Deriving the mean primary-particle diameter and related quantities from the size distribution and the gravimetric mass of spark generated nanoparticles.

Spark generated carbon and iridium nanoparticles were characterised by their electrical-mobility diameter D and by the mass of particulate matter collected in parallel on filter. The particles exhibited slightly skewed lognormal size distributions with mean mobility diameters between 18 and 74 nm. The masses calculated from the measured distributions under the assumption that the particles were spherical (diameter D) and of bulk mass density turned out to be much higher than the gravimetric mass, by factors between 8 and as high as 340. This very pronounced difference initiated a search for an improved relation between particle size and mass. Data analysis suggested that the mass increases linearly with increasing D. Hence the measured distributions were evaluated under the assumption that the spark generated matter was composed of spherical primary nanoparticles of mean diameter d, aggregated in the form of chains of joint length beta D, with beta > 1. Using reasonable values of beta between 2 and 4, the mean diameter of carbon primary particles turned out to be 10 +/- 1.8 nm, in excellent agreement with size data recently obtained by transmission electron microscopy (TEM). The primary iridium particles were found to be distinctly smaller, with diameters between 3.5 +/- 0.6 nm and 5.4 +/- 0.9 nm. The comparatively small uncertainty is due to the fact that the primary-particle diameter is proportional to the square root of beta. The calculated volume specific surface areas range between 500 and 1700 m(2)/cm(3). These numbers are close to the 'active' surface areas previously measured by the BET method. The good agreement with TEM and BET data suggests that the novel approach of nanoparticle characterisation is meaningful. Accordingly, the number concentrations of all individual primary particles rather than the concentrations measured by the mobility analyser should be considered the correct dose metric in studies on animal exposure to spark generated nanoparticles. The evaluated data imply that the numbers quoted in the literature must be enlarged by factors ranging between about 10 and a maximum as high as 80