Interactive comment on “ Investigating the influences of SO 2 and NH 3 levels on isoprene-derived secondary organic aerosol formation using conditional sampling approaches ”

This manuscript describes field measurements aimed at understanding the potential role of ambient SO2 and NH3 concentrations in affecting the formation of isoprenederived SOA tracers. Aerosol samples were collected for low and high SO2 and NH3 levels and analyzed by gas chromatography electron ionization mass spectrometry (GC/EI-MS) and ultra-performance liquid chromatography electrospray ionization high resolution quadrupole time of flight mass spectrometry (UPLC/ESI-HR-Q-TOFMS). Six isoprene-derived tracers were quantified and were found to be enhanced under high


Introduction
Tropospheric fine aerosols (PM 2.5 , with aerodynamic diameter ≤ 2.5 µm) have been recognized to have significant influences on regional air quality, climate change, and human health (Kanakidou et al., 2005;Hallquist et al., 2009).Organic aerosol constituents that usually account for a large fraction (i.e., 20-90 %) of the total PM 2.5 mass are of particular concern, especially due to their high degree of chemical complexity that results in limited understanding of their sources, sinks, and chemical structure, thus yielding major uncertainties for air quality modeling and human health risk assessment.
Secondary organic aerosol (SOA) that originates from the photochemical oxidation of biogenic volatile organic compounds (BVOCs) is thought to be the largest contributor 664 Figure 1.Proposed chemistry leading to isoprene SOA: NO x -dependent pathways (Surratt et al.,665 2010; Lin et al., 2013).For simplicity, only the cis--IEPOX isomer is shown for IEPOX in the 666 low-NO x (or NO-limited) pathway.667 Fig. 1.Proposed chemistry leading to isoprene SOA: NO x -dependent pathways (Surratt et al., 2010;Lin et al., 2013).For simplicity, only the cis-β-IEPOX isomer is shown for IEPOX in the low-NO x (or NO-limited) pathway.
to the global SOA burden, owing to large emissions and efficient SOA formation processes (Chung and Seinfeld, 2002;Kanakidou et al., 2005;Henze and Seinfeld, 2006).The atmospheric significance of biogenic secondary organic aerosol (BSOA) has been inferred from recent remote sensing data over the southeastern United States, which suggests BVOC emissions combine with anthropogenic pollutants (e.g., SO 2 , NO x , and primary OC emissions), leading to substantial amounts of SOA observed in this region (Goldstein et al., 2009).The enhancement of SOA from BVOCs in this region highlights the need for further research aimed at understanding the attributable sources and detailed mechanisms leading to BSOA formation in order to develop effective control strategies.
Recently, organosulfate formation was reported through reactive uptake of BVOC oxidation products onto acidified sulfate seed aerosols, providing a likely link between anthropogenic pollutants and the enhanced BSOA formation (Iinuma et al., 2007(Iinuma et al., , 2009;;Surratt et al., 2007aSurratt et al., , 2008)).Moreover, the effects of acid-catalyzed enhancement on BSOA formation have also been observed in several laboratory studies (Iinuma et al., 2004;Surratt et al., 2007b;Offenberg et al., 2009;Chan et al., 2011).In the presence of anthropogenic pollutants, such as nitric acid and sulfuric acid produced from the oxidation of NO x and SO 2 , SOA mass yields from isoprene under high-and low-NO x conditions, respectively, have been shown to increase substantially (i.e., from 1-3 % to 3-30 %) with preexisting acidified sulfate aerosols in the laboratory (Chan et al., 2010a;Surratt et al., 2010).Figure 1 displays the NO x -dependent isoprene SOA formation mechanisms.Under high-NO x conditions, isoprene SOA is enhanced with increasing NO 2 /NO ratios (Chan et al., 2010a;Surratt et al., 2010).This enhance-ment is explained by the formation and subsequent photooxidation of methacryloylperoxynitrate (MPAN) (Surratt et al., 2010), forming methacrylic acid epoxide (MAE) (Lin et al., 2013), which leads to 2-methylglyceric acid (2-MG) formation and its corresponding oligoesters (Surratt et al., 2006(Surratt et al., , 2010)).Notably, 2-MG has been used as one of the isoprene SOA tracer compounds in the organic tracer-based source apportionment method developed by Kleindienst et al. (2007) to estimate the contributions of BVOCs to SOA formation.On the other hand, under low-NO x (i.e., NO-limited) conditions, isoprene SOA has been observed to be enhanced in the presence of acidified sulfate seed aerosols (mass yield ∼ 29 %) over that in the presence of neutral seed aerosols (mass yield ∼ 1 %) (Surratt et al., 2010).Increased uptake of gaseous isoprene epoxydiols (IEPOX) by enhanced particlephase acid-catalyzed oxirane ring-opening reaction rates has been proposed and demonstrated to explain this enhancement (Minerath et al., 2009;Eddingsaas et al., 2010;Surratt et al., 2010;Lin et al., 2012).Identification of known isoprene low-NO x SOA tracers through reactive uptake of IEPOX onto acidified sulfate seed aerosols also supports this hypothesis (Surratt et al., 2010;Lin et al., 2012).Although clear evidence of acid-catalyzed enhancement for BSOA formation has been shown in laboratory studies, the importance of aerosol acidity on ambient BSOA formation remains unclear, owing to the fact that acidified sulfate seed aerosols used in the laboratory studies were usually much more acidic than the acidity measured in ambient aerosols (Edgerton et al., 2007;Tanner et al., 2009).In addition, ambient aerosol acidity is also likely modulated by other environmental factors, such as the atmospheric alkaline species (e.g., NH 3 ) that leads to the neutralization of acidic aerosols (Huntzicker et al., 1980;McMurry et al., 1983), but the uptake processes and the kinetics in the atmosphere have yet to be fully elucidated (Huang et al., 2011;Yao et al., 2011;Liggio et al., 2011).
In the present study, PM 2.5 samples were collected to investigate the effects of ambient aerosol acidity on BSOA formation in a rural atmosphere under the influences of anthropogenic SO 2 and NH 3 emissions.Filter samples were collected from Yorkville, GA, a rural site located within the Southeastern Aerosol Research and Characterization Study (SEARCH) network during the summer of 2010.This site is characterized by high isoprene emissions during summertime, and is influenced by SO 2 point sources from local coal-fired power plants (Edgerton et al., 2006a) as well as NH 3 emissions from nearby poultry operations (Edgerton et al., 2007).Conditional sampling strategies were employed in this study to collect PM 2.5 samples under predefined environmental thresholds (i.e., the mixing ratios of SO 2 or NH 3 ) to distinguish the influences of ambient SO 2 and NH 3 levels on ambient aerosol acidity and BSOA formation.More specifically, known isoprene SOA tracers (which include organosulfate derivatives) were chemically characterized by UPLC/ESI-HR-Q-TOFMS (ultra performance liquid chromatography coupled to electrospray ionization highresolution quadrupole time-of-flight mass spectrometry) and GC/EI-MS (gas chromatography coupled to electron ionization mass spectrometry) techniques to measure SOA constituents at the molecular level.The effects of acid enhancement on BSOA formation were examined by comparing paired samples collected under high and low SO 2 or NH 3 scenarios.Even though some of these BSOA tracers have been previously characterized from PM 2.5 samples collected from the SEARCH network in a time-integrated manner (Chan et al., 2010b;Gao et al., 2006;Surratt et al., 2007aSurratt et al., , 2008)), using conditional sampling approaches to collect PM 2.5 in this study is to our knowledge one of the first attempts to systematically examine if BSOA formation is enhanced or suppressed due to anthropogenic emissions in this region.
2 Experimental section 2.1 Collection of PM 2.5 by conditional sampling PM 2.5 samples were collected from Yorkville (YRK), GA, a rural site located ∼ 55 km west northwest of Atlanta, GA, within the SEARCH network during summer 2010.The detailed site descriptions are provided in the SEARCH overview papers (Hansen et al., 2003;Edgerton et al., 2005Edgerton et al., , 2006b)).In the present work, paired quartz filter samples were collected by conditional sampling approaches based on the measured SO 2 or NH 3 mixing ratios.Concentration thresholds were set to (1) distinguish the influences of SO 2 or NH 3 levels on ambient aerosol acidity and isoprene SOA formation; (2) ensure that sufficient material was collected for chemical analysis; and (3) avoid motor burnout by activating the samplers too frequently.The pre-defined high and low SO 2 or NH 3 thresholds were determined by the review of historical ambient data at the sampling site.For SO 2 , data from June to August 2009 were analyzed and the thresholds were set to correspond to the lowest 40 % and highest 40 % of observed 1 min concentrations.We focused on 2009 because installation of emission controls on nearby power plants caused concentrations subsequent to 2008 to be significantly lower than 2008 and before.For NH 3 , we analyzed the same months of data for three consecutive years (2007)(2008)(2009) and set thresholds to correspond approximately to the lowest 20 % and highest 40 % of observed 1 min values.

SO 2 conditional sampling
SO 2 conditional samples were collected from 25 June 2010 until 14 July 2010 between 09:00 and 18:59 local standard time.Two Tisch Environmental (Cleves, OH) Model TE-6070V-2.5 high-volume PM 2.5 air samplers were operated side-by-side to collect aerosol samples at a flow rate of 1 m 3 min −1 during the day (09:00-18:59 LT).One highvolume PM 2.5 sampler was designated as the high-SO 2 sampler, which was only turned on to collect PM 2.5 samples when the measured SO 2 mixing ratios were ≥ 0.5 ppbv.The second high-volume PM 2.5 sampler was only turned on to collect aerosol samples when the SO 2 mixing ratios were ≤ 0.25 ppbv, and this sampler was designated as the low-SO 2 sampler.SO 2 was measured with 1 min time resolution using a Thermo-Environmental (Franklin, MA) Model 43c pulsed fluorescence SO 2 analyzer.

NH 3 conditional sampling
NH 3 conditional samples were collected from 29 July 2010 until 6 August 2010.The same conditional sampling approaches described above were employed to collect PM 2.5 samples based on the NH 3 mixing ratios during the day (09:00-18:59 LT).The high-NH 3 sampler only collected PM 2.5 samples when the NH 3 mixing ratios were ≥ 2 ppbv, while the low-NH 3 sampler only collected aerosol samples when the NH 3 mixing ratios were ≤ 1 ppbv.NH 3 was measured with 1 min time resolution via continuous denuder difference with a Thermo-Environmental Model 42c chemiluminescence NO-NO x analyzer.The NO-NO x analyzer was modified to measure NH 3 as described in Saylor et al. (2010).
Available collocated measurements and meteorological data during the conditional sampling periods are summarized in Table 1.

Filter extractions and chemical analyses for isoprene SOA tracers
Known isoprene SOA tracers, including 2-methyltetrols (Claeys et al., 2004), C 5 -alkene triols (Wang et al.,  2005), 3-methyltetrahydrofuran-3,4-diols (3-MeTHF-3,4diols) (Lin et al., 2012;Zhang et al., 2012), IEPOX-derived dimers (Surratt et al., 2006), and 2-MG (Edney et al., 2005) were characterized by GC/MS with prior trimethylsilylation using electron ionization (EI).A quarter of each 8 × 10-inch quartz filter sample was extracted in pre-cleaned scintillation vials with 20 mL high-purity methanol (LC-MS CHROMASOLV-grade, Sigma-Aldrich) under 45 min of sonication.The filter extracts were filtered through 0.2 µm PTFE syringe filters (Pall Life Science, Acrodisc ® ) to remove suspended quartz filter fibers and insoluble particles, and subsequently blown to dryness under a gentle N 2 stream at room temperature.Residues of the filter extracts were immediately trimethylsilylated by reacting with 100 µL of BSTFA + TMCS (99 : 1 v/v, Supelco) and 50 µL of pyridine (anhydrous, 99.8 %, Sigma-Aldrich) at 70 • C for 1 h.The derivatized samples were analyzed by GC/MS within 24 h after trimethylsilylation.GC/MS analysis was performed using a Hewlett-Packard (HP) 5890 Series II Gas Chromatograph coupled to a HP 5971A Mass Selective Detector.An Econo-Cap ® -EC ® -5 Capillary Column (30 m × 0.25 mm i.d.; 0.25 µm film thickness) was used to separate the trimethylsilyl (TMS) derivatives before MS detection.1 µL of each derivatized sample was injected onto the GC column.Operating conditions and the temperature program of the GC/MS procedure were as described previously by Surratt et al. (2010).Isoprene SOA tracers were quantified with the following base peak ion fragments: m/z 219 for 2-methyltetrols, m/z 231 for C 5 -alkene triols, m/z 262 for 3-MeTHF-3,4-diols, m/z 335 for dimers, and m/z 219 for 2-MG.Meso-erythritol (≥ 99 %, Sigma), a structurally similar analog of 2-methyltetrols, was used as a surrogate standard to quantify 2-methyltetrols, C 5 -alkene triols, dimers, and 2-MG in the PM 2.5 samples.The ion fragment of m/z 217 was used for meso-erythritol for quantification.3-MeTHF-3,4-diols were quantified using synthesized authentic standards.The details of the synthesis procedures are reported in Zhang et al. (2012).Characterization of organosulfate derivatives was performed using ultra performance liquid chromatography interfaced to a high-resolution quadrupole time-of-flight mass spectrometer (Agilent 6500 Series) equipped with an electrospray ionization source (UPLC/ESI-HR-Q-TOFMS) operated in the negative (-) ion mode.A Waters ACQUITY UPLC HSS T3 column (2.1 × 100 mm, 1.8 µm particle size) was used for chromatographic separations.Detailed UPLC/(-)ESI-HR-Q-TOFMS operating conditions can be found in Zhang et al. (2011).Quartz filter samples for UPLC/(-)ESI-HR-Q-TOFMS analyses were extracted in the same manner as those for GC/MS analyses.After the filter extracts were blown dry, the extract residues were reconstituted with 150 µL of a 50 : 50 (v/v) solvent mixture of methanol containing 0.1 % acetic acid (LC-MS CHROMASOLV-grade, Sigma-Aldrich) and water containing 0.1 % acetic acid (LC-MS CHROMASOLV-grade, Sigma-Aldrich).5 µL of each sample was injected onto the UPLC column eluted with solvent of the same composition.Isoprene-derived organosulfate species reported previously by Surratt et al. (2008) were identified and quantified.The elemental composition of target compounds was assigned based on accurate mass data.The errors of accurate mass fittings were within ±1 mDa.Owing to the lack of authentic standards, sodium propyl sulfate (electronic grade, City Chemical LLC) was used to quantify all isoprene-derived organosulfates.The use of sodium propyl sulfate to quantify isoprene-derived organosulfates was performed under the assumption of similar ionization efficiency since the retention time (and thus the mobile phase composition) of this surrogate standard is similar to our target analytes.The quantification was done based on the mass response.The detection limit of sodium propyl sulfate on UPLC/(-)ESI-HR-TOFMS was 0.01 ng µL −1 , determined by signal-to-noise ratios of 3 : 1.
The efficiency of the extraction protocols was evaluated by spiking 5 replicates of pre-baked blank quartz filters with quantifying standards.Extraction efficiencies (62-82 %) are taken into account for SOA constituents that were quantified in the field samples.

Inorganic ion measurements
Continuous sulfate (SO 2− 4 ), nitrate (NO − 3 ), and ammonium (NH + 4 ) measurements were made during the sampling periods at the site using continuous particle analyzers; detailed instrumental setup and operating conditions of these analyzers have been described elsewhere (Edgerton et al., 2006b).Briefly, continuous SO 2− 4 measurements were made using a variation of the Harvard School of Public Health approach.A high temperature (> 850 • C) stainless steel tube was used to reduce particle SO 2− 4 to SO 2 .Then SO 2 was measured by a Thermo-Environmental Instruments Model 43s or 43ctl high-sensitivity, pulsed ultraviolet fluorescence SO 2 analyzer.Continuous NO − 3 and NH + 4 measurements were made using a three-channel continuous differencing approach developed by ARA, Inc.Air samples were drawn through the inlet and series of denuders coated with sodium carbonate and citric acid followed by an activated carbon honeycomb denuder to remove interferents, and then the flow was split into three analytical channels that converted different nitrogen species to NO, depending on temperature.Channel 1 (CH1) measured the baseline gas-phase NO y for the analyzer by passing the air through a 2 µm Teflon filter, followed by a KCl-coated filter and a molybdenum (Mo) mesh catalyst heated to 350 • C. Channel 2 (CH2) measured the baseline NO y plus particle-bound NO − 3 by passing air through Mo converter at 350 • C without filtration.Channel 3 (CH3) measured NH + 4 plus baseline NO y and particle-bound NO − 3 by flowing air through ceramic tube containing platinum (Pt) catalyst heated to 600 • C, followed by another Mo converter at 350 • C. NO − 3 and NH + 4 concentrations were calculated as CH2-CH1 and CH3-CH2, respectively.Continuous inorganic data were processed in a time-weighted manner to rep-resent the real-time aerosol inorganic conditions to estimate ambient aerosol acidity.
Filter-based inorganic measurements were also performed by analyzing a 37 mm filter punch from each quartz filter sample.Filter samples were extracted with 15 mL Millipore 18.2 Megohm Ultrapure DI H 2 O.A Dionex ICS-3000 Ion Chromatography System was used to quantify SO 2− 4 , NO − 3 , and NH + 4 concentrations in filter extracts.The anion channel included a Dionex AG18 guard column, a Dionex AS18 analytical column and 10 mM KOH eluent.The cation channel included a Dionex CG16 guard column, a Dionex CS16 analytical column and 18 mM methanesulfonic acid eluent.The ICS-3000 was calibrated with NIST-traceable multielement standards covering the range of observed concentrations.Sample injection volumes were 1.0 mL and peak area was used for quantification.Analytical detection limits were in the range of 2-3 µg L −1 , and analytical uncertainties were < 5 %.Measured concentrations were compared with continuous inorganic measurements to evaluate the changes of inorganic composition during sample storage.

Calculation of aerosol acidity
Ambient aerosol acidity of the collected PM 2.5 samples was estimated by calculating the degree of stoichiometric neutralization, defined as the molar ratios of NH + 4 to the sum of SO 2− 4 and NO − 3 , assuming aerosol SO 2− 4 and NO − 3 were only associated with NH + 4 and H + .Acidic aerosols are characterized by having a neutralization degree less than unity.The neutralization degree greater than unity implies the samples are fully neutralized.
In addition, an on-line version of the extended aerosol thermodynamics model (E-AIM II: ) was used to estimate aerosol acidity by calculating the in situ aerosol pH in the particle aqueous phase (Clegg et al., 1998).Inputs of free [H + ] were calculated based on charge balance from measured NH + 4 , SO 2− 4 , and NO − 3 concentrations.Temperature and relative humidity (RH) parameters were obtained from the measurements at the sampling site.
Modeling outputs of activity coefficient and moles of H + aq in the aqueous phase, and the total volume of aqueous phase in the aerosol per m 3 air in the thermodynamic equilibrium were used to calculate in situ aerosol pH.

Results and discussion
As the SO 2 conditional sampling and NH 3 conditional sampling experiments were two independent experiments, samples collected from SO 2 and NH 3 conditional sampling experiments could not be inter-compared for the following reasons: 1. SO 2 conditional sampling experiments were conducted 25 June 2010-14 July 2010.During this time, PM 2.5 sample collections were determined by the ambient SO 2 mixing ratios.Thus, only high-and low-SO 2 samples collected on the same day (as a paired sample) could be compared, since other variables, including NH 3 , were not constrained.
2. Similarly, NH 3 conditional sampling experiments were conducted 29 July 2010-6 August 2010, and only the ambient NH 3 mixing ratios determined when the highvolume samplers were operated.We considered highand low-NH 3 samples collected on the same day as a paired sample to distinguish the effects of ambient NH 3 .

Identification and quantification of isoprene SOA tracers at YRK
UPLC/(-)ESI-HR-Q-TOFMS and GC/EI-MS with prior trimethylsilylation were used to identify and quantify isoprene SOA tracers in the PM 2.5 samples collected from the YRK site.Figure 2 (Yasmeen et al., 2011).For the TIC from GC/EI-MS analysis, the most abundant peak (labeled as Peak 1) was identified as bis(trimethylsilyl) sulfate based on the NIST MS library search.This peak, which likely originates from the extracted particle sulfate content after trimethylsilylation, was also observed by Jaoui et al. (2012) in the samples collected from chamber studies under conditions of pre-seeded ammonium sulfate aerosol and SO 2 oxidation that forms sulfuric acid.Other major peaks detected could be attributed to isoprene low-NO x SOA tracers, including 3-MeTHF-3,4-diols (two isomers grouped as Peak 2), C 5 -alkene triols (three isomers grouped as Peak 3), and 2-methyltetrols (two isomers grouped Peak 4).Strong signal intensity of C 5 -alkene triols and 2-methyltetrols in the TIC, as well as the IEPOX-derived organosulfate in the BPC from the UPLC/(-)ESI-HR-Q-TOFMS technique, shows the atmospheric significance of IEPOX chemistry leading to SOA formation in this region.SOA tracer compounds were quantified with authentic or surrogate standards.Field blanks were collected during the field study, and subsequently analyzed to subtract background signals from all aerosol samples for quantification.The average mass concentrations of isoprene low-NO x and high-NO x SOA tracers measured under high-and low-SO 2 conditions in this study are shown in Table 2.For high SO 2 conditional samples, the average mass concentration of isoprene low-NO x tracers, including 2-methyltetrols, C 5alkene triols, 3-MeTHF-3,4-diols, as well as IEPOX-derived organosulfate (m/z 215), were higher than those of the corresponding paired low SO 2 conditional samples (p = 0.012).The differences of isoprene high-NO x (MAE-derived) SOA tracers between high-and low-SO 2 conditions were not as significant (p = 0.754).This observation provides evidence for isoprene SOA formation from IEPOX chemistry being enhanced under a high-SO 2 environment.The enhancement of isoprene SOA formation, especially for the identified isoprene low-NO x SOA tracers, can be explained by the oxidation of SO 2 producing sulfuric acid that provides aerosol acidity to enhance the rate of heterogeneous oxirane ringopening reactions when the gas-phase IEPOX is taken up by preexisting acidic aerosol surfaces (Edney et al., 2005;Surratt et al., 2010;Lin et al., 2012).Moreover, the enhancement of IEPOX chemistry could also be explained by the increased aerosol sulfate concentrations under high-SO 2 conditions (Table 1).The hygroscopic nature of the particle sulfate could have provided a wet aerosol surface that facilitated the IEPOX uptake and lead to SOA formation.In order to clarify the role of SO 2 , correlations between aerosol acidity, particle sulfate loadings and isoprene low-NO x SOA tracers are discussed in subsequent sections of this manuscript.It is noteworthy that the previous detection of particle-phase IEPOX in Chan et al. (2010b), which was characterized as m/z 262 with prior trimethylsilylation using GC/EI-MS analysis, has been shown to be a misidentification of 3-MeTHF-3, 4-diols (Lin et al., 2012).This is confirmed through organic synthesis of both IEPOX and 3-MeTHF-3, 4-diols authentic standards (Zhang et al., 2012).The concentrations of 3-MeTHF-3, 4-diols detected in this study were estimated ranging from non-detactable (n.d.) to 35 ng m −3 , which were comparable to the reported particle-phase IEPOX in Chan et al. (2010b) (n.d. to 24 ng m −3 ).
One possible source of uncertainty regarding quantification of 2-methyltetrols and 2-methylglyceric acid should be noted.These two tracers could possibly be overestimated due to the hydrolysis and derivatization of IEPOX-and MAEderived organosulfates upon trimethylsilylation.To evaluate the extent of this effect, we analyzed sodium octyl sulfate using GC/MS with prior trimethylsilylation, and compared the results to octanol with trimethylsilylation.Sodium octyl sulfate and octanol were used here as surrogates owing to the lack of authentic standards for IEPOX-and MAE-derived organosulfates, as well as their hydrolyzed products (i.e., 2methyltetrols and 2-methylglyceric acid).The results indicate ∼ 1.6 % octyl sulfate was hydrolyzed during the process of trimethylsilylation.This implies likely only a small fraction of IEPOX-or MAE-derived organosulfates would be hydrolyzed during the same analytical process and quantified as 2-methyltetrols and 2-methylglyceric acid.However, while isoprene-derived organosulfates are more highly substituted with electron withdrawing groups, the potential limitation of using octyl sulfate as a proxy should be noted, since octyl sulfate is a primary sulfate with no substitution on adjacent carbons.Additionally, the MAE-derived organosulfate (detected as the [M-H] − ion at m/z 199) produced a MS/MS spectrum (Fig. 1S, see Supplement) consistent with that recently shown by Safi Shalamzari et al. (2013).
The average mass concentrations of isoprene SOA tracers measured under high-and low-NH 3 conditions are shown in Table 3. Organosulfate species, such as the IEPOXderived organosulfate (m/z 215), the organosulfate derivative of the IEPOX-derived dimers (m/z 333), and MAEderived organosulfate (m/z 199) were more abundant under low-NH 3 conditions.Other tracer compounds, however, have higher mass concentrations detected under high-NH 3 conditions, which might coincide with the higher average solar radiation under high NH 3 conditional sampling (Table 1).Figure 3 shows correlations between isoprenederived organosulfates.Interestingly, strong correlations between the IEPOX-derived organosulfate (m/z 215) and the organosulfate derivatives of the IEPOX-derived dimers  (m/z 333) (r = 0.83) support the common pathway and formation mechanisms of these two tracers, as shown in Fig. 1.In addition, strong correlations (r = 0.71) were observed between the IEPOX-derived organosulfate (m/z 215) and the MAE-derived organosulfate (m/z 199), suggesting similar formation behaviors or limiting factors, since these two species are known to form from different NO x -dependent pathways as shown in Fig. 1.Reactive uptake of epoxide precursors followed by particle-phase oxirane ring-opening reactions appears to be a common mechanism for both low-NO x and high-NO x isoprene SOA formation pathways, although yielding distinct SOA composition.Notably, the detected isoprene low-NO x SOA tracers were higher during the NH 3 conditional sampling period (late July to early August) than the SO 2 conditional sampling period (late June to mid-July), likely due to the change of meteorological conditions.For example, a higher average temperature during the NH 3 conditional sampling period might lead to higher isoprene emissions, as well as faster photochemical processes for isoprene SOA formation.In contrast, the isoprene high-NO x SOA tracers remained constant between the two sampling periods.This could be explained by the NO-limited isoprene SOA formation pathway that occurs once the NO levels in the environment are consumed, resulting in the low-NO x (RO 2 +HO 2 ) chemistry becoming the dominant pathway in the atmosphere, while isoprene is still continuously emitted during daytime.
Temporal variations of isoprene low-NO x and high-NO x SOA tracers compared to OM are shown in Figs. 4 and 5, respectively.On most days, IEPOX-and MAE-derived SOA tracers track well with the OM mass loadings.The r values range from 0.40-0.68 for IEPOX-derived SOA tracers and 0.17-0.69for MAE-derived SOA tracers.Weak correla-  tions occur under high NH 3 conditional sampling events for both IEPOX-and MAE-derived SOA tracers.It is worth noting that the sum of IEPOX-derived SOA tracers contributed substantially (12-19 %) to the total organic matter (OM) in the PM 2.5 samples collected at this site, showing the atmospheric abundance of the isoprene low-NO x (or IEPOXderived) SOA that originates from a single source and pathway (Tables 2 and 3).OM was estimated as OC × 1.8 here for summertime aerosols (Simon et al., 2011).To examine whether or not the differences of isoprene SOA formation between high and low SO 2 or NH 3 conditional sampling protocols were significant, paired-t tests were performed.The results of the paired-t tests indicate that the enhancement of IEPOX-derived SOA is statistically significant (p = 0.012) under high-SO 2 conditions.No significant enhancement of the sum of isoprene SOA tracers was observed for high or low NH 3 conditional samples (p = 0.830).

Comparisons of continuous inorganic measurements to filter-based IC data
Figure 6 shows comparisons of continuous inorganic measurement to filter-based measurements in this study.Timeweighted average of continuous inorganic data was compared to the IC data from filter samples.Strong correlations were observed for particle NH + 4 (r = 0.85) and SO 2− 4 (r = 0.79) concentrations from two data sets.The correlation for particle NO − 3 concentrations was (r = 0.49) weak, possibly due to low concentrations throughout the study and artifacts in both the filter and continuous measurements.

Ambient aerosol acidity and isoprene SOA tracers
To estimate aerosol acidity of collected PM 2.5 samples, the degree of neutralization was calculated as the molar ratio of ammonium to the sum of sulfate and nitrate ). Acidic aerosols are characterized with neutralization degrees less than unity, while neutralization degrees greater than unity imply the samples are neutralized.Figure 7 compares the degree of neutralization for PM 2.5 samples collected in this study calculated from online continuous inorganic measurements and filter-based IC data.In general, most of the aerosol samples have been fully neutralized.However, the frequency distribution from filterbased data shows that most of the samples had the neutralization degree close to unity, while data from continuous inorganic measurements had a more broad distribution.Since filter samples could have absorbed the ambient acidic or basic gases that lead to neutralization of aerosol samples over time during the sampling period, to capture the more representative aerosol acidity, continuous inorganic data were used for further data analysis.
Figure 8 shows the distribution of aerosol samples grouping based on sampling conditions.SO 2 conditional samples are acidic, while NH 3 conditional samples are more neutralized.Figure 9 shows the result from E-AIM II modeling.In situ aerosol pH can only be calculated in very few cases due to the neutralized characteristics of aerosol samples.In addition, some of the samples, although not fully neutralized, were modeled as no liquid water content (LWC) under given RH conditions.More specifically, the amount of water (LWC) in an aerosol particle calculated using the E-AIM Model II system (H ) is dependent on the inorganic materials (electrolytes) present, the ambient relative humidity, and temperature.Although there were some other high RH episodes observed during the sampling period (with RH > 0.8), these samples in fact have been fully neutralized ([H + ] free < 0), and thus have to be excluded from the data set in the first place, since a charge imbalance in the ionic composition of the system is not allowed for the model input.Thus, aerosol pH could not be calculated for those samples either.As a result, only very limited number of samples could be modeled using this approach.Only 3 samples (out of 50) could be modeled for their in situ aerosol pH, and the average was found to be 1.71, ranging from 1.69 to 1.75.However, the E-AIM model in this study was run under the assumption of no interaction between the organic phase and the inorganic phase.In a recent study by Smith et al. (2012), isoprene-derived secondary organic materials have been shown to mix miscibly with aqueous ammonium sulfate, and the resultant mixture shifted efflorescence and deliquescence points of pure ammonium sulfate.As a result, the modeling results could be inaccurate under this assumption, and thus did not capture the actual particle LWC.
Figures 10 and 11 show the correlations between the neutralization degree and the sum of isoprene low-NO x (or IEPOX-derived) and high-NO x (or MAE-derived) SOA tracers, respectively.For IEPOX-derived SOA, the results show that the tracer compounds were weakly enhanced with lessneutralized aerosols, but no clear associations were observed for NH 3 conditional samples.For MAE-derived SOA, no clear associations were observed for either SO 2 or NH 3 conditional samples.This could be due to most of the aerosol samples being fully neutralized.Notably, prior chamber work has shown that under high-NO x conditions no enhancement in SOA mass was observed due to the presence of acidified sulfate seed aerosol, whereas under low-NO x conditions SOA mass was enhanced due to the presence of acidified sulfate seed aerosol (Surratt et al., 2006).Furthermore, the aerosol samples could have formed upwind of the sampling site, and thus were associated to a more aged and regional (background) aerosol type.Thus, distinguishing weak correlations from meteorological effects on measured isoprene SOA levels is challenging, since the isoprene SOA might have been formed upwind, but not formed locally.It should also be noted that the history of ambient aerosol acidity is complicated.The aerosol could be neutralized on average with time, but start as an acidic aerosol that has already provided sufficient acidity to allow significant IEPOX processing.Our findings are in agreement with Tanner et al. (2009), who previously reported that at the YRK site no consistent positive correlations were found between changes in OC or TC levels and aerosol acidity, which was estimated by nitrate-corrected ammonium-to-sulfate ratios, even with time lag up to 6 h.Aerosol acidity at this site is relatively low due to nearby agricultural sources of NH 3 .In addition, recent research by Liggio et al. (2011) reported that the rate of aerosol neutralization by NH 3 uptake is significantly reduced in the presence of ambient organic gases on timescales ∼ 10 min to several hours, while acidic aerosol mixed with organic-free air and NH 3 was neutralized on a timescale < 1 min.This reduction in NH 3 uptake was concurrent with an increase in the amount of particle-phase organics.Thus, our NH 3 conditional sampling approaches might not be able to capture the neutralization effects on aerosol acidity and isoprene SOA formation, since this would not be an instantaneous process in the ambient environment.

Particle sulfate loadings and isoprene SOA tracers
Figures 12 and 13 show correlations between isoprene low-NO x and high-NO x SOA tracers and the particle sulfate loadings, with r = 0.51-0.58for IEPOX-derived SOA tracers and r = 0.34-0.36for MAE-derived SOA tracers, respectively.Figure 14 shows correlations between the sum of isoprene SOA tracers (combining IEPOX-and MAE-derived SOA) and particle SO 2− 4 loadings for all PM 2.5 samples collected during SO 2 conditional sampling period (r = 0.44) and the NH 3 conditional sampling period (r = 0.58) in this study.Positive correlations were observed for all conditions.These condition-independent correlations may suggest that aerosol sulfate could serve as the surface accommodation factor that facilitates the reactive uptake of IEPOX onto preexisting aerosols.Although acidified sulfate aerosol has been demonstrated to enhance heterogeneous isoprene SOA formation (Surratt et al., 2007b(Surratt et al., , 2010)), laboratory studies are lacking that systematically examine the effect of varying surface area of preexisting aerosol on isoprene SOA formation as a function of liquid water content, aerosol acidity, and chemical composition.More work is needed to understand if this is a surface-or bulk-limited process and how this changes with environmental conditions.

Conclusions
Analyses of PM 2.5 samples collected from the rural southeastern US by off-line chromatography coupled with mass spectrometry techniques show substantial contributions (12-19 %) of isoprene low-NO x SOA tracers to organic aerosol mass, revealing the importance for heterogeneous chemistry of IEPOX in this region.Conditional sampling approaches employed in this study indicate that IEPOX-derived SOA formation is enhanced under higher SO 2 conditions (p = 0.012).In contrast, conditional sampling did not show significant influence of NH 3 levels on low-NO x isoprene SOA concentrations.Thus, it is possible that the effects (or degree) of NH 3 neutralization were masked by other confounding factors or atmospheric processes occurring simultaneously.Particle sulfate loadings have moderate positive correlations with the sum of isoprene SOA tracers for all conditions, suggesting a role of surface-or bulk-limited chemistry for isoprene SOA formation.Weak correlations between isoprene SOA tracers and aerosol acidity could be attributed to the fact that information is lacking about the history of the ambient aerosol acidity.Evaporation of volatile particle constituents, such as nitrates, would affect the estimate of aerosol acidity, although NO − 3 concentrations were generally too low to make a big difference in the neutralization calculation.In addition, during the volatilization process it remains uncertain if this is primarily HNO 3 or HNO 3 plus NH 3 leaving the filters.The latter would not have any effect on acidity.
Since very few samples could be modeled using E-AIM II, our ability to elucidate potential effects of aqueousphase chemistry on isoprene SOA formation is quite limited.Regional-scale transport could have brought isoprene SOA formed upwind that masked the effects of aerosol acidity.As a result, future work is needed to differentiate the effects of enhanced BSOA formation from regional-scale transport and aqueous-phase chemistry.

Fig. 3 .
Fig. 3. Correlations between isoprene-derived organosulfates: (a) correlations between IEPOX-derived organosulfate (m/z 215) and organosulfate derivatives of IEPOX-derived dimers (m/z 333) support the common isoprene low-NO x pathway and formation mechanisms of these two tracers.(b) Correlations between the IEPOX-derived organosulfate (m/z 215) and the MAE-derived organosulfate (m/z 199) suggest similar limiting factors for organosulfate formation, as these two species have been known to form from isoprene photooxidation through low-NO x and high-NO x pathways, respectively.

Table 1 .
Summary of average meteorological data and complementary collocated measurements.

Table 2 .
Isoprene SOA tracers quantified in PM 2.5 samples (ng m −3 ) collected under high-and low-SO 2 conditions.

Table 3 .
Isoprene SOA tracers quantified in PM