Environ. Sci. Technol. 2009 43, 8794–8800
Effective Rate Constants and Uptake Coefficients for the Reactions of Organic Molecular Markers (n-Alkanes, Hopanes, and Steranes) in Motor Oil and Diesel Primary Organic Aerosols with Hydroxyl Radicals ANDREW T. LAMBE, MARISSA A. MIRACOLO, CHRISTOPHER J. HENNIGAN, ALLEN L. ROBINSON, AND NEIL M. DONAHUE* Center for Atmospheric Particle Studies, Carnegie Mellon University, Pittsburgh, Pennsylvania
Received June 14, 2009. Revised manuscript received September 22, 2009. Accepted October 8, 2009.
Hydroxyl radical (OH) uptake by organic aerosols, followed by heterogeneous oxidation, happens nearly at the collision frequency. Oxidation complicates the use of organic molecular markers such as hopanes for source apportionment, since receptor models assume markers are stable during transport. We report the oxidation kinetics of organic molecular markers (C25-C32 n-alkanes, hopanes and steranes) in motor oil and primary organic aerosol emitted from a diesel engine at atmospherically relevant conditions inside a smog chamber. A thermal desorption aerosol gas chromatograph/mass spectrometer (TAG) and Aerodyne high resolution time-of-flight aerosol mass spectrometer (HR-ToF-AMS) were used to measure the changes in molecular comosition and bulk primary organic aerosol. From the measured changes in molecular composition, we calculated effective OH rate constants, effective relative rate constants, and effective uptake coefficients for molecular markers. Oxidation rates varied with marker volatility, with more volatile markers being oxidized at rates much faster than could be explained from heterogeneous oxidation. This rapid oxidation can be explained by significant gas-phase OH oxidation that dominates heterogeneous oxidation, resulting in overall oxidation lifetimes of 1 day or less. Based on our results, neglecting oxidation of molecular markers used for source apportionment could introduce significant error, since many common markers such as norhopane appear to be semivolatile under atmospheric conditions.
1. Introduction Substantial progress has been made toward understanding the chemical aging of organic aerosols by hydroxyl radicals via heterogeneous oxidation (1, 2). OH uptake is an efficient process, with 30-100% of collisions between OH radicals and organic surfaces leading to reaction (3-5). Chemical aging can complicate the use of organic molecular markers * Corresponding author e-mail:
[email protected]. 8794
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 23, 2009
for source apportionment, because receptor models such as Chemical Mass Balance and Positive Matrix Factorization assume markers are stable during transport. Research conducted in the Los Angeles basin was interpreted to show that many organic markers were stable over transport time scales characteristic of a large urban area (6). However, it is not clear if these conclusions are appropriate over regional transport time scales (7). To better understand chemical aging of organic aerosols, experiments must replicate atmospheric conditions. Previous work has been done mostly in flow reactors with short residence times (seconds-minutes), very high OH concentrations (109-1011 molecules cm-3) and often on surfaces much simpler than ambient particles. Relatively little research has examined chemical aging of complex aerosols under controlled conditions (8) or in environmental chambers with longer residence times (9, 10). This paper examines the oxidation kinetics of organic molecular markers for motor vehicle emissions in particles with complex composition (primary organic aerosol emitted from a diesel engine and lubricating oil aerosol). Of particular interest is the oxidation of hopanes and steranes, saturated petroleum biomarkers that are widely used as tracers for motor vehicle emissions in receptor models (11, 12). The experiments feature atmospherically relevant OH radical and organic aerosol concentrations; the latter are essential to ensure atmospherically relevant phase partitioning (13). Measuring effective rate constants and uptake coefficients for these compounds under realistic conditions should help determine their utility as markers for source apportionment, and whether chemical aging needs to be explicitly treated in receptor models.
2. Materials and Methods Experiments were conducted in the Carnegie Mellon smog chamber, a 10-m3 Teflon bag (Welch Fluorocarbon) suspended in a light- and temperature-controlled room. Experiments typically lasted about 4 h and were conducted at room temperature and low relative humidity. The chamber is surrounded on all sides by blacklights with peak emission intensity at λ ) 350 nm. Prior to an experiment, the bag was heated and flushed for 24 h with clean air supply by using silica gel, activated carbon, and HEPA filters to remove water, VOCs, and particles. 2.1. Particle Generation and Measurement. Pure motor oil, a 3:1 mixture of diesel fuel and motor oil (“oil/fuel mixture”), and dilute diesel exhaust were used to examine the oxidation kinetics of molecular markers in systems with different levels of complexity. Ultra-low-sulfur diesel fuel was used in this study, which contained 73.5% saturates by volume, 25.7% aromatics, and 0.8% olefins (ASTM). Motor oil has been used as a surrogate for primary organic aerosol matrices, and the addition of diesel fuel better represents actual diesel engine emissions (14). The chamber was initially filled with clean air. In separate experiments, liquid solutions of SAE 10W-30 motor oil with and without diesel fuel were flash vaporized using a resistively heated graphite tip. Diesel exhaust was injected into the smog chamber through a heated inlet from a single-cylinder diesel engine (Yanmar L70AE) connected to a 4.5 kW generator, which used the same oil that was flash vaporized in pure motor oil and oil/fuel mixture experiments. Figure S1 shows input particle size distributions for typical diesel and motor oil aerosols, which were measured with a TSI 3080 scanning mobility particle sizer (SMPS). The aerosol was not cooled or diluted before entering the chamber and mixing with the 10.1021/es901745h CCC: $40.75
2009 American Chemical Society
Published on Web 10/26/2009
conditioned air. The initial volume-weighted mean particle diameter agreed within 5% for the two diesel exhaust experiments. The engine was not operated with any control devices, and differs from modern on-road diesel engines in power generation and emissions composition. The ability to represent fleet-average emissions with a single source is extremely difficult and beyond the scope of this paper. However, the mass spectra of the primary organic aerosol emissions from this engine are very similar to spectra from other diesel engines (15). We monitored bulk condensed-phase chemsitry with an Aerodyne HR-ToF-AMS alternating between V mode (higher sensitivity, lower mass resolution) and W mode (lower sensitivity, higher mass resolution) during the experiments with 5-min time resolution. The temperature of the AMS vaporizer was maintained at T ) 600 °C. We measured the organic-aerosol molecular composition using a thermal desorption aerosol GC/MS (TAG, Aerosol Dynamics Inc.) (16). Online GC/MS analysis with the TAG was performed using an Agilent 5890 GC coupled to an Agilent 5971 MSD, which was operated in selected ion monitoring mode to improve measurement signal-to-noise. More details about the TAG operation are given in Supporting Information. Molecular markers quantified with the TAG included C25-C32 n-alkanes, C27-C32 hopanes, and C27-C29 steranes. TAG was calibrated using liquid standards that were manually injected into the collection cell (17), including authentic standards of n-alkanes (Accustandard DRH-008S-R1), 17a(H)22,29,30-trisnorhopane, 17a(H),21b(H)-30-norhopane, 22R17a(H),21b(H)-30-homohopane (Chiron 0615,27; 1321,29; 1339,31), and 20R-5a(H),14b(H),17b(H)-cholestane (Sigma C8003). Other hopanes/steranes for which we did not have authentic standards were identified using chromatograms by Wei et al. (18). Each TAG sample from the diesel exhaust experiments was spiked with a liquid deuterated standard before analysis to correct for potential changes in thermal extraction efficiency of target analytes.This standard combined custom-prepared solutions of hexadecane-d34, eicosaned42, tetracosane-d50, triacontane-d62, hexatriacontane-d74, palmitic acid-d31, and stearic acid-d35. Examples of chromatograms from a diesel experiment are shown in Figures S2 and S3 in Supporting Information. 2.2. OH Radical Generation. In motor oil experiments, OH radicals were generated by UV photolysis of HONO or H2O2. HONO was synthesized by adding sodium nitrite to sulfuric acid, and was added to the chamber before the experiment to achieve target levels of about 800 ppbv. In experiments with diesel exhaust, photo-oxidation was initiated by exposure to UV light; no additional radical precursors were added. OH concentrations were estimated by monitoring the decay of gas-phase tracers (n -pentane or toluene) introduced to the chamber via a septum injector. The tracers were measured using gas chromatography with flame ionization detection (GC-FID, Perkin-Elmer AutoSystem XL; J&W Scientific DB-624 capillary column, 30 m × 0.530 mm coupled to a preconcentrator (Entech 7100A)) or a commercial proton-transfer reaction mass spectrometer (PTRMS, Ionicon Analytik). The inlet and drift tube temperatures in our PTR-MS were maintained at 80 °C, and the drift voltage was set at 500 V. UV photolysis of HONO generated OH concentrations of about (2-4) × 107 molecules cm-3 for the first hour which decreased to (1-3) × 106 molecules cm-3 for the rest of the experiment. UV photolysis of H2O2 produced relatively constant OH levels of about (2-3) × 106 molecules cm-3. OH levels during diesel experiments were also about (2-3) × 106 molecules cm-3. 2.3. Analysis. 2.3.1. Wall-Loss Correction. Quantifying the oxidation rates requires correcting for wall loss. In each experiment, an initial wall-loss rate was calculated from changes in SMPS
volume measurements before turning UV lights on. We corrected for size-dependent wall loss through scaling by the SMPS mass median diameter at subsequent measurements (19). 2.3.2. OH Oxidation Kinetics. OH kinetics measurements in a multiphase system are intrinsically complex. Here we compare the relative rates of removal for various compounds in the spirit of well-known gas-phase relative kinetics (20). If target compound i and reference compound r are in the particle phase, the relative kinetics expression (eq 1) compares the particle-phase loss kinetics independent of OH uptake (21). If reference compound r is in the gas phase, the expression yields an “effective” mixed-phase rate constant (kpi /kgr )m eff (eq 2) that includes heterogeneous uptake limitations lumped into the gas-phase relative kinetics expression (21, 22). ∂ln(Cip) ∂ln(Crp) ∂ln(Cip) ∂ln(Crg)
)
)
() () kip
krp
kip krg
p
(1)
eff m
(2)
eff
where Cpi and Cpr are the concentrations of two particle-phase compounds competing for a common reactant, and (kip)eff and (krp)eff are the corresponding effective particle-phase reaction rate constants. The effective relative rate constant is not necessarily equal to the actual relative rate constant, since the right-hand side of eq 1 includes a lumped correction factor influenced by spatial variation within the particlephase concentration profiles of species i and r (22). Effective particle-phase rate constants (k pi )eff are presented by transforming eq 2 to relate decay of species i directly to the integrated OH exposure time (calculated from decay in gasphase tracer concentration Crg and elapsed time in the experiment): ∂ln(C ip) ) (k ip)effCOH ∂t
(3)
Since experiments are conducted under atmospherically relevant conditions, kinetic data obtained with eq 3 should provide a direct estimate of atmospheric lifetimes. 2.3.3. OH Uptake. The uptake coefficient (γi) of OH by particle-phase species i is defined as the fraction of collisions between an OH molecule and a particle that result in removal of i, and is given by Smith et al. (5) as: γi )
p 2 k i × Dp × F × NA 3 c¯ × Mi × Fdiff
(4)
Dp is the mean geometric surface-weighted particle diameter (measured from SMPS distributions), F is the particle density (assumed to be 0.8 g cm-3 for a hydrocarbon matrix), NA is Avogadro’s number, cj is the mean speed of OH in the gas phase, and Mi is the molecular weight of species i. Fdiff is a correction factor that accounts for diffusion limitations in the transition regime (23), and is discussed in Supporting Information. By definition, γOH e 1, and the only direct cause for γi > 1 is substantial chain-reaction chemistry in the condensed phase or preferential loss of species i relative to the bulk.
3. Results and Discussion 3.1. Effective Rate Constants in Motor Oil and Diesel Primary Organic Aerosols. Figure 1a shows the wall-losscorrected decay of 17a(H),21b(H)-norhopane measured by TAG in motor oil and diesel primary organic aerosol (diesel VOL. 43, NO. 23, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8795
FIGURE 1. (a) Wall-loss-corrected norhopane decay in motor oil, oil/fuel mixture, and diesel POA. (b) Wall-loss-corrected + AMS C4H+ 9 signal from oil/fuel mixture and C6H13 from diesel POA. (c) Wall-loss-corected AMS (circles) or SMPS (triangles, diamonds) organic aerosol concentration (COA) during oil/fuel mixture and diesel POA experiments. Each trace represents a different experiment. Dashed lines indicates rates taken from the literature: 1Weitkamp et al. (10); 2Sage et al. (24). Photochemical age indicated by top x-axis assumes a constant 24-h average OH concentration of 1 × 106 molecules cm-3. POA) as a function of OH exposure. From eq 3, the effective rate constant of norhopane with OH is the slope of these data. The range of keff values measured in motor oil (with and without added diesel fuel) was (8.4-16) × 10-12 cm3 molec-1 s-1, which is in reasonable agreement with previously reported data measured at low RH (10) (indicated by dashed line in Figure 1a). An effective rate constant of 8.4 × 10-12 cm3 molec-1 s-1 corresponds to an oxidation lifetime of about 1 day at an average OH concentration of 1 × 106 molecules cm-3. In diesel POA, we measured norhopane keff values in the range of (3.6-6.2) × 10-11 cm3 molec-1 s-1, which are 2-4 times faster than in motor oil. Differences in norhopane reactivity in motor oil and diesel POA could potentially arise from uncertainty in calculated OH exposure, different aerosol surface-area-to-volume ratios, or different gas-particle phase partitioning. The largest potential uncertainty in OH exposure arose during HONO experiments with motor oil, because UV photolysis of HONO generates a temporally variable OH concentration profile during an experiment. The uncertainty 8796
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 23, 2009
in the estimated OH exposure in these experiments may be as high as a factor of 2. Particle surface-area-to-volume ratio affects oxidation rates (25) because OH has a short reactodiffusive length and will preferentially react with the surface relative to the bulk. Therefore, reactivity should be higher on particles with larger surface-area-to-volume ratios. The initial surface-area-to-volume ratio of the diesel POA was typically about 50% greater than that of motor oil particles, but this difference was quickly negated by fast condensation of secondary organic aerosol (SOA). If motor oil and diesel particle morphology were differentsfor example, spheres versus fractal aggregatesspotential errors in the geometric surface-area measurements and keff values may have been introduced. However, comparison of AMS and SMPS measurements did not suggest nonspherical motor oil or diesel particle morphology. Initial diesel and motor oil POA concentrations were similar across the set of experiments, and significantly more SOA was formed from diesel POA than motor oil (Figure 1a), resulting in higher COA. This suggests that differences in phase partitioning is not the issue. Another possibility is that norhopane (and other molecular markers) may be more surface-active in diesel POA versus well mixed in the bulk of motor-oil particles. Their effective vapor pressure and collision frequency with OH would therefore be higher in diesel POA, resulting in faster reactivity for surface-active markers than the bulk condensed phase. This hypothesis was examined by comparing reactivity of individual organic species or molecular markers measured with TAG to the oxidation of the bulk condensed-phase hydrocarbon matrix measured with the HR-ToF-AMS. Figure 1b shows the decay of the wall-loss-corrected concentration of reduced AMS fragments during the photo-oxidation of + motor oil (C4H9+) and diesel POA (C6H13 ). These fragments are markers for the bulk hydrocarbon matrix. For about the first half hour in both experiments, decay was slow because SOA formation produced compounds with reduced alkyl fragments, resulting in mass spectral interference. In motor oil, we calculated a bulk keff of (6.1 ( 0.1) × 10-12 cm3 molec-1 s-1 based on wall-loss-corrected decay of C4H9+ after the first hour of oxidation. This value is only about 20% slower than the norhopane keff measured from the same experiment. Given interference from SOA at C4H9+, the loss rate of C4H+9 associated with primary emissions may be higher. This suggests that norhopane in motor oil is well mixed in + the bulk POA. In diesel POA, decay in the C6H13 signal was -11 3 -1 -1 consistent with keff ) 1.6 × 10 cm molec s , which is similar to a previously published diesel POA loss rate of C4H 9+ associated with the primary emissions (24). The decay of the bulk POA markers in the diesel POA is about 60% slower than norhopane, which may indicate a different norhopane mixing state in diesel particles. However, the overall uncertainty is about a factor of 2. In addition, the same caveat holds here as with the oil/fuel mixture: the true keff of the diesel POA is probably higher than the measured value given mass spectral interference from SOA formation. Interestingly, the bulk reactivity of diesel POA was about 2-3 times faster than that of motor oil. Diesel POA is more volatile than motor oil (15), which may increase availability of semivolatile organics for faster gas-phase reaction with OH following evaporation. 3.2. Effective OH Uptake Coefficients. Figure 3d shows effective uptake coefficients for molecular markers in motor oil and diesel POA. In motor oil, measured γeff’s range from 1.2 ( 0.2 (22R-17a(H),21b(H)-bishomohopane) to 9 ( 2 (nhexacosane). Many γeff’s are much greater than 1, including the bulk γeff calculated from the AMS C4H9+ decay (3.5 ( 0.7 in motor oil), and γeff values in diesel POA were even greater (13-40). Possible explanations include secondary chemistry, or evaporation followed by gas-phase oxidation. Hearn et al. (26) measured effective uptake coefficients of 1.7 ( 0.3 of Cl
FIGURE 2. Evolution of unresolved complex mixture (UCM) during a an oil/fuel mixture. Significant fraction of motor oil eluting between C23 and C28 is preferentially and progressively removed relative to less volatile UCM, indicating faster evaporation or gas-phase oxidation by OH. radicals by bis(2-ethylhexyl) sebacate and proposed additional reagent loss through H-atom abstraction via solvated alkoxy (RO) radicals. Even if secondary chemistry is occurring in these experiments, many γeffs are much greater than 2, strongly suggesting secondary chemistry is not the cause. Gas-phase oxidation is the most likely explanation, given the phase partitioning expected for these molecular markers based on their saturation vapor pressures (27). This issue is considered in more detail in the discussion section. 3.3. UCM Evolution in Photooxidation of Motor Oil and Diesel POA. Measurements of the unresolved complex mixture (UCM) provide additional evidence of the important role that volatility or gas-particle partitioning plays in the aerosol oxidation. The UCM represents the majority of the mass of complex organic aerosol mixtures like motor oil and diesel POA, and is thought to contain hundreds of isomers of aliphatic branched and cyclic alkanes that are not individually resolved by GC (11). Figure 2 shows the evolution of the UCM during an oil/fuel mixture experiment. Most of the UCM elutes between n-alkanes with carbon numbers of 23-32. Figure 2 plots chromatograms of the total m/z ) 57 signal measured by TAG at hourly intervals during oxidation; signals were scaled to match the initial UCM signal at the C32 elution time to isolate relative changes in the UCM distribution. This figure shows that a significant fraction of the motor oil UCM eluting between C23 and C28 n-alkanes is preferentially removed. These species are semivolatile; therefore, gas-phase oxidation reduces vapor concentrations causing some of the particle-phase material to evaporate to maintain phase equilibrium (28). Changes in the diesel UCM during oxidation were similar, but less significant after about the first hour than in motor oil. More SOA is formed from diesel exhaust (Figure 1a), which may cause more of the UCM to partition into the particle phase and minimize preferential removal. Since equilibrium partitioning of SOA is thought to take tens of minutes to hours (29), repartitioning of UCM could conceivably occur and result in less removal of semivolatile material. Alternatively some of this SOA may contribute to the UCM signal measured by TAG. 3.4. Effective Condensed-Phase Relative Rate Constants. To calculate condensed-phase relative rate constants, we employ the single-phase relative kinetics formulation
shown in eq 1. A particle-phase relative rate analysis decouples gas-phase oxidant uptake from subsequent reaction (21). It also removes potential systematic biases from uncertainty in wall-loss-correction and bulk oxidant concentration. Figure 3a compares the decay of an n-alkane (n-hexacosane, C26), hopane (22R-17a(H),21b(H)-homohopane, C31), and sterane (20R-5a(H),14b(H),17b(H)-ergostane, C28) relative to norhopane in motor oil and diesel POA experiments. From eq 1, the slope of the regression lines is the effective rate constant for these species relative to norhopane. Results from application of eq 1 to the set of molecular markers quantified in these experiments are shown in panel b and c of Figure 3. In general, relative rate constants decrease with increasing carbon number, suggesting a volatility dependence. This result is consistent with previous work (10). Figure 3b shows a scatter plot of relative rate constants for markers in both diesel POA and the oil/fuel mixture. Even though the absolute oxidation rate in diesel POA was significantly higher than that in the oil/fuel mixture, the relative oxidation rates for specific compounds are very similar. For example, the range of effective condensed-phase relative rate constants in diesel POA is 0.6 ( 0.3 (20S5a(H),14b(H),17b(H)-stigmastane) to 1.4 ( 0.3 (18a(H)22,29,30-trisnorhopane. Figure 3c shows effective relative rate constants measured for markers in the oil/fuel mixture experiment with significant SOA formation. Here, the range of effective relative rate constants is 0.30 ( 0.13 (22R17a(H),21b(H)-bishomohopane) to 1.8 ( 0.5 (n-hexacosane), or a factor of 6 variability which is similar to the variability in diesel POA. Effective relative rate constants were higher for markers with smaller carbon numbers (e.g., 1.8 ( 0.5 and 1.2 ( 0.3 for C26 in motor oil and diesel POA, respectively) and lower for markers with higher carbon numbers (e.g., 0.30 ( 0.13 and 0.67 ( 0.25 for bishomohopane). 3.5. Effects of SOA Formation on Oxidation Kinetics. Figure 1c shows time series of the wall-loss-corrected COA measured in oil/fuel mixture and diesel exhaust experiments; substantial amounts of SOA were formed. Weitkamp et al. (10) examined the possibility of SOA formation affecting OH uptake, and no change was found in oxidation rate from forming R-pinene SOA during oxidation of motor oil aerosol. The oil/fuel mixture and diesel POA are useful systems to further examine this hypothesis. In the oil/fuel mixture experiment, about 50 µg m-3 of SOA was formed, more than doubling the POA. In the diesel POA experiments, nearly 150 and 250 µg m-3 of SOA was formed, increasing initial aerosol mass concentrations by a factor of 3-6 and doubling the surface-area-weighted mean particle diameter from ∼100 to 220 nm during the experiment. Given substantial SOA formation, there is no evidence for markers being “protected” by an SOA coating during substantial condensation of SOA because oxidation rate does not change over the course of the experiment. As proposed by Weitkamp et al. (10), SOA may form a solution with pre-existing POA rather than either coating the particles or intercepting oxidants before they reach the POA compounds. 3.6. Discussion. The most likely explanation for the rapid oxidation shown in Figures 1-3 is that these markers are semivolatile. First, the effective uptake coefficients are much greater than one and the relative rate constants trend with volatility (Figure 3c and d), which is consistent with oxidation occurring in both phases. Second, field measurements indicate that these markers are semivolatile at typical atmospheric conditions. For example, Sihabut et al. (30) quantified sampling artifacts using backup quartz filters for most of the molecular markers discussed here. They measured backup:front filter concentration ratios of 10-13% for C27-C30 hopanes, 50-70% for C23-C24 n-alkanes, 30-50% for C25-C26 n-alkanes, and 10-30% for C27-C28 n-alkanes, VOL. 43, NO. 23, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8797
FIGURE 3. (a) Effective condensed-phase relative rate constants for 22R-17a(H),21b(H)-homohopane (red), 20R5a(H),14b(H),17b(H)-ergostane (blue), and n-hexacosane (green) relative to norhopane in oil/fuel mixture (diamonds) and diesel POA (circles, triangles). Red, blue, and green lines are bounding regression fits to the data. (b) Scatter plot of effective condensed-phase relative rate constants (relative to norhopane) for molecular markers in diesel POA and oil/fuel mixture. (c) Effective relative rate constants as a function of carbon number for molecular markers in oil/fuel mixture. (d) Effective uptake coefficients as a function of carbon number for molecular markers in oil/fuel mixture and diesel POA. Both the effective rate constants in (c) and the effective uptake coefficients in (d) show that semivolatile constituents are oxidized preferentially, consistent with efficient gas-phase removal. indicating these markers are semivolatile. Fraser et al. (27) present similar results for n-alkanes. The importance of gas-phase reactions can be understood by estimating the ratio of particle- to gas-phase reactivity using eqs 2 and 4 and setting γi ) 1: ∂ln(Cia) ∂ln(Crg)
)
() kia
m
krg eff
)
c¯ × Mi 3 × × g D 2kr p × F × NA 6DOH 1+ c¯Dp 6DOH 6DOH 1 + 1.71 + 1.33 c¯Dp c¯Dp
( )
( )
2
(5)
The hard-spheres gas-phase OH rate constant (kgr ) is about 3 × 10-10 cm3 molec-1 s-1, which is essentially equal to a gas-phase uptake coefficient of 1 (every gas-phase collision results in reaction). The particle-phase rate constant of norhopane with OH never exceeds about 1% of the hardspheres gas-phase OH rate constant (see Figure S4). Therefore, even if only a minor fraction of a species exists in the 8798
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 23, 2009
gas phase, gas-phase oxidation coupled with changes in partitioning will dominate oxidation. A comparison of the measured keff and Atkinson structure-activity relationship (SAR)-predicted rate constant (31) can be used to estimate the fraction of the norhopane that exists in the gas phase. The SAR-predicted gas-phase rate constant for norhopane with OH is about 3.9 × 10-11 cm3 molec-1 s-1 ((50% uncertainty). Therefore, within the uncertainty range of the SAR-predicted rate constant, the measured rate constant of 8.4 × 10-12 cm3 molec-1 s-1 in motor oil is consistent with 4% (k ) 6 × 10-11) to 37% (k ) 2 × 10-11) of norhopane existing in the gas phase. Given that COA ) 50 µg m-3, this implies that C* for norhopane is 6-30 µg m-3, which is similar to the estimated C* ) 2 µg m-3 value in SciFinder Scholar (Table S1). In diesel POA, the range of measured keffs for norhopanes(3.6-6.2) × 10-11 cm3 molec-1 s-1sfall within the range predicted by SAR’s for gas-phase reaction with OH. Therefore, we cannot quantitatively explain the measured keffs in diesel POA from predicted equilibrium phase partitioning and predicted rate constants, but the results imply that a much larger fraction of the norhopane exists in the gas phase in the diesel experiments than the
motor oil experiments. Further, in all experiments, norhopane and other semivolatile molecular markers are consumed from both gas-phase and particle-phase reactions with OH. Based on our measured keffs and γeffs, it seems that C* must be at least a factor of 10 lower than COA for heterogeneous OH uptake to compete with gas-phase oxidation. This work has important implications for using organic molecular markers for source apportionment. Molecular markers are assumed to be chemically stable during transport, but effective rate constants measured in this work reveal oxidation that is highly efficient, that is not inhibited by significant SOA formation, and that results in oxidation lifetimes of 1 day or less for semivolatile markers at typical ambient OH levels. Even if only a small fraction of the marker is in the vapor phase (
