Atmospheric Chemistry and Physics α-pinene photooxidation under controlled chemical conditions – Part 1 : Gas-phase composition in low-and high-NO x environments

The OH oxidation ofα-pinene under both lowand high-NOx environments was studied in the Caltech atmospheric chambers. Ozone was kept low to ensure OH was the oxidant. The initialα-pinene concentration was 20–50 ppb to ensure that the dominant peroxy radical pathway under low-NOx conditions is reaction with HO 2, produced from reaction of OH with H2O2, and under high-NOx conditions, reactions with NO. Here we present the gas-phase results observed. Under low-NOx conditions the main first generation oxidation products are a number of α-pinene hydroxy hydroperoxides and pinonaldehyde, accounting for over 40 % of the yield. In all, 65–75 % of the carbon can be accounted for in the gas phase; this excludes first-generation products that enter the particle phase. We suggest that pinonaldehyde forms from RO2 + HO2 through an alkoxy radical channel that regenerates OH, a mechanism typically associated with acyl peroxy radicals, not alkyl peroxy radicals. The OH oxidation and photolysis of α-pinene hydroxy hydroperoxides leads to further production of pinonaldehyde, resulting in total pinonaldehyde yield from low-NO x OH oxidation of ∼33 %. The low-NOx OH oxidation of pinonaldehyde produces a number of carboxylic acids and peroxyacids known to be important secondary organic aerosol components. Under high-NOx conditions, pinonaldehyde was also found to be the major first-generation OH oxidation product. The high-NOx OH oxidation of pinonaldehyde did not produce carboxylic acids and peroxyacids. A number of organonitrates and peroxyacyl nitrates are observed and identified from α-pinene and pinonaldehyde.


Introduction
The emissions of biogenic volatile organic compounds (BVOCs) far outnumber those of anthropogenic VOCs (Guenther et al., 1995;Steinbrecher et al., 2009;Monks et al., 2009).Excluding methane, BVOCs are estimated to account for a flux of ∼1150 Tg C yr −1 , while anthropogenic VOCs account for only ∼140 Tg C yr −1 (Guenther et al., 1995;Goldstein and Galbally, 2007).Important BVOCs include isoprene, (flux of ∼500 Tg C yr −1 ) and the monoterpenes (∼127 Tg C yr −1 ) of which α-pinene accounts for ∼50 Tg C yr −1 (Guenther et al., 1995;Chung and Seinfeld, 2002).Because they are unsaturated, these compounds are highly reactive towards OH, O 3 , and NO 3 and thus play an important role in tropospheric chemistry.The atmospheric oxidation of BVOCs also results in the formation of secondary organic aerosol (SOA).Monoterpenes are significant sources of SOA due to their large emission rate and high SOA yield (Hoffmann et al., 1997;Pye et al., 2010).
The gas-phase oxidation of simple alkanes and alkenes is well understood and most tropospheric chemical mechanisms use laboratory studies of reactions of these species to inform the parameterization of the atmospheric oxidation of VOCs.Many BVOCs, including isoprene, which has a conjugated double bond system, and α-pinene which is a bicyclic hydrocarbon with an endocyclic double bond, however, have much more complicated chemistry than the simple alkanes and alkenes.For example, recent experimental and theoretical studies with isoprene have shown that its atmospheric chemistry is not well modeled by the reactions of simple Published by Copernicus Publications on behalf of the European Geosciences Union.
α-pinene is a ten-carbon bicyclic hydrocarbon with an endocyclic double bond and, therefore, has the potential to react in ways not represented by simple alkenes.In addition, α-pinene is highly reactive with both O 3 and OH, reacting in the atmosphere with α-pinene nearly equally, adding to the richness of its atmospheric photochemistry (Capouet et al., 2008).Despite a number of studies of OH oxidation of αpinene, large uncertainties in the identity and yields of its reaction products remain.For instance, the reported yield of pinonaldehyde from the photooxidation of α-pinene ranges from 28 to 87 % in the presence of NO x and from 3 to 37 % in the absence of NO x (Arey et al., 1990;Hatakeyama et al., 1991;Noziere et al., 1999;Jaoui and Kamens, 2001;Wisthaler et al., 2001;Aschmann et al., 2002;Lee et al., 2006).In addition, the molecular structures of major products identified by mass (e.g.184 and 200 daltons) are subject to debate (Aschmann et al., 2002;Vereecken et al., 2007).The photooxidation product with molecular mass 184 has been assigned to an unsaturated hydroperoxy carbonyl (Vereecken et al., 2007) or a dihydroxy carbonyl (Aschmann et al., 2002).A better understanding of α-pinene gas-phase chemistry will increase the accuracy of α-pinene atmospheric chemistry models and provide insight into atmospheric photooxidation mechanisms in general.
In this study, we isolate the peroxy radical reaction pathways to investigate the photochemistry of α-pinene.We have studied these reactions under low-NO x conditions similar to those found in the atmosphere, where RO 2 + HO 2 is the dominant peroxy radical reaction, and other reactions are suppressed (RO 2 + RO 2 and reactions with O 3 ).We contrast these conditions with results from the gas-phase photooxidation of α-pinene under high-NO x (with varied amounts of NO 2 to study both RO 2 + NO and RO 2 + NO 2 ).We focus our analysis on low-NO x rather than high-NO x chemistry because the low-NO x chemistry of α-pinene is less wellcharacterized yet it is more atmospherically-relevant (Pye et al., 2010).In a forthcoming paper, SOA yields and composition formed from the controlled chemical conditions described here will be presented.

Experimental
Photooxidation experiments of α-pinene were performed in the Caltech dual 28 m 3 Teflon chambers.Details of the chamber facilities have been described elsewhere (Cocker et al., 2001;Keywood et al., 2004).A few photooxidation experiments were performed in a ∼1 m 3 bag enclosed in a small, black walled chamber with UV-lights lining one wall, as described by Crounse et al. (2011).40 W black lights (Sylvania F40/350BL) with emission peak emission at 352 nm were used in both chambers.The light intensity as a function of wavelength (300-800 nm) was measured using a Licor (LI-1800) spectroradiometer.Prior to each run, the chamber was flushed for a minimum of 24 h with dry purified air.While being flushed, the chamber was irradiated with the chamber lights for a minimum of six hours.The temperature, relative humidity (RH), and concentrations of O 3 , and NO x (NO and NO 2 ) were continuously monitored.In all experiments the RH was kept below 10 %.Aerosol size distribution and number concentration were measured continuously by a differential mobility analyzer (DMA, TSI model 3081) coupled to a condensation nucleus counter (TSI model 3760).Information on aerosol measurements can be found in part 2, where aerosol formation, growth, and composition are discussed.Table 1 shows a list of conditions for all experiments used in this study.Experiments ran for 3 to 19 h.No gas phase losses to the chamber walls of either α-pinene prior to light irradiation or oxidation products after lights were extinguished was observed.
Experiments were performed under low-and high-NO x conditions.Under low-NO x conditions, photolysis of hydrogen peroxide (H 2 O 2 ) was the OH source, while for the high-NO x experiments the photolysis of nitrous acid (HONO) or methyl nitrite (CH 3 ONO) produced OH.For low-NO x experiments, 280 µL of 50 wt % H 2 O 2 was injected into the chamber resulting in an H 2 O 2 concentration ∼4 ppm.The two different OH sources used during the high-NO x experiments provided the mechanism to vary the NO to NO 2 ratio, with a higher quantity of NO 2 in the CH 3 ONO experiments.HONO was prepared daily by dropwise addition of 15 mL of 1 wt % NaNO 2 into 30 mL of 10 wt % H 2 SO 4 in a glass bulb, and then introduced into the chamber with dry air.This process produces NO and NO 2 as side products, which are also introduced to the chamber.CH 3 ONO was synthesized, purified, and stored according to the procedure outlined by Taylor et al. (1980).CH 3 ONO was warmed from liquid nitrogen temperatures and vaporized into an evacuated 500 mL glass bulb and introduced into the chamber with an air stream of 5 L min −1 .After addition of CH 3 ONO, 300-400 ppb of NO was added to the chamber to suppress the formation of O 3 .Determination of exact NO and NO 2 concentrations using the commercial NO x monitor was precluded due to interferences by both HONO and CH 3 ONO.At the start of all high-NO x experiments the total NO x reading (NO, NO x , and interference from HONO or CH 3 ONO) was 800 ppb and NO concentration throughout the experiments was such that the concentration of O 3 never exceeded 5 ppb.
α-pinene was added to the chamber to achieve a concentration of 20-50 ppb by passing dry air through a bulb containing a known volume of α-pinene.The mixing ratio of α-pinene was monitored by gas chromatography (Agilent 6890N) coupled with a flame ionization detector (GC-FID).The GC-FID was calibrated using a 55 L Teflon bag containing a known concentration of pure α-pinene.Gas phase photooxidation products were monitored using triple-quadrupole chemical ionization mass spectrometry (CIMS) (St. Clair et al., 2010).
Details of the operation of the CIMS can be found in a number of previous reports (Crounse et al., 2006;Paulot et al., 2009a;St. Clair et al., 2010) and therefore only a brief description is presented here.The CIMS was operated in negative ion mode using CF 3 O − as the reagent ion, and in the positive ion mode using H 3 O + for proton transfer mass spectrometry (PTR-MS).In negative mode, CF 3 O − is sensitive to the detection of polar and acidic compounds by either clustering with the analyte (R) resulting in an ion with a massto-charge ratio (m/z) MW+85 (R • CF 3 O − ) or via fluorine ion transfer resulting in m/z MW+19 (HF • R − −H ).The dominant ionization mechanism depends mostly on the acidity of the neutral species; highly acidic species such as nitric acid only form the fluorine transfer ion, while non-acidic species such as methyl hydrogen peroxide form only the cluster ion.This separation aids both in the determination of the structure of a molecule and in the identification of isomers.In negative mode, tandem mass spectrometry (MS/MS) was used to help identify functional groups of an analyte.In brief, a parent ion selected in the first quadrupole is exposed to an elevated pressure of N 2 resulting in collision-induced dissociation (CID) in the second quadrupole, and the resulting fragmentation ions are detected in the third quadrupole.Molecules with different functional groups have been shown to fragment differently by CID and thus the detection of certain fragment ions in quadrupole three can aid in the identification of an analyte.For instance, hydroperoxides form a characteristic fragment at m/z 63 (Paulot et al., 2009b).
Standards are not available for most of the VOCs described here and thus the sensitivity of the CIMS is related to the thermal capture rate and the binding energy of the cluster (VOC • CF 3 O − ).Details on calculating the sensitivity of the CIMS to a given analyte can be found in previous publications (Paulot et al., 2009a,b).

Results and discussion
The goal of this study is to determine the gas-phase reaction products and mechanism of the photooxidation of α-pinene by OH under both low-NO x and high-NO x conditions.In both cases, the experiment is designed such that α-pinene only reacts with OH and that a single peroxy radical reaction is dominant.Under ambient conditions, α-pinene reacts at an approximately equal rate with O 3 (k O 3 = 9.0 × 10 −17 cm 3 molecules −1 s −1 ) and OH (k OH = 5.3 × 10 −11 cm 3 molecules −1 s −1 ) (Sander et al., 2006;Atkinson et al., 2006).To isolate the OH chemistry, the formation of O 3 was suppressed.
Under low-NO x conditions, photolysis of H 2 O 2 resulted in steady state OH concentration of ∼2 × 10 6 cm −3 and an HO 2 concentration ∼1 × 10 10 cm −3 .The OH and HO 2 concentrations were determined from a kinetic molecular model of αpinene OH reaction which is described in detail in Sect.3.2 and in the Supplement.In brief, OH is produced by the photolysis of H 2 O 2 and is primarily consumed by reactions with VOCs in the chamber and with H 2 O 2 .The reaction of OH  S1).These OH concentrations are similar to those observed in the troposphere; however, the HO 2 concentrations are about an order of magnitude greater than levels typically observed in rural or remote areas (Lelieveld et al., 2008;Ren et al., 2008;Wolfe et al., 2011).The O 3 concentration at the start of each experiment was <4 ppb and typically did not exceed 12 ppb over the course of an experiment (in one experiment, conducted for 20 h, the O 3 concentration increased nearly linearly from 4 to 25 ppb).Modeling of these oxidant concentrations indicated that reaction with O 3 accounted for <3 % of the α-pinene loss.
In the oxidation of α-pinene by OH, OH adds predominantly to the endocyclic double bond, followed by addition of O 2 to the resulting β-hydroxy alkyl radicals producing a number of hydroxy peroxy radicals.OH abstraction of a hydrogen from α-pinene occurs with a yield of ∼12 %, resulting in a peroxy radical when O 2 reacts with the alkyl rad-ical (Capouet et al., 2004).Under low-NO x conditions the peroxy radicals can react with either HO 2 or with another RO 2 (self or cross-reactions).Here we seek to emulate atmospheric conditions where RO 2 + HO 2 dominates.To confirm that the RO 2 + HO 2 pathway dominated in our experiments, a kinetic model was constructed, as described in Sect.3.2 and the Supplement.For the experimental conditions (with initial concentration of α-pinene of 20-50 ppb), the kinetic model indicates that less than 1 % of the peroxy radical reactions proceed via RO 2 + RO 2 .
For high-NO x photooxidation experiments, OH is generated from the photolysis of either HONO or methyl nitrite.When HONO is synthesized, NO and NO 2 are produced as side products, which are also introduced to the chamber when HONO is injected.Methyl nitrite is synthesized pure and upon photolysis eventually produces OH and NO 2 : Fig. 2. Pathway of photooxidation of pinonaldehyde under low-NO x conditions.Species that were observed are labeled in red.Pinonic acid and pinonic peracid were observed in the negative mode as both the transfer ion (molecular mass + 19) and the complex (molecular mass + 85), pinonic peracid is also observed as the complex minus HF (molecular mass + 65), norpinonaldehyde hydroperoxide was observed as the complex ion, pinonaldehyde and norpinonaldehyde are observed in the positive mode at their molecular mass +1.
When methyl nitrite was used as the OH source, 300 to 400 ppb of NO was added to suppress O 3 production.Under the conditions used, the peroxy radicals formed from the OH reaction of α-pinene react with NO and not RO 2 or HO 2 .This is confirmed by the lack of RO 2 + HO 2 or RO 2 + RO 2 reaction products during high-NO x photooxidation.The exact ratio of NO to NO 2 is not known as both HONO and methyl nitrite interfere with the signals from the chemiluminescence NO x instrument; however, it was determined that a higher NO 2 concentration exists during methyl nitrite photolysis due to the increased production of peroxyacyl nitrates (PANs), as detected by CIMS in the negative mode, which result from the RO 2 + NO 2 reaction.More details on the detection of, as well as the specific PANs detected will be discussed in Sect.3.3.

Gas phase composition from low-NO x photooxidation of α-pinene
The principal first-generation oxidation products observed from the low-NO x photooxidation of α-pinene were a number of α-pinene hydroxy hydroperoxides and pinonaldehyde.Production of α-pinene hydroperoxide from H-abstraction and α-pinene oxide are also observed (see Fig. 1 for proposed reaction mechanism for α-pinene and Fig. 2 for pinonaldehyde).The time traces from the CIMS are shown in Fig. 3 along with those from a number of the minor oxidation products.The identification of each of these signals is discussed below.
The α-pinene hydroxy hydroperoxides are observed at m/z (−)271 (molecular weight + CF 3 O − ) by the CIMS.They are expected to be a major products of the reaction channel of RO 2 + HO 2 .Three different α-pinene hydroxy hydroperoxide isomers are formed; two β-hydroxy hydroperoxides and one ring opened hydroxy hydroperoxide containing a double bond (see Fig. 1).The subsequent reaction pathways of these three α-pinene hydroxy hydroperoxides are expected to be different and will lead to distinct reaction products, as discussed in Sect.3.2.The overall estimated initial yield of the α-pinene hydroxy hydroperoxides is ∼23 %.This estimate accounts for the small yield of pinic acid, which has the same molecular weight as these peroxides (MW = 186).Similar to other carboxylic acids, ionization of pinic acid by CF 3 O − yields approximately equal amounts of signal at m/z (−)271 and 205 (MW −H + HF).This allows us to use the signal at m/z (−)205 to insure that the pinic acid concentration does not significantly impact the hydroperoxide yield estimate.
Pinonaldehyde is neither acidic nor is the complex with CF 3 O − significantly strong to be detected in the negative mode and thus is observed only in the positive mode of the CIMS.Pinonaldehyde was synthesized to directly observe its OH oxidation; however, the synthesized sample was not of sufficient purity to calibrate the CIMS response.While we cannot report an absolute yield for pinonaldehyde, we observe that the yield under low-NO x conditions is 2/3 of that under high-NO x conditions.As pinonaldehyde is one of the main products of α-pinene oxidation by any atmospheric oxidant, it has been widely studied.Although the reported yield of pinonaldehyde under high-NO x conditions varies widely from 27-87 % (Arey et al., 1990;Hatakeyama et al., 1991;Noziere et al., 1999;Wisthaler et al., 2001;Jaoui and Kamens, 2001;Aschmann et al., 2002;Lee et al., 2006) most of the measured yields are between 27-35 %.The very high yields were measured by FTIR which may be biased by interference from other carbonyls (Noziere et al., 1999;Hatakeyama et al., 1991).Assuming that under high-NO x conditions the pinonaldehyde yield is between 27-35 %, we estimate that the low-NO x yield is about 20 %.
Pinonaldehyde has been observed from the low-NO x photooxidation of α-pinene previously; however, it is typically assigned as a product of RO 2 + RO 2 chemistry (Noziere et al., 1999;Larsen et al., 2001).In the present set of experiments, the chemistry is overwhelmingly dominated by RO 2 + HO 2 (i.e.low α-pinene concentration and relatively high HO 2 concentration).This is confirmed by the photooxidation products observed.The two main reaction channels for alkyl peroxy radicals with other RO 2 are: resulting in the formation of alcohols and carbonyls from Reaction (R4) and alkoxy radicals from Reaction (R5).In the case of α-pinene, these would be β-diols, β-hydroxy carbonyl, and hydroxy alkoxy radicals.The six membered ring of the hydroxy alkoxy radical will then open forming pinonaldehyde.There is no indication of either β-diols or β-hydroxyl carbonyl being formed during the present experiments.In the case of alkyl peroxy radicals reacting with HO 2 , the main reaction channel is: which forms hydroperoxides.For α-pinene, Reaction (R6) will produce the observed α-pinene hydroxy hydroperoxides.For many small alkyl peroxy radicals it has been shown to be the only channel (Hasson et al., 2004;Raventos-Duran et al., 2007;Noell et al., 2010).
If Reaction (R6) was the only channel, there would be no route to pinonaldehyde production.One possible route to pinonaldehyde is a radical channel (see Fig. 1): similar to those known to be important for acetyl peroxy radicals (Hasson et al., 2004;Dillon and Crowley, 2008) and possibly for the reaction of OH with toluene (Birdsall et al., 2010).If this is the route to pinonaldehyde formation, an OH recycling channel would also be of importance in the reaction of OH with α-pinene under low-NO x conditions.α-pinene hydroperoxide and α-pinene oxide are also observed products of the OH oxidation of α-pinene under low-NO x conditions.α-pinene hydroperoxide can be formed from H-abstraction by OH, which is estimated to account for about 12 % of the OH reaction with α-pinene (Capouet et al., 2004).Assuming similar CIMS sensitivity to α-pinene hydroperoxide and α-pinene hydroxy hydroperoxide, the yield of α-pinene hydroperoxide is estimated to be ∼6 %.The remainder of the H-abstraction branching ratio cannot be accounted for at this time.α-pinene oxide is neither acidic nor is the complex with CF 3 O − significantly strong to be detected in the negative mode and thus is observed only in the positive mode at m/z (+)153 and 171 as confirmed by direct injection of α-pinene oxide into the chamber.Further evidence of α-pinene oxide formation is presented in part 2 of this series of paper where SOA composition is discussed.α-pinene oxide appears to be a minor product, and the mechanism for its formation by OH oxidation is not known.The mechanism for the formation from O 3 oxidation is known and yields of a few percent have been reported (Alvarado et al., 1998;Berndt et al., 2003); however, in the present experiments O 3 accounts for less than 3 % of the oxidation of α-pinene.
The oxidation products of pinonaldehyde have been shown to be important in the formation of SOA from the photooxidation of α-pinene.Specifically, pinonic acid, 10hydroxypinonic acid, and pinic acid have been observed in ambient SOA samples (Kavouras et al., 1998(Kavouras et al., , 1999;;Yu et al., 1999;Laaksonen et al., 2008;Zhang et al., 2010) and in laboratory studies of α-pinene photooxidation (Larsen et al., 2001;Jaoui and Kamens, 2001;Librando and Tringali, 2005).The low-NO x photooxidation of pinonaldehyde was carried out to determine the contribution of pinonaldehyde oxidation products to the gas and particle phases of α-pinene photooxidation.
Photooxidation of pinonaldehyde accounts for a number of the products observed from α-pinene photooxidation.OH oxidation of pinonaldehyde occurs by H-abstraction, typically from the aldehydic group (59-86 %) (Kwok and Atkinson, 1995;Vereecken and Peeters, 2002) forming an acyl peroxy radical after the addition of O 2 .The reaction of HO 2 with an acyl peroxy radical produces carboxylic acids, peroxyacids, and acetoxy radicals that will decompose to produce CO 2 and an alkyl radical.Hasson et al. (2004) have shown that these three pathways occur in roughly equal yields.The reaction pathways and products from pinonaldehyde are shown in Fig. 2. Pinonic acid is observed by the CIMS mostly by its transfer product ion (ratio of m/z (−)203 to m/z (−)269 is ∼9:1), typical of acidic species, while the peroxyacid is observed more by its complex ion (ratio of m/z (−)219 to m/z (−)285 is ∼1:4) (Fig. 4).In addition, we have observed, using synthesized peracetic acid, that peroxyacids are also observed in the negative mode as its molecular mass +65 (complex with CF 3 O − − HF) at 10-20 % that of the complex ion (+85).A signal at m/z (−)265, ∼10 % the intensity of m/z (−)285, has been assigned to pinonic peroxyacid providing further evidence of its formation.These two products are formed in roughly equal yields.H-abstraction at carbons other than the aldehydic carbon could also form pinonaldehyde hydroperoxide which would also be detected at m/z (−)285, as would 10-hydroxy pinonic acid (product found in SOA).A signal at m/z (−)257 that has nine carbons was observed and has been assigned to the hydroperoxide formed from the alkyl radical formed from the decomposition of the acetoxy radical.The formation of norpinaldehyde was also detected.Pinic acid was observed in equal amounts at m/z (−)205 and m/z (−)271, but is a minor product.
Low-NO x photooxidation of α-pinene also leads to the pinonaldehyde oxidation products (Fig. 4).Ions at m/z (−)203 and (−)219 assigned to the transfer ions of pinonic acid and pinonic peroxyacid are observed to be second generation products from α-pinene photooxidation, as would be expected arising from the oxidation of pinonaldehyde.The ion at m/z (−)257 assigned to the C 9 hydroperoxide produced from pinonaldehyde photooxidation is also observed to be second generation.Pinonic acid and pinonic peroxyacid are also observed as the complex ion at m/z (−)269 and (−)285 respectively.The time traces for m/z (−)269 and m/z (−)285 from the photooxidation of α-pinene are not purely from first or second generation products (see Figs. 3 and 4) but a combination of the two.The second generation products observed as these two ions are pinonic acid and pinonic peroxyacid, evident from their transfer ions, while the first generation products arise from the photooxidation of α-pinene.Oxidation products of these molecular mass (184 and 200) have been observed and theoretically proposed from the photooxidation of α-pinene in the presence of NO (Aschmann et al., 2002;Vereecken et al., 2007).Under high-NO x conditions, Aschmann et al. ( 2002) observed molecules at these masses and assigned them to a dihydroxy carbonyl and a trihydroxy carbonyl that would be formed by a number of isomerization steps as well as multiple reactions of peroxy radicals with NO to form alkoxy radicals.Vereecken et al. (2007) indicate that molecules with these molecular weights could be formed by isomerization reactions after ring opening, while the molecule with MW = 200 is produced from a reaction of NO with the peroxy radical.In both mechanisms, alkoxy radicals are essential.Alkoxy radicals appear to form from the reaction of the initially formed αpinene hydroxy peroxy radical intermediates with HO 2 , and this is a possible route; however, at this point, the structure of these molecules is unknown.These are, however, clearly polyoxygenated species formed from the primary oxidation Highly oxidized species at m/z (−)301 and (−)303 are also observed as first generation products (see Fig. 3).Both signals decay similarly to the signals at m/z (−)269 and (−)285, indicating that they are also formed as later generation products.The most likely structure for the first-generation species at m/z (−)303 is an α-pinene hydroxy hydroperoxide with a bridging peroxy group that results from the ring closing channel of the ring opened hydroxy hydroperoxide as shown in Fig. 1.This mechanism has been postulated by Vereecken et al. (2007).The addition of one hydroxyl group and two hydroperoxyl groups decreases the vapor pressure about eight orders of magnitude making this product a likely aerosol phase component (Capouet and Müller, 2006).The photooxidation of α-pinene hydroperoxide will produce a hydroxy dihydroperoxide that will also be observed at m/z (−)303 thus accounting for the slow decay of the signal.We do not know the structure of the photooxidation product(s) at m/z (−)301.Given the molecular weight, the structure potentially includes one carbonyl and two hydroperoxyl groups.Whatever the structure, the vapor pressure will be greatly reduced from that of α-pinene, once again making it a potential SOA component.
Another route to the formation of the low vapor pressure oxidation products at m/z (−)301 and (−)303 is peroxy radical isomerization.Peroxy radical isomerization has recently been extensively studied for the isoprene system both theoretically (Peeters et al., 2009;da Silva et al., 2010;Nguyen et al., 2010) as well as experimentally (Crounse et al., 2011).It has been shown that peroxy radical isomerization most often occurs via a six or seven membered intermediate (Perrin et al., 1998;Blin-Simiand et al., 2001;Jorand et al., 2003).When OH adds to the tertiary carbon of the double bond the peroxy radical on the secondary carbon can form a seven membered ring with the hydrogen of either the secondary carbon in the four membered ring or one of primary carbons attached to the four membered ring.After isomerization, reaction with O 2 and HO 2 will result in the formation of α-pinene hydroxy dihydroperoxide which would be detected at m/z (−)303.See Fig. S2 in the Supplement for proposed mechanism.When OH adds to the secondary carbon of the double bond, a seven membered intermediate can form with the other tertiary carbon of the four membered ring ultimetly forming α-pinene hydroxy dihydroperoxide.Similarly, the α-pinene hydroxy hydroperoxide peroxy radical formed from isomerization from the addition of OH to the secondary carbon can form a six membered ring with the hydrogen on the carbon α to the hydroxyl group.This isomerization would result in the formation of α-pinene carbonyl dihydroperoxide which would be observed at m/z (−)301.For the species observed at m/z (−)301 and (−)303 to be formed via isomerization, the rate of peroxy radical isomerization would have to be competitive with that of RO 2 + HO 2 .In the present experiments [HO 2 ] is ∼1 × 10 10 molecules cm −3 so the rate of loss of α-pinene hydroxy peroxy radical by reaction with HO 2 is ∼0.2 s −1 .In the atmosphere the HO 2 concentration is about an order of magnitude less, so for isomerization to be competitive (10 % of peroxy radical reaction) the rate would have to be ∼0.002s −1 .In the isoprene system the isomerization rate has been found to be 0.002 s −1 with conformer specific rates being much faster (Crounse et al., 2011).At this time there have been no experimental studies on the isomerization rate in the α-pinene system so it is unknown what the rate is and therefor its atmospheric relevance.If the rate of isomerization is competitive with reaction with HO 2 in the atmosphere, more low vapor pressure species capable of partitioning into the aerosol will be formed than were observed in this study.

Kinetic model of low-NO x photooxidation of α-pinene
To gain a better understanding of the low-NO x photooxidation of α-pinene, a kinetic model was assembled and compared to the time traces of α-pinene and the reaction products measured during the photooxidation of 19.8 ppb α-pinene and 4 ppm H 2 O 2 .The kinetic model was constructed using Kintecus modeling software (Ianni, 2002).When available, rate constants from literature were used; however, a majority of the rate constants are not known.In these cases, reaction rates with respect to OH oxidation were estimated using the structural activity relationship derived by Kwok and Atkinson (Kwok and Atkinson, 1995).This method has been shown to predict rate constants to within a factor of two for most species (see Tables S1 and S2 for lists of photooxidation products, reactions, and reaction rates used in this model).
The rate constants were then modified slightly to best fit the data.The rate constants for RO 2 + HO 2 and RO 2 + RO 2 were taken from the estimates used for the Master Chemical Mechanism (Saunders et al., 2003).Photolysis was included for hydrogen peroxide, the organic hydroperoxides, and for pinonaldehyde.The photodissociation frequencies (j) were determined from the spectral radiance measured from the chamber lights and the absorption cross sections reported in the literature for hydrogen peroxide and pinonaldehyde and estimated for the organic hydroperoxides from reported cross sections of similar hydroperoxides (Atkinson et al., 2006;Sander et al., 2006).
For the model to be accurate, the concentration of OH and HO 2 need to be correct.OH is produced from photolysis of H 2 O 2 which initially reacts with hydrocarbons in the chamber as well as H 2 O 2 .It is the reaction of OH with H 2 O 2 that produces HO 2 .The value of j H 2 O 2 was confirmed by comparing the simulated loss of H 2 O 2 (photolysis and reaction with OH) to the time trace observed from the CIMS (see SI Fig. S1).The comparison of the simulated loss of α-pinene, using the recommended rate constant for the reaction of α-pinene with OH from the IUPAC database (Atkinson et al., 2006), with the the observed loss of α-pinene also confirms Represented are α-pinene (purple), α-pinene hydroxy hydroperoxide (black), pinonaldehyde (red), α-pinene hydroperoxide (blue), αpinene hydroxy dihydroperoxide (green), and the C 9 hydroperoxide from pinonaldehyde (orange).See the Supplement for structural information and kinetic model.the simulation produces the correct amount of OH (see Fig. S1 in the Supplement).The modeled OH concentration quickly reaches 2.2 × 10 6 molecules cm −3 and is approximately constant between 2.2-2.5 × 10 6 molecules cm −3 over the course of the experiment, while the HO 2 concentration rapidly plateaus at about 1 × 10 10 molecules cm −3 .In addition, the rate constant of OH with pinonaldehyde was confirmed from the experiment where pinonaldehyde was directly injected into the chamber.Pinonaldehyde is lost by reaction with OH as well as photolysis (∼10 % of overall reaction), and the best fit to the data is achieved with k pinonaldehyde+OH = 3.7 × 10 −11 cm 3 molecules −1 s −1 (Fig. S3 in the Supplement).This rate constant is within the error recommended by the IUPAC database and the latest measurement of this reaction rate by Davis et al. (2007); Atkinson et al. (2006).
Figure 5 shows the time traces of α-pinene and several of the photooxidation products along with the simulated concentrations from the kinetic model.The oxidation products in Fig. 5 are representative of first-generation products (α-pinene hydroxy-hydroperoxides, pinonaldehyde, and αpinene hydroperoxide) and a second-generation product (C 9 hydroperoxide from pinonaldehyde) (see Table S1 for structural information of each compound).The one exception is the highly oxygenated compound observed at m/z (−)303 (assigned to hydroxyl dihydroperoxide) that is both a firstand second-generation product, and due to its high degree of oxidation and low volatility is predicted to partition into the particle phase.As noted in Sect.3.1, pinic acid is a minor second generation product that is observed at m/z (−)205 and m/z (−)271 in roughly equal amounts.This small yield was accounted for in determining the concentration of α-pinene hydroxy hydroperoxides.Using the fastest RO 2 + RO 2 rate constant for α-pinene hydroxy peroxy radicals from MCM, only 0.05 % of the peroxy radical is predicted to react through the RO 2 + RO 2 channel.Including all of the identified first-generation products (these presented here along with the products at m/z (−)269, 285, and 301), 65-75 % of the carbon can be accounted for, and this does not include compounds that are expected to partition into the aerosol phase.
Pinonaldehyde, in the present set of experiments, is formed from the RO 2 + HO 2 reaction channel of two of the α-pinene hydroxy peroxy radicals, and the reaction rate with OH has been well characterized.However, based on these formation and loss mechanisms for pinonaldehyde, the simulated pinonaldehyde decays far too quickly; therefore, pinonaldehyde must also be produced as a second-generation oxidation product.Of the first-generation products, the most likely candidates to produce pinonaldehyde from a reaction with OH are the α-pinene hydroxy hydroperoxides.
There are a number of paths by which the α-pinene hydroxy hydroperoxides will form pinonaldehyde. First of all, the hydroperoxide will not only react with OH but will also undergo photolysis.Photolysis cleaves the weak O-O bond of the hydroperoxide producing an alkoxy radical that will produce pinonaldehyde.We estimate, however, that photolysis represents ∼3 % of the α-pinene hydroxy hydroperxide loss and thus does not provide mass closure between the data and the simulation.
Figure 6 shows the three hydroxy hydroperoxides that are formed from low-NO x photooxidation of α-pinene, two βhydroxy hydroperoxides and a ring-opened one that includes a double bond.The β-hydroxy hydroperoxides have a number of places where OH hydrogen abstraction could occur (Fig. 6), the carbon α to the alcohol; carbon α to the hydroperoxyl group; either of the tertiary carbons; the alcoholic hydrogen; or the hydroperoxy hydrogen.Hydrogen abstraction from an alcohol will mostly occur from the α carbon which has been shown to result in the formation of a carbonyl and HO 2 (Carter et al., 1979;Atkinson, 1986;Grosjean, 1997).Presumably this would form a hydroperoxy carbonyl, a mass analog of the known second-generation product pinonic acid.As there are first-and second-generation products with m/z (−)269, it is not clear how important this H-abstraction channel is.There have only been a few hydroperoxides whose H-abstraction reaction rates and product distributions have been studied: methyl, ethyl and tertbutyl (Vaghjiani and Ravishankara, 1989;Niki et al., 1983;Wang and Chen, 2008;Baasandorj et al., 2010).It has been shown that the OO-H bond is fairly labile and H-abstraction from both the hydroperoxide as well as the α-carbon will occur.H-abstraction from the hydroperoxyl group will result in reformation of the peroxy radical with the same branching ratio of products from the initial reaction forming, among other species, pinonaldehyde and hydroxy hydroperoxide.H-abstraction from the α-carbon of the hydroperoxy group would presumably form a hydroxy carbonyl, which is a mass analog of α-pinene hydroperoxide.In the model, the signal associated with α-pinene hydroperoxide is fit well with a production channel from α-pinene and a reaction rate with OH that is nearly as fast as α-pinene itself, suggesting that it is unlikely that there is a second-generation product formed at this mass.It is possible that following hydrogen abstraction, the ring is cleaved, ultimately resulting in the formation of pinonaldehyde.Finally, H-abstraction from either of the tertiary carbons will produce more highly oxygenated species, including a hydroxyl dihydroperoxide (m/z (−)303).Of these different H-abstraction pathways, those involving the hydroperoxide and carbons α to the hydroperoxy and hydroxyl groups are expected to dominate.The ring opened hydroxy hydroperoxide has a highly reactive double bond instead of one of the tertiary carbons.After addition of OH to the double bond, the reaction will most likely proceed in one the following ways: formation of a dihydroxy dihydroperoxide or intermolecular reaction of the alkyl radical with the hydroperoxy group forming a ring with one oxygen and releasing OH.The former product has not been observed in the present experiments; however, this highly oxidized compound has a low vapor pressure and is expected to primarily reside in the particle phase.The later product is a mass analog of the hydroxy hydroperoxide, and the two species are indistinguishable by CIMS.
In the simulation, all the pathways for reaction of the α-pinene hydroxy hydroperoxides were considered and the contribution of each was varied until the simulation best fit the data based on a least-squares analysis of α-pinene hydroxy hydroperoxide and pinonaldehyde.The best fit results when 58 % of the reaction produces pinonaldehyde directly while another 25 % reforms the peroxy radical, resulting in a net ∼33 % of α-pinene producing pinonaldehyde from low-NO x photooxidation.This is important since the oxidation products of pinonaldehyde are main SOA components.There is need for further studies on the photooxidation of α-pinene hydroxy hydroperoxides to confirm these reaction pathways and branching ratios; however, the data and kinetic model strongly indicate that pinonaldehyde is produced from the OH oxidation of α-pinene hydroxy hydroperoxides under low-NO x conditions.

Gas-phase composition from high-NO x photooxidation of α-pinene
Figure 7 shows the proposed reaction mechanism of αpinene photooxidation under high-NO x conditions.This mechanism and the following discussion of the reaction products are not intended to be all-inclusive.The photooxidation of α-pinene under high-NO x conditions produces a large variety of compounds through functionalization and fragmentation reactions.We observe signals at many masses but, for brevity, we only focus on those of the major products of α-pinene and pinonaldehyde high-NO x OH oxidation.
In the presence of NO x , the main photooxidation products of α-pinene are pinonaldehyde, organonitrates, and isomerization products from the alkoxy radical formed from the reaction of NO with the peroxy radical.The yield of pinonaldehyde is 1.5 times that formed in the absence of NO x , and, as mentioned above, previous studies show that it is 27-35 %.Capouet et al. (2004) estimates α-pinene hydroxy nitrate yield to be ∼15 %.Alkoxy radical isomerization products are observed at m/z (−)269 (MW 184) and m/z (−)285 (MW 200).These isomerization products have been observed previously and have been assigned to either a dihydroxy carbonyl and a trihydroxy carbonyl (Aschmann et al., 2002) or to ring opened substituted hydroperoxides (Vereecken et al., 2007).In the present study, the yield of the isomerization products was found to be less than 20 %.The OH addition to the double bond can also result in prompt ring opening of the four-membered ring.Further reactions of the ring-opened alkyl radical will result in fragmentation producing an alkyl radical and acetone (see Fig. 7).This is a minor pathway, as the total acetone yield observed was less than 10 %, including the production of acetone from the photooxidation of pinonaldehyde.Finally, there is a first-generation species observed at m/z (−)301, which was also present during low-NO x photooxidation.A likely molecular formula is C 10 H 16 O 5 , but the identity of this molecule is unknown.The same oxidation products were observed from high-NO x photooxidation, regardless of OH source, HONO or methyl nitrite.
The photooxidation of pinonaldehyde was studied using methyl nitrite as the OH source.Figure 8 shows the proposed mechanism and products of this reaction.There have been two theoretical studies on the H-abstraction of pinonaldehyde (Kwok and Atkinson, 1995;Vereecken and Peeters, 2002).Kwok and Atkinson (1995) indicate that 86 % of the abstraction will occur at the aldehydic hydrogen, while Vereecken and Peeters (2002) determined that 59 % will be of the aldehydic hydrogen, 23 % from the carbon β to the aldehydic carbon, and 14 % from the tertiary carbons.The products expected, and observed, of the primary channel (Habstraction of the aldehydic hydrogen) are pinonaldehyde PAN, formed from reaction of the acyl peroxy radical with NO 2 or norpinonaldehyde, formed from reaction of the acyl peroxy radical with NO (see Fig. 8 for structures and mechanisms).H-abstraction from the carbon β to the aldehydic carbon produces either pinonaldehyde nitrate or norpinonaldehyde with the loss of CO.Finally, H-abstraction from the tertiary carbons produces pinonaldehyde nitrate or a tertiary alkoxy radical.One potential reaction pathway for the tertiary alkoxy radical is the decomposition to form acetone (see Fig. 8), which is observed in low yield from high-NO x photooxidation of pinonaldehyde.Once formed, norpinonaldehyde will be oxidized in similar fashion to pinonaldehyde.The signal for pinonaldehyde PAN is 4-8 times that of pinonaldehyde nitrate (from all high-NO x experiments, α-pinene and pinonaldehyde) depending on the NO 2 concentration relative to NO.The observed signal of the norpinonaldehyde PAN to norpinonaldehyde nitrate from the photooxidation of pinonaldehyde was greater than 14 to 1. Pinonaldehyde PAN is formed from an acyl peroxy radical while pinonaldehyde nitrate will be formed from an alkyl peroxy radical.The formation of an acyl peroxy radical is favored over the formation of an alkyl peroxy radical from H-abstraction of pinonaldehyde and thus the formation of pinonaldehyde PANs is expected to be formed in higher yield than pinonaldehyde nitrate.A quantitative yield of the PAN is not known because the CIMS has not been calibrated for these compounds.The product distribution from the photooxidation of pinonaldehyde indicates that it does not add much to the total organonitrate yield from the photooxidation of α-pinene.Pinonic acid and pinic acid were not observed in the gas phase from either α-pinene or pinonaldehyde photooxidation under high-NO x conditions.An array of organonitrates and PANs are observed from the photooxidation of α-pinene.In addition to those previously discussed, nitrates were observed at m/z (−) 276, 316, 318, 320, and 332  formed in much lower yields than those identified earlier.Nitrates and PANs are distinguished from each other by the nature of their decay.PANs thermally decompose so their gas-phase concentration will continue to decrease even after OH is no longer present.Norpinonaldehyde PAN is observed at m/z (−)316; however, comparing the time traces of this ion from pinonaldehyde photooxidation and α-pinene photooxidation, there appears to be a product formed from α-pinene as well.This could be the α-pinene dihydroxynitrate proposed by Aschmann et al. (2002).One interesting observation is that α-pinene hydroxy nitrate is lost from the gas phase at nearly an equal rate as pinonaldehyde.This is in contrast to the other organonitrates, whose signals are stable once they are formed.The products of high-NO x photooxidation of α-pinene are the same regardless of NO 2 level; however, there are a few differences in concentrations of certain species depending on the level of NO 2 .The most noticeable difference in the gasphase composition depending on the [NO 2 ] is the concentration of PANs, which are greater at high NO 2 (methyl nitrite photolysis).This is expected as the NO 2 concentration is higher in the methyl nitrite experiments resulting in a greater likelihood of NO 2 reacting with the acyl peroxy radicals produced from pinonaldehyde and norpinonaldehyde.The concentrations of isomerization products at m/z (−)269 and (−)285 showed a dependence on seed particle acidity when HONO was used as the OH source but not when CH 3 ONO was used as the OH source.In the case of the HONO source, the isomerization products were observed to be lower in the presence of an acidic seed.A discussion of why the products observed at m/z (−)269 and (−)285 are dependent on aerosol acidity only for the HONO case will be presented in part 2. The only other observed difference in the gas phase was the concentration of nitric acid, which was 1.5 times greater when methyl nitrite was the OH source then when HONO was the OH source.

Atmospheric implications
Under low-NO x photooxidation of α-pinene, pinonaldehyde is produced by RO 2 + HO 2 chemistry in high yields.RO 2 + RO 2 chemistry is not the source, as is commonly assumed.While the atmospheric concentrations of HO 2 and RO 2 are similar, the rate constant of RO 2 + HO 2 is typically 1-4 orders of magnitude greater than RO 2 + RO 2 .The further OH oxidation of pinonaldehyde produces a number of low volatility carboxylic acids and peroxyacids, which are important SOA components.While the only product with higher yield than pinonaldehyde is α-pinene hydroxy hydroperoxide, pinonaldehyde is also produced from the photolysis and OH oxidation of α-pinene hydroxy hydroperoxide.Including Atmos.Chem.Phys., 12, 6489-6504, 2012 www.atmos-chem-phys.net/12/6489/2012/ the reaction channels of α-pinene hydroxy hydroperoxide that form pinonaldehyde (58 %) and assuming 19 % direct yield from α-pinene, the overall oxidation of α-pinene by OH under low-NO x conditions has a pinonaldehyde yield of ∼33 % .Pinonic acid is an important tracer for SOA formed from α-pinene oxidation and is believed to be the precursor for other SOA tracers, such as 3-methyl-1,2,3butanetricarboxylic acid (Szmigielski et al., 2007;Müller et al., 2012).Pinonic acid is not observed from the high-NO x photooxidation of α-pinene, which is not surprising since there is no clear formation mechanism for it in the gas phase.Thus, observation of pinonic acid and the tracers derived from pinonic acid likely point to either low-NO x oxidation of α-pinene by OH or ozonolysis under high-NO x conditions.The aerosol-phase composition of these carboxylic acids as well as other carboxylic acids from α-pinene photooxidation will be discussed in detail in Part 2.
The gas-phase chemistry of α-pinene and its photooxidation products is complex and not well represented by the reactions of simple alkenes.By isolating the critical peroxy radical reactions, the products and branching ratios of a number of key reactions have been better constrained.Perhaps most importantly, we suggest that the formation of pinonaldehyde from low-NO x photooxidation of α-pinene via RO 2 + HO 2 proceeds by the formation of an alkoxy radical and OH recycling.This reaction scheme has previously been shown to be important only for acyl peroxy radicals and possibly toluene.Unfortunately, the nature of the potential energy surface leading to radical recycling in the reactions of HO 2 with peroxy radicals has not been fully formulated.Is the OH recycling channel available due to the β hydroxyl group withdrawing electron density from the peroxy group, the bicyclic ring structure of α-pinene, or some other reason all together?To date, most of the detailed work on the reaction of alkyl peroxy radicals with HO 2 has focused on small, unsubstituted peroxy radicals; further work on a broader set of peroxy radicals, including atmospherically relevant VOCs such as monoterpenes and sesquiterpenes, is clearly needed to elucidate the physical chemistry.Finally, we suggest that there is a large secondary pinonaldehyde source from the photooxidation and/or photolysis of the three pinene hydroxy hydroperoxides produced from α-pinene.

Fig. 6 .
Fig. 6.Structure of the hydroxy hydroperxides formed from the low-NO x photooxidation of α-pinene, the most likely positions for reaction with OH are also shown; H-abstraction (a-e) and OH addition (f).

Fig. 8 .
Fig. 8. Pathways of photooxidation of pinonaldehyde under high-NO x conditions, H-abstraction by OH occurs predominantly at the aldehydic carbon.Ion of compounds detected by CIMS are indicated in red.

gas-phase composition of
α-pinene and their most likely molecular formulas are C 10 H 16 O 3 and C 10 H 16 O 4 .