Article pubs.acs.org/est
Enhancement in Secondary Organic Aerosol Formation in the Presence of Preexisting Organic Particle Jianhuai Ye, Catherine A. Gordon, and Arthur W. H. Chan* Department of Chemical Engineering and Applied Chemistry, University of Toronto, Toronto, Ontario M5S 3E5, Canada S Supporting Information *
ABSTRACT: Atmospheric models of secondary organic aerosol (SOA) typically assume organic species form a well-mixed phase. As a result, partitioning of semivolatile oxidation products into the particle phase to form SOA is thought to be enhanced by preexisting organic particles. In this work, the physicochemical properties that govern such enhancement in SOA yield were examined. SOA yields from α-pinene ozonolysis were measured in the presence of a variety of organic seeds which were chosen based on polarity and phase state at room temperature. Yield enhancement was only observed with seeds of medium polarities (tetraethylene glycol and citric acid). Solid hexadecanol seed was observed to enhance SOA yields only in chamber experiments with longer mixing time scales, suggesting that the mixing process for SOA and hexadecanol may be kinetically limited at shorter time scales. Our observations indicate that, in addition to kinetic limitations, intermolecular interactions also play a significant role in determining SOA yields. Here we propose for the first time to use the Hansen solubility framework to determine aerosol miscibility and predict SOA yield enhancement. These results highlight that current models may overestimate SOA formation, and parametrization of intermolecular forces is needed for accurate predictions of SOA formation. gas phase (in μg/m3), respectively, and Mo is the concentration of total organic aerosol (OA, in μg/m3), which includes both newly formed SOA and preexisting organic particle. When preexisting organic particle mixes with SOA, the particle-phase activity (concentration) of SOA is reduced. According to the Raoult’s Law, the decrease of particle-phase activity suppresses the evaporation of SOA from the condensed phase, therefore lowering its effective saturation concentration, shifting the partitioning equilibrium to the particle phase and enhancing SOA formation. This assumption has important implications on estimating the amount of SOA. For example, future increases in primary organic aerosol (POA) emissions are expected to increase SOA by 10−30%.5,11 A different modeling study has even suggested that reducing anthropogenic POA may be the most effective tool to control biogenic SOA.12 Despite the importance of this assumption, laboratory studies have failed to conclusively demonstrate such yield enhancement by organic seed particles. Song et al. showed that SOA yields from α-pinene ozonolysis were not enhanced in the presence of hydrophobic seeds or some hydrophilic seeds but were enhanced in the presence of citric acid.13,14 Using aerosol mass spectrometer data, Asa-Awuku et al. showed that SOA
1. INTRODUCTION Secondary organic aerosol (SOA) produced from atmospheric oxidation of organic vapors is a major component of ambient particulate matter and plays a critical role in global climate change1 and human health impacts.2 While atmospheric models are making progress in reducing the underestimation of SOA and better predicting spatial variability,3 these models fail to capture other aspects, such as degree of oxidation,4 temporal variability, and vertical profiles.5 This discrepancy suggests that, in addition to underestimating hydrocarbon precursors to SOA formation, the underlying processes for SOA formation are poorly understood, thereby limiting the ability to accurately predict SOA formation. One of the key assumptions in predicting SOA formation is that the oxidation products undergo semivolatile partitioning between the gas and the particle phases.6 SOA formed in laboratory studies is indeed semivolatile: the partitioning is shown to depend on temperature7 and dilution.8 Current modeling frameworks, including the traditional two-product model9 and volatility basis set,10 are based on this semivolatile assumption: A p,i 1 K p,i = = * Ag , i Mo ci (1)
Received: November 9, 2015 Revised: March 9, 2016 Accepted: March 10, 2016
where Kp,i is the partitioning coefficient (in m3/μg) of species i, c*i is the effective saturation concentration (in μg/m3), Ap,i and Ag,i are the mass concentrations of species i in the particle and © XXXX American Chemical Society
A
DOI: 10.1021/acs.est.5b05512 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
All the seed reagents were purchased from Sigma-Aldrich (HPLC or ACS grades). For flow tube experiments, α-pinene/cyclohexane solution was injected continuously into an air stream using a 1000 mL syringe (Hamilton) mounted on a syringe pump (KDS Legato 100). Ozone was produced by passing purified air through an ozone generator (UVP 97006601). Ozone concentrations were at least 5 times higher than α-pinene concentrations in order to consume as much α-pinene as possible. In humid experiments, water vapor was introduced into the flow tube using a custommade humidifier. Relative humidity was maintained at 55−60%. A diffusion dryer was used to dry particles before particle sampling inlet in order to minimize particle phase water. Particle wall loss was corrected by atomizing dry ammonium sulfate particles into the flow tube and comparing the particle concentrations at the inlet and outlet of the system. A 25% and 13% wall loss correction for the particle volume concentration was applied to the flow tube experiments under dry and humid conditions, respectively. Residence time inside the flow tube was maintained at 4 min for all the experiments. Particle volume concentrations were converted to mass concentrations using a density of 1.25 g/cm3. For chamber experiments, a 1 m3 chamber was flushed with zero air until the total particle number concentration was less than 1 #/cm3 before experiment. The chamber was prefilled with ozone at a concentration at least 4 times higher than αpinene concentration. The α-pinene/cyclohexane solution was then gently injected into a Swagelok tee and introduced into the chamber by zero air at a flow rate of ∼10 L/min to initiate SOA formation. A typical chamber experiment lasted 3 h. Particle volume concentrations were corrected for wall loss assuming a first-order loss rate. This rate was calculated from particle volume concentration decay 30 min after α-pinene concentrations fell below detection limits. In all experiments, α-pinene concentration was measured using a gas chromatography−flame ionization detector (GCFID, SRI 8610C) equipped with a Tenax TA trap. Particle size distribution and volume concentration were monitored using a scanning mobility particle sizer (SMPS) assembled in our lab. The SMPS combines a long differential mobility analyzer column (TSI 3081) with flow controls and a condensation particle counter (TSI 3772). Data inversion was performed using custom code written in Igor Pro (Wavemetrics).
mixed well with POA from diesel exhaust but not with diesel fuel and motor oil.15 It is reasonable to expect that gas-particle partitioning of semivolatile organic compounds (SVOCs) depends on both the chemical and physical properties of organic seed particles, but the underlying mechanism by which these properties affect SOA yield enhancement remains poorly understood. In addition to thermodynamic driving forces, partitioning of OA components also depends on the kinetics of mixing. Given the recent observations that most SOA is highly viscous,16−18 one might expect that OA mixing to be kinetically limited under atmospheric time scales. In this work, the hypothesis of enhanced SOA formation in the presence of organic seeds is examined. Laboratory experiments with α-pinene ozonolysis were performed in the presence of a wide variety of organic compounds, including squalane, hexadecanol, tetraethylene glycol (TEG), citric acid, erythritol, and levoglucosan, which were chosen based on their polarity and phase state at room temperature. These seeds cover a wide range of oxygen-to-carbon ratio (O/C) of atmospheric organic aerosol. Squalane has an O/C ratio of 0 and serves as a POA surrogate of vehicular exhaust; levoglucosan and erythritol have an O/C ratio of 1, and they represent biomass burning OA19 and SOA from isoprene photooxidation,20 respectively. α-Pinene was chosen as the SOA precursor because it is a representative biogenic VOC, and its SOA formation has been well studied. The composition of α-pinene SOA has also been predicted using the Master Chemical Mechanism.21 In addition, it has been shown that SOA from α-pinene ozonolysis is formed rapidly after initial oxidation without any further aging process,22,23 making it a simple model system for investigating aerosol mixing without complications from reaction kinetics. In this study, a method using the Hansen Solubility Parameters and the Flory−Huggins Equation to determine organic miscibility is also developed to understand the dependence of SOA yield enhancements and may be used as a partitioning framework for atmospheric models. Using simple, well-known seed compounds, we are able to demonstrate that Hansen solubility framework can be used to predict organic mixing.
2. EXPERIMENTAL SECTION Experiments were conducted in a 10 L quartz flow tube reactor (10.2 cm I.D. × 120 cm L.) and a 1 m3 Teflon chamber (87.6 cm × 87.6 cm × 130.8 cm), as summarized in Table S1. Reactor temperature (22−25 °C) and humidity were monitored using an Omega HX94C RH/T transmitter. αPinene (Sigma-Aldrich, 98%) was prediluted in cyclohexane (Sigma-Aldrich, 99.5%) at a ratio such that the reaction rate between OH and cyclohexane was ∼100 times higher than that between OH and α-pinene. Seed aerosols, including ammonium sulfate, tetraethylene glycol, citric acid, erythritol, and levoglucosan, were generated from a TSI 3076 Aerosol Generator in an aqueous or methanol solution with a concentration of 2−3 mg/mL. Squalane and hexadecanol particles were generated by homogeneous nucleation as described by Kolesar et al.24 Briefly, dry N2 was blown through a preheated liquid squalane or hexadecanol (∼130 °C) and the organic vapor was then cooled to room temperature to condense and form particles. Before entering the flow tube or Teflon chamber, seed aerosols were dried using a custom-made diffusion dryer. Methanol and organic vapors in the seed flow were removed by a custom-made honeycomb charcoal denuder.
3. EXPERIMENTAL RESULTS 3.1. SOA Enhancements in a Flow Tube under Dry Conditions (RH < 20%). Experiments were performed to examine SOA mass yield enhancements of α-pinene ozonolysis in the presence of a variety of organic seeds, and the yield curves for the flow tube experiments are shown in Figure 1. The enhancements were examined by plotting SOA yield (Y, mass of SOA formed per mass of hydrocarbon reacted) as a function of generated SOA concentration.9 The data were fitted to the Volatility Basis Set model10 (eq 2), using 10 μg/m3 and 100 μg/m3 as the two volatility bins of interest, denoted as c1* and c2*, respectively. Y=
∑
aiMo/ci* 1 + Mo/ci*
=
a1Mo/c1* 1 + Mo/c1*
+
a 2Mo/c 2* 1 + Mo/c 2*
(2)
Experiments with ammonium sulfate seed were conducted as a baseline for seeded experiments (Figure 1a). SOA yields with ammonium sulfate seed matched the yield curve generated B
DOI: 10.1021/acs.est.5b05512 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
suggest that organic compounds in the atmosphere do not always form a single phase and uniformly enhance SOA yields. In addition, the presence/absence of interactions cannot be simply explained by hydrophobicity, since yield enhancement was not observed for erythritol (highly polar and water-soluble) and squalane (nonpolar and water-insoluble) seeds. 3.2. SOA Enhancements in a Chamber under Dry Conditions (RH = 2−5%). Chamber experiments were performed in order to investigate SOA enhancements with longer interaction time scale (∼3 h). α-Pinene SOA formation was performed in the presence of ammonium sulfate, hexadecanol, erythritol, or levoglucosan seeds. Since the chamber is a batch reactor, the evolution of SOA can be measured as a function of reacted α-pinene (known as the SOA growth curve)22 and is shown in Figure 2. For each seed,
Figure 2. Generated SOA as a function of α-pinene reacted with different preexisting seeds in chamber. Each SOA growth curve is compared to that with ammonium sulfate seed: (a) no seed, (b) hexadecanol, (c) erythritol, and (d) levoglucosan.
Figure 1. α-Pinene SOA yields as a function of generated SOA amount with different preexisting seeds in the flow tube under dry conditions (
