Assessment of SAPRC07 with updated isoprene chemistry against outdoor chamber experiments

Atmospheric Environment 105 (2015) 109e120

Contents lists available at ScienceDirect

Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv

Assessment of SAPRC07 with updated isoprene chemistry against outdoor chamber experiments Yuzhi Chen, Kenneth G. Sexton, Roger E. Jerry, Jason D. Surratt, William Vizuete* Department of Environmental Sciences and Engineering, Gillings School of Global Public Health, University of North Carolina at Chapel Hill, 166 Rosenau Hall CB#7431, Chapel Hill, NC 27599, USA

h i g h l i g h t s
We evaluated the updated CMAQ version of SAPRC07 against chamber experiments.
The mechanism shortens NO to NO2 conversion time and increases O3 production.
In lower-NOX conditions the mechanism increases the bias of O3 from 4.9% to 9.4%.
New Chemistry increases the reaction rate of VOCs and enhances radical production.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 18 August 2014 Received in revised form 13 December 2014 Accepted 20 January 2015 Available online 20 January 2015

Isoprene, the most emitted non-methane hydrocarbon, is known to influence ozone (O3) formation in urban areas rich with biogenic emissions. To keep up with the recent advance on isoprene oxidation chemistry including the identification of isoprene epoxydiols (IEPOX) as a precursor to secondary organic aerosol (SOA), Xie et al. (2013) updated the SAPRC (Statewide Air Pollution Research Center)-07 chemical mechanism. It is currently unknown how the Xie modification of SAPRC07 impacts the ability of the model to predict O3. In this study we will evaluate the Xie mechanism with simulations of 24 isoprene experiments from the UNC Dual Gas-phase Chamber. Our results suggest that the new mechanism increases NOx (nitrogen oxides) inter-conversion and produces more O3 than SAPRC07 for all experiments. In lower-NOx experiments, the new mechanism worsens O3 performance in the wrong direction, increasing bias from 4.92% to 9.44%. We found increased NOx recycling from PANs accounts for that. This could be explained by more PANs made due to increased first generation volatile organic compound (VOC) products and hydroxyl radical (OH) production. © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Atmospheric chemical mechanism Isoprene Smog chamber SAPRC07

1. Introduction Isoprene (2-methyl-1, 3-butadiene, C5H8) is the most abundant non-methane hydrocarbon emitted from vegetation (Guenther et al., 2006) and has a significant impact on atmospheric chemistry. Isoprene is known to influence ground-level ozone (O3) formation in urban areas rich with biogenic emissions (Chameides et al., 1988; Li et al., 2007). In recent years new discoveries have been made concerning isoprene oxidation chemistry leading to secondary organic aerosol (SOA) or particulate matter (PM) formation (Claeys et al., 2004; Edney et al., 2005; Hallquist et al., 2009; Kroll et al., 2006; Lin et al., 2013; Paulot et al., 2009; Surratt et al.,

* Corresponding author. E-mail address: [email protected] (W. Vizuete).

2006, 2007, 2010). By combining organic synthesis, computational calculations, smog chamber studies, and field measurements, researchers have recently characterized reactive epoxides that are produced from the photochemical oxidation of isoprene and are significant for SOA formation (Claeys et al., 2004; Edney et al., 2005; Paulot et al., 2009; Surratt et al., 2010). These gas-phase oxidation products include methacrylic acid epoxide (MAE) and isomeric isoprene epoxydiols (IEPOX). From recent work it is clear that anthropogenic pollutants, such as oxides of nitrogen (NOx) and sulfur dioxide (SO2), significantly enhance these isoprene-derived epoxides as a source of PM2.5 (Claeys et al., 2004; Edney et al., 2005; Lin et al., 2013; Paulot et al., 2009; Surratt et al., 2010). This is of great public health and regulatory importance since isoprene is primarily emitted from terrestrial vegetation, and thus, is not controllable, whereas anthropogenic emissions (e.g., NOx, SO2, or pre-existing primary aerosol) are controllable.

http://dx.doi.org/10.1016/j.atmosenv.2015.01.042 1352-2310/© 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

110

Y. Chen et al. / Atmospheric Environment 105 (2015) 109e120

Table 1 Characterization runs e initial concentration. Date/side

Compound injected

Injection (ppm)

Initial NOx (ppm)

Initial NO (ppm)

Initial NO2 (ppm)

OC0984BLUE ST3097RED ST2396RED ST2396BLUE JL1588RED JL1588BLUE AU2497RED AU2497BLUE AU2393BLUE AU1688BLUE AU1092RED AU1092BLUE AU0197RED AU0197BLUE ST3097BLUE AU3093BLUE

HCHO HCHO HCHO HCHO HCHO HCHO ETHENE ETHENE ETHENE ETHENE METHANE METHANE METHANE METHANE CO CO

0.96 2 0.5 1 0.79 0.43 1.92 1.84 0.49 0.97 500 250 500 250 250 100

0.5 0.32 0.33 0.33 0.31 0.3 0.32 0.32 0.33 0.41 0.35 0.35 0.35 0.34 0.33 0.32

0.35 0.27 0.28 0.28 0.22 0.21 0.29 0.29 0.28 0.34 0.28 0.28 0.3 0.29 0.28 0.28

0.15 0.05 0.05 0.05 0.09 0.09 0.03 0.03 0.05 0.07 0.07 0.07 0.05 0.05 0.05 0.04

Note: Date/side trailing code ‘RED’ and ‘BLUE’ denote the side of Dual chamber where the experiment was operated.

The chemical mechanisms in current regulatory models, however, do not have the gas-phase chemistry needed to predict isoprene-based SOA precursors. Isoprene gas-phase oxidation chemistry is currently represented in air quality models (AQMs) in a condensed form. It is designed to represent the chemical formation of O3 while incorporating simplification and approximation for computational efficiency. To evaluate future control strategies, new gas- and particle-phase isoprene chemistry must be incorporated in AQMs to study the importance of isoprene emissions on both ground-level O3 and SOA formation. Xie et al. (2013) have developed a more explicit isoprene chemical mechanism with additional hydroxyl radical (OH) and nitrate radical (NO3) oxidation pathways that produce SOA precursors. The mechanism is based on SAPRC07T (Hutzell et al., 2012), which adds hazardous air pollutants and secondary aerosol precursors to the original SARPC07 (Carter, 2010a,b). In the original

Table 2 Isoprene runs e initial injected species and concentration. Date/Side

ISOP/NOx (ppm/ppm)

Initial ISOP Initial NOx Initial NO (ppm) (ppm) (ppm)

Initial NO2 (ppm)

JN1793BLUE JN1793RED JL1780RED JN2381BLUE JL2381RED JN2381RED JL1780BLUE JN2697BLUE JN2592RED OC1596BLUE OC1596REDa AU0897RED AU1597BLUE JN0298BLUE ST0799BLUE ST1199BLUE JN2996RED AU1196BLUE JN2996BLUE ST2496RED JN2592BLUE JN2697RED OC0697RED ST1199RED

0.18 0.35 0.44 0.58 0.65 1.04 1.11 1.13 1.61 1.78 1.84 1.89 1.9 2.13 2.31 2.47 2.53 2.91 3.12 3.2 3.33 3.73 4.85 9.29

0.1 0.19 0.2 0.26 0.28 0.46 0.52 0.38 0.58 0.58 0.6 1.28 1.58 1.5 1.62 1.5 1.02 0.98 1.26 2.06 1.2 1.29 3.12 1.56

0.07 0.07 0.1 0.13 0.08 0.13 0.11 0.07 0.05 0.04 0.04 0.09 0.06 0.12 0.01 0.01 0.07 0.04 0.07 0.07 0.04 0.07 0.08 0.01

a

0.54 0.54 0.46 0.45 0.43 0.44 0.47 0.34 0.36 0.33 0.33 0.68 0.83 0.7 0.7 0.61 0.4 0.34 0.4 0.64 0.36 0.35 0.64 0.17

200 ppm CO also injected in this experiment.

0.47 0.47 0.36 0.32 0.35 0.31 0.36 0.27 0.31 0.29 0.28 0.59 0.78 0.58 0.69 0.6 0.33 0.29 0.34 0.57 0.32 0.28 0.56 0.16

SAPRC07 mechanism, the reaction of isoprene with OH gives a product mixture including methacrolein (MACR) and methyl vinyl ketone (MVK) as shown below in R.1:

ISOP þ OH/0:986RO2C þ 0:093RO2XC þ 0:093zRNO3 þ 0:907xHO2 þ 0:624xHCHO þ 0:23xMACR þ 0:32xMVK þ 0:357xIPRD þ yR6OOH þ 0:167XC (R.1) In the Xie modification, bulk hydroxyl-peroxy isoprene radicals (ISOPO2) are formed (R.2), which in turn produce MACR and MVK by reactions with other species, namely, NO, HO2, MEO2, RO2C, MECO3 and ISOPO2 itself (see Supplemental material for explanation of undefined model species here and henceforth).

ISOP þ OH/ISOPO2 þ ISOPRXN

(R.2)

In the original SAPRC07 mechanism, reaction with NO3 does not yield either MACR or MVK, but in the Xie modification the product NISOPO2 is created (R.3).

ISOP þ NO3/NISOPO2

(R.3)

NISOPO2, like its non-nitrated analog, reacts with NO3, NO, MEO2, RO2C, and MECO3 to give small yields of MACR and MVK. Other reactions of isoprene, with O3 and Cl radicals, are unchanged in the Xie modification. Isoprene-derived nitrates are treated explicitly. First generation isoprene nitrates include ISOPN (¼ISOPND þ ISOPNB), NIT1, and NISOPOOH. ISOPND and ISOPNB are produced in the Xie mechanism instead of RNO3 in the standard SAPRC07. ISOPN reacts an order of magnitude faster with OH and has almost 100% higher recycling of NO2 than RNO3 does. NIT1 formed in NISOPO2 þ NO/NO3/RO2 reactions reacts with OH and NO3 and forms respective peroxy radical species, which react with other radicals similarly as the initially formed peroxy radical species (ISOPO2/NISOPO2) to yield little NOx and other products. Xie et al. (2013) also assumes 70% NOx recycling efficiency from NIT1 þ O3 oxidation. NISOPOOH is the product of NISOPO2 þ HO2 reaction. The OH oxidation of the above first-generation nitrates forms secondary isoprene nitrates including: short-lived MVKN, MACRN, ETHLN, RNO3I, and longer-lived PROPNN. These products can either react with OH or photolyze to give NO2 back. In the Xie mechanism, isoprene oxidation chemistry under low-NOx conditions was updated. These updates include IEPOX formation from ISOPO2 þ HO2 channel (R.4) and HPALD formation from ISOPO2 isomerization (R.5). Subsequent reactions to both channels produces HOx (OH and HO2).

ISOPOOH þ OH/IEPOX þ OH

(R.4)

ISOPO2/HO2 þ HPALD

(R.5)

Xie et al. (2013) integrated the updated SAPRC07T chemical mechanism into the CMAQ model version 4.7 and simulated from 1 July to 16 August 2004 across the entire continental US and a portion of Canada and Mexico. These additions to the CMAQ model allow for explicit predictions of OH reformation from isoprene peroxy radicals, NO2 recycling from isoprene nitrates and IEPOXSOA tracers (and thus total SOA mass from isoprene oxidation). Current AQMs estimate that isoprene contributes 27% (Hoyle et al., 2007) ~48% (Henze and Seinfeld, 2006) to the global burden of SOA (Carlton et al., 2009), yet under predict summertime isoprene SOA especially in areas like southeastern U.S. (Foley et al., 2010). They found that, compared to the base case simulation with SAPRC07T, their updated mechanism improves the simulation of aircraft measurement for gas-phase compounds including NOx (bias

Y. Chen et al. / Atmospheric Environment 105 (2015) 109e120

111

Fig. 1. Ozone peak (a) under low isoprene:NOx ratio and (b) under high ISOP:NOx ratio. Note: Each point (circle or cross) represents the crossover time of an experiment, with the predicted value on the y axis and measured value on the x axis. The diagonal line suggests perfect modeling agreement with the observation. Red Circle stands for ozone concentration simulated by SAPRC07 and blue cross stands for predictions by the Xie mechanism. The area enclosed by two dashed lines at ±25% is the acceptable range of bias. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

from 7% to 1%), O3 (increases 1e2 ppbv with bias still within ±5%), HCHO (bias from 12% to 9%), and isoprene (bias from 26% to 4%). The researchers also reported biogenic SOA increased by 15% compared to the base case. The CMAQ results reported by Xie et al. (2013) demonstrated improved model performance for several species, but large uncertainties still lie in the new gas-phase chemistry. These uncertainties include isoprene nitrates yield from isoprene þ OH/NO pathway and NOx recycling efficiency from first-generation nitrates. Further, it is difficult to evaluate a chemical mechanism in AQMs, where other processes like transport, deposition and emissions act synergistically. The compensating errors from those processes might result in good agreement between observations and predictions and thus veil the real problems within the mechanism. To resolve this problem, smog chamber experiments are traditionally used to test and refine a new mechanism, or evaluate an existing mechanism (Azzi et al., 2010). The smog chamber is a closed and controlled system allowing for chemistry to be the main process that influences concentration. Thus, the discrepancy or agreement between observations and predictions are directly correlated with the mechanism being used. Condensed gas-phase mechanisms are finely tuned engineering approximations for atmospheric chemistry. Xie et al. (2013) have added new reactions and species to the base mechanism and consequently have altered the radical budgets and nitrogen cycling. Xie mechanism has yet to be evaluated against smog chamber experiments before being widely used in AQMs. Thus, it is currently unknown how the Xie modification of SAPRC07T impacts the ability of the model to predict O3, isoprene decay, and its oxidation products. In this paper we will evaluate the Xie mechanism with simulations of experiments from the UNC outdoor smog chamber. A rich archive of chamber experiment data (Jeffries et al., 2013) provides this study with reliable measurements from 24 experiments conducted with isoprene and NOx. These experiments were carried out for isoprene to NOx concentration ratios (ISOP:NOx) ranging from 0.18 to 9.29 in ppm/ppm, with initial NOx concentrations from

0.17 ppm to 0.83 ppm. The focus will be on changes in model predictions from the SAPRC07 mechanism that may have been introduced by Xie et al. (2013) updates and evaluating the mechanism's ability to predict O3. 2. Methods 2.1. Experimental Overall, 40 experiments (16 characterization runs and 24 isoprene runs) were conducted in the UNC Dual Gas-phase Chamber (Pittsboro, NC), where the real-time concentration of NOx, VOCs and O3 can be measured accurately. Environmental parameters including temperature, relative humidity, and light intensity were also monitored to compute chemical and photochemical reaction rates. Detailed descriptions of the chamber and measurement instruments employed can be found elsewhere (Jeffries et al., 1976). Sixteen characterization runs were chosen to evaluate the light model and wall chemistry parameters represented in the auxiliary mechanism. Explanation of the auxiliary mechanism developed by Jeffries et al. (2002) for the UNC smog chamber is available elsewhere (Jeffries et al., 2002; Parikh et al., 2013). Species in these runs include carbon monoxide (CO), methane (CH4), ethene (C2H4) and formaldehyde (HCHO). These are explicit species in the chemical mechanism whose kinetic information is well constrained. The initial concentrations of these runs are outlined in Table 1.

Table 3 Summary of O3 peak model performance statistics. Experiment

Condition

Mechanism

NMB (%)

R2

P value

N¼8

High NOx

N ¼ 16

Low NOx

SAPRC07 Xie SAPRC07 Xie

9.80 2.00 3.42 9.44

0.13 0.08 0.85 0.79

0.38 0.32 3.09E-07 4.00E-06

112

Y. Chen et al. / Atmospheric Environment 105 (2015) 109e120

Fig. 2. NOeNO2 Crossover time. Note: Each point (circle or cross) represents the crossover time of an experiment, with the predicted value on the y axis and measured value on the x axis. Circle stands for crossover time simulated by SAPRC07 and cross stands for times predicted by Xie mechanism. The diagonal line suggests perfect modeling agreement with the observation. Red indicates the experiments with low initial isoprene:NOx and blue for high isoporene:NOx. Note that the time shown here is the crossover time relative to the start of the experiment. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Twenty-four isoprene photooxidation experiments were selected and shown in Table 2. In this study, runs in which the isoprene:NOx concentrations have a ratio less than 1.25 (ppm/ppm) were arbitrarily classified as having a low VOC:NOx ratio or a highNOx experiment, while those over 1.25 were considered to have a high VOC:NOx ratio or a lower-NOx experiment. Note that the term high- or lower-NOx in this manuscript is an indication of the relative initial NOx level to initial isoprene of an experiment. The absolute amounts of NOx injections are larger than ambient level. 2.2. Modeling Two mechanisms compared in this study, standard SAPRC07 (Carter, 2010a,b) and the Xie mechanism, were implemented in the UNC Morpho Photochemical Reaction Simulation System (Jeffries and Singh, 1995). SAPRC07 source code was created based on Dr. Carter's report (Carter, 2010a,b). The Xie mechanism was contained in the CMAQ files provided by Dr. Ying Xie, and converted to the

Morpho format (Jeffries et al., 2002). For both mechanisms, we use the rate proposed by Mollner et al. (2010) for this reaction: NO2 þ OH / HNO3. The UNC Auxiliary Mechanism (version-aadg) was used to account for chamber-dependent wall mediated effects; specifically these reactions are provided in the Supplementary material. In fact, same chamber-dependent reactions were used in a previously published research (Parikh et al., 2013). Simulation results of 16 characterization runs used in this study also have satisfactory modeling performance. Based on this evidence we are confident in the representation of our auxiliary mechanism. In addition to the auxiliary mechanism, however, each experiment has unique injection values of initial loadings of wall water, HONO and wall HNO3. These are dependent on the amount of time the chamber has been vented with dry air before the experiment and recent history of experiments conducted in the chamber before that experiment.

Table 5 Selected study cases initial condition and O3 peak concentration.

Table 4 Summary of crossover time statistics. Experiment

Condition

Mechanism

NMB (%)

R2

P value

N¼8

High NOx

N ¼ 16

Low NOx

SAPRC07 Xie SAPRC07 Xie

5.45 14.94 22.12 29.73

0.02 0.03 0.76 0.67

0.73 0.68 1.06E-05 9.62E-05

Experiment Initial ISOP:NOX (ppm/ppm)

Initial isoprene (ppm)

Initial NOX O3 peak (ppm) (ppm)

O3 max (ppm)

JN2381BLUE 0.58 JN2697RED 3.73

0.26 1.29

0.45 0.35

0.85 0.75

0.85 0.57

Note: Here only the measured values of the first ozone peak are shown.

Y. Chen et al. / Atmospheric Environment 105 (2015) 109e120

113

Fig. 3. Ozone and NOx concentration time profile for: (a) JN2381BLUE (High NOx) and (b) JN2697RED (Lower NOx). Note: Observation values are plotted in black with measured data points. Simulation by SAPRC07 is plotted in red and Xie mechanism in blue. “Obs” stands for observation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3. Results and discussion 3.1. Model performance Simulations with SAPRC07 and the Xie mechanism were statistically compared with observational data in terms of peak O3 concentration and NOeNO2 crossover time derived from temporal concentration profiles of NO, NO2 and O3. The NOeNO2 crossover time implies how fast a mechanism is converting NO to NO2 and propagating the autocatalytic process of O3 production. Ozone peak/maximum value is a direct indicator of regulatory interest as reflected by 8-h ozone standard in NAAQS. Both indicators are reported in the form of normalized mean bias (NMB) computed using Eqn. (1).

NMB ¼

1 X simulation  observation  100% n observation

(1)

3.1.1. O3 peak Simulated O3 peak concentration values were plotted against observational values shown in Fig. 1. For chamber VOC(s) þ NOx daytime experiments, the observed pattern of O3 temporal profile

changes with relative NOx abundance. A lower-NOx experiment is often characteristic of two O3 peaks, of which the second peak is due to photolysis of NO2 recycled back from reservoir species like peroxy acyl nitrates (PAN). Here we only show the graphical results of the first/morning O3 peak due to direct ozone photochemistry (Fig. 1). Under both conditions the Xie O3 peak was consistently higher than the peak predicted with SAPRC07 by as much as 20%. The difference was most pronounced under high-NOx conditions (Fig. 1a). Both mechanisms show consistent over-prediction under lower-NOx conditions as shown in Table 3. It is important to note that under lower-NOx conditions, the Xie mechanism is pushing model performance in the wrong direction increasing bias of first ozone peak from 4.92 to 9.44%; bias of second peak from 1.36% to 20.51%. Although there is insufficient number of runs in our highNOx experiments to conclude a statistically significant difference in the two mechanisms, it is clear that Xie et al.'s modifications made to SAPRC07 have resulted in an elevated magnitude of peak O3. 3.1.2. NOeNO2 crossover Fig. 2 shows the observed NOeNO2 crossover time versus simulated results across 8 high-NOx experiments and 16 lower-NOx experiments. We found that both mechanisms under-predict NOeNO2 crossover time for lower-NOx experiments. The Xie mechanism always had an earlier crossover time than SAPRC07 and

Fig. 4. Isoprene concentration time profile for: (a) JN2381BLUE (High NOx) and (b) JN2697RED (Lower NOx). Note: Observation values are plotted in black. Simulation by SAPRC07 is plotted in red and the Xie mechanism in blue. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Y. Chen et al. / Atmospheric Environment 105 (2015) 109e120

was even earlier for lower-NOx experiments. Overall, SAPRC07 over-predicts for 3 out of 24 experiments while Xie only overpredicts 1 experiment. Detailed statistical results are summarized in Table 4.

k[BE06] k[BR07] k[BR07] k[BE06] k[BR07]

Rate

114

Note: Rate constant IDs are provided here and their values can be found in the original report of SAPRC07 mechanism (Carter, 2010a,b).

ISOP þ OH / 0.986*RO2C þ 0.093*RO2XC þ 0.093*zRNO3 þ 0.907*xHO2 þ 0.624*xHCHO þ 0.23*xMACR þ 0.32*xMVK þ 0.357*xIPRD þ yR6OOH þ 0.167*XC xHO2 þ NO / HO2 þ NO RO2C þ NO / NO2 ISOP þ OH / ISOPO2 þ ISOPRXN ISOPO2 þ NO / 0.40*MVK þ 0.26*MACR þ 0.883*NO2 þ 0.07*ISOPND þ 0.047*ISOPNB þ 0.66*HCHO þ 0.10*HC5 þ 0.043*ARO2 þ 0.08*DIBOO þ 0.803*HO2 SAPRC07

Xie

Reactions Mechanism

Table 6 The initial steps of isoprene oxidation by OH for the SAPRC07 and Xie mechanisms.

3.2. Process analysis In the following analysis, our focus shifts from model-toobservation comparison to mechanism inter-comparison under different initial isoprene:NOx ratios. Chemical reaction process analysis is applied in this step using PERMM (Python-based Environment for Reaction Mechanisms/Mathematics) (Henderson et al., 2009). Process analysis is based on the concept of integrated reaction rate (IRR) analysis (Jeffries and Singh, 1995; Crouse, 1990). The IRR over each time step of each reaction in the mechanism is output to .irrt file, and the IRR over the course of simulation and its corresponding reaction is output to .irrmg file by MORPHO. PERMM reads in these MORPHO outputs and provides an interface to obtain net reactions, and to quantify radicals and NOx budgets. To account for the pattern observed in model performance assessment, we conducted reaction process analysis on two selected cases e one high-NOx case and one lower-NOx case (Table 5). The difference between SAPRC07 and the Xie mechanism is reported in the form of relative difference (RD) using Eqn. (2). In this case, a positive RD value would suggest the Xie mechanism has a higher value, vice versa.

RD ¼

Xie  SAPRC07  100% SAPRC07

(2)

3.2.1. Case study The concentration profiles of O3, NOx, and isoprene are displayed in Figs. 3 and 4. Ozone temporal concentration profiles are characteristic of distinctive observed patterns under different initial VOC(s) and NOx concentration levels. For a typical high-NOx (low VOC:NOx ratio) chamber experiment (Fig. 3a), the system runs out of NO late in the day compared with the lower-NOx (high VOC:NOx ratio) experiment (Fig. 3b) and gradually makes ozone until 5 pm for our high-NOx experiment JN2381BLUE. In comparison, a typical lower-NOx experiment has a shorter-lasting photochemical chain of making O3. Taking our lower-NOx experiment JN2697RED as an example, the second O3 peak (750 ppb) occurs at 2 pm due to NO2 released from its reservoir species NOz. For JN2381BLUE (high-NOx), both mechanisms over-predict the NOeNO2 crossover time and under-predict maximum O3 concentration. Neither mechanism is able to predict the actual O3 peak around 5 pm but the Xie mechanism is closer to the measurement. For JN2697RED (lower-NOx), both under-predict the NOeNO2 crossover time and the morning O3 peak to a similar extent. The Xie mechanism, however, produces O3 faster in the afternoon increasing final O3 concentrations by 210 and 350 ppb compared to the observation and SAPRC07. Regardless of the difference between their abilities to reproduce the observation, the Xie mechanism predicts an earlier crossover time and a higher O3 peak concentration, which is visually represented in Fig. 3. Table 6 lists the reactions of OH with isoprene and its resulting products. Fig. 4 shows the isoprene decay for the two experiments with the Xie mechanism oxidizing more isoprene faster. The loss of isoprene due to the reaction with OH is consistent with the OH concentration levels shown in Fig. 5. Because of that, the first generated ISOPO2 in the Xie mechanism dominates over RO2C (peroxy radical operator in SAPRC07) in NO-to-NO2 conversion, and produces more formaldehyde and HO2. There was up to a 1.5-fold increase in HO2

Y. Chen et al. / Atmospheric Environment 105 (2015) 109e120

115

Fig. 5. OH concentration time profile for: (a) JN2381BLUE (High NOx) and (b) JN2697RED (Lower NOx). Note: Simulation by SAPRC07 is plotted in red and the Xie mechanism in blue. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 6. HO2 production rate from aldehydes time series for: (a) JN2381BLUE (High NOx) and (b) JN2697RED (Lower NOx). Note: For SAPRC07 “aldehydes” include formaldehyde (HCHO), acetaldehyde (CCHO) and lumped aldehyde (RCHO). In addition to those three species in SAPRC07, another two are added in the Xie mechanism: glycoaldehyde (HOCCHO) and hydroxyl-peroxy aldehyde (HPALD).

production rate from aldehydes photolyses, most of which is from formaldehyde (Fig. 6). HO2 from aldehydes converts NO to NO2 and produces OH radicals that participate in a new round of VOC oxidation and propagation reactions, thus amplifying the entire

chain. Accordingly, more OH is predicted by the Xie mechanism as shown in Fig. 5. Increased OH production permits the Xie mechanism to continuously oxidize VOCs until the end of the experiment (Fig. 7).

Fig. 7. VOCs and isoprene reaction rate against OH time series for: (a) JN2381BLUE (High NOx) and (b) JN2697RED (Lower NOx). Note: Total VOCs reaction rate is plotted in solid steps and isoprene reaction rate is plotted in dashed steps. Curves of SAPRC07 are drawn in red and those of the Xie mechanism are drawn in blue. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

116

Y. Chen et al. / Atmospheric Environment 105 (2015) 109e120

Fig. 8. Detailed comparison of mass flowing through various VOC þ OH reaction pathways for JN2381BLUE (High NOx). Note: The integrated reaction rate of each VOC with OH is plotted in separate bar. The upper plot contains the VOCs that two mechanisms share and have actual mass flow in the simulations by both mechanisms; the bottom plot contains the VOCs that have mass flow only in the simulations by Xie mechanism. Right axes are for species right to the vertical dashed lines.

Integrating reaction rate of VOCs þ OH (solid lines in Fig. 7) over the entire courses of simulations proves that the Xie mechanism has greater amount of IRR for VOCs þ OH reactions due to both the introduction of new species and more existing species reacted. A complete list of VOCs reacting with OH are represented in their own identities in Figs. 8 and 9. Note that there is a mass shift within certain species because of changed pathways in Xie mechanism. For example, in the Xie mechanism additional explicit species are isolated from the generalized RNO3 group. These new species include: ISOPND, ISOPNB, NIT1, NISOP2OOH, MACRN, MVKN, and ETHLN, RNO3I. The reaction rates of some of those species with OH is 2e6 times greater than that of RNO3, providing competition for available OH radicals. A summary of the OH budget is tabulated in Table 7. OH cycle number indicates the efficiency of the system in using OH radicals. In the simulations by the Xie mechanism, OH cycle is about 7% different in the high-NOx case, but 32% different in the lower-NOx case. In the high-NOx case, the Xie mechanism was less efficient in use of OH even though it produced 1.4 times the amount of new OH (Table 7). The increased OH concentration in this experiment, coupled with higher NO2 concentrations, increased the competition for OH through termination reaction (R.6) instead of through VOCs þ OH reactions. For the lower-NOx case, this termination pathway is not as significant so the difference in OH cycle between the two mechanisms is less.

NO2 þ OH/HNO3

(R.6)

The photolysis of NO2 is the dominant pathway to make O3 in the troposphere. Table 8 shows that for the high-NOx case

JN2381BLUE the reactions of NO with XO2 (HO2 and RO2C) and NOz account for 41% of the NO2 formation in the Xie mechanism compared to 31% in SAPRC07. Around half of NO2 comes from reaction of NO with O3 in the Xie mechanism, compared to 65% in SAPRC07. Therefore, the Xie mechanism is able to produce more O3. For the lower-NOx case, there is 57% difference between the Xie mechanism and SAPRC07 in terms of the second O3 peak concentration. Table 8 also shows that about 74% more NO2 is recycled from NOz in the Xie mechanism than in SAPRC07, causing this increase in ozone production. Table 9 shows the NO cycles for both experiments. In high-NOx experiment, enhanced nitrogen termination reactions result in a 11% decrease in the NO cycle. In the lower-NOx experiment, Xie et al.'s reconstruction of the pathways, which recycle NO2 from NOz species, results in an increasing rate of NO2 production in the afternoon (Fig. 11b), and therefore a higher NO cycle. In this experiment, the NO cycle difference between two mechanisms is 32%. This explains the sustained O3 production, resulting in increased O3 concentrations shown in Fig. 3b. The consistency remains for experiments having higher VOC:NOx ratios, in which the second peak differences between two mechanisms are even larger. We found that the majority of increase in NO2 recycling comes from PANs (Fig. 12). The Xie mechanism predicts 65% more NO2 from PANs (PAN, MAPAN and PAN2) than SAPRC07 and that accounts for 85% of the total increase in NO2 recycling. This is due to increased first-generation VOC products and NO2 concentration, thus more PANs are being made. The Xie mechanism has changed the radical balance of SAPRC07 and directed more NOx to its temporary reservoir species instead of terminating it through nitrogen

Y. Chen et al. / Atmospheric Environment 105 (2015) 109e120

117

Fig. 9. Detailed comparison of mass flowing through various VOC þ OH reaction pathways for JN2697RED (lower NOx).

loss (XN) (Fig. 10b). Fig. 13 shows the percent difference between the two mechanisms of the OH and NO cycle for the entire experimental set. In high-NOx experiments, OH cycle is less concentrated around the mean than lower-NOx experiments. When the experimental conditions are at lower-NOx concentrations, the Xie mechanism tends to have a higher OH cycle. A similar trend is observed in NO cycle difference (Fig. 13b). The Xie mechanism has the ability to produce more OH attack on VOCs and convert more NO to NO2 than the SAPRC07 mechanism. Under high-NOx conditions, this results in higher NO2 and OH concentrations and thus increases termination

reactions. Although more NO2 and ultimately O3 are produced, the OH and NO cycle are lowered. For lower-NOx experiments, the termination reactions are not as large and the recycled NO2 late in the day accounts for the differences in the cycle numbers with the SAPRC07 mechanism. 3.3. Sensitivity runs Model performance results under lower-NOx conditions prompted several sensitivity runs exploring low NOx chemistry and NO2 recycling. Again, the lower-NOx experiment selected for

Fig. 10. Nitrogen loss rate time series through deposition of HNO3 and XN (SAPRC07 model species for nitrogen loss): (a) JN2381BLUE (High NOx) and (b) JN2697RED (Lower NOx).

118

Y. Chen et al. / Atmospheric Environment 105 (2015) 109e120

Table 7 OH radical budget for: (Upper Row) JN2381BLUE (High NOx) and (Lower Row) JN2697RED (Lower NOx) experiments. Experiment

JN2381BLUE JN2697RED

Mechanism

SAPRC07 Xie SAPRC07 Xie

New OH (ppb) Inorganic

Organic

Total

22.6 27.2 45.5 66.0

56.2 80.8 476 538

78.8 108 522 604

OH þ VOC (ppb)

OH þ NO2 (ppb)

OH cycle (unitless)

532 592 1382 1749

141 195 13.9 29.9

8.9 8.3 2.5 3.3

Note: OH cycle number ¼ amount of OH reacted/new OH.

Table 8 NO2 production IRR (ppb) from NO conversion and recycling from NOZ: (Upper Row) JN2381BLUE (High NOx) and (Lower Row) JN2697RED (Lower NOx). Experiment

JN2381BLUE JN2697RED

Mechanism

SAPRC07 Xie SAPRC07 Xie

NO / NO2 (ppb) NO þ XO2

NO þ NO3

NO þ O3

Total

1191 1357 2138 2776

2884 3257 93 335

27,575 23,866 4537 5972

30,208 26,851 6721 8915

NOz / NO2 (ppb)

Total NO2 (ppb)

12,061 16,678 9686 16,860

42,269 43,529 16,407 25,775

Note: NO to NO2 conversion includes conversion by XO2 (¼HO2 þ RO2), NO3 and O3.

Table 9 NO cycle calculation specifics: (Upper Row) JN2381BLUE (High NOx) and (Lower Row) JN2697RED (Lower NOx). Experiment

JN2381BLUE JN2697RED

Mechanism

SAPRC07 Xie SAPRC07 Xie

New NO (ppb) Injection

HONO

WHNO3

Total

445 445 342 342

8 8 1 1

4 4 0 0

457 457 343 343

NO / NO2 (ppb)

NO cycle (unitless)

30,215 26,886 6723 8955

66.2 58.9 19.8 26.1

Note: NO cycle number ¼ amount of NO to NO2 conversion/new NO.

process analysis, JN2697RED, will be used as our testing case. Recent chamber experimental work by Fuchs et al. (2013) confirmed OH regeneration from HPALD produced in ISOPO2 unimolecular reactions (1,6-H shift) but the rate constant (k1,6isom ¼ 8.5  108exp(5930/T) cm3/s) in the Leuven isoprene mechanism (LIM) is too large (Peeters and Müller, 2010). They evaluated the values of k1,6isom against isoprene photo-oxidation experiments and proposed a rate of 6.20  108exp(7700/T) cm3/s that was able to reproduce observed OH (Fuchs et al., 2013). However, their experiments were not able to give a more accurate quantification of k1,6isom because of uncertainties in the rate coefficient value for 1,5-H shift isomerization of ISOPO2 and

additional recycling of unknown peroxy radicals. Xie et al. (2013) used 4.07  108exp(7694/T) cm3/s (Crounse et al., 2011; da Silva et al., 2010). Here we simulated with a value of 2.35  108exp(7694/T) cm3/s. Our second sensitivity analysis will focus on changing the isoprene nitrates (ISOPN) yield (stoichiometry) from ISOPO2 þ OH/NO pathway. The original value used in Xie et al.'s paper (2013) is 12%. The value in the CMAQ files Dr. Xie provided was reduced to 6% and we used this value in our simulations so far. Here we zeroed it out to evaluate its effect on NO2 recycling of the lower-NOx case. In Run A, halving ISOPO2 isomerization rate constant has no impact on NOeNO2 crossover time and O3 maximum concentration

Fig. 11. NO2 recycling rate time series from oxidized form of NO2: (a) JN2381BLUE (High NOx) and (b) JN2697RED (Lower NOx).

Y. Chen et al. / Atmospheric Environment 105 (2015) 109e120

119

performance at lower-NOx experiments showed limited influence by ISOPN yield from ISOPO2 þ OH/NO pathway. A shut-off of ISOPN yield merely resulted in a 4.5% decrease in ozone maximum concentrations. This confirms that changes in NO2 recycling efficiency should be attributed to increased PANs production, which is a result of increased initial VOC(s) þ OH reactions and NO2 production. The Xie mechanism adds important chemistry that is conceptually correct and lends itself for SOA simulations. The model performance described in this study indicates a same direction of changes in concentrations of ozone and isoprene as demonstrated by Xie et al.'s CMAQ simulation results. This also applies to other gas-phase species such as MACR and MVK. Compared to SAPRC07, Xie mechanism predicts a greater maximum and a faster degradation of both MACR and MVK. Statistical analysis for these two species was not available but maxima for each experiment are provided in Table S1. It is likely the result of a combination of incorrect kinetic rate constants and missing chemistry. For Fig. 12. NO2 recycling rate time series from PANs for JN2697RED (Lower NOx).

Fig. 13. Radical cycle relative difference with regard to initial isoprene:NOx ratios: (a) OH cycle (b) NO cycle. Note: Each dots stands for an experiment. The vertical dotted lines separate the high NOx experiments on the left and low NOx experiments on the right. The black dashed trendlines are for visual aids only. A positive value means the Xie mechanism has a higher cycle number, vice versa.

(1.0% less). In Run B, the shut-off of ISOPN yield from ISOPO2 þ OH/ NO pathway reduced O3 maximum by only 4.5%. This confirms the key role of PANs in NO2 recycling efficiency changes under lowerNOx conditions. 4. Conclusion We evaluated against outdoor chamber experiments, the CMAQ version of SAPRC07 mechanism, which is a combination of improved isoprene oxidation pathways by Xie et al. (2013) and base mechanism SAPRC07T by Hutzell et al. (2012). Our model performance results suggest that the Xie mechanism produces more O3 and predicts an earlier NOeNO2 crossover time than SAPRC07 for all experiments. Under lower-NOx conditions, both mechanisms over-predict O3 observations; the Xie mechanism worsens performance and increases the bias of O3 from 4.92% to 9.44%. Overall, the Xie mechanism reacts more VOCs due to a more explicit representation of isoprene oxidation products and therefore increases subsequent OH formation. The increased reaction rate of VOCs results in more NO to NO2 conversions by peroxy radicals and more production of aldehyde. The Xie mechanism also increases NO2 recycling from NOz species, which accounts for the increase in the afternoon O3 peak concentrations for lower-NOx experiments. The increase in NO2 recycling from PANs accounts for 85% of the total increase in NO2 recycling. Attempts to improve ozone model

example, although Xie et al. separated isoprene proxy radical (ISOPO2) from lumped RO2 and added pathways for it under different NOx levels, the rate of its reaction with NO is the same as that of RO2C þ NO (bold texts in Table 6). This could be problematic as this reaction is important early in the day and therefore kicks off a stronger propagation and positive radical feedback. If ozone, isoprene, and other gas-phase species cannot be simulated correctly in our chambers, then the simulations of SOA could also be problematic. This suggests a continued focus on the productions of radical and SOA precursor species to improve the ability of the mechanism to simulate chemistry in air quality models. The larger than ambient initial amounts of NOx in our experiments allows for all day oxidation. Our results, however, may not be applicable for ambient conditions at low NOx concentrations. Chamber experiments at even lower NOx concentrations are needed to obtain a complete understanding of the performance of the Xie mechanism under ambient-relevant conditions. Acknowledgements The preparation of this report was financed through a grant from the Texas Commission on Environmental Quality (TCEQ), administered by the University of Texas through the Air Quality Research Program (Project #14-003). The contents, findings, opinions and conclusions are the work of the author(s) and do not

120

Y. Chen et al. / Atmospheric Environment 105 (2015) 109e120

necessarily represent findings, opinions or conclusions of the TCEQ. We thank Dr. Ying Xie and Deborah Luecken at EPA for providing the CMAQ source code of the mechanism under evaluation and corresponding CSQY file. We are also grateful to Blaine Heffron for technical assistance in modeling, and our former colleague Dr. Haofei Zhang, who provided insight and expertise that greatly assisted this research. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.atmosenv.2015.01.042. References Azzi, M., White, S.J., Angove, D.E., Jamie, I.M., Kaduewela, A., 2010. Evaluation of the SAPRC-07 mechanism against CSIRO smog chamber data. Atmos. Environ. 44 (14), 1707e1713. http://dx.doi.org/10.1016/j.atmosenv.2010.02.013. http:// linkinghub.elsevier.com/retrieve/pii/S1352231010001196. Carlton, A.G., Wiedinmyer, C., Kroll, J.H., 2009. A review of Secondary Organic Aerosol (SOA) formation from isoprene. Atmos. Chem. Phys. Discuss. 9 (2), 8261e8305. http://dx.doi.org/10.5194/acpd-9-8261-2009. http://www.atmoschem-phys-discuss.net/9/8261/2009/. Carter, W.P.L., 2010a. Development of a condensed SAPRC-07 chemical mechanism. Atmos. Environ. 44 (40), 5336e5345. http://dx.doi.org/10.1016/j.atmosenv.2010.01.024. http://www.sciencedirect.com/science/article/pii/ S1352231010000622. Carter, W.P.L., 27 January, 2010b. Development of the SAPRC-07 Chemical Mechanism and Updated Ozone Reactivity Scales. Report to the California Air Resources Board, Contract No. 03-18, 06-408, and 07-730. Available at: http:// www.engr.ucr.edu/~carter/SAPRC/. Chameides, W.L., Lindsay, R.W., Richardson, J., Kiang, C.S., 1988. The role of biogenic hydrocarbons in urban photochemical smog: Atlanta as a case study. Sci. New York N. Y. 241, 1473e1475. http://dx.doi.org/10.1126/science.3420404. Claeys, M., Graham, B., Vas, G., Wang, W., Vermeylen, R., Pashynska, V., Cafmeyer, J., Guyon, P., Andreae, M.O., Artaxo, P., Maenhaut, W., 2004. Formation of secondary organic aerosols through photooxidation of isoprene. Science 303, 1173e1176. http://dx.doi.org/10.1126/science.1092805. Crounse, J.D., Paulot, F., Kjaergaard, H.G., Wennberg, P.O., 2011. Peroxy radical isomerization in the oxidation of isoprene. Phys. Chem. Chem. Phys. 13, 13607e13613. http://dx.doi.org/10.1039/c1cp21330j. Crouse, R., 1990. Integrated Reaction Rate Analysis of Ozone Production in a Photochemical Oxidant Model (Master's thesis). University of North Carolina at Chapel Hill. da Silva, G., Graham, C., Wang, Z.-F., 2010. Unimolecular beta-hydroxyperoxy radical decomposition with OH recycling in the photochemical oxidation of isoprene. Environ. Sci. Technol. 44 (1), 250e256. http://dx.doi.org/10.1021/es900924d. http://www.ncbi.nlm.nih.gov/pubmed/19943615. Edney, E., Kleindienst, T., Jaoui, M., Lewandowski, M., Offenberg, J., Wang, W., Claeys, M., 2005. Formation of 2-methyl tetrols and 2-methylglyceric acid in secondary organic aerosol from laboratory irradiated isoprene/NOX/SO2/air mixtures and their detection in ambient PM2.5 samples collected in the eastern United States. Atmos. Environ. 39 (29), 5281e5289. http://dx.doi.org/10.1016/ j.atmosenv.2005.05.031. http://linkinghub.elsevier.com/retrieve/pii/ S1352231005004875. Foley, K.M., Roselle, S.J., Appel, K.W., Bhave, P.V., Pleim, J.E., Otte, T.L., Mathur, R., Sarwar, G., Young, J.O., Gilliam, R.C., Nolte, C.G., Kelly, J.T., Gilliland, A.B., Bash, J.O., 2010. Incremental testing of the Community Multiscale Air Quality (CMAQ) modeling system version 4.7. Geosci. Model Dev. 3, 205e226. http:// dx.doi.org/10.5194/gmd-3-205-2010. Fuchs, H., Hofzumahaus, A., Rohrer, F., Bohn, B., Brauers, T., Dorn, H.-P., H€ aseler, R., Holland, F., Kaminski, M., Li, X., Lu, K., Nehr, S., Tillmann, R., Wegener, R., Wahner, A., 2013. Experimental evidence for efficient hydroxyl radical regeneration in isoprene oxidation. Nat. Geosci. 10e13. http://dx.doi.org/10.1038/ ngeo1964. http://www.nature.com/doifinder/10.1038/ngeo1964. Guenther, A., Karl, T., Harley, P., Wiedinmyer, C., Palmer, P.I., Geron, C., 2006. Estimates of global terrestrial isoprene emissions using MEGAN (Model of Emissions of Gases and Aerosols from Nature). Atmos. Chem. Phys. Discuss. 6, 107e173. http://dx.doi.org/10.5194/acpd-6-107-2006. Hallquist, M., Wenger, J.C., Baltensperger, U., Rudich, Y., Simpson, D., Claeys, M., Dommen, J., 2009. The formation, properties and impact of secondary organic aerosol: current and emerging issues. Atmos. Chem. Phys. 9, 5155e5236. http://

dx.doi.org/10.5194/acp-9-5155-2009. Henderson, B.H., Vizuete, W., Jeffries, H.E., 2009. The python environment for reaction mechanism/mathematics (permm). In: Poster at the 8th Annual CMAS Conference, Chapel Hill, NC. Henze, D.K., Seinfeld, J.H., 2006. Global secondary organic aerosol from isoprene oxidation. Geophys. Res. Lett. 33 http://dx.doi.org/10.1029/2006GL025976. Hoyle, C.R., Berntsen, T., Myhre, G., Isaksen, I.S.A., 2007. Secondary organic aerosol in the global aerosol e chemical transport model Oslo CTM2. Atmos. Chem. Phys. 7, 5675e5694. http://dx.doi.org/10.5194/acp-7-5675-2007. :// 000251239500014. Hutzell, W.T., Luecken, D.J., Appel, K.W., Carter, W.P.L., 2012. Interpreting predictions from the SAPRC07 mechanism based on regional and continental simulations. Atmos. Environ. 46, 417e429. http://dx.doi.org/10.1016/j.atmosenv.2011.09.030. http://linkinghub.elsevier.com/retrieve/pii/S135223101100971X. http://www. sciencedirect.com/science/article/pii/S135223101100971X. Jeffries, H.E., 1995. Photochemical air pollution. In: Singh, H.B. (Ed.), Composition, Chemistry, and Climate of the Atmosphere. John Wiley and Sons, Inc. Jeffries, H.E., Fox, D.L., Kamens, R.M., 1976. Outdoor smog chamber studies: light effects relative to indoor chambers. Environ. Sci. Technol. 10, 1006e1011. http:// dx.doi.org/10.1021/es60121a016. Jeffries, H.E., Voicu, I., Sexton, K., 2002. Experimental Tests of Reactivity and Reevaluation of the Carbon Bond Four Photochemical Reaction Mechanism: Final Report for Cooperative Agreement r828906. USEPA (Tech. rep.). Jeffries, H.E., Kamens, R.M., Sexton, K., 2013. Early history and rationale for outdoor chamber work at the University of North Carolina. Environ. Chem. 10 (4), 349. http://dx.doi.org/10.1071/EN13901. http://www.publish.csiro.au/? paper¼EN13901. Kroll, J.H., Ng, N.L., Murphy, S.M., Flagan, R.C., Seinfeld, J.H., 2006. Secondary organic aerosol formation from isoprene photooxidation. Environ. Sci. Technol. 40, 1869e1877. http://dx.doi.org/10.1021/es0524301. Li, G., Zhang, R., Fan, J., Tie, X., 2007. Impacts of biogenic emissions on photochemical ozone production in Houston, Texas. J. Geophys. Res. 112 (D10), D10309. http://dx.doi.org/10.1029/2006JD007924. http://doi.wiley.com/10. 1029/2006JD007924. Lin, Y., Zhang, H., Pye, H.O.T., Zhang, Z., Marth, W.J., Park, S., Arashiro, M., 2013. Epoxide as a precursor to secondary organic aerosol formation from isoprene photooxidation in the presence of nitrogen oxides. Proc. Natl. Acad. Sci. U. S. A. http://dx.doi.org/10.1073/pnas.1221150110. http://www.pnas.org/content/110/ 17/6718.short. Mollner, A., Valluvadasan, S., Feng, L., 2010. Rate of gas phase association of hydroxyl radical and nitrogen dioxide. Science 330 (6004), 646e649. http:// dx.doi.org/10.1126/science.1193030. http://www.ncbi.nlm.nih.gov/pubmed/ 21030650. http://www.sciencemag.org/content/330/6004/646.short. Parikh, H.M., Jeffries, H.E., Sexton, K.G., Luecken, D.J., Kamens, R.M., Vizuete, W., 2013. Evaluation of aromatic oxidation reactions in seven chemical mechanisms with an outdoor chamber. Environ. Chem. 10 (3), 245. http://dx.doi.org/10.1071/ EN13039. http://www.publish.csiro.au/?paper¼EN13039. Paulot, F., Crounse, J.D., Kjaergaard, H.G., Kürten, A., St Clair, J.M., Seinfeld, J.H., Wennberg, P.O., 2009. Unexpected epoxide formation in the gas-phase photooxidation of isoprene. Sci. New York, N. Y. 325 (5941), 730e733. http:// dx.doi.org/10.1126/science.1172910. http://www.ncbi.nlm.nih.gov/pubmed/ 19661425. Peeters, J., Müller, J.-F., 2010. HO(x) radical regeneration in isoprene oxidation via peroxy radical isomerisations. II: experimental evidence and global impact. Phys. Chem. Chem. Phys. 12 (42), 14227e14235. http://dx.doi.org/10.1039/ c0cp00811g. http://www.ncbi.nlm.nih.gov/pubmed/20882226. Surratt, J.D., Murphy, S.M., Kroll, J.H., Ng, N.L., Hildebrandt, L., Sorooshian, A., Szmigielski, R., Vermeylen, R., Maenhaut, W., Claeys, M., Flagan, R.C., Seinfeld, J.H., 2006. Chemical composition of secondary organic aerosol formed from the photooxidation of isoprene. J. Phys. Chem. A 110, 9665e9690. http:// dx.doi.org/10.1021/jp061734m. Surratt, J.D., Lewandowski, M., Offenberg, J.H., Jaoui, M., Kleindienst, T.E., Edney, E.O., Seinfeld, J.H., 2007. Effect of acidity on secondary organic aerosol formation from isoprene. Environ. Sci. Technol. 41, 5363e5369. http:// dx.doi.org/10.1021/es0704176. Surratt, J.D., Chan, A.W.H., Eddingsaas, N.C., Chan, M., Loza, C.L., Kwan, A.J., Hersey, S.P., Flagan, R.C., Wennberg, P.O., Seinfeld, J.H., 2010. Reactive intermediates revealed in secondary organic aerosol formation from isoprene. Proc. Natl. Acad. Sci. U. S. A. 107 (15), 6640e6645. http://dx.doi.org/10.1073/ pnas.0911114107. http://www.pubmedcentral.nih.gov/articlerender.fcgi? artid¼2872383&tool¼pmcentrez&rendertype¼abstract. Xie, Y., Paulot, F., Carter, W.P.L., Nolte, C.G., Luecken, D.J., Hutzell, W.T., Wennberg, P.O., Cohen, R.C., Pinder, R.W., 2013. Understanding the impact of recent advances in isoprene photooxidation on simulations of regional air quality. Atmos. Chem. Phys. 13 (16), 8439e8455. http://dx.doi.org/10.5194/acp13-8439-2013. http://www.atmos-chem-phys.net/13/8439/2013/.