**3. Results**

Some preliminary SERS measurements on Rhodamine 6G diluted in deionized water at different concentrations were performed in order to quantify the effect of the Raman response enhancement. SERS spectra were collected on the Rhodamine–water solutions in similar conditions, by using an acquisition time of 30 s. The obtained spectra are reported in Figure 1a, arbitrarily shifted along the y-axis in order to improve the readability. They refer to Rhodamine 6G solutions with concentrations of

*c* = 0.0025, 0.005, 0.01, 0.1, 0.5, 1 and 5 mM, respectively. The intensities of the spectra referring to lower concentrations (red lines) are reported as being magnified ten times with respect to the remaining spectra (blue lines). The spectra observed are consistent with data reported in literature [20] and prove the high sensitivity of the method. The dependence of SERS signal intensity on the Rhodamine 6G concentration is shown in Figure 1b, where the intensity values of the spectral mode at 1504 cm−<sup>1</sup> are reported as a function of the solution concentration *c*. The data have been obtained by fitting the Raman spectra of Figure 1 by Lorentzian functions. For low-concentration values the SERS signal intensity increases approximately linearly with concentration, as visible in Figure 1b where the linear dependence is represented by a dotted line. When concentration is larger than 10 μM SERS signal is lower than the value expected in the case of a linear dependence on *c*, indicating a saturation trend probably related to a filling effect of GNP surfaces involved in the SERS mechanism. A SERS efficiency of about four times 10<sup>3</sup> has been estimated by comparing the SERS data of *c* = 5 μM solution with the signal obtained by conventional Raman spectroscopy on the *c* = 5 mM solution.

**Figure 1.** (**a**) SERS of Rhodamine aqueous solution at different concentration in the range of 0.0025–5 mM. The intensity of spectra reported in red are amplified by a factor 10. The spectra are reported arbitrarly shifted along the y-axis. (**b**) dependence of SERS signal intensity (SERS mode at 1504 cm−1) on the Rhodamine 6G concentration (log scale). The linear dependence is represented by the dotted line.

SERS signals from tears were collected in conditions similar to those used for the preliminary tests on Rhodamine 6G. In this case, the acquisition time was typically 180 s. Three or more acquisitions were performed for each sample, using different points of the sample area. The relative standard deviation of the SERS signals obtained from each sample was typically lower than 5%. The spectrum reported in Figure 2a is obtained by averaging SERS responses of tear samples obtained from the eight considered healthy patients. In order to quantify similarities or possible differences among the samples and the repeatability of the measurement, the standard deviation with respect to the average values was calculated for each point of the average SERS signal, and the variation range of SERS response, due to patients' individual peculiarities, is reported in Figure 2b. A mean standard deviation *σ* = 6.3 ± 3.3% was calculated with respect to averaged signals (bottom spectrum of Figure 2). The signal dispersion is generally lower than 10% of the SERS signal, even if a larger signal deviation range occurs in some points of the spectra, at about 1050, 1336, 1523 and 1624 cm<sup>−</sup>1. Amide I and Amide III Raman bands are clearly seen in the reported spectrum at the wavenumber Raman-shifts of ∼1600 cm−<sup>1</sup> and ∼1250 cm<sup>−</sup>1, respectively. In Figure 3a, the result of the spectrum deconvolution in terms of Lorentzian components (reported as blue lines) obtained by a numerical fitting procedure is reported. The main component modes are determined and listed in Table 2, with a temptative assignment of the vibrational sources, based on SERS data of proteins available in the literature [12,21,22]. The SERS signal of Figure 2a is compared with the conventional Raman spectrum measured on a human tear dried drop (Figure 2b). Acquisition times and modalities were similar to those used for SERS measurements, but a significant signal intensity increase and improvement in the spectral resolution are noticed in the SERS signal when the two spectra are compared. It is worth to note that the spectroscopy measurements here reported employed lower laser power and/or photon energy (i.e., a lower Raman scattering efficiency) with respect to those used in the works of Filik [14] and Hu [11] on dried tear drops. This evidences that the observed spectra are due to SERS mechanisms instead of a mere optimization of the Raman spectroscopy collection, as in the case of DCDRS experiments.

**Table 2.** Principal contributions in SERS spectrum of human tears and their assignment according to Refs. [12,21,22].


(def.: deformation; wag.: wagging; str.: stretching; sciss.:scissoring).

**Figure 2.** (**a**) Human tear SERS signal. (**b**) SERS signal standard deviation with respect to the average signal (bottom spectrum) of tears. The red box indicated the mean value of the signal standard deviation (6.3 ± 3.3%).

**Figure 3.** (**a**) SERS spectrum of human tears. The experimental data were fitted by a convolution of Lorentzian functions (numbered blue peaks) representing the main Raman modes occurring in the sample (see Table 2 for a temptative assignment of the Raman modes. (**b**) Conventional Raman spectrum of dried human tears (the signal in (**a**,**b**) are arbitrary scaled).

### **4. Discussion**

The modes featured by the spectrum of tears reported in Figure 3 and listed in Table 2 are mainly concerning proteins and can be temptatively assigned to amino acids [22]. The main components occurring in tears are immunoglobulins (IgA), lactoferrin, lysozyme, lipocalin and albumine [10,23]. The molecular weights and the typical concentrations in human tears of these substances are listed in Table 3. The concentrations are of the order of few mg/mL [10], thus, compatible with SERS sensitivity estimated by Rhodamine 6G measurements, even if each substance could have a different response efficiency regard to SERS mechanisms.


**Table 3.** Main components of human tears.

Among the tear components, lactoferrin (LF) and lysozyme (LZ) have an important role for eye functionality providing defence mechanisms against infective agents [1]. A significant decrease of their levels has been reported in patients suffering from inflammatory Dry Eye Disease (DED) [24–26]. LF is an iron-binding protein present in almost all body biofluids. It has a proven anti-bacterial and anti-inflammatory ability [27]. The main components of the LF are Glutamic and Aspartic acids, Leucine, Arginine, Lysine, Valine and Phenylalanine [28]. LZ is an antimicrobial enzyme constituted by a single chain polypeptide. Its main components are Aspartic acid, Alanine, Glycine, Arginine, Serine, Leucin and threonine [29]. A direct correlation of SERS signal to the LF and LZ is not seen due to the complex and rich composition of tears. However, the peak assignments done in Table 2 for Glutamic acid (at 1243 cm−<sup>1</sup> and 1434 cm−1) and for Aspartic acid (1342 cm−1) provide potential

markers for LF and LZ, respectively. These modes are in agreement with Raman spectra reported for single-component solutions of LF and LZ [11,14] and with SERS response of LZ [30]. In particular, the SERS spectrum of LZ reported by Jun Hu et al. [30] exhibited a particularly intense peak at 1358 cm−<sup>1</sup> stronger than the ones observed at 1280 and 1432 cm−<sup>1</sup> . Furthermore, modes at 730 cm−<sup>1</sup> and 1567 cm−<sup>1</sup> should be related to LZ, in agreement with Ref. [14,30]. In a recent SERS study on human tears, W.S. Kim et al. found a correlation between SERS signal intensity ratio at 1342 cm−<sup>1</sup> and 1242 cm−<sup>1</sup> and the infection state of eye [13]. The authors noticed an increase of the *I*1342/*I*<sup>1242</sup> ratio value in patients affected by Adenovirus (*I*1342/*I*<sup>1242</sup> = 1.13) or Herpes Simplex (*I*1342/*I*<sup>1242</sup> = 3.73) diseases. These values are significatively higher than the *I*1342/*I*<sup>1242</sup> = 0.8 estimated for reference healthy patients. This feature could be explained in terms of a decrease of the LF level originated by the infection state. LZ level should not change significantly because, in agreement with oculist phenomenology, patients with blepharitis, conjunctivitis, and keratitis had normal mean LZ content of tears while patients with herpes simplex keratitis had low LZ values [31]. In our case, we estimated an average value of *I*1342/*I*<sup>1242</sup> = 0.9 ± 0.3 for the considered height healthy patients.

### **5. Conclusions**

The potentiality of SERS for characterizing tears has been investigated by using home-made fabricated Gold-nanoparticle-based substrates. The method has been previously tested and characterized on water diluted Rhodamine 6G samples. Human tear fluids from eight healthy patients have been considered. The SERS response results are mainly related to amino acids and provided a valuable source of information even if the interpretation is not immediate and more work should be done in order to have a more complete and exhaustive data comprehension. Nevertheless, assignments for two components of tears have been determined at the spectral positions of 1243 cm−<sup>1</sup> and 1434 cm−<sup>1</sup> (lactoferrin) and 1342 cm−<sup>1</sup> (lysozyme). As widely reported in the literature and discussed above, the concentrations of both these components are affected by eye health state, and a change in SERS intensity is expected in the case of pathologies. A quantitative assessment of main tear components (in particular, Lysozyme and Lactoferrin) is undoubtedly a demanding issue and a challenge for the future progress of SERS as a diagnostic method. The promising results that are reported here allow us to estimate a future development of research activity towards the implementation of methods suitable to a widespread application of SERS methods in oculistic practice. The development of cheap and friendly methods for SERS implementation, as the one considered in this work, is a first step towards this aim. An interesting future perspective is constituted by the development of soft substrates for SERS implementation, properly designed by using paper or tissues embedded in metallic nanoparticles [13,32,33].

**Author Contributions:** Conceptualization, C.C., M.L. (Mikhail Lisitskiy), D.M. and G.C.; Formal analysis, C.C., M.L. (Mikhail Lisitskiy) and M.L. (Maria Lepore); Investigation, C.C., M.P., S.D.P. and G.C.; Methodology, M.L. (Maria Lepore) and M.P.; Resources, D.M., S.D.P. and G.C.; Writing—original draft, C.C. and M.L. (Maria Lepore); Writing—review & editing, M.L. (Mikhail Lisitskiy), M.P., D.M., S.D.P. and G.C.

**Conflicts of Interest:** The authors deckare no conflict of interest.
