Comparative study of atmospheric water-soluble organic aerosols composition in contrasting suburban environments in the Iberian Peninsula Coast

This study investigates the structural composition and major sources of watersoluble organic matter (WSOM) from PM2.5 collected, in parallel, during summer and winter, in two contrasting suburban sites at Iberian Peninsula Coast: Aveiro (Portugal) and Coruña (Spain). PM10 samples were also collected at Coruña for comparison. Ambient concentrations of PM2.5, total nitrogen (TN), and WSOM were higher in Aveiro than in Coruña, with the highest levels found in winter at both locations. In Coruña, concentrations of PM10, TN, and WSOM were higher than those from PM2.5. Regardless of the season, stable isotopic δ 13 C and δ 15 N in PM2.5 suggested important contributions of anthropogenic fresh organic aerosols (OAs) at Aveiro. In Coruña, δ 13 C and δ 15 N of PM2.5 and PM10 suggests decreased anthropogenic input during summer. Although excitation-emission fluorescence profiles were similar for all WSOM samples, multi-dimensional nuclear magnetic resonance (NMR) spectroscopy confirmed differences in their structural composition, reflecting differences in aging processes and/or local sources between the two locations. In PM2.5 WSOM in Aveiro, the relative distribution of nonexchangeable proton functional groups was in the order: H-C (40-43%) > H-C-C= (31-39%) > HACCEPTED MANUSCRIPT


Introduction
The study of the water-soluble fraction of organic aerosols (OAs) has been in the spotlight of atmospheric research community due to its effects on aerosol optical depth (Andreae and Gelencsér, 2006;Mladenov et al., 2010;Moise et al., 2015), cloud formation and properties (Martin et al., 2013;Padró et al., 2010;Sun and Ariya, 2006;Wonaschütz et al., 2013), radiation balance (Bond et al., 2013;Laskin et al., 2015;Moise et al., 2015), and atmospheric chemistry (George et al., 2015;Laskin et al., 2015;Mellouki et al., 2015). Atmospheric deposition (wet and dry) is the major pathway for removal of organic carbon (OC) from the atmosphere, thus A C C E P T E D M A N U S C R I P T 5 local wind pattern is mainly driven by the land-sea breeze. North-westerly synoptic winds are dominant and generally carry relatively clean air from the sea, but other wind directions are also recorded, with a significant contribution to air pollution levels at this site.
In both sampling locations, a total of eight high-volume PM 2.5 samples (particles with aerodynamic diameter less than 2.5 µm) were simultaneously collected on quartz fibre filters, on a weekly basis (7 days in continuum), during September-October 2016 [n = 4, Summer (SU2016)] and January-February 2017 [n = 4, Winter (WI2017)] in order to collect enough material for subsequent WSOM characterization. In A Coruña, eight high-volume PM 10 samples (particles with aerodynamic diameter less than 10 µm) were concomitantly collected in both seasons, following the same sampling procedure. One field blank was collected in each sampling period in order to correct for ambient background PM 2.5 and PM 10 mass, total carbon (TC), water-soluble organic carbon (WSOC), total nitrogen (TN), isotopic (δ 13 C and δ 15 N), and water-soluble trace metals levels. This sampling procedure is similar to those adopted in previous studies of advanced structural characterization of aerosol WSOM from low sample size groups (Duarte at al., 2015;Duarte et al., 2017b;Duarte et al. 2005;Duarte et al., 2007;Duarte et al., 2008;Lopes et al., 2015;Matos et al., 2017;Matos et al., 2015b). Additional details on aerosol sampling procedure in Aveiro and A Coruña are available in section S1, in SI. After sampling, filter samples were folded in two, with the exposed side face to face, wrapped in aluminum foil and immediately transported to the laboratory in charge of the sampling site, where they were weighted and stored frozen (up to 6 months) until further analysis (section S1, SI). The sampled filters and filter blanks were divided into two fractions, enclosed into heated treated aluminum foil, and one of the fractions were sent by express mail to the laboratories participating in this study.

A C C E P T E D M A N U S C R I P T
6 extracted with 150 mL of ultra-pure water (18.2 MΩ cm) by mechanical stirring for 2 min followed by ultrasonic bath for 15 min. This same extraction methodology was applied to the aerosol samples collected in A Coruña using a volume of ultra-pure water of 75 mL. Each final aqueous slurry was filtered through a hydrophilic polyvinylidene fluoride (PVDF) membrane filter (Durapore ® , Millipore, Ireland) of 0.22 µm pore size. At the end of this filtration step, the slurry residue was washed twice with 5 mL of ultrapure water in order to remove any WSOC still loosely bound to the filter residues. At least two different blanks were performed for each of extractions. Concentrations of blanks were below the detection limits. Also, to avoid metal contamination during filters pretreatment and analysis, all plastic ware and glassware were washed with ultrapure water of 18 MΩ cm resistance and kept for 48 h in 10% (v/v) nitric acid (ultraclean nitric acid 69-70 %), and then rinsed several times with ultrapure water before use.
After collection, sample manipulation and analysis were carried out in a class-100 clean room. followed by a freeze-drying procedure and kept on a desiccator over silica gel until NMR analysis (section 2.5).

EEM fluorescence spectroscopy
The EEM fluorescence spectrum of each PM 2.5 and PM 10 aqueous extract was recorded on a spectrophotometer JASCO (Tokyo, Japan), model FP-6500 using a 1 cm path-length quartz cuvette. Excitation and emission wavelength ranges were set from 220 to 450 nm and 250 to 600 nm, respectively, and their scanning intervals were set at 10 nm and 5 nm, respectively.
The excitation and emission slit widths were fixed at 10 nm and the scan speed was set at

ACCEPTED MANUSCRIPT
A C C E P T E D M A N U S C R I P T 7 100 nm/min. For each day of analysis, a spectrum of a sample of ultra-pure water was acquired under the same experimental conditions and used as blank.

Solution-state 1D and 2D NMR spectroscopy
All NMR spectra were acquired using a Bruker Avance-500 spectrometer operating at 500.13 and 125.77 MHz for 1 H and 13 C, respectively, and equipped with a liquid nitrogen cooling CryoProbe Prodigy TM . All 1D and 2D spectra were run at 295.1 K, and additional details on NMR data acquisition can be found in Section S2, in SI. The dried WSOM samples were dissolved in deuterated methanol (MeOH-d 4 , ~1 mL) and transferred to 5 mm NMR tubes. The identification of functional groups in the NMR spectra was based on their chemical shift relative to the central solvent (MeOH-d 4 ) peak set at δ H 3.31 ppm and δ C 49.0 ppm. The interpretation of the spectral regions and structural assignments were based on the NMR chemical shift data described in the literature for standard organic compounds and for natural organic matter (NOM) from different environmental matrices (Duarte et al., 2008;Hertkorn and Kettrup, 2005;Lopes et al., 2015;Matos et al., 2017;Simpson et al., 2001), as well as on data generated by NMR simulators software's and databases (namely, Perkin Elmer ChemBioDraw® Ultra 14.0 and nmrdb.org (Banfi and Patiny, 2008).

ACCEPTED MANUSCRIPT
A C C E P T E D M A N U S C R I P T  N data, respectively. The total mass of WSOM was estimated as 1.6  WSOC, based on elemental analysis performed on the WSOC aerosol samples collected at Aveiro . In Aveiro, this factor ranged between 1.5 in winter and 1.7 in summer, yielding an average WSOM-to-WSOC ratio of 1.6 . It is very likely that this ratio would vary from site to site; however, due to the lack of additional information for the region of A Coruña, a value of 1.6 was used here to calculate the total mass of particulate WSOM at both sites.

A C C E P T E D M A N U S C R I P T
As depicted in Table 1, the ambient concentrations of PM 2.5 and its TC and WSOC components were consistently higher in Aveiro than in A Coruña, with the highest levels being found during the winter season at both locations. This seasonal trend has been already quite well documented in this and other regions (Duarte et al., 2017b(Duarte et al., , 2007Kiss et al., 2002;Shakya et al., 2012), although an opposite trend has been observed in North America [e.g., (Park et al., 2003;Wozniak et al., 2012)]. Moreover, in summer, the contribution of WSOC to TC in PM 2.5 is higher in Aveiro than in A Coruña, whereas in winter, the WSOC/TC ratios are of same order of magnitude in both locations and higher than those found during warmer conditions. The scatter plot of WSOC versus TC in PM 2.5 [ Figure 1(a)] also indicates a relationship between these two carbonaceous fractions, which is higher in Aveiro (R 2 = 0.84, n = 8) than in A Coruña (R 2 = 0.54, n = 8). In Aveiro, this trend suggests that both TC and its water-soluble organic fraction are probably derived from the same primary emission source(s) and/or are influenced by similar secondary processes in the atmosphere. In previous studies carried out at this suburban site, it has been shown that contributions of biomass burning combined with less warm weather conditions (favoring the particulate phase of semi-volatile organics) are important contributors to ambient TC and water-soluble OA levels during winter [e.g., (Duarte et al., 2017bLopes et al., 2015;Matos et al., 2017)]. In summer, secondary OAs from fossil fuel combustion may prevail over primary sources as an important contributor to fine particulate TC and WSOC fractions (Lopes et al., 2015). In A Coruña, on the other hand, the large spread of WSOC/TC ratios in PM 2.5 (8.1 to 46%, Table 1), combined with the low correlation between WSOC and TC, and low ambient concentrations of PM 2.5 , TC, and WSOC further suggests dissimilar seasonal and spatial variability in emission sources, their strength, and contribution from aging processes at this suburban site. Interestingly, the range and median values of the WSOM/PM 2.5 ratio (Table 1) are rather similar between the two coastal locations regardless of the seasonal period, with the highest values being again found during winter. This finding suggests that the mass contribution of the WSOM fraction to PM 2.5 do not portray the differences in specific water-soluble organic compounds and their sources which were identified in these two locations (additional details in section 3.4).

A C C E P T E D M A N U S C R I P T
10 Publications describing WSOC concentrations in aerosols are most often concentrated on PM 2.5 size-range. When the size distribution of WSOC is discussed [e.g., (Contini et al., 2014;], this carbonaceous fraction is typically dominant in the size interval up to 2.5 μm. In aerosols collected in A Coruña, the particulate matter in suspension is enriched in PM 10 coarse fraction (i.e., particles with aerodynamic diameter between 2.5 and 10 µm), especially during summer (Table 1) samples. This feature suggests that both TC and WSOC in PM 10 coarse fraction are being influenced by similar emission sources [e.g., biomass burning (Reid et al., 2005)] and local conditions, particularly during the winter season, whereas their presence in PM 2.5 may be influenced by different emission sources and/or removal processes.
The TN content at each studied location (Table 1)    Indeed, throughout the summer campaign at A Coruña, the backward air masses trajectories are mostly characterized by air masses that traveled over the Atlantic Ocean (Table S1, Section S3, in SI). On the other hand, regardless of the seasonal period, the isotopic δ 13 C and δ 15 N data obtained in Aveiro may be mainly associated with the presence of anthropogenic fresh OAs.

A C C E P T E D M A N U S C R I P T
12 Water-soluble metal contents in PM 2.5 and PM 10 collected in Aveiro and A Coruña during winter and summer seasons are shown in Water-soluble metal content is generally impacted by the particulate matter origin (see principal component and cluster analysis in Section S5, in SI).  (Chen et al., 2016;Fan et al., 2016;Fu et al., 2015;Mladenov et al., 2011). However, one should be careful when associating the spectral features of these fluorophores to those of humic-like substances occurring in water and soils, since these are unlikely to resemble the WSOM from atmospheric aerosols, both in origin/transformation and compositional terms (Duarte et al., 2007;Matos et al., 2015a). In fact, fluorophore α' is also similar to the peaks in the EEM profiles of secondary

Fluorescence properties of aerosol WSOM
OAs from the ozonolysis of α-pinene (Lee et al., 2013). This difficulty in assigning fluorophores α and α' to specific organic species indicates that a final conclusion on the structural nature of the whole aerosol WSOM samples cannot be withdrawn based solely on their EEM fluorescence profiles.

A C C E P T E D M A N U S C R I P T
13 The spatial and temporal variations of the intensity of fluorophores α and α' seems to be also consistent with those of TC and WSOC, with the highest intensities being found for PM 2.5 samples collected in Aveiro during the winter season. Of notice, the remarkable decrease in the intensity of fluorophore α in the EEM spectra of PM 10 samples collected in A Coruña during winter. The fluorescence feature of PM 10 samples in winter resemble that of WSOM from diesel exhaust particles (Mladenov et al., 2011), being consistent with the isotopic δ 13 C and δ 15 N data, which suggest a higher influence of anthropogenic emissions of fresh OAs during this period at this Spanish location.

Structural and molecular characterization of aerosol WSOM
The solution-state 1D

A C C E P T E D M A N U S C R I P T
15 The potential contribution of fresh biomass burning emissions to winter aerosol samples can also be inferred from the aromatic content of WSOM samples collected in Aveiro (13% for PM 2.5 ) and A Coruña (9.0% for PM 2.5 , and 10% for PM 10 ) as compared to those collect in summer (5.2, 2.2 and 5.9%, respectively). Also, the presence of an intense sharp resonance at δ 1 H 5.3 ppm attributed to anhydrosugars in all winter samples further confirms the presence of smoke particles during this period. Interestingly, the relative content of aromatic protons in the WSOM from PM 2.5 samples collected during summer in Aveiro is higher than those in A Coruña.
A possible explanation can be related to the primary emissions from traffic sources (Heal and Hammonds, 2014), which are expectedly to be more enhanced close to the sampling site in Aveiro than in A Coruña.
Additional details on the structural differences between the aerosol WSOM samples with respect to the two suburban locations were further explored using 2D NMR spectroscopy. ppm, including anomeric carbons), and aromatic (δ H 6.5 -8.5 ppm / δ C 107 -160 ppm) regions.
The distribution of the 2D NMR cross peaks across these chemical shift areas are consistent with other 2D NMR spectral profiles found in literature related to WSOM fractions from field OAs samples (Duarte et al., 2008;Matos et al., 2017;Schmitt-Kopplin et al., 2010). By combining chemical shift information, for known organic structures (Section 2.5) and for other aerosol WSOM samples described in the literature (Duarte et al., 2008;Matos et al., 2017), with homonuclear (COSY) and heteronuclear (HSQC and HMBC) connectivity data (Figures S5 to S10), it was possible to describe the most important substructures within the WSOM components that are likely to be present in the aerosol samples collected in Aveiro and A Coruña. Figure 7 discriminates the substructures common to all WSOM samples, from those typical of each suburban location. The 2D NMR spectral assignments for each substructure are described in Table S5 (section S7), in SI.

<FIGURE 4> & <FIGURE 5> & <FIGURE 6> & <FIGURE 7>
Overall, 20 polyfunctional aliphatic and aromatic substructures (labeled as (1) to (19) in Figure   7, and Table S5 in SI) were identified in this study as being common to all aerosol WSOM samples collected in Aveiro and A Coruña. Of those, aliphatic substructures (1) to (8) were identified in both summer and winter samples, suggesting that their sources mostly remain identical in both suburban areas, regardless of the seasonal period. Such type of aliphatic substructures has been recognized as first-and/or second-generation photochemical oxidation products of different gas-phase precursors (e.g., alkanes, isoprene, carbonyl, epoxides, and anhydrides) emitted from both anthropogenic [e.g., biomass burning, fossil fuel combustion, and meat cooking (Kundu et al., 2010;Liu et al., 2011)] and natural sources [e.g., sea-to-air emission of marine organics, and terrestrial vegetation (Decesari et al., 2011;Facchini et al., 2008;Liu et al., 2011;Russell et al., 2011;Schmitt-Kopplin et al., 2012)]. For example, the persistent in both seasons and locations of molecular signatures characteristic of dimethylammonium (DMA + ), diethylammonium (DEA + ), and methanesulfonic acid (MSA) [substructures (5) to (7) (Figure 7 and Table S5), respectively], indicate the contribution of marine aerosols originating from the Atlantic Ocean. These three WSOM constituents are wellknown tracers of marine aerosols -DMA + and DEA + have a biogenic oceanic source and are produced through the reaction of gaseous amines with sulfuric acid or acidic sulfates, whereas MSA is a photochemical product from marine dimethylsulfide (Facchini et al., 2008). In terms of functional group distribution, the NMR resonances assigned to these three WSOM constituents are more pronounced in the aerosol WSOM samples collected in summer with respect to those collected in winter ( Figure 3). This feature is likely associated to the enhanced marine biological activity during summer as opposed to winter period, when plankton blooming is at its lowest (Cavalli et al., 2004;O'Dowd et al. 2004). The marine origin of these aerosol WSOM constituents also agree with the principal component and cluster analyses (PCA and CA, respectively) performed for major water-soluble ions and metals present in the PM 2.5 and PM 10 samples (Section S5, in SI), which identified a marine source dominated by ions Na + and Mg 2+ .
On the other hand, substructure (8) (Figure 7 and Table S5), known as aliphatic methyl esters, and whose NMR resonances are more prominent in aerosol WSOM from Aveiro, could be a secondary product derived from compounds emitted by various combustion sources (Gordon et Schnelle-Kreis et al., 2007). The existence of a traffic road source, dominated by elements such as Cu, Cr, Ni and V (Section S5, in SI), favors the interpretation of the secondary formation of aliphatic methyl esters from anthropogenic precursors. Photo-oxidation of gasphase N-containing organics from traffic emissions under the presence of atmospheric OH radicals (Barnes et al., 2010;Tong et al., 2016), could also explain the presence of substructure (21) (Figure 7 and Table S5) in WSOM samples from Aveiro. In a similar fashion, the NMR resonances assigned to structure (22) (Figure 7 and Table S5) are more prominent in summer aerosol WSOM from Aveiro. This structure closely resemble those of secondary OAs derived from green leaf volatiles, which are unsaturated, oxygenated hydrocarbons emitted in large quantities by stressed plants (e.g., grass cutting or local weather changes) (Jain et al., 2014). Its presence in summer aerosol WSOM from Aveiro could be due to the occurrence of grass cutting activities, which took place during two of the sampling days in the surrounding areas.

A C C E P T E D M A N U S C R I P T
The aerosol WSOM samples collected at both locations also exhibit the ubiquitous presence of anhydrosugars (such as levoglucosan and mannosan -structures (9) and (10), respectively, in Figure 7 and Table S5) and disaccharides (such as trehalose and maltose -structures (11) and (12), respectively, in Figure 7 and Table S5). Levoglucosan, and to a minor extent mannosan, are well-known organic molecular markers of biomass-burning emissions [ (Matos et al., 2017) and references therein]. Their contribution to the aerosol WSOM load, particularly during the winter season, confirms that biomass burning for house heating is an important source of these compounds in the studied suburban OAs. Their presence in summer samples can be due to the occurrence of forest fire events in the surrounding areas during the sampling campaign. Dimeric sugars, such as maltose, has previously been identified in aerosol samples influenced by biomass burning (Matos et al., 2017;Nolte et al., 2001). The trehalose is a fungal metabolite usually referred as a tracer for the resuspension of surface soil and unpaved road dust and associated microbiota (Simoneit et al., 2004). Hence, resuspension of soil from agricultural activities in areas nearby the sampling locations could be a plausible source of this compound, particularly at A Coruña. This finding is in agreement with PCA and CA results for major watersoluble ions and metals (Section S5, in SI), which indicate the existence of a crustal source dominated by elements such as Al or Ca 2+ or Fe. Interestingly, molecular signatures characteristic of amino sugar derivatives [structure (26), Figure 7, Table S5] were exclusively

A C C E P T E D M A N U S C R I P T
18 found in both PM 2.5 and PM 10 WSOM samples collected in A Coruña, probably reflecting the contribution of fungal-derived microbial residues in resuspended soil material (Joergensen and Wichern, 2008). These results clearly indicate that soil and/or dust resuspension is an important source for aerosol WSOM at this suburban site.
Eight aromatic substructures [(13) to (20) in Figure 7] carrying neutral (aliphatic carbon), NO 2 , and/or oxygen-containing (namely, OCH 3 , OH, COOR, and COR, where R = H or alkyl group) substituents were consistently found in the aerosol samples collected at both suburban locations, particularly during the winter season. One exception is the terephthalic acid [structure (14) in Figure 7, Aveiro (Matos et al., 2017), being associated with the oxidation of aromatic hydrocarbons from urban traffic emissions (Chalbot et al., 2014;Lee et al., 2014). In a similar fashion, nitrophenylderived compounds [substructure (23) in Figure 7, Table S5], whose NMR resonances were exclusively found in WSOM samples from Aveiro, have been mainly attributed to traffic emissions (Tong et al., 2016). Smog chamber studies have also suggested that nitroaromatic compounds in the aerosol phase may also originate from the photo-oxidation of anthropogenic volatile organic compounds, such as toluene (Kelly et al., 2010) and benzene (Borrás et al., 2012), under different NO x concentrations. In this regard, the city of Aveiro distances 15 km from industrial sources producing aniline and nitrobenzene compounds (see Section 2.1). The photo-oxidation of gas-phase N-containing aromatics emitted from those sources cannot be excluded as an SOA source in this urban region. Overall, these findings reflect a notable influence of anthropogenic emissions to the secondary (and more water-soluble) OAs formation at Aveiro. Moreover,substructures (15) to (20) have been usually used as tracers for biomass burning emissions (Duarte et al., 2008;Matos et al., 2017). Once in the atmosphere, these structures can also undergo photooxidation, originating highly oxidized SOA species. For example, substructures (16) and (17)

Conclusions
This study employed a multidimensional non-targeted analytical strategy to investigate and compare the chemical composition of water-soluble OAs from two contrasting suburban The noteworthy structural findings and source signatures reported in this study for aerosol WSOM pinpoint the need to build, in the future, a consistent assessment on the seasonal and spatial variability of the water-soluble OAs composition within and between different Iberian coastal locations. To accomplish this goal, one needs to expand this study both in time and spatial dimensions, and include other atmospheric parameters related to air quality.

A C C E P T E D M A N U S C R I P T A C C E P T E D M A N U S C R I P T
A C C E P T E D M A N U S C R I P T

A C C E P T E D M A N U S C R I P T
A C C E P T E D M A N U S C R I P T         ).

A C C E P T E D M A N U S C R I P T
33 Table 1