Coastal New Particle Formation: A Review of the Current State-Of-The-Art

Transcription

1 CSIRO PUBLISHING Review C. D. O Dowd and T. Hoffmann, Environ. Chem. 5,, doi:1.171/en577 Coastal New Particle Formation: A Review of the Current State-Of-The-Art Colin D. O Dowd A,C and Thorsten Hoffmann B A Department of Experimental Physics & Environmental Change Institute, National University of Ireland, Galway, University Road, Ireland. B Johannes Gutenberg-Universität Mainz, Institut für Anorganische und Analytische Chemie, Duesbergweg 1-1, 5518 Mainz, Germany. C Corresponding author. colin.odowd@cmas.demon.co.uk Environmental Context. Atmospheric aerosols play an important role in determining the earth s radiative budget, climate change and air quality levels. Much effort has been spent on quantifying the impact of aerosols on climate change; however, the largest gap in our knowledge relates to quantifying natural aerosol systems and the new particle formation process associated with these systems. The marine aerosol system is of particular interest due to the 7% ocean coverage of the earth s surface. Coastal new particle formation events are thought to be more frequent and of stronger intensity compared with open ocean events and thus have been studied in detail to identify possible processes leading to open ocean new particle production. Abstract. New particle formation via secondary gas-to-particle conversion processes over the oceans is one of the main mechanisms controlling the marine aerosol number population; however, despite extensive effort over the years, this phenomenon is still not well quantified. Coastal new particle formation events are more frequent than open ocean events and consequently have been studied in greater detail. This review article summarizes the recent studies into coastal new particle formation events and summarizes the linkage of these events to iodine emissions and ultimate particle formation via iodine oxide nucleation processes. The current state of knowledge may be summarized by concluding that, in general, coastal nucleation events are driven by biogenic emissions of iodine vapours that undergo rapid chemical reactions to produce condensable iodine oxides leading to nucleation and growth of new particles. The primary source of the condensable iodine vapours is thought to be molecular iodine (I ). The role of iodine oxides in open-ocean new particle production still remains an open question and is the most pressing next step to undertake. Keywords. aerosol nucleation algae iodine oxides marine aerosol organoiodine compounds Manuscript received: 7 September 5. Final version: 1 October 5. Colin O Dowd s research focuses on aerosol formation and transformation, aerosol cloud interactions and impacts of aerosols on climate change and air quality. He is based at the Department of Experimental Physics and the Environmental Change Institute at the National University of Ireland, Galway. He is joint Editor-in-Chief of the Journal of Geophysical Research Atmospheres and is Associated Editor of the International Journal of Oceans and Oceanography. He is currently Co-Chair of the International Committee on Nucleation & Atmospheric Aerosols and is Chair of the American Geophysical Union Technical Committee and Aerosols and Clouds. In he was awarded the Smoluchowski Award for achievements in atmospheric aerosol research. Thorsten Hoffmann is Professor of Analytical Chemistry at the University of Mainz. His research focuses on atmospheric chemistry of biogenic VOCs, formation of secondary organic aerosols from hydrocarbon oxidation and iodine oxides and chemical aerosol characterization of aerosol species. In particular, the research focuses on development of new analytical techniques for aerosol chemical analysis. He is Associated Editor of the Journal of Geophysical Research Atmospheres and Atmospheric Chemistry and Physics. CSIRO /5/5

2 C. D. O Dowd and T. Hoffmann Introduction Natural new particle formation via gas-to-particle conversion is an important process determining the concentration of atmospheric aerosols, and ultimately, the concentration of cloud condensation nuclei. [1] Although there have been many studies of new particle formation in pristine environments, the most studied new particle formation process has been that observed in the coastal environment. Coastal new particle formation events have been observed more frequently than other new particle events over the ocean, possibly in part because of the more frequent measurements undertaken at coastal stations compared to that over the ocean and partly because of the greater intensity and frequency of these events resulting from increased biological activity in coastal regions. A few locations have been identified as possibly important boundary layer sources of natural biogenic aerosols: the most obvious are over forests, [,3] where biogenic volatile organic compounds emissions are high, and around coastal regions during exposure of shore biota directly to the atmosphere during low-tide conditions. [,5] Coastal nucleation events might occur in most coastal environments. This is certainly true, at least, in western Europe, with frequent observations of particle production events being observed around the Scottish, [6,7] Irish, English and French coastlines. [8] Furthermore, coastal nucleation has been observed in the southern hemisphere around the Antarctic coastline [9] and in the Tasmania coastal region. [1] New Particle Formation Processes, Nucleation and Growth The first step of new particle formation is typically homogeneous nucleation of condensable vapours to form a critical cluster of the order of.5 nm in size and comprizing of the order of tens of molecules. [11] Clusters are forming continuously in the atmosphere; however, if they do not reach the critical cluster size, they will evaporate, whereas if they do form at the critical size, they will continue to grow via condensation processes. Under atmospheric conditions, it has long been considered that sulphuric acid and water are the two most likely species to drive nucleation but the conditions for such binary nucleation to occur are limited to low temperatures and high concentrations of sulphuric acid typically higher than that encountered in the lower troposphere or boundary layer. The most recent studies show that the addition of ammonia to this system (ternary nucleation) is likely to result in nucleation occurring under most atmospheric conditions and is likely to be ubiquitous in space and time. [11] Although significant advances have been achieved in understanding the ternary nucleation of H SO -H O-NH 3, it is clear from atmospheric measurements of aerosol properties that new particle formation does not necessarily result from the occurrence of nucleation. This has led us to the decoupling of nucleation from particle formation. Particle formation is somewhat operationally defined since we cannot measure nucleated clusters and the smallest particle size Time (h) Mar-1998 (68) to 31-May-1998 (151), CN concentration Day of year Fig. 1. Hour-average particle concentration at Mace Head (colour contour) plotted as a function of Day ofyear and Time of Day. The + marks the occurrence of low tide at the station. Reproduced/modified by permission of American Geophysical Union, O Dowd et al., J. Geophys. Res., 17, doi:1.19/1jd555. Copyright [] American Geophysical Union. that can be measured is of the order of 3 nm. Although models predict the relative ease with which ternary nucleation can occur, they cannot predict subsequent growth of the nucleated clusters to aerosol sizes of 3 nm, since there is apparently sufficient sulphuric acid availability to nucleation but insufficient availability to grow the clusters to 3 nm particle sizes. This leaves two options: either the clusters are scavenged by the pre-existing aerosol population through coagulation; or an additional condensable vapour is required to contribute to the growth after nucleation has occurred. For example, this can occur through the activation of sulphate clusters in a super-saturated organic vapour field in the same way that water clouds are produced by the activation of aerosol nuclei in a super-saturated water vapour field. Of course, if there is a sufficient source of another condensable vapour that can both nucleate and provide sufficient additional condensable vapours for continued growth into aerosol sizes, then the nucleation and particle formation system would be totally coupled. Physical Characteristics of Coastal New Particle Events The most studied coastal nucleation events have been at the Mace Head Atmospheric Research Station in the North-east Atlantic, and the most comprehensive study was under the PARFORCE (New Particle Formation and Fate in the Coastal Environment) project, [5] which assembled a large team of European researchers into a specifically focused research project aimed at elucidating the coastal nucleation process. At Mace Head, it was observed that when air had advected over exposed tidal areas under conditions of solar radiation, massive aerosol production events were observed with the produced aerosol residing in the sub-1 nm size range

3 Coastal New Particle Formation: A Review of the Current State-Of-The-Art 1 Particles Tide amplitude JO( 1 D) Sulphuric acid e 5 6 1e 6 Sulphuric acid (1 6 mol cm 3 ) 8 6 Particle concentration (cm 3 ) 8e 5 6e 5 e 5 e 5 e 5 1e 5 5e 6 JO( 1 D) s Tide amplitude (m) e e Julian day (1999) Fig.. Total particle concentration, sulphuric acid concentration, JO( 1 D) and tide height during a typical new particle production event at Mace Head. Reproduced/modified by permission of American Geophysical Union, O Dowd et al., J. Geophys. Res., 17, doi:1.19/1jd555. Copyright [] American Geophysical Union. dn/d log Dp (1/cm 3 ) Particle diameter (m) Type I, 1999 Type I, 1998 Type II Type III Maritime background Fig. 3. Size distributions during different coastal nucleation event types. Reproduced/modified by permission of American Geophysical Union, O Dowd et al., J. Geophys. Res., 17, doi:1.19/ JD6. Copyright [] American Geophysical Union. Figure 1 illustrates the mean hourly average particle concentration at sizes larger than 3 nm for a 3-month period. Particle concentration is significantly increased almost on a daily basis, reaching concentrations in excess of cm 3. This aerosol increases coincide with the low-tide mark during daylight hours. Figure expands on a typical case encountered during the PARFORCE campaign. The minimum in the tidal height (low tide) occurred around 8: hours and prior to low tide, the background aerosol concentration was typical of clean marine levels at 5 cm 3. When low tide occurs, the particle concentration, over a period of a couple of hours, increases to in excess of 1 6 cm 3 and then slowly decreases back to the background levels as the tide height increases again. Also shown on the figure is the concentration of sulphuric acid and J(O( 1 D)) as a measure of photochemical activity. What is clear is that the peak in particle concentration occurs a number of hours before the peak in sulphuric acid concentration suggesting that sulphuric acid is not driving the process. The maximum sulphuric acid concentration observed was of the order of molecules cm 3, whereas during the peak aerosol production period, the acid concentration was less than 1 6 molecules cm 3. These results suggest that sulphuric acid is not driving the nucleation and growth process; however, it appears that the increased condensation sink associated with the increased surface area of the new particles seems to lead to a local depletion of the acid vapour, which would indicate that sulphuric acid may contribute to the particle growth after the fact. Size distributions observed during the events clearly indicate that the increased concentrations are the result of a nucleation mode typically at sizes smaller than 1 nm. Figure 3 illustrates the observed size distributions taken for different classification of events. [1] Type I events are those where the air has advected over a source region 1 m upwind of the station; type II events are those where air has advected over multiple source regions between.1 and 1 km; and type III events are typically continentally-influenced air that has advected over tidal regions about 3 5 km from the station. Also shown is the typical bi-modal (Aitken and Accumulation mode) background marine distributions. The peak in the nucleation mode occurs between 3 and 5 nm for type I events and between 5 and 1 nm for type II and III events, resulting from the longer growth times associated with type II and III conditions. Dal Maso (), [13] using a novel analytical tool (using the measured aerosol micro-physical 7

4 C. D. O Dowd and T. Hoffmann GF at 9% RH Type II Fig.. Hygroscopic growth factors during nucleation events. Reproduced/modified by permission of American Geophysical Union, Väkevä, et al., Hygroscopic properties of nucleation mode and Aitken mode particles during nucleation bursts and in background air on the west coast of Ireland, J. Geophys. Res., 17, doi:1.19/ JD76. Copyright [] American Geophysical Union. changes to determine the required nucleation rate, condensable vapour concentration and condensable vapour source rate), calculated growth rates for these events and found the nucleation mode growth rate to be of the order of nm hour 1 and that estimated concentration of 1 nm particles was of the order of cm 3 with a source rate of 3 1 cm 3 s 1. Chemical Characteristics of Coastal Aerosols Quantification of the chemical characteristics of coastal nucleation mode aerosols provides a significant challenge to current experimental methods and analytical techniques owing to the difficulties in separating the nucleation mode from the background aerosol and analyzing such small amounts of aerosol mass. Nevertheless, two techniques were applied during PARFORCE: (1) separation of nucleation mode particles and evaluating their hygroscopic growth response to increasing humidity fields; [1] and () electrostatic-precipitation of the nucleation mode to TEM stubs followed by subsequent X-ray dispersive analysis. [15] The hygroscopic growth factor is an indirect method to estimate the chemical composition of aerosols since standard aerosol species such as ammonium sulphate and sea-salt have well-known growth factors. For sub-1 nm particles, the growth factor resulting from an increase in relative humidity from to 9% is greater than 1. (see Väkevä et al. [1] for exact figures as a function of particle size and composition). Figure shows the growth factor (GF) for the nucleation event displayed in Fig.. It is seen that at the start of the event, the GF is of the order of 1.1, which indicates a very low solubility in water. Such low growth factors rule out sulphate salts and sea-salt and point to a nucleating species with a low affinity to water. What is interesting to note is that the GF increases around the same time there is an apparent reduction in sulphuric acid concentration, as discussed in Fig.. This provides additional evidence that sulphuric acid does not appear to be driving the nucleation to aerosol-formation step of the process, but can contribute to the further growth thereafter, as evidenced by the coincident increase in GF to 1.3. Mäkelä et al. () [15] separated 6 8 nm particles onto a TEM stub for subsequent X-ray dispersive analysis and found that the particles contained iodine in all samples and sulphur in most. In the same study, they measured the biogenic halocarbon emissions from various types of seaweeds encountered in the coastal region and they found strong emissions of CH I from Fucus vesiculosus in particular. CH I [16] and IO [17] were also observed at Mace Head and these species had a peak concentration at low tide and minimum concentration at high tide. The tidal cycle of the iodo-vapours, aerosol formation events, the direct emission measurements from algae cuvette experiments, combined with the finger printing of iodine in the nucleation mode aerosols tentatively provided a link between the emission and subsequent photo-oxidation of CH I and formation of coastal particles through a mechanism involving iodine. Estimates of Precursor Vapour Concentration, Source Rates and Particle Formation Rates Without knowledge of the main vapour responsible for nucleation and growth and without the ability to measure nucleation rates, two theoretical approaches were taken to elucidate these issues. The first was an evaluation of the nucleation potential of sulphuric acid, based on observed acid concentrations and using an improved ternary nucleation model. [18] Kulmala et al. [18] found that for sulphuric acid concentrations of cm 3 (i.e. the typical concentrations occurring during production events) and ammonia concentrations of 5 5 ppt ( cm 3 ), the nucleation rate was predicted to be of the order of >1 8 cm 3 s 1. Although the limitation of classical nucleation modes must be recognized (i.e. they can sometimes produce unrealistically high nucleation rates of >1 1 cm 3 s 1 ), this study clearly suggested that sulphuric acid nucleation was highly likely to be occurring, particular at elevated sulphuric acid concentrations. However, although it could be argued for some events that sulphuric acid nucleation was playing a role, there were many events occurring during periods of very low sulphuric acid production (i.e. dawn and dusk). The main result is that even if sulphuric acid is not driving the particle production, it most likely can, for a large majority of the event cases, provide sufficient thermodynamically stable sulphate clusters for additional vapour to condense on. The second approach involved utilizing the AEROFOR aerosol nucleation and dynamics model to reproduce the observed aerosol size distributions during the nucleation events with the objective of given an estimate of the vapour precursor concentration and source rate required to explain the observations. [19,] O Dowd et al. [19] first concluded that sulphuric acid could not account for the observations and invoked the necessity of an additional fourth species X. Pirjola et al. [] further developed this approach using an additional generic condensable vapour in an attempt to explain the observations. For the strongest events, the criteria that the model simulations [] had to meet was that after 5 s from the source region, the total concentration at sizes larger than 3 nm had to be 1 cm 3 and the concentration of particles larger than nm had to be within the range 9 1 cm 3. During the simulations, the model was initiated with nucleation parameterization based 8

5 Coastal New Particle Formation: A Review of the Current State-Of-The-Art Number concentration (cm 3 ) Number: CPC 35 ( 3 nm) CPC 31 ( 1 nm) SEMS ( 5 nm) Volume and mass: SEMS volume ( 5 nm) AMS mass ( nm Aerod.) 8 6 SEMS volume concentration ( m 3 cm 3 ) 3 1 AMS aerosol mass (uncalibrated) Time after inception of photolysis (min) Fig. 5. Time evolution of aerosol number and mass in the Caltech smog chamber following photolysis of CH I in the presence of ozone. Reproduced/modified by permission of American Geophysical Union, J. L. Jimenez, R. Bahreini, D. R. Cocker III, H. Zhuang, V. Varutbangkul, R. C. Flagan, J. H. Seinfeld, C. D. O Dowd and T. Hoffmann, J. Geophys. Res. 3, 18, doi:1.19/jd5. Copyright [] American Geophysical Union. on ternary sulphuric acid nucleation and an additional generic biogenic condensable vapour source rate was added. For the range of conditions, it was found that a nucleation rate of cm 3 s 1 was required and that the condensable vapour source rate needed to be of the order of cm 3 s 1, leading to a condensable vapour concentration of the order of cm 3. The conclusions from these modeling studies were that sulphuric acid concentrations could explain the nucleation of sufficient stable seed clusters to produce the number concentration associated with such events; however, there was typically three orders of magnitude too low sulphuric acid availability to explain the growth of these clusters to detectable particle sizes (typically of 3 nm). Ruling out sulphuric acid as the driver, and given that organic vapour production is also significantly lower than the required production rates, the focus narrowed on investigation of iodine emissions as the possible explanation of the phenomenon. Laboratory Studies As mentioned above, algal emission studies showed that CH I was the most abundant single iodine-containing compound released by macroalgae in the region. Further, gas-phase ambient air measurements of CH I indicated a tidal cycle with peak concentrations of the order of 1 ppt. [16] In addition to a tidal signal being observed for CH I,as mentioned above, a known oxidation product, iodine monoxide (IO) was also observed at Mace Head, possessing also a tidal cycle with the peak concentrations occurring during low tide. Combining the available knowledge that iodocarbons are readily emitted from macroalgae, that a tidal cycle is observed in both CH I and IO and that iodine was identified in all nucleation mode particles analysed by TEM analysis, that the hygroscopic growth factor indicated an insoluble substance and, finally, that sulphuric acid concentrations were insufficient to explain the events, the iodine system was highlighted for examination in the context of particle production. Hoffmann et al. (1) [1] conducted laboratory experiments on the aerosol formation potential of CH I using CH I concentrations of the order of 17 ppb. They found the occurrence of rapid particle production in the presence of O 3 and mass spectrometry analysis of the produced aerosol indicated iodine oxide fragments in the form of [IO], [IO ] and [IO 3 ]. Hoffmann et al. suggested that CH I was rapidly photolysed and that a series of reactions with ozone and IO self reaction led to the production of OIO, which, in turn, self reacted to from the condensed particulate phase. They proposed the following reactions: CH I + hν I + CH I I + O 3 IO + O IO + IO OIO + I The self reaction of OIO to produced polymers was suggested as the nucleation mechanism following: OIO + OIO I O (or [IO] + [IO 3 ] ) I O + noio [ I O IO ] 1+n/ This polymeric structure has a characteristic low solubility in many solvents including water, thus is corroborated by the growth-factor measurements also indicating low solubility. The laboratory studies of Hoffmann et al. (1) [1] were repeated in a more comprehensive manner by Jimenez et al. (3) [] at the Caltech Smog Chamber facilities. The Caltech studies were conducted over a broader range of CH I concentrations including down to near atmosphere levels of 15 ppt. They confirmed the particle production system involving CH I and that a combination of solar radiation and ozone was required for particle production to proceed. An example of one of the chamber experiments is shown in Fig. 5, where total number concentration, measured by condensation particle counters (CPC) at sizes larger than 3 and 1 nm are shown 9

6 C. D. O Dowd and T. Hoffmann I 6 ( Hz) h HOI Aerosol OH OH HO O CH I h 3 IO I h IO OIO IO IO h OH h IO I I h I I O I I O Ion signal (Hz) 5 HI IO 3 HIO IO HIO HIO 3 I 1 IO 5 I O I O 3 Photolysis Reaction Aerosol uptake MoO MoO 15 m/z (amu) 5 3 Fig. 6. Aerosol mass spectra following the formation of aerosol from the photolysis and oxidation products of CH I. Reproduced/modified by permission of American Geophysical Union, J. L. Jimenez, R. Bahreini, D. R. Cocker III, H. Zhuang, V. Varutbangkul, R. C. Flagan, J. H. Seinfeld, C. D. O Dowd and T. Hoffmann, J. Geophys. Res. 3, 18, doi:1.19/jd5. Copyright [] American Geophysical Union. along with particle concentration larger than 5 nm measured by a scanning electrical mobility sizer (SEMS). Also shown is total volume from the SEMS system and total aerosol mass larger than nm derived from an aerosol mass spectrometer (AMS). After the inception of photolysis, a rapid increase in concentration larger than 3 nm is seen, and progressively later, similar increases are seen in the 1 nm and 5 nm sizes. After the initial burst, reaching a concentration of almost cm 3, the concentration rapidly falls again (due to coagulation) and aerosol mass continues to rise for a number of hours. Hygroscopic growth studies revealed a similar picture to that seen at Mace Head [1] in that during the dry chamber experiment, no growth was seen; however, during the wet humidified experiment, reverse growth down to GF =.65, indicative of a collapse of the particles, was observed followed by a recovery of the GF from.65 back up to 1. The aerosol composition was also measured using the Aerodyne AMS and the resultant fragments are illustrated in Fig. 6. The main fractions were indicative of iodine oxides and/or oxy-acids. These results confirmed more conclusively than the previous studies that iodine oxides, derived from CH I, result in particle formation. Additional analysis of the aerosol microphysics concluded that under dry conditions, the particles formed were fractal agglomerates with a mass fractal dimension of D f = 1.8.5, whereas under the humidified experiments, D f was.7 indicating a more compact structure. Similar studies were performed by Burkholder et al. () [3] corroborating the main features seen in the Caltech experiments. Theoretical Predictions of Iodine Oxide Gas-Phase Processes and Nucleation From the Caltech experiments, Jimenez et al. proposed a more complex production system following on from Hoffmann Fig. 7. Chemical reaction scheme for the formation of condensable iodine vapours from CH I in the presence of ozone. Reproduced/modified by permission of American Geophysical Union, J. L. Jimenez, R. Bahreini, D. R. Cocker III, H. Zhuang, V. Varutbangkul, R. C. Flagan, J. H. Seinfeld, C. D. O Dowd and T. Hoffmann, J. Geophys. Res. 3, 18, doi:1.19/jd5. Copyright [] American Geophysical Union. et al. and using the current-knowledge from the gas-phase community relating to the gas-phase iodine cycle. The schematic diagram of the pathways leading to aerosol production from photolysis of CH I in the presence of ozone is shown in Fig. 7. The system involved rapid conversion between iodine reservoirs consisting of I, I O,I O, OIO, IO, and HOI and involve various photolysis reactions, thermal decomposition, reactions with OH and HO, reactions with IO and the self reaction of OI OI and OIO OIO. The modelling studies of Jimenez et al. did not include actual OIO nucleation so to speak; however, to replicate the observed measurements, it was concluded that the accommodation coefficient for iodine vapours to the aerosol phase had to be 1 or close to 1. Bases on the timescales for growth and the ability of the modeling studies to replicate the observations, Jimenez et al. suggested that although it was clear that CH I could produce significant amounts of particles, an additional source of condensable vapours other than those derived from CH I might be required to explain the observations in the atmosphere. Burkholder et al. () [3] were the first to attempt to completely model the iodine system from photolysis, to iodine oxide nucleation and on to aerosol growth. In their studies, they used a kinetic nucleation approach coupled to an aerosol dynamics model. The treatment of OIO nucleation involved use of available thermodynamic data and included both nucleation and evaporation of OIO clusters to achieve consistency with the experimental results. Using their aerosol formation model, Burkholder et al. evaluated the possibility of OIO particle formation in the coastal zone and the open ocean through comparisons of measurements of both IO and OIO concentrations in oceanic regions. The IO measurements were undertaken at Mace Head and Tenerife where peak measured values ranged from to 5 ppt, [17,] whereas the OIO measurements taken by Allen et al. (1) were the order of.5 ppt. The simulations concluded that steady state concentrations of 1 ppt OIO would lead to a particle concentration of 1 6 cm 3 after 1 minutes, while concentrations of the order of 5 1 ppt IO would be required to produce 5

7 Coastal New Particle Formation: A Review of the Current State-Of-The-Art 1e 9 Nano-SMPS BIOFLUX chamber study: Fucus e 6 From I From CH I Diameter ( m) e 3 1e 1: : 1: : 1: Fig. 8. Measured I flux compared to calculated CH I flux during new particle production events at Mace Head. Taken from McFiggans et al., Atmos. Chem. Phys.,, Copyright [] European Geophysical Union. significant particles. These concentrations are significantly higher than the reported observations of both IO and OIO. It should, however, be noted that IO and OIO measurements at Mace Head were conducted using long-path differential optical absorption spectroscopy (DOAS) measurements and thus, spanned a horizontal distance of 15 km. Given that the tidal regions are only a fraction of a kilometre, both IO and OIO concentrations could easily reach an order of magnitude higher that reported by the DOAS measurements and they concluded that iodine oxide particle production could be feasible in local hot spots where elevated concentrations of CH I could drive the system. Closing the Loop on Iodine Oxide Particle Formation All the above studies point to iodine being involved in the coastal new particle production process and, in particular, as the main ingredient over and above any other species, including sulphuric acid or condensable organics. However, none of the studies have been able to fully underpin the processes with CH I as the primary precursor. The lack of closure had implications, not only for the aerosol formation processes, but also for the ozone depletion process, which is regarded as significant in terms of gas-phase chemistry. McFiggans et al. () [5] conducted chamber experiments on the aerosol formation potential of macroalgae at Mace Head and found that the resulting particles had a similar mass spectra (Fig. 1) to that observed to those produced in the Caltech chamber by CH I and also to that produced in the lab by I. They and Saiz-Lopez et al. (, 5) [6,7] also reported very high concentrations of I of the order of ppt during the day time and increasing to >8 ppt at night. OIO concentrations were also measured and reported at 3 ppt levels. Given the lifetime of OIO, concentrations of 3 ppt suggest a very large source rate of OIO, which could be accounted for by the elevated I concentrations. McFiggans et al. [5] also reported the measured molecular iodine photolysis rate at Mace Head and found that the iodine flux from molecular iodine was three orders of magnitude larger that the flux from CH I (see Fig. 8). Thus, it seems that molecular iodine Particle concentration (cm 3 ) Number of day Fig. 9. Nucleation pulses during chamber experiments for naturally occurring seaweed in the vicinity of Mace Head. could be the missing additional source of condensable iodine vapours necessary to explain coastal new particle production. The most recent studies on coastal new particle formation events again occurred at Mace Head during the QUEST-BIOFLUX and the NAMBLEX experiments. During BIOFLUX, the main aim was to further explore the potential relationship between molecular iodine and new particle formation. BIOFLUX focused on Mace Head and regions around Mace Head, which were regarded to be biological hot spots and where the algae biomass was significantly greater than at the Mace Head station. Further, in these nearby hot spots, strong nucleation events were observed from aircraft while simultaneously only very week or non-existent events were seen at Mace Head (C. D. O Dowd, Y. J. Yoon, W. Junkerman, P. P. Aalto, H. Lihavainen, unpublished data). BIOFLUX also focused on chamber experiments of new particle formation events using realistic biomass loadings. The experimental set up and main findings during BIOFLUX are reported in detail in Sellegri et al. (5). [8] The chamber experiments were carried out in a m 3 Perspex reactor with 5% UV transmission. Ambient air was scrubbed of background aerosol and was allowed to flow through the chamber and the natural solar radiation was used for photolysis. Background levels of ozone were of the order of ppb. During the chamber experiments, massive nucleation and growth bursts were observed with peak particle concentrations exceeding 1 7 cm 3. A typical series of nucleation pulses is shown in Fig. 9. The chamber experiments were modified (by increasing the flow-through of air) to produce pseudo-steady-state aerosol fields in order to remove the evolution of the size distribution as seen in Fig. 9. For the steady-state chamber experiments, particle concentration and molecular iodine concentration measurements were made for a range of seaweed mass loadings ranging from.5 to 7 kg m. A linear correlation was found between molecule iodine concentrations and biomass and between particle concentration and molecular iodine concentrations. 51

8 C. D. O Dowd and T. Hoffmann (a) Particle concentration (# cm 3 ) nm particles Total number particles 1 R.975 a 79 8e b I (ppt) dn/d log Dp (cm 3 ) I flux (cm 3 s 1 ) in legend 1.5e 9 1e 9.5e 9 (b) Particle concentration (# cm 3 ), I 1 (ppt) nm particle concentration Total number particle concentration I Seaweed mass (kg) Fig. 1. Relationship between seaweed mass, molecular iodine and particle concentration for the chamber experiments. Regression lines on upper plot refer to (a) relationship between total particle concentration and I and (b) relationship between 3 3. nm particle concentration and I concentration. These results provided the first direct correlation between new particle production and molecular iodine emissions. The relationship between iodine, seaweed mass and particle concentration is shown in Fig. 1. Seaweed mass loadings of 5kgm would be considered typical of regions around Mace Head. Molecular iodine concentrations ranged from 1 to > ppt for the experiments, whereas particle concentration ranged from cm 3 to 1 7 cm 3. From the chamber experiments, the flux of particles in the size. nm was of the order of m s 1 for a seaweed loading of.5 kg m. For these conditions, aerosol growth rates were calculated to be 1. nm min 1. The flux values derived from the chamber experiments are in excellent agreement from direct measurements of new particle flux using eddycovariance techniques. [9] The direct measurements report fluxes of the order of m s 1. The chamber reactions were simulated with an updated version of the AEROFOR nucleation and dynamics model. [3] The model was updated to include the current state-of-the art in iodine gas-phase chemistry and included dimer nucleation of iodine oxides, ternary nucleation of stable sulphate clusters and fractal coagulation. The model was able to predict well the chamber experiments using known reaction schemes and confirmed that I emissions were Diameter Dp (m) Fig. 11. Modelled (solid lines) and observed aerosol size distributions (dashed lines) for the chamber experiments. I flux in cm 3 s 1 is given in the legend. After Pirjola et al. (5). the dominant driver of these nucleation events. The model used a derived range of I fluxes from cm 3 s 1 to cm 3 s 1. The calculations showed that for the lower I flux, I reached a steady state concentration of 1 9 cm 3 and for the higher flux, a concentration of cm 3, which were in very close agreement with the observations. The corresponding IO and OIO concentrations ranged from cm 3 to cm 3 and cm 3 to cm 3 respectively. The modelled growth rate was in the range of nm min 1, again in agreement with the observations. Using condensable iodine vapours as the only condensable species, the model was able to predict most the main features of the chamber experiments; however, it slightly over-predicted the particle concentration in the size range of 3 1 nm and under-predicted the concentration in the size 3 nm. In an attempt to better predict the observations, a condensable organic vapour source rate ranging from cm 3 to cm 3 was included and better agreement was observed. The results for these simulations are shown in Fig. 11. Molecular iodine concentrations were also measured at Mace Head and the hot spots nearby. [8] Typical concentrations were of the order of 3 ppt at Mace Head and 11 ppt at the hot spots. Peak concentrations in the chamber reached 8 ppt. Saiz-Lopez et al. (5) [7] also measured I concentrations and found a peak concentration at Mace Head of 9 ppt, whereas IO and OIO concentrations were of the order of 7 ppt of 1.5 ppt respectively. Saiz-Lopez et al. [7] also used an aerosol and chemistry box model to simulate the Mace Head events and they also found good agreement with the observed aerosol measurements and the authors confirm that I is the major precursor of the observed particle bursts at Mace Head. The model also showed that that the iodide oxide nanoparticles can grow to make an important contribution, at least on the regional scale, to CCN. 5

9 Coastal New Particle Formation: A Review of the Current State-Of-The-Art Vapour concentration (cm 3 ).e 7 1.5e 7 1.e 7 5.e 6 Particle concentration (cm 3 ) With CIVs Q cm 3 s 1 Q cm 3 s 1 Q cm 3 s 1 Q cm 3 s 1 Without CIVs N 3 (Q cm 3 s 1 ) N 3 (Q cm 3 s 1 ) N 3 (Q cm 3 s 1 ) N 3 (Q cm 3 s 1 ) N 3 (Q 1 3 cm 3 s 1 ) N Total ( 1 nm) Q cm 3 s Time of day (h) Fig. 1. Simulations of condensable iodine vapour concentration and resulting new particle production for open ocean conditions. Q is the source rate of condensable iodine vapours. N tot is the total number of particles and clusters larger than 1 nm, N 3 is particle concentration larger than 3 nm. Taken from O Dowd et al., Marine particle formation from biogenic iodine emissions, Nature, 17, So far, the measurements have focused on coastal events and chamber experiments. With respect to new particle formation over the open ocean, there has only been one modelling study in terms of iodine oxide contributions to particle formation. [31] In their study, they modelled ternary nucleation of sulphuric acid, ammonia and water, based on the DMS cycle, and concluded that this system on its own could not produce new particles even though nucleation was readily predicted. They then examined the effect of additional condensable iodine vapours and found that the condensable iodine vapours could carry the sulphate clusters over a coagulation loss barrier and that estimated open ocean iodine vapours sources could increase the survival probability of freshly nucleated clusters by more than 1 times, thus making these more likely to eventually contribute for cloud condensation nuclei formation. In this situation, without condensable iodine vapours, the newly formed clusters will coagulate to the pre-existing aerosol size distribution and thus will be removed from the system. As the cluster size increases, its coagulation loss rate decreases as 1/d and thus has a higher probability of survival. Without the condensable iodine vapours, none of the clusters will grow to detectable sizes of 3 nm before they are captured by coagulation, whereas with the additional condensable vapours, a significant fraction grow rapidly enough to overcome this coagulation loss and reach pseudo-stable aerosol sizes. The simulations of particle production over the open ocean are shown in Fig. 1. The issue of iodine oxide and new particle formation over the ocean, and possible processes (a) (b) leading to such production, has been the subject of debate recently [3,33] and clearly, further work is required to elucidate the role of iodine oxides in producing new particles over the open ocean. Aspects of this debate are highlighted in the following section outlining future research needs. Future research needs Although numerous studies about the release of reactive organic iodocarbons (e.g. CH I,CH ClI, CH BrI, CHBr I) from marine biota into the atmosphere have been performed (see e.g. Laturnus 1; [3] Carpenter 3 [35] ), many open questions remain. One is connected with the release of iodocarbons from phytoplankton. Their oceanwide distribution could make a much larger contribution to ocean atmosphere fluxes than macroalgal species. Up to now, research on the open ocean iodocarbon source has focused on laboratory studies with a few selected phytoplankton species. [36] Likewise, most of the work concentrated on active microalgal biomass with only a few isolated studies considering alternative sources such as bacteria [37,38] or photochemistry. Certainly, more knowledge is needed here to enable better predictions how much organoiodine is emitted to the atmosphere. In fact, vertical concentration profiles measurements of chlorophyll A and selected iodocarbons (e.g. CH I,CH ClI) in open seawater indicate the formation of these compounds by phytoplanktonic processes, [39] although rapid photolysis of CH I in the surface water has to be taken into account. [] However, the magnitude of the sea air flux of these compounds is not quantified yet. Other investigations of mixed marine microbial communities indicated that their productivity in respect to organoiodine production is much higher than that of algae monocultures typically investigated in previous studies. [1] Both observations suggest potentially significant sources of reactive organic iodocarbons also in the open ocean. Much less is known about the release mechanism of other volatile iodine compounds, such as elemental iodine (I ). Whether the pathway of elemental iodine release into the atmosphere is exclusively and directly connected with biological processes (i.e. linked to the occurrence of marine biota) or also represents an abiotic process at the air sea interface is currently not known. McFiggans et al. [5] suggested that the coastal appearance of I is related to air-exposed macroalgae that suffer oxidative stress and releases iodine by a haloperoxidase-mediated oxidation: I + H O HOI + OH HOI + I + H + I +H O Obviously, this release mechanism would be restricted to the costal atmosphere. Another biogenic formation pathway for molecular iodine which might also play a role in the open ocean are iodide-oxidizing bacteria, as observed by Amachi et al. in iodide-enriched seawater samples. [] Beside biological processes, there are other possibilities for the release of iodine from the sea surface. One might be the direct photochemical oxidation of iodide at the sea air interface 53

10 C. D. O Dowd and T. Hoffmann suggested by Miyake and Tsunogai already in 1963: [3] I +.5 O + H O + hν I +OH Although these kind of iodine volatilization reactions have been intensively investigated for the evaluation of potential nuclear reactor accidents, [] the relevance and chemical mechanism of such a photochemically induced iodine release from sea surface iodide oxidation under ambient conditions is unclear. Possibly, radical initiated liquid phase reactions of I, based on mechanisms similar to the Cl formation from aqueous NaCl particles, [5,6] have to be taken into account. Another potential source of elemental iodine is the oxidation of seawater iodide by ozone followed by the release of I into the gas phase: [7] I + O 3 + H + I +O + H O Evidently this iodine release mechanism is directly connected to the dry deposition of ozone to the sea surface. However, since most of the studies mentioned above were laboratory investigations, the extrapolation of their results to the ocean surface is problematic without further fieldwork. Likewise, even if seawater iodide is oxidized to molecular iodine, the rapid reaction of I with organic matter in seawater requires that the production must take place in the air water interface for volatilization of I to occur. Nevertheless, since molecular dynamics simulations for the series of halide ions have shown that there is a trend to increasing availability at the interface for larger, more polarizable ions such as iodide, the concentration of iodide ions are enhanced at the air water interface relative to the bulk. [8,9] Thus, iodide may be preferentially located at the interface and the I volatilization therefore preferred relative to its liquid phase reactions. Obviously, these processes might occur not only in the open ocean surface layer, but also at the surface of sea-salt particles, [5] although the iodine enrichment in atmospheric aerosol represents evidence against the latter assumption. It remains to be investigated what impact this has on the chemistry and photochemistry of iodine species. Not only the source strength and the release mechanism of reactive iodine into the atmosphere are not sufficiently known, but also the nature of the chemical species finally nucleating. As mentioned above, the first suggestions were based on the OIO self reaction. However, several other iodine oxides and oxy-iodine acids might be involved. Although several studies focused on the investigation of new particle formation from reactive iodine precursors, [,3,51] the detailed mechanism leading to the formation of iodine oxide nanoparticles is still an open question. To change this situation both is needed, a better understanding of the gas phase chemistry of IO and OIO leading to higher oxidized or larger molecules and the knowledge of the composition of the newly formed particle phase. Beside the understanding of the underlying chemistry, one of the most important aspects about particle formation processes in the coastal or marine atmosphere is the question about their influence on the CCN number concentration. Up to now, most of the studies have been performed in Mace Head, Ireland. Therefore, measurements in other coastal environments are needed to estimate how widespread coastal particle sources are. At the same time, more observational evidences for the connection between particle formation and iodine-containing aerosol precursor concentrations at these sites are necessary. Finally, both natural coastal contributions to CCN and open ocean iodine sources that contribute to particle growth to CCN sizes have to be quantitatively understood. The large enrichment of iodine in the fine aerosol fraction in comparison to the iodine fraction in seawater is long been known. [5] Unfortunately, most of the existing measurements of iodine compounds in marine aerosol are based on bulk measurement of total iodine, i.e. the individual chemical iodine species that contribute to particulate iodine and that might give insight into the aerosol formation pathways have rarely been measured (e.g. differentiation between I,IO 3, I x O y ). One reason for this is the fact that the measurement of individual iodine species is still an analytical challenge, especially at the relatively low natural concentration levels. Recently, Baker measured inorganic soluble iodine species (i.e. I,IO 3 ) in size-fractionated aerosol samples collected from the tropical Atlantic Ocean. [53] Surprisingly, he found high concentrations if iodide and often relatively low concentrations of iodate, which was believed to be the opposite due to iodide recycling reactions (I + HOX + H + IX + H O) and accumulation of iodate in aerosol particles as it ages. [5,55] Consequently, these results raise serious questions about the current understanding of aerosol iodine chemistry. Clearly, more measurements of the individual iodine species in submicron particles in the marine boundary layer are required, ideally also including the measurement of higher iodine oxides. This kind of information might also be useful for the interpretation of chamber-based laboratory studies, where in the past the focus laid on the investigation of the gas-phase composition. Possibly, the development and application of new analytical techniques, such the rapidly developing field of aerosol mass spectrometry, will also give new insight into the chemical processes leading to biogenic particles in the marine atmosphere. Acknowledgements The authors acknowledge for Science and Engineering Technology (IRCSET) and European Commission under contract EVK-CT-1-17 (QUEST). References [1] R. J. Charlson, J. E. Lovelock, M. O. Andreae, S. G. Warren, Nature 1987, 36, 655. doi:1.138/36655a [] M. Kulmala, K. Hämeri, P. P. Aalto, J. M. Mäkelä, L. Pirjola, E. D. Nilsson, G. Buzorius, Ü. Rannik, M. Dalmaso, W. Seidl, T. Hoffmann, R. Janson, H.-C. Hansson, Y. Viisanen, A. Laaksonen, C. D. O Dowd, Tellus B 1, 53, 3. doi:1.13/j x [3] C. D. P. O Dowd, P. P. Aalto, K. Hämeri, M. Kulmala, T. Hoffmann, Nature, 16, 97. doi:1.138/1697a [] C. D. O Dowd, M. Geever, M. K. Hill, S. G. Jennings, M. H. Smith, Geophys. Res. Lett. 1998, 5, doi:1.19/ 98GL15 5

11 Coastal New Particle Formation: A Review of the Current State-Of-The-Art [5] C. D. O Dowd, K. Hämeri, J. M. Mäkelä, L. Pirjola, M. Kulmala, S. G. Jennings, H. Berresheim, H.-C. Hansson, G. de Leeuw, G. J. Kunz, A. G. Allen, C. N. Hewitt, A. Jackson, Y. Viisanen, T. Hoffmann, J. Geophys. Res., 17, 818. doi:1.19/ 1JD555 [6] J. A. Aitken, Trans. R. Soc. Edinburgh 1897, XXXIX, 5. [7] C. D. O Dowd, M. H. Smith, J. A. Lowe, B. M. Davison, C. N. Hewitt, R. M. Harrison, in Proceedings of the 1th International Conference on Nucleation and Atmospheric Aerosols (Eds M. Kulmala, P. E. Wagner) 1996, pp [8] J. Y. Paugam, J. Rech. Atmos. 1975, 9, 67. [9] C. D. O Dowd, B. Davison, J. A. Lowe, M. H. Smith, R. M. Harrison, C. N. Hewitt, J. Geophys. Res. 1997, 1, doi:1.19/96jd79 [1] E. K. Bigg, D. E. Turvey, Atmos. Environ. 1978, 1, 16. [11] M. Kulmala, L. Pirjola, J. M. Mäkelä, Nature,, 66. doi:1.138/35355 [1] C. D. O Dowd, K. Hämeri, J. M. Mäkelä, M. Väkeva, P. P. Aalto, G. de Leeuw, G. J. Kunz, E. Becker, H.-C. Hansson, A. G. Allen, R. M. Harrison, H. Berresheim, C. Kleefeld, M. Geever, S. G. Jennings, M. Kulmala, J. Geophys. Res., 17, 817. doi:1.19/jd6 [13] M. Dal Maso, M. Kumala, K. E. J. Lehtinen, J. M. Mäkelä, P. Aalto, C. D. O Dowd, J. Geophys. Res., 17, 897. doi:1.19/1jd153 [1] M. Väkevä, K. Hämeri, P. P. Aalto, J. Geophys. Res., 17, 81. doi:1.19/jd13 [15] J. M. Mäkelä, T. Hoffmann, C. Holzke, M. Väkevä, T. Suni, T. Mattila, P. P. Aalto, U. Tapper, E. I. Kauppinen, C. D. O Dowd, J. Geophys. Res., 17, 811. doi:1.19/1jd58 [16] L. J. Carpenter, W. T. Sturges, P. Liss, A. Alicke, K. Hebestreit, U. Platt, J. Geophys. Res. 1999, 1, doi:1.19/ 98JD76 [17] B. Alicke, K. Hebestreit, J. Stutz, U. Platt, Nature 1999, 397, 57. doi:1.138/1758 [18] M. Kulmala, P. Korhonen, I. Napari, A. Karlsson, H. Berresheim, C. D. O Dowd, J. Geophys. Res., 17, doi:1.19/ 1JD9 [19] C. D. O Dowd, G. McFiggens, L. Pirjola, D. J. Creasey, C. Hoell, M. H. Smith, B. Allen, J. M. C. Plane, D. E. Heard, J. D. Lee, M. J. Pilling, M. Kulmala, Geophys. Res. Lett. 1999, 6, 177. doi:1.19/1999gl9335 [] L. Pirjola, C. D. O Dowd, M. Kulmala, J. Geophys. Res., 17, 898. doi:1.19/jd13 [1] T. Hoffmann, C. D. O Dowd, J. H. Seinfeld, Geophys. Res. Lett. 1, 8, 199. doi:1.19/gl1399 [] J. L. Jimenez, R. Bahreini, D. R. Cocker III, H. Zhuang, V. Varutbangkul, R. C. Flagan, J. H. Seinfeld, C. D. O Dowd, T. Hoffmann, J. Geophys. Res. 3, 18, 318. doi:1.19/ JD5 [3] J. B. Burkholder, J. Curtius, A. R. Ravishankara, E. R. Lovejoy, Atmos. Chem. Phys.,, 19. [] B. J. Allan, J. M. C. Plane, G. McFiggans, Geophys. Res. Lett. 1, 8, 195. doi:1.19/gl168 [5] G. McFiggans, H. Coe, R. Burgess, J. Allan, M. Cubison, M. R. Alfarra, R. Saunders, A. Saiz-Lopez, A. R. Rickard, P. S. Monks, Atmos. Chem. Phys.,, 71. [6] A. Saiz-Lopez, R. W. Saunders, D. M. Joseph, S. H. Ashworth, J. M. C. Plane, Atmos. Chem. Phys.,, 13. [7] A. Saiz-Lopez, J. M. C. Plane, G. McFiggans, P. I. Williams, S. M. Ball, M. Bitter, R. L. Jones, C. Hongwei, T. Hoffmann, Atmos. Chem. Phys. 5, 5, 55. [8] K. Sellegri, Y. J. Yoon, S. G. Jennings, C. D. O Dowd, L. Pirjola, S. Cautenet, H. Chen, T. Hoffmann, Environ. Chem. 5,, 6. [9] R. J. Flanagan, M. Geever, C. D. O Dowd, Environ. Chem. 5,, 56. [3] L. Pirjola, C. D. O Dowd, Y. J. Yoon, K. Sellegri, Environ. Chem. 5,, 71. [31] C. D. O Dowd, J. L. Jimenez, R. Bahreini, R. C. Flagan, J. H. Seinfeld, L. Pirjola, M. Kulmala, S. F. G. Jennings, T. Hoffmann, Nature, 17, 63. doi:1.138/ NATURE775 [3] G. McFiggans, Nature 5, 33, E13. doi:1.138/ NATURE337 [33] C. D. O Dowd, J. L. Jimenez, R. Bahreini, R. C. Flagan, J. H. Seinfeld, L. Pirjola, M. Kulmala, S. F. G. Jennings, T. Hoffmann, Nature 5, 33, E13. doi:1.138/ NATURE3373 [3] F. Laturnus, Environ. Sci. Poll. Res. 1, 8, 13. [35] L. J. Carpenter, Chem. Rev. 3, 13, 953. doi:1.11/ CR665 [36] R. M. Moore, in The Handbook of Environmental Chemistry 3, pp (Springer-Verlag: Berlin). [37] S. Amachi, Y. Kamagata, T. Kanagawa, Y. Muramatsu, Appl. Environ. Microbiol. 1, 67, 718. doi:1.118/aem [38] C. Hughes, G. Malin, C. Turley, A. B. J. Keely, P. D. Nightingale, P. S. Liss, SOLAS Science conference, Halifax, Canada, October,. [39] H. Yamamoto, Y. Yokouchi, A. Otsuki, H. Itoh, Chemosphere 1, 5, 371. doi:1.116/s5-6535()51-5 [] M. Martino, P. S. Liss, J. M. C. Plane, Environ. Sci. Technol. 5, 39, 797. doi:1.11/es8718s [1] C. Hughes, PhD thesis. University of East Anglia, Norwich, UK,. [] S. Amachi, Y. Muramatsu, Y. Akiyama, K. Miyazaki, S. Yoshiki, S. Hanada, Y. Kamagata, T. Ban-nai, H. Shinoyama, T. Fujii, Microb. Ecol. 5, 9, 57. doi:1.17/s [3] Y. Miyake, S. Tsunogai, J. Geophys. Res. 1963, 68, [] J. C. Wren, J. M. Ball, G. A. Glowa, Nucl. Technol., 19, 97. [5] E. M. Knipping, M. J. Lakin, K. L. Foster, P. Jungwirth, D. J. Tobias, R. B. Gerber, D. Dabdub, B. Finlayson-Pitts, Science, 88, 31. doi:1.116/science [6] E. M. Knipping, D. Dabdub, J. Geophys. Res., 17, 36. doi:1.19/1jd867 [7] J. A. Garland, H. Curtis, J. Geophys. Res. 1981, 86, [8] P. Jungwirth, D. Tobias, J. Phys. Chem. 1, 15, [9] L. Vrbka, M. Mucha, B. Minofar, P. Jungwirth, E. Brown, D. Tobias, Curr. Opinion Coll. Interf. Sci., 9, 67. doi:1.116/j.cocis..5.8 [5] B. J. Finlayson-Pitts, Chem. Rev. 3, 13, 81. doi:1.11/ CR653T [51] N. Begovic, Z. Markovic, S. Anic, L. Kolar-Anic, Environ. Chem. Lett.,, 65. [5] A. Baker, D. Thompson, M. Campos, S. Parry, T. Jickells, Atmos. Environ., 3, 331. doi:1.116/s135-31()8-9 [53] A. R. Baker, Geophys. Res. Lett., 31, L3S. doi:1.19/ GL1 [5] R. Vogt, R. Sander, R. von Glasow, P. J. Crutzen, J. Atmos. Chem. 1999, 3, 375. doi:1.13/a: [55] G. McFiggans, J. M. C. Plane, B. J. Allan, L. J. Carpenter, H. Coe, C. O Dowd, J. Geophys. Res. Atmos., 15, doi:1.19/1999jd