Gas/particle partitioning of seven volatile polycyclic aromatic hydrocarbons in a heavy traffic urban area.

Air samples (vapor- and particle-phase) were taken for 19 sampling events during the period from December 1997 to July 1998 in an urban site in the center of Athens. The urban site is densely populated and characterized by heavy traffic circulation and elevated concentrations of VOCs, NO(x), CO and smoke. Seven volatile polycyclic aromatic hydrocarbons (PAHs) were determined in samples. The temperature dependence of gas-phase atmospheric concentration of PAHs, C(g), was investigated using diagrams of natural logarithm of partial pressures (lnP) vs. reciprocal mid-point temperatures. For the six of seven volatile PAHs, the temperature dependence of lnP was statistically significant (at least at the 90% confidence level) and the temperature accounted for 21-67% of the variability in gas-phase concentrations. The gas-phase concentration C(g) of the very volatile PAHs was affected more significantly by changes in temperature, but the variation of the less-volatile PAHs fluoranthene and pyrene C(g), was better explained by changes in temperature. The temperature dependence of gas/particle partitioning constant K(P) was also examined. Regressions of log(K(P))(-1) vs. T(-1) for fluorene, fluoranthene and pyrene were classified into two different temperature ranges. The gas/particle partitioning of PAHs was studied by correlating the partition constant to the sub-cooled liquid saturation vapor pressure (P(L)(o)). The Junge adsorption model underestimated the particle fraction of volatile PAHs probably due to the presence of non-exchangeable fraction. Slopes (m(r)) of the regressions logK(P) vs. logP(L)(o) were different from the value -1 as Pankow's theory predicts. The short distance between the sampling point and the emission sources is also estimated to be a factor that causes deviations from the theoretical value. Evidence that atmospheric conditions favorable for secondary aerosol formation coincide with higher value of m(r), was provided by limited sampling events. An interrelation was found to exist between the m(r) values, allowing the prediction of the gas/particle partitioning of a series of seven PAHs by the measurement of a single PAH partitioning.