Two-Dimensional Self-Assembly of Au@Ag Core-Shell Nanocubes with Different Permutations for Ultrasensitive SERS Measurements.

Different self-assembly methods not only directly change the arrangement of noble metal particles on the substrate but also indirectly affect the local electromagnetic field distribution and intensity of the substrate under specific optical excitation conditions, which leads to distinguished different enhancement effects of the structure on molecular Raman signals. In this paper, first, the gold species growth method was used to prepare the silver-coated gold nanocubes (Au@Ag NCs) with regular morphology and uniform size, and then the two-phase and three-phase liquid-liquid self-assembly and evaporation-induced self-assembly methods were used to obtain the substrate structure with different NC arrangement patterns. The optimal arrangement of NCs was found by transverse comparison of Raman signal detection of probe molecules with the same concentration. Subsequently, surface-enhanced Raman scattering (SERS) measurements of Rhodamine (Rh6G) and aspartame (APM) were carried out. Furthermore, the finite element method (FEM) was employed to calculate the local electromagnetic fields of the substrates with different Au@Ag NC arrangements, and the calculated results were in agreement with the experimental results. The experimental results show that the SERS-active substrate was largely associated with the different arrangements of Au@Ag NCs, and the island membrane Au@Ag NCs array substrate obtained by evaporation-induced self-assembly can generate a strong local electromagnetic field due to the edge and corner bonding gap between the tightly arranged NCs; this endows the substrate with benign sensitivity and reproducibility and has great potential in molecular detection, biosensing, and food safety monitoring.

Introduction

In recent years, noble metallic nanostructures (gold, silver, platinum, etc.) have attracted wide attention due to their interesting optical properties. Precious metal nanostructures can concentrate light into subwavelength volume when they are excited under specific excitation conditions, thus enhancing the local electromagnetic field near the nanostructures. This phenomenon will increase the Raman scattering signal of the attached molecules on its surface by multiple orders for magnitude, which is called surface-enhanced Raman scattering (SERS).[1−3] In addition, precious metal nanostructures play a very important role in the surface-enhanced fluorescence (SEF) effect.[4] The SERS effect is widely used in medical diagnosis and treatment,[5] optics,[6] biological analyses,[7] environmental pollution monitoring,[8] catalysis,[9] and many other fields, based on its unique advantages, including simultaneous detection for multiple samples, high sensitivity, and fingerprint identification.

A large number of studies have shown that the absorption and scattering frequencies of surface plasmon resonance (SPR) can be selectively adjusted by changing the morphology, size, structure, and arrangement of noble metal nanoparticles and the local electromagnetic field on the substrate surface can be further enhanced.[10,11] On the one hand, researchers improved their SERS effect by preparing metal nanoparticles with different morphologies and sizes, such as spheres, triangles, cubes, tetrahedrons, regular octahedrons, etc.[12−16] On the other hand, changing the structure of nanoparticles can better adjust the optical properties of the sample. For example, the core–shell structure can effectively adjust the sample’s position of the absorption peak, which can be better coupled with the excitation light.[17] Moreover, the inert shell can indirectly change the spacing of the internal nanoparticles and form an isolation layer between the probe molecules and metal particles, which can effectively avoid the fluorescence quenching effect.[18] However, in the construction of an alloy structure, advantage can be taken of the bimetallic synergistic effect combining the characteristics of the two metals, so that the performance of the alloy structure is much better than that of the single-component sample structure.[19] In addition, the nanoparticle self-assembly into a large area of densely ordered monolayer film will provide a stable and effective substrate for further spectroscopic study. Wang[20] used l-cysteine (Cys) as a cross-linking agent to prepare the gold nanocube (NC) dimer structure. Professor Li[21] systematically analyzed the Raman activity of Au@SiO2 NCs by regulating the shell thickness of SiO2; although fluorescence quenching could be effectively avoided, the Raman signal was also weakened. Park[22] prepared a cube-in-cube hollow structure with super-high photoluminescence intensity and photonic quantum yield by in situ replacement and reduction methods and studied its photoluminescence characteristics, but the influence mechanism of the structure on SERS enhancement was not studied. Lin[8] quantitatively measured the SERS intensity of the remaining thiram in the soil with the help of the Au@Ag NC substrate but did not carry out an in-depth analysis of the influence of the substrate size change on SERS intensity. Although Au@Ag NCs have been extensively studied in the field of SERS measurement, analysis on the distribution and intensity of plasmonic “hot spots” on the surface of the substrate structure formed by different arrangements of NC and an in-depth research on subsequent spectral measurements are lacking. Therefore, further studies are required on how to obtain a single-layer high-quality structural substrate with a uniform structural arrangement and even particle distribution using different time-saving and labor-saving self-assembly methods.

In this paper, we used the crystal seed growth method and epitaxial growth method to prepare Au@Ag NCs with controllable size and uniform morphology. The Au@Ag NC monolayer was transferred to the surface of monocrystalline silicon in the sequence of three-phase and two-phase liquid–liquid self-assembly and evaporation-induced self-assembly methods, and the substrate structure with different Au@Ag NC arrangement patterns were obtained, as shown in Figure . The influence of particle spacing and arrangement on the distribution and strength of “hot spots” was calculated by finite element simulation while sample characterization techniques were carried out, and then the enhancement factors of the three substrates for the Raman signal with the same concentration probe molecule (Rh6G) were calculated under the same conditions. The best Au@Ag NC arrangement of the corresponding self-assembly methods was determined, and the detection limit was explored through SERS measurement of the Rh6G concentration gradient. After that, the SERS activity of aspartame in sugar-free drinks was detected, too. Combined with numerical simulation and experimental results, it is shown that different Au@Ag NCs exhibit admirable SERS detection sensitivity and reproducibility. It further expands the engineering application of Raman spectroscopy technology and has well scientific research value and potential engineering application.

Figure 1

Preparation of different Au@Ag NC substrates with different arrangements and the subsequent SERS detection flow chart.

Results and Discussion

Characterization of Au@Ag NCs

Nanocubes have been widely used in the study of nonlinear optics,[17] SEF, Raman scattering effects, and catalytic reaction[23−26] due to their easy coupling of eight vertices and 12 edges under specific excitation conditions. It can generate strong local electromagnetic fields due to the characteristics of core–shell structure and bimetallic synergism. As a result, more and more researchers have begun to pay attention to the application of Au@Ag NCs in the related area.

We used the modified seed growth method of Prof. Xia’s group to prepare Au@Ag NCs.[27] In simple terms, first, we reduced the gold precursor solution with sodium borohydride to obtain gold nanoseeds with a size of 3–5 nm. Second, we used CTAC, AA, and HAuCl4 solutions to grow gold nanoparticles of 30 nm size in two steps. Finally, a silver shell was coated on the surface of the gold nanospheres by the epitaxial growth method to form Au@Ag NCs; the specific sample preparation process is shown in the Supporting Information (SI).

Figure a shows the SEM image of the sample substrate at a small field of view, indicating that Au@Ag NCs are evenly distributed. The image in the inset of Figure b shows its colloidal solution in orange-red color. Figure b shows the SEM image under the wide field of view, with the area shown in Figure a enclosed within the red frame. Figure c shows the histogram of the particle size distribution of Au@Ag NCs in Figure b, with a calculated RSD of 4.24%; the relative deviation formula can be expressed as follows: RSD = SD/Sm, where SD is the standard deviation intensity of the peak, Sm is the average size, about 44 nm, with a regular morphology and uniform size distribution. Figure d shows the normalized UV–vis absorption spectrum of the aqueous solution of NCs. The characteristic absorption peaks of the core–shell structure can be adjusted indirectly by changing the thickness of the core–shell.[28] In addition, we replaced the surfactant from CTAB to CTAC in the subsequent sample preparation because a large amount of AgNO3 will be added to the system when the gold core is coated with a silver layer, and CTAB (Br–) will generate silver bromide (AgBr) with it. AgBr precipitates (solubility equilibrium constant Ksp = 7.7 × 10–13 at 25 °C) will reduce the amount of AgNO3 added to some extent due to the limited solubility of AgBr in water, while AgCl (Ksp = 1.56 × 10–10) formed by CTAC (Cl–) will improve this problem.[29]

Figure 2

Au@Ag NCs prepared by the seed growth method: (a) SEM image with a small field of view, and the image of the colloidal solution in the inset. (b) SEM of a large field of view. (c) Particle size distribution of Au@Ag NCs in (a). (d) UV–vis absorption spectrum of NCs in an aqueous solution.

Here, to prove that the as-prepared Au@Ag NCs have a core (gold)–shell (silver) structure, transmission electron microscopy (TEM) and energy-dispersive spectroscopy (EDS) characterization was performed. Through TEM characterization (Figure a), the structural information of the sample can be clearly understood, and it is indeed a core–shell structure. The EDS energy spectrum (Figure b) and mapping images (Figure c,d) were used to determine the distribution of gold and silver elements, which further indicated that the sample had a core (gold)–shell (silver) structure. We have constructed the structure of gold and silver to obtain both the stability of gold and the excellent plasmonic properties of silver. Studies have shown that the elements Au and Ag have very similar lattice constants (Au = 0.408 nm, Ag = 0.409 nm) and the same face-centered cubic (fcc) crystal structure, and so they can form stable solutions. In addition, the thickness of outer elements at the bimetallic interface will determine the overall structure. Ag is deposited on the surface of gold in the order of {110} > {100} > {111}. Thin silver layers will undergo rapid alloying, while thick silver layers will form core–shell structures.[30]

Figure 3

(a) TEM image of Au@Ag NCs. (b) X-ray energy-dispersive spectroscopy (EDS) of Au@Ag NCs. (c, d) Distribution of elements Au and Ag in panel (b), respectively.

Self-Assembly and Characterization of the Au@Ag NC Monolayer Film

In the self-assembly method, a long-term stable and ordered structure is formed through specific interactions of nanoparticles previously existing in a disordered system, which plays a critical role in subsequent substrate preparation and spectral measurement. Traditional self-assembly methods include gas–liquid, liquid–liquid, evaporation, and electrically controlled self-assembly methods. Among these, the former two mentioned are called fluid interface self-assembly, which is realized using amphiphilic molecular characteristics and the method of increasing interface pressure or Marangoni effect. Typical methods include Langmuir–Blodgett (LB) membrane assembly,[31] two-phase liquid–liquid self-assembly,[32] three-phase liquid–liquid self-assembly,[33] etc. In evaporation-induced self-assembly, a colloidal solution is placed under appropriate temperature and humidity conditions, resulting in the formation of nanoparticles in an ordered array of nanostructures through the interaction of electrostatic repulsion force, van der Waals force, and energy dissipation among nanoparticles in the solution.[34] In electronically controlled self-assembly, an ordered electric field is built near the colloidal solution, resulting in the formation of a chainlike arrangement of particles by adjusting the relevant parameters of the applied electric field.[35] In addition, centrifugal force regulation,[36] magnetic field regulation,[37] and optical regulation[38] have also been widely studied as self-assembly methods. Nanoparticles with different arrangement patterns obtained using different self-assembly methods will directly affect the distribution and intensity of the sample’s local electromagnetic field under excitation conditions, which is of vital importance for SERS.

The detailed assembly process can be found in the Supporting Information (SI), including liquid–liquid two-phase, liquid–liquid three-phase, and evaporative self-assembly methods.

Two-Phase Self-Assembly of the Au@Ag NC Monolayer Film

In the process of liquid–liquid two-phase self-assembly method, achieving the spontaneous migration of a large number of particles to the liquid–liquid interface (LLI) is difficult due to the electrostatic repulsion between Au@Ag NCs, the colloidal solution, and the oil phase mix. The general solution is to remove the replication Au@Ag surface charge to reduce the repulsive force between particles. Here, we add a kind of common solvents with low-polarity ethanol to solve this problem; although it is necessary for Au@Ag NC assembly, the destruction of the surface charge will cause substrate surface defects.[39] In addition, the use of surfactants to assist the screening of particles is also one of the methods to solve the assembly problem. There are a large number of CTAC micelles on the surface of Au@Ag NCs prepared and in the colloidal solution, which will induce the solid host NCs to adsorb more easily to the LLI. This process has little contact with the morphology of particles and the types of substances because of the physical process.[40] The experimental process is shown in Figure a. The Au@Ag monolayer film is transferred from the LLI to the surface of the silicon substrate. As shown in Figure b, it is clearly seen that the Au@Ag NCs have an overall nondense distribution due to a low colloidal concentration or defective surface ligand in a small field of view, and under the wide field, the particles are not tightly arranged and the spacing is large. So, we built a model and simulated and calculated the distribution and intensity of the local electromagnetic field of the structure (Figure c); it can be seen that there are many hot spots distributed between cubes, and most of them are distributed at the corners of cubes. However, it is difficult to form a nanogap due to the large distance between the cubes, which results in the local electromagnetic field being spliced between the individual cubes rather than plasmonic coupled.

Figure 4

(a) Schematic diagram of the two-phase self-assembly process. (b) SEM image of the Au@Ag NC array substrate. (c) Simulated electric field distribution of Au@Ag NCs arranged on the Si substrate by two-phase self-assembly; the incident light is polarized in the vertical direction.

Liquid–Liquid Three-Phase Self-Assembly of the Au@Ag NC Monolayer Film

As we know, liquid–liquid self-assembly is realized through the Marangoni effect. Therefore, we added another solvent with lower surface tension on the basis of two-phase to construct the oil/water/oil (O/W/O) three-phase system, which improves the surface tension gradient at different interfaces. It has the following advantages compared with the former: (1) the concentration and volume of the required colloidal sample solution can be less; (2) a higher amount of the nanoparticles in the colloidal solution can be carried to the LLI; (3) the formed gold film is more dense and easy to transfer to the LLI. However, it also has the disadvantage of easily forming a multilayer structure.[33] The specific experimental process is shown in Figure a. As the gold film at the LLI self-heals before the removal of cyclohexane on the surface, the beaker should not be shaken after removal to avoid damaging its compactness. By characterizing the Au@Ag NCs transferred to the silicon substrate (Figure b), it is demonstrated that the three-phase self-assembled gold films exhibit large-scale, monomer, dense, and uniform characteristics. Although Figure c shows that the strength of the local electromagnetic field on the substrate surface increases as the NC distance decreases, the distribution of plasmonic “hot spots” is disordered due to the nonuniform arrangement of the cubes. Furthermore, we can verify the particle transfer process in the assembly process by the following formula derivationwhere E is the particle energy, L is the side length, σ is the surface tension, and Z is the distance to the O/W interface; we can obtain the following equation according to Young’s equationwhere θ is the contact angle of the nanoparticles at the interface. In the process of self-assembly, the nanoparticles stably stay on the two-phase interface, and the contact angle is 90°, i.e., σp/w = σP/O, Z = 0.Therefore, when particles are transferred from the dichloromethane/water interface (bottom) to n-hexane/water interface (top), the interface energy can be described asIt is clear that the energy between the interfaces is only positively correlated with the gradient difference of surface tension. Therefore, it is also verified that the particles transfer in the direction that minimizes the Helmholtz free energy in the three-phase self-assembly processes.[41]

Figure 5

(a) Schematic diagram of the three-phase self-assembly process. (b) SEM image of the Au@Ag NC array substrate prepared by three-phase self-assembly. (c) Simulated electric field distribution of Au@Ag NCs arranged on the Si substrate by three-phase self-assembly; the incident light is polarized in the vertical direction.

It is worth noting that we need to replace the CTAC on the surface of Au@Ag NCs with PVP before conducting (O/W/O) three-phase self-assembly. The reason is that CTAC as a lipid bimolecular material will be tightly coated on the surface of the NCs during sample preparation, thus making it positively charged. In the subsequent formation of the three-phase systems, the particles are difficult to be induced at the LLI due to the large electrostatic repulsion between NCs and the formation of irreversible agglomeration of NCs.[42] In addition, the morphology and structure of the PVP-coated samples are basically unchanged under the irradiation of the near-range high-energy electron beam (TEM),[43] and so it is necessary to replace the surface ligands of Au@Ag NCs with PVP for the self-assembly process.

Evaporation-Induced Self-Assembly of the Au@Ag NC Monolayer Film

Although we built three systems by adding the liquid phase and improved the phenomenon of two-phase assembly of a small and nondense gold film, the specific arrangement of Au@Ag NCs is still worthy of further research, inspired by the preparation of a vertical substrate of gold rods in our group.[44] We chose to construct a tightly arranged Au@Ag NC array using an evaporation-induced self-assembly method.

A colloidal solution in the process of evaporation can cause a capillary stream that runs from the center to the edge of the droplet due to the difference in the droplet evaporation rate between the margin and center. In the case of fixed gas–liquid–solid contact line, this flow will also carry the solvent deposited near the contact line, resulting in the irregular distribution of particles in multiple layers; the “coffee ring” effect is contrary to the preparation of a single ordered film.[45] In general, the coffee ring effect can be inhibited by regulating the internal dynamics of the droplet or controlling the movement of the contact line,[46,47] and the former is chosen here. The specific experimental process is shown in Figure a. As mentioned above, in the process of sample preparation, we use a lipid-like substance, CTAC, which is present in large quantities of molecules or micelles in the colloidal solution. When the sample evaporates under a constant temperature and humidity, there will naturally be marginal capillary flow, which will bring Au@Ag NCs and CTAC to the edge of the droplet at the same time. With the increase of CTAC concentration near the contact line, a centripetal capillary flow will be generated due to the difference in concentrations between the edge and center of the droplet. However, the CTAC at the liquid/air interface causes the surface tension at the edge of the droplet less than that at the center, resulting in the centripetal Marangoni vortex.[48] These two centrally directed flows will bring the Au@Ag NCs to the center of the droplet, thus eliminating the “coffee ring” effect to an extent. With the evaporation of the solution, Au@Ag NCs in the droplet will be affected by the interaction of van der Waals force, induced depletion force, and electrostatic repulsion force. The relationship between the three can be expressed aswhere Evdm represents the van der Waals energy, Edep represents the depletion energy, Eele represents the electrostatic energy, and Etotal represents the total energy. The van der Waals forces and depletion forces try to bring adjacent NCs close to each other. The electrostatic repulsion stabilizes the NCs at a distance and prevents them from clumping randomly.

Figure 6

(a) Schematic illustration of the evaporation-induced self-assembly process. (b) SEM image of the Au@Ag NC array substrate prepared by evaporation-induced self-assembly. (c) Simulated electric field distribution of Au@Ag NCs arranged on the Si substrate by evaporation-induced self-assembly; the incident light is polarized in the vertical direction, and the wavelengths are (i) 532 nm and (ii) 785 nm, respectively.

So, using the evaporation-induced self-assembly method, the closely arranged Au@Ag NC array substrate can be prepared, as shown in Figure b. The island film of NCs meets the characteristics of large-scale, single-layer, and uniform distribution, which can be observed under a large field of view. Au@Ag NCs form a tight “one after the other” pattern when viewed in a small field. Based on the simulation calculation (Figure c) and the theoretical study,[49] we can clearly see that the plasmonic “hot spot” distribution of Au@Ag NCs is mainly concentrated in the corner position due to the regular arrangement of NCs. In addition, we simulate the local electromagnetic field distribution under the excitation wavelengths of 532 and 785 nm, respectively, which are consistent with the excitation conditions of the probe molecules Rh6G (532 nm) and APM (785 nm) in the subsequent spectral measurements. Meanwhile, it shows that the center position of the closely arranged cubic array in the substrate of the NCs island film exhibits a stronger local electromagnetic field intensity. It provides an excellent substrate and theoretical support for our subsequent spectral measurement.

SERS Measurement of the Au@Ag NC Monolayer Film with Different Arrangements

Through the two-phase and three-phase self-assembly and evaporation-induced self-assembly method described above, Au@Ag NCs with different arrangements are deposited on a clean silicon substrate to form a thin-film substrate; the substrate is immersed in a Rh6G ethanol solution with a concentration of 10–7 M under the same conditions for 30 min. It is thoroughly cleaned to ensure that a monolayer of Rh6G molecules is adsorbed to the substrate surface. Single-layer thin-film substrates are composed of three different Au@Ag NC arrangement modes that are used to measure the SERS spectrum, as shown in Figure a. In the experiment, the excitation wavelength was 532 nm, the laser power was 2.5 mW, and the integration time was 3 s. Each group of data is measured and averaged from 10 positions on the same substrate. By comparison of the SERS intensity of these three substrates, it is shown that the island membrane Au@Ag NC array substrate obtained through evaporation-induced self-assembly has the strongest effect on the Raman signal enhancement of Rh6G (gray line). Based on this substrate, we measure Rh6G Raman spectra with concentrations of 10–8–10–12 M to explore the sensitivity of the island membrane Au@Ag NC array substrate and the detection limit of probe molecules (Rh6G), as shown in Figure b. In the range of 400–1800 cm–1, the characteristic peaks of Rh6G at 614, 772, 1185, 1310, 1359, 1507, 1576, and 1650 cm–1 can be clearly seen, which is consistent with the literature reports.[32] At the same time, we also observed that the Raman strength weakened with a decrease in the concentration of probe molecules until a concentration of 10–12 M was reached, which was the substrate detection limit under the current test conditions. Therefore, it can be shown that the enhancement of the local electromagnetic field at the edge and corner binding gap between four adjacent Au@Ag NCs can effectively improve the Raman scattering signal of the probe molecule and thus greatly improve the detection sensitivity.

Figure 7

(a) SERS detection intensity of Rh6G at 10–8 M concentration by different self-assembled structural substrates. (b) SERS detection of Rh6G at a concentration gradient of 10–8–10–12 M by evaporative self-assembled structural substrates.

APM is a synthetic noncarbohydrate sweetener that occupies a significant position in the current food market due to its carbohydrate-free and fat-free properties. APM is widely used as a food sugar substitute in more than 100 countries. However, APM can be rapidly decomposed into phenylalanine, aspartic acid, and methanol in the human body. As methanol is highly toxic to the human body and excessive intake will lead to blindness, liver disease, and even death, it is important to establish a system for rapid detection of APM in food for the field of food hygiene and safety.[50] Here, we used the island membrane Au@Ag NC array substrate to conduct SERS spectrum detection of different concentrations of APM in the beverage.

We prepared APM water solutions at concentrations of 0.03125, 0.0625, 0.125, 0.25, 0.5, 1, and 2 g/L. The island membrane Au@Ag array substrates described above were prepared via the self-assembly method and used to analyze the SERS measurements with different concentrations of APM using an excitation wavelength of 785 nm, a laser power of 50 mW, and an integration time of 3 s. The results are shown in Figure a; the characteristic peak intensity of APM at 1007 cm–1 decreased with a decrease in the concentration of APM in solution, which is similar to the SERS result of this substrate for Rh6G above, and the characteristic peak at 523 cm–1 represents the Si substrate, which could be confirmed by the bare Si substrate control group; experimental results show that the detection limit of the substrate for APM is 0.0325 g/L, which is lower than the internationally recognized safe daily intake of APM of 0.05 g/L. Figure b shows the linear fitting of the Raman scattering intensity of the peak at 1007 cm–1 with different concentrations of APM. The regression coefficient R2 was 0.986, indicating that the SERS intensity of APM molecules is linearly dependent on the molecular concentration. The enhancing factor (EF) is one of the significant factors to evaluate the SERS substrate. We used the following formula to calculate the EF of the APM Raman signal of the island membrane installed Au@Ag NC array substrate.[32]where IRS and CRS are the Raman response strength and APM concentration of the control sample, respectively. ISERS and CSERS are the Raman response strength and target molecular concentration, respectively, deposited on the island membrane Au@Ag NC array substrate. Here, the substrate soaked with 10 g/L APM is selected as the reference sample, and the Raman peak at 1007 cm–1 is used as the reference characteristic peak. As seen in Figure a, we can obtain the Raman signal intensity of 0.5 g/L APM and 10 g/L APM at 1007 cm–1, which are 135.3 and 1627.4, respectively. Thus, the island film substrate is calculated to have an EF of about 240.56 for the APM Raman signal. At the same time, we also calculated the EF of the substrate for Rh6G by the Raman signal intensity at the 1356 cm–1 wavenumber obtained from the control group (10–3 M) and the experimental group (10–12 M), which is about 2.2 × 108. This strong enhancement is attributed to the plasmonic “hot spots” between the tightly packed Au@Ag NCs. In the study of SERS, the reproducibility of the substrate is considered as a principal standard to measure the SERS substrate in addition to cracking sensitivity. To measure the substrate’s reproducibility, 10 points are randomly selected from a substrate soaked at a concentration of 0.5 g/L APM. As shown in Figure c,d, the RSD value of Raman strength at 1007 cm–1 is calculated using the formula RSD = SD/IM from the 10 points measured, in which SD is the standard deviation intensity of the peak and IM is the average Raman intensity of the main peak, which is about 10.2%. This indicates that the island membrane Au@Ag NC array substrate has excellent reproducibility.

Figure 8

(a) SERS detection of the evaporation-induced self-assembled structure substrate for APM at a concentration gradient of 0.1, 0.2, 0.5, and 1 g/L. (b) Linear dependence of APM strength and molecular concentration. (c) Ten different detected locations are randomly selected on the substrate soaked with APM at a concentration of 0.5 g/L for SERS measurement. (d) Taking the characteristic peak at APM 1007 cm–1 as an example, the Raman intensity histogram of 10 locations was drawn, and the RSD was calculated as 10.2%.

Stability is used as another principal factor to evaluate the quality of SERS substrates. To verify the substrate with high stability. Figure a shows the Raman spectra of Rh6G at a concentration of 10–12 M on the island membrane Au@Ag NCs array substrate after 45 days; the SERS signal intensities of Rh6G molecules decrease to some extent after 45 days because of the loss of SERS activity with time. The intensities of the characteristic peaks at 776, 1363, and 1513 cm–1 are presented in Figure b for different periods, respectively. Even though the substrates soaked with Rh6G are exposed to air for 45 days, the substrate still exhibits excellent Raman activity. This substrate possesses great potential in sensing and detection due to the advantages mentioned above. There are also related articles that studied the factors affecting the SERS activity of the coffee ring lining obtained after the colloidal natural evaporation.[51] Meanwhile, Zhang made a detailed study on the SERS activity of the substrate obtained by electrostatic self-assembly,[52] which was a good comparison and proof for the SERS detection results obtained in this paper.

Figure 9

(a) Raman spectra of 10–12 M Rh6G on the island membrane Au@Ag NC array substrate on different days. (b) Comparison of the intensities of SERS signals of peaks at 776, 1363, and 1513 cm–1. (c) Raman spectrum of fruity beer using the island membrane Au@Ag NC array as the substrate.

To study the application of the island membrane Au@Ag NC array substrate in liquid food molecular detection, we soaked the substrate directly in untreated fruity beer; the sample was taken out after 1 h, and SERS measurements were carried out after drying. The Raman spectrum is shown in Figure c (average spectral data obtained from 20 different positions randomly selected). It can be clearly seen that there is a single and very obvious characteristic peak at a wavenumber of 1007 cm–1. This characteristic peak corresponds to APM due to the Raman spectrum of the 0.25 g/L on the Si substrate (the inset shows an image of fruity beer). Later, we used the internal standard method to carry out quantitative SERS detection to obtain the accurate concentration of the measured substance. This result indirectly proves the admirable development prospect of our work in the field of food safety detection.

Conclusions

In a nutshell, we obtained plasmonic nanostructures with different arrangement patterns using two-phase and three-phase liquid–liquid self-assembly and evaporation-induced self-assembly methods. Combined with the FEM simulation, the electromagnetic field distribution of the different substrate structures and its effect on Raman signals of different probe molecules were systematically studied. Both experimental and theoretical results showed that the SERS activity of the substrate is largely related to the different arrangements of Au@Ag NCs; the compact Au@Ag NC monolayer obtained by three-phase self-assembly lacks the ordered arrangement, while the island membrane array substrate of NCs formed by evaporation-induced self-assembly produced a strong local electromagnetic field due to the edge and corner bonding gap between the tightly arranged NCs. The Raman signal of Rh6G and APM can be significantly enhanced by the substrate, and the detection of 10–12 M and 0.03125 g/L concentration limits can be achieved, respectively. The Raman scattering intensity with different concentrations of APM shows good linear fitting (R2 = 0.986). Therefore, the substrates’ unexceptionable uniformity and stability can be proved by the spectroscopic measurement of the sample, and the substrate has been initially used in the detection of food molecules. As a result, the island membrane Au@Ag NC array substrate based on evaporation-induced self-assembly shows a potential application prospect in biological sensing, food safety detection, SERS detection, and many other fields.