm ±5) nm can be observed in the scattering spectrum of the InAsSb ribbon, corresponding to $S_{2\text{nm}}^{*}=60\text{nm}{\text{RIU}}^{-1}$, the spectrum of the gold ribbon presents no detectable difference in terms of simulation with and without the SAM. We presume that this is due to the different field enhancement factors and field profiles associated to the semiconductor and the gold structure: in Fig. 7(c), we contrast the electric field profile at 10 µm for the single InAsSb ribbon and the single gold ribbon. The mode profiles resemble those observed for gratings. Note the different spatial extensions of the resonators, which allow for high densification of the InAsSb ribbons. The electric field strength maximum of the semiconductor structure is clearly higher. Moreover, the electric field is strongly confined to the the corners of the ribbon, making the structure thus highly sensitive for refractive index changes in the close proximity. Figure 7(d) underlines the different field enhancement and confinement by displaying the evanescent field extracted from a cut through the field profiles 1 nm above the substrate-resonator interface. The resonator side wall has been placed at the position y = 0 to compare the evanescent fields for the semiconductor and the gold resonator. In the considered spectral range, gold has a large and negative real part of the permittivity. The electric field is less confined as for materials with a small negative permittivity, and the enhancement factor is lower. Consequently, the resolution of this gold structure is not sufficient to detect the refractive index difference within 2 nm thickness.

 

Fig. 7 (a) Scattering spectra of a single InAsSb ribbon on GaSb substrate with and without a 2 nm thin layer with refractive index n = 1.61. A wavelength shift of (34 nm ± 5) nm is introduced by the refractive index change as indicated by the dashed, vertical lines. (b) Scattering spectra of a single gold ribbon on GaSb substrate with and without the monolayer. (c) Electric field profile at 10 µm for the single InAsSb ribbon (left side) and the single gold resonator (right side) on GaSb substrate. The InAsSb resonator allows for high densification, as shown by the spatial extensions, while leading to a one order of magnitude higher electric field strength. (d) Cut through the field profile 1 nm above the substrate-resonator interface (x = −1 nm). The side wall of the resonator has been placed at the position y = 0 in order to compare the evanescent field of the semiconductor and the gold resonator.

Download Full Size | PDF

3.5. Experimental validation of the simulations

To validate the above shown simulation results, an InAsSb grating with pitch Λ = 500 nm, ribbon thickness t = 100 nm and width w = 360 nm, was fabricated by interferential lithography and dry etching as described elsewhere [28]. Typical Drude parameters of the grown InAsSb film can also be found therein. Also, a gold grating with Λ = 2.6 µm, t = 100 nm and w = 1.88 µm was fabricated. To provide for the larger dimensions, UV lithography was used to pattern a photoresist bilayer (LOR A5 and AZ MIR 701) prior to the deposition of 3 nm Ti adhesion layer and 100 nm gold by electron beam evaporation. The lift-off of the spare material was done in an acetone bath. Scanning electron microscope (SEM) images of the samples are shown in Figs. 8(a) and 8(b).

 

Fig. 8 (a) SEM image of the InAsSb grating. (b) SEM image of the gold grating. The scale bars are 4 µm in (a) and (b). (c) Experimental reflectance spectra of an optimized structure with and without a 200 nm thick layer of PMMA A4 photoresist. The experimentally observed red shift of 480 nm (±70 nm) is indicated by the dashed vertical lines. (d) Experimental reflectance spectra of a gold grating (Λ = 2.6 µm, t = 100 nm, w = 1.88 µm) with and without a 200 nm thick layer of PMMA A4 photoresist.

Download Full Size | PDF

Reflectance measurements were performed using a Hyperion 3000 microscope coupled to a Vertex 70 FTIR spectrometer with a globar lightsource, a KBr beam splitter and the microscope’s internal MCT detector. A 15x objective with NA = 0.4 was used. The beam path was purged with nitrogen and the space between the objective and the sample was enclosed by a plexiglas housing in order to minimize environmental fluctuations. Spectra were taken with a resolution of 1 cm⁻¹ in a wavenumber range from 600 to 5500 cm⁻¹. The acquired sample spectra were normalized by the spectrum of a gold mirror serving as reference. The polarization of the quasi-normal incident light was either parallel or perpendicular to the ribbons’ width axis. We use the symbols ‖ and ⊥ to name the polarization with respect to the width axis.

First, reflectance spectra of the samples as prepared were taken, shown as the black spectra in Figs. 8(c) and 8(d). A strong plasmonic feature at 9.76 µm can be observed when illuminating the InAsSb grating sample with ‖ light. In the case of ⊥ polarization, no LSPR is excited in the investigated spectral range. For the gold grating, diffractive modes appeared in the spectrum taken under ‖ polarization. In the considered wavelength range between 6 and 14 µm, we attribute the major peak at 10.2 nm to the first order diffraction from the gold grating-GaSb interface. For the ⊥ polarization, high reflectivity near unity is observed as the electric field vector points along the homogeneous gold ribbons. The electric field probes consequently a uniform material and not a periodic variation of the refractive index.

Subsequently, a 200 nm thick layer of PMMA A4 photoresist was deposited onto the samples by spin coating before measuring again the reflectance. The PMMA introduces the refractive index modification necessary to investigate the sensing properties of the samples. A red shift of the LSPR of (480 nm ±70) nm was measured when the InAsSb grating is covered with PMMA A4, with n = 1.49. This yields a thickness dependent sensitivity $S_{200\text{nm}}^{*}=980\text{nm}{\text{RIU}}^{-1}$. We find similar tendencies as those resulting from simulations. The deviations in terms of the resonance wavelength and the even higher sensitivity might be due to differences between the idealized rectangular geometry used in simulations and the real samples whose cross section is influenced by the etching procedure. The experimental results confirm that a plasmonic grating made of InAsSb operated in the mid-IR spectral range is highly sensitive to refractive index changes. For the gold grating under ‖ polarization, the first order diffraction mode shifts less than 12 nm on deposition of PMMA. This is smaller than the order of magnitude of the error. We observe a variation in the reflectance intensity on the long wavelength side of the peak probably related to the modified interface.

PMMA presents absorption lines within the investigated spectral range. To determine their spectral position, absorption measurements of PMMA on a gold surface were performed under grazing incidence using the Hyperion 3000 microscope equipped with a grazing angle objective with light incidence between 54 and 84. The absorption spectrum is presented in Fig. 9. This measurement serves as reference as it is taken on the same film as the one on the gold grating and deposited under same conditions as the one on the InAsSb grating. We find marked absorption lines at 7.87 µm (1271 cm⁻¹), 8.02 µm (1246.9 cm⁻¹), 8.36 µm (1196.2 cm⁻¹) and 8.67 µm (1153 cm⁻¹). The features can also be seen in both polarizations on the InAsSb grating. We observe furthermore two weak absorption lines at 10.1 µm (990 cm⁻¹) and 10.35 µm (966 cm⁻¹). Owing to their overlap with the LSPR, these weak feature are enhanced and are visible as dips close to the center of the LSPR peak whereas they can not be observed in the spectrum taken with ⊥ polarization. This evidences that the highly doped InAsSb grating provides sufficiently strong electric field enhancement to enhance weak vibrational lines when they are spectrally overlapping with the LSPR. A quantitative analysis of the vibrational signal enhancement will be subject of a further specific study on SEIRA.

 

Fig. 9 Left y-axis: Experimental absorption spectrum of PMMA on a smooth gold surface measured under grazing incidence (black line). Absorption features of interest are indicated by dashed lines. Right y-axis: Experimental reflectance spectra of the InAsSb grating covered with PMMA photoresist. The most intense absorption features in the spectral range can be observed in both polarization while the weaker ones at 10.1 µm (990 cm⁻¹) and 10.35 µm (966 cm⁻¹) are sufficiently enhanced by the plasmonic grating when the LSPR is excited under ‖ polarization but not under ⊥ polarization. The dashed lines indicate the absorption lines at 7.87 µm (1271 cm⁻¹), 8.02 µm (1246.9 cm⁻¹), 8.36 µm (1196.2 cm⁻¹), 8.67 µm (1153 cm⁻¹), 10.1 µm (990 cm⁻¹) and 10.35 µm (966 cm⁻¹).

Download Full Size | PDF

4. Conclusion

We presented the tunability of the plasmon resonance wavelength λ_R of 1-D periodic InAsSb gratings on GaSb substrate by variation of the geometrical parameters of the grating in FDTD simulations. Focusing on targeted molecular absorption around the wavelength 10 µm, a large range of structures with a resonance peak at this wavelength was found. All resonant structures with at least 100 nm thick ribbons and correspondingly chosen ribbon width showed strong electric field enhancement. The theoretical sensitivity for bulk material refractive index sensing of 900 nm RIU⁻¹ underlines the interest in these structures for sensor applications. Investigating one optimized structure with ribbon thickness t = 100 nm and width w = 260 nm in detail, we reported a remarkable wavelength shift of 45 nm for an only 2 nm thin material layer in simulations. Moreover, we compared in simulation and experiment the InAsSb plasmonic grating to a gold grating. While the InAsSb grating showed easily recognizable wavelength shifts and corroborates the bulk material refractive index sensitivity value found in simulations, the gold grating is mainly dominated by diffraction phenomena due to the dimensions necessary to work in the mid-IR with a simple geometry as the here proposed gratings. We also observed the enhancement of weak vibrational signals spectrally overlapping with the LSPR mode of the InAsSb grating. Our comparative study confirms that InAsSb is an appropriate material for mid-IR sensing devices. It can be used to create a platform for SPR sensing and SEIRA.