Theoretical versus observed gas-particle partitioning of carbonyl emissions from motor vehicles

A state-of-the-science thermodynamic model describing gas-particle absorption processes was used to predict the gas-particle partitioning of mixtures of approximately 60 carbonyl compounds emitted from low-emission gasoline-powered vehicles, three-way catalyst gasoline-powered vehicles, heavy-duty diesel vehicles under the idle-creep condition (HDDV idle), and heavy-duty diesel vehicles under the five-mode test (HDDV 5-mode). Exhaust was diluted by a factor of 120-580 with a residence time of approximately 43 sec. The predicted equilibrium absorption partitioning coefficients differed from the measured partitioning coefficients by several orders of magnitude. Time scales to reach equilibrium in the dilution sampling system were close to the actual residence time during the HDDV 5-mode test and much longer than the actual residence time during the other vehicle tests. It appears that insufficient residence time in the sampling system cannot uniformly explain the failure of the absorption mechanism to explain the measured partitioning. Other gas-particle partitioning mechanisms (e.g., heterogeneous reactions, capillary adsorption) beyond the simple absorption theory are needed to explain the discrepancy between calculated carbonyl partitioning coefficients and observed partitioning. Both of these alternative partitioning mechanisms imply great challenges for the measurement and modeling of semi-volatile primary organic aerosol (POA) species from motor vehicles. Furthermore, as emitted particle concentrations from newer vehicles approach atmospheric background levels, dilution sampling systems must fundamentally change their approach so that they use realistic particle concentrations in the dilution air to approximately represent real-world conditions. Samples collected with particle-free dilution air yielding total particulate matter concentrations below typical ambient concentrations will not provide a realistic picture of partitioning for semi-volatile compounds