*3.2. Mixture of Model RSS: 3-MPA and Na2S as Representatives for Organic and Inorganic RSS*

Although under some experimental conditions the RS–Hg and HgS reduction peaks merge [9,10,40], in our case at given experimental conditions: By selecting the 3-MPA as a representative of the organic RSS; by selecting the deposition potentials at E*<sup>d</sup>* = −0.2 and −0.4 V, and by the different deposition accumulation times (t*<sup>a</sup>* = 0−120 s), separate reduction peaks were clearly revealed as shown in Figure 3. Different accumulation times at different deposition potentials influence the position and intensities of the recorded reduction peaks in the mixture of the 3-MPA and Na2S. Deposition at E*<sup>d</sup>* = −0.2 V would produce only one peak, while deposition at E*<sup>d</sup>* = −0.4 V allows appearance of both the 3-MPA–Hg and HgS reduction.

**Figure 3.** CSSWV curves of the solution containing 65.0 nM 3-MPA and 32.5 nM Na2S, recorded at the different deposition potentials with different accumulation times, as illustrated. Green solid (bold) line is recorded after purging the mixture solution with N<sup>2</sup> when all added sulfide was stripped out. Experimental conditions: E*<sup>d</sup>* = −0.2 and −0.4 V, t*<sup>a</sup>* = 60 and 120 s, A = 25 mV, f = 80 Hz.

Herein, note that with longer accumulation time at E*<sup>d</sup>* = −0.2, only one peak appears. While by deposition at E*<sup>d</sup>* = −0.4 V, and longer deposition, the HgS reduction is also expressed. The acidification step assures that all volatile RSS (sulfide here) will be removed from the mixture, and only non-volatile RSS will remain.

However, the accumulation effect is more expressed in the case of HgS. As already shown, the HgS reduction is not influenced by changing the deposition potentials from 0 to −0.5 V [13], and the HgS reduction peak remains the same whether it is accumulated at −0.2 or −0.4 V. In other words, by selecting these potentials it is possible to make a rough distinction between organic and inorganic S compounds [2,13,16] since majority of organic RSS would not deposit HgS at Hg.

Furthermore, there is a possibility to further characterize and define recorded peaks separately by detailed convolution analyses (Autolab GPES and/or ECDSOFT software, [43]; http://gss.srce.hr/ pithos/rest/omanovic@irb.hr/files/Software/.) However, if the acidification and purging step by N<sup>2</sup> is additionally applied, it is possible to completely remove sulfide (i.e., all volatile RSS) from the mixture and detect one reduction peak assigned to 3-MPA–Hg, as shown in Figure 3 by the green, bold line. Therefore, 3-MPA belongs to a pool of non-volatile and non-acidification-step sensitive RSS, as already discussed [2].

Based on the above-presented results, we conclude that the mixture characterization should be done by measuring at deposition potentials of −0.2 and −0.4 V, with variation of accumulation times (t*<sup>a</sup>* = 0–120s). The acidification step (pH = 2–3), and the measurement of the voltammetric peaks before and after acidification and purging, followed by adjustment to original pH, will indicate the presence of volatile and/or non-volatile RSS.

### *3.3. Electrochemical RSS Responses in WS Fraction of Fine Marine Aerosols*

Electrochemical responses recorded in the WS fraction of the fine marine aerosols are shown in Figure 4. In Figure 4a, the electrochemical response is shown for three consecutive samples from the spring season (March 29–April 4, 2016), denoted as S1, S2, and S3. Figure 4b shows electrochemical responses for four early autumn samples denoted as A1, A2, A3, and A6 (October 8–14, 2016 and October 18–20, 2016, respectively). In spring samples, the presence of mainly RS–Hg reduction was observed, while both, the RS–Hg and HgS peaks were detected in early autumn samples. Peak identification was additionally confirmed by considering the standard 3-MPA solution spiked with Na2S as shown within Figure 4a.

Electrochemical responses recorded in the WS fraction of the fine marine aerosols are shown in Figure 4. In Figure 4a, the electrochemical response is shown for three consecutive samples from the spring season (29 March–4 April 2016), denoted as S1, S2, and S3. Figure 4b shows electrochemical responses for four early autumn samples denoted as A1, A2, A3, and A6 (October 8–14, 2016 and October 18–20, 2016, respectively). In spring samples, the presence of mainly RS–Hg reduction was observed, while both, the RS–Hg and HgS peaks were detected in early autumn samples. Peak identification was additionally confirmed by considering the standard 3-MPA solution spiked with Na2S as shown within Figure 4a.

Observed RSS peaks in the studied WS aerosol samples were recorded in the same range of potentials (−0.58 to −0.76 V) as recorded in the model mixtures of the 3-MPA and sulfide. In spring samples, there was no difference in the peak appearance and its height by measuring at different (E*<sup>d</sup>* = −0.2 and −0.4 V) deposition potentials, behaving more like the HgS reduction, even the peak position was more positive than expected for the typical HgS reduction (Figure 4a). However, in autumn samples, deposition at −0.4 V during t*<sup>a</sup>* = 120 s revealed only the HgS peak (Figure 4c). In these samples, the accumulation effect was also more expressed through increasing of the HgS peak as observed in the model mixtures. Standard addition of the 3-MPA directly into the electrochemical cell containing the A1 sample after the acidification and purging step caused an increase of the first more positive peak, RS–Hg around −0.6 V (Figure 4d), pointing to similar electrochemical behavior to the 3-MPA-Hg electrode process, while the addition of sulfide caused the increase of the more negative HgS peak. It is important to stress that all recorded RSS peaks (RS–Hg and HgS-type electrode reaction) were not visible under diffusion-controlled conditions (t*<sup>a</sup>* = 0s) and were not sensitive to the purging and acidification step, indicating the presence of non-volatile RSS.

The appearance of the negative HgS peak in autumn samples could be likely associated with enhanced release of volatile S compounds during water layer mixing in the nearby marine Rogoznica Lake [44]. That is, during intense vertical lake water layers mixing, millimolar concentrations of sulfide present in the anoxic bottom layer reach the surface where it is rapidly oxidized (mainly to S 0 ) and further lost into the atmosphere. Such a scenario is supported by decreasing the HgS peak in aerosol samples collected five days and later from October 5/6, when mixing of the water layers started (Figure 4b and Table 1). In the same WS aerosol samples, the decrease of the RS–Hg peaks was also noticed.

According to the electrochemical behavior and position of the recorded RSS peaks in the studied ambient aerosol samples, corresponding to the 3-MPA–Hg and HgS reduction, the detected peaks presumably can be evidence of non-volatile mercapto-type RSS and sulfide and/or S<sup>0</sup> or other non-volatile and labile RSS-like compounds that deposit HgS at the Hg surface. The existence of similar compounds has already been proved for the oxic water layers in the North Adriatic as well as marine Rogoznica Lake [2,7,13]. Similar electrochemical response, implying similar RSS were reported as well for precipitation samples (the concentration range for detected RSS was between 2 and 5 nM) [16], where mainly volatile RSS were associated with the RS–Hg peak detected at around −0.55 V (vs. Ag/AgCl).

**Figure 4.** CSSWV curves recorded in the WS fraction of (**a**) three consecutive Spring 2016 aerosol samples (S1, S2, and S3). Red-solid line corresponds to CSSWV scan of the electrolyte containing 19.8 nM 3-MPA spiked with Na2S; (**b**) four consecutive Autumn 2016 aerosol samples (A1, A2, A3, A6); (**c**) effect of different deposition potential (E*<sup>d</sup>* = −0.4 V) for Autumn samples A1, A2, A3, A6. t*<sup>a</sup>* = 120 s for **a**–**c**; (**d**) addition of the standard 3-MPA in Autumn sample A1, after the acidification and purging step, t*a* = 30 s. Please note that pH in this sample was readjusted to 10 after acidification, which influenced the peak position. The best response in samples S2 and A1 was obtained after 2× dilution of original WS aerosol fractions. These samples have the highest surface-active substance (SAS) content. Other experimental conditions: E*<sup>d</sup>* = −0.2 V, *A* = 25 mV, f = 80 Hz.

In the studied spring samples, I*<sup>p</sup>* of non-volatile RS–Hg peaks ranges from 22 to 130 nA, which corresponds to concentrations of the mercapto-type compounds ranging from 10.0 to 65.0 nM in analyzed WS filter aliquots (calibrated with the 3-MPA). The concentration of non-volatile RSS, expressed in relation to the volume of the sampled air, ranges from 2.60 to 15.40 ng m−<sup>3</sup> (mean value = 7.40 ng m−<sup>3</sup> , *N* = 10). In autumn samples, detected peaks range from 12 to 63 nA, corresponding to concentrations of the mercapto-type compounds ranging from 6.0 to 30.0 nM, i.e., from 0.48 to 2.23 ng m−<sup>3</sup> (mean value = 1.26 ng m−<sup>3</sup> , *N* = 6). The I*<sup>p</sup>* of the second HgS peak in autumn samples calibrated with the sulfide, ranged from 7 to 117 nA, which correspond to RSS concentration ranging from 0.75–11.89 nM in analyzed WS filter aliquots, i.e., from 0.02 to 0.26 ng m−<sup>3</sup> (mean value = 0.07 ng m−<sup>3</sup> , *N* = 6) if expressed in relation to the volume of the sampled air.

In addition, concentration of WSOC and surface-active substances (SAS) was determined for all aerosol samples discussed above (S1–S3 and A1–A3, A6), and their values are given in Table 1. Relatively high organic matter content, especially its surface-active fraction that strongly adsorbs on the hydrophobic Hg surface [2,36,45], could potentially suppress the reduction process of organic RSS [2,12,45] and mask the real concentration of RSS present in the ambient samples. On the other side, recorded RSS peaks in solutions with such relatively high SAS presence indicate relatively strong interaction between the present RSS and Hg. Recently published work on voltammetry of reduced glutathione, in its analytical protocol recommends a reduction of natural organic matter concentration to at least 1 mg L−<sup>1</sup> [12] for getting the signal for Hg-glutathione reduction, implying that, in the samples studied here, RSS with a stronger stability constant with Hg than that with Hg-glutathione are possible present.

Presence of SAS is also known to influence the appearance and shape of the HgS peak [45], as was the case in WS aerosol samples studied here. For example, in samples S2 and A1, the most evident RSS response was obtained after double dilution of the original WS aerosol fractions. Moreover, it appears that the HgS peak is positioned more negatively in WS aerosol and precipitation samples [16] in comparison with seawater samples, which can be a consequence of more hydrophobic SAS presence that are blocking electrode redox reaction. Furthermore, the same organic material could have important role in increasing the solubility of organic compounds as well as stabilization of RSS in the aqueous phase and could modify the dissolution rate of aerosol particles in the atmosphere. It is worth noting that higher values of WSOC (0.81–3.45 µg m−<sup>3</sup> ) were recorded in the spring samples when higher amount of non-volatile mercapto-type RSS (2.5−15.8 ng m−<sup>3</sup> ) was also detected. Such seasonal difference could be attributed to biological activity, as already shown for the WSOC fraction in the marine aerosols [46]. The observed seasonal variations of the sulfate concentration shown in Table 1 also lead to the same conclusion.

Results from this work are supportive to studies reporting a significant concentration of non-sulfate sulfur species in fine aerosols [33–35,47]. Besides organosulfates, additional sulfur species such as methanesulfonates, hydroxymethanesulfonates, sulfites, sulfides, polycyclic aromatic sulfur heterocycles, and primary biological particles were reported to contribute to non-sulfate sulfur species. Depending on location and season, the likely presence of such compounds was reported for locations dominated by biogenic emissions, such as forested and agricultural regions and marine environment. As in our case, the signal for such S aerosols was most pronounced during the summer. In addition, preliminary source apportionment results indicate sulfur-containing organic aerosols as an important WSOC fraction in the same studied aerosol samples [48]. However, these S species may also have an anthropogenic source [47].

It is also important to note how the hydrophobic part of WSOC, called humic-like substances (HULIS) [49,50] as they share similar spectroscopic properties to macromolecular humic substances in terrestrial and aquatic environments [51], could have electrochemically active sulfur. That is, a high level of electrochemically active sulfur is confirmed for fulvic and humic material isolated from the lagoon and marine sediments [7,52]. In the previous paper of Frka et al. [36], a broad voltammetric peak that was recorded around a potential of −0.70 V in continental, coastal, and urban aerosol samples as well as in the isolated atmospheric HULIS, could be possibly associated with the presence of RSS. These RSS contribute to the surfactant activity, since HULIS material, i.e., HULIS concentration can be regarded as a rough upper-limit estimate of aerosol water-soluble SAS [53]. Moreover, RSS are an important part of the total S pool in aerosol samples. On the other side, HULIS could contribute to higher solubility and stabilization of RSS in the atmospheric aqueous phase, similarly to that obtained for natural organic macromolecules and elemental sulfur in seawater [7].

In addition, sulfur-containing compounds in the class of carbohydrates and proteins have been recently identified in HULIS extracts of aerosols collected in the coastal area of South Korea [54]. Moreover, protein-like components were found dominant in marine aerosols sampled over the Amundsen Sea as a result of biological activity [55]. Similar compounds are shown to be electroactive on Hg [56–58]. Likewise, a persistent behavior of the revealed voltammetric peaks, suggest that, in this study, detected RSS are from secondary processing in the atmosphere and not from primary emissions.

**Table 1.** Concentrations of reduced sulfur species (RSS), water-soluble organic carbon (WSOC), surface-active substances (SAS), and sulfate (SO<sup>4</sup> <sup>2</sup>−) in the selected Spring and Autumn 2016 aerosol samples, whose electrochemical characterization is shown in Figure 4.

