How comparable are size-resolved particle number concentrations from different instruments?

The need for comparability of particle size resolved measurements originates from multiple drivers including: (i) Recent suggestions that air quality standards for particulate matter should migrate from being mass-based to incorporating number concentrations. This move would necessarily be predicated on measurement comparability which is absolutely critical to compliance determination. (ii) The need to quantify and diagnose causes of variability in nucleation and growth rates in nano-particle experiments conducted in different locations. (iii) Epidemiological research designed to identify key parameters in human health responses to fine particle exposure. Here we present results from a detailed controlled laboratory instrument inter-comparison experiment designed to investigate data comparability in the size range of 2.01-523.3 nm across a range of particle composition, modal diameter and absolute concentration. Particle size distributions were generated using a TSI model 3940 Aerosol Generation System (AGS) diluted using zero air, and sampled using four TSI Scanning Mobility Particle Spectrometer (SMPS) configurations and a TSI model 3091 Fast Mobility Particle Sizer (FMPS). The SMPS configurations used two Electrostatic Classifiers (EC) (model 3080) attached to either a Long DMA (LDMA) (model 3081) or a Nano DMA (NDMA) (model 3085) plumbed to either a TSI model 3025A Butanol Condensed Particle Counting (CPC) or a TSI model 3788 Water CPC. All four systems were run using both high and low flow conditions, and were operated with both the internal diffusion loss and multiple charge corrections turned on. The particle compositions tested were sodium chloride, ammonium nitrate and olive oil diluted in ethanol. Particles of all three were generated at three peak concentration levels (spanning the range observed at our experimental site), and three modal particle diameters. Experimental conditions were maintained for a period of 20 minutes to ensure experimental stationarity and in the data analysis only the middle 18 minutes of data are analyzed. Because of variations in the discretization of the different instrumental configurations, the data are analyzed both after being transformed onto a common size resolution and in terms of a fitted modal distribution. Diagnostic analysis are conducted to assess the impact of SMPS configuration on total number concentration, model geometric mean diameter and distribution dispersion. Preliminary results indicate that selection of DMA exerts the larger control over instrument response.