Secondary organic aerosol (SOA) particles are formed in the atmosphere from condensable oxidation products of anthropogenic and biogenic volatile organic compounds (VOCs). On a global scale, biogenic VOCs account for about 90% of VOC emissions and of SOA formation (90 billion kilograms of carbon per year). SOA particles can scatter radiation and act as cloud condensation or ice nuclei, and thereby influence the Earth's radiation balance and climate. They consist of a myriad of different compounds with varying physicochemical properties, and little information is available on the phase state of SOA particles. Gas-particle partitioning models usually assume that SOA particles are liquid, but here we present experimental evidence that they can be solid under ambient conditions. We investigated biogenic SOA particles formed from oxidation products of VOCs in plant chamber experiments and in boreal forests within a few hours after atmospheric nucleation events. On the basis of observed particle bouncing in an aerosol impactor and of electron microscopy we conclude that biogenic SOA particles can adopt an amorphous solid-most probably glassy-state. This amorphous solid state should provoke a rethinking of SOA processes because it may influence the partitioning of semi-volatile compounds, reduce the rate of heterogeneous chemical reactions, affect the particles' ability to accommodate water and act as cloud condensation or ice nuclei, and change the atmospheric lifetime of the particles. Thus, the results of this study challenge traditional views of the kinetics and thermodynamics of SOA formation and transformation in the atmosphere and their implications for air quality and climate.

The formation of new atmospheric particles with diameters of 3-10 nm has been observed at a variety of altitudes and locations. Such aerosol particles have the potential to grow into cloud condensation nuclei, thus affecting cloud formation as well as the global radiation budget. In some cases, the observed formation rates of new particles have been adequately explained by binary nucleation, involving water and sulphuric acid, but in certain locations--particularly those within the marine boundary layer and at continental sites--observed ambient nucleation rates exceed those predicted by the binary scheme. In these locations, ambient sulphuric acid (H2SO4) levels are typically lower than required for binary nucleation, but are sufficient for ternary nucleation (sulphuric acid-ammonia-water). Here we present results from an aerosol dynamics model with a ternary nucleation scheme which indicate that nucleation in the troposphere should be ubiquitous, and yield a reservoir of thermodynamically stable clusters 1-3 nm in size. We suggest that the growth of these clusters to a detectable size (> 3 nm particle diameter) is restricted by the availability of condensable vapour. Observations of atmospheric particle formation and growth from a continental and a coastal site support this hypothesis, indicating that a growth process including ternary nucleation is likely to be responsible for the formation of cloud condensation nuclei.

Taking advantage of only the measured aerosol particles spectral evolution as a function of time, a new analytical tool is developed to derive formation and growth properties of nucleation mode aerosols. This method, when used with hygroscopic growth-factors, can also estimate basic composition properties of these recently-formed particles. From size spectra the diameter growth-rate can be obtained, and aerosol condensation and coagulation sinks can be calculated. Using this growth-rate and condensation sink, the concentration of condensable vapours and their source rate can be estimated. Then, combining the coagulation sink together with measured number concentrations and apparent source rates of 3 nm particles, 1 nm particle nucleation rates and concentration can be estimated. To estimate nucleation rates and vapour concentration source rates producing new particle bursts over the Boreal forest regions, three cases from the BIOFOR project were examined using this analytical tool. In this environment, the nucleation mode growth-rate was observed to be 2-3 nm hour−1, which required a condensable vapour concentration of 2.5-4×107 cm−3 and a source rate of approximately 7.5-11×104 cm−3 s−1 to be sustained. The formation rate of 3 nm particles was #1 particle cm−3 s−1 in all three cases. The estimated formation rate of 1 nm particles was 10-100 particles cm−3 s−1, while their concentration was estimated to be between 10,000 and 100,000 particles cm−3. Using hygroscopicity data and mass flux expressions, the mass flux of insoluble vapour is estimated to be of the same order of magnitude as that of soluble vapour, with a soluble to insoluble vapour flux ratio ranging from 0.7 to 1.4 during these nucleation events.

[1] The formation and growth of new particles has been evaluated using a revised version of a simple, but novel, theoretical tool. The concentration of condensable vapors and their source rates has been estimated using the aerosol condensation sink together with the measured particle growth rate. Also, by adding the coagulation sink and the measured formation rate of 3 nm particles, the formation rate of 1 nm particles and their concentration can be estimated. Condensation and coagulation sinks can be obtained from ambient aerosol size distribution data. The method has been applied to analyze the particle formation and growth rates observed during coastal and boreal forest nucleation events. The condensation sinks are typically 4-7 Â 10 À3 s À1 in the forest and 2 Â 10 À3 s À1 under coastal conditions, while the coagulation sinks for 1, 2, and 3 nm particles are typically smaller by factors 1.5-2, 5-7, and 11-15, respectively. The measured growth rates are 2-10 nm/h for the boreal forest and range from 15 to 180 nm/h at the coast, corresponding to a vapor concentration of 2-13 Â 10 7 cm À3 and 10 8 cm À3 to 10 9 cm À3 , respectively. The vapor source rate was 1-2 Â 10 5 cm À3 s À1 in the boreal forest and 2-5 Â 10 6 cm À3 s À1 in the coastal environment. The estimated formation rate of 1 nm particles in the forest environment was 8-20 cm À3 s À1 and 300-10,000 cm À3 s À1 at the coast. The concentration of 1 nm particles was estimated to be 2000-5000 and 4 Â 10 4 -7 Â 10 6 particles cm À3 in forest and at coast, respectively.

We have measured submicron particles rather continuously since January 31st 1996 at a forest site in Southern Finland. Number size distribution data from the size range of 3-500 nm (particle diameter) are obtained every 10 minutes. This article attemps to give an overall view over diurnal pattern of submicron size distribution. We will show the typical ambient submicron particle size distribution as it is monitored at the site and, furthermore, we want to highlight some certain events that have been observed during the nine month period. Since the measurements are carried out inside forest, and there are no local sources of pollution nearby of the site, we have also to include the possibility to observe the formation of aerosol particles of biogenic origin.

[1] Atmospheric concentrations of gaseous sulfuric acid (H 2 SO 4 ), methane sulfonic acid (MSA), and hydroxyl radicals (OH) were measured by chemical ionization mass spectrometry (CIMS) during the second New Particle Formation and Fate in the Coastal Environment (PARFORCE) campaign in June 1999 at Mace Head, Ireland. Overall median concentrations in marine background air were 1.5, 1.2, and 0.12 ϫ 10 6 cm Ϫ3 , respectively. H 2 SO 4 was also present at night indicating significant contributions from nonphotochemical sources. A strong correlation was found between daytime OH and H 2 SO 4 levels in clean marine air suggesting a fast local production of H 2 SO 4 from sulfur precursor gases. Steady state balance calculations of ambient H 2 SO 4 levels agreed with measured concentrations if either very low H 2 SO 4 sticking coefficients (0.02-0.03) or sources in addition to the SO 2 ϩ OH reaction were assumed. Overall, variations in ambient H 2 SO 4 levels showed no correlation with either the tidal cycle or ultrafine particle (UFP) concentrations. However, on particular days an anticorrelation between H 2 SO 4 and UFP levels was occasionally observed providing evidence for the contribution of H 2 SO 4 to new particle formation and/or particle growth. Gaseous MSA concentrations were inversely correlated with dew point temperature reflecting a highly sensitive gas-particle partitioning equilibrium of this compound. The present observations seriously question the general use of MSA as a conservative tracer to infer the relative production yield of H 2 SO 4 from dimethylsulfide (DMS) oxidation. MSA/H 2 SO 4 concentration ratios typically ranged between 0.06 and 1.0 in marine air at ground level. Measured diel OH profiles showed a significant deviation from concurrent variations of the ozone photolysis frequency. They also showed up to 1 order of magnitude lower values compared to OH concentrations calculated with a simple photochemical box model. These differences were most pronounced during particle nucleation events occurring on sunny days around noon and at low tide. The present results suggest that both the oxidation capacity and the particle formation potential in the coastal boundary layer were significantly affected by reactions of unknown compounds prevailing in this type of environment.

Abstract. Aerosol number size distributions have been measured since 5 May 1997 in Helsinki, Finland. The presented aerosol data represents size distributions within the particle diameter size range 8-400 nm during the period from May 1997 to March 2003. The daily, monthly and annual patterns of the aerosol particle number concentrations were investigated. The temporal variation of the particle number concentration showed close correlations with traffic activities. The highest total number concentrations were observed during workdays; especially on Fridays, and the lowest concentrations occurred during weekends; especially Sundays. Seasonally, the highest total number concentrations were observed during winter and spring and lower concentrations were observed during June and July. More than 80% of the number size distributions had three modes: nucleation mode (D p <30 nm), Aitken mode (20-100 nm) and accumulation mode (D p >90 nm). Less than 20% of the number size distributions had either two modes or consisted of more than three modes. Two different measurement sites were used; in the first (Siltavuori, 5.5.1997(Siltavuori, 5.5. -5.3.2001, the arithmetic means of the particle number concentrations were 7000 cm −3 , 6500 cm −3 , and 1000 cm −3 respectively for nucleation, Aitken, and accumulation modes. In the second site (Kumpula, 6.3.2001(Kumpula, 6.3. -28.2.2003) they were 5500 cm −3 , 4000 cm −3 , and 1000 cm −3 . The total number concentration in nucleation and Aitken modes were usually significantly higher during workdays than during weekends. The temporal variations in the accumulation mode were less pronounced. The lower concentrations at Kumpula were mainly due to building construction and also the slight overall decreasing trend during these years. During the site changing a period of simultaneous measurements over two weeks were performed showing nice correlation at both sites.

No abstract