The formation of secondary organic aerosol in the atmosphere remains one of the most poorly understood aspects of the air pollution problem in urban areas. Photooxidation of gas-phase emissions produces low vapor pressure species that are converted to the aerosol phase either by homogeneous nucleation of new particles or by condensation onto existing particles. One of the goals in studying aerosol dynamics in atmospheric systems is to determine the factors that govern which of these two pathways dominates in the conversion of gas-phase species to the aerosol phase.
We have conducted an extensive series of experiments aimed at elucidating the physics of atmospheric organic aerosol formation. An outdoor smog chamber was used to study the formation and growth of secondary aerosol resulting from the photooxidation of aromatic hydrocarbons (toluene, m-xylene, ethyl benzene, and 1,3,5-trimethyl benzene) in the presence of NOₓ. In the experiments, particular emphasis was given to the effect of primary aerosol on the subsequent aerosol evolution in the system. We observed that with a sufficient number concentration of initial seed particles in the system, homogeneous nucleation could be suppressed and all gas-to-particle conversion occurred via condensation onto the seed particles.
Aerosol yields by mass from the gas phase were calculated for each experiment. These yields were somewhat dependent on the initial hydrocarbon/NOₓ ratio in each experiment, which is an indication of the system reactivity. Average yields for each aromatic species were: toluene - 4.8%, m-xylene - 3.5%, ethyl benzene - 1.9%, and 1,3,5-trimethyl benzene - 2.4%. These results are in good agreement with previous determinations of aerosol yield for the toluene and m-xylene systems.
Several models were used to describe the observed aerosol dynamics. An integral model assuming a monodisperse aerosol, developed by Warren and Seinfeld (1984, 1985b), was used to determine apparent saturation vapor pressures of condensible species from the observations of nucleation events. Overall predictions of final number concentrations with the integral model, based on these saturation vapor pressures, were fairly close to the experimentally observed number concentrations.
An analysis of the rate of aerosol growth was carried out for those experiments exhibiting uniform condensational growth. This analysis provided estimates for the gas-phase partial pressures of the condensible species, which could be compared with the integral model vapor pressures to give approximate saturation ratios during these periods of growth.
Full aerosol size distribution simulations were performed using the sectional model ESMAP (Warren and Seinfeld, 1985a), based on the work of Gelbard et al. (1980). Number concentrations resulting from these predictions were higher than those of the integral model, since the condensation rate on a polydisperse aerosol is smaller than that on a monodisperse distribution, leading to a higher nucleation rate. Comparisons of predicted and observed size, distributions during the course of each experiment were limited in accuracy by the numerical diffusion associated with current versions of the sectional model.
