Remarkable SERS Detection by Hybrid Cu2O/Ag Nanospheres.

Cu2O nanospheres (NSs) were synthesized by modifying the glucose reduction method. Based on this method, Cu2O/Au (Ag) NSs were further prepared by in situ reduction of HAuCl4 (via electron beam evaporation of Ag). With Rhodamine 6G (R6G) as probe, the surface-enhanced Raman scattering (SERS) characteristics of the three samples were systematically studied. The experiment results showed that the enhancement factor (EF) of Cu2O/Au (Ag) NSs as 1.25 × 108 (2.74 × 109) and the ultralow detection limit (LOD) as 8.07 × 10-12 (1.13 × 10-13) M for R6G. The excellent performance of SERS may be due to the charge transfer (CT) between metal-semiconductor (MS) molecules and the strong electromagnetic field (E-field) of each hot spot. In addition, discrete dipole approximation (DDA) simulations were performed to simulate the E-field enhancement of the Cu2O and Cu2O/Au (Ag) NSs in a three-dimensional (3D) configuration. These further supported that the high SERS performance for R6G is because of the powerful E-field coupling between neighboring Au (Ag) NPs and the surface plasmon resonance (SPR) effect. The Cu2O/Ag NSs have potential in applications such as biomedicine, food safety, and environmental monitoring because of their high sensitivity and good reproducibility.

Introduction

Surface-enhanced Raman scattering (SERS) is a no-label, lossless, and ultrasensitive spectral detection technique for identifying trace analytes and enables many applications, such as medicine, chemistry, biology, and so on.[1−3] It is widely accepted that the SERS effect comes from two main factors: chemical enhancements (CM) and electromagnetic (EM) effect.[4,5] CM is based on the enhancement effect due to the charge transfer (CT) caused by the photoinduction of adsorbed molecules and surfaces.[6−8] In contrast, the core of the EM effect is that the local surface plasma oscillations on the noble metal surface under the action of an EM field greatly enhance the Raman signal of the target molecule.[9−11] To date, various noble metallic nanoparticles (NPs), for instance, Ag, Au, Pd, and Pt, have been extensively used as SERS substrates.[12−19] However, these metal nanostructures do not find widespread use in realistic applications because of their high cost. Naturally, scientists have developed semiconductor materials as the SERS substrate,[20−22] but the drawback is their lower SERS enhancement effect. Therefore, a precious metal was intentionally added to semiconductors to form hybrid semiconductor materials as SERS substrates, which combine the merits of both materials: the high SERS sensitivity of noble metals and the low cost of semiconductor materials. For example, Au/CdS, SiO2/Au, ZnO/Ag, Fe3O4/Au, and TiO2/Ag composites were employed as effective SERS substrates with high sensitivity,[23−26] excellent detection limit, rapid response, and fingerprint effect. Unfortunately, the SERS study of such synergetic contribution of noble metals and semiconductors in hybrid nanocomposites is still relatively rare, especially for Cu2O.

Cu2O has a direct band gap of 2.2 eV and has good electrical conductivity as well as wide applications in magnetic field radiation, batteries, solar conversion, gas sensing, magnetic storage media, and catalysis.[27−29] The SERS ability of Cu2O hydrosol was first put forward by Kudelski et al.[30] Shortly after, structures based on Cu2O NSs became a research hot spot because of their significant advantages with dense hot spots. However, the study on the synergetic effect of different noble metal (Au or Ag) hybrid cuprous oxides as SERS substrates has seldom been reported. In this work, we prepared highly uniform Cu2O, Cu2O/Au, and Cu2O/Ag NSs as SERS substrates, and explored the SERS performance of R6G molecules. It is found that these substrates have ultrahigh SERS activity and reproducibility, on which the lowest detected threshold of R6G can be as low as 1.0 × 10–12 M. DDA simulation was performed to plot the EM field around Cu2O, Cu2O/Au, and Cu2O/Ag NSs. It was further certified that the high Raman sensitivity due to EM enhancement mechanisms exhibited a strong electrical field surrounding the metal NPs and the region between Au (Ag) NPs and Cu2O NSs. In addition, CM was explored and the contribution of the CT process of MS molecules to SERS was investigated by the Herzberg–Teller theory.[31] The results show that the coupling between the semiconductor Fermi level and the metal work function is related to the CT process. This research is important for extending the applicability of Raman spectroscopy to various charge transfer problems on MS molecular surfaces and for developing novel MS Raman probes.

Results and Discussion

Morphology and Microstructure

The scanning electron microscope (SEM) image of the Cu2O NSs is presented in Figure a. The samples were observed to have a uniformly spherical morphology with an average diameter of about 200 nm. The surface of the Cu2O NS is rough and its corresponding high-resolution electron microscopy (HRTEM) image confirms its high-quality polycrystalline nature, as shown in Figure b,c. The lattice spacing between the neighboring planes is 0.244 nm, corresponding to the (111) crystal plane of Cu2O. Its selected-area electron diffraction (SAED) (inset of Figure c) exhibits diffraction rings, indicating that the Cu2O NS has a polycrystalline structure.

Figure 1

SEM, TEM, and HRTEM images of the Cu2O NSs. (a) SEM, (b) TEM, and (c) HRTEM images corresponding to the framed area of (b); the inset is the SAED pattern of an individual NS.

SEM, Gene">TEM, and HRTEM images of the Cu2O NSs. (a) SEM, (b) TEM, and (c) HRTEM images corresponding to the framed area of (b); the inset is the SAED pattern of an individual NS.

Figure a shows the SEM image of the Cu2O/Au NSs. It is seen that their surfaces are rough with many Au NPs attached to them. Figure b illustrates the TEM image of the Cu2O/Au NSs; it was revealed that Au NPs with a diameter of about 16.76 nm were adsorbed on the surface of the Cu2O NSs. The HRTEM image further depicts that the lattice spacing between contiguous planes is 0.231 and 0.244 nm, corresponding to the (111) crystal plane of Au and the (111) plane of Cu2O respectively, as displayed in Figure c.

Figure 2

SEM, TEM, and HRTEM images of the Cu2O/Au NSs. (a) SEM, (b) TEM, and (c) HRTEM images recorded on the framed area of (b).

SEM, Gene">TEM, and HRTEM images of the Cu2O/Au NSs. (a) SEM, (b) TEM, and (c) HRTEM images recorded on the framed area of (b).

Next, we analyzed the formation mechanism of the Cu2O/Au NSs. Cu2O can directly reduce AuCl4– at room temperature because the standard reduction potential of Cu2+/Cu2O is 0.203 V and the standard reduction potential of AuCl4–/Au is 1.002 V.[32,33] In consequence, the following reaction occurs[36]

The reduction reaction occurs on the surface of the NSs, so the resulting gold NPs can be in situ adsorbed on the surface of the Cu2O NSs. Subsequently, Cu2O/Au NSs were obtained.

The SEM image of the Cu2O/Ag NSs is visualized in Figure a,b. It is seen that the surface of the Cu2O/Ag NSs is rough and the distribution of Cu2O/Ag NSs on the Si substrate is uniform. The TEM image of the Cu2O/Ag NSs is presented in Figure c. It should be noted that the sputtered Ag NPs with a diameter of 20.62 nm are evenly adsorbed on the upper surface of the Cu2O NSs. The HRTEM image further elucidates that the lattice spacing between neighboring planes is 0.230 and 0.244 nm, respectively, matching the (111) crystal plane of Ag and the (111) plane of Cu2O, as highlighted in Figure d.

Figure 3

SEM, TEM, and HRTEM images of the Cu2O/Ag NSs. SEM (a, b) images of different magnifications, (c) TEM, and the (d) HRTEM image recorded on the framed area of (c).

SEM, Gene">TEM, and HRTEM images of the Cu2O/Ag NSs. SEM (a, b) images of different magnifications, (c) TEM, and the (d) HRTEM image recorded on the framed area of (c).

Figure describes the X-ray diffraction (XRD) patterns of the Cu2O, Cu2O/Au, and Cu2O/Ag NSs. All of the peaks can be indexed to the cubic structure of Cu2O with lattice constants of a = b = c = 4.268 Å (JCPDS card no. 05-0667). The other four peaks can be assigned to the planes (111), (200), (220), and (311) of the Ag/Au cubic structure with a = b = c = 4.086 Å/a = b = c = 4.079 Å (JCPDS no. 04-0783/04-0784). Except these, no other impurities are detected.

Figure 4

XRD patterns of the Cu2O, Cu2O/Au, and Cu2O/Ag NSs.

The composition of the obtained Cu2O/Au and Cu2O/Ag NSs was further analyzed by the EDX spectrum, as shown in Figure . It proved that the former is made up of Au, Cu, and O elements and the latter consists of Ag, Cu, and O elements. The Si signal comes from the silicon substrate on which the sample is placed.

Figure 5

EDX patterns of the NSs. (a) Cu2O/Au and (b) Cu2O/Ag.

EDX patterns of the NSs. (a) pan class="Chemical">Cu2O/Au and (b) pan class="Chemical">Cu2O/Ag.

The elemental composition and chemical states of the Cu2O, Cu2O/Au, and Cu2O/Ag NSs were explored by X-ray photoelectron spectroscopy (XPS), as shown in Figure . Calibration was based on C (1s) at 284.5 eV for XPS analysis to obtain accurate binding energy. The high-resolution Cu 2p XPS spectrum (Figure b) shows two peaks of 952.11 and 932.18 eV, which are ascribed to Cu+ 2p1/2 and Cu+ 2p3/2, respectively, indicating the presence of Cu2O.[34] In addition, the Cu 2p binding energy of the Cu2O/Au and Cu2O/Ag NSs shifts to a higher energy than that of the bare Cu2O NSs, and the peak sites of Au 4f and Ag 3d differ from those of monometallic Au0 and Ag0.[35] This indicates that the charge distribution at the interface of Cu2O–Au (Ag) NSs has changed.

Figure 6

XPS spectra of the Cu2O/Au and Cu2O/Ag NSs. (a) Survey scan, (b) Cu region, (c) Au region, and (d) Ag region.

SERS Study of the Cu2O, Cu2O/Au, and Cu2O/Ag NSs

A schematic diagram of the Cu2O (Cu2O/Au or Cu2O/Ag) NSs for SERS is illustrated in Figure . To study the SERS properties of these substrates, R6G was selected as the probe. Figure shows the SERS spectra of R6G with different concentrations on the different substrates. It was found that the SERS signals of R6G on the Cu2O (Cu2O/Au or Cu2O/Ag) NS substrate were still observable when its concentration was down to 1.0 × 10–6 M (1.0 × 10–11 or 1.0 × 10–12 M). The peaks were located at 611, 771, 1182, 1310, 1367, 1420, 1509, 1538, 1571, 1600, and 1650 cm–1, all of the peaks were consistent with the characteristic peak of R6G.[36]

Figure 7

Schematic illustration of SERS to target molecules on the substrates. (a) Cu2O NSs, (b) Cu2O/Au NSs, and (c) Cu2O/Ag NSs.

Figure 8

SERS spectra of different concentrations of the R6G molecule on the different substrates. (a) Cu2O NSs, (b) Cu2O/Au NSs, (c) Cu2O/Ag NSs, (d, e), and (f) linear fit of the SERS intensity of the peak at 611 cm–1 vs R6G concentrations.

To evaluate its limit-of-detection (LOD) on these substrates, the calibration curve of the SERS intensity and concentration of the R6G solution at 611 cm–1is shown in Figure d–f. The LOD values for R6G were identified to be 1.01 × 10–7, 8.07 × 10–12, and 1.13 × 10–13 M (S/N = 3)[37] on the Cu2O, Cu2O/Au, and Cu2O/Ag NS substrates, respectively. In addition, for practical application, the enhancement factor is another important parameter. Therefore, the characteristic peak at 611 cm–1 for R6G was also employed to assess it. The EF of our proposed SERS substrates was calculated using the equation[38]where ISERS and I0 are the SERS intensity of the R6G molecule and the Raman signal intensity of solid R6G under the same experimental conditions, respectively. N0 and NSERS represent the number of R6G probe molecules in solid and SERS samples under laser irradiation during the test (details in the Supporting Information Figure S1). Based on our experimental results, the EFs for R6G are 3.87 × 103, 1.25 × 108, and 2.74 × 109 on the Cu2O, Cu2O/Au, and Cu2O/Ag NS substrates, respectively. The calculated details are shown in Table S1 (Supporting Information). In comparison with other hybrid semiconductor substrates, SERS detection parameters of various target molecules such as R6G and other analytes on the other substrates and our substrates are listed in Table S2 (Supporting Information). The results show that the EF and LOD on Cu2O/Ag NS substrate are obviously higher than those of other substrates. (Table S2, Supporting Information). The higher EF and lower LOD are related to the morphology of the Cu2O/Ag NSs with the rough surface because, given a certain volume, the sphere has the largest surface area, which is favorable for the capture of probe molecules.

In addition, 40 points were randomly selected from the Cu2O (Cu2O/Au or Cu2O/Ag) NS substrate to collect 3D Raman signals with an R6G concentration of 1.0 × 10–4 M (1.0 × 10–8 or 1.0 × 10–11 M) to validate the repeatability and uniformity of the substrate, as shown in Figures a, 10a, and 11a. All of the characteristic peaks are neatly arranged. Moreover, the relative standard deviations (RSDs) of the intensity are shown in Figures b–d, 10b–d, and 11b–d. RSDs of the peaks at 1367, 1509, and 1650 cm–1 are 14.42% (17.08 or 14.23%), 15.62% (16.93 or 14.07%), and 15.45% (17.37 or 15.15%), respectively. All calculated RSD values are lower than 18%,[39] which demonstrates the reproducibility of the Cu2O (Cu2O/Au or Cu2O/Ag) NS substrate.

Figure 9

(a) 3D Raman spectra of 10–4 M R6G molecules at 40 randomly selected spots on the Cu2O substrate and (b–d) RSD values of the selected peaks at 1367, 1509, and 1650 cm–1.

Figure 10

(a) 3D Raman spectra of 10–8 M R6G molecules at 40 randomly selected spots on the Cu2O/Au substrate and (b–d) RSD values of selected peaks at 1367, 1509, and 1650 cm–1.

Figure 11

(a) 3D Raman spectra of 10–11 M R6G molecules at 40 randomly selected spots on the Cu2O/Ag substrate and (b–d) RSD values of selected peaks at 1367, 1509, and 1650 cm–1.

(a) 3D Raman spectra of 10–4 M R6G molecules at 40 randomly selected spots on the pan class="Chemical">Cu2O substrate and (b–d) RSD values of the selected peaks at 1367, 1509, and 1650 cm–1.

(a) 3D Raman spectra of 10–8 M R6G molecules at 40 randomly selected spots on the pan class="Chemical">Cu2O/Au substrate and (b–d) RSD values of selected peaks at 1367, 1509, and 1650 cm–1.

(a) 3D Raman spectra of 10–11 M R6G molecules at 40 randomly selected spots on the pan class="Chemical">Cu2O/Ag substrate and (b–d) RSD values of selected peaks at 1367, 1509, and 1650 cm–1.

SERS Mechanism

Wang et al. previously studied the pan class="Chemical">SERS mechanism of Ge and Si nanomaterials and reported that the pan class="Chemical">SERS activity of Ge and Si substrates was mainly caused by the CT effect between the semiconductor and the molecule.[40]

For our case, involving the semiconductor-molecule model, the specific CT process is shown in Figure a. In this process, from the Cu2O NSs to R6G, first the electrons in Cu2O NSs VB are stimulated by incident light. This creates the electrons in CB of Cu2O NSs CB and holes in VB. The excited electrons quickly transfer from the Cu2O NSs to the matching energy levels in the R6G mesosphere. They then go back to the Cu2O NSs and recombine with the hole. In this process, R6G molecules emit a Raman photon in some vibrational state.[41] Similar is the case for the CT process from R6G to the Cu2O NSs.[42] To further verify our SERS model above, the R6G molecules on Cu2O NS substrates with an excitation wavelength of 785 nm were tested, as shown in Figure . The results show that under an excitation of 785 nm, the enhancement effect of R6G molecules on the Cu2O NSs is significantly lower than 532 nm because the excitation frequency (785 nm; 1.58 eV) is lower than the energy gap between the LUMO and HOMO of the probe molecule (R6G; (2.3 eV) and the band gap (2.2 eV) of the Cu2O NSs. Consequently, the CT process in R6G is prohibited under the excitation of 785 nm (Figure b).

Figure 12

Schematic of light-induced CT between the Cu2O NSs and the R6G molecule at different excitation wavelengths. (a) 532 nm and (b) 785 nm.

Figure 13

SERS of the R6G spectrum (1.0 × 10–3 M) absorbed on the Cu2O NSs at 532 nm (red line) and 785 nm (black line) excitation wavelength.

Schematic of light-induced CT between the pan class="Chemical">Cu2O NSs and the R6G molecule at different excitation wavelengths. (a) 532 nm and (b) 785 nm.

pan class="Chemical">SERS of the R6G spectrum (1.0 × 10–3 M) absorbed on the pan class="Chemical">Cu2O NSs at 532 nm (red line) and 785 nm (black line) excitation wavelength.

To determine the CM mechanism on the SERS of the Cu2O substrate modified by noble metal, the SERS spectra of R6G on different substrates were compared. According to the Herzberg–Teller selection rule[31] (the details are provided in the Supporting Information), the (a″) vibration at 611 cm–1 is more selectively enhanced. The experimental results are shown in Figure .

Figure 14

Comparison between the SERS spectrum of the R6G molecule adsorbed on the Cu2O NSs, Cu2O/Au NSs, and Cu2O/Ag NSs at 532 nm excitation wavelength. (a) Comparison of the intensity of each peak position on different substrates, (b) comparison of the ratio of the peak intensity in the a″ vibration mode to the peak intensity in the a′ vibration mode on different substrates, and (c) comparison of the SERS spectrum on different substrates.

We can see that the a″ vibration mode at 611 cm–1 is enhanced relatively less compared to the a′ ones at 1367, 1509, and 1650 cm–1. The enhancement of the vibration mode (a″) is mainly caused by the CT process. More importantly, we compared the SERS intensity ratio of the a″ mode at 611 cm–1 to the a′ modes at 1367, 1509, and 1650 cm–1. It reveals that the ratio for Cu2O (Cu2O/Ag) NSs is significantly higher than that for Cu2O/Au NSs, as displayed in Figure b. It illustrates that the contribution of the CT effect in the SERS spectrum is in the order Cu2O/Ag NSs > Cu2O NSs > Cu2O/Au NSs. Furthermore, UV–visible absorption spectroscopy (UV–vis) absorption spectra of the R6G molecule adsorbed on different substrates (Cu2O, Cu2O/Au, and Cu2O/Ag NSs) were compared with those of the pristine counterparts and R6G, which is in accordance with the Raman intensity ratio (Figure S2, Supporting Information).

We further elucidate the CT contribution to SERS by the energy level diagram, as shown in Figure a–b. The work function of Au (Ag) is 5.10 (4.26) eV, and the Fermi level of Cu2O (4.84 eV) is higher than that of Ag but lower than that of Au. Thus, the charge transfers from Cu2O to Au and from Ag to Cu2O until the Fermi levels of the two systems are consistent.[43,44]

Figure 15

(a, b) Charge transfer process between the interface of Cu2O and Au (Ag). (c, d) Schematic diagram of the light-induced CT process of Cu2O/Au (Ag) NSs and the R6G molecule under 532 nm excitation.

Based on the above discussion, we return to explain the CT process in SERS and the established MS-molecule model, as depicted in Figure c,d. For the Ag (Au)–Cu2O-R6G system, the electrons in Ag (Au) were first stimulated by the laser, then the excited charge quickly transferred from Ag (Au) to the CB state of the Cu2O NSs. However, Ag (4.26 eV) and Au (5.1 eV) have different work functions, and the Fermi level of Cu2O is lower than that of Ag but higher than that of Au. In this case, the CT from Cu2O to Au dominates the CT process for the Au–Cu2O system, whereas the CT from Ag to Cu2O dominates the CT process for the Ag—Cu2O system. Thus, the CT process of Cu2O and R6G in the Au–Cu2O-R6G system is inhibited, while the CT process of Cu2O and R6G in the Ag–Cu2O-R6G system is enhanced. Sequentially, the two systems cause a significant difference in CT efficiency in the R6G molecular system.

To understand our experimental results, DDA simulation was performed to determine the electromagnetic field distribution (E-field) around the Cu2O NSs, Cu2O/Au NSs, and Cu2O/Ag NSs. According to the experimental results, Cu2O NSs with a diameter of about 200 nm were used as a model; this is consistent with the SEM image in Figure . For Cu2O/Ag (Au) NSs, Ag (Au) NPs cover the surface of Cu2O NSs (see Figures and 3). Figure S3 shows the UV–vis spectra of Cu2O, Cu2O/Au, and Cu2O/Ag NSs. It can be seen that there are two absorption peaks at 479 and 619 nm for Cu2O NSs, at 482 and 616.5 nm for Cu2O/Au NSs, and at 507.5 and 615 nm for Cu2O/Ag NSs. On account of the dispersion properties of Cu2O that neither the refractive index nor the dielectric function is fixed with the wavelength and fluctuates greatly. Therefore, we choose the corresponding wavelengths of the two absorption peaks as the incident plane wave. The incident EM wave enters along the y-axis (k vector). The value of the dielectric constant is taken from Palik.[45]

The E-field intensity distribution is displayed in Figure . It varies greatly with the wavelength of the incident plane wave, which is consistent with the above experimental results. It is observed in Figure c–f that due to the synergetic effect of the CT process, there is a highly localized strong E-field between the Ag (Au) NPs and Cu2O NSs. Significantly, the high-intensity E-field regions are distributed at the top of each Ag (Au) NP and in the junction region between the Ag (Au) NP and Cu2O NSs. This high-intensity localized E-field is very important for ultrasensitive Raman detection of molecules. Therefore, the SERS signal of R6G on the Cu2O/Ag (Au) NSs can be observed even with a concentration down to 1.0 × 10–12 (1.0 × 10–11) M.

Figure 16

DDA simulations of the E-field distribution on (a, b) the Cu2O NSs at incident wavelengths of 479 and 619 nm, (c, d) the Cu2O/Au NSs at incident wavelengths of 482 and 616.5 nm, and (e, f) the Cu2O/Ag NSs at incident wavelengths of 507 and 615 nm, respectively. The light source is incident from the y-axis. The figure shows the xz-plane including the polarization direction in logarithmic coordinates.

This pan class="Chemical">SERS substrate can be applied for the detection of other molecules, such as pan class="Chemical">acridine orange, Sudan III, and so on.

Conclusions

In summary, we have successfully prepared Cu2O NSs and Cu2O/Ag (Au) NSs by vacuum sputtering deposition (in situ reduction). The Cu2O NSs, Cu2O/Au NSs, and Cu2O/Ag NSs showed excellent SERS performances for detecting R6G with LODs of 1.01 × 10–7, 8.07 × 10–12, and 1.13 × 10–13 M, respectively. The extraordinary SERS phenomena were caused by two factors: (i) influence of the connection between the metal work function and semiconductor work function on the CT process of metal–semiconductor-molecular systems and (ii) the synergetic contributions of the strong E-field based on the E-field coupling effect between Ag (Au) and Cu2O NSs, and the surface plasmon resonance (SPR) effect of each Ag (Au) NP. The good SERS signal consistency and sensitive SERS detection performance revealed that Cu2O/Ag NSs possess great application prospects for biomedicine, environmental science, food safety, and so on.

Experimental Details

Materials

CuSO4·5H2O, NaOH, glucose, poly(vinylpyrrolidone) (PVP), ethylene glycol, and HAuCl4·4H2O were acquired from Tianjin Sailboat Chemical Reagent Technology Co., Ltd. The Ag particle was obtained from Beijing Rui New Material Technology Co., Ltd. All chemicals were analytically pure and could be used without any further purification.

Synthesis of the Samples

Synthesis of Cu2O Nanospheres (NSs)

Cu2O NSs were synthesized by improving the glucose reduction method. The process is as follows: first, under an ultrasonic oscillation, 0.2496 g of CuSO4 and 0.04 g of PVP were sequentially dissolved in 50 mL of the ethylene glycol solution for 30 min. Second, 25 mL of an aqueous solution containing 0.1 g of NaOH was added to the above solution and stirred for 10 min to completely mix them. Then, a solution of glucose (6 g in 25 mL deionized water) was added for 15 min with slow stirring. Third, the obtained mixture was transferred to a water bath and kept at 80 °C for 1 h, then naturally cooled to room temperature. Finally, the resulting orange sediment was collected by centrifugation, washed several times thoroughly with deionized water and anhydrous ethanol, and then dried for 5 h in a vacuum oven at 55 °C. The product was Cu2O NSs.

Synthesis of Cu2O/Au NSs

The Cu2O NSs obtained by 0.01 g were dispersed in 10 mL of deionized water and then 3 mL of the HAuCl4 solution (0.1 wt %) was added to it under magnetic agitation. The solution color could be observed gradually changing from orange to red to black in a few seconds. The products were collected by centrifugation, washed several times with deionized water and anhydrous ethanol, and then dried in vacuum at 55 °C for 5 h.

Preparation of Cu2O/Ag NSs

The above-prepared Cu2O NSs (0.01 g) were dispersed into 50 mL of absolute ethanol in a beaker under ultrasonic oscillation for 30 min. Then, 30 μL of Cu2O NS ethanol solution was pipetted on the silicon wafer (1.0 ×1.0 cm2). After that, silver NPs were deposited on this silicon wafer by electron beam evaporation under vacuum condition (5 × 10–4 Pa in the cavity). Thus, Cu2O/Ag NSs were obtained.

SERS Measurements

For the SERS measurements, the aforementioned Cu2O/Au and Cu2O/Ag NSs were dissolved in a certain amount of absolute ethanol to form a suspension. Afterward, 20 μL of the suspension was taken to disperse on the Si piece (0.5 × 0.5 cm2) and dried in air. Then, the substrate for SERS measurements was obtained by dripping 5 μL of Rhodamine 6G aqueous solution into the treated Si substrate and drying it.

Characterization

Energy-dispersive spectroscopy (EDS) and SEM (Quanta 250ESEM, FEI) were used to investigate the surface morphology and elemental composition of the samples. The microstructure of the samples was further analyzed by using a JEOL JEM-2000EX TEM. XRD of the D/max-3B Rigaku model with Cu Kα radiation (λ = 1.5406 Å) was used to determine the phase identification of the sample. At room temperature, XPS from the PHI 5000 Versaprobe of UlVAC-PHI (Japan) was used to detect the sample chemical composition and chemical valence states.

The samples were dispersed in a certain amount of pan class="Chemical">ethanol solution, and the UV–vis spectrometer (U-4100, Hitachi) was used to detect the absorption spectra of the samples in the wavelength range of 200–800nm.The probe molecular Raman spectral signal was measured using a confocal Raman spectrometer (Andor, England) with a laser excitation source wavelength of 532 nm and a laser power of approximately 4 mW. The acquisition time of each test was 10 s, and the stacking time was 3.

Simulation

DDA is a method developed in recent years to calculate the absorption, scattering, and electromagnetic field distribution of particles of arbitrary shapes and sizes, which has a strong advantage in calculating the interaction between the light and nanoparticles.[46] The DDA method considers the nanoparticles as cubes with N-polarizable points. Under the induction of the electric field of light, each small cube is polarized, and the induced electric field generated further affects the adjacent cubes.[47] The optical properties of the material can be obtained by self-consistent calculation of the whole system. There is no limit to which the grid is polarized, which means that the DDA method can represent multiple substances of arbitrary shape and composition. This paper used the DDA (DDSCAT code 6.0) algorithm to calculate the electromagnetic wave (light wave) and the arbitrary shape of the particle. It was written in Fortran by Draine & Flatau of Princeton University and runs on Linux. The program provides a variety of particle shapes (triangular prism, hexagonal prism, sphere, cuboid, cylinder, ellipsoid, and regular tetrahedron). For the particle shapes not set in the program, the user can customize the dipole array to generate this shape. The direction of polarization when light interacts with particles can be set by the angle. The parameters to be set in the program include the particle shape, effective radius of particles, number of dipoles, complex dielectric constant of particles varying with wavelength, etc. For Cu2O NS and Au and Ag NPs, the complex permittivity values of all simulation parameters are derived from Palik.[45]