1. Introduction

Surface-enhanced Raman scattering (SERS) has been found as an analytical technique with ultra-high sensitivity and high selectivity. SERS can be extensively used in biochemical analysis, medical diagnosis, food safety and environmental monitoring and other fields [1–3]. For the mechanism of SERS enhancement, SERS enhancement is generally considered the result of the combined action of an electromagnetic mechanism (EM) and a chemical mechanism (CM) [4]. EM uses the surface plasmon of the metals to enhance the electromagnetic field, and it is capable of enhancing the Raman signal more than 10⁸ times [5,6]. CM is produced by the charge transfer between the substrate and the molecule adsorbed on its surface, only 10–100 enhancement factors can be provided [7].

Two types of EM play a significant role in Raman enhancement [8]. One is localized surface plasmon (LSP), typically excited in the gap between metal nanoparticles or sharp protrusions (commonly termed the “hot” spot) through conducting collective oscillations of electrons in metal nanoparticles (NPs) [9,10]. The other is surface plasmon polariton (SPP), which is essentially the collective oscillation wave of electrons propagating at the interface between a metal film and a dielectric film. The weak electromagnetic field generated is restricted to the interface between metal and dielectric, which improves the electromagnetic field at the interface [11]. In multilayered metal/dielectric film structures, adjacent SPPs are coupled to each other, and the bulk plasmon polariton (BPP) is generated within multilayer film (MLF) structures, which supports the propagation of interlamellar evanescent wave [12,13]. However, the generation of BPP in the visible light range requires the incident light and SPP to achieve momentum conservation, which causes BPP to be difficult to stimulate and use [14,15]. As indicated by recent research, BPP can be successfully excited and coupled with LSP through specific structural design and material selection to further enhance the hot spot and improve the sensitivity of SERS substrate [16,17]. For instance, Lai et al. developed a hybrid system composed of star-shaped Au/Ag NPs and MLF, which indicated that the MLF more significantly enhanced SERS than the simple metal layer [18]. Li et al. proposed an Al nanoparticle-film system. As revealed by the simulation results, the SPPs in the respective metal layer could be coupled with each other to stimulate a collective response in the system, and the BPP in the MLF further improved the electric field intensity between the NPs [19]. Typical SERS substrates are metal nanostructures or rough metal surface structures. Nevertheless, when the above metal substrate structures bring about significant Raman enhancement, they are generally accompanied by low uniformity and low reproducibility of SERS substrate. For typical SERS substrates composed of metal particles, to achieve significant electric field enhancement and strong SERS effect, the small nanogap of the substrate should be ensured. Furthermore, only a few molecules fall into the nanogap, and they are excited by a strong electric field in tests. Thus, the Raman signal detected at different points on the substrate has varying intensity, and the substrate has poor uniformity. As reported by Fang’s team, less than 1% of molecules could contribute to 70% of SERS signals by detecting Raman signals on SERS substrates of metal NPs [20]. In fact, in the development process of SERS substrate, the problem of metal–molecule contact induced signal variations has become an increasingly important subject. The unfavorable disturbances primarily include carbonization effect on the metal that leads to a strong spectral background, photoinduced damage, metal instability as well as strong catalytic activity [21]. The defects of such metal substrates limit the practical application of SERS analysis technology.

With the continuous development of SERS technology, rough metal substrates can no longer meet higher requirements. Two-dimensional (2D) materials (e.g., graphene) have good application potential in SERS enhancement [22,23]. Graphene has been reported as an ultrathin 2D layered material with many advantages (e.g., great chemical inertness, high light transmittance and biocompatibility, as well as high stability), which indicates that it can be used in SERS substrate [24]. When graphene shows CM-based Raman enhancement by charge transfer between molecules and graphene, extensive studies have been conducted on hybrid nanostructures containing graphene and plasmonic metal components [25,26]. For instance, Qiu et al. developed a sandwich structure of graphene and sandwiched Ag-nanoflowers. They demonstrated that graphene layers could significantly inhibit the oxidation of Ag NPs, and the stability of the structure was improved [27]. Zhang et al. designed a graphene-mediated SERS flexible substrate and conducted the in situ measurement, which found the advantages of graphene in reducing fluorescence background and light pollution [21]. The development of a convenient, environmentally friendly and low-cost method to prepare highly sensitive and reliable graphene/Ag NPs structures has also been reported [28–30]. The application of graphene in SERS substrate above can make up for the defects (e.g., an easy chemical reaction between metal substrate and molecules to be measured and strong spectral background). However, the graphene-metal hybrid structure cannot achieve high Raman enhancement since it has simple nanoparticles. Accordingly, it is necessary to prepare a SERS substrate with high sensitivity, high uniformity, less interference, strong affinity and good stability for developing SERS technology.

This paper developed a composite SERS substrate of monolayer graphene/Ag NPs coupled with multilayer Au/Al₂O₃ film. The MLF was separated by Al₂O₃ because of its high light permeability and elimination of charge disorder in the dielectric environment. Ag NPs were prepared through direct evaporation. Impacted by the surface energy and the atomic rate, a dense layer of Ag NPs could be formed under a small evaporation thickness [31]. The evenly distributed hemispherical nanoparticles could ensure the uniform distribution and effective utilization of hot spots [21]. Moreover, the application of graphene has endowed the composite SERS substrate with several advantages including reduction of fluorescence background and good biocompatibility. The electric field distribution of the composite SERS substrate was simulated using finite element method, the electric field intensity at Au/Al₂O₃ interface of the respective layer was analyzed, which indicated that the BPP was successfully excited and coupled with the LSP. Lastly, the SERS performance of the composite SERS substrate was measured with rhodamine 6G (R6G) and malachite green (MG), with the detection limits of 10⁻¹¹ M and 10⁻⁸ M, respectively, and the substrate had high sensitivity, uniformity and reproducibility. As revealed by the study, the composite SERS substrate had excellent SERS performance and good application potential in the field of low concentration molecular analysis and detection.

2. Experimental section

2.1 Preparation of multilayer film structure of Au/Al₂O₃ (MLF)

The Si wafer was washed with acetone, ethanol and deionized water (DI water) consecutively using an ultrasonic cleaner for 1 h to remove surface contaminants. Subsequently, an 8 nm thick Au film was deposited on the Si wafer by a vacuum thermal evaporation system (0.1 Å/s, current 40 A, 7×10⁻⁵ Pa). After the deposition of the Au film, 3 nm Al film was deposited (0.8 Å/s, current 72 A, 7×10⁻⁵ Pa). The obtained Au/Al film was transferred to a vacuum kettle and then evacuated to 1×10³ Pa, and pure oxygen was introduced for 10 min to oxidize the Al film. Lastly, a layer Au/Al₂O₃ film structure was obtained. By repeating the deposition and oxidation processes above, a 6-layer Au/Al₂O₃ film structure was successfully prepared (Fig. 1).

2.2 Preparation of monolayer graphene/Ag NPs

The graphene/Cu foil prepared with CVD was transferred to the surface of FeCl₃ solution with a concentration of 1 M/L to remove Cu foil. After standing for 4 h, the Cu foil was completely eroded in FeCl₃ aqueous solution, which left graphene floating on the solution surface. Subsequently, it was transferred to the surface of DI water solution using an ultrasonic cleaner and plasma treated slides for 3 h. The above process was repeated 3 times to remove FeCl₃ molecule. The graphene was taken out with thin crystal sheets of pure NaCl and then transferred to the thermal vacuum evaporation system. Subsequently, a 3.5 nm thick Ag (1 Å/s, current 65 A, 7×10⁻⁵ Pa) was deposited on graphene. After the deposition, a dense nano hemispherical particle layer of Ag was formed on the graphene surface, and the monolayer graphene/Ag NPs were successfully prepared.

2.3 Combination of monolayer graphene/Ag NPs and MLF

Thin crystal sheets of pure NaCl with graphene/Ag NPs were slowly placed into the DI water solution. The graphene/Ag NPs were gradually separated from the NaCl thin crystal sheet during the process, and only the graphene/Ag NPs floated on the DI water solution surface. Subsequently, graphene/Ag NPs were transferred to the prepared multilayer Au/Al₂O₃ film, and the composite SERS substrate of monolayer graphene/Ag NPs coupled with multilayer Au/Al₂O₃ film was well prepared.

2.4 Characterization and SERS experiments

The cross-section and surface morphology of the composite structure were observed under the scanning electron microscopy (SEM, Zeiss Gemini Ultra-55). We used atomic force microscopy (AFM) to characterize the thickness of the multilayer structure, and elemental analysis was conducted with energy dispersive spectrometer (EDS). To test the SERS effect of the composite structure, Raman test was performed with probe molecule R6G (dissolved in DI water, concentration 10⁻⁶−10⁻¹⁰ M). The conditions of Raman signal acquisition by Raman spectrometer (Horiba HR Evolution) included 532 nm exciting laser, 0.48 mW laser power, gratings with 600 grooves per 1 mm, ×50 objective lens, 1 µm laser spot, as well as 4 s acquisition time. 5 µL probe molecules were dropped onto the sample, and Raman tests were performed after the sample was dried.

 

Fig. 1. Schematic representation of the processes for preparing SERS substrate.

Download Full Size | PDF

3. Results and discussion

3.1 Structure characterization

SEM images were used to characterize the morphology of the different structures prepared through the experiment. Figure 2(a) and 2(b) present the cross-section and surface of the prepared MLF. To be specific, the brighter layers were Au film, and the darker layers were Al₂O₃ film. There was a clear interface between the two different materials, and the thickness of the film could be on a nanometer scale, which could ensure the propagation of SPP in the MLF. As indicated by the atomic force microscope (AFM) height profile (Fig. 2(c)) of MLF surface, the height profile showed a thickness of 68.1 nm for the MLF, which was nearly consistent with the expected thickness of 66 nm. Figure 2(d) and 2(e) present the cross-section and surface of the prepared composite structure. As indicated by the cross-section diagram, the monolayer graphene/Ag NPs completely covered the MLF. Graphene was tough and flat, and graphene was also used as a “protection” to avoid the oxidation of Ag NPs, which improved the stability of the composite SERS substrate. The morphology and size distribution of the hemispherical particles of silver nanoparticles are illustrated in the surface figure. The distribution of Ag NPs was uniform, and the diameter distribution histogram was calculated according to the SEM images. Their diameters were largely distributed from 25 to 35 nm, with the average diameter close to 30 nm, and the gap between the Ag NPs was mostly 8 nm. The uniform distribution of silver nano hemispherical particle ensured the uniform distribution and utilization efficiency of the hot spot. The arrow marks a tiny break in the graphene, where the Ag NPs were in contrast to the rest of the particles to indicate the integrity of the graphene. Furthermore, the composite structure was analyzed by EDS, as shown in Fig. 2(f), to confirm the composition of elements (e.g., Au, Ag, Al and O), and the uniformity of the Ag NPs, Au films and Al₂O₃ film could be found by observing the color distribution of different elements.

 

Fig. 2. (a) The cross-sectional SEM image of the 6LF. Inset shows the local-magnification SEM image. (b) Top view SEM image of the MLF. (c) Atomic force microscopy (AFM) image of the MLF and the corresponding height profile. (d) The cross-sectional SEM image of the composite SERS substrate. Inset presents the local-magnification SEM image. (e) SEM images and size distribution histograms of Ag NPs. (f) the EDS spectrum of the composite SERS substrate. Inset shows the EDS elemental maps.

Download Full Size | PDF

3.2 Simulation and theory

To deeply understand the SERS performance of the composite structure, the electric field distribution of the structure was simulated by Finite Element Method Simulation. This paper used the structural characteristics observed by the electron microscope to set the simulation parameters. According to Fig. 3(a) of composite structure, multiple silver hemispheres were uniformly adhered below monolayer graphene to simulate the monolayer graphene/Ag NPs structure, Ag NPs had a diameter of 30 nm, and the particle spacing was 8 nm. The incident laser wavelength of 532 nm was used, which was incident by the air above the composite structure, consistent with the actual Raman test. The electric field simulations of three different structures were compared with each other. First, the electric field distribution of graphene/Ag NPs structure was simulated as presented in Fig. 3(b). LSP is a coupling excitation mode between incident electromagnetic waves and free electrons in a metal nanostructure. It is capable of satisfying momentum matching in the case of light incident, and it can be excited directly by an incident electromagnetic wave, so strong electromagnetic hot spots are generated between particles. Subsequently, the electric field distribution in MLF was simulated (Fig. 3(c)). SPP refers to an electromagnetic wave generated by the coupling of free electrons and photons in the dielectric and metal surface region. At the same frequency, the wave vector of SPP is always higher than the wave vector of light, so it cannot be excited directly by incoming light in the air. In addition, the weak electric field in the figure demonstrated that the SPP is not generated in the MLF [32]. When the two structures were combined to form a composite structure (Fig. 3(d)), the electromagnetic field between the particles was further enhanced. This is because after the incident light is scattered by the Ag NPs, the composite structure can match the momentum conditions and successfully excite the SPP mode and the BPP mode at optical frequencies [13,33]. To more effectively reflect the strong Raman enhancement effect of the composite structure, according to Fig. 3(e), the Raman test results of the three structures were compared with the enhancement factors. For the enhancement factor of the electric field simulation, where the composite SERS substrate is the strongest, graphene/Ag NPs is the second strongest and MLF is the weakest. The Raman test results show the same trend that the composite structure has the strongest Raman enhancement effect. And the enhancement of the composite substrate could reach 6 times that of the graphene/Ag NPs and 28 times that of the MLF. In addition, as presented in Fig. 3(d) and 3(f). For the MLF of composite structure, we investigated the field intensity of Au/Al₂O₃ interface from the bottom to the top layer. For the analysis of electric field distribution in different MLF in Fig. S1 (see Supplement 1). With the increase in the number of layers, the electric field intensity slowly increased and reached the highest at the top layer, which indicated that BPP in the MLF was successfully excited and transmitted energy to the surface. As revealed by the large increase in the field intensity at the top layer, a good coupling between BPP and LSP was generated. Compared with SPP, BPP can be coupled with LSP more effectively, and the existence of MLF can more significantly enhance the electric field between Ag NPs [34].

 

Fig. 3. (a) Schematic diagram of the composite SERS substrate. The electric field distribution for (b) Ag NPs and (c) MLF. (d) The electric field distribution for the composite SERS substrate and each layer Au/Al₂O₃ interface. (e) The comparison of Raman intensity of 613 cm⁻¹ peak of 10⁻⁶ M R6G and Enhancement factor of different substrates. (f) The field intensity variation of Au/Al₂O₃ interface in layers 1–6.

Download Full Size | PDF

3.3 Structural optimization experiment

In order to optimize the structure of composite SERS substrate, two types of substrate structures were prepared by transferring graphene/Ag NPs onto the MLF by inverted transfer and forward transfer respectively. That is, the former (inverted transfer) is composite structure with graphene surface and Ag NPs spacer. The latter (forward transfer) has Ag NPs metal surface and graphene intermediate layer. Figure 4(a) illustrates the original Raman spectra of the two structures, and it was significantly found that the peak intensity measured in the graphene surface structure was higher than that of the Ag NPs surface structure. Interestingly, the intensity of the fluorescence background obtained by the graphene surface structure was only about 8,000, while the intensity of the fluorescence background for the Ag NPs surface structure was up to 30,000. The peaks at 613cm⁻¹ and fluorescence background of the two spectra were compared, as presented in Fig. 4(b), and the hot spot distribution and molecular detection of the two structures were analyzed. According to Fig. 4(b), in the graphene surface structure, Ag NPs tightly adhered on the backside of monolayer graphene. The strong electromagnetic hot spot was located at the gap in the plane with the graphene, and hot spot penetrated the graphene well, thus spreading from the central strongest hot spot to the vicinity and exposing the graphene surface. During SERS detection, the probe molecules could fall into the range of strong electromagnetic hot spots with a higher probability, which increased the utilization efficiency of hot spots. It was therefore indicated that the close vicinity of these hot spots from Ag NPs could make the monolayer graphene a hot surface. Thus, localized electromagnetic enhancement could be achieved on the atomically flat surface. In the Ag NPs surface structure, the electromagnetic hot spot was located at the gap of the lower surface of Ag NPs. However, the electromagnetic field on the hemispherical surface a with larger area was weaker, and the hemispherical surface was farther from the hot spots. When the probe molecules fell evenly onto the structure, most of the molecules would be distributed in the hemispherical surface with weak electromagnetic field, and only a few molecules would fall into the strong electromagnetic hot spot region. Moreover, molecules landing directly on the Ag NPs surface may have some adverse effects. (1) Certain chemical reactions were performed when metals had direct contact with probe molecules [35]. (2) Some probe molecules might undergo photocatalytic reactions under laser excitation [36]. (3) The fluorescence of dyes and the photoluminescence of metal produced strong fluorescence background [37]. The mentioned factors can significantly interfere with the intrinsic Raman signal of probe molecules. Graphene at the top of the composite structure has high biocompatibility and huge chemical inertness. It is capable of enriching probe molecules through π-π interaction and effectively avoiding the contact between metal and probe molecules, which prevents the chemical reaction between metal and probe molecules [38,39]. In addition, composite SERS substrate provides a cleaner baseline of the Raman spectrum since graphene can suppress interference by quenching background fluorescence through energy resonance transfer with the target molecule [40,41], which improves the signal-to-noise ratio and ensures the authenticity of the signal. Furthermore, the ultrafast carrier transfer mechanism in the graphene-plasmonic metal structure can generate greater localized electromagnetic field enhancement in the composite structure, which further improves the sensitivity of SERS substrate [42].

 

Fig. 4. (a) Raman spectra of 10⁻⁶ M R6G of the composite SERS substrate and the metal surface substrate. (b) Comparison of the peak (613 cm⁻¹) relative intensities and fluorescence background of the above spectrum. The inset shows the schematic diagram of hot spot distribution and molecular detection of the two substrates.

Download Full Size | PDF

In addition, the composite SERS substrate with only one layer of Ag NPs has a higher Raman enhancement effect compared with other graphene/Ag NPs structures [26,29], further demonstrating that the presence of MLF makes the Raman effect enhanced.

3.4 Performance exploration

To verify the sensitivity and enhancement effect of the composite SERS substrate, Raman tests were performed with R6G at different concentrations (10⁻¹¹−10⁻⁶ M) on the substrate. According to Fig. (5a), the Raman signal intensity decreased with the decrease of the R6G concentration, which indicated that concentration was proportional to SERS intensity. As for detecting limitation, tiny peaks were still observed at 613 cm⁻¹ and 1364 cm⁻¹ for sample concentration of 10⁻¹¹ M. Figure 5(b) presents a good linear relationship between signal intensity and R6G concentration at 613 cm⁻¹ and 1364 cm⁻¹. The linear equation is log(I)= 0.29 Log(C)+ 6.02 and log(I)= 0.31 Log(C)+ 5.72, where I and C denote the Raman intensity and the concentration of R6G, respectively. To ensure the reliability of the data, ten sampling points were randomly selected from the respective concentration sample to calculate its average signal intensity. The values of correlation coefficient (R²) of the 613 cm⁻¹ and 1364 cm⁻¹ were 0.991 and 0.996, respectively, demonstrating a high credibility. According to the formula: LOD = 3 (SD/S) [43], the LOD of R6G was reported as 6.2×10⁻¹⁰ M. To evaluate the SERS performance of the composite SERS substrate more accurately, the enhancement factor was calculated by (EF) [44].

 

Fig. 5. (a) Raman spectra of R6G on the composite SERS substrate with different concentrations (10⁻⁶ to 10⁻¹¹ M). (b) The interrelationship between the intensity at 613 cm⁻¹ and 1362 cm⁻¹ vs different R6G concentrations at log scale. (c) Raman spectrum of 10⁻⁹ M R6G on SERS substrate and that of 10⁻³ M R6G on Si substrate. (d) 15 Raman spectra of R6G at a concentration of 10⁻⁶ M were randomly collected from the SERS substrate. The inset shows the peak intensity distribution at 613 cm⁻¹. (e) The average Raman intensity of R6G at 613 cm⁻¹ from 10 different batches of the SERS substrates.

Download Full Size | PDF

3.5 Application for actual molecules

MG ion has been found as a significant cation in the human body. It participates in a wide variety of physiological activities of the human body, and it is closely correlated with the occurrence and prevention of many common and frequently-occurring diseases [45]. In the actual production and life, MG used as a bactericidal and parasitic chemical agent is widely decomposed in aquaculture to solve water mold [46]. For the human body, high MG ion content will cause a series of diseases. To further study the application of the composite SERS substrate in practical life, MG molecules were selected to verify the performance of the substrate in toxic substance detection. MG solutions at different concentrations (10⁻⁸−10⁻⁴ M) were dropped on the composite SERS substrate, and the corresponding Raman spectra were determined using a Raman spectrometer. According to Fig. 6(a), the characteristic Raman peak of MG molecule could be clearly observed. The Raman signal intensity of different MG concentrations decreased with the decrease in the concentration, with detecting limitation about 10⁻⁸ M. Figure 6(b) shows a good linear relationship (R² = 0.989) between Raman signal intensity (peak at 1617 cm⁻¹) and MG concentration, which contributes to quantitative detection of trace substances. For uniformity in application, 15 spectra collected from 50×50 µm² areas of the substrate were selected randomly to draw the Raman spectrogram, and the Raman signal intensity of the characteristic peak did not change significantly (Fig. 6(c)). According to Fig. 6(d), the characteristic peak intensities of the above 15 MG spectra were compared, the blue dot represents the relative intensity of the peak at 1617cm⁻¹, and the red dashed line indicates the average intensity with a relative standard deviation (RSD) of 6.40%. Thus, the composite SERS substrate had high uniformity. Accordingly, the composite SERS substrate of monolayer graphene/Ag NPs coupled with MLF can be used to detect toxic substances in practice.

 

Fig. 6. (a) Raman spectra of MG under different concentrations on the composite SERS substrate. (b) The interrelationship between the intensity at 1617cm⁻¹ vs different MG concentrations at log scale. (c) 15 Raman spectra of MG at a concentration of 10⁻⁴ M were randomly collected from the composite SERS substrate. (d) The peak (1617 cm⁻¹) relative intensities were collected from the mentioned Raman spectra.

Download Full Size | PDF

4. Conclusion

In this paper, a composite SERS substrate of graphene/Ag NPs combined with MLF was proposed. Through the combination of Ag NPs and MLF, and the analysis of the field intensity variation at the Au/Al₂O₃ interface of the respective layer, it was proved that BPP in the MLF was successfully excited and coupled with LSP. As indicated by the simulation and experimental results, the MLF significantly enhanced the electric field between Ag NPs. Moreover, graphene had the advantage of reducing the fluorescence background and avoiding chemical reactions between metals and molecules, and SERS substrate had high stability and biocompatibility. Besides, the SERS signal of R6G could be clearly detected even at the very low concentration of 10⁻¹¹ M, the experimental EF for 10⁻⁹ M R6G was examined as 2.8×10⁻⁷. The substrate also had prominent quantitative detection ability for MG molecules (LOD = 3.7×10⁻⁹ M), which could show extremely high sensitivity and good uniformity in practical detection. Based on the mentioned research, we consider that the composite SERS substrate is expected to be extensively used in practice such as medical diagnosis, food safety and environmental monitoring.