Fabrication and characterization of SERS composites
Ag nanoparticles used in this study were prepared in aqueous solution by reduction of Ag(I) salts with strong reducing agent sodium borohydride and subsequently phase transferred in organic solution. SERS probes were constituted by dense layers of Ag nanoparticles immobilized on transparent rigid (glass) and flexible (PDMS) substrates from self-assembled chlorobenzene solutions. The use of organic solutions was motivated by previous studies carried out by our group and others showing that direct deposition of aqueous plasmonic colloidal solutions on glass substrates resulted in dis-homogeneous nanoparticle distribution (i.e. accumulation of nanoparticle along the perimeter of the droplet), due to coffee stain effects [27, 28]. In contrast, homogeneous arrays were obtained by surface transfer of self-assembled superstructures obtained from evaporation of organic droplets. The process of SERS probe fabrication is schematically shown in Fig. 1.
A small droplet (10 μL) of concentrated Ag nanoparticle in chlorobenzene solution was deposited on a clean glass substrate and left undisturbed for 5–10 min (1a). Following the evaporation of small amount of solvent, Ag nanoparticles spontaneously self-assembled at the solvent/air interface, as evidenced by formation of a metallic lustre (1b). At this point a receiving substrate (glass or PDMS) was brought in contact with the nanoparticle droplet (1c). Capillary effects promoted the transfer of the metallic layer on the receiving substrate (1d). Following transfer, the glass/PDMS substrate containing Ag nanoparticles was allowed to dry and subsequently immersed overnight in isopropanol to remove unbound particles and excess surfactants present in the organic solution. Figure 1e, f show photographs of Ag nanoparticles deposited on glass and PDMS substrates, respectively.
Figure 2 show SEM images of (a) as-synthesized Ag nanoparticles in aqueous solutions showing formation of spherical particles with average size of 15 ± 5 nm and (b) Ag nanoparticles transferred on glass substrates from self-assembled chlorobenzene droplets, following the method described in Fig. 1. The images showed that particles retained their spherical size through the phase transfer/self-assembly/substrate transfer process and that the final average size of SERS substrate deposited particles was 20 ± 3 nm. More details on UV–vis response and size distribution are reported in Additional file 1: Figures S1, S2).
Minimal invasiveness tests
Prior attempting SERS analysis on real works of art tests were performed in order to establish the capability of fabricated SERS structures for minimal-invasive analysis. To this end, Ag nanoparticles/glass and Ag nanoparticle/PDMS substrates were repetitively placed for few minutes on marked areas of commercial white paper sheets. Close contact between the Ag nanoparticle/glass substrates and the analytical white paper sheet was obtained by placing two glass slides on the sides of the SERS substrate (see Fig. 4a). During SERS measurements this procedure ensured that the substrate was held in place and also maintained close contact between the Ag nanoparticle moiety and the analytical surface necessary for the SERS effect to occur. Optical microscopy images of white paper sheets were taken prior SERS probe deposition (Fig. 3a) and after each deposition (Fig. 3b–f) to assess the accidental release of loosely bound nanoparticles on the analytical surface. As shown by Fig. 3b, the first Ag nanoparticle/glass substrate placement (deposition 1) on the paper substrate left nanoparticle residues; Fig. 3 shows that four repetitive placements had to be performed before no residual traces of Ag nanoparticles could be detected on the paper sheet. In contrast, when Ag nanoparticle/PDMS substrates were placed on the paper no residual traces of Ag nanoparticles were observed already after the first placement (Fig. 3g, h). It is important to stress that Ag nanoparticle/PDMS substrates were applied to the analytical surface just by slight pressure. The substrates showed good adhesion with the paper surfaces and, in contrast with the Ag nanoparticle/glass substrates, did not require the use of additional weights to hold the SERS substrate in place or to maximize SERS effects. These data show that the adhesion of Ag nanoparticles to PDMS was stronger than the adhesion to glass substrates. One reason could be the strong hydrophobic affinity between PDMS and the hydrophobic ODA molecules crafted on the surface of Ag nanoparticles following their phase transfer in chlorobenzene. It should be pointed out that the strong adhesion of the Ag nanoparticle/PDMS composite to the paper analytical surface was enough to ensure close contact between the Ag nanoparticles and the analytical surface to promote the SERS effect. This effect was due to the use of PDMS and was promoted by the PDMS/Ag nanoparticle area ratio in the composite. When transferred on PDMS the Ag nanoparticles covered an approximately circular area of 5 mm in diameter. However, the size of the used PDMS was larger (ca. four times in size). That means that the adhesion to the substrate was not provided by the nanoparticle area but by the PDMS area, which stuck to the surface and ensured adhesion of the overall composite.
SERS tests with Ag/PDMS composites
The capability of Ag nanoparticle/PDMS composites to perform SERS analysis was initially assessed by taking SERS spectra of model molecule 4-aminobenzene thiol (4-ABT). This particular molecule was selected the occurrence of a SERS effect is easily assessed by the occurrence of additional b2 in plane, out of phase vibrational modes not present in the Raman spectrum and attributed to a SERS chemical enhancement effect (plasmonic metal–molecule charge transfer process) [29]. The spectrum of 4-ABT was taken by placing the composite on a dry droplet (1 × 10−6 M, MeOH) of 4-ABT on a glass slide with the Ag nanoparticle facing the 4-ABT. SERS spectra were obtained by back laser illumination and showed characteristics a1 and b2 modes of 4-ABT (see Additional file 1: Figure S5 for detailed description of the 4-ABT SERS spectrum obtained with Ag nanoparticle/PDMS composites).
SERS analysis on model ink-colored paper samples
In parallel, capabilities of fabricated substrates for in situ SERS analysis of commercial BIC ballpoint inks were tested. All spectra were taken in situ, by applying the Ag nanoparticle/glass substrates on pen colored squares with the Ag nanoparticle facing the analytical surface followed by direct back laser illumination (see Fig. 4a for photographic details of the setup). All spectra were background subtracted. Figure 4b–d show Raman and SERS spectra of colored squares drawn on commercial paper by Medium red, Crystal purple and Medium green BIC ballpoint pens, respectively. A bench size Raman spectrograph with 514 nm excitation wavelength was used for analysis of the red and purple squares. Both colored squares showed featureless Raman spectra saturated by fluorescence. The generation of high fluorescence interference was not surprising as UV–vis spectra (see Additional file 1: Figure S3) showed that both red and purple colored inks were characterized by strong absorption centered at 554 nm, close to the selected excitation wavelength of 514 nm, which resulted in concomitant generation of interference fluorescence emission, and consequent masking of Raman signals. In contrast, the SERS spectra of red and purple colored squares showed dramatically enhanced spectral features (Fig. 4b, c blue lines). The red square showed clear presence of Rodhamine B (Pigment Violet 1, C.I. 45,170), evidenced from the presence of the following vibrational bands: 1647 (C–C bending and C=C stretching of xanthene aromatic ring), 1530, 1505 (aromatic C–H bending), 1398–1278 (aromatic C–C bending), and 620 (xanthene ring puckering) cm−1 [30]. Rhodamine B and/or rhodamine 6G had been previously identified in red ballpoint pen formulations through Easy Ambient Sonic-spray Ionization Mass Spectrometry (EASI-MS) analysis [31], and Laser Desorption/Ionization Time of Flight Mass Spectrometry (LDI-TOF-MS) [32]. Interestingly, while clear distinction between Rhodamine B and Rhodamine 6G could not been achieved with mass spectrometry techniques, the reported SERS analysis clearly showed presence of Rhodamine B in this particular Medium red BIC ink formulation (Fig. 4c, blue line). The SERS spectrum of the Crystal purple square displayed spectral features similar to the red square (main bands at 1650, 1505 and 1358 and 620 cm−1), therefore also suggesting presence of Rhodamine B in the ink mixture. However, also small additional bands centered at 1623 (stretching of benzene rings), 938, 914 (asymmetric stretching and bending of C–Ccenter–C bonds), 565, 525, 443 and 420 (bending of the C–N–N bonds) cm−1 were observed, indicative of the presence of triarylmethane dye crystal violet (Methyl violet 10B, CI 42,555) in the ink mixture [33]. For comparison reference spectra of Rodhamine B and crystal violet are shown in Fig. 4b. The presence of crystal violet was further confirmed by the intense purple color of the ink, its occurrence in blue and black BIC pen ink formulations, and by previous Raman analysis performed on the Crystal purple ink at other excitation wavelengths [19]. It is important to stress that the blank SERS spectrum recorded by placing a Ag nanoparticle/glass substrate on a white paper substrate was featureless and did not display any spectral feature attributable to the excess of stabilizing agents used to promote Ag nanoparticle phase transfer from aqueous to chlorobenzene phases (see Additional file 1: Figure S4).
Previous Raman analysis carried out on colored BIC pens showed that green inks possess good Raman response at 514 nm illumination wavelengths, making the application of SERS not necessary for identification and analysis [19]. However, in view of analytical restrictions imposed by the preciousness and limited mobility of works of art, often requiring the use of portable instrumentation, it is important to assess the capabilities of low-cost handheld Raman instrumentation for analysis of artistic drawings. Figure 4c shows the Raman and SERS spectra of a green colored square measured with a handheld Raman spectrograph with excitation at 785 nm. The Raman spectrum of the Medium green pen ink resulted featureless. However, the SERS spectrum showed strong enhanced spectral features with peaks centered at 1539, 1342, 758 and 499 cm−1. From previous work carried out in our group, it is known that green inks are often constituted by a mixture of blue and yellow dyes and that blue phthalocyanine Blue 38 dye is used in BIC pen formulations [15]. In this case, identification of the yellow component could not be achieved with portable instrumentation. However, the presence of phthalocyanine Blue 38 was confirmed from peaks at 1344 and 1454 cm−1, which were both attributed to internal vibrations of copper phthalocyanine (CuPh) macrocycle [19]. Figure 4e, f show optical images of colored papers taken after removal of SERS probes following analysis showing that no residue was left on the analytical surface.
SERS analysis on real works of art
Finally, Ag nanoparticle/glass and Ag nanoparticle/PDMS composites were used for SERS analysis of real works-of-art. Figure 5a show photograph of a drawing made in red BIC ballpoint pen by contemporary French artist Anne-Flore Cabanis. The peculiarity of the drawing was its overall circular pattern constituted by one continuous line bent at 90° angles. As the pen was never lifted from the paper during the drawing, the continuous line showed areas of high ink accumulation, generated by the occasional harder pressure of the pen on the paper at resting and corner points, respectively. In order to maximize spectral response, the laser was focused on one of these high ink concentration areas for SERS analysis. SERS spectra were taken in situ, by applying the Ag nanoparticle/PDMS substrate on the drawing with the Ag nanoparticle facing the analytical surface followed by direct back laser illumination. The SERS spectrum of Cabanis’ red drawing is shown in Fig. 5b and was characterized by one strong band at 1648 cm−1 and medium intensity bands at 1532,1505, 1429, 1352, 1282, 1198 and 623 cm−1 all attributed to Rodhamine B. Figure 5c shows a photograph of a Japanese wood print characterized by areas of strong purple and red colorations. Figure 5d show SERS spectrum of the Japanese print marked purple area recorded by application of Ag nanoparticle/PDMS substrate followed by in situ back laser illumination. The SERS spectrum showed clear bands attributable to Crystal violet at 1622 and 1587 cm−1 (stretching of benzene rings), 1171 and 912 cm−1 (asymmetric stretching and bending of C–Ccenter–C bonds), 442 and 419 cm−1 (bending of the C–N–N bonds). Rodhamine B and Crystal violet reference spectra were also added to Fig. 5b, d, respectively.