Gas–particle partitioning of polycyclic aromatic hydrocarbons (PAHs) in an urban traffic site in Eskisehir, Turkey

Gas and particle-phase ambient air samples were collected at a traffic site in Eskisehir over a six-month period between January and October 2006. Gas/particle concentrations of 15 PAHs were determined by analyzing integrated glass fiber and polyurethane foam samples. On average, 66% of the total concentrations of PAHs were found in the gas-phase in heating and 69% in the non-heating period samples. Heating period PAH concentrations were found to be 7 to 8 times greater than concentrations in non-heating period samples. Multiple linear regression analysis was carried out to determine the effect of meteorological parameters on measured individual PAH concentrations. Temperature, wind speed and wind direction explained 43% (dibenzo[a,h]anthracene and benzo[a]pyrene) to 78% (phenanthrene) of the variability in atmospheric PAH concentrations. Results of the multiple linear regression analysis indicated that temperature and wind speed were statistically significant factors for the measured concentrations of PAHs. Gas–particle partitioning coefficients, (Kp) and particle-phase fractions (Φ) of PAHs, were correlated with supercooled vapor pressures (PLo). Application of non-linear fitting for Φ versus log PLo plots was found to be more robust than linear logarithmic regressions of log Kp versus log PLo plots. Particle-phase fractions (Φ) for each PAH were also calculated using octanol–air and soot–air partitioning models. Soot model resulted better predictions of Φ for fluorene, phenanthrene, anthracene, fluoranthene and pyrene. Both two models yielded very similar particle-phase fractions having very close results to experimentally obtained Φ values for the rest of PAHs measured.