Advances in characterization of size-resolved organic matter in marine aerosol over the North Atlantic

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 109,, doi: /2004jd005137, 2004 Advances in characterization of size-resolved organic matter in marine aerosol over the North Atlantic F. Cavalli, M. C. Facchini, S. Decesari, M. Mircea, L. Emblico, and S. Fuzzi Istituto di Scienze dell Atmosfera e del Clima, CNR, Bologna, Italy D. Ceburnis, Y. J. Yoon, and C. D. O Dowd Department of Experimental Physics and Environmental Change Institute, National University of Ireland, Galway, Ireland J.-P. Putaud and A. Dell Acqua Institute for Environment and Sustainability, Joint Research Centre, European Commission, Ispra, Italy Received 17 June 2004; accepted 12 October 2004; published 30 December [1] Size-segregated marine aerosols were collected at Mace Head Atmospheric Research Station (Ireland) during spring and autumn 2002 corresponding with the phytoplankton bloom periods in the North Atlantic. Strict control of the sampling, air mass back trajectory analysis, and analysis of pollutant tracers allowed the selection of a set of samples representative of clean marine conditions. A comprehensive chemical characterization of both (1) water-soluble and water-insoluble organic fraction and (2) water-soluble inorganic ions was performed. The selected samples illustrated a consistent picture in terms of chemical composition. The supermicron mode predominantly comprises sea-salt aerosol with a mass concentration of ± 0.80 mg m 3, the remainder being non-sea-salt (nss) sulphate, 0.03 ± 0.01 mg m 3, and nitrate, 0.13 ± 0.04 mg m 3. By comparison, the mass of sea salt, nss sulphate, and nitrate in the submicron mode is found to be 0.39 ± 0.08 mg m 3, 0.26 ± 0.04 mgm 3, and 0.02 ± 0.01 mg m 3, respectively. Water-soluble organic carbon (WSOC) is observed in the submicron mode with a mass concentration of 0.25 ± 0.04 mg m 3, comparable to that of nss sulphate, and in the supermicron mode with a mass concentration of 0.17 ± 0.04 mg m 3. The WSOC to total carbon (TC) ratio is found to be 0.20 ± 0.12 for the submicron fraction and 0.29 ± 0.08 for the supermicron fraction, while the black carbon (BC) to TC ratio is, on average, ± for both aerosol modes. The remaining carbon, water-insoluble organic carbon, contributes 0.66 ± 0.11 mg m 3 and 0.26 ± 0.06 mg m 3 to the submicron and supermicron modes, respectively and, thus, represents the dominant submicron aerosol species. Furthermore, the WSOC chemical composition comprises mainly aliphatic and only partially oxidized species and humiclike substances, resulting in appreciable surface-active properties. The observed organic matter chemical features (size-dependent concentration, hydrophobic nature of a substantial fraction of the organic matter, and low oxidized and surface-active WSOC species) are consistent with the hypothesis of a primary marine source; bubble-bursting processes, occurring at the surface of the North Atlantic Ocean during phytoplankton blooms, effectively transfer organic matter into marine aerosol particles, particularly enriching the fine-aerosol fraction. INDEX TERMS: 0305 Atmospheric Composition and Structure: Aerosols and particles (0345, 4801); 0315 Atmospheric Composition and Structure: Biosphere/atmosphere interactions; 0365 Atmospheric Composition and Structure: Troposphere composition and chemistry; KEYWORDS: organic marine aerosol, size-dependent composition, North Atlantic Citation: Cavalli, F., et al. (2004), Advances in characterization of size-resolved organic matter in marine aerosol over the North Atlantic, J. Geophys. Res., 109,, doi: /2004jd Introduction [2] Marine particles play a potentially important role in the climate system, directly by backscattering and absorbing Copyright 2004 by the American Geophysical Union /04/2004JD solar radiation and indirectly by acting as cloud condensation nuclei (CCN), thus influencing cloud droplet size distribution, cloud albedo and lifetime. Their chemical composition as a function of aerosol size is one of the critical parameters defining aerosol properties relevant to radiative effects. 1of14

2 [3] There has been a general consensus in literature that, under clean marine conditions, non-sea-salt sulphate (nss SO 4 2 ), formed by gas-to-particle conversion of dimethyl sulphide (DMS)-derived SO 2 [Charlson et al., 1987], dominates submicron aerosol composition, while sea salt, produced by the mechanical disruption of the sea surface, largely controls the chemical composition of the supermicron aerosol fraction [Blanchard, 1954; Quinn et al., 1998, 2000]. Direct and indirect evidence [O Dowd and Smith, 1993; Murphy et al., 1998] shows, however, that the salt mode could well extend into the submicrometer size range. It has also long been known that marine aerosols contain organic material [i.e., Blanchard, 1964; Hoffman and Duce, 1976]. Recent field measurements clearly document the presence of organic matter in individual particles [Middlebrook et al., 1998], and a significant contribution of organic species to the fine-aerosol mass in the unperturbed marine boundary layer (MBL) [Putaud et al., 2000]. However, the role of organic compounds in the remote marine aerosol, although recognized as potentially important [Novakov et al., 1997; Matsumoto et al., 1997], remains largely uncertain, mainly because of the lack of quantitative measurements of their size-dependent composition. Moreover, despite a large number of studies on the occurrence of individual compounds or classes of compounds, e.g., carboxylic acids [Kawamura and Sakaguchi, 1999; Mochida et al., 2003a], a comprehensive characterization of the chemical nature of organic matter in marine aerosols remains currently unavailable. [4] This paper reports the size-segregated composition of clean marine aerosol samples collected at the Mace Head Atmospheric Research Station ( N, W), ideally situated on the west coast of Ireland for measurements of background northeast Atlantic aerosols [Jennings et al., 2003]. To obtain information on aerosol properties of relevance to aerosol-cloud interaction, a complete chemical characterization of the water-soluble organic and inorganic aerosol fraction was performed, with particular emphasis on the organic component of the aerosol population. 2. Experimental Setup 2.1. Particle Sampling [5] The measurements were carried out at Mace Head station during spring (April, May and June) and autumn (late September and October) 2002, coinciding with the phytoplankton bloom periods occurring in the North Atlantic. A Berner low-pressure impactor (flow rate 1.8 m 3 h 1 ), sampling air at ambient relative humidity (RH) (80 ± 8%), was used to collect aerosol particles in eight size fractions, between 0.03 and 16.0 mm according to the following 50% equivalent aerodynamic cutoff diameters: 0.063, 0.125, 0.25, 0.5, 1, 2, 4, 8, 16 mm. Tedlar foils were used as the collection substrate in the impaction stages. The impactor was mounted at approximately 50 m from the shoreline at a height of approximately 3 m above ground level. In parallel, aerosol particles smaller than 10 mm were collected, at ambient RH, on quartz fiber filters, using a Sierra-Andersen high-volume cascade impactor (flow rate 60 m 3 h 1 ), located at ground level 50 m from the shoreline. [6] An automated sectorized sampling system was used to control the aerosol sampling and each sample was deployed for a period of 7 days. The system ensured sampling only of air (1) reaching the site from a controlled oceanic clean sector between 190 and 300 and (2) with a total particle concentration (d 50% = 14 nm) below 900 cm 3. Previous analysis has shown that, in general, air masses meeting such criteria can typically be regarded as clean marine air masses, having passed over thousands of kilometers of open sea, free from the influence of major anthropogenic sources [Cooke et al., 1997]. [7] Submicron aerosol size distributions were measured over the range of 10 nm to 263 nm using a scanning mobility particle sizer (SMPS) [Wang and Flagan, 1990] located at 8 m height. The SMPS was operated at a RH of approximately 40% with a temporal resolution of 2 min. The estimated uncertainty in particle size (based on mobility) is ±5% Postsampling Selection Criteria [8] All sampling periods were analyzed after the fact to gain further information on pollution influences, and to confirm the integrity of the labeled clean marine air masses. The first-level check was to ensure that the black carbon (BC) concentration did not exceed that of clean marine air. The second-level check was to examine all back trajectories on a 2-hour timescale while the third-level check was to ensure that CO concentrations remained typical of the background level and revealed no influence of biomass burning Black Carbon [9] Black carbon has previously been used over the northeast Atlantic as a tracer for anthropogenic influences, and as a tool for distinguishing between clean marine, modified marine and continental air [O Dowd et al., 1993]. O Dowd et al. [1993] indicated that BC loadings of ng m 3 are typical of the cleanest marine air over the northeast Atlantic. Using an Aethalometer (MaGee Scientific), BC was measured continuously throughout the experiment with a temporal resolution of 10 min. The range of BC concentrations during activated sampling periods was between 17 and 70 ng m 3, with a median value of 35 ng m 3. This value is in excellent agreement with the clean air values observed by O Dowd et al. [1993] and with those predicted for remote marine sites [Chung and Seinfeld, 2002] Back Trajectories [10] The 120-hour backward trajectories were calculated, using the National Oceanic and Atmospheric Administration (NOAA) HY-SPLIT model (R. R. Draxler, and G. D. Rolph, HY-SPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) model access via NOAA ARL READY website ( html), NOAA Air Resources Laboratory, Silver Spring, Maryland, 2003), every 2 hours for all sampling periods. Overall, the majority of the trajectories had passed over northern Canada and Newfoundland, northern and mid- Atlantic and the Arctic regions 120 hours previously. Trajectories over two typical sampling periods are shown in Figure 1 for the periods 2 9 October 2002 and April In the first example, the majority of the trajectories arrive along one route having traversed Newfoundland 120 hours previously at an altitude of km above sea level. For these trajectories, the air has spent at 2of14

3 Figure 1. Typical air mass trajectories for the clean marine sample during 2 9 October 2002 and April 2002 representing Arctic, northern Canadian, and mid-atlantic air masses. See color version of this figure at back of this issue. least 4 days over the North Atlantic prior to arriving at Mace Head in the prevailing westerly airflow. A small number of trajectories (5 6) originate between Greenland and Newfoundland with virtually no contact with land for 120 hours, and a similar amount originates in the subtropics, again with no land contact for 120 hours. One trajectory indicates a brief passage over the United Kingdom and northwest France; however, BC and total particle concentration criteria for clean marine air were not breached and the sample was deemed valid. [11] In the second example, the prior transport pathways to Mace Head exhibit a greater diversity of air mass origin; however, four main source regions and transport patterns are observed. One cluster of trajectories is associated with polar air masses emerging out of the Arctic region along the east of Greenland, while another cluster is associated with 3of14

4 Table 1. Meteorological and Other Related Parameters for Clean Marine Samples a Variable April May 30 May to 6 June 7 14 June 2 9 October 9 16 October Average Wind direction, deg 249 ± ± ± ± ± ± ± 11 Wind speed, m s ± ± ± ± ± ± ± 0.7 Temperature, C NA 11.2 ± ± ± ± ± ± 1.5 RH,% NA 82±6 82±6 84±8 86±7 67±7 80±8 CPC (TSI 7610), >14 nm 440 ± ± 430 NA 490 ± ± ± ± 205 BC, ng m 3 17 ± ± ± ± ± ± ± 23 a Values are ±1 standard deviation. Values are compiled from a 1-min data set, except BC values, which are derived from a 5-min data set. NA indicates data are not available. CPC is condensation particle counter. Arctic air masses emerging out from the oceanic region between the southwest tip of Greenland and the north of Newfoundland/Canada. A third cluster emerges overland initially from the Canadian/Newfoundland region before advecting out over the mid North Atlantic. The fourth cluster can be viewed as air that advected from the mid North Atlantic, down into the near subtropics and then to Mace Head, all without any contact with land. Analysis of trajectory levels shows at least 4 5 days below 1.5 km and, as in the first case, trajectories advecting over the extreme north of North America do so initially at 2 3 km before reaching the Atlantic, after which almost all trajectories remain below 1.5 km. As in the previous case, two trajectories advect over the southern United Kingdom and northern France while the other selection criteria indicate the typical conditions of marine air Carbon Monoxide [12] During the activated sampling periods, carbon monoxide, a useful indicator of biomass burning, showed mixing ratios consistent with the background value for Mace Head, 130 ± 5 ppbv (P. Simmonds, personal communication, 2003). Holloway et al. [2000] demonstrated that this value is representative of the remote Arctic and North Atlantic environment and supports the absence of local biomass burning influences. However, low carbon monoxide mixing ratios [e.g., Forster et al., 2001] were also sometimes observed to be associated with Canadian forest fire emissions. This potential influence was evaluated for those samples (e.g., collected during 2 9 October 2002 and 9 16 October 2002) whose air masses originated over northern Canada. A permanent snow cover exceeding 1 cm in depth (Meteorological Service of Canada), already present during these periods, rendered the occurrence of forest fires over the northern Canadian areas very unlikely. Furthermore, since the submicron aerosols released from Canadian forest fires are more efficiently removed than CO by wet deposition during the transport across Atlantic, any influence on the aerosol mass measured at Mace Head 5 6 days later is likely to be negligible. [13] A general summary of meteorological data, total particle concentration and black carbon concentration for the selected sampling periods is presented in Table 1. Wind speeds in the range of m s 1, encountered during the measurements, are representative of low-pressure systems passing over western Ireland. [14] On the basis of the above-mentioned selection criteria and on sampling times (i.e., frequency of clean sector occurrence during 7 days) sufficiently long to produce meaningful data for all the chemical species at each size interval, six samples were selected as valid clean marine aerosol. In addition, blank substrates of the sampling procedure were taken on a regular basis throughout the measurements (13 blank tedlar foils and 10 quartz filters) and analyzed for all chemical components Analyses and Data Processing [15] Berner tedlar foil water extracts were used to determine the mass size distribution of inorganic ions and watersoluble organic carbon content by ion chromatography and a TOC liquid analyzer, respectively. A detailed description of the analytical procedure used for sample treatment, extraction and analyzes has been published elsewhere [Matta et al., 2003]. Sea salt and nss SO 4 2 aerosol concentrations were calculated using Na + as sea salt tracer and a standard sea salt composition. [16] Hi-vol quartz filters were appropriately combined into a fine (aerodynamic particle diameter, Dae < 1.5 mm) and a coarse (Dae > 1.5 mm) aerosol fraction, and used to determine concentrations of total, organic and black carbon by a thermal evolved gas analysis, EGA analysis [Putaud et al., 2000] and WSOC by TOC analysis. The amount of inorganic carbon measured directly by TOC analysis was found to be negligible in all samples. The average ratio WSOC/TC and BC/TC established from Hi-vol samples, and WSOC concentrations from impactor samples were combined to derive the size distribution of water-insoluble organic carbon (WIOC) and BC. [17] In order to compare the contribution of the carbonaceous species to that of inorganic ions, a molecular mass to carbon mass ratio of 1.8, with an estimated random uncertainty of ±0.1, was adopted for WSOC, based on the speciation of WSOC performed on the samples (see section 3.2) [Fuzzi et al., 2001]; a conversion factor of 1.2, with an arbitrary random uncertainty ±0.1, was applied for WIOC [Gray et al., 1986; Zappoli et al., 1999]. The BC was considered to consist of pure carbon, and a conversion factor of 1.0, without any significant uncertainty [Cachier et al., 1989], was therefore used. [18] The mean random uncertainty for each aerosol component and each size range was computed using the procedure of error propagation described by Putaud et al. [2000]. For correlated parameters, absolute uncertainties were conservatively added. [19] Table 2 shows the relative random uncertainties of the mean aerosol component concentrations for each impactor stage obtained, including (1) uncertainty in the sampled air volume, ±3%; (2) precision of the extraction water volume, ±0.04 ml; (3) uncertainties in ion chroma- 4of14

5 Table 2. Overall Relative Random Uncertainties for the Average Aerosol Sample a Stage Cutoff, mm Sea Salt + NH 4 2 nss SO 4 NO 3 WSOC WIOC BC Sum a Uncertainties are given as percentages. WSOC is water-soluble organic carbon, WIOC is water-insoluble organic carbon, and BC is black carbon. tography and WSOC measurements, ±5%; (4) uncertainties in the molecular mass to carbon ratios for carbonaceous species and (5) the blank variabilities, ±52, ±15, ±144, ±27, ±15, and ±53 mg L 1 for Na +,NH 4 +,Cl,NO 3,SO 4 2, and WSOC, respectively. The main sources of uncertainty are (1) blank variability for sea salt in stages 1 3, NH 4 + in stages 1 and 7 8, SO 4 2 in stages 1 2 and 8, NO 3 in stages 1 4 and 7 8, and WSOC in all stages and (2) analytical precision for sea salt, NH 4 +,SO 4 2, and NO 3 in stages 4 8, 2 6, 3 7, and 5 6, respectively. [20] Table 3 shows the overall random uncertainties of the mean WSOC, TC, BC concentrations and the mean WSOC/ TC and BC/TC ratios observed in the Hi-vol fine- and coarse-aerosol fractions calculated, including (1) uncertainty in the sampled air volume, ±10%; (2) precision of TC and BC determination, ±8%; (3) uncertainties in the molecular mass to carbon ratios for carbonaceous species and (4) the blank variabilities ±0.022, ±0.062, and ±0.011 mgc cm 2 for WSOC, TC, and BC, respectively. The main sources of uncertainty are analytical precision for TC and blank variability for WSOC and BC. The uncertainty in WSOC/TC and BC/TC ratios derives mainly from their variability observed in the Hi-vol samples. [21] A characterization of WSOC was performed adopting the procedure proposed by Decesari et al. [2000]. According to this procedure, the WSOC mixture was separated by ion exchange chromatography into three main classes of compounds: (1) neutral/basic compounds, NBC; (2) mono- and di-carboxylic acids, MDA; (3) polycarboxylic acids, PA; a functional group analysis was performed by proton Nuclear Magnetic Resonance ( 1 HNMR). Surface tension measurements of the water extracts of fine- and coarse-aerosol samples were performed by means of an oscillating bubble tensiometer [Facchini et al., 2000]. 3. Results 3.1. Size-Segregated Chemical Composition [22] Figure 2 reports the average size-segregated concentrations of the main constituents of the aerosol samples. The uppermost stage is excluded because of the very low efficiency of the impactor inlet in collecting particles with diameter larger than 8 mm at wind speed of 5 m s 1. The contributions of each species to the submicron and supermicron modes are summarized in Table 4. In this paragraph and Figure 2, we refer to average concentration values ±1 standard deviation observed. Consistent with recent observations [O Dowd et al., 1997; Murphy et al., 1998; Quinn et al., 1998], our results indicate the presence of at least a small amount of sea salt in the aerosol submicrometer range down to the lowest size cut and reveal that particles larger than about 1 mm are almost exclusively composed of sea salt, suggesting that primary marine aerosol sources contribute to the entire aerosol size range investigated. The submicron sea salt mass is 0.39 ± 0.08 mg m 3, while the supermicron fraction amounts to ± 0.80 mg m 3. [23] The nss sulphate is primarily found to contribute 0.26 ± 0.04 mg m 3 to the submicron size fraction and 0.03 ± 0.01 mg m 3 to the supermicron mode, and is in agreement with previous observations in the unperturbed MBL, by Huebert et al. [1998], Putaud et al. [2000] and Kleefeld et al. [2002]; such low sulphate concentrations are in a range close to those expected from natural sources (generally less than 300 ng m 3 )[Savoie et al., 1989]. By contrast, nitrate is found predominantly in the supermicron mode, with a mass of 0.13 ± 0.04 mg m 3, compared to 0.02 ± 0.01 mg m 3 in the submicron mode. [24] From the impactor data, WSOC exhibits a bimodal distribution with a submicron mode and a supermicron mode, with average mass concentrations of 0.25 ± 0.04 mg m 3 and 0.17 ± 0.04 mg m 3, respectively. The bulk WSOC concentration of 0.42 ± 0.06 mg m 3 confirms the value reported by Kleefeld et al. [2002] for bulk aerosol samples collected previously at Mace Head in background conditions. [25] The WSOC concentrations were also measured from the Hi-vol samples, finding average mass concentrations of 0.20 ± 0.08 mgm 3 and 0.15 ± 0.11 mgm 3 for the fine- and coarse-fractions, respectively. These results are in agreement with those from the impactor samples. The average bulk TC mass concentration, from Hi-vol samples, is 0.98 ± 0.64 mg Cm 3, partitioned in 0.68 ± 0.38 mg Cm 3 and 0.33 ± 0.30 mg C m 3 for the fine- and coarse-aerosol fractions, respectively. The average ratio of WSOC to TC is therefore 0.20 ± 0.12 for the fine fraction and 0.29 ± 0.08 for the coarse fraction. The average bulk BC concentration of 36.5 ± 4.4 ng m 3, from Hi-vol samples, is in excellent agreement with the range of BC encountered for background air masses in the North Atlantic [O Dowd et al., Table 3. Overall Relative Random Uncertainties for the Average High-Volume Aerosol Sample a Aerosol Fraction WSOC TC BC WSOC/TC BC/TC Dae < 1.5 mm Dae > 1.5 mm a Uncertainties are given as percentages. TC is total carbon, and Dae is aerodynamic particle diameter. 5of14

6 Figure 2. Average concentrations of some main chemical components as a function of particle size (geometric mean aerodynamic diameter) in units of mg m 3. Variability bars of ±1 standard deviation have been added. Table 4. Average Submicron and Supermicron Mass Concentrations for the Main Aerosol Components a Range Sea Salt + NH 4 2 nss SO 4 NO 3 WSOC WIOC BC Submicron (24) (4) (16) (1) (15) (39) (1) Supermicron (94) (0) (0) (1) (2) (2) (0) a Concentrations are given in mg m 3. Corresponding percentage contributions to the total measured mass are reported in parentheses. 6of14

7 Figure 3. Nuclear Magnetic Resonance ( 1 HNMR) spectra of (a) fine and (b) coarse fractions of the sample collected on 2 9 October The main functional group categories are marked on the 1 HNMR spectra. Definitions are the following: H-Ar, aromatic protons; H-C=C, protons bound to unsaturated carbon atoms; H-C-O, protons bound to oxygenated aliphatic carbon atoms; other H-C-X, protons bound to aliphatic carbon atoms bearing heteroatoms (DMSO and MSA excluded); H-C-C=, protons bound to aliphatic carbon atoms adjacent to unsaturated groups; and H-C, purely alkylic protons. 1993; Cooke et al., 1997]. The BC/TC ratio is significantly low, on average ± for both fine and coarse aerosols. WIOC represents therefore the dominant carbon species, for both fine and coarse size ranges, amounting to 0.66 ± 0.11 mg m 3 and 0.26 ± 0.06 mg m 3, respectively. [26] The results also show that, on average, WSOC concentrations are comparable to those of nss SO 4 2, over the whole fine size range, and that WIOC is the main individual component of the submicron aerosol. This represents, to our best knowledge, the first definite evidence of an enrichment of carbonaceous species in fine remote marine aerosol Water-Soluble Organic Compound Analysis [27] Analysis of organic functional groups was performed by 1 HNMR on both fine and coarse water-soluble fractions of the aerosol [Decesari et al., 2000]. Figures 3 and 4 report the 1 HNMR spectra of the fine and the coarse fractions for 7of14

8 Figure 4. Average concentrations (in mmol H m 3 units) of the main functional group categories described in the text for the fine- and coarse-aerosol fractions. sample 2 9 October and the functional group concentrations averaged for the analyzed set of samples, respectively. The 1 HNMR spectra show that purely alkylic groups, CH- CH, account for the main part of the WSOC hydrogen content, 55% (percentage contribution to the sum of the identified functional groups) in both fine and coarse fractions. Aromatic protons are visible between 7.0 and 8.3 ppm of chemical shift; an additional band of H-C=C protons between 5.2 and 6.2 ppm can be attributed to alkenes. The aromatic/alkene content of WSOC in Mace Head samples is particularly low, accounting for only 1 4% of the total organic hydrogen. This value is also significantly lower than the 7 10% of aromatic fraction found at continental polluted sites using the same 1 HNMR technique [Matta et al., 2003; Mayol-Bracero et al., 2002]. The scarce aromatic/ alkene content of the samples indicates that the H-C-C= protons, i.e., aliphatic hydrogen atoms in alpha position to unsaturated carbon atoms, are mainly in alpha to C=O groups (i.e., carbonyls) instead of C=C double bonds: either ketones, carboxylic groups or amides. Aliphatic carbonyls H-C-C=O (e.g., carboxyls and ketones) account for 25 30% of the total organic hydrogen in the samples and are the most abundant polar functionalities. Aldehydes are unlikely to occur in significant concentrations because the spectral region of H-C=O aldehydic protons (9 10 ppm) contains virtually no signals. These chemical features indicate that aliphatic moieties carrying oxidized functional groups, mainly carboxyls and ketones, are the most common WSOC constituents for the fine and coarse fractions. The H-C-O groups from alcohols and ethers are present mainly in the coarse fraction (16%) of the aerosol. [28] Furthermore, the fine-aerosol samples contain a variable amount of aliphatic C-H groups bound to a heteroatom, either nitrogen, sulphur or a halogen atom, in correspondence to a series of signals between 2.7 ppm and 3.2 ppm of chemical shift. This group may account for 5 20% of the total organic hydrogen content in the fine fraction, depending on the sample, and only for 2% in the coarse fraction. In particular, the sharp peaks at 2.72 and 2.82 ppm are attributable to dimethyl sulphoxide (DMSO) and methane-sulphonic acid (MSA), respectively. The most intense system of peaks, visible between 3.00 and 3.15 ppm, arises from unidentified compounds which are quantified collectively as other H-C-X in Figures 3 and 4. [29] The overall chemical composition observed for the WSOC in the Mace Head aerosol samples cannot be fully explained by previous studies on aerosol characterization in clean marine air masses. The well-known very polar C 2 C 4 di-carboxylic acids [e.g., Kawamura and Sakaguchi, 1999] may contribute to the H-C-C= moieties; however, the most abundant aliphatic groups, i.e., the purely alkylic moieties H-C, rest substantially unaccounted for if we exclude the very few examples of water-soluble organic acids C 7 C 13 with residual aliphatic chains [Kawamura and Gagosian, 1990]. The MSA and DMSO were found in significant concentrations in the fine fraction, but most organic compounds carrying a heteroatom remain uncharacterized at the molecular level. [30] Although there is evidence that organic aerosol particles of continental origin may be transported over long distances across the oceans [Cachier et al., 1986; Mochida et al., 2003b], several spectral features, arising from the 1 HNMR analysis of Mace Head aerosols, cannot be easily reconciled with the available 1 HNMR libraries of WSOC from continental sources. In particular, the band of the H-C-O groups in the coarse aerosols is not 8of14

9 associated to a peak of anomeric protons (chemical shift 5.4 ppm) which identifies the anhydrosugars produced by biomass burning processes. Moreover, the spectral region including the purely alkylic groups ( H-C ) in the fine fraction is rich in signals from methyl groups (chemical shifts: , 1.1, 1.2 ppm) and is more intense between 1.1 and 1.5 ppm. The same spectral region from samples collected in continental polluted areas instead shows relatively weak signals attributable to methyls and peaks always between 1.3 and 1.7 ppm. Finally, the fine-aerosol samples from Mace Head show quite significant concentrations of the functional groups containing a heteroatom (nitrogen, sulphur or halogens), whereas these groups are only minor components of the WSOC composition in continental areas. [31] High-performance liquid chromatography (HPLC) [Decesari et al., 2000] reveals the occurrence of NBC, MDA and PA in the fine- and coarse-aerosol fractions. The air concentrations of the three HPLC fractions show the following ranges of variation (minimum maximum): , , mg cm 3 for NBC, MDA and PA, respectively, in the fine fraction and , , mg cm 3 for the same classes in the coarse fraction. The concentrations of the three chemical classes are positively correlated to total WSOC, with r 2 = 0.90, 0.61, 0.89, respectively, for NBC, MDA and PA in the fine fraction. Therefore the large intervals of variation are attributable mainly to differences in total WSOC concentrations among samples. Mono-/di-carboxylic acids (MDA) are the most abundant identified species, representing over 49 51% of the total WSOC classes, while PA contribute a further 22 27% (the first indicated values refer to the fine fraction and the latter to the coarse one). These findings are in agreement with the 1 HNMR data on polar functionalities. Previous studies on marine aerosol WSOC have shown mainly the presence of MDA, whereas the occurrence of PA in marine aerosol has never been documented before, and their mechanisms of formation are not currently known. A possible source of aerosol PA is the humic substances dissolved in seawater [Stuermer and Harvey, 1977; Oppo et al., 1999]. Humic substances account for the most hydrophobic fraction of marine dissolved organic carbon and are characterized by both polar functional groups and purely aliphatic moieties. In particular, HNMR analysis of marine humic substances isolated from surface water shows a minor contribution from H-C-O protons (3 24%) and a large fraction (72 94%) of aliphatic moieties saturated or bound to C=O groups [Harvey et al., 1983]. This pattern of functionalities encompasses the one observed for the WSOC aerosol samples from Mace Head. It should be noted that humic substances account for only 5 25% on average of the marine dissolved organic carbon, which is dominated by polysaccharide compounds [Hansell and Carlson, 2002]. A possible mechanism of selective transport of humic material into the aerosol is discussed in section Surface Tension Measurements [32] Surface tension measurements were carried out on water extracts of both fine- and coarse-aerosol samples. The average surface tension decreases as a function of WSOC, following the Szyskowski-Langmuir equation s = T ln ( C), where T is the temperature in kelvins and C is the dissolved organic carbon concentration in mol L 1 units, as observed in other different environments [Facchini et al., 1999; Charlson et al., 2001]. Following the results of WSOC chemical characterization, it is reasonable to attribute the observed surface tension decrease to the compounds carrying both polar groups and aliphatic chains, the latter providing a partial hydrophobic character. This fraction of WSOC comprises aliphatic MDA as well as humic-like substances Potential Sampling Artifacts [33] Several studies show that positive and negative artifacts can be very significant especially during filter sampling [Mader et al., 2001, and references therein] and organic aerosol analysis. Positive artifacts can result from adsorption of gaseous species to substrate surface, while negative artifacts arise from particle losses on walls and volatilization of semivolatile species. [34] In this study, WSOC was measured from Berner impactor samples and WIOC was derived from TC and WSOC, as WIOC = TC WSOC BC, both measured from Sierra Hi-vol impactor substrates and back up filters. The meteorological conditions encountered during the sampling periods, i.e., a relative high RH, 68 85%, daytime temperatures generally quite low and never exceeding 15 C and very limited temperature variation <10 C, prevented substantial evaporative losses to occur. Further, the use of the impactor minimized biases of carbonaceous concentration values since, in such sampler, volatilization of organic carbon is generally limited to <4% [Hering et al., 1997] and the condensation of organic vapors on Tedlar impactor substrate is minimal [Turpin et al., 1997; Saxena and Hildemann, 1996]. Consistently, the size-segregated WSOC concentrations are found to be significant with respect to random uncertainties and cannot be due to sampling artifacts in the Berner impactor. [35] As already discussed in section 3.1, the bulk WSOC concentration measured in this study confirms the one previously observed at Mace Head by Kleefeld et al. [2002] and agreement is found between WSOC concentrations from Hi-vol and those from Berner impactor for both fine and coarse fractions; one can therefore conclude that WSOC concentrations are not likely to be distorted in this study. Nevertheless, it is useful to explore the likelihood of artifacts, in particular, positive artifacts associated with the impact-derived results. [36] In order to evaluate any bias potentially associated with impactor data, and to increase the confidence level of the results, the volume distributions derived from the impactor chemical data (originally based on aerodynamic diameters) were compared with the correspondent ones (based on mobility diameters) obtained using a Scanning Mobility Particle Sizer. Significant positive or negative artifacts eventually associated with the impactor data would be reflected in notable differences between the two volume distributions. There are three regions of overlap between the two instruments, stage 1, stage 2 and partially stage 3, the last two being the impactor stages with the least uncertainty in the chemical data, and covering most of the accumulation 9of14

10 Figure 5. Comparison of scanning mobility particle sizer (SMPS) and impactor-derived volume distributions. SMPS data are representative of 40% relative humidity (RH), while impactor data are corrected for density and Cunningham slip. The impactor volume distribution is shown for RH of 40% (assuming growth factor equal to 1) and for 80% RH (growth factors calculated for each stage based on chemical composition). See color version of this figure at back of this issue. mode size range (up to 263 nm for the SMPS and 250 nm for the impactor). As a number of corrections had to be made to ensure a correct comparison, the approach is presented here step by step. [37] The impactor mass size distribution was converted into volume distribution, dividing by the appropriate densities calculated according to the actual chemical composition of individual stages (WSOC, 1.38; WIOC, 1.00; BC, 1.00; sea salt, 2.20; ammonium sulphate/nitrate, 1.75 g cm 3 ). A density of 1.41 and 1.40 was obtained for stages 1 and 2, respectively, close to the values generally accepted for particles of organic composition. The SMPS number size distribution was also converted into volume distribution. [38] Furthermore, it was decided to present the volume distributions in mobility diameter scale to keep the SMPS volume distribution unchanged, since the impactor distribution is a more coarse step function. The conversion from aerodynamic into mobility diameter involves two corrections, the first one accounting for the aerosol particle growth factor (GF). The SMPS operational RH was approximately 40%, while the impactor operated close to the ambient RH of approximately 80%. The calculation of the GF for stages 1 and 2 was based on the actual chemical composition of the individual stages and the growth factors for chemical species as derived from Berg et al. [1998] and Swietlicki et al. [2000] accounting for the difference in RH (WSOC, 1.2; WIOC, 1.0; BC, 1.0; sea salt, 1.8; ammonium sulphate/nitrate, 1.33). Growth factor of 1.20 and 1.21 was obtained for stages 1 and 2, respectively. It was assumed that particles were internally mixed. The second correction concerns the change in density associated with the wet aerosol size. The new or wet density was calculated as: r wet ¼ r dry þ r 0 ðgf 3 1Þ GF 3 ; ð1þ where r dry is the density of a dry internally mixed particle, r 0 is the density of water, and GF is the growth factor of internally mixed particles. In particular, when GF is close to unity the wet density equals dry density, and when GF is close to 2.0 and larger, the wet density approaches unity and no diameter correction is necessary to convert from aerodynamic into mobility diameter or vice versa. [39] The new wet densities for stages 1 and 2 were calculated to be 1.24 and 1.26, respectively. Using the new densities, a diameter correction was made according to C 1=2 aerod D mob ¼ D aerod ; ð2þ rc mob where D mob and D aer are the mobility and aerodynamic diameters, r is the new density obtained by equation (1), and C mob and C aer are Cunningham slip correction factors. [40] The direct comparison of derived aerosol volume distributions is shown in Figure 5 over the SMPS size intervals corresponding to the new, rebined, impactor size intervals according to equations (1) and (2). For stage 1, volumetric loadings of mm 3 m 3 and mm 3 m 3 were obtained for the impactor and SMPS, respectively, leading to an overestimation of 53% for the impactor 10 of 14

11 respect to SMPS. Similarly for stage 2, volumetric loadings of mm 3 m 3 and mm 3 m 3 leaded to an overestimation of 29%. The larger difference for stage 1 can be explained by the higher uncertainties associated to the chemical data because of the extremely low mass loading (very often just few micrograms in total). Furthermore, although the SMPS does not completely cover stage 3 of the impactor, a negligible difference was observed in the range of overlapping (Figure 5). In general, the differences in the volumes derived by the SMPS and impactor are similar to the magnitudes of the random uncertainties associated with each impactor stage as listed in Table 2. Figure 5 also reports the comparison using GF = 1; the agreement is even closer and the difference becomes smaller than 16% for stage 2 and drops to 45% for stage 1. [41] In conclusion, in the accumulation mode range (i.e., stages 2 and 3), the agreement between SMPS and impactor-derived volume distributions is within a few percent, with differences increasing as size decreases; the comparison allows therefore a high degree of confidence in impactor-derived data (especially for those of the accumulation mode), and any significant positive artifacts to be ruled out. 4. Discussion [42] The work reports new evidence on the contribution of organic matter to the clean marine aerosol population. The total organic carbon fraction is observed to increase dramatically as particle diameter decreases, ranging from 3 to 83% of the analyzed mass over the size range investigated. Detailed analyses further show quite similar chemical and physical properties for both fine and coarse organic carbon modes, i.e., the hydrophobic nature (WIOC) of a substantial fraction of the organic matter, and a WSOC composition comprising aliphatic and only partially oxidized species and humic-like substances, resulting in surface-active properties. [43] If we exclude significant anthropogenic influences on the carbon loading of the samples (see section 2.2), the contribution of WIOC to clean marine aerosol samples may be mainly attributed to hydrophobic organic matter which accumulates in the surface film of the ocean and is effectively transferred into the atmosphere through the bursting of entrained air bubbles during whitecap formation [Blanchard, 1983; Oppo et al., 1999]. In the northeast Atlantic Ocean, the biological activity reaches its maximum during the periods studied, corresponding to the spring and autumn plankton blooms [Ducklow and Harris, 1993]. It may therefore account for the large sea-air transfer of hydrophobic organic matter observed. The chemical features and the surface-active properties of the watersoluble organic species further corroborate the hypothesis of a primary emission from the ocean. Indeed, also marine surface-active dissolved organic compounds are enriched in the microlayer at the air-sea interface and may be efficiently scavenged and incorporated into aerosol particles by bubble bursting processes [Blanchard, 1983]. Conversely, other constituents of marine dissolved organic carbon, which are not surface-active (e.g., sugars), are not expected to undergo the same mechanism of enrichment and transport into the aerosol phase. Additionally, the enrichment of WSOC carrying heteroatoms (either nitrogen, sulphur and halogens) in fine particles as shown by the 1 HNMR analysis provides another indication of the potential biological sources for the fine fraction of the aerosol. [44] There exists a great deal of evidence of the occurrence of lipidic species in marine aerosol, for which marine biogenic origin has been postulated [Gogou et al., 1998; Mochida et al., 2002; Tervahattu et al., 2002a, 2002b]; a strong confirmation to our hypothesis on a dominant primary source from the biological active waters of the North Atlantic derives from the laboratory work of Gershey [1983]; the experiments demonstrate that the production of particles by bubbling in North Atlantic seawater discriminates against the more soluble low molecular weight compounds in favor of the more surface-active high molecular weight compounds in periods of blooming plankton. [45] Moreover, the enrichment of carbonaceous species observed in the fine fraction of the aerosol samples is supported by the thermodynamic model calculations of Oppo et al. [1999] on the bubble bursting process. Under conditions where the sea surface microlayer becomes concentrated of surface-active matter, bursting of bubbles is predicted to generate spray drops covered by a saturated organic adsorption film, thus producing a transition in the chemical composition of particles from almost pure inorganic salt to organic enriched aerosol, as the particle size decreases. [46] The high organic fraction observed for the fine-mode sizes supports previous suggestions of the enrichment of the organic compounds compared to nss sulphate in marine particles of D < 100 nm [Novakov et al., 1997; Middlebrook et al., 1998]. However, the fine aerosols at Mace Head possess quite different attributes compared to aerosol particles characterized for their hygroscopic properties in other clean marine areas [Berg et al., 1998; Swietlicki et al., 2000; Zhou et al., 2001; Maßling et al., 2003]. In particular, the high insoluble (organic) fraction of particles of D < 250 nm (Figure 2) cannot be reconciled with the average growth factor of 1.6 most frequently observed by HTDMA measurements of marine particles in the same size range. Indeed, such growth factor would correspond to a soluble fraction of 100%, assuming the soluble material is all nss sulphate [Berg et al., 1998], or 50% in the case of sodium chloride [Ming and Russell, 2001]. However, the CCN concentrations calculated on the basis of such measured hygroscopic growth factors (showing a high soluble fraction) tend to overestimate measured CCN concentrations [Covert et al., 1998], suggesting a lower solubility in marine aerosol than that inferred from the growth factor data. Therefore the high insoluble organic fraction observed in clean marine aerosol can, at least partially, explain the failure of many attempted closure experiments on the marine aerosol CCN system, where large discrepancies still exist between measurements and model predictions [Covert et al., 1998; Wood et al., 2000; Snider and Brenguier, 2000]. It should be noted, however, that in some maritime regions, closure has been obtained for assumptions of predominant ammonium sulphate nuclei composition [VanReken et al., 2003], indicating that organic fraction in marine aerosol may vary in time and location. 11 of 14

12 [47] While our chemical composition results remain inconsistent with hygroscopic growth factor data over other oceanic regions, it should be noted that the northeast Atlantic is one of the most biologically active oceanic regions, and that the aerosol sample collection took place during spring and autumn phytoplankton blooms; in this context, the results presented here are not to be unexpected. [48] Finally, since the sampling was performed at a coastal site, the possibility of coastal influences should also be raised. Recent studies during a shoreline experiment in Hawaii demonstrated that breaking waves could be identified from a coastal sampling tower [Clarke et al., 2003] and that the breaking wave plumes were evident above the background concentrations. Analysis of Mace Head data using 1-Hz condensation particle counters revealed no such evidence of breaking waves contributing to the general aerosol signal. The difference appears to be due to the significantly smaller surf zone around the rocky shores of Mace Head compared to the flat sandy beaches of Hawaii. In addition, during a separate study, an Aerodyne aerosol mass spectrometer (AMS) was operated at Mace Head during the NAMBLEX experiment, cycling between 7 and 22 m levels on an hourly cycle. Results from these two sampling levels indicated no noticeable difference in either particle concentration or composition between the two levels (J. Allen, personal communication, 2004), suggesting that the local surf zone had a minimal effect on the aerosol fields measured. Further material corroborating the argument that coastal influences are not significant is included as auxiliary material Conclusions [49] This paper reports the results of comprehensive sizesegregated analyses of clean marine aerosol samples collected at Mace Head during periods of phytoplankton bloom (spring and autumn), providing new evidence on the chemical composition of marine aerosol. The supermicron mode predominantly comprises sea salt aerosol with a mass concentration of ± 0.8 mg m 3 ; to the remaining, nss sulphate contributes 0.03 ± 0.01 mg m 3, while nitrate contributes 0.13 ± 0.04 mgm 3. By comparison, the mass of sea salt, nss sulphate and nitrate in the submicron mode is found to be 0.39 ± 0.08 mg m 3, 0.26 ± 0.04 mg m 3 and 0.02 ± 0.01 mg m 3, respectively. [50] WSOC is observed in the submicron mode with a mass concentration of 0.25 ± 0.04 mg m 3, comparable to that of nss sulphate, and with a mass concentration of 0.17 ± 0.04 mg m 3 in the supermicron mode. The WSOC to TC ratio is found to be 0.20 ± 0.12 for the submicron fraction and 0.29 ± 0.08 for supermicron fraction, while the BC to TC ratio is found to be on average ± for both aerosol modes. The remaining carbon, WIOC, contributes 0.66 ± 0.11 mg m 3 and 0.26 ± 0.06 mg m 3 to the submicron and supermicron modes, respectively, and represents the main aerosol constituent in the submicron mode, accounting alone for 39% of the total 1 Auxiliary material is available at ftp://ftp.agu.org/apend/jd/ 2004JD measured mass. Detailed analyses show WSOC chemical and physical features common to both fine- and coarseaerosol ranges, with a substantial contribution of aliphatic and humic-like substances and appreciable surface-active properties. [51] The observed size-dependent concentration of organic matter, together with its chemical and physical properties (water solubility, WSOC chemical composition and surface tension properties), indicate an important biogenic source of these particles, produced by bubble bursting processes over the biologically active waters of the northeast Atlantic. The relative importance of sea salt and nss SO 2 4 in determining the direct and indirect radiative properties of aerosol in the MBL has been the 2 subject of extensive debate. However, because nss SO 4 is not the dominant component of submicron aerosol mass, as our results suggest, both soluble and insoluble organic compounds, along with sea salt, will largely contribute to aerosol cloud nucleating properties in the North Atlantic ocean during periods of blooming plankton. Further effort is required to better quantify the components of accumulation mode and Aitken mode marine aerosols with improved aerosol-chemical instrumentation and to improve the speciation of both the WSOC and WIOC components. [52] Acknowledgments. The authors would like to thank Philip O Brien for generating the air mass back trajectory maps, Peter Simmonds for providing CO data, and Carsten Junker for providing the Aethalometer BC data. We gratefully acknowledge Erik Swietlicki for his constructive criticism during the manuscript preparations. This work was partly supported by the European Commission (Projects QUEST and PHOENICS), Italian Ministry of Environment (Italy USA Cooperation on Science and Technology of Climate Change), Irish Higher Educational Authority and the Irish Research Council for Science Engineering and Technology Embark Initiative. References Berg, O. H., E. Swietlicki, and R. Krejci (1998), Hygroscopic growth of aerosol particles in the marine boundary layer over the Pacific and Southern oceans during the First Aerosol Characterization Experiment (ACE 1), J. Geophys. Res., 103, 16,535 16,545. Blanchard, D. C. (1954), Bursting of bubbles at an air-water interface, Nature, 173, Blanchard, D. C. (1964), Sea to air transport of surface active material, Science, 146, Blanchard, D. C. (1983), The production, distribution and bacterial enrichment of sea-salt aerosol, in Air-Sea Exchanged of Gases and Particles, edited by P. S. Liss and W. G. N. Slinn, pp , D. Reidel, Norwell, Mass. Cachier, H., P. Buat-Ménard, M. Fontugne, and R. Chesselet (1986), Longrange transport of continentally-derived particulate carbon in the marine atmosphere: Evidence from stable carbon isotope studies, Tellus, Ser. B, 38, Cachier, H., M. P. Bremond, and P. Buat-Menard (1989), Determination of atmospheric soot carbon with a simple thermal method, Tellus, Ser. B, 41, Charlson, R. J., J. E. Lovelock, M. O. Andreae, and S. G. Warren (1987), Oceanic phytoplankton, atmospheric sulphur, cloud albedo and climate, Nature, 326, Charlson, R. J., J. H. Seinfeld, A. Nenes, M. Kulmala, A. Laaksonen, and M. C. Facchini (2001), Reshaping the theory of cloud formation, Science, 292, Chung, S. H., and J. H. Seinfeld (2002), Global distribution and climate forcing of carbonaceous aerosols, J. Geophys. Res., 107(D19), 4407, doi: /2001jd Clarke, A. D., V. Kapustin, S. Howell, K. Moore, B. Lienert, S. Masonis, T. Anderson, and D. Covert (2003), Sea-salt size distributions from breaking waves: Implications for marine aerosol production and optical extinction measurements during SEAS, J. Atmos. Oceanic Technol., 20, of 14

13 Cooke, W. F., S. G. Jennings, and T. G. Spain (1997), Black carbon measurements at Mace Head, , J. Geophys. Res., 102, 25,339 25,346. Covert, D. S., J. L. Gras, A. Wiesensohler, and F. Stratmann (1998), Comparison of directly measured CCN with CCN modeled from the numbersize distribution in the marine boundary layer during ACE 1 at Cape Grim, Tasmania, J. Geophys. Res., 103, 16,597 16,608. Decesari, S., M. C. Facchini, S. Fuzzi, and E. Tagliavini (2000), Characterization of water-soluble organic compounds in atmospheric aerosol: A new approach, J. Geophys. Res., 105, Ducklow, H., and R. Harris (Eds.) (1993), JGOFS North Atlantic Bloom Experiment (NABE), Deep Sea Res., Part II, 40, Facchini, M. C., M. Mircea, S. Fuzzi, and R. J. Charlson (1999), Cloud albedo enhancement by surface-active organic solutes in growing droplets, Nature, 401, Facchini, M. C., S. Decesari, M. Mircea, S. Fuzzi, and G. Loglio (2000), Surface tension of atmospheric wet aerosol and cloud/fog droplets in relation to their organic carbon content and chemical composition, Atmos. Environ., 34, Forster, C., et al. (2001), Transport of boreal forest fire emissions from Canada to Europe, J. Geophys. Res., 106(D19), 22,887 22,906. Fuzzi, S., S. Decesari, M. C. Facchini, E. Matta, and M. Mircea (2001), A simplified model of the water soluble organic components of atmospheric aerosols, Geophys. Res. Lett., 28, Gershey, R. M. (1983), Characterization of seawater organic matter carried by bubble-generated aerosols, Limnol. Oceanogr., 28, Gogou, A. I., M. Aposolaki, and E. G. Stephanou (1998), Determination of organic molecular markers in marine aerosols and sediments: One-step flash chromatography compound class fractionation and capillary gas chromatographic analysis, J. Chromatogr. A, 799, Gray, H. A., R. G. Cass, J. J. Huntzicker, E. K. Heyrdahl, and J. A. Rau (1986), Characteristics of atmospheric organic and elemental carbon particle concentration in Los Angeles, Environ. Sci. Technol., 20, Hansell, D. A., and C. A. Carlson (Eds.) (2002), Biogeochemistry of Marine Dissolved Organic Matter, Academic, San Diego, Calif. Harvey, G. R., D. A. Boran, L. A. Chesal, and J. M. Tokar (1983), The structure of marine fulvic and humic acids, Mar. Chem., 12, Hering, S., A. Eldering, and J. H. Seinfeld (1997), Bimodal character of accumulation mode aerosol mass distributions in southern California, Atmos. Environ., 31, Hoffman, E. J., and R. A. Duce (1976), Factor influencing the organic carbon content of marine aerosols: A laboratory study, J. Geophys. Res., 81, Holloway, T., H. Levy II, and P. Kasibhatla (2000), Global distribution of carbon monoxide, J. Geophys. Res., 105, 12,123 12,147. Huebert, B. J., S. G. Howell, L. Zhuang, J.-A. Heath, M. R. Litchy, D. J. Wylie, J. L. Kreidler-Moss, S. Cöppicus, and J. E. Pfeiffer (1998), Filter and impactor measurements of anions and cations during the First Aerosol Characterization Experiment (ACE1), J. Geophys. Res., 103, 16,493 16,509. Jennings, S. G., C. Kleefeld, C. D. O Dowd, C. Junker, T. G. Spain, P. O Brien, A. F. Roddy, and T. C. O Connor (2003), Mace Head atmospheric research station Characterization of aerosol radiative parameters, Boreal Environ. Res., 8, Kawamura, K., and R. B. Gagosian (1990), Mid-chain ketocarboxylic acids in the remote marine atmosphere: Distributions patterns and possible formation mechanisms, J. Atmos. Chem., 11, Kawamura, K., and F. Sakaguchi (1999), Molecular distributions of watersoluble dicarboxylic acids in marine aerosols over the Pacific Ocean including tropics, J. Geophys. Res., 104, Kleefeld, S., A. Hoffer, Z. Krivacsy, and S. G. Jennings (2002), Importance of organic and black carbon in atmospheric aerosols at Mace Head, on the west coast of Ireland, Atmos. Environ., 36, Mader, B. T., R. C. Flagan, and J. H. Seinfeld (2001), Sampling atmospheric carbonaceous aerosols using a particle trap impactor/denuder sampler, Environ. Sci. Technol., 35, Maßling, A., A. Wiedersohler, B. Busch, C. Neusüß, P. Quinn, T. Bates, and D. Covert (2003), Hygroscopic properties of different aerosol types over the Atlantic and Indian oceans, Atmos. Chem. Phys. Discuss., 3, Matsumoto, K., H. Tanaka, I. Nagao, and Y. Ishizaha (1997), Contribution of particulate sulfate and organic carbon to cloud condensation nuclei in marine atmosphere, Geophys. Res. Lett., 24, Matta, E., M. C. Facchini, S. Decesari, M. Mircea, F. Cavalli, S. Fuzzi, J.-P. Putaud, and A. Dell Acqua (2003), Chemical mass balance of size-segregated atmospheric aerosol in an urban area of the Po Valley, Italy, Atmos. Chem. Phys., 3, Mayol-Bracero, O. L., P. Guyon, B. Graham, G. C. Roberts, M. O. Andreae, S. Decesari, M. C. Facchini, S. Fuzzi, and P. Artaxo (2002), Water-soluble organic compounds in biomass burning aerosols over Amazonia: 2. Apportionment of the chemical composition and importance of the polyacidic fraction, J. Geophys. Res., 107(D20), 8091, doi: /2001jd Middlebrook, A. M., D. M. Murphy, and D. S. Thomson (1998), Observation of organic material in individual particles at Cape Grim during the First Aerosol Characterization Experiment (ACE 1), J. Geophys. Res., 103, 16,475 16,483. Ming, Y., and L. M. Russell (2001), Predicted hygroscopic growth of sea salt aerosol, J. Geophys. Res., 106, 28,259 28,274. Mochida, M., Y. Kitamori, K. Kawamura, Y. Nojiri, and K. Suzuki (2002), Fatty acids in the marine atmosphere: Factors governing their concentrations and evaluation of organic films on sea salt particles, J. Geophys. Res., 107(D17), 4325, doi: /2001jd Mochida, M., A. Kawabata, K. Kawamura, H. Hatsushika, and K. Yamazaki (2003a), Seasonal variation and origins of dicarboxylic acids in the marine atmosphere over the western North Pacific, J. Geophys. Res., 108(D6), 4193, doi: /2002jd Mochida, M., K. Kawamura, N. Umemoto, M. Kobayashi, S. Matsunaga, H-J. Lim, B. J. Turpin, T. S. Bates, and B. R. T. Simoneit (2003b), Spatial distribution of oxygenated organic compounds (dicarboxylic acids, fatty acids, and levoglucosan) in marine aerosol over the western Pacific and off the cost of East Asia: Continental outflow of organic aerosols during Ace-Asia campaign, J. Geophys. Res., 108(D23), 8638, doi: / 2002JD Murphy, D. M., J. R. Anderson, P. K. Quinn, L. M. McInnes, F. J. Brechtell, S. M. Kreidenweis, A. M. Middlebrook, M. Pòsfai, D. S. Thomson, and P. R. Buseck (1998), Influence of sea-salt on aerosol radiative properties in the Southern Ocean marine boundary layer, Nature, 392, Novakov, T., C. E. Corrigan, J. E. Penner, C. C. Chuang, O. Rosario, and O. L. Mayol Bracero (1997), Organic aerosols in the Caribbean trade winds: A natural source?, J. Geophys. Res., 102, 21,307 21,313. O Dowd, C. D., and M. H. Smith (1993), Physicochemical properties of aerosol over the northeast Atlantic: Evidence for wind-speed-related submicron sea-salt aerosol production, J. Geophys. Res., 98, O Dowd, C. D., M. H. Smith, and S. G. Jennings (1993), Submicron aerosol, radon and soot carbon characteristics over the northeast Atlantic, J. Geophys. Res., 98, O Dowd, C. D., M. H. Smith, I. E. Consterdine, and J. A. Lowe (1997), Marine aerosol, sea salt, and marine sulphur cycle: A short review, Atmos. Environ., 31, Oppo, C., S. Bellandi, N. degli Innocenti, A. M. Stortini, G. Loglio, E. Schiavuta, and R. Cini (1999), Surfactant components of marine organic matter as agents for biogeochemical fractionation and pollutant transport via marine aerosols, Mar. Chem., 63, Putaud, J.-P., et al. (2000), Chemical mass closure and assessment of the origin of the submicron aerosol in the marine boundary layer and the free troposphere at Tenerife during ACE-2, Tellus, Ser. B, 52, Quinn, P. K., D. J. Coffman, V. N. Kapustin, T. S. Bates, and D. S. Covert (1998), Aerosol optical properties in the marine boundary layer during ACE 1 and the underlying chemical and physical aerosol properties, J. Geophys. Res., 103, 16,547 16,563. Quinn, P. K., et al. (2000), Surface submicron aerosol chemical composition: What fraction is not sulfate?, J. Geophys. Res., 105, Savoie, D. L., J. M. Prospero, and E. S. Saltzman (1989), Nitrate, nonseasalt sulphate and methansulfonate over the Pacific Ocean, in Chemical Oceanography, vol. 10, pp , edited by J. P. Riley, R. Chester, and R. A. Duce, Academic, San Diego, Calif. Saxena, P., and L. M. Hildemann (1996), Water-soluble organics in atmospheric particles: A critical review of the literature and application of thermodynamics to identify candidate compounds, J. Atmos. Chem., 24, Snider, J. R., and J.-L. Brenguier (2000), Cloud condensation nuclei and cloud droplet measurements during ACE-2, Tellus, Ser. B, 52, Stuermer, D., and G. Harvey (1977), The isolation of humic substances and alcohol soluble organic matter from seawater, Deep Sea Res., 24, Swietlicki, E., et al. (2000), Hygroscopic properties of aerosol particles in the north-eastern Atlantic during ACE-2, Tellus, Ser. B, 52, Tervahattu, H., K. Hartonen, V.-M. Kerminen, K. Kupiainen, P. Aarnio, T. Koskentalo, A. F. Tuck, and V. Vaida (2002a), New evidence of an organic layer on marine aerosols, J. Geophys. Res., 107(D7), 4053, doi: /2000jd Tervahattu, H., J. Juhanoja, and K. Kupiainen (2002b), Identification of an organic coating on marine aerosol particles by TOF-SIMS, J. Geophys. Res., 107(D16), 4319, doi: /2001jd of 14

14 Turpin, B. J., P. Saxena, G. Allen, P. Koutrakis, P. McMurry, and L. Hildemann (1997), Characterization of the southwestern desert aerosol, Meadview, AZ, J. Air Waste Manage. Assoc., 47, VanReken, T. M., T. A. Rissman, G. C. Roberts, V. Varutbangkul, H. H. Jonsson, R. C. Flagan, and J. H. Seinfeld (2003), Toward aerosol/cloud condensation nuclei (CCN) closure during CRYSTAL-FACE, J. Geophys. Res., 108(D20), 4633, doi: /2003jd Wang, S. C., and R. C. Flagan (1990), Scanning electrical particle sizer, Aerosol Sci. Technol., 13, Wood, R., D. W. Johnson, S. R. Osborne, B. J. Bancy, M. A. Andreae, C. D. O Dowd, P. Glantz, and K. J. Noone (2000), Boundary layer, aerosol and chemical evolution during the third Lagrangian experiment of ACE-2, Tellus, Ser. B, 52, Zappoli, S., et al. (1999), Inorganic, organic and macromolecular components of the fine aerosol in different areas of Europe in relation to their water solubility, Atmos. Environ., 33, Zhou, J., E. Swietlicki, and O. H. Berg (2001), Hygroscopic properties of aerosol particles over the central Arctic Ocean during summer, J. Geophys. Res., 106, 32,111 32,123. F. Cavalli, S. Decesari, L. Emblico, M. C. Facchini, S. Fuzzi, and M. Mircea, Istituto di Scienze dell Atmosfera e del Clima, CNR, Via Gobetti 101, Bologna, Italy. (f.cavalli@isac.cnr.it) D. Ceburnis, C. D. O Dowd, and Y. J. Yoon, Department of Experimental Physics and Environmental Change Institute, National University of Ireland, University Road, Galway, Ireland. A. Dell Acqua and J.-P. Putaud, Institute for Environment and Sustainability, European Commission, Joint Research Centre, Ispra, Italy. 14 of 14

15 Figure 1. Typical air mass trajectories for the clean marine sample during 2 9 October 2002 and April 2002 representing Arctic, northern Canadian, and mid-atlantic air masses. 3of14