A volatility tandem differential mobility analyzer (VTDMA) was developed that offers high size resolution with high temporal resolution: the eight heating columns of the VTDMA-8 allow rapid measurement of the non-volatile component of aerosol particles within a narrow particle size range in about 15 min. The non-volatile composition of these quasi-monodisperse particles is described in terms of number fraction Phi(N) and volume fraction Phi(V) of the non-volatile material. Laboratory studies were made to determine the transfer efficiency and the volatile behavior of sulfate particles in the VTDMA-8. Its applicability to atmospheric aerosols was confirmed during three field measurements. Since the distribution of carbonaceous material depends on the distance to combustion sources and varies with region (i.e., urban or rural) and with time, we made continuous surface measurements in air masses polluted to different levels: (a) in Falkenberg, a moderately polluted, rural region in north-eastern Germany, (b) in the city of Leipzig, a strongly polluted environment, and (c) directly at a combustion source in the exhaust of a diesel-fueled vehicle. The results show that the refractory material contained in 150 nm atmospheric particles (at Falkenberg and Leipzig) ranged between 8% and 25% in volume and represents an external mixture of particles containing various amounts of non-volatile material. However, the vehicle-emitted 50 nm particles at a single combustion source were internally mixed, and that non-volatile fraction ranged between 75% and 98% in volume, depending on the operating conditions of the vehicle (speed, engine revolution). The results show that throughout all measurements the total number of particles remained unchanged after evaporation of the volatile material. Our results demonstrate that the VTDMA-8 can rapidly measure the non-volatile fractions of size-segregated pollution aerosols and that it is applicable to study the mixing state of aerosols and the effect of carbonaceous material on aerosol aging processes. (C) 2003 Elsevier Ltd. All rights reserved.