The silver (>99% pure) and copper targets (>99% pure) with thickness of 1 mm, tetrachloroauric (III) acid (HAuCl₄ 4H₂O) ≥99.9% trace metals basis, methanol (reagent grade), acetone (reagent grade), methylene blue (C₁₆H₁₈ClN₃S·xH2O) were purchased from Sigma Aldrich. The explosive molecules of 2,4,6-trinitrophenol (PA), 2,4-dinitrotoluene (DNT) were provided by HEMRL (Pune, India) for SERS detection.

Pure Ag and Cu substrates were cleaned with acetone, ethanol and deionized water sequentially in an ultrasonic cleaner for 15 min to remove any organic dopants from the surface. Then the samples were ablated with a regenerative fs amplifier (Ti: Sapphire, LIBRA) delivering ~50 fs at 800 nm with a repetition rate of 1 kHz. An input pulse energy of ~300 μJ was used for the ablation experiments. The input laser beam (diameter of ~14 mm) was focused onto the sample using a convex lens with a focal length of 100 mm. The intensity profile of the laser pulses was Gaussian, with an estimated <140 μm beam waist on the target surface. We have estimated the effective spot size (beam waist) according to the method proposed by Barcikowski et al. [12]. This value is slightly over-estimated since the ablation process in the present case is a result of several overlapping pulses. Accurate value of the spot size will be obtained when the ablation is performed with a single pulse. Therefore, we believe that the fluences estimated using these values are slightly under-estimated. Since we are not quantitatively measuring any quantity in this work, the reported values of fluences could be considered with an error of ±20% due to the uncertainty in the spot size estimation. The target (Ag/Cu) was placed at the bottom of the glass dish and was covered with a layer of 10 ml HAuCl₄ (5 mM) with liquid height above the sample being 6 mm. The surface of the target was adjusted by observing the intense plasma plume generated by the focused laser pulses and by listening to the sound produced. Verification experiments were performed initially using a He-Ne laser to exactly obtain the plane where the focus is located. However, when we shift to 800 nm, there will be a small change in the focal plane. The Ag and Cu targets immersed in a liquid were placed normal to the laser beam and is rotated to avoid deposition of NPs onto a substrate and is expected to result in a higher production rate of the NPs in the liquid. We observed that the liquid in which the ablation was carried out turned into dark pink and dark brown color due to the replacement reaction between ablated Ag/Cu ions from target and Au ions from the precursor solution. Following the ablation procedure, the colloidal solutions were centrifuged at 3,500 rpm for 15 min. Later, the NPs were washed with deionized water for several times and dispersed in 5 ml distilled water using in an ultrasonic bath for 5 min. The prepared NPs were preserved in air tightened glass vials and stored to prevent the oxidation.

The formation of bimetallic NPs was confirmed by studying the absorption spectra using an UV–Visible absorption spectrophotometer (Perkin Elmer, Lambda 750). The X-ray diffraction patterns were recorded using a powder X-ray diffractometer (Discover D8 diffractometer of Bruker, Germany) with Cu Kα radiation (λ = 0.15408 nm). The NPs morphology was investigated using TEM (FEI Tecnai G2 S-Twin) with an electron-accelerating voltage of 200 kV. Further, elemental mapping on the fabricated NPs was achieved using high-resolution energy dispersive x-ray spectroscopy (EDS) unit connected to field emission scanning electron microscope (FESEM) (Carl Zeiss model Merlin compact 6027) operated with beam voltage 30 kV.

The analyte molecules 2,4,6-trinitrophenol (Picric Acid, PA dissolved in methanol), 2,4-dinitrotoluene (DNT, dissolved in acetone) and a dye methylene blue (MB, dissolved in methanol) were prepared in stock solution with the concentration of 0.1 M. The lower concentrations of analyte molecules (10⁻³ to 10⁻⁹ M) were prepared by further dilution of the stock solution.

Silicon (Si) was taken as a base to prepare SERS substrate because the metal NPs are tightly immobilized to prevent any aggregation. Before drop casting the NPs, Si wafers were cleaned with acetone and deionized water in an ultrasonic bath for 15 min. The SERS substrates were prepared by dropping the synthesized Ag@Au and Cu@Au NPs (typically 10 μl) on a cleaned Si wafer and was subsequently dried on the hot plate at 50°C. Later the analyte molecules such as PA (5 μM)/2, 4-DNT (1 μM)/MB (5 nM) were drop costed (typically 10 μl) on NPs films (NPs were earlier drop casted on Si surface). The Raman spectra were acquired with a portable Raman system (B&W Tek, USA) operated with a laser source at 785 nm and a maximum power of 300 mW. The collected SERS data was an average of 3 spectra and the integration time 5 s. Before the Raman measurements the intensity of the Raman peak at 520 cm⁻¹ from silicon was normalized before each data acquisition.

Figure 1 illustrates the normalized optical absorption spectra of the prepared NPs, and insets show the coloration of NPs. The appearance of a single SPR peak in the UV–Visible absorption spectrum clearly indicates the bimetallic Ag@Au and Cu@Au NPs formation. The absence of two plasmon peaks in the spectrum indicates that these NPs were not a mixture of individual metal NPs and also there is no core-shell formation [36, 37]. The SPR peak of Ag@Au NPs was observed at ~566 nm along with a redshift and broadening. This is not an intermediate position of pure Ag (~400 nm) and pure Au (~520 nm) NPs and this could be attributed to the variation in NPs shapes/sizes [38] and the concentration of precursor solution [6]. The SPR peak shift is primarily dependent on the size, shape and composition of nanoparticles [39]. The SPR peak of Cu@Au NPs was observed at 551 nm which was within the SPR peaks of Cu (~620 nm) and Au (~520 nm) NPs. The broad absorbance indicates the presence NPs with a wide size distribution by non-spherical shape, surface morphology and possible formation of nanoaggregates. Based on the previous reports, the absorption peak of bimetallic NPs shifted to a longer wavelength side due to the (i) concentration of precursor [32], (ii) irradiation energy, and (iii) time of exposure. Xu et al. fabricated the bimetallic Ag@Au nanocrystals and observed the red shift in the SPR peak (~858 nm) achieved using laser-induced photo-oxidation. Kuladeep et al. have recently demonstrated the formation of metal alloy NPs under direct laser irradiation of metal precursor solutions and observed a shift in absorption peak with different concentrations [40].

For XRD measurements, the prepared NPs were drop cast on the Si wafer and dried using a hot plate. Figure 2 shows XRD spectra of synthesized bimetallic Ag@Au and Cu@Au NPs. The bimetallic phase of the NPs was confirmed by XRD peaks. In Figure 2 red color spectra indicate the XRD pattern of Ag@Au NPs and the main peaks are located at 38.4°, 44.6°, 64.8°, and 77.7°. These peaks are assigned to the crystal planes of (1 1 1), (2 0 0), (2 2 0) and (3 1 1). Due to the similarities in lattice constants of Ag and Au [Au (0.40786 nm) and Ag (0.40862 nm)] one cannot distinguish the phases of Ag@Au alloy/bimetallic NPs based on only XRD data (JCPDS file number Ag: 03-0921, Au: 04-0784). Figure 2 depicts the XRD spectra of Cu@Au NPs (green color) with the peaks observed at 38.3°, 44.6°, 64.7°, and 77.7° are attributed to the Au planes of (1 1 1), (2 0 0), (2 2 0) and (3 1 1) [JCPDS file: 04-0784].The other peaks positioned at 41.1°, 50.5°, 74.3° were corresponding to crystal planes (1 1 1), (2 0 0), and (2 2 0) of Copper (JCPDS file: 03-1005). The additional peaks of 27.9° and 31.1° show the presence of CuClO₄ (JCPDS: 38-0594). The FESEM EDAX mapping also revealed that the presence of chlorine in Cu@Au NPs. XRD spectra from the drop cast NPs (i.e., NPs on Si) shows only some peaks with the small shift with respect to that of powder due to the preferred orientation of the lattice planes of the sample surface.

The morphology and crystallographic phases of Ag@Au and Cu@Au NPs was studied with TEM and HRTEM patterns. Figures 3A–D illustrate the TEM & HRTEM images of bimetallic Ag@Au and Cu@Au NPs, respectively. From the low magnification images of NPs presented in Figures 3A,C it is evident that the some of the particles are in quasi-spherical and few of them are interconnected with a chain like structure. This could be attributed to the high-pressure laser-induced plasma near the target-liquid surface. Lattice plane separations of Ag@Au NPs obtained through the HRTEM images are depicted in Figure 3B with an interplanar spacing of 0.14 nm for Ag, 0.204 nm for Au. Similarly, the d-spacing values of Cu@Au NPs were 0.13 nm, 0.207 nm for Cu and 0.223, 0.23 nm for Au NPs (in Figure 3D), respectively. The corresponding SAED patterns are shown in the inset of Figures 3B,D. The possible mechanism for the growth of bimetallic Ag@Au and Cu@Au NPs can be explained as follows. Briefly, the bimetallic NPs are controlled by laser-induced replacement reaction, rapid nucleation and growth processes. The fs laser pulses interact with precursor (HAuCl₄) resulting in the formation of Au ions because of the higher reduction potential of AuCl⁴⁻ species (0.93 V). When an intense laser pulse interacts with the solid target (Ag/Cu) huge amount of electrons are pulled out due to the inverse Bremsstrahlung and thermionic emission effects. Due to this the material will be short of electrons leading to the presence of excess ions. The ejected electrons oscillate and collide with atoms of target transferring some of the energy to surrounding lattice. At the focal point, plasma (atoms, clusters, electrons, ions, etc.) formed as a consequence of the ablated target surface heats up suddenly and vaporizes. Simultaneously, the pressure wave (shockwave) propagates below and above the target surface. As the time progresses, plasma plume cools down and transmission of energy to the surrounding medium triggers the formation of the cavitation bubble [containing both the ablated matter (Ag/Cu) and the liquid vapor (Au)], which expands to a maximum radius (typically ~1 mm) which then consumes the energy ending up in a collapse. The nanoparticles are ejected into the surrounding liquid following the collapse and nanostryctures are leftover on the target. With the presence of extreme conditions of high pressure and high temperature for a short period of time there is possibly a replacement reaction between ablated Ag and Au ions. The nucleation and growth rapidly occur leads to the formation of bimetallic/alloy NPs [41]. Due to high-intensity of the laser pulse the formation of bimetallic/alloy NPs is favored while suppressing the formation of metal segregation.

To reveal the elemental distribution of synthesized bimetallic NPs, Energy Dispersive X-ray Spectroscopy (EDS) technique was employed along with FESEM characterization. In EDS, each element has a unique atomic structure with a unique set of X-ray emission peaks were produced by the interaction of a highly energetic electron beam with the sample. Figures 4A–F shows the FESEM analysis of Ag@Au NPs, formed due to the laser-induced replacement reaction between the ablated Ag ions ejected from the target and Au ions from the precursor solution. The EDS mapping conducted on a single particle is depicted in Figures 4C–E. The images clearly illustrate that the formation of bimetallic Ag@Au NPs wherein Ag (green color) is seen to be decorated on the surface of Au (red color). The line scan on a single particle is indicated in Figure 4F, which reveals that the NP is composed of Au rich center and decorated by Ag NPs. The Au mole fraction in bimetallic NPs depends on the concentration of the HAuCl₄ solution. The atomic ratio of Ag@Au determined by FESEM - EDS measurements for the particle is 38:61 and the data shown in Figure 4B. The high weight percentage of gold may play a lead role in the stability of NPs. Due to the laser-induced replacement reaction between Cu and Au ions, the bimetallic Cu@Au NPs can possibly form. FESEM image and EDS mapping of Cu@Au NPs are depicted in Figures 5A–F, in which Cu and Au elemental mapping are shown in red and green colors (shown in Figures 5C–E) and we also observed a small amount of chlorine (data shown in Figure 5B). We believe that this could be due to the interaction of fs pulses with gold precursor solution resulting in the formation of Au, Cl radicals in the gold salt solution (HAuCl4). The presence of chlorine in Cu@Au NPs could possibly be due to the fast exchange reaction between the copper and chlorine compared to nobel metals (Ag, Au), where the latter are least reactive. This also supports the absence of chlorine in Ag@Au NPs. However, further detailed studies are essential to clearly identify the mechanism of this formation. EDS mapping revealed that the center rich Au covered Cu NPs are homogeneously distributed over the whole surface and the atomic ratio of the particle is 17:78 (Figure 5B). Thus, the EDS mapping results confirmed the formation of Ag@Au and Cu@Au alloys.

For practical applications reproducibility is one of the essential parameters for fabricated SERS substrates. The reproducibility of the SERS substrates was investigated by collecting the SERS signals from >10 different sites on Ag@Au and Cu@Au NP films. Figures 8A,B depict the SERS spectra of MB (5 μM) and their corresponding histogram. The relative standard deviation (RSD) of SERS intensity for 1,622 cm⁻¹ peak is 7.45% for Ag@Au NPs. Similarly, the reproducibility of Cu@Au NPs substrate was examined with DNT (1 μM) and the histogram of Raman peak at 849 cm⁻¹ shown in Figures 8C,D. The 3D SERS spectra are provided in the supplementary information file (Figure S1). The estimated relative standard deviation (RSD) for SERS intensity of 849 cm⁻¹ modes is ~12%. These RSDs indicates the reproducibility and suggest that these Ag@Au and Cu@Au NPs SERS substrates were good candidates for detection. The observed small variation in SERS intensity could be attributed to the morphology of NPs and the distance between the analyte and NPs. Even higher reproducibility can be achieved by placing these NPs in periodic templates of nanostructures.

Pure metal (Ag and Cu) NPs are easily oxidized under ambient conditions and, therefore, reduce their efficacy. We believe that the presence of Au in these bimetallic NPs (SERS substrates) displays stable properties. To test the sensitivity of Ag@Au and Cu@Au NPs substrates a series of SERS measurements were conducted for PA and MB (from 5 mM to 5 μM and 5 μM to 5 nM, respectively). Figures 9A,C illustrate the variation in the SERS intensity with varying analytes [(a) PA (b) MB] concentration and recorded using Ag@Au NPs. The SERS intensity decreased as the concentration of analyte decreased and further we did not notice any shifts in the Raman modes. The Raman modes for MB were clearly identified even at the concentration of 10⁻⁹ M with a good signal-to-noise ratio. Further, statistical analysis of the data revealed that there exists a linear dependence between the logarithmic concentrations (log C) and the logarithmic intensity (log I) (for the characteristic peak of PA - 829 cm⁻¹ and MB - 1,622 cm⁻¹) using Ag@Au NPs (shown in Figures 9B,D). The data presented in the graph confirms the linearity at lower concentrations. However, at higher concentrations a slight saturation was observed and the reasons for this needs to be investigated. The observed SERS signals are related to the hotspots due to closely packed plasmonic metal NPs. We believe that Ag@Au has taken the advantages of the strong plasmonic electric field enhancement of Ag and the chemical stability of Au beneficial for SERS capability.

Figures 10A,C show the concentration-dependent SERS studies with the analyte [(a) PA (c) MB] using Cu@Au NPs. The SERS spectral intensity decreases as the concentration decreases. The small shifts in the characteristics peaks are due to the interaction of analyte molecule with the substrate, affected by the orientation of the adsorbed molecule [47]. The Raman modes for MB are clearly recognized even at the concentration of 10⁻⁹ M with a good signal-to-noise ratio. Furthermore, statistical analysis of the data revealed that here too there exists a linear dependence between the logarithmic concentrations (log C) and the logarithmic intensity (log I) (for the characterstic peak of PA - 821 cm⁻¹ and MB - 1,620 cm⁻¹) using Cu@Au NPs (shown in Figures 10B,D). This graph confirms a linear correlation between SERS intensities and logarithmic concentration within the concentration range investigated.

The enhancement factor (EF) was calculated using the relation EF=ISERSIRCRCSERS, where I_SERS Raman Intensity using NPs, I_R is the Raman intensity on Si wafer (without using NPs), C_SERS the concentration of sample on NPs substrate (low concentration) and C_R is the concentration of sample on Si wafer (0.1 M) which produce Raman signal (I_R). For the reference, the Raman spectra of PA, DNT and MB with concentration 0.1 M on Si substrate was recorded and the plots were presented in the supplementary information (Figure S2). The calculated EFs for Ag@Au NPs and Cu@Au NPs are 3.45 × 10⁷, 1.44 × 10⁷ for MB (1,622 cm⁻¹ mode), 8.22 × 10⁴, 4.47 × 10⁴ for PA (829 cm⁻¹ mode) and 7.5 × 10⁵, 6.0 × 10⁵ for DNT (855 cm⁻¹ mode), respectively. The calculated EFs for all the molecules are summarized in Table 1. The comparison plot of obtained EFs for PA, DNT, and MB recorded from Ag@Au and Cu@Au substrates is depicted in Figure 11. From the presented results, Ag@Au NPs demonstrated more than two orders of magnitude stronger EFs than the Cu@Au NPs with best detection amount in terms of mass being 16 pg of MB, 11.45 ng of PA and 1.82 ng of DNT respectively. The detection limit of MB (nM) is higher compared to the explosive molecules (μM) and this could possibly be due to partial attachment of the explosive molecules electromagnetically with the NPs [47]. The EF critically depends on the analyte molecule adsorption, orientation of molecules on the NP surface, NP surface morphology (nanostars usually demonstrate superior performance compared to spherical nanoparticles), the availability of number of hotspots and the excitation wavelength. Our earlier work reported the detection of 25 μM DNT (explosive molecule) and 5 nM MB [30] in the case of Ag-Au NPs and nanostrcutures (NSs) prepared through the ablation of Ag-Au targets in acetone. Similarly, we could achieve the detection of explosive molecules of PA at 5 μM concentration levels and ammonium nitrate at 50 μM concetration levels using gold NPs and NSs [30]. In the present work, though, we achieved the detection of 1 μM DNT and 5 μM PA. The EFs obtained in the present case could possibly be further improved by increasing the number of hot spots through simple procedures such as (a) drop casting more number of NPs of same type or of different metals (b) drop casting these NPs in nanostructured metals or even Si/SiO₂ and subsequently perform the SERS measurements. Our earlier report [30] has already demonstrated one of this possibility. This is conceivable since the NPs will be localized on these NSs and by increasing the concentration of the present NPs one could possible achieve higher number of hotspots. Our initial results on similar idea (present set of NPs on Si nanostructured surface) have indeed demonstrated superior SERS enhancements. The results of those studies are beyond the scope of this work. It is well established that pure Ag/Cu NPs have been reported to be easily susceptible to oxidation and aggregation in air. The presence of stable Au in bimetallic NPs composition with its rich and excitation properties leads to stable and reproducible SERS substrates which are favorable for on-site and field-based detection of a variety of molecules in combination with portable Raman spectrometers.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.