Importance of biogenic precursors to the budget of organic nitrates : observations of multifunctional organic nitrates by CIMS and TD-LIF during BEARPEX 2009

Alkyl and multifunctional organic nitrates, molecules of the chemical form RONO 2, are products of chain terminating reactions in the tropospheric HO x and NOx catalytic cycles and thereby impact ozone formation locally. Many of the molecules in the class have lifetimes that are long enough that they can be transported over large distances. If the RONO2 then decompose to deliver NO x to remote regions they affect ozone production rates in locations distant from the original NOx source. While measurements of total RONO2 (6ANs) and small straight chain alkyl nitrates are routine, measurements of the specific multifunctional RONO2 molecules that are believed to dominate the total have rarely been reported and never reported in coincidence with ambient6ANs measurements. Here we describe observations obtained during the BEARPEX 2009 experiment including 6ANs and a suite of multifunctional nitrates including isoprene derived hydroxynitrates, oxidation products of those nitrates, 2-methyl-3-buten-2-ol (MBO) derived hydroxynitrates, and monoterpene nitrates. At the BEARPEX field site, the sum of the individual biogenically derived nitrates account for two-thirds of the 6ANs, confirming predictions of the importance of biogenic nitrates to the NO y budget. Isoprene derived nitrates, transported to the site, are a much larger fraction of the 6ANs at the site than the nitrates derived from the locally emitted MBO. Evidence for additional nitrates, possibly from nocturnal chemistry of isoprene andα-pinene, is presented.


Introduction
Nitrogen oxides affect rates of ozone production throughout the troposphere.They are emitted from soils, lightning, and natural and anthropogenic combustion, primarily as NO but also as NO 2 and HONO.In the troposphere, these compounds are oxidized to a mix of compounds including peroxynitrates (RO 2 NO 2 ), alkyl and multifunctional nitrates (RONO 2 ), and HNO 3 .The individual components of this NO y family are transported throughout the troposphere, where they can deposit, enriching soil N content, or undergo photochemistry releasing NO y and contributing to enhanced production of ozone far from their source.
Alkyl nitrates are formed from the reaction of organic peroxy radicals (RO 2 ) with NO.The major channel of this reaction forms NO 2 , Reaction (R1), the photolysis of which leads to ozone formation.A minor reaction pathway of RO 2 radicals with NO forms organic nitrates, RONO 2 , Reaction (R2).
Published by Copernicus Publications on behalf of the European Geosciences Union.
The yield (α = R2 / (R1 + R2)) of this minor, radical terminating pathway depends on the carbon chain length and structure of the VOC oxidized, and can range from a fraction of a percent for methane to 35 % for large molecules (Atkinson and Arey, 2003b).Nocturnal nitrate radical chemistry also forms organic nitrates often with large yields up to 80 % (Hallquist et al., 1999;Fry et al., 2009).The lifetime and yield of these organic nitrate species with respect to deposition and further oxidation, along with the fate of the products from oxidation, control the effect of RONO 2 on global NO x redistribution and local ozone formation.
The importance of alkyl nitrates to both global and regional ozone production has been recognized.Different chemical representations of alkyl nitrates in models, specifically isoprene nitrates, have been found to explain differences in global simulations of tropospheric ozone (Ito et al., 2007;Wu et al., 2007).Depending on the chemical mechanism used, i.e. the formation yield, lifetime, and NO x recycling of these nitrates, ozone production behavior can vary widely (Paulot et al., 2012).Treatment of alkyl nitrates as permanent sinks for NO x can disguise the sensitivity of ozone production to biogenic emissions such as isoprene, and causes much of the simulated global atmosphere to be VOC saturated (Wu et al., 2007).Farmer et al. (2011), using Mexico City as a case study, have recently demonstrated alkyl nitrate formation can complicate efforts to engineer improved air quality due to non-linear interactions of VOC and NO x emissions.
Isoprene nitrates are a central focus of studies investigating ozone production since globally, isoprene (2-methyl-1,3butadiene) emissions dwarf those of other biogenic volatile organic compounds.440-660 Tg C yr −1 are emitted to the atmosphere as isoprene (Guenther et al., 2006).However, regionally and locally, other VOCs such as monoterpenes and oxygenated VOCs can dominate reactivity.Isoprene has a lifetime of approximately 3 h at typical atmospheric hydroxyl radical concentrations of 1 × 10 6 molecules cm −3 (Atkinson et al., 2006).However, organic nitrates function to increase the spatial scale where relatively short-lived VOCs can impact NO x concentrations and ozone formation.The formation of organic nitrates from biogenic VOCs couples anthropogenic NO x emissions with naturally occuring VOC emissions.
The branching ratio between pathway (R2) and (R1) for isoprene has been studied extensively, and a range of yields for Reaction (R2) between 4.4 and 15 % have been reported (Tuazon and Atkinson, 1990;Chen et al., 1998;Chuong and Stevens, 2002;Sprengnether et al., 2002;Patchen et al., 2007;Lockwood et al., 2010;Paulot et al., 2009a).This yield has also been found to be temperature and pressure dependent (Crounse et al., 2011;Carter and Atkinson, 1989).Ambient observations over the continental US during ICARTT/INTEX-NA of alkyl nitrates attributed to isoprene are consistent with a similar range of Reaction (R2) yield (4.4 to 12 %), depending on the subsequent fate of the ni-trates (Perring et al., 2009a).Another analysis of this dataset using the MOZART chemical transport model suggested that a nitrate yield from isoprene of 4 % was the most consistent with the sum of ANs to ozone ratios (Horowitz et al., 2007).A deposition rate of the isoprene nitrates equal to nitric acid deposition was also necessary to explain boundary layer concentration observations.Horowitz et al. (2007) used, however, a rate for the reaction of HO 2 with isoprene peroxy radicals a factor of two lower than now recommended by MCM (Saunders et al., 2003) thus substantially overestimating the fraction of isoprene peroxy radicals reacting with NO.
Predictions of global ozone are very sensitive to the yield for isoprene nitrate formation.Wu et al. (2007), using the GEOS-Chem model, found a 10 % decrease in global ozone production rates by increasing the yield for isoprene nitrate formation from 4.4 % (Chen et al., 1998) to 12 % (Sprengnether et al., 2002).Paulot et al. (2012) similarly illustrated the remarkable sensitivity of tropical ozone to this yield.
The fate of isoprene hydroxynitrates, ISOPN, has been the subject of recent research.The lifetime of ISOPN, with respect to OH and ozone is relatively short due to the presence of a double bond.For example, the E-1,4-ISOPN isomer is estimated to have a lifetime of 2.9 h (Paulot et al., 2009a) and 5.33 h (Lockwood et al., 2010) at OH = 1 × 10 6 molecules cm −3 and 40 ppb O 3 , respectively.Lifetimes vary greatly amongst the possible ISOPN isomers, for the same oxidant conditions, the 1,2-IN has a lifetime of 28 h and 2.7 h with respect to OH and O 3 .The oxidation of isoprene hydroxynitrates can either release NO x as NO 2 or retain the organic nitrate functionality.Paulot et al. (2009a) have shown that approximately 50 ± 15 % (weighted average of the 8 isomers) of the isoprene nitrates recycle NO x upon further photooxidation by OH.Horowitz et al. (2007) found NO x recycling of 40 % agreed with ambient observations, using a low yield of nitrates, and a fast deposition (similar to nitric acid).A fraction of isoprene hydroxynitrates retain the nitrate functionality upon oxidation by forming methyl vinyl ketone nitrate (MVKN), methacrolein nitrate (MACRN), propanone nitrate (PROPNN), and ethanal nitrate (ETHLN) (Paulot et al., 2009a).NO x recycling levels from the ozonolysis of isoprene hydroxynitrates are unclear (Lockwood et al., 2010).
Organic nitrate yields from biogenic VOCs other than isoprene have not been studied as extensively.Alvarado et al. (1999) found a yield of 5.2 % for the dihydroxynitrate from MBO, MBON.Chan et al. (2009) report a higher yield of ∼ 10 % (±4 %).MBON contains no double bond, and the lifetime with respect to OH, assuming an ambient OH concentration of 1×10 6 molecules cm −3 is approximately 6 days (Saunders et al., 2003).Yields and lifetimes of organic nitrates from monoterpenes such as α-pinene have also been reported.Currently, these range from 1-18 % yield (Aschmann et al., 2002;Noziere et al., 1999).
Reactions of nitrate radical with biogenic VOCs, including isoprene (Rollins et al., 2009;Perring et al., 2009b), and αpinene and β-pinene (Hallquist et al., 1999;Fry et al., 2009) are also sources of organic nitrate species.From an analysis of ICARTT field data from the Eastern US, Horowitz et al. (2007) found half of the isoprene nitrates were formed from isoprene + NO 3 chemistry, despite a small fraction (∼ 6 %) of isoprene oxidized by NO 3 chemistry.Brown et al. (2009) using ambient data from flights over the Northeast US, found that 73 % of the loss of the nitrate radical was due to the oxidation of isoprene.Using the same dataset, they inferred 2-9 % of total reactive nitrogen was composed of organic nitrates formed from this reaction, using a nitrate yield of 80 % for the isoprene + NO 3 reaction (Barnes et al., 1990).
Ambient measurements of specific individual RONO 2 have, with few exceptions (Fischer et al., 2000;Grossenbacher et al., 2001Grossenbacher et al., , 2004)), been limited to simple alkyl nitrates.As a result of the development of thermal dissociation laser induced fluorescence (TD-LIF) by Day et al. (2002), measurement of total alkyl-nitrates ( ANs) as a class has become routine.In this study, we use chemical ionization mass spectrometry (CIMS) measurements interpreted as organic nitrates in combination with the sum of total alkyl nitrates ( ANs) measurements from TD-LIF to determine the absolute concentrations and speciation of organic nitrates at an urban-impacted, forested measurement site.The dataset was collected as part of the Biosphere Effects on AeRosol and Photochemistry EXperiment, BEARPEX 2009.The measurement site is locally surrounded by dominantly non-isoprene emitting pine species; however, the site receives advected isoprene and isoprene oxidation products from the oak-rich region between the measurement site and Sacramento (Dreyfus et al., 2002).One modeling study has predicted organic nitrates from isoprene and other biogenic VOCs dominate the alkyl nitrate budget at the site (Perez et al., 2009).We report measurements of first and second generation organic nitrates from isoprene and an MBO derived nitrate.We also propose chemical structures for other nitrogen-containing compounds measured at the site.Overall, these measurements confirm that the alkyl nitrate budget at this rural site is dominated by nitrates derived from biogenic VOC precursors and that RONO 2 is an important fraction of NO y .

Site description
The measurements presented here were obtained during the Biosphere Effects on AeRosol and Photochemistry EXperiment, BEARPEX 2009.One science goal of the collaborative research campaign was to understand forest-atmosphere interactions and how the urban plume of Sacramento influences these interactions.The measurement site is located in the Western foothills of the Sierra Nevada at 38.90 • N, 120.63 • W, at 1315 m elevation, approximately 75 km northeast of Sacramento (Goldstein et al., 2000).The immediate vegetation is a Ponderosa pine forest plantation managed by Sierra Pacific Industries.Average canopy height in 2009 was 8.7 m.The dominant summertime meteorology affecting the site is a thermally driven upslope flow that brings air from California's Central Valley into the Sierra Nevada foothills; at night, this flow reverses, and returns to the Valley (Dillon et al., 2002).The major anthropogenic inputs to the plume are from vehicular traffic in Sacramento and its surrounding suburbs.Between Sacramento and the measurement site, there is a 30 km wide oak tree belt which contributes significant emissions of isoprene to the urban plume (Goldstein et al., 2000;Dreyfus et al., 2002).The measurement site contained two scaffolding towers used to deploy a variety of chemical and meteorological instrumentation along with sampling inlets.The Caltech CIMS instrumentation suite was located at the top of the North scaffolding tower, with an inlet height of 17.8 m, approximately 9 m above the top of the forest canopy.

Instrumentation
Organic nitrates were quantified using the Caltech triple quadrupole tandem chemical ionization mass spectrometer.This instrument utilizes CF 3 O − as the reagent ion, and is described in detail elsewhere (St. Clair et al., 2010).In general, this ionization chemistry has been found to be selective for acidic species, peroxides, and multifunctional (hydroxy-and nitrooxy-) molecules (Crounse et al., 2006).This instrument was also used to measure oxidation products of isoprene during the summer NASA ARCTAS campaign in 2008 (Paulot et al., 2009b).Briefly, ambient air was pulled through the inlet at 30 m s −1 , then the center of this flow was sampled into the single quadrupole CIMS instrument's reduced pressure (∼ 35 hPa) flow tube.The bypass flow was then sampled by the triple quadrupole CIMS, similar to the sampling arrangement described by St. Clair et al. (2010).The ambient flow was diluted to a 1 : 7 ratio with dry N 2 to lower the water vapor concentration in the flow tube.
The triple quadrupole CIMS instrument was operated alternately in two mass filtration modes: single mass spec (sMS) and tandem (MSMS) modes.When operating in sMS mode, the first quadrupole selected the m/z ratio to be observed, then the remaining two quadrupoles were used as ion guides.The data presented here were measured in sMS mode, since no known mass interferences exist.Each m/z ratio was observed for 0.5 s once approximately every 4-6 min.The remainder of the duty cycle was spent in MSMS mode.Instrumental background signals were measured at each mass, once approximately every 1.6 h, by scrubbing the analytes from ambient air using a two-stage filter containing palladium coated alumina pellets and sodium bicarbonate treated nylon wool (Spencer et al., 2011).The background signals were measured at approximately the same water vapor concentration as the ambient data.To determine ambient concentrations, an interpolation between these background signals was subtracted from the ambient signals.Also, starting at julian day 199, approximately every 1.6 h, ambient scans and background scans up to m/z 325 were conducted.Since this technique provides no prior separation step, no isomeric information is determined, and measurements are reported as the sum of the possible isomers (geometric or structural).
The CF 3 O − CIMS technique induces minimal fragmentation, therefore nitrates are observed at the m/z of the parent molecule + 85 (CF 3 O − ).The lack of fragmentation aids in the determination of the molecular weight of the species, since authentic standards of the organic nitrates presented here are unavailable.We use theoretically derived sensitivities to determine concentrations.Relative sensitivities of the organic nitrates to nitric acid were assumed to be equal to the relative rates of the molecule's collision with CF 3 O − .The collision rates with CF 3 O − were calculated according to Su and Chesnavich (1982) and Garden et al. (2009).Polarizibilites and dipole moment calculations were conducted using quantum chemical computations using the B3LYP/6-31G(d) level of theory (Garden et al., 2009).Comparisons of theoretical sensitivity calculations to measured sensitivities has been shown to compare well (Paulot et al., 2009b).For some analytes, the sensitivity of this technique is known to depend on water (Crounse et al., 2006;St. Clair et al., 2010).To determine the sensitivity of the nitrate clusters with the CF 3 O − reagent ion, environmental chamber experiments were performed to produce the isoprene hydroxynitrates, similar to the conditions outlined in Paulot et al. (2009a).The CIMS sampled from the sample chamber and water vapor was varied across the relevant range for this campaign, and sensitivity remained constant.Therefore, no water dependent sensitivity was applied to the ambient data.Nitrate formation yields determined from these experiments were in agreement with Paulot et al. (2009a).The estimated uncertainty for the CIMS concentrations presented here is ±40 %, dominated by the uncertainty in the theoretical predictions of sensitivity.
The signal at any m/z can contain an interference from the naturally occurring 13 C isotope signal produced from the species at the m/z −1 amu.The higher the number of carbon atoms in a molecule, the larger the contribution to the m/z signal from the 13 C isotope.It is possible to correct for this interference if a molecular formula for the species producing the m/z −1 ion is known or assumed.Specifically, the interference contribution has been removed by calculating 1.1 % * the number of carbons in the molecule that produces the m/z −1 ion * the signal at m/z −1, and subtracting this signal from the m/z signal.We evaluated this correction method using the large ambient signal at m/z 159 due to hydroxyacetone.The ratio of the signal at 160 to 159 was 0.047, consistent with the expectation from the m/z +1 isotopologue of hydroxyacetone cluster with the reagent ion.For the nitrates presented here, this correction was applied only to the PROPNN signal, observed at m/z 204, due to the large ambient signal observed at m/z 203 (known to be due to isoprene hydroxyhydroperoxide + isoprene dihydroxyepoxide) (Paulot et al., 2009a).For the other even masses, the isotopologues of the m/z −1 signals are too small to contribute to the nitrogen signals.
The sum of total alkyl nitrates, ANs, measurement was made by the University of California Berkeley using a thermal dissociation laser induced fluorescence, TD-LIF, instrument similar to that of Day et al. (2002).Briefly, the instrument measures NO 2 by LIF, and thermally dissociates most other constituents of NO y into NO 2 then measures their contribution by the difference between channels.Specifically, in this instrument ANs characteristically dissociate into NO 2 and a companion radical completely at temperatures greater than or equal to 380 • C. The difference in the NO 2 signal between ambient air passing through the 380 • oven ( ANs + PNs + NO 2 ) and the 200 • oven ( PNs + NO 2 ) is attributed to ANs.Measurements in the lab and under field conditions show that HNO 3 is not detected in the AN channel.The instrument is routinely calibrated using an NO 2 gas standard (4.9 ± 0.2 ppm NO 2 in N 2 , Praxair), diluted with zero air, and added to the system at the inlet tip.The accuracy of the NO 2 detection is ±5 % due the uncertainty in the concentration of the diluted calibration standard.The accuracy for the for the ANs measurement also includes terms for the completeness of thermal dissociation and the transmission efficiency through the inlet and is estimated following Day et al. (2002) to be 15 %.Measurement comparisons between PTR-MS and TD-LIF of an isoprene nitrate standard show the techniques to be consistent to within 10 % (Perring et al., 2009b) and similar accuracy is reported for PNs when compared to various instruments (Wooldridge et al., 2010).The minimum detectable ANs signal depends on the concentration of the adjacent channel (NO 2 + PNs) and the associated uncertainties of both channels and is estimated as 63 ppt at 1 ppb of NO 2 + PNs and 20 ppt for 100 ppt of NO 2 + PNs at S/N = 2 on an averaging time of 10 s (Day et al., 2002).The TD-LIF ANs measurement includes molecules in both the gas and aerosol phase.The sampling frequency for each class of measured NO y species is 5 Hz and the heated inlets for this instrument were also located on the top level of the North measurement tower.

Suite of organic nitrates observed
Chemical structures for the nitrates measured and presented here are shown in Fig. 1 including PROPNN, MVKN, MACRN, and ETHLN were also observed.MVKN and MACRN are isobaric, therefore, our measurement is assumed to be a sum of these species.MVKN and MACRN can also form from the oxidation of MVK and MACR.A signal attributed to MBON, a dihydroxynitrate, was also measured.Each of these species has been previously observed and identified using CIMS in laboratory experiments (Paulot et al., 2009a;Chan et al., 2009).Time series and average diurnal profiles for each of the nitrates in Fig. 1 and ANs are shown in Fig. 2. All data are presented using Pacific Standard Time (PST).Although isoprene is not a major local emission, the isoprene hydroxynitrate is the most abundant of the nitrates observed at the BEARPEX site.MBON was less abundant than all other organic nitrates except the two carbon ETHLN.Diurnal profiles shown in red are mean concentrations on days with maximum daytime temperatures of 26 °C and higher.Blue diurnal profile is the mean concentration on days with maximum temperatures less than 26 °C.This division roughly corresponds to the noticeable increase in nitrate concentrations observed around Julian day 195.Biogenic precursor emissions should be higher at the warmer temperatures; also, more rapid photochemistry is likely at the warmer temperatures (Day et al., 2008).For the isoprene hydroxynitrates and its degradation products, the diurnal profiles look very similar to advected chemical species such as anthropogenic VOCs, CO, and nitrogen oxides (Murphy et al., 2007;Day et al., 2009;Lamanna and Goldstein, 1999).These chemical species have afternoon or evening maxima between the hours of 16 and 21.MBON peaks at noon consistent with more local VOC photochemistry.The secondary MBON peak around hour 8 is similar to observations of morning peaks in NO x (Day et al., 2009) and biogenic VOCs (Schade et al., 2000).These morning peaks have been attributed to downward mixing of the nocturnal residual layer.The differences in the diurnal patterns of this suite of organic nitrates reflect the different VOC sources relative to the site location.Also, the observations of the second generation ETHLN and PROPNN indicate the ISOPN likely form upwind of the BEARPEX forest location and are oxidized during the transport to the site.
Ambient measurements of isoprene nitrates are rare, but have been conducted over two forested locations in Michigan and Tennessee using gas chromatography followed by pyrolytic conversion to and subsequent detection of NO 2 (Grossenbacher et al., 2004(Grossenbacher et al., , 2001)).This technique speciates the isomers of the isoprene nitrates, in contrast to the measurement made here which is the sum of the isomers.The two locations were characterized by vastly different concentrations of isoprene nitrates, with average, midday concentrations at the Tennessee site 5-10 times greater than observed at the Michigan forest site (mean, midday ∼ 10 pptv).The authors explained these datasets by the differences in the photochemical age of the two different airmasses.The isoprene emissions were much closer in proximity, therefore fresher and less processed at the Michigan forest site when compared with the Tennessee site.Our measurements are generally much higher than observed at either the Michigan or Tennessee sites, which can likely be explained by differences in location, season, oxidative environment, and measurement methods (individual isomers vs. the sum of the isomers).

Alkyl nitrates budget
Though ANs have been measured at this site previously, speciation of the hydroxynitrates has not been performed.A comparison of the speciated nitrates measured using CIMS and the ANs is shown in Fig. 3a.The mean ANs measurement during JD 199-212 is shown as the dashed line, with the speciated organic nitrates summed in individual colors.The diurnal profile of the total alkyl nitrate meaurement also looks similar to compounds known to be transported to the site, though the maximum does occur at noon when local chemistry and emissions would be expected to be a larger contribution.When summed, the identified organic nitrates measured with the CIMS comprise two-thirds of the total alkyl nitrates at the site, averaged from noon-06:00 p.m.The fraction shown in gray is discussed below.A scatter plot of ANs vs. the sum of the speciated nitrates (ISOPN + PROPNN + MVKN + MACRN + MBON + ETHLN) is shown in Fig. 3b.These two measurement sets are well correlated with an R 2 of 0.85.
The influence of transport on the speciation of the alkyl nitrates can be clearly seen by comparison with predictions based on local photochemical production.The production of alkyl nitrates is calculated according to Eq. ( 1), where α is the branching ratio to the formation of the organic nitrate (Rosen et al., 2004;Perring et al., 2010), and F i is the fraction of peroxy radicals reacting with NO.Data used to calculate the contributions of each VOC to alkyl nitrate production are shown in Table 1.
VOC concentrations were measured using GC-FID, and the afternoon mean from 12:00-18:00 PST is used in this analysis.This type of analysis is limited by using only the VOC measurements at the site.It is possible these VOC measurements are missing a VOC which is a large local source of nitrates or a VOC that is a large nitrate source upwind is no longer present in the air mass once it reaches the forest and is measured.
The fraction of peroxy radicals from each VOC precursor expected to react with NO, F i , has been estimated based on local, average midday NO (65 pptv) and HO 2 (20 pptv) (Mao et al., 2012) concentrations at the site, along with the ratio of rate constants for the peroxy radical reactions with NO and HO 2 .F i is calculated using the ratio of the concentrations of NO to NO plus HO 2 , weighted by the ratio of rate constants for reaction with NO and HO 2 .For both isoprene and methacrolein, recently proposed peroxy radical isomerizations compete, to varying degrees, with NO and HO 2 reactions (Peeters et al., 2009;Crounse et al., 2011Crounse et al., , 2012)).For isoprene, using the experimentally derived peroxy radical isomerization rate constant derived by Crounse et al. (2011), the fraction of isoprene-derived peroxy radicals reacting with NO decreases to 58 % (65 % is estimated if only reactions with NO and HO 2 are considered), with approximately 9 % of the peroxy radicals undergoing isomerization reactions.Using the theoretically-calculated rate of Peeters and Muller (2010), as adopted in the recent study by (Taraborrelli et al., 2012), only 10 % of the peroxy radicals would react with NO as the vast majority of these radicals are predicted to undergo isomerization, decreasing the expected contribution of isoprene nitrates to total alkyl nitrates at the site; these values are shown in parentheses in Table 1.Crounse et al. (2012) has recently shown the yield of nitrates following addition of OH to MACR is about 6 %.Because there are two channels in the OH chemistry, the effective yield from reaction of OH with MACR is only ∼ 50 % of this value (e.g. 3 %).However, Crounse et al. (2012) show that for atmospheric conditions like those at BEARPEX ([NO] < 1 ppb), essentially none of the peroxy radicals formed following addition of OH will react with NO due to fast peroxy radical isomerization reaction.Thus, we have used zero for the fraction of methacrolein derived peroxy radicals reacting with NO.The effects of these peroxy radical isomerization reactions on oxidation product distribution will be more fully explored in a future publication.
Branching ratios, α, are generally taken from MCM (v3.2) (http://mcm.leeds.ac.uk/MCM) except where noted (Saunders et al., 2003).The value of α for alkyl nitrates (to account for the formation of dinitrates or nitrate retaining oxidation products) is dependent on the distribution of the specific alkyl nitrates present.Here, we have assumed an α of 0.15 to reflect a mixture of ISOPN and other alkyl nitrates.Our calculation of the AN distribution is not very sensitive to this assumption.Daytime NO concentrations at the site were typically ∼ 65 pptv.If the alkyl nitrates at the site are mostly affected by the local production, the atmospheric con-centrations should be reflective of the alkyl nitrate production rates.Overall, based on the VOCs measured at the site, MBO is calculated to be the largest contributor (52 %) to alkyl nitrate production, with isoprene as the second most important, contributing 35 %.However, the CIMS observations show that ISOPN (38 %) and second generation nitrates from ISOPN (21 %) are a much larger contribution to the alkyl nitrate budget than MBON (7 %).Perez et al. (2009), using an MCM (v3.1) based Lagrangian model of the Sacramento urban plume, predicted 68 % of the alkyl nitrates at the measurement site to be lumped hydroxynitrates from the OH oxidation of ISOPN, 11 % ISOPN, and 17 % MBON.This highlights that the alkyl nitrates at the site are likely a mix of transported first and second generation isoprene nitrates as well as locally produced multifunctional nitrates such as MBON.Indeed, MBON is an even smaller component of the alkyl nitrate budget than predicted by the Lagrangian model of Perez et al. (2009).Both the measurements and the model predict the dominance of isoprene-derived nitrates, however, the model predicts the oxidation products of isoprene nitrates to dominate over the first generation isoprene nitrates.

Possible sources of unidentified alkyl nitrates
The speciated nitrates observed by CIMS account for approximately 65 % of the total organic nitrates observed with TD-LIF at the BEARPEX site.Two avenues were explored to investigate the possible sources of other organic nitrates contributing to ANs at the site.These include the investigation of the contributions of: (1) shorter chain, C 1 -C 3 , alkyl nitrates as measured during the BEARPEX 2007 campaign and C 1 -C 5 alkyl nitrates as measured from an airborne platform during the NASA TC 4 campaign over the BEARPEX site in 2007, (2) unidentified, nitrogen containing mass spectral signals from our CIMS data.(Atkinson et al., 2006); d (Atkinson and Arey, 2003a); e estimated; f OH = 6 × 10 6 molecules cm −3 (Bouvier- Brown et al., 2009;  Wolfe et al., 2011; Dreyfus et al., 2002); g MCMv3.2 (Saunders et al., 2003); h Chan et al. (2009); i assumed to have the same yield as the pinenes; j (Crounse et al., 2012); k estimated; l (Crounse et al., 2011); m rate constant scaled to reflect only OH addition pathways, from MCMv3.2 (Saunders et al., 2003); n values given in parentheses show the fractional contribution of each VOC to alkyl nitrate production if an isoprene peroxy radical isomerization rate constant from Peeters and Muller (2010)  Though the CIMS technique has been shown to measure two and three carbon multifunctional nitrates, the technique is not sensitive to shorter carbon chain, monofunctional alkyl nitrates such as methyl, ethyl, and propyl nitrate; it is possible some fraction of the nitrates unidentified by the CIMS are composed of these alkyl nitrates.These compounds are derived from anthropogenic VOCs, and are formed in or near Sacramento and transported to the site.A GC-MS was used during the BEARPEX 2007 (August-October) campaign to measure C 1 -C 3 alkyl nitrates.The sum of these species was generally < 20 pptv (J.deGouw and J. Gilman, personal communication, 2009).During TC 4 , the NASA DC8 aircraft flew over the BEARPEX location (10 August 2007) before landing in Sacramento.Whole air samples were collected inflight, and analyzed off-line using GC-FID.The aircraft was at a pressure altitude of < 3.5 km when near the site, and decreased in altitude before landing in Sacramento.During this transect, the C 1 -C 5 alkyl nitrates peak concentration was 25 pptv, at the lowest altitude, just before landing (D.Blake, personal communication, 2010).Thus, neither of these measurements indicate small chain, alkyl nitrates are a large component of the alkyl nitrate budget at or near the BEARPEX site.
Mass spectral identification of nitrogen containing compounds benefits from the odd number of valence electrons in nitrogen.Parent ions or fragments of organic compounds containing an odd number of nitrogen atoms have an odd molecular weight (McLafferty and Turecek, 1993).Thus, with the CIMS technique used here, the CF 3 O − cluster with a nitrogen containing compound has an even mass.Since dinitrates do not have an odd molecular weight, any dinitrates would not be identified using this analysis.All identified or- ganic nitrates presented here have even m/z.As deployed during this campaign (starting with julian day 199), the instrument was set to scan all m/z ratios up to m/z 325 approximately once every two hours.In Fig. 3a, gray portion, we sum the signals from the nine even m/z that carry essentially all the additional intensity from nitrogen containing compounds.In the absence of sensitivity calculations for each species, a CIMS sensitivity equal to the sensitivity for ISOPN has been applied to these signals.The summed m/z are listed in Table 2. Possible structures for four of the nine even m/z compounds are shown in Fig. 4. Signals at m/z 230 and m/z 248 have been observed in smog chamber isoprene + NO 3 experiments (Ng et al., 2008).Since vegetative emission of isoprene is known to be  2. m/z 248 could also be due to MBO + NO 3 chemistry, but since the NO 3 rate constant is ∼ 500 times lower than with α-pinene, and ∼ 60 times lower than isoprene, it is expected to be a minimal contribution.light-dependent, no isoprene is emitted at night, when nitrate radical chemistry would be expected to be active.However, isoprene emitted late in the day when photochemical oxidants are low has been found to be oxidized by the nitrate radical (Brown et al., 2009;Starn et al., 1998).Near the end of the day, as well as during cloudy and stable conditions, reactions with BVOCs and other hydrocarbons can be a significant loss for nitrate radicals (Geyer et al., 2003).Specifically, the proposed structure for m/z 230 is a product from isoprene + NO 3 chemistry, forming a nitrooxyperoxy radical, which upon further reaction with RO 2 or NO 3 can form the C 5nitrooxycarbonyl.This compound, the C 5 -nitrooxycarbonyl, could also result from OH-initiated oxidation of ISOPN via H-abstraction from the carbon alpha to the OH group, however, this product has not been observed to be a significant product of ISOPN photooxidation in smog chamber experiments (Paulot et al., 2009a).The structure proposed for m/z 248 is also a product of isoprene + NO 3 chemistry; to form this product, the nitrooxyperoxy radical reacts with HO 2 to form the C 5 -nitrooxyhydroperoxide.The signal at m/z 248 could also result from MBO + NO 3 chemistry, however, the lifetime of MBO with respect to NO 3 , for conditions relevant to the site is expected to be longer than a month (Atkinson and Arey, 2003a).
We propose a nitrooxyhydroperoxide structure for the m/z 316 signal orginating from α-and β-pinene, and two possible structures resulting from α-pinene are shown in Fig. 4.These monoterpenes are generally modeled as temperature dependent emissions, and can be emitted at night when NO 3 chemistry is important (Sakulyanontvittaya et al., 2008).Nighttime HO 2 radical concentrations were generally 3-4 pptv (Mao et al., 2012).Nighttime concentrations of NO 3 of 1 pptv have been predicted at the measurement site (Bouvier- Brown et al., 2009).The lifetime of α-pinene with respect to NO 3 (1 pptv) and O 3 (55 ppbv) is 220 min and 170 min, respectively, so nitrate radical chemistry is expected to compete with ozonolysis for α-pinene oxidation at night.It is hypothesized this product might form an isobaric epoxide compound upon oxidation by OH (Paulot et al., 2009b).The proposed structure for the signal at m/z 300 results from the photooxidation of α-or β-pinene, and has been observed in smog chamber photooxidation experiments of α-pinene in the presence of NO x (Eddingsaas et al., 2012).The diurnal profiles of these signals (solid lines) are shown in Fig. 5 in comparison to the ISOPN nitrate signal (dashed line).
The diurnal profiles of the m/z 316 signal peaks at night, suggesting the formation of this product from nocturnal chemistry, combined with a long atmospheric lifetime.The profiles of m/z 230 and 248 also peak at night, supporting the proposal of the role of nitrate radical chemistry in their formation.If these four tentatively identified nitrates, and the five other unidentified nitrates listed in Table 2 are included in the comparison between the CIMS and TD-LIF, the two measurements are in close agreement, Fig. 3b.The timeseries of all CIMS identified nitrates, including these additional nitrates, are included on the bottom panel of Fig. 2.
Several of the proposed structures highlight the importance of nocturnal, nitrate radical chemistry at this rural, forested location.The formation of two of these structures requires reaction with HO 2 after reaction with NO 3 .This chemistry is challenging to study in nitrate radical driven chamber experiments that can often be dominated by RO 2 + RO 2 reactions.Therefore, even at much lower HO 2 concentrations than encountered during peak photochemistry, HO 2 contributes to radical reactivity at night.Another possible explanation for the unidentified organic nitrates measured by the TD-LIF instrument and not by the CIMS could be a difference in phase (gas or aerosol) of the organic nitrogen.The TD-LIF technique was operated in the field using no filter/denuder on the inlet, so the measurement should be considered the sum of gas and aerosol phase alkyl nitrates.The CIMS technique does not convert aerosol particles to the gas phase, therefore the technique only measures the gas phase contribution to the organic nitrates.Rollins et al. (2010) have recently shown the TD-LIF technique measures both gas and aerosol phase organic nitrates, and found 15-18 % of the SOA formed from α-pinene photooxidation, initiated by OH oxidation, under high NO x conditions was composed of alkyl nitrates.Fry et al. (2009) found 32-41 % of the aerosol mass formed in the β-pinene + NO 3 reaction was from organic nitrogen.It is possible that daytime and nocturnal oxidation of the monoterpene species at the site result in organic nitrates that partition into the aerosol phase, therefore the TD-LIF would be sensitive to the compounds and the CIMS would not.However, no aerosol organic nitrogen measurement during this campaign is available to evaluate this possibility.

Summary and conclusions
Observations of speciated and total alkyl and multifunctional nitrates over a ponderosa pine forest in the Sierra Nevada during the summer show total alkyl nitrates to be composed primarily of biogenically produced nitrates.First and second generation nitrates from isoprene along with a nitrate from MBO account for two-thirds of the alkyl nitrates at the site.Based on these measurements, the organic nitrates at the site are from a mix of locally produced and transported nitrates formed from photochemical and nighttime chemistry.These measurements highlight the interaction of the Sacramento urban plume with the forested areas downwind.Similar results would be expected from other forested areas downwind of urban areas.As areas of the tropics are developed, the impact of biogenically derived organic nitrates on ozone formation will become ever more significant.

Fig. 2 .
Fig. 2. Time series (left) and mean diurnal profiles (right) of organic nitrates and ANs measured during BEARPEX 2009.Diurnal profiles shown in red are mean concentrations on days with maximum daytime temperatures of 26 °C and higher.Blue diurnal profile is the mean concentration on days with maximum temperatures less than 26 °C.Vertical lines indicate local sunrise.
Fig. 3. (a) Comparison of CIMS and TD-LIF measurements during the last two weeks of the campaign (Julian days 199-212).ANs as measured by TD-LIF are shown as the dashed line.Speciated organic nitrates are shown in the colors.The gray fraction represents the sum of other organic nitrate species observed with the CIMS (Table 2).Red = ISOPN, Green = PROPNN, Black = MVKN+MACRN, Magenta = MBON, Blue = ETHLN.(b) In blue, ANs (measured with TD-LIF) vs. the sum of individual nitrates (measured with CIMS), not including the data in the gray portion.Equation for the linear fit is y = 1.25x +0.096, R 2 = 0.85.In red, ANs (measured with TD-LIF) vs. the sum of individual nitrates (measured with CIMS), including the data in the gray portion.Equation for the linear fit is y = 0.91x + 0.052, R 2= 0.89.There are fewer data points for the data shown in red due to the frequency of data points (approximately once every 1.6 h).

Fig. 4 .
Fig. 4. Proposed structures for even mass signals listed in Table2.m/z 248 could also be due to MBO + NO 3 chemistry, but since the NO 3 rate constant is ∼ 500 times lower than with α-pinene, and ∼ 60 times lower than isoprene, it is expected to be a minimal contribution.

Fig. 5 .
Fig.5.Average, normalized diurnal signal of the m/z 316 product, proposed as the nitrooxyhydroperoxide from α-pinene (solid black line); m/z 300 product, proposed as the hydroxynitrate from α-pinene (solid blue line); m/z 248 product, proposed as the nitrooxyhydroperoxide from isoprene (solid red line); m/z 230, proposed as the nitrooxycarbonyl from isoprene (solid cyan line).For reference, the signal for ISOPN is shown as the dashed line.The solid gray line indicates the approximate time of local sunset.
. The most abundant nitrate was the first generation isoprene hydroxynitrates; this measurement is a sum of isomers.Second generation nitrates from isoprene Organic nitrates from isoprene and MBO photooxidation observed in this study.Only one isoprene hydroxynitrate from the eight possible isoprene nitrate isomers is shown here.Only ETHLN forms from the isoprene hydroxynitrate isomer shown here.The other second generation nitrates can form from the isomers indicated by the superscripted letters (a

Table 2 .
Significant nitrogen-containing m/z observed with the CIMS.The m/z column indicates the signal in the mass spectrum.The m.w.column is the inferred molecular weight (m/z −85).The letters in the proposed structure column designated the corresponding structures shown in Fig. 4.