Plasmonic metal nanostructures have received much research attention due to their size- and shape-dependent surface plasmon resonance (SPR)-derived optical properties ^[1–5]. These unique SPR effects show great promise for a variety of applications in biomedical engineering ^[6,7] and sensor technology ^[8,9]. The SPR extinction spectrum consists of two accumulated parts: absorption and scattering. The intensities of the scattering and absorption depend on the size and composition of the nanostructure in question ^[10,11]. Noble metal (such as Au and Ag) nanoparticles have excellent optical features with regard to extinction, resulting in strong electromagnetic (EM) fields and high sensitivity as surface-enhanced Raman scattering (SERS) substrates.

SERS is a powerful and attractive spectroscopic technique for the sensitive detection of low-concentrated analyte molecules on metal surfaces, which benefits greatly from the strong EM enhancement associated with the molecular excitation generating local plasmonic fields on such surfaces ^[12–14]. SERS is highly dependent on the interaction between the adsorbed molecules and the surface of plasmonic nanostructures. Currently, EM enhancement can be directly observed on metal nanoparticles with SPR in the visible near-infrared (Vis-NIR) region, such as Au ^[15], Ag ^[13], and Cu nanostructures ^[16,17]. This optical flexibility and strong EM field enhancement lend several advantages to plasmonic nanoparticles. For example, plasmonic nanoparticles have been widely used as sensors ^[18,19] because their SPR frequency can be greatly affected by slight changes to the surrounding medium ^[20]. The EM field of plasmonic nanoparticles is useful in enhancing different EM signals, such as Raman scattering ^[21], fluorescence ^[22], and Rayleigh scattering ^[10]. For this reason, SERS is a research topic of great interest in nanotechnology and allows for important applications in materials and biological sciences.

The physicochemical properties of metal nanostructures are strongly correlated with their shapes. Motivated by fundamental research and technical interest, plasmonic nanostructures with different shapes have been prepared to study these shape effects ^[23–27]. One such structure, gold nanorods (AuNRs), has been successfully synthesized, and its utility has been demonstrated in diverse applications ranging from catalysis to plasma and optoelectronics ^[28,29]. As face-centered cubic structured noble metals, Au and Ag crystals have very similar lattice constants (4.0786 and 4.0862 Å, respectively). More recently, Ag nanostructures have attracted extensive interest for their potentially superior optical properties ^[30]. Although both the intensity of SERS and the quality factor of SPR for Ag nanoparticles (AgNRs) ^[31] are higher than those of AuNRs, research regarding AgNRs has seldom been reported due to issues of quality and yield ^[32–34], which severely limits further fundamental studies and technical applications. Therefore, it is important to develop effective methods for fabricating highly pure and size-controlled AgNRs, and there is still much room to explore in the study of their physicochemical properties.

In this paper, we adopt an improved method for the preparation of high-purity and size-controlled Au/AgNRs (core/shell) through anisotropic Ag overgrowth on Au nanobipyramids (AuNBPs). The purification process of Au/AgNRs is simplified and conducted directly by using a static precipitation method, without an additional step of redispersing in some auxiliary organic solvent (CTAB) ^[35]. Au/AgNRs with different lengths/diameters can be produced by changing the amount of Ag precursor and the size of the AuNBP core. The formation progress of the Au/AgNRs was investigated, and the influence of AuNBP on the SPR properties of the Au/AgNR was discussed based on the experimental results and finite-difference time-domain (FDTD) simulations. The tunable plasmon properties of the Au/AgNRs, FDTD simulated results on the electric field enhancement, and their superior SERS are described.

Sodium tetrahydridoborate (${\mathrm{NaBH}}_4$, 99%), hydrochloric acid (HCl, 37%), and ascorbic acid (AA, $\ge 99.7\%$) were obtained from Sinopharm Chemicals. Cetyltrimethyl ammonium bromide (CTAB, 99%) and hexadecyltrimethyl ammonium chloride (CTAC, 99%) were obtained from Nanjing Robiot Co. Ltd. Silvernitrate (${\mathrm{AgNO}}_3$, 99.8%) and tetrachloroaurate (${\mathrm{HAuCl}}_4·4{\mathrm{H}}_2\mathrm{O}$, 99.9%) were obtained from Shanghai Chemicals. Trisodium citrate (${\mathrm{Na}}_3{\mathrm{C}}_6{\mathrm{H}}_5{\mathrm{O}}_7·2{\mathrm{H}}_2\mathrm{O}$) was obtained from Chengdu Kelong Chemicals. Deionized water (18.25 MΩ) was used in all experiments. All reagents were used without further purification.

AuNBPs were prepared using a seed-mediated method ^[35–37]. Briefly, the seed solution was prepared by addition of fresh, chilled ${\mathrm{NaBH}}_4$ (0.15 mL, 0.01 mol/L) to an aqueous solution composed of ${\mathrm{HAuCl}}_4$ (0.125 mL, 0.01 mol/L), trisodium citrate (0.25 mL, 0.01 mol/L), and water (10 mL) with vigorous stirring. The resultant seed solution was aged for 2 h before use. The growth solution was obtained by adding ${\mathrm{AgNO}}_3$ (0.4 mL, 0.01 mol/L), ${\mathrm{HAuCl}}_4$ (2 mL, 0.01 mol/L), HCl (0.8 mL, 1 mol/L), and AA (0.32 mL, 0.1 mol/L) into 40 mL of a 0.1 mol/L CTAB aqueous solution. For the preparation of AuNBPs of different sizes, various amounts of seed solution (0.1–2 mL) were added to growth solutions under gentle stirring for 30 s. The products were centrifuged at 6500 rpm for 10 min. To synthesize Au/AgNRs ^[25,38,39], the as-prepared AuNBPs were redispersed in a CTAC solution (12 mL, 0.08 mol/L), followed by subsequent addition of ${\mathrm{AgNO}}_3$ growth solution (0.5–2 mL, 0.01 mol/L) and ascorbic acid (0.25–1 mL, 0.1 mol/L). The reaction system was placed in an oscillator and shaken for 4 h at 333 K, during which Ag grew on the surface of the AuNBPs to form Au/AgNRs.

The as-prepared Au/AgNRs colloid was purified through an effective and simple method, conducting directly by a static precipitation method without centrifugation or filtration process. During the static process, the Au/AgNRs precipitated naturally to the bottom of the container without the need of centrifugation or filtration due to electrostatic aggregation, while the by-products (AgNRs etc.) and excess chemicals remained dispersed in the supernatant.

Absorption spectra of the samples were collected using an ultraviolet Vis-NIR spectrometer UV-6300 (200–1100 nm) and UV-3600 (200–3300 nm). Clean samples were deposited on copper grids covered by an amorphous carbon film and micro grids for transmission electron microscopy (TEM: JEOL-100CX) and high-resolution TEM (JEOL-2011) measurements.

Rhodamine-6G (R6G) and as-prepared AgNRs were used as the probing molecule and SERS active substrate, respectively. To prepare the SERS active substrate, the as-prepared Au/AgNRs were dispersed in R6G solutions of different concentrations (${10}^{-6}–{10}^{-12}\mathrm{m}\mathrm{o}\mathrm{l}/\mathrm{L}$). After 1 h of gentle shaking, the mixture was sealed and stored in the dark for 5 h, followed by drying under ${\mathrm{N}}_2$ flow. For SERS measurements, the samples were scanned for 5–20 s with a 785 nm laser (200 mW).

FDTD Solutions (Lumerical, Inc.) was utilized to study the near-field EM responses and to evaluate the plasmon-resonant properties of the metal by solving Maxwell’s curl equations on a discrete grid. Figure 1 shows the simplified calculation model of the Au/AgNRs. In the simulation, the AuNBP was modeled using two circular cones, and with spherically rounded tips, the two cones were connected to each other at their bottom bases ($t=2\mathrm{nm}$). The diameter (${\mathrm{D}}_1$) and length (${\mathrm{L}}_1$) of the AuNBP were set to 15–35 nm and 45–105 nm, respectively. The AgNR was modeled as a cylinder with two hemispherical caps with the AuNBP in the center. The diameter (${\mathrm{D}}_2$) and length (${\mathrm{L}}_2$) of the Au/AgNR were 20–40 nm and 120–400 nm, respectively. The total-field scattered-field source was launched into the boundary to simulate a propagating plane wave interacting with the target, where the wavelength of the incident light was 300–2500 nm with a 1 nm mesh size. The dielectric functions of the AuNBP and the AgNR were adopted from Johnson and Palik, and the electric field polarizations were perpendicular to the short and long axes of the Au/AgNR. The surrounding medium was water with a refractive index of 1.33. Through adjusting the parameters of the model, absorption spectra and near-field enhancement of the Au/AgNR were obtained.

A simple and effective method for separating Au/AgNRs from a mixture of nanorods and nanoparticles is presented using a CTAC-assisted approach, which is quite different from typical centrifugation with the addition of an organic solvent ^[40] or extra surfactant ^[41]. As a commonly used surfactant in the growth of noble metal nanorods, CTAB is difficult to dissolve and easy to crystalline under room-temperature because of low solubility, which is not favorable for further sample purification and characterizations. The obtained sample was directly purified, avoiding the precipitation of dissolved CTAB and simplifying the purification step. The key point of this purification strategy lies in the aggregation potential between Au/AgNRs and nanoparticles, which is dominated mainly by the electrostatic repulsion between the ${\mathrm{CTA}}^{+}$ bilayer along the Ag surface and steric exclusion ^[42]. On the other hand, Au/AgNRs with semicurved geometries more easily come into contact with each other compared to spherical or spherical-like nanoparticles having greater curvature. For these two cases, Au/AgNRs spontaneously aggregated into precipitates, while the by-products (such as amorphous nanoparticles) were well dispersed in the supernatant solution. Detailed synthetic instructions for these Au/AgNRs are shown in Fig. 2.

Figure 3(A) shows a TEM image of the as-obtained AuNBPs prepared using our seed-mediated method. Particles of highly uniform shape and size ($\sim 25\mathrm{nm}$ in width and $\sim 75\mathrm{nm}$ in length) are obtained, together with several spherical Au nanoparticles. Anisotropic Ag overgrowth was carried out on the surfaces of the as-prepared AuNBPs using different amounts of ${\mathrm{AgNO}}_3$ (80, 110, 200, 1000 μL) in the presence of CTAC, as shown in Figs. 3(B)–3(E). Highly pure Au/AgNRs were obtained by taking advantage of electrostatic aggregation and shape effects, separated directly by the static precipitation in the growth solution. The obtained Au/AgNRs are monodisperse and highly uniform with AuNBP cores. Statistical analysis of the Au/AgNR dimensions (based on the TEM images) shows that the diameter of the Au/AgNRs does not change obviously ($30\pm 2\mathrm{nm}$), while the length increases from 150 to 800 nm with increasing amounts of Ag precursor.

Figure 4(A) shows the absorption spectra of AuNBPs and Au/AgNRs corresponding to the TEM images in Fig. 3. The SPR at $\sim 530\mathrm{nm}$ corresponds to the transverse (${\mathrm{SPR}}_{\mathrm{T}}$) mode of AuNBPs and of spherical nanoparticles that may be present, while the SPR at $\sim 785\mathrm{nm}$ is ascribed to the longitudinal (${\mathrm{SPR}}_{\mathrm{L}}$) mode of AuNBPs, as shown by curve a in Fig. 4(A). Curves b–e in Fig. 4(A) show the corresponding absorption spectra of Au/AgNRs obtained after overgrowth of Ag on the AuNBPs. As the amount of ${\mathrm{AgNO}}_3$ used increases, new resonance peaks appear in the region $>500\mathrm{nm}$. The absorption shoulder at $\sim 350$ and $\sim 390\mathrm{nm}$ corresponds to the plasmon resonance mode of bulk Ag ^[43,44] and the transverse Au/AgNRs, respectively; ${\mathrm{SPR}}_{\mathrm{T}}$ undergoes no obvious shifts. The resonance peaks that redshift from 830 to 1335 nm are ascribed to the dipole ${\mathrm{SPR}}_{\mathrm{L}}$ of the Au/AgNRs. In addition to this strong dipole ${\mathrm{SPR}}_{\mathrm{L}}$ mode, weak multipolar ${\mathrm{SPR}}_{\mathrm{L}}$ is present at 550–800 nm [curves d and e in Fig. 4(A)].

To better understand the formation of Au/AgNRs and the detailed changes of their SPR properties, optical spectra of Au/AgNRs obtained using different Ag precursor volumes were recorded, as shown in Figs. 4(B) and 4(C). When less ${\mathrm{AgNO}}_3$ is added, ${\mathrm{SPR}}_{\mathrm{T}}$ does not obviously shift, while ${\mathrm{SPR}}_{\mathrm{L}}$ blueshifts from $\sim 745$ to $\sim 680\mathrm{nm}$ with increasing ${\mathrm{AgNO}}_3$ up to 30 μL [see curves b–e in Fig. 4(B)], indicating that the Au core dominates the absorption spectra ^[36,45]. The observed blueshift is possibly caused by the increased electron density of the AuNBPs produced by the enhanced restoring force ^[46–48]. Further increasing the ${\mathrm{AgNO}}_3$ contents from 35 to 140 μL causes the ${\mathrm{SPR}}_{\mathrm{L}}$ to redshift from $\sim 680$ to $\sim 1060\mathrm{nm}$ [curves a–k in Fig. 4(C)], corresponding to the ${\mathrm{SPR}}_{\mathrm{L}}$ mode of Au/AgNRs with increasing aspect ratio (AR). The intensity of the new absorption peak at $\sim 390\mathrm{nm}$ increases in this same manner, which is ascribed to the ${\mathrm{SPR}}_{\mathrm{T}}$ of Au/AgNRs; this indicates that the Ag shell is dominant when more than 35 μL ${\mathrm{AgNO}}_3$ is used. Meanwhile, an absorption shoulder appears at $\sim 350\mathrm{nm}$, which is ascribed to the plasmon resonance mode of bulk Ag.

In addition, the diameters of the Au/AgNRs can be controlled using AuNBPs of different sizes. The obtained Au/AgNRs are also uniform in length with average diameters of $\sim 15$, 20, 25, and 30 nm, as shown in Fig. 5, wherein the longitudinal plasmon wavelengths of AuNBPs can be tuned from 680–830 nm.

Figure 6(A) shows the FDTD-calculated SPR data (${\mathrm{SPR}}_{\mathrm{T}}$ and ${\mathrm{SPR}}_{\mathrm{L}}$) of different AuNBPs based on the core sizes in Fig. 5, wherein ${\mathrm{D}}_1/{\mathrm{L}}_1$ are 15/45 nm, 20/60 nm, 25/75 nm, and 30/90 nm, respectively. The simulated results indicate that the ${\mathrm{SPR}}_{\mathrm{L}}$ redshifts significantly with increasing the AuNBP size (solid line b–e), which is accordant with the experiment results. The calculated spectrum of AuNBPs with ${\mathrm{SPR}}_{\mathrm{L}}$ at $\sim 765\mathrm{nm}$ agrees well with the experimental results (dashed line a). The clear ${\mathrm{SPR}}_{\mathrm{T}}$ at $\sim 540\mathrm{nm}$ in the experimental result is likely due to the presence of AgNRs in these samples.

Figure 6(B) shows the FDTD-calculated optical absorption spectra (including ${\mathrm{SPR}}_{\mathrm{T}}$ and ${\mathrm{SPR}}_{\mathrm{L}}$) of Au/AgNR; here, the AuNBP is 75 nm in length and 25 nm in diameter, while the diameter and AR of the Au/AgNR are 30 nm and 4–10, respectively. The simulated results indicate that the ${\mathrm{SPR}}_{\mathrm{T}}$ of the Au/AgNR is located at $\sim 435\mathrm{nm}$ with no obvious shift at different ARs. Conversely, ${\mathrm{SPR}}_{\mathrm{L}}$ redshifts from 700 to 1680 nm as the Au/AgNR AR increases from 3 to 10, with an accompanying enhancement in intensity. The ${\mathrm{SPR}}_{\mathrm{L}}$ absorption intensities are about twice that of ${\mathrm{SPR}}_{\mathrm{T}}$, similar to pure AgNRs ^[10]. The relationship between the AR and the ${\mathrm{SPR}}_{\mathrm{L}}$ position of Au/AgNR has a linear trend, which agrees well with the experimental vibration trend. Moreover, some multiple plasmon resonance peaks (500–750 nm) can be observed for Au/AgNR of high AR, in accordance with the experimental results [Fig. 4(A)]. The differences in ${\mathrm{SPR}}_{\mathrm{L}}$ position and spectral width between the experimental and FDTD results potentially arise from the inevitable differences between the two setups, such as surrounding medium, structural model, and AR distribution.

Additionally, based on the schematic in Fig. 1, the influence of the different size dimensions of the AuNBP core (such as ${\mathrm{D}}_1$, $L_1$, and $\alpha$) on the ${\mathrm{SPR}}_{\mathrm{L}}$ properties of the AgNRs was calculated by FDTD, to compare AgNRs with and without the AuNBP core. The results display that the geometric changes to the AuNBP core (such as size, cone angle or the position) have little effect on the plasma behavior of the Au/AgNRs, as shown in the inset of Fig. 6(B). The simulated spectra (absorption, scattering, and extinction) and field distributions of the Au/AgNRs are consistent with those of pure AgNRs ^[40]. Therefore, it is safe to treat the stable Au/AgNRs as pure AgNRs with poor stability in plasma ^[32].

SERS benefits greatly from the strong electric field enhancement at regular tips and asperities due to the strong power dependence. To determine the Raman enhancement ability of Au/AgNRs, the Raman band intensities of R6G absorbed on Au/AgNR surfaces were measured.

Figure 7(A) shows the concentration-dependent SERS spectra of R6G excited by a 785 nm laser with 5 s accumulation times and 200 mW powers. SERS spectra were obtained using samples immersed in solutions with R6G concentrations of ${10}^{-6}–{10}^{-12}\mathrm{m}\mathrm{o}\mathrm{l}/\mathrm{L}$. SERS spectra of R6G at different molar concentrations produce several well-resolved Raman signals at $607.9{\mathrm{cm}}^{-1}$ (C-C-C in-plane bending), $769.1{\mathrm{cm}}^{-1}$ (C-H out-of-plane bending), $1181.5{\mathrm{cm}}^{-1}$ (C-H in-plane bending), $1308.3{\mathrm{cm}}^{-1}$ (C-O-C stretching), and at 1361.6 and $1510.8{\mathrm{cm}}^{-1}$, which correspond to C-C stretching modes for aromatic ring vibrations [Fig. 7(A)]. Slight Raman shifts in these spectra arise from the weak interaction between the analyte R6G molecule and Au/AgNRs substrate, as shown by the inset in Fig. 7(A), which examines the band at $1510.8{\mathrm{cm}}^{-1}$. The plot of intensity versus log (R6G concentration) [Fig. 7(B)] shows the gradual decrease of intensity with decreasing R6G concentration, in good agreement with previous reports of similar systems ^[49].

Despite the low laser power (200 mW) and short exposure time (5 s), the Au/AgNRs displayed excellent SERS performance. Additionally, Raman enhancement using Au/AgNRs is stronger than that of pure AuNRs of similar size, indicating that Au/AgNRs are far more SERS sensitive than AuNRs. Studies on the high stability of Au/AgNRs and further surface functionality are currently being planned.

Au/AgNRs have been successfully synthesized with a stepwise combination of seed-mediated growth of AuNBPs with Ag overgrowth. Highly pure Au/AgNRs can be obtained through static precipitation in the growth solution, and Au/AgNRs with different lengths/diameters can be produced by changing the amount of Ag precursor and the size of AuNBP core. The formation of the Au/AgNRs was investigated, and the influence of AuNBP on the SPR properties of Au/AgNR was discussed based on both experimental results and FDTD simulations. The AuNBP core has a minimal effect on the plasma behavior of Au/AgNRs. Compared with homogeneous AuNRs or AgNRs with tunable SPR properties, the heterogeneous Au/AgNRs not only exhibit excellent SPR properties but are also highly stable. The unique optical properties and strong EM effects of Au/AgNRs, along with their superior SERS signal enhancement, show that Au/AgNRs are an attractive candidate for plasmonic applications in plasmon-enhanced spectroscopies, biomolecular detection, and plasmon sensors.