Bifunctionalized Silver Nanoparticles as Hg2+ Plasmonic Sensor in Water: Synthesis, Characterizations, and Ecosafety

In this work, hydrophilic silver nanoparticles (AgNPs), bifunctionalized with citrate (Cit) and L-cysteine (L-cys), were synthesized. The typical local surface plasmon resonance (LSPR) at λ max = 400 nm together with Dynamic Light Scattering (DLS) measurements (<2RH> = 8 ± 1 nm) and TEM studies (Ø = 5 ± 2 nm) confirmed the system nanodimension and the stability in water. Molecular and electronic structures of AgNPs were investigated by FTIR, SR-XPS, and NEXAFS techniques. We tested the system as plasmonic sensor in water with 16 different metal ions, finding sensitivity to Hg2+ in the range 1–10 ppm. After this first screening, the molecular and electronic structure of the AgNPs-Hg2+ conjugated system was deeply investigated by SR-XPS. Moreover, in view of AgNPs application as sensors in real water systems, environmental safety assessment (ecosafety) was performed by using standardized ecotoxicity bioassay as algal growth inhibition tests (OECD 201, ISO 10253:2006), coupled with determination of Ag+ release from the nanoparticles in fresh and marine aqueous exposure media, by means of ICP-MS. These latest studies confirmed low toxicity and low Ag+ release. Therefore, these ecosafe AgNPs demonstrate a great potential in selective detection of environmental Hg2+, which may attract a great interest for several biological research fields.


Introduction
Nanosized inorganic particles, in simple or composite formulation, have unique physical and chemical properties and represent an increasingly important material in the development of novel nanodevices that can be used in numerous fields such as catalysis, energy, optoelectronics, biomedicine, signals intensity, we selected a photon energy value of 630 eV (impinging at 60 • ) for all elements except S; in order to maximize the intensity of S2p signals, that was expected to be very low due to element dilution, the S2p core level was measured with photon energy = 350 eV. Charging correction of binding energies (BEs) was done using as a reference the aliphatic C1s (BE 285.0 eV) [57]. To fit core level spectra, we subtracted a Shirley background and then used Gaussian peak functions as signals components [58,59].
NEXAFS experiments were carried out at the BEAR (Bending magnet for Emission Absorption and Reflectivity) beamline, installed at the left exit of the 8.1 bending magnet and located the ELETTRA third generation storage ring. The beamline optics can deliver photons having energy comprised between 5 eV and 1600 eV with selectable degree of ellipticity. The carbon K-edge spectra were recorded at normal (90 • ) and grazing (20 • ) incidence relative to the sample surface of the linearly polarized photon beam; however, no angular effects were detected. Calibration of the photon energy and resolution was carried out at the K absorption edges of Ar, N 2 and Ne. In order to normalize the spectra, a straight line fitting the part of the spectrum below the edge was subtracted and the value recorded at 330.00 eV was assessed to 1.

AgNPs Synthesis
The AgNPs stabilized with citrate (Cit) and L-Cysteine (L-cys) were prepared and characterized in analogy to literature reports [60,61]. 1.47 g of sodium citrate were dissolved in 50 mL of distilled water (0.01 M), 0.006 g of L-cys in 25 mL of distilled water (0.002 M) and 0.21 g of AgNO 3 in 25 mL of distilled water (0.05 M). Then, 25 mL of L-cys solution, 10 mL of Cit solution and 2.5 mL of AgNO 3 solution were added sequentially in a 100 mL flask, provided with a magnetic stir. The mixture was degassed with Argon for 10 min, then 4 mL of NaBH 4 solution (0.016 g in 4 mL distilled water) were added. The mixture was allowed to react at room temperature for 2 h and at the end the brown solution was recollected and purified by centrifugation (13,000 rpm, 10 min, 2 times with deionized water). AgNPs main characterizations: UV-Vis (λ max [nm], H 2 O) = 400 nm; < 2R H > ([nm], H 2 O) = 8 ± 1; TEM Ø = 5 ± 2 nm.

AgNPs Sensing Tests
A fixed volume of AgNPs in water (typical concentration 1.6 mg/mL) was mixed with a fixed volume of water solution containing the heavy metal ions at specific concentration. After five minutes of interaction of the nanoparticles with the metal ions, the optical absorption spectra were collected in order to verify possible changes (shape, energy and intensity) of the typical localized surface plasmon resonance (LSPR). The response to several metal ions at different concentrations from 25 ppm down to 1 ppm was tested by UV-Vis spectroscopy [27].

AgNPs Ecosafety Tests
The intrinsic toxicity of AgNPs was investigated through standardized 72 h algal growth inhibition tests [62,63] using two different algal species, belonging to the freshwater and to the marine water environment, Raphidocelis subcapitata and Phaeodactylum tricornutum. Algae were cultured, respectively, in TG 201 medium and F/2 medium, in axenic and exponential growth conditions, at 18 ± 1 • C and 16/8 h light-dark cycle photoperiod. In order to reduce the introduction of chelating agents which have been shown to interfere with heavy metals toxicity [64,65], algal toxicity tests were run with the same medium used for algal culturing modified only in the concentration of ethylenediaminetetraacetic acid (EDTA) (0.05 mg/L and 0.8 mg/L respectively as previously proved to ensure acceptable algal growth). PS single-use sterile multiwell were used for toxicity tests, with 2 mL volume capacity for each well. Algae were acclimated 72 h before running the test under the following conditions: 22 ± 2 • C and 16/8 light-dark photoperiod. Initial algal concentration was 1 × 10 4 cells/mL and tests were manually aerated eVery 24 h using a pipette with sterile tips. Exposure concentrations were as follows: 10, 25, 50, 100, 200, 500 µg AgNPs/L. AgNO 3 was used as positive control at the following concentrations of 3.5, 7, 14, 21, 35 µg/L. In order to exclude any possible role of coating components the toxicity of the AgNPs, citrate and L-Cysteine were tested separately at the following concentrations of 0.5, 1, 2, 5, 10 mg/L.

AgNPs Synthesis and Characterizations
The synthesis of AgNPs by wet chemical reduction is a useful method to obtain spherical nanoparticles with tuned sizes and opportune capping agent [60]. In this work, AgNPs have been prepared by the reduction in aqueous solution of silver nitrate with sodium borohydride as strong reducing agent. Two different capping agents have been chosen: Cit, crucial to induce a high hydrophilic behavior, and L-cys, to induce selective interaction with the environment.
In fact, it is well known that the amino group can easily interact with the chemical environment and in particular with Hg 2+ ions [10]. In order to avoid the excessive presence of L-cys, leading to nanoparticles aggregation, the molar ratio Au/Cit/L-cys = 1/4/2 was chosen. The obtained AgNPs scheme is shown in Figure 1a. After careful purification, AgNPs have been characterized by means of UV-Vis, FTIR and XPS spectroscopies, by DLS and TEM studies. UV-Vis spectra have been carried out together with DLS measurements in water, as reported in Figure 1b,c. The UV-Vis spectrum showed the typical LSPR band, at λ max = 400 nm, confirming the nanodimensions. DLS measurements in water showed hydrodynamic diameter <2R H > = 8 ± 1 nm, as expected. Moreover, FTIR investigations showed the presence of Cit and L-cys on the particles, as reported in Figure 1d.
When the ATR-FTIR spectrum of AgNPs bifunctionalized with Cit and L-cys was compared with reference spectra of these capping agents Figure 1d, it is immediately eVident the absence of the band at~1582 cm −1 (ν as (COO − )) in AgNPs spectrum [66]. This could be interpreted as eVidence that most carboxylate groups of the citrate and cysteine are attached to the surface of the AgNPs. In fact, similar results was observed by Frost et al. [67], who studied the ATR-FTIR spectra of citrate-capped AgNPs and Au-NPs; only for AgNPs, they observed the disappearance of the ν as (COO − ) peak and interpreted this result in terms of adsorption geometry of the citrate molecule, with all the three carboxylate groups interacting with the AgNPs surface: the ν as (COO − ) lying parallel to the AgNPs surfaces results in an infrared-inactive transition. Moreover, in the AgNPs spectrum is present a broad peak at~1547 cm −1 that could be due to the overlapping between the N-H bending of cysteine and the residual carboxylate groups not attached to the AgNPs surface. The functionalization with Cit and L-cys can be also supported by the presence in the AgNPs spectrum of a broad peak centered at 1379 cm −1 indicating the COO − symmetric stretching vibrations; the peak position is slightly shifted to lower wavenumber compared to the free carboxylate anion, as already eVidenced by Frost et al. [68]. The main contributions to the symmetric and asymmetric stretching vibration of the carboxylate group are obviously mainly due to the citrate molecule, that is present in higher concentration on the AgNPs; however, it seems reasonable to hypothesize a similar reactivity for the different carboxylate groups. The broad peak centered at 3300 cm −1 is attributed to the stretching vibrations of the hydroxyl group (ν(O-H)) (data not shown) probably coming from water adsorbed onto nanoparticles surface. We were not able to detect any S-H band around 2550 cm −1 , that would indicate the presence of free L-cys on the AgNPs surface [68,69]. The intensity of this peak in the spectrum of L-cys is actually rather low (see Figure S2 in Supplementary Material), however, XPS results confirm the absence of S-H groups.
To support FTIR data, AgNPs were also investigated by NEXAFS spectroscopy. C k edge spectra of AgNPs were recorded at normal, magic and grazing incidence; however, no angle-dependent effect was detected; therefore, the C k-edge spectrum of AgNPs is shown in Figure 2. The main feature in the spectrum is the 1s→π* consisting of two peaks located at 287.7 and 288.8 eV, with a shoulder at 285.5 eV. According to literature, the 1s→π* transition related to the C=O bond in the carboxylate of L-cys is expected at 288.6 eV [70]; similar values are expected for carboxyl groups [71]. Moreover, L-cys is expected to show a 1s→σ* peak related to the C-S bond at about 287.3 eV. Therefore, we can assign this complex band to overlap between π*C=O and σ*C-S transitions. The broad bands located at about groups [71]. Moreover, L-cys is expected to show a 1sσ* peak related to the C-S bond at about 287.3 eV. Therefore, we can assign this complex band to overlap between π*C=O and σ*C-S transitions. The broad bands located at about 294 and 300 eV are related to 1sσ* transitions of C-H and C=O bonds respectively [72]. The overall spectrum yields prove of successful capping of the AgNPs surface by both Cit and L-cys. Moreover, the TEM studies carried out on diluted samples show AgNPs with narrow size distribution (TEM Ø 5 ± 2 nm) but with a certain tendency to aggregation, which could be due only to drying (see Figure S1). DLS measurements show a monodisperse population of NPs in solution These data are in agreement with DLS data. It can be noticed that the dimensions obtained from DLS are greater than those obtained from TEM images, since DLS estimated the hydrodynamic radius < 2R H > of the particles in the aqueous environment and it is the Z-average value, which is the mean diameter weighted over the scattered light intensity. Therefore, the DLS data are not directly comparable with dimensions obtained from TEM images, as reported in the literature [7,56]. Anyway, the data confirmed in both cases the nanodimension of the AgNPs.
Measurements of Ag concentration in fresh and marine aqueous media showed an almost absent Ag + ions release from AgNPs (Table 1) as opposed to what observed in the literature for other types of AgNPs [45][46][47][48][49]53,55]. Table 1. Ag + concentrations (expressed as µg/L) in freshwater and marine waters with algal medium solution (CTRL), as well as algal medium solutions with AgNPs (500 µg/L) and with AgNO 3 (7 µg/L). In fact, Ag + concentrations in both TG 201 medium (freshwater) and F/2 medium (marine water) with 500 µg AgNPs/L were low (up to 0.4 µg/L) and comparable with those measured in controls (up to 0.36 µg/L; Table 1). However, it is to be noted that in the presence of AgNPs, Ag + levels slightly increased after 72 h, from 0.19 to 0.4 µg/L in freshwater, and 0.15 to 0.37 µg/L in marine water. Since AgNPs dissolution was demonstrated to be independent from particles aggregation state [73], the reason for the lack of ion release is probably to be found in the Cit/L-cys coating.

AgNPs Sensing Tests
Sensing tests were made by checking the optical absorption spectra of the contaminated systems with respect to the reference AgNPs water solution (see also Figure S3). In Figure 3a  The limit of detection (LOD) of the nanosensor for the analysis of Hg 2+ was determined using calibration curves. The LOD can be estimated as three times the standard deviation of the blank signal, obtaining a value equal to 0.6 ppm. The obtained LOD was found to be comparable with other AgNPs systems and the experimental results in the determination of Hg 2+ obtained by some other methods are listed in Table 2 for comparison [8,10,40,[74][75][76][77]. Table 2. Comparison between dimension and limit of detection (LOD) of some metal NPs used as a colorimetric sensor for the detection of Hg 2+ .

Hg 2+ Detection Limit (LOD) (M) References
Citrate/Lcysteine-AgNPs 5 ± 2 3.0 × 10 −6 (0.6 ppm) This work −7 The limit of detection (LOD) of the nanosensor for the analysis of Hg 2+ was determined using calibration curves. The LOD can be estimated as three times the standard deviation of the blank signal, obtaining a value equal to 0.6 ppm. The obtained LOD was found to be comparable with other AgNPs systems and the experimental results in the determination of Hg 2+ obtained by some other methods are listed in Table 2 for comparison [8,10,40,[74][75][76][77]. Table 2. Comparison between dimension and limit of detection (LOD) of some metal NPs used as a colorimetric sensor for the detection of Hg 2+ .

SR-XPS Characterization of AgNPs and AgNPs-Hg 2+
Synchrotron radiation-induced X-ray photoelectron spectroscopy (SR-XPS) measurements allowed to probe the interaction between AgNPs and Hg 2+ ions. Spectra were collected at C1s, N1s, O1s, Ag3d, S2p and Hg4f core levels (all BE (eV), FWHM (eV), relative intensity values and proposed signals assignments are reported in Supplementary Materials file, Table S1). All the individuated spectral components confirm AgNPs stability, and Cit and L-cys capping efficiency. C1s spectrum has three components at 285.00, 286.46 and 288.45 eV BE, respectively associated with aliphatic carbons (mainly impurities, that are always observed in samples prepared in air by liquid solutions); O1s spectra, reported in Figure 4c, also show three different kinds of oxygen atoms, respectively belonging to carbonyl (C=O) functional groups (532.00 eV BE), hydroxyl moieties (-OH, 533.00 eV BE) and physisorbed water (small contribution at about 534.5 eV BE). The atomic ratio between C=O and -OH is C=O/-OH = 1/1.4, very close to the C=O/-OH = 1/1.3 that is theoretically expected for a Cit/L-cys stoichiometry = 2/1, as in the synthetic procedure reported in Materials and Methods. N1s spectrum was also collected, showing a single component centred at 400.24 eV, as expected for amine-like nitrogens. Indeed, C1s, O1s, and N1s data analysis confirm the molecular stability of the AgNPs, also after interaction with Hg 2+ ions.  The occurrence of a signal indicative for metallic Hg is in excellent agreement with the findings reported in [68], where for AgNPs stabilized by Cit molecules and interacting with Hg(II) ions, a direct interaction between Hg and Ag atoms at the nanoparticle surface was envisaged. To better understand the chemistry and geometry of Hg(II)-AgNPs interaction, XAS experiments at the Hg LIII-edge are in programme. Actually, X-ray absorption experiments will allow to directly probe the local coordination chemistry of the metal ion, providing information complementary to the SR-XPS data and allowing for a precise description of the Hg coordination site. Generally speaking, the most indicative signals for the surface-structure analysis of metal nanoparticles capped with thiols are M (Au4f, Ag3d) and S2p core levels; for this purpose, Ag3d and S2p spectra are reported in Figure 4 and will be here discussed in detail.
Ag3d spectra Figure 4a are asymmetric at high BE, a common feature in capped nanoparticles [20,27], indicating that at least two different kinds of silver atoms compose the nanoparticle: the spin-orbit pair at lower BE values (Ag3d5/2 = 368.08 eV) is associated with metallic silver at the NPs core; the signal at higher BE, of very small intensity (about 9% of the whole signal) is due to positively charged silver atoms at the NP surface, interacting with the capping molecule [20,27]. S2p spectra are also composite, showing two spin orbit pairs of very similar intensity (atomic percents are 54.4% lower BE-45.6% higher BE, as reported in Table S1 in the Supplementary Material); interestingly, the two signals are both indicative for sulphur atoms covalently bonded to silver, but with two different hybridizations: the spin-orbit pair at lower BE (S2p3/2 = 161.05 eV) is indicative for S-Ag bonds with sp hybridized sulphur; the signal at higher BE (S2p3/2 = 162.09 eV) suggests S-Ag bonds with sp3 S atoms [78]. It is noteworthy that no physisorbed thiol moieties appear (R-SH S2p3/2 signals are expected around 163-164 eV BE); this finding is in excellent agreement with the hypothesis made by A Majzik et al. [61] based on 1 H NMR studies and suggesting that in metal nanoparticles stabilized by mixed Cit and L-cys capping agents the L-cys molecules preferentially tend to directly bond the metal surface, inducing the most part of Cit molecules to form a shell around the first L-cys layer, in a "layer-by-layer"-like arrangement. It is noteworthy that XPS data do not allow excluding that some Cit molecules could intercalate between L-cys and directly interact with Ag atoms at the NP surface, as eVidenced by IR spectroscopy results. The last signal reported in Figure 4d is the Hg4f spectrum. As eVidenced in the figure, two spin-orbit pairs can be individuated; the first signal (Hf4f 7/2 BE = 100.10 eV) can be associated with metallic Hg atoms [79]; the components at higher BE (Hg4f 7/2 = 100.85 eV) are consistent with literature data reported for Hg 2+ ions in oxides or coordination compounds [72].
The occurrence of a signal indicative for metallic Hg is in excellent agreement with the findings reported in [68], where for AgNPs stabilized by Cit molecules and interacting with Hg(II) ions, a direct interaction between Hg and Ag atoms at the nanoparticle surface was envisaged. To better understand the chemistry and geometry of Hg(II)-AgNPs interaction, XAS experiments at the Hg L III -edge are in programme. Actually, X-ray absorption experiments will allow to directly probe the local coordination chemistry of the metal ion, providing information complementary to the SR-XPS data and allowing for a precise description of the Hg coordination site.

AgNPs Ecosafety
Algal toxicity tests with R. subcapitata and P. tricornutum exposed to AgNPs showed no inhibition of growth rate at all tested concentrations, as reported in Figure 5. Furthermore, algae showed a slight increase in growth rate compared to the control (negative growth rate inhibition), at almost all tested concentrations. Such increase, however, was not observed by positive control tests carried out with Cit and L-cys, which showed no effect on algal growth.
Reference toxicant (AgNO 3 ) showed a 60% inhibition of growth rate compared to the control of R. subcapitata (freshwater), already at the lowest concentration tested (3.5 µg AgNO 3 /L), confirming the toxicity of Ag for microalgae. On the opposite, no effect on growth was observed for P. tricornutum (marine water) exposed in the same conditions, confirming the of drop in free Ag + ions in solution due to complexation with Cl − ions in seawater as measured by Gunsolus et al. [73].
LIII-edge are in programme. Actually, X-ray absorption experiments will allow to directly probe the local coordination chemistry of the metal ion, providing information complementary to the SR-XPS data and allowing for a precise description of the Hg coordination site.

AgNPs Ecosafety
Algal toxicity tests with R. subcapitata and P. tricornutum exposed to AgNPs showed no inhibition of growth rate at all tested concentrations, as reported in Figure 5. Furthermore, algae showed a slight increase in growth rate compared to the control (negative growth rate inhibition), at almost all tested concentrations. Such increase, however, was not observed by positive control tests carried out with Cit and L-cys, which showed no effect on algal growth. Figure 5. Percentage of growth rate inhibition compared to control of R. subcapitata and P. tricornutum exposed to AgNPs (10,25,50,100,200, 500 µg/L) for 72 h. Data are shown as mean ± standard deviation.
Molecules with a high reduced sulphur content are known to bind metal ions; L-cys, thanks to the presence of a thiol group, is able to act as a strong Ag + ligand [82]. Thiol groups could act both either binding Ag + ions, blocking their release in the media, or by excluding oxidizing agents to come in contact with the particles surface, preventing their dissolution and consequent release of Ag + ions in the medium [83]. Gunsolus et al., [73] reported a significant reduction of Ag + ions release and of bactericidal activity of citrate-AgNPs upon incubation with natural organic matter rich in thiol groups. Some studies observed a reduction in the toxic effect of Ag-releasing AgNPs after addition of L-cys in solution, probably as a consequence of L-cys complexation of free Ag + ions [45][46][47].
However, information on L-cys effects on AgNPs stability and dissolution are conflicting. The presence of free L-cys in solution might also enhance Ag + ions release from AgNPs [84]. In our study, however, the absence of toxicity to microalgae and low Ag levels in exposure media confirmed that Cit/L-cys coating prevents the release Ag + ions from the NP. Similar results were obtained by Mu et al. [85] for graphene oxide (GO) nanosheets with covalently-bound L-cys, compared to pristine GO nanosheets. In terms of toxicity, for particles smaller than 20 nm an additional dissolution-independent effect seems to play a role [44,53]. However, despite being very small, AgNPs further confirm their integrity in terms of low Ag + release. Since the main concern regarding the environmental application of AgNP is due to potential toxic effect to non-target aquatic species [44], our findings clearly showed the ecosafety of AgNPs and promote their use in the aquatic environment.

Conclusions
In this work AgNPs functionalized with hydrophilic capping agents, i.e., Cit and L-cys, were synthesized and characterized by means of UV-Vis, FT-IR, SR-XPS and NEXAFS spectroscopies, confirming the surface functionalization. Their nanodimensions were studied by DLS and TEM analysis, showing diameter less than 10 nm. AgNPs showed high selectivity and sensitivity for Hg 2+ in water (concentration range:1-10 ppm) respect to 16 different metal ions investigated. The AgNPs-Hg 2+ system was deeply investigated by means of DLS, ICP-MS, TEM and SR-XPS. Measurements of Ag concentration in fresh and marine aqueous media showed low Ag + ions release, probably due to the good Cit/L-cys covering, also confirmed by SR-XPS data. Moreover, AgNPs ecosafety was confirmed by ecotoxicity tests which showed no effects on algal growth of both freshwater R. subcapitata and marine P. tricornutum algae in the range of tested concentrations (10-500 mg/L). Our results further support the hypothesis that the Cit/L-cys coating of AgNPs prevent dissolution of Ag + ions in both fresh and saltwater. These results open new ways for AgNPs sensing applications in environmental tests on more complex biological systems, up to tests on real environmental aquatic scenarios.
Author Contributions: I.V. designed and made the chemical experiments and synthesis; P.P. and L.B. performed sensing tests; I.C., G.P., C.F., and A.B. carried out the AgNPs biological eValuation and TEM, ICP-MS and DLS measurements; G.I. and L.T. performed measurements and the data analysis by ATR-FTIR; G.I. and V.S. performed measurements and the data analysis, by NEXAFS; C.B. performed measurements and the data analysis, by SR-XPS; S.F. provided technical support in SR-XPS measurements. All authors contributed to the paper writing.
Funding: This research received no external funding.