1. Introduction

The interest in study subwavelength plasmonic structures begin with the discovery of Ebbesen et al., 1998 [1], about the extraordinary optical transmission (EOT) phenomenon, in which is obtained transmitted light more intense than could be expected based on ray optics. The EOT was explained in terms of coupling of the incident light with localized surface plasmons (LSPs) or propagating surface plasmons (SPPs) generated on the nanostructures [2]. This phenomenon opened the doors to the development of new nanoplasmonic devices, which are currently very attractive to sensing applications, especially to chemical and biological detection [3–5]. The effects of the EOT can also be evidenced in the reflectance spectra of these nanostructures, and are explained by the same mechanisms as in the transmission spectra [1,6].

Plasmonic devices are constituted mainly by surfaces with periodic 1D or 2D metallic arrays, in which an incident light produce coherent oscillations of conduction electrons. If these arrays are fabricated with a periodicity value that matches with the SPPs wavelength, propagating charge oscillations are excited on the nanostructured surface of metal films. These SPPs are also responsible for modulation of radiation in plasmon-enhanced fluorescence and Surface-Enhanced Raman Spectroscopy (SERS) [7]. Additionally, subwavelength arrays promotes the generation of collective but non-propagating oscillations of surface electrons, LSPs, which strongly depends on the refractive index of the surrounding medium, and this fact makes possible to design refractometric biosensors able to detect slight refractive index change of the surrounding medium caused by adsorption of molecules on the metallic surface [8,9]. On the other hand, LSPs concentrates the incident electromagnetic field around the nanostructures, which under certain conditions, can promote the production of resonant effects, like Förster resonance energy transfer and Purcell effect [7,8].

Improvements in the nanofabrication techniques allowed fabricate plasmonic nanostructures increasingly complex, with higher quality and with more varied geometries [6,10], enabling the rapid development of the nanoplasmonics [11]. Metallic nanostructures, mainly AuNHs of different shapes, sizes and periodicity have been fabricated by electron beam lithography (EBL) [12], focused ion beam (FIB) [13] or nanosphere lithography (NSL) [14], which in combination with graphene or graphene oxide (GO) layers have been successfully applied as platform substrates to study SERS effect or plasmonic properties. For instance, Hao et al. [15] have presented a SERS study on graphene coated AuNHs nanostructures fabricated by EBL, using methylene blue (MB) as the probe molecule, given an enhancement factor (EF) of ~120. Zhao et al. [16] have also deposited graphene on the surface of silver nanohole (AgNHs) arrays fabricated by EBL to study SERS effect. They have shown that the 3D gold nanoparticle-graphene-AgNHs array hybrid structure manifests ultrahigh SERS sensitivity with a detection limit of 10⁻¹³ M for Rhodamine 6G (Rh6G) molecules, as well as good reproducibility and stability. On the other hand, GO is chemically treated graphene and has most recently emerged as a potential alternative to graphene [17,18]. GO is considered as a promising material for biological applications owing to its excellent aqueous processing ability, amphiphilicity, surface functionalizing ability, fluorescence quenching ability [19,20] and plasmonic properties [21]. Recently, Tabassum et al. [21,22] have used the enhanced gas adsorption ability of GO and the strong light-matter interaction, at the surface of lithographic plasmonic nanoposts, as an optical gas sensor. Their plasmonic nanostructures were composed of an array of nanoposts with Au disks at the top and perforated nanoholes in an Au thin film at the bottom. The optical response (reflectance) of the plasmonic nanostructure was altered due to absorbing different concentrations of ethanol, water vapor and ethylene in presence of GO coating [21,22]. In addition to conventional procedures, the interference lithography (IL) technique [23,24] is another convenient approach to fabricate AuNHs and AuNSs array for application in optical studies.

Therefore, in order to explore the optical and biosensor properties of graphene oxide, it is important and necessary to identify the number of GO layers, when deposited on different substrates, mainly on the surface of AuNHs and AuNSs arrays. In this type of substrates the optical properties are enhanced. It is worth to emphasize that the few studies that show and identify the number of GO layers are performed on silicon substrates [25–27] and not on nanostructured substrates, mainly, manufactured by the interferometric lithography technique.

In this work, Raman spectroscopy is used to make a local analysis (in a specific circular region of ~3.0 μm²) of GO flakes of different thicknesses, and to show its potential applications to identify, in a very reliable way, the number of GO layers (n_GO) on nanostructured periodic substrates generated by interferometric lithography technique. On the other hand, optical reflectance measurements were performed, taking into account a greater area than for Raman measurements, and revealed the generation of plasmons in the AuNSs/GO and AuNHs/GO samples. Results revealed that this kind of system has potential applications as biosensing devices, based on visible fluorescent radiation.

2. Samples and experimental details

2.1 Nano-structures fabrication

For the fabrication of the two types of metal nanostructures, high aspect ratio photoresist (1D and 2D) structures were patterned on glass substrates using interference lithography technique [28,29] followed by thermal deposition of Au and finally a lift-off process to remove the photoresist. A Lloyd-mirror interferometer setup was chosen to generate the photoresist template, due to its optical simplicity and low cost [30]. Details of the experimental process can be found in Menezes et al. [31]. Briefly, a spatially filtered laser beam (458 nm) impinges onto a stage consisting of a highly reflective mirror perpendicularly attached to a substrate holder. One portion of the wavefront is reflected by the mirror toward the sample, while the other portion is directly incident on the substrate. The interference pattern is generated by the superimposition of these two beams. The sample, coated with photosensitive material, is positioned in the interference region. A positive 500 nm thick photoresist layer (SC-1827, Room and Haas) was exposed twice to the same interference fringe pattern. The sample was rotated by 90° for the second exposure to produce the two-dimensional template. After the two exposures, with a dose of 200 mJ/cm² each, the photoresist was developed for 60 seconds with the Microposit 351 Developer (Rohm and Haas, 1:4 diluted in water). The photoresist template was coated with an 80 nm gold film by e-beam evaporation, and then the photoresist was removed using acetone (lift-off process). The gold nanohole arrays surface has a surface area of around 2 cm². In Fig. 1 are represented schematically, not in scale, a top-view and cross section (in the region indicated by the red dash line) of the three used substrates. The geometrical parameters of the structures were determined by AFM and are similar to the projected values: a = b = 275 nm, t = 80 nm, d = 250 nm, D = 300 nm. A 1μm layer of SU-8 photoresist was used between the glass and Au to improve the adhesion.

 

Fig. 1 Schematic diagram of the Au (a), AuNSs (b) and AuNHs (c) substrates. The right side images are the cross section for each substrate in the red dash line. The parameter values are: t = 80 nm, a = b = 275 nm, d = 250 nm, D = 300 nm.

Download Full Size | PDF

2.2 Preparation, deposition and topography characterization of GO layers

The GO layers were deposited on Au, AuNSs and AuNHs substrates by spin coating method. A controlled volume of GO dispersed in H₂O (provided by Sigma-Aldrich, product code: 763705-100L) was diluted from 2.0 mg/mL to 0.5 mg/mL adding deionized water and then sonicated for 5 minutes at room temperature. The GO suspension was dropped to cover the entire substrates (Au, AuNSs and AuNHs) and let it stand for 5 min and finally turn-on the spinner at 500 RPM for 30 seconds.

Figure 2 shows the AFM images of the substrates before (a, c, e) and after (b, d, f) GO layers deposition. The images were obtained from an area with a high concentration of GO layers. It is observed that the GO layers are distributed in different ways on each substrate. Both nanostructured substrates have the same periodicity, P = 550 nm. The scale bar for all images is equal to 2 μm. The inset between the Figs. 2(c) and 2(d) shows a region with a partial deposition of GO on AuNSs.

 

Fig. 2 AFM images of the Au (a), AuNSs (c) and AuNHs (e) substrates and GO layers deposited on Au (b), AuNSs (d) and AuNHs (e) substrates. The inset between (c) and (d) shows a region of the AuNSs with a partial deposition of GO layers. The scale bar in all images is 2 μm.

Download Full Size | PDF

2.3 Raman spectroscopy and optical reflectance

The Raman spectra were acquired using the Horiba Raman Spectrometer with 600 lines/mm grating, a short integration time of 3.0 s and a laser with power of 0.12 mW and λ = 532 nm to excite the samples. The 100x objective lens (NA = 0.9) was used, which provided a spot size of approximately 2 μm, allowing in this way to obtain information of circular regions of GO layers of ~3 μm². Considering the incident beam diameter and focal length of the used objective lens, it was estimated that the incident beam angle is dispersed within ± 2°, therefore, it is possible to consider that the incidence angle was normal. On the other hand, the optical reflectance measurements were performed using a modular spectrometer Ocean Optics model FLAME-S-XR1 in a scanning range of 250 to 850 nm, and the light beam with a spot size of ~15 mm². This means that the reflectance spectrum has a contribution of the gold film, GO flakes of various thicknesses and SU8 photoresist.

3. Results and discussions

3.1 Raman spectroscopy measurements

Figures 3(a)-3(c) presents the optical images of GO layers deposited by spin coating method on Au, AuNSs and AuNHs substrates, respectively. Raman spectra for these samples are shown in the Figs. 3(d)-3(f). For each sample it was chosen five regions of different optical contrast, which are likely due to different thicknesses and are identified with the numbers {1}, {2},…, {5} in ascending order with the number of GO layers. Thus, {1} corresponds to the thinner GO layers (~1 - 2 layers) and {5} corresponds to the thicker GO layers (bulk).

 

Fig. 3 Optical image (a, b, c) and Raman spectra (d, e, f) of GO layers on the surface of Au (d), AuNSs (e) and AuNHs (f) substrates. The numbers {1}, {2}, {3}, {4} and {5}, indicated in the optical images, are in correspondence with the numbers in the Raman spectra. These numbers were chosen in increasing order to optical contrast. All spectra were obtained with laser power of 0.12 mW.

Download Full Size | PDF

The Raman spectra show an increasing intensity as the number of GO layers increase. The two more intense peaks correspond to G and D peaks. The very intense D peak is expected for this type of systems due to the several sources of structural defects, as diverse functional groups in the structure and disordered stacking of individual layers. The bands 2D at ~2700 cm⁻¹ is a second order Raman process originating from the in-plane breathing-like of the carbon rings; D + G at ~2940 cm⁻¹ and 2D’ at ~3240 cm⁻¹ (two phonon process and the overtone of the D^’ at ~1620 cm⁻¹), which is a two phonon process and the overtone of the D^’ at ~1620 cm⁻¹ are significantly weaker than the D and G bands [32]. Similar Raman spectra were observed on GO layers when deposited on the surface of SiO₂ substrates [25,27].

It is worth to emphasize that in spite of using a short integration time of 3 s, it was not possible to avoid the effect of laser-induced heating, because the GO layers were burnt and/or partially converted in rGO (see black regions in Figs. 3(b) and 3(c)). This burn was also observed, in another samples, even using a lower laser power of 0.01 mW (not shown here), but the Raman spectra had the same behavior. These results are in accordance with the literature, for instance, Huaping et al. [25] have reported that in using a laser power of 0.1 mW and integration times of 3 s and 0.5 s, the GO layers burnt during the acquisition of its Raman spectrum [25].

For each sample, the relative intensity ratio of the D and G bands (I_D/I_G) and the full width half maximum of the G peak (FWHM(G)) were analyzed as a function of the G peak intensity (I_G). The obtained results are shown in Figs. 4(a) and 4(b). It was chosen I_G as the horizontal axis because this parameter is directly dependent of the thickness of the analyzed GO layers. Additionally, Fig. 4(a) shows that for Au substrate the I_D/I_G ratio decrease slowly with the number of GO layers, however, for the nanostructured substrates, AuNSs and AuNHs, the I_D/I_G values increase as the number of GO layers increase. The obtained values, considering from the thinner to thicker GO layers, are from 1.89 to 1.78 for Au, from 0.77 to 1.79 for AuNSs and from 1.21 to 1.83 for AuNHs substrates. Similarly, from Fig. 4(b) it is possible to observe that for AuNHs and AuNSs substrates, the FWHM(G) parameter decreases as the number of GO layers increases. For AuNSs substrate from 95.8 (thinner GO) to 91 cm⁻¹ (thicker GO) and for the AuNHs substrate presented an even greater variation, from 148.6 cm⁻¹ to 91.4 cm⁻¹, whereas for Au substrate remains almost constant (from 90 cm⁻¹ to 90.8 cm⁻¹). Moreover, it is possible to note that the plasmonic substrate effect on the Raman spectra is negligible for bulk GO, that is, for the three substrates the I_D/I_G and FWHM(G) parameters tend to the same value, (I_D/I_G)_Bulk-GO ~1.8 and FWHM(G)_Bulk-GO ~91 cm⁻¹, in complete concordance with bulk-GO reported in others works [33]. Therefore, from these results, we can to see that the effect of gold nanostructures substrates is decrease (increase) the I_D/I_G (FWHM(G)) parameter when compared with the parameter on the surface of a flat gold film. This comparison reveals that the I_D/I_G ratio is more sensitive to the number of GO layers on AuNSs substrates, while the FWHM(G) parameter presents a greater sensitivity to the number of GO layers on AuNHs substrates as shown in Figs. 4(a) and 4(b).

 

Fig. 4 Dependence of the intensity ratio I_D/I_G (a) and FWHM(G) (b) of the Raman peaks of GO over the Au, AuNSs and AuNHs substrates, with respect to I_G.

Download Full Size | PDF

It is well-established that the I_D/I_G ratio is a good parameter for estimating the degree of graphitization of carbonaceous materials, as it is normally used for measuring the amount of defects. However, in the Raman spectrum of graphene oxide, the presence of the D′ (~1600-1620 cm⁻¹) and D* (~1150-1200 cm⁻¹) bands influences not only the I_D/I_G ratio but also its position and FWHM (G). We can argue that the I_D/I_G ratio is lower for thinner GO layers (~1 - 2 layers) on AuNSs and AuNHs substrates than on non-nanostructured Au substrate, due to the evolution of the D and D′ bands produced by disorder effect. According to Alice et al. [34] the D′ peak is present in all defective graphene and is therefore attractive as a measure of quality. However, due to the superposition of the G and D′ modes, it is impractical to measure directly the position or intensity of the D′ mode by Raman spectroscopy. Lopez et al. [35] have interpreted the evolution of the D′ band with structural defects in terms of the double-resonance mechanism for defective graphene. According to this mechanism, an increase in the number of defects affects the electron lifetime, decreasing the band intensities (I_D, I_D′ or I_2D). Moreover, the intensities I_D and I_D′ are also proportional to the defect concentration and susceptible to the types of defects, grain boundaries, vacancies, or sp³ hybridization [36]. Consequently, we expect that for thinner GO layers the AuNSs behave like line-defects and the AuNHs as point-defects, as can be seen in Figs. 2(d) and 2(f).

These defects would be responsible for the broadening of the D, G and D′ bands in thinner GO layers, therefore, decreasing the I_D/I_G ratio for AuNSs ~0.8 and for AuNHs ~1.2 (during this broadening I_D ≤ I_G and increase of D′ due to disorder effect). On the other hand, as the Au substrate is free of line and point defects the broadening is less than in AuNSs and AuNHs substrates, then, the I_D/I_G ratio increase to ~1.9 (during this broadening I_D > I_G and decrease of the D′ band due to disorder effect). It is interesting to note that for thicker GO layers the I_D/I_G ratio and FWHM(G) are the same in Au, AuNSs and AuNHs substrates, indicating that AuNSs and AuNHs behave as line and point defects only for thinner GO layers. Patrick et al. [37] have shown that the presence of point and line defects (grain boundaries) produce in graphene layers a broadening of the D and G bands, as in our samples. At the same way, Lopez et al. [35] have reported an I_D/I_G ratio of ~1.8 for GO layers synthesized by oxidation of GANF carbon nanofibers (non-graphitized), on a flat substrate, for a Csp² percentage of 40%. Additionally, they have shown for other graphene oxide materials that the I_D/I_G ratio decrease as the Csp² percentage increase, which means that the disorder decrease, as the sp² percentage increase.

The traditional methods to determine number of graphene layers ($n_G$) involve to analyze the G band position, FWHM(G), format of the 2D band and the I_2D/I_G ratio [38–40]. But, most of these parameters are also sensible to strain, doping or defects [41–44] and the GO can be consider a system of graphene in which was insert a lot of structural defects and strain. In this sense, the strong dependence of the parameters I_D/I_G and FWHM(G) of the Raman spectra of GO, previously deposited on AuNSs and AuNHs substrates, with the thickness of GO layers can be harnessed as a new methodology to determine with good precision the number of GO layers ($n_{GO}$) over AuNSs and AuNHs substrates. It would be very useful in manufacturing of plasmonic devices based on GO or rGO layers where the number of layers is an important parameter to consider [45].

3.2 Optical reflectance measurements

In order to investigate the generation of plasmons on 1D and 2D gold nanostructures, bare and covered with GO layers, optical reflectance spectra were obtained for the Au, AuNSs and AuNHs substrates, before and after GO layers deposition, and these spectra are shown in Figs. 5(a)-5(c). The Fig. 5(a) shows the reflectance spectrum of Au substrate. The spectrum present a characteristic format of flat shiny gold substrates [46] with an abrupt increasing of the reflectance from 480 to 600 nm. It is also shown the spectrum for Au substrate after GO deposition, Au/GO, which presents a slightly lower reflectance due to the absorption of GO layers. These two spectra are used as reference to compare with the spectra obtained using the nanostructured substrates.

 

Fig. 5 Reflectance spectra for samples of Au and Au/GO (a), AuNSs and AuNSs/GO (b) and AuNHs and AuNHs/GO (c).

Download Full Size | PDF

Figure 5(b) shows the reflectance spectra for the 1D nanostructured substrates before (AuNSs) and after (AuNSs/GO) the GO deposition. For AuNSs substrate it is observed the increasing of reflectance from 480 to 600 nm, as shown for the Au substrate. It is also observed a dip at ~639 nm, which corresponds to absorption due to complex modes originates from the coupling of localized surface plasmons (LSPs) and surface plasmon polaritons (SPPs) [47,48]. These modes are commonly named of SPP–Bloch waves (BW-SPP) because are SPP modes generated in a periodic structure and can be expressed by Bloch wave functions [49,50]. In general, the position of this dip depend of the periodicity of array and dielectric constants of metal and dielectric medium [51], and when the dielectric constant of the surrounding medium is slightly altered after to GO deposition on AuNSs substrate, it produce a widening and a slight upshift (Δλ ~7 nm) of the dip.

Other effect produced after to GO deposition is the increase of reflectance in the region from 500 to 600 nm, evidenced by the broad peak at λ_F = 540 nm as shown in Fig. 5(b). This peak is generated by two fluorescent effects, in which, the fluorophores contained on GO are randomly distributed above and below of GO layers. These fluorophores are formed jointly by the non-oxidized carbon regions (-C = C-) and the boundary of oxidized carbon regions (where the functionalized groups C-O, C = O and O = C-OH participate [52]). The first mechanism of fluorescence is known as Förster resonance energy transfer (FRET), in which the radiative process is produced by an energy transfer between the LSPs and fluorophores. This process occurs with efficiency ~100% for fluorophores very close (d_separation < 8 nm [7]) to LSPs produced at the edges of the AuNSs. The second mechanism involves the Purcell effect [7], in which the resonant cavity formed by AuNSs and GO layers amplifies the radiation emitted by the fluorophores inside the cavity and further away of the edges of AuNSs (8 nm < d_separation < 137.5 nm, for width of nanoslits equal to 275 nm) [7,53].

On the other hand, the Fig. 5(c) shows the reflectance spectra for the 2D nanostructured substrates, before (AuNHs) and after (AuNHs/GO) deposition of GO. It is also observed the characteristic variation of reflectance for Au substrates from 480 to 600 nm. The reflectance spectrum for the AuNHs substrate presents two dips at 643 nm and 738 nm, which correspond to absorption by BW-SPP modes with energies 1.93 eV and 1.68 eV, respectively. For AuNHs substrates after deposition of GO, AuNHs/GO sample, is only observed one wider dip due to BW-SPP modes at ~646 nm (1.92 eV). A fluorescence peak is also originated after to GO deposition on AuNHs substrate, more intense that the obtained for AuNSs/GO sample and positioned at λ_F = 468 nm. The origin of this radiative radiation is also ascribed to the same fluorescence mechanisms as in the AuNSs.

The mechanism of interaction between fluorophores and LSPs generated in nanostructured substrates can be used in optical sensors, to detect the concentration of several types of fluorescent particles doped with some molecule which is desired to detect. In addition, since optical reflectance is a relatively simple detection technique, this would be an interesting method to project an optical biosensor taking advantage of the high intensity reflected with certain wavelengths.

4. Conclusions

In summary, in this paper we report a strong dependence of the parameters I_D/I_G and FWHM(G) of the Raman spectra of GO layers deposited on AuNSs and AuNHs substrates, fabricated by the interference lithography technique, in relation to number of GO layers (in a very rough approximation, $n_{GO}\le 10$). The optical reflectance revealed the generation of plasmons in the AuNSs/GO and AuNHs/GO samples, which absorb at specific wavelengths and produce visible fluorescent radiation by two different resonant mechanisms. Future work includes obtaining a quantitative relation of the I_D/I_G and FWHM(G) parameters with $n_{GO}$, depositing GO on nanostructured substrates by drop-casting and/or by exfoliation. Systems like these present high potential for detecting molecules by absorption of fluorescent radiation, but a prior adjustment of the geometric parameters of substrates to produce fluorescent radiation at suitable wavelengths is required.