Environ. Sci. Technol. 2003, 37, 4113-4121
Gas-Particle Partitioning of Semivolatile Organic Compounds (SOCs) on Mixtures of Aerosols in a Smog Chamber BHARADWAJ CHANDRAMOULI, MYOSEON JANG, AND RICHARD M. KAMENS* Department of Environmental Sciences and Engineering, University of North Carolina, Chapel Hill, North Carolina 27599-7431
The partitioning behavior of a set of diverse SOCs on two and three component mixtures of aerosols from different sources was studied using smog chamber experimental data. A set of SOCs of different compound types was introduced into a system containing a mixture of aerosols from two or more sources. Gas and particle samples were taken using a filter-filter-denuder sampling system, and a partitioning coefficient Kp was estimated using Kp ) Cp/(CgTSP). Particle size distributions were measured using a differential mobility analyzer and a light scattering detector. Gas and particle samples were analyzed using GCMS. The aerosol composition in the chamber was tracked chemically using a combination of signature compounds and the organic matter mass fraction (fom) of the individual aerosol sources. The physical nature of the aerosol mixture in the chamber was determined using particle size distributions, and an aggregate Kp was estimated from theoretically calculated Kp on the individual sources. Model fits for Kp showed that when the mixture involved primary sources of aerosol, the aggregate Kp of the mixture could be successfully modeled as an external mixture of the Kp on the individual aerosols. There were significant differences observed for some SOCs between modeling the system as an external and as an internal mixture. However, when one of the aerosol sources was secondary, the aggregate model Kp required incorporation of the secondary aerosol products on the preexisting aerosol for adequate model fits. Modeling such a system as an external mixture grossly overpredicted the Kp of alkanes in the mixture. Indirect evidence of heterogeneous, acid-catalyzed reactions in the particle phase was also seen, leading to a significant increase in the polarity of the resulting aerosol mix and a resulting decrease in the observed Kp of alkanes in the chamber. The model was partly consistent with this decrease but could not completely explain the reduction in Kp because of insufficient knowledge of the secondary organic aerosol composition.
Introduction The gas/particle partitioning of semivolatile organic compounds (SOCs) on aerosols, both ambient and chamber * Corresponding author fax: (919)966-7911; e-mail: kamens@ unc.edu. 10.1021/es026287c CCC: $25.00 Published on Web 08/15/2003
2003 American Chemical Society
TABLE 1. Experimental Conditions during the Smog Chamber Experiments date
temp (K)
08/23/2001 295-294 10/09/2001 284-281 06/04/2002 300-295
RH (%)
aerosol sources
45 75 46
wood + diesel road dust + diesel wood + diesel + R-pinene-O3
generated, has been studied using a variety of models (1-6). Studies focusing on predicting the ambient partitioning coefficient Kp of various compound classes have either used the subcooled vapor pressure (p0L) (2, 3) or the octanol-air partitioning coefficient (KOA) (7, 8) as a correlator. Predictive models have been developed both for absorptive partitioning onto aerosols with a liquid layer (5, 9) and for adsorptive partitioning (6, 10-12) on solid aerosols. However, most of these models have involved predicting the partitioning behavior of SOCs on single-source aerosols. There has been little work in the area of mechanistically predicting the partitioning behavior of SOCs in the presence of multiple aerosol sources. The influence of the mixing state of the aerosol on the Kp of SOCs is also unclear. Recent work (13) has indicated that in a chamber experiment involving partitioning of PAHs on a mixture of wood combustion particles and secondary organic aerosol (SOA), the Kp could be predicted assuming that the system was externally mixed and the aggregate Kp was a mass-weighted sum of the individual Kp of the polynuclear aromatic hydrocarbons (PAHs) in the two aerosols. However, it is not clear whether this can apply to other classes of compounds or other types of aerosol systems. The sensitivity of mixture Kp predictions to the mixing state of the aerosol is especially important because recent work (14) indicates that the mixing regime of aerosol in the atmosphere is varied and dependent on meteorological conditions. The formation of secondary organic aerosol on preexisting particles has the potential to alter the partitioning behavior of SOCs already present in the aerosol. Since SOA can comprise a significant portion of the ambient aerosol load (15, 16), the effects of SOA on the partitioning of SOCs need to be studied. In the present work, the gas/particle partitioning of a set of SOCs of different compound classes on various mixtures of organic and inorganic aerosols was studied. The objectives were 2-fold, first, to develop techniques to predict the aggregate Kp of the SOCs on the aerosol mixture, and second, to test the sensitivity of the partitioning coefficient to the mixing state employed for modeling the aerosol. Wood and diesel combustion aerosols, Arizona road dust, and secondary organic aerosol from the R-pinene-O3 reaction were used as aerosol sources to simulate a variety of particle types, organic layer activities, and mixture regimes.
Experimental Section All gas-particle partitioning data was obtained from chamber experiments carried out at the University of North Carolina at Chapel Hill outdoor smog chamber facility located near Pittsboro, NC (17) in twin, 25 m3 Teflon film chambers. The experimental conditions in the outdoor chamber are shown in Table 1. The experiments were conducted after sunset to preclude photochemical effects. To study partitioning of SOCs on a binary mixture of wood and diesel combustion aerosols, wood combustion aerosol was initially introduced into both the chambers by burning dry yellow pine in an Arrow catalytic wood stove operated in catalyst bypass mode (5). The SOCs VOL. 37, NO. 18, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4113
TABLE 2. Model Composition of Secondary Organic Aerosol during the 6-4-2002 Experiment at 0% Relative Humidity compound
mole fraction
pinic acid pinonic acid hydroxy-pinonaldehyde pinalic acid pinonaldehyde hydrate
0.04 0.02 0.66 0.06 0.22
(for list of SOCs injected, see Table 4) were then injected into the gas phase of the West chamber using a gently heated injector (5, 18, 19). Gas and particle phase samples were then collected to establish a “before mixture” Kp. A filterfilter-denuder sampling train consisting of two 47 mm Teflon coated glass fiber filters (type T60A20, Pallflex Products Corp., Putnam, CT), and a 40 cm, 5-channel annular denuder (University Research Glassware, Carrboro, NC) (20) was used to collect the samples for 20 min at 20 L min-1 as described in previous studies (21). Diesel combustion aerosol was then injected into the West chamber using a 1980 Mercedes Benz 300SD engine. Gas and particle samples were then taken. The East chamber (which contained only wood combustion aerosol) was used as a tracking chamber, and particle samples were taken at the same time as in the West chamber. For the dust + diesel experiment, Arizona road dust (supplied by the AC Spark Plug division of the General Motors Corporation, Flint, MI, U.S.A.) was injected into both the chambers using a high-pressure venturi aerosol generator (9, 22). After injection of the particles into the chamber, a wait period of 3 h was needed to ensure that the larger particles settled out and that the bulk of surface area in the chamber was associated with particles smaller than 2.5 µm. The SOCs were then injected into both the chambers,
followed by the injection of diesel exhaust aerosol into the West chamber. Gas and particle phase samples were taken in both chambers prior to and after diesel exhaust injection. For the wood + diesel + R-pinene-O3 SOA mixture experiment, wood combustion aerosol was first injected into both the chambers, and the SOCs were injected into the East chamber. Diesel combustion aerosol was then injected into both chambers, and sufficient ozone was injected to establish a concentration of 0.3 ppm. R-Pinene (0.32 ppm) was then injected into the East chamber in order to produce secondary organic aerosol. The West chamber was used as a tracking chamber. Particle size distributions were measured using a scanning mobility particle sizer (DMA 3936 TSI, Shoreview, MN) coupled with a condensation nuclei counter (3025A, TSI) and an Electrical Aerosol Analyzer (Model 3030, TSI) for the fine particles (
