Abstract
This work reports on highly efficient surface enhanced Raman spectroscopy (SERS) constructed on low-cost, fully recyclable and highly reproducible cardboard plates, which are commonly used as disposable packaging material. The active optical component is based on plasmonic silver nanoparticle structures separated from the metal surface of the cardboard by a nanoscale dielectric gap. The SERS response of the silver (Ag) nanoparticles of various shapes and sizes were systematically investigated, and a Raman enhancement factor higher than 106 for rhodamine 6G detection was achieved. The spectral matching of the plasmonic resonance for maximum Raman enhancement with the optimal local electric field enhancement produced by 60 nm-sized Ag NPs predicted by the electromagnetic simulations reinforces the outstanding results achieved. Furthermore, the nanoplasmonic SERS substrate exhibited high reproducibility and stability. The SERS signals showed that the intensity variation was less than 5%, and the SERS performance could be maintained for up to at least 6 months.
Export citation and abstract BibTeX RIS
1. Introduction
When metal nanoparticles (MNPs) are excited by electromagnetic (EM) radiation, their free electrons collectively oscillate, resulting in a localized surface plasmon resonance (LSPR). At this resonance, the MNPs can produce particularly intense localized near-field and propagating far-field scattering; however, these features depend intrinsically on the size, shape and spacing of the MNPs as well as on their surrounding environment [1–3]. MNP-far-field scattering has recently been considered a promising light-trapping mechanism since it can improve the optical path length in photovoltaic devices [2, 4]. Surface-enhanced Raman spectroscopy (SERS), on the other hand, takes advantage of near-field scattering, which is mainly associated to the enhanced local electric field intensity in the MNPs' vicinity [5, 6]. Such localized field enhancement can lead to highly amplified Raman scattering signals at the surface of the nanoparticles, resulting in an increase of the Raman signals from molecules that have been adsorbed onto or are in the vicinity of the nanometer-sized metallic particles [7].
Currently, the most frequent method for producing MNPs with nanoscopic controlled geometry and interparticle distance is the employment of advanced lithographic processes such as e-beam lithography [8]. However, this method has crucial drawbacks, such as a high patterning time and elevated costs, which limit its extensive use in macroscopic scale systems. Alternative large-scale processes, such as chemical synthesis [9–12], physical deposition with post-deposition annealing (PDA) [4] or template and electrodeposition [13], are also commonly applied. However, several processing steps are required despite the fact that PDA, for example, requires elevated temperature conditions. Merlen et al [14] showed the potential of fabricating MNP structures at room temperature using a sputtering technique with a short deposition time (a few seconds) under a primary vacuum without post-annealing. However, this process does not allow for fine control of the size, shape and organization of the nanoparticles, which hinders the direct correlation between the NPs' morphology and the resulting SERS signal intensities; however, it can be advantageous for the fabrication of novel SERS nanostructures such as metal-coated dielectric nanorod arrays [15].
The methodology employed in this work (thermal evaporation assisted by an electron beam) results in the direct arrangement of individual nanoparticles with good control of their size and shape without any post-deposition thermal procedures. This methodology consists of one step only. Silver (Ag) NPs are formed in situ during the thermal evaporation of thin films of silver onto cardboard packaging substrates heated at 150 °C. The low-temperature nanoparticle formation method developed in this work is largely applicable to flexible substrates, for instance made of polyethylene (Text S1 of the Supporting Information available at stacks.iop.org/NANO/25/415202/mmedia), which would not be compatible with standard post-annealing dewetting methods carried out at high temperatures (300–500 °C) [4, 16, 17]. Nevertheless, suitable substrate heating is imperative in order to favor material surface diffusion during deposition and to avoid worm-like structures [4, 18]. This process was revealed to be relatively simple and inexpensive for the preparation of SERS substrates with improved stability, uniform morphology and pronounced Raman enhancement effects throughout large-scale areas.
Intense efforts have been made to find cost-efficient SERS substrates such as paper substrates [11, 19–24]. In fact, paper has been widely used as a low-cost platform for medical diagnostics [25–27], analysis and/or quality control devices [28, 29].
In this work, we developed a new kind of cost-efficient SERS substrate. This particular substrate is an aseptic and biodegradable material that is widely used in several applications such as food packaging, and it is also used in the beverage industry. The cardboard substrate, like common paper, is cost-efficient, highly sensitive and amiable to several different environments and target analytes. In addition, it is extremely robust when compared to common paper substrates. By using this device, we demonstrate the feasibility of nanoplasmonic cardboard substrates for reproducible and stable SERS substrates with a tunable visible resonance. By using Rhodamine 6G (R6G) as a Raman probe, we show that Ag nanoplasmonic cardboard substrates prepared by an alternative one-step thermal evaporation method for arrangement of individual SERS active nanoparticles exhibit a very strong SERS activity.
2. Experimental
2.1. Chemicals
The tetraethylrhodamine hydrochloride (Rhodamine 6G) was purchased from Sigma-Aldrich. The R6G was used as received without further purification. Distilled water passed through a Millipore water system (ρ = 18.2 MΩ), which was used in all of the experiments.
2.2. Substrate
The cardboard packaging substrates used in this study consist of pressed cellulose fibers, polymeric coatings and evaporated aluminium. The polymeric portions of the paper promote the proper adhesion of the metalized layer to the cellulose and protect the thin metalized layer from scraping or other environmental damage. Moreover, a thin native oxide (AlxOy) layer is present on top of the aluminium layer. The presence of such a surface oxide is essential for the formation of metallic nanoparticles by the solid-state dewetting method.
2.3. Preparation of silver nanoparticles
Ag NPs were obtained by the deposition of metal layers directly on the cardboard substrate (2.5 × 2.5 cm2) using an electron gun-assisted thermal evaporation technique. The substrate was kept at 150 °C during the thermal evaporation, while the thicknesses of the Ag films ranged from 2 to 12 nm. The deposition was carried out with a working pressure of 10−5 mbar and a deposition rate of 0.07 nm s−1. The thicknesses and growth rates of the films were inferred by a calibrated quartz crystal detector.
2.4. Preparation of samples for SERS measurements
R6G was chosen as a model analyte to investigate the performance of the AgNPs-coated cardboard substrate for SERS detection because R6G has been extensively studied and well characterized by SERS [19, 24, 30, 31]. The SERS substrates were prepared by dropping 100 μl of R6G solution with a concentration of 10−6 M onto the substrates.
2.5. Characterization
2.5.1. Morphological characterization
The scanning electron microscopy (SEM) observations of the cardboard substrate before and after the NPs deposition were carried out using a Carl Zeiss AURIGA CrossBeam (FIB-SEM) Workstation. The surface roughness of the cardboard substrate and the average height of the NPs were measured with an Asylum MFP3D atomic force microscope (AFM) in ac mode. The statistical parameters of the NPs were derived from the digital analysis of the SEM-based images, which was conducted using the ImageJ [32] software package. The crucial point of the analysis was the determination of the contrast threshold that defines the distinction between the particles and the background. The accuracy of this approach was estimated at two pixels. The longest in-plane axis is called particle size (D) for simplicity.
2.5.2. Optical characterization
The absorptance spectra of the cardboard substrates after the NPs deposition and the following rhodamine 6G (R6G) adsorption were obtained with a double-beam UV–vis-NIR spectrometer (Lambda 950) equipped with an integrating sphere in the range of 350–1500 nm.
2.5.3. Raman measurements
To characterize the SERS activity of the cardboard SERS substrates, a drop (100 μl) of R6G with a concentration of 10−6 M was cast onto the substrates. Raman measurements were carried out with a Labram 300 Jobin Yvon spectrometer equipped with an air-cooled CCD detector and a HeNe laser operating at 17 mW and at 632.8 nm. The spectra were recorded as an extended scan. The laser beam was focused with a 50× Olympus objective lens. The laser power at the surface of the samples was fixed with the aid of a neutral density filter (optical densities 0.3). All of the measurements were made with five scans of 25 s laser exposure. The 521 cm−1 peak of a silicon wafer was used between the different Raman sessions to calibrate the spectrograph for possible fluctuations of the Raman system.
3. Results and discussion
As demonstrated in previous studies, MNPs have a high SERS effect [9, 13, 30, 33, 34]. So, in this work, we fabricated the metal nanostructures to enhance the Raman signal of the R6G by thermal evaporation assisted by the electron beam method using an alternative SERS substrate. Scanning electron microscopy (SEM) and atomic force microscopy (AFM) analyses of the cardboard substrates before and after the NPs deposition were carried out (figure 1). Figures 1(A) and (B) show the surface morphology of the cardboard substrates before the Ag NPs deposition. The surface roughness of the cardboard substrate lies in the micrometer range. Nevertheless, at the nanoparticle level, the root mean square (RMS) is 2.37 (±0.05) nm, as determined by AFM (figure 1(B)).
Figure 1. (A) SEM and (B) AFM images showing the cardboard packaging surface at two distinct magnifications. (C) SEM images showing the uniformly dense surface of cardboard substrates covered with Ag NPs obtained from different Ag mass thicknesses and the (D) corresponding AFM images.
Download figure:
Standard image High-resolution imageThe Ag films were successively deposited onto a cardboard surface by a similar method. The surface morphology evolution of the nanoparticles for an Ag mass-equivalent layer thickness (hereafter designated as mass thickness) ranging from 2 to 8 nm is shown in figure 1(C). Despite the inherent heterogeneity of the cardboard packaging substrate, the SEM images revealed a highly dense and uniform distribution of Ag nanoparticles without large-scaled agglomerates along the substrates. The uniformity of the nanostructures produced by the thermal evaporation greatly contributes to the high reproducibility of SERS, as the laser spot covers a range of tens of microns that contain several thousands of particles. Thus, the resulting signal is affected by a large ensemble of illuminated nanoparticles.
Figure 2 presents colour plots with the distribution of the particles in the in-plane semi-axes for four different Ag mass thicknesses. The 2 nm mass thickness nanoparticles exhibited mostly semi-spherical shapes. When mass thickness increases from 2 to 6 nm, the NPs shape becomes more irregular, elongated (with a semi-ellipsoidal shape) and larger in size, as observed by the SEM images (figure 1(C)). Recently, Eshwar Thoutis et al [17] reported a similar thickness-dependent morphology evolution of nanostructures fabricated by thermal evaporation with a post annealing step at 200 °C. The Ag nanoparticles evolved into an elongated shape and became larger as the mass thicknesses increased.
Figure 2. Histogram distributions of the Ag NPs for the 2 (A), 4 (B), 6 (C) and 8 (D) nm mass thicknesses. The values for the average size (in-plane major axis, D) of the deposited Ag NPs are indicated for samples with 2, 4 and 6 nm mass thicknesses. Mass thicknesses of 8 nm or higher resulted in heterogeneous dispersions of NP sizes, for which the average size is not presented.
Download figure:
Standard image High-resolution imageAFM measurements of the nanoparticles with a mass thickness of 4 and 6 nm revealed an average height of approximately 30 and 46 nm, respectively (figure 1(D)), with typical in-plane ellipsoid axes ratios of ∼1.3. For mass thicknesses of 8 nm or higher, heterogeneously distributed NP sizes were observed, which is evidenced by the strong increase of particles smaller than 40 nm and larger than 80 nm. This heterogeneity is clearly visible in figure 1(C) and originates a double-peak in the corresponding distribution histogram.
To study the LSPRs' spectral positions of the SERS substrates and, hence, the wavelength range in which the structures can be useful for SERS applications, the absorptance spectra (figure 3), determined from the total reflectance (Rtot) spectra before and after the Ag NPs deposition, was calculated using the following equation:
Figure 3. UV–vis-NIR absorption spectra of the cardboard packaging substrates after Ag NPs deposition with 2, 4, 6 and 8 nm Ag mass thicknesses. Photographs of the NP-coated substrates are presented above the spectra.
Download figure:
Standard image High-resolution imageThe SERS substrates present two characteristic peaks. The shorter wavelength peaks correspond to quadrupole resonances (sharp peaks at ∼400 nm), whereas the broader peaks at longer wavelengths (centered approximately around the used laser wavelength) correspond to the dipolar resonances of the Ag NPs [1]. The increase of the Ag mass thickness leads to an increase in the average particle size and to more elongated semi-ellipsoidal shapes, as previously mentioned, which in turn results in a red shift and in the broadening of the LSPR peaks [35]. These nanoplasmonic SERS substrates that provide visible different plasmon resonances allow a direct correlation between the NPs' plasmonic properties, the used excitation wavelength and the observed SERS signal intensities. Photographs of the SERS substrates with 2, 4, 6 and 8 nm mass thicknesses (see insets in figure 3) reveal that a preliminary assessment of the NPs' differences and the LSPR spectral position can be estimated by simple visual inspection of the substrate colours.
Standard optical measurements, as those of figure 3, can only detect the far-field light extinction caused by the particles; however, they do not allow probing of their near-field light scattering, which is responsible for SERS [36]. To investigate the localized near-field light enhancement that originates in the vicinity of Ag NPs, electromagnetic models can be used such as the single-particle Mie theory formalism [36] method. This analytical method is based on a spherical particle immersed in an infinite homogeneous medium. This condition does not contemplate the semi-hemispherical Ag NPs deposited on the cardboard packaging substrates. Nevertheless, the Mie theory can still be used to estimate the order of magnitude of the electric field intensity scattered at the vicinity of each particle. For that, an Ag nanosphere surrounded by a homogeneous medium with an effective refractive index between the alumina (native oxide layer at the surface of the aluminum-coated cardboard) and air [37] has been considered.
When the particle is illuminated by a plane wave incident from the top (depicted in the inset of figure 4) with a wavelength higher than the particle diameter (D), a dipolar-like near-field pattern is produced with an intensity that is maximized when the wavelength matches the LSPR of the nanoparticle [5]. The solid curve in figure 4 shows the scattered electric field intensity (|ES|2) at the LSPR wavelength of the particle at the point of maximum field enhancement, located in the intersection between its equator and the incident field (E0 ) direction.
Figure 4. Solid line: The maximum scattered electric field (ES) intensity in units of the incident electric field (E0) intensity produced at the LSPR of an Ag nanosphere as a function of particle diameter (D). The sphere is illuminated by a plane wave with wavevector K0 and is immersed in a uniform medium with an effective refractive index between the alumina (Al2O3) and air. Dashed line: Integral of |ES2|/E02| along the solid line on the surface of the sphere in the E0, K0 plane.
Download figure:
Standard image High-resolution imageFor small nanoparticle sizes (D < 30 nm), the near-field intensity remains almost invariable, as particles scatter in the purely-dipolar electrostatic regime. When the diameter increases, retarded potentials start to influence the plasmonic oscillations, causing deviations from the fundamental dipolar mode. Thus, higher-order LSPRs appear in the scattering spectra together with the highly localized dipolar near-field enhancement exhibited by the Ag NPs, which is, in turn, reduced.
For molecular detection in SERS, it is vital not only to allow a strong electric field enhancement in the vicinity of the NPs but also to have the largest area with such an enhancement. Therefore, a determination of the optimal particle size must account for the fact that smaller MNPs exhibit higher ES intensities, while larger MNPs have more surface area for Raman-active molecules to attach and be illuminated by the scattered near-field. The preferable particle size for SERS detection can be estimated by the integral of the scattered field enhancement along the line segment depicted on the surface of the sphere in figure 4. Such integration is performed only along a quarter of the perimeter of the sphere due to the symmetry of the dipolar scattering pattern and to the unpolarized nature of the illuminating light, with E0 at any direction orthogonal to the direction of incidence. The integral values presented in figure 4 increase almost linearly in the electrostatic small-particle size range (D < 30 nm), with a maximum at D = 59.3 nm, and they decrease for larger particle sizes. Hence, according to the morphological characteristics of the MNP structures presented in figure 1, the desirable particle sizes for molecular SERS detection correspond to those with the 6 nm Ag mass thickness in which the majority of the MNPs have sizes around 60 nm.
It is well known that the plasmonic resonances of the particles are strongly dependent on their surrounding dielectric environment (including the substrate, solvent and adsorbates) [38–41]. Therefore, the LSPR spectral positions will change upon the deposition of the molecule. For that reason, the produced LSPR shift has been optically reanalyzed before performing the Raman measurements. R6G molecules were adsorbed onto the SERS substrates by drop-casting a 10−6 M R6G aqueous solution onto the substrates, followed by drying at 80 °C for 10 min. The absorption spectra (figure 5) revealed that the LSPR peaks were red-shifted and broadened when compared to those of figure 3 due to the increase of the effective refractive index surrounding the NPs. In the case of the 6 nm mass thickness structure, the LSPR position red-shifts 90 nm, resulting in λLSPR = 660 nm.
Figure 5. Absorptance spectra of SERS substrates with 4 nm (A), 6 nm (B) and 8 nm (C) mass thicknesses of silver after the deposition of R6G. The vertical solid line indicates the ideal wavelength (λmax) for maximum SERS intensity with excitation wavelength λexc.
Download figure:
Standard image High-resolution imageRecent works have investigated the SERS enhancement factor (EF) as a function of both the LSPR spectral location and the input laser wavelength [42–44]. The condition for maximum Raman enhancement occurs when the LSPR wavelength is located at the midpoint between the laser excitation wavelength (λexc) and the chosen Raman-scattered photon wavelength, which enhances both the incident and Raman-scattered fields. For an excitation light with a wavelength of λexc = 633 nm, the outgoing Raman-scattered photons, associated with stretches in the 1000–1800 cm−1 frequency region, have wavelengths in the 675–714 nm range (marked in figure 5). Therefore, the maximum SERS enhancement is predicted for LSPRs located within the 654–674 nm range (also marked in figure 5), which precisely matches the LSPR peak of the optimal NPs structures with 6 nm of Ag mass thickness.
The SERS spectra of the RG6 on the cardboard substrates are reported in figure 6. Figure 6(A) clearly shows that the intensity of the Raman signal is pronouncedly amplified when the mass thickness increases from 2 to 6 nm, while it reduces as the mass thickness increases from 6 to 12 nm. We also provide an estimation of the SERS enhancement achieved in each substrate at the same laser excitation source by calculating the enhancement factor (Text S2, Supporting Information). In our case, the EF has been calculated by the ratio between the area under the Raman peak at 1360 cm−1 and was used to compare substrates with and without Ag NPs (Reference curve in figure 6(A)). The average EF values and relative standard deviation as a function of the Ag mass thicknesses are reported in figure 6(B). Each point in figure 6(B) represents the average value from five individual spectra measured at distinct locations on the samples. The error bars show standard deviations of about 5% for each set of five measurements.
Figure 6. (A) SERS spectra of the cardboard packaging substrates coated with and without (Reference) nanoparticles in the presence of R6G. (B) Enhancement factor (EF) as a function of mass thickness.
Download figure:
Standard image High-resolution image
The EF data show that an average enhancement as high as 1 × 106 is achieved for the nanoplasmonic cardboard SERS substrate obtained from 6 nm Ag mass thickness. This is attributed to both the optimal spectral matching of the plasmonic resonance (λLSPR = 660 nm), as depicted in figure 5, and to the high local electric field enhancement that produced the 60 nm-sized silver nanoparticles, as predicted by Mie theory analysis (figure 4).
According to the maximum scattered electric field curve in figure 4, individual Ag nanoparticles that sustain LSPRs can produce scattered electric field intensities with two orders of magnitude higher than the incident intensity close to their external surface. Therefore, such particles would be able to generate SERS enhancements up to ∼104 when the LSPR is matched with the frequencies of both the incident and the Raman-scattered photons. However, the localized near-field enhancement produced by the NP structures fabricated in this work can be considerably higher due to the their elongated character and to the overlap of the near-field regions between adjacent particles, creating the so-called 'hot-spots' and thereby yielding EF values one or two orders of magnitude higher than those that could be attained with single Mie nanospheres. As such, these results support the fact that the observed SERS mechanism involves the plasmonic enhancement of both the incident and Raman-scattered photon intensities, particularly with the best-performing 6 nm Ag mass thickness structure.
Apart from its highly sensitive detection, the uniformity and stability of the SERS substrate is crucial for its use as a routine analytical tool. To test the ability to give reproducible SERS signals, we collected additional SERS spectra of R6G molecules using the best-performing 6 nm Ag mass thickness from 6 randomly selected spots on the same substrate (a 2.5 × 2.5 cm2 area) separated by a distance of at least 1 cm (Text S3, Supporting Information). The Raman spectra profiles are almost identical, indicating a good uniformity. In addition, the SERS measurements were performed on different substrates that were produced in separate batches under the same best-performing conditions, which yielded quite similar results; this proves the reproducibility of the method.
The stability of the materials has been tested by storing the samples for six months. The SERS spectra of 10−6 M R6G aqueous solutions were posteriorly reanalyzed (Text S4, Supporting Information), which showed that the Raman peaks profile is similar to that of the newly prepared sample, suggesting a rather stable SERS substrate.
4. Conclusions
In summary, the implementation of cardboard packaging for highly efficient SERS substrates has been demonstrated. The efficiency of the nanoplasmonic cardboard-based SERS substrates has been assessed through Raman measurements in which rhodamine (R6G) was employed; under optimal fabrication conditions, an enhancement factor of ∼106 was also achieved. Electromagnetic simulations performed with the Mie theory indicate that the desirable Ag particle sizes for molecular SERS detection should be ∼60 nm. The optimal spectral matching of the plasmonic resonance for maximum Raman enhancement together with the high local electric field enhancement produced by the 60 nm-sized silver nanoparticles enabled the achievement of a remarkable maximum SERS enhancement factor. Furthermore, high reproducibility and stability were also demonstrated. Thus, the high tunability of the surface plasmon resonance combined with the advantage of high stability and reproducible EF demonstrate that these cost-efficient Ag nanoplasmonic substrates, which were fabricated by a low-temperature methodology, are efficient large area platforms for SERS.
Associated Content Supporting Information.
A thermal robustness analysis of cardboard substrates was performed to measure the mass changes and thermal effects of the cardboard packaging substrate, the determination of the SERS enhancement factor and the determination of the SERS substratesʼ stability by comparing the Raman signals produced by freshly prepared samples with those measured after six months of storage.
Acknowledgments
This work was financed by the European Commission under projects INVISIBLE (FP7 ERC Advanced Grant no. 228144) and APPLE (FP7-NMP-2010-SME/262782-2) and by the Portuguese Science Foundation (FCT-MEC) through the projects PEst-C/CTM/LA0025/2013-14, PEst-C/EQB/LA0006/2013, EXCL/CTM-NAN/0201/2012, PTDC/CTM-POL/1484/2012, COMPETE and EXPL/CTM-NAN/0754/2013 and through grant SFRH/BD/85587/2012 to A Araújo. C Caro acknowledges Junta de Andalucía through Grant P10-FQM-06615 and through post-doctoral fellowship P07-FQM-02595. M J Mendes acknowledges funding by the EU FP7 Marie Curie Action FP7-PEOPLE-2010-ITN through the PROPHET project (Grant No. 264687). The authors thank Ms Vanessa Otero (REQUIMTE and DCR, FCT/UNL) for her help with the use of the Raman spectrometer. We thank our colleague J Pinto and A Gonçalves (CENIMAT) for the AFM and STA measurements, respectively, and we also thank Stora Enso for the substrate supply.