Office paper decorated with silver nanostars - an alternative cost effective platform for trace analyte detection by SERS

For analytical applications in portable sensors to be used in the point-of-need, low-cost SERS substrates using paper as a base, are an alternative. In this work, SERS substrates were produced on two different types of paper: a high porosity paper (Whatman no. 1); and a low porosity paper (commercially available office paper, Portucel Soporcel). Solutions containing spherical silver nanoparticles (AgNPs) and silver nanostars (AgNSs) were separately drop-casted on hydrophilic wells patterned on the papers. The porosity of the paper was found to play a determinant role on the AgNP and AgNS distribution along the paper fibres, with most of the nanoparticles being retained at the illuminated surface of the office paper substrate. The highest SERS enhancements were obtained for the office paper substrate, with deposited AgNSs. A limit of detection for rhodamine-6G as low as 11.4 ± 0.2 pg could be achieved, with an analytical enhancement factor of ≈107 for this specific analyte. The well patterning technique allowed good signal uniformity (RSD of 1.7%). Besides, these SERS substrates remained stable after 5 weeks of storage (RSD of 7.3%). Paper-induced aggregation of AgNPs was found to be a viable alternative to the classical salt-induced aggregation, to obtain a highly sensitive SERS substrates.


Determination of the diameter and molar concentration of AgNPs by UV-Vis spectroscopy
To calculate the average diameter and the molar concentration of AgNPs the Paramelle et al. 1 method was used. Previously, Haiss et al. 2 reported a similar method to calculating these parameters for spherical AuNPs coated with citrate using the respective absorption spectrum.
First, the average diameter of AgNPs is determined by relating this parameter ( ) with maximum absorbance ( ) by the following equation: A= ɛ 402 nm bc Equation (3) Bastús et al. method provides an improved procedure for the AgNPs synthesis over the method of Lee and Meisel. 3,4 Well distributed AgNPs with a desired diameter were obtained.
Moreover, the lack of use of strong surfactants promotes the accessibility of the surface of the NPs, a crucial factor for efficient SERS. 5 3 S2.

Determination of the diameter and concentration of AgNSs in colloidal solution using Nanoparticle Tracking Analysis
Nanoparticle Tracking Analysis (NTA) methodology was first described in 2006 by Malloy and Carr 6 . The technique is based on the analysis of the Brownian motion of nanoparticles in colloidal solution. A laser beam goes through the sample chamber and the light scattered by the nanoparticles that cross the laser path is captured by a microscope-like assembly that directs the light to a digital image detector. A set of short videos (e.g. 60 seconds) is recorded and the light scattered by individual particles (bright circular shapes) can be easily seen on it.
Video analysis of the path of the individual particles, allows the calculation of the mean squared displacement (MSD) for each particle. From these MSD values, diffusion coefficient (D) can be obtained and, related to the hydrodynamic radius by the Stokes-Einstein equation: (4) where kB is the Boltzmann's constant, T the temperature,  the solvent viscosity, and r is the hydrodynamic radius.
At the same time, the number of nanoparticles per volume can be calculated, since the volume of the analysed solution is known, and the number of particles is counted by the software.
The synthesis of anisometric nanoparticles was confirmed also by TEM ( Figure S 1), proving the homogeneity of the sample relative to the star content. Other morphologies are detected in the samples, namely irregular spheres and rods, but these consistently represent less than 10% of the total nanoparticles.

S3. Vibrational lines assignment for rhodamine (R6G)
Vibrational lines assignments for R6G are presented in Table S1. The areas of the vibrational Raman lines at 1360 cm -1 and 1509 cm -1 were used to calculate spectral intensity as usually reported in the literature 7,8 .

Enhancement Factor calculation
The average SERS enhancement factor (EF) is calculated according to the equation: 11

SERS Raman
Raman SERS Where, ISERS is the SERS intensity of a particular Raman line of the analyte (in this case R6G) and IRaman is the normal (not enhanced) Raman intensity of the R6G measured over a nonplasmonic reference substrate. The Raman signal of the reference is, in most cases, too weak to be detected when measuring small analyte concentrations on the surface. Therefore, the analyte concentration applied to measure the reference Raman spectra is usually higher than that used in the SERS spectra, so a correction factor (NRaman /NSERS) is introduced in the EF expression to take that into account. In the present measurements, NSERS corresponds to the estimated number of molecules contributing to the SERS signal, while NRaman is the number of molecules contributing to the reference Raman signal (from non-SERS surface). Both values are related with the available area of the SERS substrate and the laser spot focus. They are determined by the relation: (6) and (7) Where NA is the Avogadro number, V is total volume of solution spread on the substrate (

S6. Hydrophobic barriers on paper substrates
The fabrication of paper SERS substrates involved the Lab-on-paper technology, based on printing hydrophobic wax patterns and barriers on paper 13 . These patterns were then advantageously used to support the metal nanoparticles deposited by drop-casting method.
Due to the confinement, the diffusion outside the wells is prevented which allowed a uniform distribution and a higher concentration of the nanoparticles (Figure S 3). In addition, the barriers prevent contamination of adjacent samples, even in the more hydrophobic office paper 14 , allowing multiple assays.

Characterization of the synthesized nanoparticles
Solutions of spherical AgNPs presented a yellow colour and its UV-Vis spectrum showed a band centred at around 400 nm, which was assigned to their LSPR ( Figure S 4 -Line A). The width and position of this band can be used to determine the average size and molar concentration of nanoparticles. Their optical properties depend on its size and morphology, verifying the red-shift phenomenon, related to a decrease in the plasmonic frequency. Spherical AgNPs were aggregated in solution by adding 50 mM of NaCl. This process could be followed by UV/vis spectroscopy, by a red-shift (to ≈745 nm) and broadening of the LSPR band Figure S 4 -Line B). This band (≈745 nm) presumably indicates nanoparticle dimers and higher aggregates. 15 Silver nanostars (AgNSs) give rise to a spectrum displaying one strong peak at ≈379 nm (Figure S 4 -line C). At longer wavelengths, an extinction background arises from very different LSPR bands in the suspension giving a wide range of wavelengths in the visible and near-IR. Wavelengths at the near-IR region with a broad width are characteristic of the structures with sharper corners that lead to higher enhancements. The several distinct LSPR bands derive from different morphological shapes of the AgNSs, with different number of arms and varying tip sharpness. 16 Silver NSs with larger vertex angles or a higher number of arms, are probably involved in the more modest multi-pole resonance peak and narrower bandwidth at lower wavelengths (blue-shift of resonant wavelength). 16   8.4x10 8 NPs/mm 2 ; (E and F) 1.7x10 9 NPs/mm 2 ; (G and H) 3.4x10 9 NPs/mm 2 ; (I and J) 6.8x10 9 NPs/mm 2 ; (K and L) 1.4x10 10 NPs/mm 2 . More NPs were observed in the paper surface for higher volumes added, regardless of the type of paper.

Distribution of AgNS on Whatman no.1 and Office papers
However, for office paper, the amount of AgNSs that is necessary to cover the same surface area of the paper, is much lower than for Whatman no.1 paper.

Office paper fluorescence elimination by silver nanoparticles
Office paper samples showed fluorescence emission with excitation under the Raman laser (red beam). However, this fluorescence was eliminated after the addition of silver nanoparticles (Figure S 6). This can be explained by the fact that in the one hand, metal nanoparticles tend to quench fluorescence emission signals. On the other hand, the deposited metal nanoparticles on the surface can shield any fluorescence emission from the paper material 19 . Raman lines observed for AgNSs (C), are assigned to its capping-agent, citrate (Figure S 7). The insets in A and C refer to fluorescence images of paper before and after AgNSs addition.

S10. Interference signals in SERS spectra of AgNSs
Contaminants even in a lower concentration may be present in the sample. They can be selectively enhanced through a resonance Raman scattering mechanism, overlapping the bands of the adsorbate under study. 20 Therefore, the positions of the vibrational lines were evaluated and confronted with components that were known to be present on the paper SERS substrate: (1) paper; and (2) reagents used in the NPs synthesis, such as hydroxylamine and sodium citrate. The spectra of paper alone (data not shown) and citrate ( Figure S 7 -A) were analysed according to the standard Raman spectrum. The Raman data of citrate oxidation products (acetonedicarboxylic and acetoacetic acid) and hydroxylamine were obtained from the literature. 20,21 The SERS signal of the residual nitrate (NO3 -) from the AgNO3 (the precursor for NPs synthesis) was also compared to the literature data. 22 However, no Raman lines from NO3could be detected.   22,23 The bands in Raman spectrum of citrate ( Figure S 7), are consistent with published values. 22,24 The slight deviations between the two spectra may be indicative of interactions between the citrate molecules and the Ag surface. 24 Nevertheless, SERS spectra of the observed interfering bands can be clearly attributed to citrate, since they correlate very well with the Raman spectrum of citrate in aqueous solution. In fact, Yaffe et al. found that in a salt-aggregated AgNP colloid, the band patterns for the citrate did not disappear after addition of the analyte. 24 The bands related to the appearance of citrate oxidation products were not identifiable, hence, it is possible that they do not interfere in R6G SERS spectra. 25 The presence of interferences in paper SERS substrates underscores the importance of the interpretation of the spectra profiles, even before the addition of the analyte.

S11. Reproducibility between different AgNSs synthesis batches
To test the reproducibility of AgNSs substrates in office paper, nine plasmonic wells, from three independent AgNSs synthesis batches, were randomly selected and spectra were collected at room temperature for three different R6G concentrations 10 -7 , 10 -8 and 10 -9 M (amounts of 1×10 -1 , 1×10 -2 and 1×10 -3 ng). Results of relative standard deviation (RSD) were calculated by variation of the 1509 cm -1 Raman spot-to-spot area (Figure S 8). It can be concluded that drop-casting of AgNSs on the office paper substrate yields uniform nanostructured surfaces over a scale of several microns, leading to a homogenous distribution of hot spots, and resulting in highly reproducible SERS responses.

S12. Time stability of the plasmonic paper substrate
The stability of the optimized paper SERS substrate was systematically investigated over a period of 5 weeks and between measurements, it was stored at 4 ᵒC in a desiccator. The spectra were collected at room temperature for three different R6G concentrations 10 -6 , 10 -8 and 10 -9 M (amounts of 1×10 -1 , 1×10 -2 and 1×10 -3 ng). Results of RSD were calculated by variation of the 1509 cm -1 Raman spot-to-spot area ( Figure S 9). The signals of SERS spectra after 5 weeks were very similar to the one obtained in freshly prepared substrates.