Secondary organic aerosol formation from intermediate-volatility organic compounds: cyclic, linear, and branched alkanes.

Intermediate volatility organic compounds (IVOCs) are an important class of secondary organic aerosol (SOA) precursors that have not been traditionally included in chemical transport models. A challenge is that the vast majority of IVOCs cannot be speciated using traditional gas chromatography-based techniques; instead they are classified as an unresolved complex mixture (UCM) that is presumably made up of a complex mixture of branched and cyclic alkanes. To better understand SOA formation from IVOCs, a series of smog chamber experiments was conducted with different alkanes, including cyclic, branched, and linear compounds. The experiments focused on freshly formed SOA from hydroxyl (OH) radical-initiated reactions under high-NO(x) conditions at typical atmospheric organic aerosol concentrations (C(OA)). SOA yields from cyclic alkanes were comparable to yields from linear alkanes three to four carbons larger in size. For alkanes with equivalent carbon numbers, branched alkanes had the lowest SOA mass yields, ranging between 0.05 and 0.08 at a C(OA) of 15 μg m(-3). The SOA yield of branched alkanes also depends on the methyl branch position on the carbon backbone. High-resolution aerosol mass spectrometer data indicate that the SOA oxygen-to-carbon ratios were largely controlled by the carbon number of the precursor compound. Depending on the precursor size, the mass spectrum of SOA produced from IVOCs is similar to the semivolatile-oxygenated and hydrocarbon-like organic aerosol factors derived from ambient data. Using the new yield data, we estimated SOA formation potential from diesel exhaust and predict the contribution from UCM vapors to be nearly four times larger than the contribution from single-ring aromatics and comparable to that of polycyclic aromatic hydrocarbons after several hours of oxidation at typical atmospheric conditions. Therefore, SOA from IVOCs may be an important contributor to urban OA and should be included in SOA models; the yield data presented in this study are suitable for such use.