Modeling the Formation of Secondary Organic Aerosol. 1. Application of Theoretical Principles to Measurements Obtained in the α-Pinene/, β-Pinene/, Sabinene/, Δ^3-Carene/, and Cyclohexene/Ozone Systems

Secondary organic aerosol (SOA) forms in the atmosphere when volatile parent compounds are oxidized to form low-volatility products that condense to yield organic particulate matter (PM). Under conditions of intense photochemical smog, from 40 to 80% of the particulate organic carbon can be secondary in origin. Because describing multicomponent condensation requires a compound-by-compound identification and quantification of the condensable compounds, the complexity of ambient SOA has made it difficult to test the ability of existing gas/particle (G/P) partitioning theory to predict SOA formation in urban air. This paper examines that ability using G/P data from past laboratory chamber experiments carried out with five parent hydrocarbons (HCs) (four monoterpenes at 308 K and cyclohexene at 298 K) in which significant fractions (61−100%) of the total mass of SOA formed from those HCs were identified and quantified by compound. The model calculations were based on a matrix representation of the multicomponent, SOA G/P distribution process. The governing equations were solved by an iterative method. Input data for the model included (i) ΔHC (μg m^(-3)), the amount of reacted parent hydrocarbon; (ii) the α values that give the total concentration T (gas + particle phase, ng m^(-3)) values for each product i according to Ti = 10^3 αiΔHC; (iii) estimates of the pure compound liquid vapor pressure P^o_L values (at the reaction temperature) for the products; and (iv) UNIFAC parameters for estimating activity coefficients in the SOA phase for the products as a function of SOA composition. The model predicts the total amount M_o (μg m^(-3)) of organic aerosol that will form from the reaction of ΔHC, the total aerosol yield Y (= M_o/ΔHC), and the compound-by-compound yield values Y_i. An impediment in applying the model is the lack of literature data on P^o_L values for the compounds of interest or even on P^o_L values for other, similarly low-volatility compounds. This was overcome in part by using the G/P data from the α-pinene and cyclohexene experiments to determine P^o_L values for use (along with a set of 14 other independent polar compounds) in calculating UNIFAC vapor pressure parameters that were, in turn, used to estimate all of the needed P^o_L values. The significant degree of resultant circularity in the calculations for α-pinene and cyclohexene helped lead to the good agreement that was found between the Y_i values predicted by the model, and those measured experimentally for those two compounds. However, the model was also able to predict the aerosol yield values from β-pinene, sabinene, and Δ^3-carene, for which there was significatly less circularity in the calculations, thereby providing evidence supporting the idea that given the correct input information, SOA formation can in fact be accurately modeled as a multicomponent condensation process