PAHs, PCBs and organochlorine pesticides in the atmosphere of Eastern Mediterranean: Investigation of their occurrence, sources and gas-particle partitioning in relation to air mass transport pathways

Abstract

Thirty-four air samples were collected from 2013 to 2015, at a semi-rural site in Eastern Mediterranean (Island of Crete), to study the atmospheric occurrence of polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs) and organochlorine pesticides (OCPs) in the gas and particulate phase. Average levels (gaseous and particulate phase) of PAHs (36 compounds, 11–18 ng/m3), PCBs (49 congeners, 77–93 pg/m3) and OCPs (23 substances, 77–140 pg/m3) were comparable to those reported for similar locations worldwide. Multiple-linear regression analysis, performed to relate atmospheric concentrations with meteorological conditions, revealed as main controlling factors local sources for PAHs and long-range transport (LRT) for PCBs and OCPs. The consideration of parent-metabolite ratios for most OCPs excluded fresh inputs. The application of the potential source contribution function (PSCF) identified Black Sea and eastern Balkans as likely sources for PCBs and OCPs. Significant linear correlations (R2 = 0.79–0.98) were determined between the partitioning coefficients (logKp) and partial vapor pressures (logPL0) for most air samples for PAHs and PCBs excepting OCPs. Slope mr values were close to −1 for PAHs and OCPs indicating gas/particle partitioning close to equilibrium. The corresponding mr values for PCBs were shallower (<-0.60) denoting non-equilibrium conditions and potential sampling artefacts. The octanol-air partition coefficient absorption model, logKp-logKoa, did not offer robust evidence for the evaluation of the atmospheric partitioning of the studied compounds. Experimentally determined particle fractions (ϕ) fitted better with the typical remote and rural curves as predicted by the Junge-Pankow model for most PAHs and PCBs but not for OCPs. The Koa-fom absorptive model could not adequately simulate the measured ϕ values for the majority of the compounds.