No abstract

[1] We carried out a detailed size-resolved chemical characterization of particle emissions from the combustion of European conifer species, savanna grass, African hardwood, and German and Indonesian peat. Combustion particles were sampled using two sets of five-stage Berner-type cascade impactors after a buffer volume and a dilution tunnel. We determined the emission factors of water-soluble organic carbon (WSOC, 46-6700 mg kg À1 , sum of five stages), water-insoluble organic carbon (WISOC, 1300-6100 mg kg À1 ), (apparent) elemental carbon (ECa, 490-1800 mg kg À1 ), inorganic ions (68-400 mg kg À1 ), n-alkanes (0.38-910 mg kg À1 ), n-alkenes (0.45-180 mg kg À1 ), polycyclic aromatic hydrocarbons (PAHs) (1.4-28 mg kg À1 ), oxy-PAHs (0.08-1.0 mg kg À1 ), lignin decomposition products (59-620 mg kg À1 ), nitrophenols (1.4-31 mg kg À1 ), resin acids (0-110 mg kg À1 ), and cellulose and hemicellulose decomposition products (540-5900 mg kg À1 ). The combustion and particle emission characteristics of both of peat were significantly different from those of the other biofuels. Peat burning yielded significantly higher emission factors of total fine particles in comparison to the other biofuels. Very high emission factors of n-alkanes and n-alkenes were observed from peat combustion, which may be connected to the concurrently observed ''missing'' CCN in peat smoke. A highlevel ofmonosaccharide anhydrides, especiallylevoglucosan, wasdetectedfromall types of biofuel combustion. The fractions of monosaccharide anhydrides in the emitted total carbon were higher in smaller particles (aerodynamic diameter, D pa < 0.42 mm).

[1] The aerosol mixing state was investigated with an optical closure study at Xinken, Pearl River Delta of China in 2004. On the basis of in situ aerosol microphysical and chemical measurements and a two-component aerosol optical model an internal consistency algorithm was developed to model the mass ratio (r) of externally mixed elemental carbon (EC) to total EC, which minimized the discrepancies between measured and calculated optical properties. The rest of EC was assumed to be internally mixed. A time series of r was retrieved. Good agreement between model and observation was found, on the order of ±15% for total/back scattering coefficients and ±10% for absorption coefficient. The EC mixing state was strongly dependent on the local wind patterns. When north/northeasterly winds prevailed, the air came from the urban and industrial areas of mainland China, and EC was mainly externally mixed with an average r of 85 ± 12%. When the airflow was controlled by a weak local wind system, the mixing state showed a pronounced diurnal variation. During daytime the wind speed was nearly zero. This favored the increase of local pollution, and the average r was about 95%. However, during nighttime the EC mixing state transformed to be internally mixed apparently with an average r of 53 ± 15%, which can be explained by a more aged air mass. The south/ southeasterly winds coming from the sea were found to have the most important effect on the transformation of EC mixing state in the night, but fairly rapid local aging processing was also observed. The uncertainties of the model were explored by a Monte Carlo simulation.

Abstract. Atmospheric aerosol particle size distributions at a continental background site in Eastern Germany were examined for a one-year period. Particles were classified using a twin differential mobility particle sizer in a size range between 3 and 800 nm. As a novelty, every second measurement of this experiment involved the removal of volatile chemical compounds in a thermodenuder at 300 • C. This concept allowed to quantify the number size distribution of nonvolatile particle cores -primarily associated with elemental carbon, and to compare this to the original non-conditioned size distribution. As a byproduct of the volatility analysis, new particles originating from nucleation inside the thermodenuder can be observed, however, overwhelmingly at diameters below 6 nm. Within the measurement uncertainty, every particle down to particle sizes of 15 nm is concluded to contain a non-volatile core. The volume fraction of non-volatile particulate matter (non-conditioned diameter < 800 nm) varied between 10 and 30% and was largely consistent with the experimentally determined mass fraction of elemental carbon. The average size of the non-volatile particle cores was estimated as a function of original non-conditioned size using a summation method, which showed that larger particles (>200 nm) contained more non-volatile compounds than smaller particles (<50 nm), thus indicating a significantly different chemical composition. Two alternative air mass classification schemes based on either, synoptic chart analysis (Berliner Wetterkarte) or back trajectories showed that the volume and number fraction of non-volatile cores depended less on air mass than the total particle number concentration. In all air masses, the non-volatile size distributions showed a more and a less volatile ("soot") mode, the latter being located at about 50 nm. During unstable conditions and in maritime air masses, smaller values were observed compared to Correspondence to: W. Birmili (birmili@tropos.de) stable or continental conditions. This reflects the significant emission of non-volatile material over the continent and, depending on atmospheric stratification, increased concentrations at ground level.

No abstract