Figure 1 illustrates a step-by-step fabrication procedure for AuNP@BDMT@AuNSs sensors. Au thin film is deposited on the Si surface via sputtering then subjected to thermal annealing to produce 2D AuNSs. After BDMT functionalization of the AuNS surface, AuNPs are immobilized on the BDMT@AuNSs. BDMT has a chain length of approximately 1 nm and therefore provides a vertical nanogap between the AuNSs and AuNPs29.
Figure 2 shows the morphological, chemical, and optical characteristics of AuNPs according to the different growing steps. As shown in Fig. 2a, the diameter of the AuNPs increased from 6.3 ± 1.6 nm (AuNP0), to 16.1 ± 4.5 nm (AuNP1), to 34.7 ± 3.6 nm (AuNP2), to 44.9 ± 4.3 nm (AuNP3), to 50.2 ± 3.7 nm (AuNP4) according to the different growing steps. The lower-magnification TEM images of AuNPs were depicted in Fig. S2. To demonstrate the chemical elements of the AuNPs, EDS analyses were conducted. The line-scan analysis across the AuNPs proved that Au L-edge distribution was well defined as shown in Fig. 2b. Figure 2c shows the diameter of AuNPs as an increase of the growing steps. The detailed size distributions of AuNPs were depicted in Fig. S1. The TEM image analysis of the AuNPs was performed using freeware ImageJ (ver. 1.53e) program30. The optical properties of the AuNP solutions measured with UV-vis spectroscopy are displayed in Fig. 2d. The absorption peak denotes the photon absorption due to the LSPR of the AuNPs, and the red-shift is observed as a successive particle growth, which is consistent with Mie theory (inset of Fig. 2d)31.
Since the diameters of the AuNP3 and the AuNP4 are similar as depicted in Fig. 2c, subsequent experiments were performed except for the AuNP3. Figure 3 shows the morphological and chemical properties of the AuNP@BDMT@AuNSs. Figures 3a and b present SEM images of the AuNSs and the AuNP2@BDMT@AuNSs, respectively. After the thermal annealing process of the Au thin film, the larger grain size in the film were formed and as a result, 2D AuNSs were obtained. Following the immobilization of the AuNPs, the AuNPs including AuNP0, AuNP1, and AuNP2 were rather uniformly distributed on the BDMT@AuNS surface as shown in Fig. 2b and d-f. However, the AuNP4 with a diameter of ~ 50 nm was not well immobilized on the BDMT@AuNS surface (Fig. S3). In order to verify the functionalization of the well-ordered BDMT SAMs, the surface-enhanced Raman spectrum of the BDMT@AuNSs was measured. The characteristic peak at 1586 cm-1 can be identified as the ring stretching mode, and those at 1218 and 1167 cm-1 as CH2 wagging mode and βC-H vibration mode, respectively32,33. Figures 3d–f show AFM images of the AuNP0@BDMT@AuNSs, the AuNP1@BDMT@AuNSs, and the AuNP2@BDMT@AuNSs, respectively. The 3D topographic image and height profile along the line were inserted in each AFM image. From the AFM images, it can be seen that AuNPs including AuNP0, AuNP1, and AuNP2 were well immobilized on the surface of BDMT@AuNSs.
To evaluate the SERS performance of the proposed AuNP@BDMT@AuNSs, the SERS signal of the AuNP@BDMT@AuNS sensor was measured using R6G at a range of concentrations (10− 5 to 10− 9 M). The R6G spectra exhibited strong peaks for vibrational bands at 611, 773, 1181, 1310, 1360, 1507, 1574, and 1648 cm-1, corresponding to the Raman characteristic peaks of R6G, as shown in Fig. S4 and Table S2. Figure 4a presents the Raman spectra for the AuNSs and the AuNP@BDMT@AuNSs with differently sized AuNPs at 10 µM R6G. It is obvious that the Raman intensities of the AuNP2@BDMT@AuNSs sensor are strongly enhanced, which is about eight times higher than that of the AuNSs sensor. Figure 4b shows the SERS spectra of the AuNP2@BDMT@AuNSs sensor at R6G concentrations of 10− 5 to 10− 9 M. Figure S5a-c show the SERS spectra of the AuNSs, the AuNP0@BDMT@AuNSs, and the AuNP1@BDMT@AuNSs sensors at R6G concentrations of 10− 5 to 10− 7 M, respectively. Figure 4c presents the Raman intensity at 611 cm-1 as a function of the logarithmic concentration of R6G for the AuNSs, the AuNP0@BDMT@AuNSs, and the AuNP2@BDMT@AuNSs sensors (SERS data of the AuNP1@BDMT@AuNSs sensor is separately presented in Fig. S5d for clarity). For the AuNP2@BDMT@AuNSs sensor, the signal increase (ΔI) of the Raman peak at 611 cm-1 shows a good linear relationship with R6G logarithmic concentration CR6G, which can be expressed by having an R2 of 0.986. The SERS signal of the AuNP2@BDMT@AuNSs sensor at 611 cm-1 increased by about 12 times compared to the AuNSs sensor at 100 nM of R6G concentration. The limit of detection (LOD) was computed according to the formula of where meanblank and SDblank are the average and the standard deviation of (ΔI) without analyte, respectively, and SDanalyte is the standard deviation of (ΔI) for the lowest analyte concentration measured 34. The obtained LOD was about 10-10.07 M (~ 84 pM). To investigate the reproducibility of the AuNP2@BDMT@AuNSs sensor, the Raman intensity at 611 cm-1 was evaluated for 11 different regions at 10 µM of R6G concentration as shown in Fig. 4d, and the relative standard deviation (RSD) was about 5.3%. Figure S6a shows the Raman intensity at 773 cm-1 as a function of the logarithmic concentration of R6G for the AuNP2@BDMT@AuNSs sensors with an R2 of 0.931, and the intensity distribution was depicted in Fig. S6b. The RSD was about 5.2%.
To evaluate the electric field distribution of the proposed AuNP@BDMT@AuNSs sensor, the finite element method (FEM) simulation was conducted using COMSOL Multiphysics (ver. 4.3a). Figure 5a presents the simulation set-up. The AuNP with different diameters of 6.3 nm (AuNP0), 16.1 nm (AuNP1), and 34.7 nm (AuNP2) and a 1 nm gap was placed on the Au thin film/Si substrate along the z-direction. The light with a wavelength of 532 nm polarized to x-axis was incident on the sensor along the z-axis. Figure 5b-d shows the electric field distributions of the AuNP0, the AuNP1, and the AuNP2 on the Au thin film/Si substrate, respectively. It can be clearly observed that highly localized electric fields can be generated in the small gaps between the Au thin film and the AuNP. Furthermore, as the diameter of the NPs increases, the localized electric fields in the nanogap are enhanced.
To demonstrate the practical utility of our proposed sensing platform, we investigated the SERS performance of the AuNP2@BDMT@AuNSs with thiram. Figure 6a shows the SERS spectra of the AuNP2@BDMT@AuNSs sensor at thiram concentrations of 10− 3 to 10− 7 M. Figure 6b presents the signal increase (ΔI) at 558 cm-1 as a function of the logarithmic concentration of thiram described by with the R2 of 0.972. The LOD was found to be ~ 13 nM. Compared to the recently reported SERS substrates, this LOD value is compatible to that of other SERS substrates as depicted in Table S3. In addition, the LOD value is about 1000 times lower than the maximal residue limit of 7 ppm (~ 29 µM) in fruit 35. These results indicate that our proposed SERS platform shows satisfactory performance, and suggests an alternative approach for highly sensitive SERS sensor.