A high-volume Andersen sampler has been used to measure particle size distributions for atmospheric Pb, Br, and Cl at urban control sites and near secondary lead refineries in Toronto, Canada. Bromine and Cl were measured by instrumental neutron activation analysis, while Pb was determined both by absorption spectrophotometry and by instrumental photon activation analysis. The Pb aerosol near roadsides was predominantly submicrometer in size, whereas elevated Pb concentrations in the vicinity of the refineries were attributed mainly to increases in Pb associated with larger particles which settle more rapidly. The strong correlation between Pb and Br from automobile emissions allowed an indirect estimate of nonautomotive Pb in the atmosphere near the refineries. Chlorine, although also associated with automobile emissions, appears to be derived from several additional sources. The localized contamination of residential areas, which has resulted from refinery emissions, emphasizes the importance of particle size considerations in establishing air quality standards.An increasing amount of information on the multielement content of urban aerosols has led several investigators to deduce possible trace element sources and to estimate the relative importance of such sources to ambient air quality (1-3). The additional information provided by particle size sampling done in conjunction with multielement trace analysis greatly increases the possibility of source identification, occasionally allowing one to distinguish between different sources of the same element. This has been clearly demonstrated by Martens et al. (4), who have used aero-

Developing high fidelity quantitative models of solid state reaction systems can be challenging, especially in deposition systems where, in addition to the multiple competing processes occurring simultaneously, the solid interacts with its atmosphere.In this work, we develop a model for the growth of a thin solid film where species from the atmosphere adsorb, diffuse, and react with the film. The model is mesoscale and describes an entire film with thickness on the order of microns. Because it is stochastic, the model allows us to examine inhomogeneities and agglomerations that would be impossible to characterize with deterministic methods. We demonstrate the modeling approach with the example of chalcopyrite Cu(InGa)(SeS) 2 thin film growth via precursor reaction, which is a common industrial method for fabricating thin film photovoltaic modules. The model is used to understand how and why through-film variation in the composition of Cu(InGa)(SeS) 2 thin films arises and persists. We believe that the model will be valuable as an effective quantitative description of many other materials systems used in semiconductors, energy storage, and other fast-growing industries.

No abstract