Localized surface plasmon resonances, which arise from the collective oscillations of the near-Fermi-level electrons in noble metal nanostructures, have received intense attention in recent years due to their rich, intriguing, and complex optical properties. The localized plasmon resonances of Au and Ag nanostructures are spectrally located in the visible to near-infrared range. In addition, Au and Ag nanostructures are relatively stable in ambient environments. They have, therefore, been intensively studied from the perspectives of both fundamental sciences1–3 and potential applications in optics,4–6 sensing,7–10 imaging,11, 12 information processing,13 photothermal therapeutics,11, 12 solar energy harvesting,14, 15 and photocatalysis.16, 17
Gold nanocrystals, especially Au nanorods, have been widely used as detection agents for analytical and biochemical applications as well as photothermal therapy,11, 12 owing to their chemical stability, synthetically tunable plasmon resonance bands within the visible and near-infrared spectral regions, and facile surface functionalization using thiol-containing molecules. Moreover, due to their chemical stability and various well- developed preparation methods, Au nanocrystals have also been frequently employed in fundamental plasmonic studies, such as plasmon resonance coupling,3, 18 plasmonic−molecular resonance coupling,19 and plasmon-controlled fluorescence.20, 21 On the other hand, Ag nanocrystals have been mostly utilized to enhance Raman scattering and fluorescence. Their performance in enhancing Raman and fluorescence signals has been found to be generally better than that of Au nanocrystals.22–24 This is because Ag nanocrystals exhibit much stronger electric field enhancements than Au nanocrystals. Figure 1 shows the finite-difference time-domain (FDTD) simulation results of the electric field intensity enhancements of the same sized Ag and Au nanorods of 30 nm in length and 20 nm in diameter. The longitudinal plasmon wavelengths of the Ag and Au nanorods are 463 and 564 nm, respectively. The field intensity enhancement contours were calculated under the excitation at their longitudinal plasmon wavelengths. The field intensity enhancement factors along the central length axis are plotted in Figure 1C. They decay rapidly away from the nanorod surface. The enhancement factors of the Ag nanorod are 3.5−6.8 times those of the Au nanorod, depending on the distance from the nanorod surface. Besides the larger electric field enhancements, Ag nanocrystals also exhibit higher refractive index sensitivities25, 26 and larger solar energy conversion efficiencies than Au nanocrystals.14 Moreover, compared to gold, silver has a higher interband transition energy and is able to support localized plasmon resonances in the near-ultraviolet spectral range. Ag nanocrystals are therefore very attractive in various plasmonic applications. However, although the growths of Ag nanocrystals with highly symmetric shapes, such as spheres, cubes, octahedrons, and triangular plates, have been widely demonstrated, the preparation of elongated anisotropic Ag nanocrystals of uniform shapes and narrow size distributions has proven challenging. As revealed by their extinction spectra, previously prepared anisotropic Ag nanorods are not as good as Au nanorods in terms of the shape and size uniformity.27, 28 In addition, Ag nanorods are rather instable and change automatically into nanospheres in aqueous solutions.28, 29 Moreover, the synthetic control of the aspect ratio of Ag nanorods and thus their longitudinal plasmon wavelengths has remained difficult. These obstacles severely hinder the use of the attractive plasmonic properties of Ag nanocrystals in many plasmon-based applications.
Gold nanorods, owing to their attractive advantages, including facile growth methods, tunable longitudinal plasmon wavelengths, and chemical stability, can function as excellent supports for the formation of Ag shells. The resultant (Au core)−(Ag shell) nanorods are expected to exhibit plasmonic properties that are determined by silver. At the same time, the plasmon wavelengths of the core−shell nanorods can be tailored in the visible to near-infrared spectral regions by varying the Au core size and the Ag shell thickness. On the basis of this reasoning, (Au core)−(Ag shell) nanostructures in different shapes, including dumbbells,30, 31 rods,32–35 and octahedrons,36 have been prepared by different methods. Upon Ag coating, the lowest-energy plasmon band is generally blue-shifted and new plasmon bands are generated simultaneously at higher energies.32–35 The lowest-energy plasmon band is usually attributed to the longitudinal mode. The number of the newly generated plasmon bands is dependent on the shape of the core−shell nanostructures. However, the exact nature of the newly generated plasmon bands has not been investigated systematically and still remains controversial. Unraveling the nature of these plasmon bands is of vital importance because different plasmon modes exhibit distinct optical properties. For example, dipolar plasmon modes can radiate efficiently to the far field while quadrupolar plasmon modes only take effect in the near-field region mostly. Here we started with two differently sized Au nanorod samples and coated them with Ag shells systematically at varying thicknesses. The evolution of the plasmon bands, their peak wavelengths and extinction intensities as functions of the shell thickness was studied. The nature of each plasmon band was ascertained unambiguously with FDTD simulations. So far this has been the first systematical study on the plasmon resonances of (Au core)−(Ag shell) nanorods. Our results will be useful for applying (Au core)−(Ag shell) nanorods in the construction of optical devices and in various plasmon-enhanced spectroscopies.
The uncoated Au nanorod samples were prepared using a seed-mediated growth method in aqueous solutions and are stabilized with cetyltrimethylammonium bromide (CTAB). The longitudinal and transverse plasmon resonance wavelengths of the large Au nanorod sample dispersed in aqueous solutions are 868 and 512 nm, respectively, as revealed by the extinction spectrum shown in Figure 2A. The particle concentration of the large nanorod sample is estimated according to its extinction coefficient37, 38 at the longitudinal plasmon peak to be 2.3 nM. Figure 2B shows the transmission electron microscopy (TEM) image of the uncoated large Au nanorods. The nanorods have narrow shape and size distributions. Their average length, diameter, and aspect ratio determined from the TEM images are 73 ± 8 nm, 18 ± 2 nm, and 4.1 ± 0.6, respectively.
The (Au core)−(Ag shell) nanorods were prepared according to a previously reported method35 with slight modifications. Briefly, 2 mL of the uncoated Au nanorod solution was centrifuged and redispersed into an aqueous cetyltrimethylammonium chloride (CTAC) solution at the same volume. Varying volumes of aqueous AgNO3 and ascorbic acid solutions were then added to produce Ag coating at different thicknesses. Figure 2C−G show the TEM images of five representative (Au core)−(Ag shell) nanorod samples. The shapes and sizes of the nanorod samples are relatively uniform. As the amount of AgNO3 is increased, the (Au core)−(Ag shell) nanorods gradually change from a cylindrical to a cuboidal shape, with the edges and corners becoming sharper and sharper. The Ag shells at the side of the nanorods are seen to be thicker than those at the ends, resulting in a decrease in the aspect ratio. The extinction spectra of the (Au core)−(Ag shell) nanorod samples grown with increasing amounts of AgNO3 are shown in Figure 2A. For these extinction spectra, the dilution caused by the addition of the AgNO3 and ascorbic acid solutions have been corrected. Therefore, the particle concentrations in the solutions of all the samples are the same. Four plasmon resonance bands are observed. They become stronger as the supply of the Ag precursor is increased. We define the plasmon bands from low to high energies sequentially as peaks 1−4, respectively. The peak 1, which has been ascribed to the longitudinal dipolar plasmon mode in previous studies,33–35 blue-shifts. The peak 2, which has been ascribed to the transverse dipolar plasmon mode,33–35 first blue-shifts and then stays at a nearly constant wavelength. The peaks 3 and 4 appear after the amount of AgNO3 reaches a certain value. They only red-shift very slightly as the amount of AgNO3 is further increased. Since the plasmon resonance wavelength of a Ag nanocrystal is generally shorter than that of a Au nanocrystal of the same shape and size, the blue shift of the peak 1 can be understood as arising from both the reduction in the aspect ratio of the nanorods and the increasing effect of the optical properties of silver over those of gold. The initial blue shift of the peak 2 can be attributed to the coating of silver, where the material on the exposed nanorod surface is changed from gold to silver gradually.
To quantify the dependences of the plasmon resonance peak wavelengths and intensities on the sizes of the (Au core)−(Ag shell) nanorods, we measured the overall length and width as well as the Ag shell thickness of each sample from the TEM images. For each sample, about 150 particles were measured, their average values were calculated. Since the thickness of the Ag shell at the side is different from that at the ends, the shell thicknesses were measured at both the side and ends and averaged separately. The wavelengths and extinction intensities of the four plasmon resonance peaks were determined from the extinction spectra (Figure 2A). All of the obtained values are listed in Table S1 in the Supporting Information. Figure 3A shows the variations of the thicknesses of the Ag shell at the side and ends versus the volume of the AgNO3 solution for the core−shell nanorods prepared with the large Au nanorods as the cores. The thicknesses of the Ag shell at both the side and ends increase with increasing amounts of AgNO3, but the increase in the thickness of the Ag shell at the side is faster than that at the ends. This results in the reduction of the aspect ratio of the (Au core)−(Ag shell) nanorods.
Figure 3B illustrates the wavelength evolution of each plasmon band with the volume of the AgNO3 solution. The peak 1, initially located at 868 nm, blue-shifts first rapidly and then slowly to ∼590 nm. The peak 2, initially positioned at 512 nm, first blue-shifts and then red-shifts very slightly around 460 nm. The turning point occurs at a volume of 0.24 mL, where the peak wavelength is 458 nm. The average thicknesses of the (Au core)−(Ag shell) nanorods at the turning point are 5.0 and 2.5 nm for the Ag shell at the side and ends, respectively. As mentioned above, the initial blue shift of the peak 2 is mainly caused by the change of the exposed material from gold to silver. In addition, the peak 2 has been known to arise from the transverse dipolar plasmon mode. Therefore, the occurrence of the turning point suggests that the effect of gold on the plasmon bands is completely screened when the Ag shell becomes thicker than ∼5 nm. The peaks 3 and 4 appear on the extinction spectra when the AgNO3 volume reaches 0.12 mL, where the average thicknesses of the Ag shell at the side and ends are 3.0 and 2.0 nm, respectively. The appearance of these two peaks implies that the plasmonic properties of silver are manifested when the Ag shell gets thicker than ∼3 nm. As the Ag shell thickness is increased, the peaks 3 and 4 only show very small red shifts around 400 and 340 nm, respectively. The extinction intensities of all the four plasmon bands increase with increasing AgNO3 volumes (Figure 3C). The peak 2 is weaker than the peak 1 at small volumes of AgNO3. It becomes stronger than the peak 1 when the AgNO3 volume is above 0.72 mL, where the average shell thickness is 11.0 nm in the transverse direction and 6.0 nm in the longitudinal direction, even though the overall lengths of the core−shell nanorods are still larger than the overall widths. This result demonstrates that the (Au core)−(Ag shell) nanorods can support longitudinal and transverse dipolar plasmon modes with comparable intensities, which is usually not observable in other elongated metal nanostructures. This property makes the (Au core)−(Ag shell) nanorods useful for some particular plasmonic applications. For example, in plasmon-controlled fluorescence, because of the usual large Stokes shifts of the emission from the absorption of common fluorophores, the peak 2 can be used to enhance the excitation process and the peak 1 can be utilized to enhance the emission process simultaneously.20, 21
Figure 3D depicts the changes of the plasmon resonance wavelengths versus the aspect ratio of the nanorods. The peak 1 exhibits a nonlinear dependence of the peak wavelength on the aspect ratio. As the aspect ratio is reduced, the peak 2 first blue-shifts and then red-shifts very slightly. The shift behaviors of the peaks 1 and 2 verify again the role of the material change on the plasmonic properties of the (Au core)−(Ag shell) nanorods, because the longitudinal dipolar plasmon wavelength usually exhibits a linear relationship with the aspect ratio for nanorods made of a single metal.11 The peaks 3 and 4 exhibit very small red shifts as the aspect ratio is reduced. We also obtained the changes of the extinction intensities as functions of the particle volume (Figure 3E). The overall variation trends of the extinction intensities of the four plasmon bands versus the particle volume are very similar to those of the extinction intensities as functions of the AgNO3 volume. This result suggests that the supplied Ag precursor is converted to silver on the core−shell nanorods at a nearly constant conversion yield irrespective of the nanorod size.
We performed FDTD simulations to unravel the nature of the four plasmon modes of the (Au core)−(Ag shell) nanorods, especially the peaks 3 and 4. The average sizes of the sample 8 were chosen. The FDTD simulations were performed in a homogeneous medium of water, which has a refractive index of 1.33. The entire (Au core)−(Ag shell) nanorod was modeled as a cuboid with slightly rounded corners and edges, and the Au nanorod core was modeled as a cylinder capped with a hemi-ellipsoid at each end. Five polarization directions were considered in the simulations. They are along the longitudinal edge (LE), the longitudinal face diagonal (LD), the transverse edge (TE), the transverse face diagonal (TD), and the body diagonal (BD), respectively. The calculated absorption, scattering, and extinction spectra for each polarization direction are shown in Figure S1 in the Supporting Information. The peak 1 can be excited with the LE polarization, and the other three peaks can be excited with the TE or TD polarization. A comparison between the measured and calculated extinction spectra indicates that the experimental spectrum can be roughly reproduced by adding together the LE-polarized spectrum plus twice the TE-polarized spectrum or adding together the LE-polarized one plus twice the TD-polarized one (Figure 4A). The relatively large difference in the plasmon wavelength of the peak 1 can be ascribed to that the experimental spectrum is measured from the ensemble sample while the calculated ones are based on the average sizes. The plasmon wavelength of the peak 1, the longitudinal plasmon mode, is more sensitive to the nanorod size variation than those of the other three peaks (Figure 3B and D). We also find that the TE-polarized extinction spectrum appears slightly different from the TD-polarized one in the spectral region between the peaks 3 and 4 (Figure S1C and D, Supporting Information). Because of this difference, the BD-polarized spectrum is better reconstructed from the LE- and TD-polarized spectra than from the LE- and TE-polarized ones. In addition, there is a shoulder on the higher-energy side of the peak 3 on the extinction spectrum reconstructed from the LE- and TE-polarized ones. This shoulder is absent on the experimental spectrum (Figure 4A). The slight difference between the TE- and TD-polarized spectra is caused by the interference between the collective electron oscillations excited with the two perpendicular TE polarizations, which will be discussed below.
The nature of the different plasmon modes can be identified from their charge distributions. Figure 4B−E shows the charge distribution contours of the four plasmon modes. For the peak 1, the excitation is along the LE polarization. For the peaks 2−4, the excitation is along the TE polarization. Small amounts of charges are observed at the interface between the Au core and Ag shell. The presence of the interface charges has not been observed before. In order to find out whether the plasmonic properties are affected by the interface charges as well as by the Au core, we performed FDTD simulations on a pure Ag cuboid and a (H2O core)−(Ag shell) cuboid with the same sizes. The simulation results indicate that the extinction spectra of the pure Ag cuboid and the (H2O core)−(Ag shell) cuboid are very close to that of the (Au core)−(Ag shell) nanorod (Figure S2A and B, Supporting Information). Therefore, the interface charges and the Au core play a minor role on the plasmonic properties of the (Au core)−(Ag shell) nanorod. The plasmonic properties are mainly determined by the Ag shell. Figure S2C−F in the Supporting Information shows the charge distribution contours of the four peaks for the pure Ag cuboid. They look nearly the same as those of the (Au core)−(Ag shell) nanorod except the absence of the interface charges. According to the charge distribution contours, the peaks 1 and 2 can be unambiguously attributed to the longitudinal and transverse dipolar modes, respectively. The peaks 3 and 4 can both be assigned to an octupolar mode. In addition, the charge distribution contours obtained under the two perpendicular TD polarizations are also octupolar. Previous studies have shown that pure Ag nanocubes also exhibit four plasmon modes.26 In the order from low to high energies, the four plasmon modes of pure Ag nanocubes have been attributed to a dipolar, two quadrupolar, and an octupolar mode, respectively. The difference in the nature of the plasmon modes between the (Au core)−(Ag shell) nanorod and the pure Ag nanocube can be understood by considering the different symmetries possessed by a cuboid and a cube. The symmetry of a cuboid belongs to the D4h point group, whereas a cube has the Oh symmetry. The nature of the plasmon modes possessed by a metal nanocrystal has been known to be dependent on the symmetry of the nanocrystal.39, 40 In addition, four plasmon modes have also been observed on (Au core)−(Ag shell) nanocubes.41 They are ascribed to be dipolar, quadrupolar, multipolar, and multipolar, respectively, in the order of low to high energies, with the exact nature of the two high-energy modes unidentified.
As mentioned above, the calculated extinction spectra under the TE and TD polarizations exhibit a slight difference in the spectral region between the peaks 3 and 4. In order to understand this difference, we scaled down the core−shell nanorod to 9 nm × 5.2 nm. For simplicity, a pure scaled Ag cuboid was considered. The scaling-down technique has been proven to be effective in studying the eigenmodes of plasmon resonances, because it diminishes the damping and retardation effects caused by large nanocrystal sizes.42 Figure 4F shows the calculated extinction spectra of the small pure Ag cuboid excited under the LE, TE, and TD polarizations. Clearly, the LE-polarized light only excites the peak 1, and the other three peaks can be excited by the TE or TD polarization. The TE- and TD-polarized extinction spectra are different. Under the TE-polarized excitation, the spectrum is composed by one broad peak (peak 2) and two barely discernible weak peaks (peaks 3 and 4). In comparison, two distinct strong peaks (peaks 2 and 3) and one weak peak (peak 4) are observed under the TD-polarized excitation. Since the TD polarized excitation field can be decomposed into the two perpendicular TE-polarized fields with the same magnitude, the difference between the TE- and TD-polarized extinction spectra for the small Ag cuboid suggests that the interference between the collective electron oscillations along the two perpendicular TE directions is important for the observation of the peak 3. The peak 4 of the small Ag cuboid is very weak under both the TE- and TD-polarized excitation. The fact that the peak 4 of the (Au core)−(Ag shell) nanorod is stronger can therefore be reasonably ascribed to the retardation effect. To verify this speculation, we calculated the extinction spectrum of the small Ag cuboid under the TE-polarized excitation in a medium having a higher refractive index of 1.7. Under this situation, the peak 4 is considerably intensified (Figure 4F). Immersing a metal nanocrystal in a medium with a higher refractive index can increase the retardation effect due to the shortening of the excitation light wavelength. The retardation effect is therefore crucial for the clearly observable peak 4 on the (Au core)−(Ag shell) nanorod. Moreover, the charge distribution contours of the four peaks for the small Ag cuboid are nearly the same as those for the (Au core)−(Ag shell) nanorod (Figure 4G−J). This result further confirms our assignment of the different plasmon modes to the four peaks on the extinction spectra of the (Au core)−(Ag shell) nanorods.
We prepared an additional series of (Au core)−(Ag shell) nanorod samples using a small Au nanorod sample as the core in order to ascertain the dependence of the different plasmon modes on the sizes of the Au core. The transverse and longitudinal plasmon wavelengths of the small Au nanorod sample dispersed in aqueous solutions are 511 and 723 nm, respectively (Figure 5A). The particle concentration of the small Au nanorod sample is estimated according to the longitudinal plasmon peak extinction value to be 1.6 nM. Its average length and diameter are determined from the TEM images to be 48 ± 7 and 16 ± 1 nm, respectively (Figure 5B). The average aspect ratio is reduced from ∼4 for the large Au nanorod sample to ∼3 for the small one. Figure 5A shows the extinction spectra of the (Au core)−(Ag shell) nanorod samples that were grown from 2 mL of the small Au nanorod sample with increasing amounts of AgNO3. The evolution of the extinction spectra is very similar to that observed for the core−shell samples prepared from the large Au nanorod sample. Eventually four plasmon bands are clearly observed. The TEM images of three representative core−shell samples are provided in Figure 5C−E. The core−shell nanorods turn gradually from a cylindrical to a cuboidal shape, with the edges and corners becoming relatively sharper, as the supply of AgNO3 is increased. These results suggest that the growth behavior and the evolution of the plasmonic properties are relatively insensitive to the sizes of the starting Au nanorods within a certain range. On the other hand, we find that some small core−shell nanorods are oriented vertically on the TEM grids (Figure 5C−E). They look like nanocubes containing nanosphere cores under TEM imaging. The vertical alignment is rarely seen for the large core−shell nanorods. It is mainly due to the shorter lengths of the small core−shell nanorods. In addition, the Au cores are seen to be deviated from the center for many small core−shell nanorods. This deviation is not observed at all for the large core−shell nanorods. We also prepared another small Au nanorod sample and repeated the Ag coating. The deviation of the Au cores from the center was also observed. More experiments will be needed to understand the underlying reason for this deviation.
Figure 6 quantitatively illustrates the evolutions of the sizes, extinction peak wavelengths and intensities of the small (Au core)−(Ag shell) nanorod samples. As the amounts of AgNO3 is increased, the Ag shell at both the side and ends becomes thicker, with the thickness of the shell at the side increasing faster than that at the ends (Figure 6A). The wavelength of the peak 1, initially at 723 nm, decreases first rapidly and then slowly to ∼540 nm (Figure 6B). The peak 2, initially positioned at 511 nm, first blue-shifts and then red-shifts slightly around 435 nm. The turning point occurs at a AgNO3 volume of 0.24 mL, where the peak wavelength is 423 nm and the average shell thicknesses are 7.5 and 5.0 nm at the side and ends, respectively. This result suggests again that the effect of gold on the plasmonic properties is completely screened when the Ag shell thickness is larger than ∼5 nm. The peaks 3 and 4 appear when the AgNO3 volume reaches 0.12 mL, where the shell is 4.5-nm thick at the side and 2.5-nm thick at the ends. The wavelengths of the peaks 3 and 4 are ∼390 and ∼350 nm, respectively, nearly independent on the amount of AgNO3. A comparison of the peak wavelengths between the large and small (Au core)−(Ag shell) nanorods indicates that the wavelength of the peak 1 of the small core−shell nanorods is considerably shortened and that the wavelengths of the peaks 2−4 are respectively close to each other for the two series of core−shell nanorod samples. The extinction intensities of all the four peaks of the small (Au core)−(Ag shell) nanorods increase with the amount of AgNO3, with those of the peaks 1 and 2 increasing faster than those of the peaks 3 and 4 (Figure 6C). Figure 6D shows the plots of the wavelengths of the four peaks versus the aspect ratio. As the aspect ratio is reduced, the peak 1 exhibits a general blue shift, the peak 2 first blue-shifts and then red-shifts slightly owing to the material change at the exterior surface, and the peaks 3 and 4 remain nearly unchanged in terms of the spectral position. The extinction intensities of the four peaks all increases with the particle volume (Figure 6E). A comparison between the large and small (Au core)−(Ag shell) nanorods indicates that the overall variation trends for the shell thicknesses and the peak wavelengths are similar in the two cases, except that there are slight differences in the variation trends of the peak intensities.
In summary, (Au core)−(Ag shell) nanorods of varying Ag shell thicknesses were prepared using Au nanorods as the cores. Four plasmon bands are observed on the (Au core)−(Ag shell) nanorods. The evolution of the peak wavelengths and extinction intensities of the plasmon bands were studied quantitatively as functions of the nanorod sizes and the supplied amount of the Ag salt. Electrodynamic simulations reveal that the lowest-energy peak belongs to the longitudinal dipolar plasmon mode, the second-lowest-energy peak is the transverse dipolar plasmon mode, and the two highest-energy peaks can be ascribed to octupolar plasmon modes. The retardation effect and the interference between the excitations along the two perpendicular transverse-edge directions are important for the observation of the highest-energy and second-highest-energy octupolar plasmon modes, respectively. There are charges at the interface between the Au core and Ag shell, but the Au core does not affect the plasmonic properties of the core−shell nanorods when the Ag shell thickness is more than ∼5 nm. Generally, as the Ag shell becomes thicker, the longitudinal plasmon mode blue-shifts, the transverse plasmon mode first blue-shifts and then red-shifts slightly, and the two octupolar plasmon modes remain nearly unchanged in terms of their spectral position. The extinction intensities of all the four plasmon bands increase with the overall particle size. The results obtained from our study will help in understanding the localized plasmon resonances of (Au core)−(Ag shell) nanorods. They will also be useful for utilizing the different plasmon modes of (Au core)−(Ag shell) nanorods either individually or simultaneously in various plasmon-based applications.