1. Introduction

Surface-enhanced Raman scattering (SERS) as a promising spectral technique has found widespread applications [1–5], due to its intriguing feature to produce significant enhancement of Raman signal for the ultrasensitive detection of analytes. It is now widely recognized that the SERS effect is mainly based on the strongly enhanced electromagnetic field near the metallic nanostructures due to localized surface plasmon resonance (LSPR), especially, occurring in nanoscale gaps (so-called “hotspots”) narrower than 10 nm between the adjacent plasmonic nanostructures [6]. By taking full advantage of the hotspots, single-molecule detection sensitivity can be achieved [7–11]. Thus, high-density and homogeneous hotspots are prerequisite for reliable SERS applications [12]. Many fabrication methods such as electron beam lithography, nanosphere lithography, focused ion-beam patterning, two-photon polymerization lithography, and self-assembly, have been employed to achieve such controllable nanogaps [13–17]. Although high-performance SERS substrates can be achieved by these techniques, for commercial applications, cost-effective and reliable fabrication of wafer-scale SERS substrates with reproducible hotspots over the entire area remains a challenge.

Semicontinuous metal films (i.e., metallic nanoisland films) with interstitial nanogaps have been demonstrated as high-quality SERS substrates both theoretically and experimentally [18–21]. The advantages of semicontinuous metal films include the facile fabrication process, controllable LSPR wavelength, and robust SERS performance. A seed-mediated overgrown approach has been employed to fabricate centimeter-scale gold nanoislands; however, large-scale homogeneity in distribution remains a challenge for this approach [22]. While solid-state dewetting of metal films allows for the fabrication of metal nanoislands arrays over a large area, the balance between small interstitial distance and large nanoisland size is still challenging [23]. Compared to the above-mentioned techniques, physical deposition methods such as electron-beam evaporation [24], DC sputtering [25], and thermal evaporation [26,27], are easier to fabricate large-area metallic nanoisland films. However, all the above-mentioned physical deposition techniques usually require high temperature annealing process, and simultaneously remain a challenge to achieve the concurrent fabrication of large nanoisland size and small interstitial nanogaps via one-step silver deposition process. Beside these techniques, one of the most versatile techniques is pulsed laser deposition technique (PLD) [28,29]. For nanoislands films fabrication, the PLD’s advantages include high homogeneity, efficient stoichiometric transfer, adherent films to the substrate, and precise controllability of the film growth [30]. Even though several studies have been reported on PLD fabrication of plasmonic metallic films, there are only a few studies discussing the design and fabrication of the controllable nanoisland film morphology. Until now, reliable fabrication of wafer-scale SERS substrates of nanogap-rich metal nanoisland films with high sensitivity and reproducibility through a facile and cost-effective route remains a challenge.

In our previous work, we reported that aluminum cloud-like nanostructures arrays can be homogeneously decorated with spherical silver nanoparticles through LMBE technique to generate 3D SERS platforms, wherein we demonstrated the reproducibility of 13.1% in terms of RSD for the optimized 3D-structured SERS substrate [31]. Here, we fabricated nanogap-rich silver nanoislands films on silicon at wafer level (6×6 cm²) for SERS substrates via one-step deposition process by LMBE technique. By controlling the deposition time, enlarged silver nanoislands with abundant and homogeneous ∼5 nm interstitial nanogaps can be prepared. The adsorption of the optimized silver nanoisland film was higher than 40% in the range from 400 to 800 nm. With optimized deposition parameters, the produced SERS substrates exhibited high AEF as much as 1.17×10⁵, excellent reproducibility with RSD of 7.76%, and good stability of less than 10% in air for 21 days. In addition, the electromagnetic field distribution of the optimized silver nanoisland film was simulated and illustrated using the software COMSOL Multiphysics. Furthermore, the SERS activity was tested using malachite green. This simple method can be readily employed to produce wafer-scale SERS substrates of silver nanoisland films with high sensitivity and good reproducibility for practical applications.

2. Experiments

2.1 Materials and instruments

The silicon wafers [n-type (100)] with a native oxide layer was purchased from Hangzhou Jingbo Technology Co. Ltd., China. Silver target (99.99%) was purchased from Zhongnuo Advanced Material Technology Co. Ltd., China. Rhodamine 6G (R6G) was purchased from Sigma-Aldrich. Malachite green (MG) was obtained from Aladdin Ltd. in Shanghai.

The surface morphology and element of the silver nanoisland films were characterized using a JEOL JSM-6380 field emission scanning electron microscopy. The surface topography of the optimized silver nanoisland film was analyzed using a Bruker Dimension Icon AFM. The UV-vis reflectance spectra of the silver nanoisland films were measured by the HITACHI U-3310 spectrophotometer. Silver nanoisland films deposited on the silicon wafer were achieved by using the laser molecular beam epitaxy (LMBE) growth system (model LMBE 450, SKY Company, China). The Raman spectra of analytes were measured with the Renishaw (Gloucestershire, UK) RM-2000 Laser Raman Spectrometer.

2.2 Preparation of SERS substrates

The silicon wafers were firstly cleaned in acetone, ethanol and deionized water successively, each for 10 min, to remove residues on the surface, followed by drying in air and segmenting into 2×2 cm² or 6×6 cm² (the optimized SERS substrate). Then, the as-prepared silicon wafers were decorated with silver nanoisland films at different deposition times (15, 20, 25, and 30 min) in the LMBE growth system. During the deposition process, the pulse energy of the laser beam with wavelength 248 nm was up to 170 mJ, the laser pulse repetition rate was 5 Hz, and the vacuum was 2.0 × 10⁻⁴ Pa. In our experiments, high-purity silver targets were sputtered under the same conditions except with different deposition times.

2.3 Raman measurements

For SERS measurement, the Raman signals of R6G and MG were obtained after the 5 μL droplet of aqueous solution on the different silver nanoisland films evaporated naturally, and were measured at room temperature under the same conditions. The SERS measurements were performed using a Renishaw (Gloucestershire, UK) RM-2000 Laser Raman Spectrometer based on a 785 nm laser and a 20× objective lens. The laser power at the sample was about 5 mW and the laser spot was ∼ 2 μm. The integration time was 10 s.

3. Results and discussions

3.1 Controllable size-gap of the silver nanoisland films

The SERS substrates of nanogap-rich silver nanoislands on silicon were prepared through LMBE technique. It is well known that the enlarged metal nanoislands with small nanogaps apparently increase the number of electromagnetic hotspots and simultaneously support the tunability of plasmon resonance wavelength [23]. Here, the sizes and separation distances of the silver nanoislands can be moderately controlled by the deposition time. The silver film deposited on the silicon substrates using this technique grows according to Volmer-Weber growth mode [32]. Small isolated silver nanoparticles are firstly formed on the silicon substrate, followed by the coalescence of small silver nanoparticles via prolonging the deposition time, constructing various sizes of large silver nanoparticles and their aggregates. These results are displayed in Fig. 1, which shows the scanning electron microscopy (SEM) images of the silver nanoisland films obtained at different deposition times ranging from 15 to 30 min with an interval of 5 min. It can be clearly observed that all the silver films consist of plenty of nanoislands and the morphology changes with the deposition time. The silver films at low deposition time (Fig. 1(a)) do not form continuous layer, instead, they induce the formation of small, isolated nanoislands. As the deposition time increases the nanoislands become bigger (Fig. 1(b) and 1(c)), producing nanoislands of a variety of shapes and sizes. For higher deposition time, the sizes of the silver nanoislands are further increased (Fig. 1(d)), as a result of coalescence of adjacent nanoislands, but they were not large enough to form a smooth film. Thus, it can be anticipated that the larger silver nanoislands may support a more intensive localized field due to plasmonic excitation, achieving even stronger SERS enhancement [33].

 

Fig. 1. SEM images of silver nanoisland films with different deposition times: (a) 15 min, (b) 20 min, (c) 25 min, and (d) 30 min.

Download Full Size | PDF

Furthermore, using ImageJ software, the average nanogap of the silver nanoisland films with different deposition times was calculated. For statistical analysis, the nanogap size distribution was estimated based on 50 independent measurements on each sample. As shown in Fig. 2(a), when the deposition time was 15 min, the average nanogap between adjacent nanoislands was approximately 7.613 ± 0.160 nm. As the deposition time increased, larger and irregular silver nanostructures were formed and the average nanogap tuned to 6.464 ± 0.140 nm (20 min), 5.323 ± 0.212 nm (25 min), 4.594 ± 0.075 nm (30 min). These results clearly demonstrated that controlling of deposition times allows for the fabrication of enlarged silver nanoislands with ∼5 nm interstitial gaps. Under illumination of a laser source, these silver nanoisland films can provide large numbers of electromagnetic hotspots, which are located at the tips [34], and nanogaps between the adjacent nanoislands [35], because of the electrostatic lightning-rod effect and plasmonic coupling effect [36,37]. In order to quantify the nanogap/silver ratio of each film, we used Matlab to transform the SEM images to binary images and then determined the ratio of nanogap area over silver covered area. The images generated by the binary reconstruction method are shown in Fig. 2(b)-(e), where the light grey regions are the silver nanoislands and the dark grey regions are interstitial space in between. By increasing the deposition time from 15 to 20, 25, and 30 min, the corresponding area ratio decreased from 50.58% to 46.60%, 42.06%, 37.88%, respectively. Therefore, the smaller deposition time favors a higher density of hotspots sites. Thus, a compromise is required. Although an increase in deposition time results in reducing the number of hotspots per unit area, the plasmonic excitation of larger nanoislands is stronger. Therefore, there exists an optimized combination of the size of silver nanoislands and density of hotspots.

 

Fig. 2. (a) The relationship between the nanogaps with the deposition times. Binarized SEM images of silver nanoisland films with different deposition times: (b) 15 min, (c) 20 min, (d) 25 min, and (e) 30 min. (f) Reflectance spectra of fabricated different silver nanoisland films.

Download Full Size | PDF

In order to investigate the LSP effect in these SERS substrates, the reflectance spectra of four different silver nanoisland films were recorded by UV-vis spectroscopy, as shown in Fig. 2(f). One can see that all the silver nanoisland films exhibit similar shapes of the reflectance spectra with one isolated peak centered at about 350 nm and a lower reflectance in the range of 400 to 800 nm. It should be noted that all the silver nanoisland films have absorption of higher than 40% in a broad spectral range from 400 to 800 nm, wide enough to cover both the excitation frequency and the Raman vibrational spectrum [38,39]. The broad adsorption of our SERS substrates can be explained by plasmon resonance between silver nanoisland structures by the dipolar and multipolar coupling mechanism.

3.2 SERS performance of the silver nanoisland films

To investigate the SERS enhancement of these substrates, rhodamine 6G, a typical analyte for SERS activity evaluation, was chosen as the probe. Figure 3(a) shows the Raman spectra of 5 μL 10⁻⁵ M R6G aqueous solution adsorbed on four different silver nanoisland films. From the signal profiles, one can see that all the silver nanoisland films exhibit strong Raman signals of R6G molecule with little background noises. The characteristic peaks of the R6G fingerprint can be clearly observed, e.g., at 611 cm⁻¹, 780 cm⁻¹, 1190 cm⁻¹, 1312 cm⁻¹, 1365 cm⁻¹ and 1510 cm⁻¹. In addition, the SERS intensity first goes up with increasing deposition time, and then drops after the deposition time is larger than 25 min. Specifically, we plot the Raman intensity at a 1510 cm⁻¹ Raman shift as a function of the deposition time, as shown in the inset of Fig. 3(a). The result indicates that the silver nanoisland film with deposition time of 25 min yields the strongest SERS enhancement, whose intensity at the 1510 cm⁻¹ peak is approximately 3.18 times higher than that with deposition time of 15 min. According to the SEM images (Fig. 1) and binarized images (Fig. 2), that the optimized silver nanoisland film (25 min) exhibits the maximum SERS intensity can be explained as follows: at the deposition time of 25 min, larger silver nanoislands induce stronger localized field than smaller ones obtained at deposition times less than 25 min, and the number of small nanogaps (∼ 5 nm) producing hotspots is larger than that obtained at the deposition times more than 25 min.

 

Fig. 3. (a) Raman spectra of 10⁻⁵ M R6G solution collected with four different silver nanoisland films. The inset shows the intensity at the 1510 cm⁻¹ peak varies with the deposition time. (b) SERS spectra of R6G solution with different concentrations ranging from 10⁻⁸ to 10⁻⁴ M collected with the 25-min-depostited silver nanoisland film. (c) The correlation between the SERS intensity of R6G at 1510 cm⁻¹ and the R6G concentration. The error bars represent the standard deviation, and the number of measurements, n=5. (d) The Raman spectra of 10⁻⁷ M R6G solution dropped on the 25-min-depostited silver nanoisland film (red line), and 10⁻³ M R6G solution dropped on silicon (green line).

Download Full Size | PDF

To evaluate the sensitivity of the optimized silver nanoisland film, the SERS spectra of R6G solution with different concentrations ranging from 10⁻⁴ to 10⁻⁸ M were measured, as depicted in Fig. 3(b). The SERS signal weakens with decreasing R6G concentration, but the characteristic peaks of the R6G molecule can be still observable even at concentration as low as 10⁻⁸ M. We regard this value as the practical limit of detection (LOD) of the optimized silver nanoisland film. Figure 3(c) shows the linear fitting correlation of the SERS peak intensity at 1510 cm⁻¹ with R6G concentration, which can be expressed through the following equation: logI=0.32logC+6.07, where I is the SERS intensity and C is the R6G concentration in M. The Raman intensity and the logarithmic concentration demonstrate a very good linear correlation coefficient of 0.985. The analytical enhancement factor (AEF) for the silver nanoisland film substrate was calculated using the equation: AEF= (I_SERS/ I_norm)×(C_norm/C_SERS) [40], where the I_SERS and I_norm are the peak intensities at 1510 cm⁻¹ in the SERS and norm Raman spectra, C_SERS and C_norm are the concentrations of R6G aqueous solution dropped on the optimized SERS substrate and silicon substrate, as shown in Fig. 3(d). From a practical standpoint, AEF can provide a very conservative estimate of the enhancement factor, because it is based on actual measurements of the normal Raman signal and SERS signal for an initial and diluted analytes rather than on intermediated assumptions and calculations [22]. The silicon substrate exhibits very weak Raman signal for the 10⁻³ M R6G aqueous solution, with the intensity of 955 arbitrary units at the 1510 cm⁻¹ peak; however, the optimized SERS substrate displays strong Raman signal for the 10⁻⁷ M R6G aqueous solution, with the intensity of 11171 arbitrary units at 1510 cm⁻¹ peak. Therefore, the EF of the optimized SERS substrate structures was estimated to be 1.17×10⁵ for the 10⁻⁷ M R6G solution in the SERS experiment and 10⁻³ M R6G solution in the norm Raman experiment. These results indicate the potential of the optimized silver nanoisland film for quantitative and sensitive SERS analysis.

Figure 4(a) shows the Raman spectra of 10⁻⁵ M R6G collected at 60 random spots in four different regions of the SERS substrate, which gradually move from the middle to the edge with an interval of ∼0.9 cm. In each region, we have randomly obtained 15 SERS spectra. One can see that all the Raman spectral curves in various regions display similar intensities for all the characteristic peaks of R6G molecules. The peak intensity at 1510 cm⁻¹ derived from the spectra is presented in the Fig. 4(b). According to the formulas in the literature [41], the corresponding RSD is calculated to be about 7.76%, demonstrating reasonable spot-to-spot reproducibility across the entire substrate. To further illustrate the point-to-point reproducibility of this method, a Raman mapping experiment was carried out at a concentration of 10⁻⁵ M R6G in the central and marginal regions of the silver nanoisland film substrate, as displayed in the Fig. 4(c) and (d). It should be noted that the SERS mapping was achieved by a Raman spectrometer (inVia, Renishaw) based on a 785 nm laser and a 50× objective lens. Eighty-one spectra were collected at a laser power of 3.25 mW and integration time of 10 s. Typically, the RSDs of the intensities of the main vibration at 1510 cm⁻¹ in the central and marginal regions were calculated to be 8.64% and 9.57%, respectively. Although this value is larger than that of less than 5% [42,43], our observed SERS intensity variation (less than 20%) is acceptable for SERS quantitative studies [44]. Table 1 displays the comparison of RSD and EF for silver nanoisland film with previously reported works. The morphologies of semicontinuous metal films affect the RSD and EF of SERS detection [21,24–27], and the enlarged silver nanoislands with abundant and high-density nanogaps as SERS substrates in our work can simultaneously achieve lower RSD and better SERS enhancement, and thus significantly improve the consistency and intensity of the Raman signal. From the energy-dispersive spectrometry (EDS) line and mapping images of silver nanoisland film shown in Fig. 4(e), we can see that silver element is uniformly distributed on the silicon substrate, which is indirectly attributed to the formation of homogeneous silver nanoisland structures. In addition, the fabrication of the SERS substrate based on silver nanoisland film across a ∼6 cm × 6 cm silicon wafer is given in Fig. 4(f). From the above discussions, the excellent spot-to-spot reproducibility of our wafer-scale SERS substrate demonstrates that the nanogap-rich silver nanoisland film is capable of affordable practical SERS analysis.

 

Fig. 4. (a) The Raman spectra of R6G molecules collected at 60 random sites from four regions of the optimized SERS substrate. (b) The histogram characterization of the intensities at the 1510 cm⁻¹ peak derived from the Raman spectra. (c) and (d) are Raman mapping images of R6G at 1510 cm⁻¹ peak collected in the central and marginal region of the silver nanoisland film substrate, respectively. (e) The EDS line and mapping images of the optimized SERS substrate. (f) The photograph of the SERS substrate (6×6 cm²).

Download Full Size | PDF

Table 1. Comparing the RSD and EF value of silver nanoisland films prepared by different deposition techniques.

View Table

Furthermore, the stability of the SERS substrate is an important parameter for achieving routine analysis. Therefore, in order to investigate the SERS stability of the silver nanoisland film substrate, a set of comparative experiments were carried out under ambient conditions using a Raman spectrometer (inVia, Renishaw). Figure 5(a) shows the SERS spectra of 10⁻⁵ M R6G obtained from the same optimized silver nanoisland film substrate stored for 7, 14 and 21 days. To make the contrast even more clear, the intensity attenuation of the measured R6G SERS signals at peak 1510 cm⁻¹ as a function of the storage time is displayed in Fig. 5(b). Three weeks later, the average intensity of peak at 1510 cm⁻¹ dropped only 10%. Thus, in the short term, using the silver nanoisland film substrate to detect the target molecule, the SERS signal will not change significantly.

 

Fig. 5. (a) The Raman spectra of R6G adsorbed on the optimized SERS substrate after different storage times. (b) Raman intensity of 10⁻⁵ M R6G at 1510 cm⁻¹ as a function of the storage times (the error bars were calculated based on six independent measurements).

Download Full Size | PDF

To determine the SERS intensity enhancement caused by the interstitial nanogaps in the silver nanoisland film, theoretical simulations were performed using the software COMSOL Multiphysics to model and calculate the electric field distribution of the optimized SERS substrate. In the simulation model, the incident light at λ = 785 nm with x-polarization or y-polarization propagated along the –z direction, and the magnitude was 1 V/m. In contrast to metal-insulator-metal nanostructure model [27], a portion of a SEM image was imported in COMSOL software to build highly consistent geometrical models for the irregular island-like nanostructures. Figure 6(a) shows a typical SEM image of a portion of the optimized SERS substrate (Fig. 1(c)). The thickness of silver nanoisland film was estimated to be about 10 nm for the simulations, which is confirmed using atomic force microscopy (AFM) measurement, as shown in Fig. 6(b). Figure 6(c) and (d) show the calculated electric field distribution when the silver nanoisland film is illuminated by an incident light polarized along the x-axis and y-axis, respectively. As can be seen, the local electric field can be greatly enhanced near the edges of irregular nanostructures, especially in the regions of nanogaps inside the silver nanoisland film that are perpendicular to the axis of polarization of the incident light, due to plasmonic coupling between the adjacent island-like nanostructures. These simulations show that the silver nanoisland film can produce the electromagnetic hotspots, and the maximal field enhancement (E_loc/E₀) was about 229. According to the electromagnetic enhancement theory, the maximum enhancement factor in our model simulation data is 2.75×10⁹.

 

Fig. 6. (a) SEM image of the optimized silver nanoisland film at a deposition time of 25 min. (b) AFM image of the optimized silver nanoisland film at a deposition time of 25 min to visualize dimensions of island-like nanostructures. (c) and (d) are COMSOL simulations of the electric field distribution inside the silver nanoisland film illuminated by an x-polarized plane wave and a y-polarized plane wave at λ = 785 nm, respectively.

Download Full Size | PDF

3.3 SERS detection of malachite green

In order to demonstrate practical applications of the optimized SERS substrate, malachite green was chosen as the analysts. MG is a kind of triphenylmethane dye. Due to its genotoxicity and carcinogenicity [45,46], MG is prohibited in many countries. Figure 7(a) shows the SERS spectra of 5 μL diluted MG solution with concentrations ranging from 10⁻⁴ to 10⁻⁸ M. The prominent peaks of MG were in line with previous studies [47]: 1177 cm⁻¹ (the inplane vibrations of ring C–H), 1219 cm⁻¹ (the rocking of C–H), 1394 cm⁻¹ (the stretching of N-phenyl) and 1616 cm⁻¹ (the stretching of ring C–C). The SERS intensity significantly increases with the increase of MG concentration. For minimal MG concentration, 10⁻⁸ M, the SERS peaks can still be resolved. Therefore, the LOD of malachite green can be roughly as 10⁻⁸ M. The SERS intensity at 1616 cm⁻¹ peak was used for the quantification. As shown in Fig. 7(b), good linearity was obtained with a correlation coefficient of 0.995. Therefore, the silver nanoisland film can be used for sensitive and quantitative detection of malachite green in environmental water.

 

Fig. 7. (a) The Raman spectra of MG with different concentrations obtained from the optimized SERS substrate. (b) The calibration curve of the logarithms of Raman intensity at 1616 cm⁻¹ Raman shift and MG concentration.

Download Full Size | PDF

4. Conclusions

In summary, we have fabricated the wafer-scale and broadband SERS substrate of nanogap-rich silver nanoisland film on silicon through a versatile LMBE technique. By precisely controlling the deposition time, the enlarged silver nanoisland structures consisting of abundant and homogeneous ∼5 nm nanogaps have been prepared. The optimized SERS substrate with high-density hot spots exhibits high AEF of 1.17×10⁵, large-area reproducibility with RSD of 7.76%, and good stability in air for 21 days. Using COMSOL Multiphysics software, we theoretically verify that the electric field is greatly enhanced in the regions of interstitial nanogaps inside the silver nanoisland film. Furthermore, malachite green can be detected down to 10⁻⁸ M using our optimized SERS substrate. The wafer-scale and broadband silver nanoisland films provide many potential applications, including sensitive and quantitative SERS analysis, solar cells and energy harvesting.