High temporal resolution measurements of roadside particle size distributions and their implications for exposure

High time resolution, size segregated studies of ultrafine particle number concentrations (PNC) were performed at two UK roadside locations using a DMS500 particle spectrometer including sampling with and without a thermodenuder. Intercomparisons between the DMS500, an SMPS and a CPC, indicate that the SMPS can underestimate PNC due to the transient nature of excursions in concentrations and the longer scan time of the instrument. Short duration transient periods of high number concentrations made considerable contributions to hourly average PNC. The transient nature of changes in PNC potentially affects the type of exposure experienced in the roadside environment, as well as indoor outdoor exchange. The transient shifts were dominated by Aitken mode particles during high traffic periods, with the Aitken mode comprised of internally mixed particles with a low volatility core (thought to be primarily soot) and a surface of condensed and absorbed material. Transient periods of high nucleation mode PNC were also found, but were relatively more important during night-time and afternoon periods, due to secondary formation processes and lower concentrations of pre-existing particles. The thermodenuder studies indicated the presence of highly volatile liquid droplets in the nucleation mode which were lost on heating. Transient events and complexities in particle mixing state within the ultrafine fraction, coupled with discrepancies between instruments, make the potential adoption of a total number based air quality standard more complicated than first appears. Nucleation mode particle numbers are highly dependant on atmospheric and traffic conditions as well as pre-existing particle surface area and may be toxicologically less significant if they are predominantly liquid droplets. Aitken mode particles have been shown in previous studies to have higher toxicological significance due to their chemical composition. Their hourly average PNC correlated well with PM2.5 and estimated PM1 with lower significance of equivalent correlation of nucleation mode particles.