Individual particles produced from atomized rainwater samples collected in California and the Indian Ocean were analyzed with an aerosol time-of-flight mass spectrometer (ATOFMS) to investigate the chemical composition of the individual rain residue particles. Insoluble residue particle types were determined on the basis of a comparison of the rainwater particle mass spectra with ambient particle spectra. Major particle types found in rainwater include dust, organic carbon with sodium, aromatic organic carbon, vegetative detritus, and an internally mixed sea salt and elemental carbon class. A unique internally mixed sea salt–elemental carbon particle type was detected in both the ambient and rainwater samples, suggesting this particle type was most likely formed by cloud processing occurring during long-range transport. The presence of this particle type in remote marine locations has important climate ramifications as it is anticipated it will be strongly absorbing on the basis of the combination of an absorbing particle (elemental carbon) mixed with a high refractive index material (sea salt). Most of the particle types detected in rainwater were detected in the ambient particles with the exception of a unique aromatic particle type detected in rainwater samples from both locations. The presence of the aromatic type coupled with the absence of biomass particles in the rainwater samples leads to the hypothesis the aromatic components were originally associated with atmospheric biomass burning particles. The ubiquitous presence of this aromatic type in rainwater samples highlights the potential importance of biomass burning and/or humic-like substances (HULIS) compounds in cloud formation and rain processes.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Aerosols play an important role in the hydrological cycle, climate, and supply of nutrients to the oceans. They scatter and absorb sunlight, act as CCN, and provide surfaces for heterogeneous chemical reactions to occur upon. Ambient particles typically have lifetimes of one to two weeks during which time they can be transported long distances before they are removed from the atmosphere through dry or wet deposition. By characterizing the removal processes and particle lifetimes, the overall effects of aerosols on climate will be better understood.
 Although aerosols play an important role in the radiative balance of the earth, it is challenging to include them explicitly in global models [Kiehl and Trenberth, 1997]. Aerosols can affect the climate both directly and indirectly. The direct effect of aerosols involves scattering or absorbing incoming solar radiation. The indirect effects include changes in cloud cover due to the availability of cloud condensation nuclei. An increase in aerosol concentration due to anthropogenic activity leads to smaller, more numerous cloud droplets [Twomey, 1977; Wilcox et al., 2006]. This change may reduce precipitation efficiency [Albrecht, 1989; Rosenfeld, 1999] and make clouds brighter, and thus more reflective/cooling. The indirect effect of atmospheric aerosols is expected to have a considerable impact on the global climate system; this could be large enough in some model scenarios to cancel the estimated warming effect by all greenhouse gases [Intergovernmental Panel on Climate Change, 2001]. Large uncertainties in the indirect effect result from the quite limited fundamental knowledge of aerosol particles, especially concerning their global abundance and distribution, physical and chemical characteristics, and interaction with cloud processes. Current global climate models predict a larger warming of the atmosphere near the earth's surface than is currently observed [Haywood and Boucher, 2000]. This discrepancy between prediction and observation is likely linked to the fact that the models do not adequately describe the radiative effect of aerosols.
 Aerosols are removed from the atmosphere through either wet or dry deposition. Here we discuss the processes involved in wet deposition. In wet deposition, particles can be removed either by acting as the nucleus of a cloud droplet which then grows large enough to precipitate out as a raindrop, or through scavenging by the raindrop as it is falling. Studies show that the probability of scavenging is low for particles in the 0.2–2.0 μm size range, because as the drop is falling, it tends to push air and particles out of the way rather than collecting and absorbing them [Horn et al., 1988; Chate, 2005]. This suggests the nucleation mechanism serves as the dominant removal pathway for particles, a process which depends on particle size and chemistry. Thus the composition of rainwater may provide insight to the composition of CCN and cloud droplets.
 Few previous studies have described insoluble individual particles in rainwater. Schutz and Kramer  found a possible size dependence in the removal process of insoluble particles (d = 0.2–200 μm) in rural rainwater samples. Ro et al.  used low-Z electron probe X-ray microanalysis to investigate the chemical composition of Asian dust and found carbonaceous particles were associated with the dust particles. Matthias-Maser et al.  measured aerosols in cloud water and found 25% of the total insoluble particles were biological. Additionally, Ma et al.  investigated insoluble individual particles in rainwater using X-ray fluorescence and suggested that dust became incorporated into cloud droplets during transport and washed out during rainfall. These single particle studies provide valuable information on the links between ambient aerosols and rain samples, however many conclusions were drawn from a limited sample size (i.e., tens of particles). Furthermore, information was obtained on the inorganic components only.
 In addition to studies of inorganic components of insoluble single particles in rainwater, several studies have investigated both water soluble and insoluble organic matter occurring in fog and cloud water to determine the potential sources and optical properties. A substantial fraction of organic carbon, 23% on average, was found to be insoluble in fog water [Herckes et al., 2002]. Also, biomass burning particles have been shown to consist of soluble and insoluble carbonaceous components [Pósfai et al., 2004]. A large fraction of the biomass burning aerosol consists of water soluble organic carbon and HULIS [Mayol-Bracero et al., 2002]. The HULIS has been hypothesized to form during biomass burning from the incomplete breakdown of polymeric carbohydrates [Mayol-Bracero et al., 2002] or formed by aqueous phase processing of smaller phenol compounds and other precursors in fog and cloud water [Zappoli et al., 1999; Graham et al., 2002; Gelencser et al., 2003]. HULIS characterized in cloud water, and hence CCN precursors, were found to be produced from atmospheric polymerization from low molecular weight organics [Feng and Moller, 2004]. In addition to their importance in cloud formation, the HULIS in cloud water can absorb sunlight [Gelencser et al., 2003]. Single particle mass spectrometry can provide additional information including improved statistics by analyzing a greater number of particles, direct evidence of chemical mixing state of the insoluble compounds, information on organic compounds, and most significantly, the ability to directly compare rainwater with atmospheric aerosol measurements made over the course of the rain event with the same instrumentation.
 Here we present the first results involving the use of single particle mass spectrometry for the analysis of atomized rainwater particles, showing a direct comparison of the chemical composition of aerosols suspended from rainwater to concurrently sampled ambient atmospheric particles. Rainwater samples were compared from two field campaigns that aimed to characterize the impacts of long-range transport on aerosol chemistry conducted in the United States and India. Results from these studies indicate this approach may offer a unique way to gain insight into the relationship between ambient aerosols and precipitation processes.
2. Experimental Overview
 The field transportable single particle aerosol time-of-flight mass spectrometry instrument (ATOFMS) measures aerodynamic size (Da) and chemical composition of 0.2 to 3.0 μm diameter aerosol particles in real time [Gard et al., 1997]. The ATOFMS system consists of a differentially pumped sampling interface, light scattering region, desorption/ionization region, and a dual polarity mass spectrometer. Aerosols enter the sampling interface and pass from atmospheric pressure through a converging nozzle (d = 0.340 mm) (10−2 hPa), two skimmers (10−5 hPa), and are accelerated to a terminal velocity dependent on their aerodynamic diameter. Smaller particles are accelerated by the gas stream to a higher velocity. The sampling interface allows particles to pass from atmospheric pressure to the high vacuum needed for the mass spectrometer (10−7 hPa). Particles are sized with two orthogonal continuous wave green lasers (λ = 532 nm) separated by 6 cm. Scattered light is collected with an elliptical mirror and focused on a photomultiplier tube. The particle time of flight (TOF) is related to its aerodynamic size and determined empirically by performing a calibration using standard polystyrene latex spheres of a known diameter. The particle TOF across the 6 cm laser spacing is used to calculate the precise firing time for the desorption/ionization laser (λ = 266 nm). The LDI laser generates 5 ns pulses of ∼1.2 mJ, which are focused to ∼400 μm spot sizes for a power density of ∼1 × 108 W/cm2). For each individual particle, positive and negative ions are simultaneously extracted orthogonally and analyzed with a time of flight mass spectrometer.
 Particles were generated from the rainwater samples using a Collison type atomizer. The generated aerosol was dried using silica diffusion driers before entering the ATOFMS. Particles generated by this method may be soluble or insoluble [Wiedensohler et al., 1994]. Ambient aerosols and atomized rainwater samples were directly sent to an ATOFMS instrument and analyzed to determine their aerodynamic size and chemical composition.
 The ATOFMS raw data were calibrated and a list of peaks was generated, and subsequently imported into MATLAB 188.8.131.520 Release 12.1 (The MathWorks, Inc.) using YAADA 1.20 (J. O. Allen, Yet another ATOFMS data analyzer (YAADA): Software toolkit to analyze single-particle mass spectral data, available at http://www.yaada.org/). Mass spectral data were analyzed either by searching for specific compounds, using selective mass-to-charge markers, or by using the ART-2a cluster analysis method which groups chemically similar aerosols together into clusters. ART-2a was run using the following parameters: vigilance factor = 0.8, learning rate = 0.05, 20 iterations, and dual polarity [Song et al., 1999]. Ambient particle data in both studies were combined, and ART-2a was run on the joint set to allow direct comparison between locations. The rainwater samples were analyzed with ART-2a separately for each study as the numbers of particles in each study were quite different. Further classification was done by hand grouping chemically similar clusters together to reduce the total number of clusters from ∼40 to less than 10. It is important to note that throughout this paper, mass spectral peak assignments represent the most likely ion peak based on the complete mass spectrum and standards run in previous laboratory studies [Silva et al., 1999, 2000].
 The Cloud Indirect Forcing Experiment (CIFEX) 1–21 April 2004 was conducted at Trinidad Head, a coastal site in northern California chosen to represent a clean marine monitoring site with periodic long-range transport of aerosols from the Asian continent across the Pacific. A primary goal was to study the influence of aerosols on cloud properties [Roberts et al., 2006; V. Ramanathan, Cloud Indirect Forcing Experiment (CIFEX), available at http://borneo.ucsd.edu/cifex]. The sampling site was located on the summit of Trinidad Head, (41.05°N, 124.15°W, 107 m) on the coast of northern California. Ambient aerosols were collected through a 10 m stack installed by National Oceanic and Atmospheric Administration's Climate Monitoring and Diagnostic Laboratory. The relative humidity was controlled to 55% by heating. We have found in marine locations that controlling the RH is important for reducing the negative impact of water on the mass spectra. The hit rate (ratio of the number of particles with mass spectra to total sized particles) averaged 13% over the entire study. The flow from the conditioner was split using a stainless steel sampling manifold and sent to the various instruments.
 During CIFEX there were two main periods of precipitation after a long dry period of more than 10 d. Rain occurred from DOY 105–107 and DOY 111–113. CIFEX rainwater samples were collected in a glass beaker at ground level for several hours on 4 d in April 2004 at Trinidad Head. The volume of rain collected ranged from 25 to 100 mL (Table 1). Samples less than 100 mL were diluted with Milli-Q water to provide enough volume to atomize. The total number of particles measured per sample ranged from 266 to 601 (Table 1) and the relative number concentration of different particle types varied among the samples. A total of 535 single particle mass spectra were obtained from the four samples, giving an average hit fraction of 22%.
Table 1. Rainwater Sample Collection Volumes and Particles Detected
 The Atmospheric Brown Cloud (ABC) Post Monsoon Experiment (APMEX) was conducted in the fall of 2004 to investigate the seasonal monsoonal transition period. Instruments were housed in the ABC-Super Observatory on Hanimaadhoo Island, Republic of Maldives in the Indian Ocean (6.77°N, 73.18°E, 1 m) (V. Ramanathan, Atmospheric Brown Cloud Post Monsoon Experiment (APMEX), available at http://www-abc-asia.ucsd.edu/APMEX/october2004campaign.htm). The APMEX site was chosen to follow up on Indian Ocean Experiment (INDOEX) conducted in 1999 [Ramanathan et al., 2001] which found surprisingly high pollution levels over the northern Indian Ocean [Lelieveld et al., 2001]. Hanimaadhoo is an ABC Supersite in a clean marine environment seasonally influenced by outflow from the Indian subcontinent [Corrigan et al., 2006]. A major objective of the APMEX campaign was to understand the long-range transport of black carbon.
 During APMEX, ambient aerosols were sampled through a 15 m high community inlet. The ATOFMS instrument was housed in an air conditioned room immediately outside the aerosol sampling room. To reduce sampling losses and possibilities of condensation in sampling lines, a 5 L/min flow rate was used, with 1 L/min going to the ATOFMS instrument, and 4 L/min exhausting through a tee before the inlet. Sampling lines after the stainless steel sampling manifold consisted of 1/4″ stainless steel tubing (∼2.5 m long), with a short section of black conductive tubing used at either end for convenience.
 The APMEX campaign was conducted toward the end of the wet season and the beginning of the transition to the dry monsoon season. The ambient sampling period was from 15 October to 5 November 2004. During this period, the relative humidity was 79 ±6%, and the ATOFMS hit fraction was 10%. Ten rainwater samples were collected in glass beakers atop the 15 m tower between 19 and 29 October 2004. There were three periods of precipitation DOY 294–295, 297–298, and 301–306. During the second two periods, rainfall was quite intense, up to 115 mm/d. Collection times ranged from 30 min to 16 h with most samples collected in a couple of hours. Because of limited instrument availability, seven samples were analyzed and results are presented here. As shown in Table 1, a total of 2,317 single particle mass spectra were obtained from these seven rainwater samples, with an average hit fraction of 39%.
 Rainwater samples collected during two field campaigns were analyzed using ATOFMS to investigate the chemical composition of single aerosol particles suspended in rainwater. The rainwater residue particles and their relative amounts are presented in conjunction with ambient aerosol measurements.
 In addition to sampling rainwater, ambient measurements were made before, during, and after the rain events. The rainwater samples were analyzed after the rain events. This strategy allows direct comparison of rainwater samples acquired concurrently with ambient measurements. Not surprisingly, in both the CIFEX and APMEX studies, the composition of rainwater particles differed from the ambient particles, as shown in the relative number concentrations presented in Figure 1. More diversity is seen in the CIFEX rainwater particles compared to the APMEX rainwater particles. This is consistent with the ambient aerosols being more complex during CIFEX also, particularly in the submicron fraction. In CIFEX, the single largest class (an internal mixture of dust and sea salt) represented 43% of the total particles sampled, whereas in APMEX, the largest class (organic carbon with sodium) represented 73% of the particles.
 The Collison atomization process used for the rainwater samples can generate soluble particles, insoluble particles, or insoluble particles with a coating of soluble material [Wiedensohler et al., 1994]. Because the ATOFMS does not measure the solubility directly, the water solubility was inferred from the composition as determined from the ions forming the mass spectra. The insoluble particles were either composed of water insoluble elements, or those that were soluble, but bound together into a single residue particle. Additionally, particle types from rainwater samples were compared to ambient particles. The identification of chemically similar particles observed in rainwater and ambient aerosols provided further evidence as to which ones were insoluble.
Figure 1a shows a pie chart representing all of the chemical types from the sum of all CIFEX rainwater samples. The most common rainwater particle types observed during CIFEX (Figure 1a) were dust internally mixed with sea salt (Dust (SS)) which represented 43% of the total particles, an organic carbon type with mass spectral ion peaks indicative of an aromatic/HULIS compounds at m/z 51, 63, 77, 91, and 115 (OC Aromatic) (23%), vegetative detritus (16%), dust (10%), and a mixed sea salt and elemental carbon class (SS-EC) (5%). The APMEX rainwater samples (Figure 1d) were composed of an organic carbon with sodium class OC (Na) (73%), 49% of which was associated with sulfate, SS-EC (8%), K-rich biomass burning (5%), OC Aromatic (4%), and vegetative detritus (3%). The rainwater particles are grouped together and not separated by size into submicron and supermicron classes, because of the lower number of counts and the fact that the size most likely had more to do with the particle generation process than the original ambient particle size. The rainwater particle types common to both APMEX and CIFEX include the SS-EC type, OC Aromatic, Dust, and Dust (SS) classes.
 For comparison, Figure 1 also presents an overview of ambient atmospheric aerosol particle types sampled during the respective rain events and the relative fractions from both CIFEX and APMEX. Ambient particles were separated into submicron (0.2–1.0) and supermicron (1.0–3.0) aerodynamic diameters and classified using ART-2a. During CIFEX, the submicron mode (Figure 1b) was composed of biomass burning particles (35%), followed by sea salt (32%), elemental carbon (EC) (14%), and finally organic carbon (OC) (8%). The remaining particle classes represented no more than five percent of the total particles. The supermicron mode (Figure 1c) was composed of ∼3/4 sea salt particles, while the biomass burning particles represented 8%.
 The APMEX campaign yielded ambient particles chemically similar to INDOEX [Guazzotti et al., 2001], a chemically uniform submicron particle type with more diversity in the chemical composition of supermicron particles. During APMEX, a total of 61,744 submicron particles (0.2 < Da < 1.0 μm) were detected. Approximately 59% of the submicron mode particles (Figure 1e) were an elemental carbon particle type mixed with sulfate. Biomass burning was the next most common class (34%) and was characterized by a large potassium peak, minor OC and EC peaks, and often sulfates. Sea salt particles made up 35% of the 229,686 particles detected in the supermicron mode (Figure 1f). Fly ash was also common (28%), followed by biomass burning (13%), EC (8%), and SS-EC (7%). Other particle types represented five percent or less. A full description of the ambient particles during APMEX is given by Spencer et al. .
 This study represents the first direct comparison between the mixing states of ambient and rainwater particle spectra. The mass spectra of the most common particle types observed in rainwater samples (both CIFEX and APMEX) are shown in Figure 2. Figure 2a shows the negative and positive ion weight matrices for m/z = −140 to +140 Daltons for the top five most common rainwater particle types observed, SS-EC, OC-aromatic, vegetative detritus, Dust (SS), and OC (Na). The Dust (SS) class suggests either an internally mixed, insoluble dust/salt particle formed through cloud processing of dust particle during transport [Ma et al., 2004; Zhang et al., 2006], or a salt coating formed on the dust particle after being in solution. The Dust (SS) types observed in rainwater during CIFEX may be from long-range transport of Asian dust across the Pacific. During the CIFEX study, the Chemical Weather Forecast System CFORS [Uno et al., 2003] predicted the arrival of Asian dust, and this aged dust was measured to be an effective CCN despite its low soluble species content, because of its large size [Roberts et al., 2006]. Also shown is the potassium phosphate particle weight matrices, which are representative of biogenic material either bacteria or vegetative detritus [Casareto et al., 1996; Silva, 2000]. The OC-aromatic particle class is characterized by ions indicative of aromatic compounds (+77 and +91 m/z) [McLafferty, 1993; Silva, 2000] which were sometimes mixed with sulfate. A likely source of these aromatic compounds is biomass burning [Pósfai et al., 2003]. A comparison of Figures 2a and 2b illustrates the similarity between particle types observed in air at the same time the rainwater samples were collected. Figure 2a shows the weight matrix of elemental carbon (with sodium and potassium) particles in rainwater are very similar to atmospheric elemental carbon particles (Figure 2b), demonstrating the ability to measure insoluble compounds in rainwater water without change. Figures 2a and 2b show the weight matrices for matching vegetative detritus particle classes. Because the same mass spectral signatures are observed in rainwater and ambient aerosols, this suggests they are insoluble. The mixed SS-EC particles are hypothesized to represent SS and EC particles that have coagulated in clouds and acted as CCN. As the cloud dissipates, the water evaporates and leaves the mixed SS and EC residue behind. Alternatively, the cloud droplets may continue to grow and precipitate. The observation of SS-EC in rainwater samples suggests they were deposited after acting as CCN. This is a unique finding and may indicate an important atmospheric processing mechanism which can affect the rate of EC removal and hence its lifetime and total radiative impact. For example, Fuller et al.  have shown that when EC (soot) is an inclusion in a high refractive index material (such as sulfate or sodium chloride), the absorption is extremely strong. Hence these mixed SS-EC particle types are predicted to have unique optical properties and significant climate forcing potential [Fuller et al., 1999].
 Finally, ATOFMS can provide insight into other organic carbon classes in rainwater in addition to aromatic carbon species. Specifically, there is an interest in organic nitrogen deposition via wet scavenging [Cornell et al., 2003]. Organic nitrogen observed in rainwater is believed to be from precursors of water soluble organic nitrogen in the ambient aerosol particles. In Figure 2a, 26CN−, a common ATOFMS marker ion for organic-N species [Silva, 2000], can be seen as associated with the SS-EC, OC-aromatic, and Dust (SS) particles. Furthermore, 42CNO− is observed in all five of the most common rainwater particle types. Thus organic nitrogen species are commonly associated with particles in rainwater.
 A comparison of common insoluble compounds in rainwater and ambient data is shown in Figure 3. The numbers of particles in each pie chart are as follows: CIFEX rainwater 284, CIFEX ambient 2138, APMEX rainwater 52, APMEX ambient 1039. By comparing the concentration of only the common insoluble particles (as described above) in rainwater and ambient data, a relative concentration can be determined. From the relative concentrations in rainwater versus concurrent ambient sampling, chemically selective nucleation processes can be investigated. Comparing the relative fraction of insoluble particle types, more vegetative detritus particles were found in rainwater (3 to 7 times more) than ambient air. SS-EC was constant but a slightly lower fraction of dust (i.e., the sum of dust, dust (SS) and fly ash) was observed in rainwater than in ambient air. While the dust was a small fraction of total ambient particles relative to SS, the dust comprised a large fraction of the total residue particles from rainwater during CIFEX.
Figure 4 shows the ambient submicron and supermicron particle concentrations during 10 rain sample collection periods, along with particles of all sizes from associated rain samples. Deposition rates are observed to vary from one precipitation event to another [Volken and Schumann, 1993; Chate et al., 2003; Laakso et al., 2003]. Is this previously observed variation in deposition rate due to variations in ambient aerosol concentrations, or other factors like rain rate, droplet size, gas phase, etc.? To address this question, the data in this study were examined for a corresponding variation in the deposition from the single particle data. Statistical information from the samples in Figure 4 is found in Table 2. This lists the mean number fraction, standard deviation, and theoretical error for the top five classes of particles. The mean number fraction is the mean of the four samples for CIFEX, and five samples in APMEX (Samples 1, 2 shown in the first bar was excluded as it represents a distinct meteorological condition). The standard deviation represents the amount of variability, and it is the combination of natural variability and variability due to error from poor counting statistics. The observed variability in rainwater is typically twice the calculated error, indicating significant variations in the types of particles were observed during different rain events. Furthermore, greater variability was observed between sequential rainwater samples than sequential ambient samples in both studies. The variations observed in the chemistry of the rainwater samples were not correlated with fluctuations in ambient aerosol concentrations at ground level.
Table 2. Statistics of Ambient and Rainwater Samples, Mean Number Fraction, Standard Deviation, and Error
 A unique and potentially insightful finding of this study is the group of aromatic particles detected at both locations. On the basis of results from the clusters, a detailed peak search was performed to identify all particles that contained aromatic ion markers (ion peak areas > 100 for m/z +51, +63, +77, +91, and +115), which may have been classified into other clusters on the basis of other, larger, ion peaks. Interestingly, many aromatic particles were detected in rainwater but only a few particles with this combination of aromatic signature were detected in ambient sampling. During APMEX, approximately 7% (152 of 2317) of the rainwater particles contained aromatic markers. During CIFEX, aromatic-containing particle types were even more prevalent (157 of 534 total particles, ∼29%) in rainwater samples. As described below, we hypothesize these species are HULIS or water soluble aromatic compounds dissolved from the biomass particles in the rainwater samples [Zappoli et al., 1999; Graham et al., 2002; Mayol-Bracero et al., 2002].
 This initial study shows how ATOFMS can provide a rapid view of the single particles generated from rainwater which can then be compared to ambient aerosol data. Elemental carbon particles are observed in both the ambient and rainwater data. Insoluble compounds including dust, vegetative detritus, and elemental carbon particles appear to be effectively incorporated in rainwater most likely as cloud condensation nuclei or possibly scavenged by the falling raindrop.
 During CIFEX, ambient particles (0.2 < Da < 3 μm) measured concurrently with rain events are primarily fresh sea salt and biomass burning. During APMEX, ambient particles were primarily elemental carbon, biomass burning, fly ash, and sea salt. Comparing ambient aerosols measured concurrently during the rain events with particles generated from rainwater, several particle types are found to be similar, particularly SS-EC and a vegetative detritus type. This demonstrates the ability to measure individual insoluble particles in rainwater, which show similarity to atmospheric particles. Sea salt particles, while prevalent in ambient samples, were not commonly observed in rainwater samples. One possibility for their absence is they dissolved and became too dilute. Alternatively, sea salt ions (Na+ and Cl−), which were also found complexed with the top dust type (Dust-SS) as well as the other particle types, could have been dried residue on the insoluble dust cores formed by the particle generation process.
 An important finding in this study is the OC Aromatic class which contains ions indicative of HULIS [Qin and Prather, 2006], water soluble macromolecular compounds common in continental aerosols. These particles composed of aromatic compounds were present at both CIFEX and APMEX locations which further supports the important role of aromatic biomass species in cloud formation [Roberts et al., 2002; Li et al., 2003; Andreae et al., 2004]. Many aromatic-containing particles are observed in rainwater, yet similar particles were not frequently observed in ambient air. Biomass burning generates particles with both soluble and insoluble carbonaceous components [Pósfai et al., 2004] and represents a large source of carbonaceous particles in the northern Indian Ocean [Guazzotti et al., 2003; Spencer et al., 2007]. A large fraction of the biomass burning aerosol consists of water soluble organic carbon and HULIS [Mayol-Bracero et al., 2002]. The HULIS has been hypothesized to form during biomass burning from the incomplete breakdown of polymeric carbohydrates [Mayol-Bracero et al., 2002] or formed by aqueous phase processing of smaller phenol type compounds and other precursors in fog and cloud water [Zappoli et al., 1999; Graham et al., 2002; Gelencser et al., 2003]. Notably missing from the rainwater samples were biomass burning particles. The coupled observation of the absence of the aromatic types in ambient particles and the lack of biomass particles in the rainwater samples suggests the biomass particles were dissolved or transformed during the analysis procedure forming the aromatic particle types observed in the rainwater samples. From this analysis alone, we cannot determine whether the OC aromatic particle class is due to insoluble residue particles, or soluble species that dissolved from biomass and reformed individual particles in the atomization process [Wiedensohler et al., 1994]. Previous studies of water-soluble organic compounds (WSOC) have identified HULIS as representing 33 to 60% of the WSOC [Hoffer et al., 2005; Kiss et al., 2005]. Kiss et al.  reports that model calculations of droplet activation of organic aerosols are extremely sensitive to input parameters. More research on the properties of these aerosols is needed to better understand their role in cloud formation.
 The association observed between the aromatic species and sulfate may suggest that sulfate was encased in an insoluble organic shell. These organic shell and sulfate core particle types have been observed in biomass burning particles using transmission electron microscopy [Pósfai et al., 2004]. Alternatively the sulfate may have been dissolved in solution and coated the particle upon resuspension. While some sulfate was observed on several particle types, it was mainly associated with the OC aromatic class, suggesting the OC-sulfate core-shell picture may be the best explanation for this type.
 One of the challenges with this type of analysis is determining whether compounds are soluble, insoluble, or mixed. On the basis of the identification of similar particles in the atmosphere, we suggest the dust/sea salt, vegetative detritus, and elemental carbon particles types are insoluble. The aromatic compounds may not have been soluble. Because of their highly enhanced concentration in rainwater, it is more likely they are at least partially water soluble.
 Future sampling of rainwater will include a high time resolution rain gauge to allow comparisons of changes in atmospheric concentrations before and during rain events. By collecting several samples from a single rain event, more can be learned about the removal mechanism. The question remains as to whether the aerosols detected from the rainwater samples formed in the clouds or were scavenged by the rain. In situ single particle physical and chemical measurements from airborne platforms would help sort this out by allowing sampling at various altitudes. Additionally, vertical profiles may help unravel the link between the enhanced variability in rainwater samples compared to ambient. That is, the raindrops may have scavenged different particle types at distinct altitudes which may or may not be present at the same concentrations at ground level. Finally, unique aromatic compounds were observed only in the rainwater samples collected at two quite different locations, Trinidad Head, CA and the Republic of Maldives, indicating the possibility of broad spatial importance of aromatic carbon compounds acting as CCN.
 The authors thank the National Oceanic and Atmospheric Administration, grant NOAA/NA17RJ1231, and V. Ramanathan for the support of this research. C. E. Corrigan provided invaluable assistance setting up sampling at the MCOH. The manuscript benefited from the comments of two anonymous reviewers.