INTERCOMP2000: the comparability of methods in use in Europe for measuring the carbon content of aerosol

We tested the comparability of methods, in use in Europe, for measuring the concentration of aerosol-carbon. The study was performed in the framework of INTERCOMP2000, a field campaign organised in the first two weeks of April 2000 in the vicinity of the village of Melpitz, 45 km NE of Leipzig, Germany. Sampling of the aerosol took place with 19 samplers. Total carbon (TC) was measured with eleven methods and the mean concentration found for PM 10 was close 3 to 3 mugm(-3), of which 90% was in the PM2.5 fraction. Carbon and associated compounds constituted approximately 25% of the total aerosol mass in this fine fraction. Starting point of the evaluation of the data were the concentrations stemming from two identical PM2.5 high-volume samplers, of which the samples were analysed by four institutes. The daily values of TC were highly correlated and the (random) differences were of an absolute nature, with an overall standard deviation (SD) of 0.22 ( +/-0.12) mugm(-3). This value for the SD is only twice that in the data of the individual institutes for a similar number of samples. In the series of four PM10 filter-samplers, the highest concentration was obtained with the sampler that collected the lowest amount per unit of surface area. This difference is indicative of an artefact caused by adsorption of gaseous carbon containing compounds. Evidence for such an adsorption artefact was provided by the presence of carbon in back-up filters, in amounts up to 20% of those of front filters. Adsorption can also be (part of) the explanation for a difference of 30% observed between concentrations obtained with filters and cascade impactors as samplers, because the latter collectors do not suffer from an adsorption artefact. Samples were also analysed for elemental carbon (EC). While differences of more than a factor of three between methods were observed, the correlation of the daily EC-concentrations over the period was high, better than 0.9 (R-2). Two institutes analysed samples from filters and cascade impactors in the same way. Both observed that the EC-values from the filters were a factor of 1.4 higher than those from the impactors. Three methods were used to determine black carbon (BC) by optical means. Differences between the methods were large, but apparently of a systematic nature, because the correlation of the daily BC-values was very good. The correlation of daily BC and EC concentrations was also high. (C) 2004 Elsevier Ltd. All rights reserved.