JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112, D21302, doi:10.1029/2007JD008756, 2007 Click Here for Full Article
Nighttime nitrate radical chemistry at Appledore Island, Maine during the 2004 International Consortium for Atmospheric Research on Transport and Transformation J. L. Ambrose,1,2 H. Mao,2 H. R. Mayne,1 J. Stutz,3 R. Talbot,2 and B. C. Sive2 Received 5 April 2007; revised 22 June 2007; accepted 7 August 2007; published 2 November 2007.
[1] Trace gases including nitrogen dioxide (NO2), nitrate radical (NO3), ozone (O3), and a suite of volatile organic compounds (VOCs) were measured within the New England coastal marine boundary layer on Appledore Island (AI), Maine, USA as part of the International Consortium for Atmospheric Research on Transport and Transformation (ICARTT) field campaign. These measurements, together with local meteorological records and published kinetic data were used to investigate nighttime NO3 chemistry at AI during the period of 8–28 July 2004. Among the VOCs, isoprene, monoterpenes and dimethylsulfide (DMS) were the dominant NO3 reactants; on average, DMS accounted for 51 ± 34% of the total reactivity. For three case studies, NO3 mixing ratios were calculated from measured parameters with resultant uncertainties of 30%. Discrepancies with measured NO3 appeared to result primarily from input parameter variability and exclusion of heterogeneous dinitrogen pentoxide (N2O5) chemistry. We indirectly determined that nighttime NO3 and NOx (=NO + NO2) removal via N2O5 chemistry (gas-phase + heterogeneous) was on average 51–54% and 63–66% of the total respectively. Our analysis suggested that the minimum average NO3 and NOx removal via heterogeneous N2O5 chemistry was 10% of the total. Reducing gas-phase N2O5 reactivity in accord with Brown et al. (2006a) increased the importance of heterogeneous N2O5 chemistry substantially. It is plausible that the latter pathway was often comparable to gas-phase removal of NO3 and NOx. Overall, 24 h-averaged NOx removal was 11 ppbv, with nighttime chemical pathways contributing 50%. Citation: Ambrose, J. L., H. Mao, H. R. Mayne, J. Stutz, R. Talbot, and B. C. Sive (2007), Nighttime nitrate radical chemistry at Appledore Island, Maine during the 2004 International Consortium for Atmospheric Research on Transport and Transformation, J. Geophys. Res., 112, D21302, doi:10.1029/2007JD008756.
1. Introduction and is highly photo-labile with limited oxidative capacity during the daytime due to rapid photolysis via [2] The hydroxyl radical (OH), nitrate radical (NO3) and ozone molecule (O ) are the most important gas phase 3 ðR2aÞ NO þ hu ! NO þ O ðÞ 10% oxidants of volatile organic compounds (VOCs) and nitro- 3 2 gen oxides (NOx =NO+NO2) in the troposphere. Both OH and O3 have primary photochemical sources, and OH is the ðR2bÞ NO3 þ hu ! NO2 þ O ðÞ 90% most important daytime oxidant. NO3 is produced exclu- sively by the reaction of NO with O , 2 3 and reaction with NO,
ðR3Þ NO3 þ NO ! 2NO2 ðR1Þ NO2 þ O3 ! NO3 þ O2 [Atkinson, 2000; Geyer et al., 2001; Geyer and Platt, 2002; Brown et al., 2003a]. At night, the production rate of NO3 far exceeds that of OH, and NO3 is generally several orders 1Department of Chemistry, University of New Hampshire, Durham, of magnitude more reactive toward VOCs and NOx than O3 New Hampshire, USA. [e.g., Atkinson, 2000]. 2Climate Change Research Center, Institute for the Study of Earth, Oceans and Space, University of New Hampshire, Durham, New [3] During the nighttime, NO3 is directly removed from Hampshire, USA. the atmosphere by reactions with VOCs and NO. For 3Department of Atmospheric Sciences, University of California, Los saturated VOCs, dimethylsulfide (DMS), the oxygenated Angeles, California, USA. VOCs (OVOCs), and aromatics, NO3-initiated degradation proceeds mostly via hydrogen-atom abstraction to generate Copyright 2007 by the American Geophysical Union. 0148-0227/07/2007JD008756$09.00 nitric acid (HNO3) and peroxy radicals. For unsaturated
D21302 1of19 D21302 AMBROSE ET AL.: NIGHTTIME NITRATE RADICAL CHEMISTRY D21302
VOCs including isoprene, the alkenes and monoterpenes, ambient conditions vary from those characteristic of the reaction with NO3 proceeds mostly via initial NO3 addition remote Atlantic to being dominated by strong continental at unsaturated sites with the products being nitrooxy peroxy biogenic and anthropogenic sources. Measurements of NO3 radicals. Biogenic VOCs, such as isoprene, monoterpenes in this region were previously limited to the 2002 New and DMS [Allan et al., 2000; Geyer et al., 2001; Warneke et England Air Quality Study [Brown et al., 2004; Warneke et al., 2004; Aldener et al., 2006] are particularly important al., 2004; Aldener et al., 2006], which demonstrated that reactants for NO3. Because NO is rapidly oxidized to NO2 during the summer months nighttime NO3 chemistry com- by reaction with O3 after dark [Allan et al., 2000], it is petes with daytime OH chemistry in controlling the NOx generally an important reactant for NO3 only in the prox- budget [Brown et al., 2004] and isoprene oxidation [Warneke imity of sources [Platt and Janssen, 1995; Brown et al., et al., 2004]. The nighttime abundances of NO3 and mono- 2003a; Stutz et al., 2004]. Also important to nighttime NO3 terpenes were also closely coupled [Warneke et al., 2004; chemistry is the reversible reaction of NO3 with NO2 to Aldener et al., 2006]. Hydrolysis of N2O5 and reactions of generate dinitrogen pentoxide (N2O5), NO3 with VOCs were shown to contribute roughly equally to nighttime NO3 removal [Aldener et al., 2006]. kf ðR4aÞ NO3 þ NO2 þ M ÀÀÀ! N2O5 þ M [6] In this study we utilize measurements of a suite of trace gasses, including NO2,NO3,O3, and VOCs, and relevant meteorological parameters within the coastal Gulf kr of Maine MBL at the University of New Hampshire (UNH) ðR4bÞ N2O5 þ M ÀÀÀ! NO2 þ NO3 þ M AIRMAP Observing Station on Appledore Island (AI) (http://www.airmap.unh.edu) during the 2004 International where kf and kr are the rate coefficients for reactions (R4a) Consortium for Atmospheric Research on Transport and and (R4b), respectively. The NO3-NO2-N2O5 system Transformation (ICARTT). (See Fehsenfeld et al. [2006] for equilibrates rapidly after dark [Atkinson, 2000; Geyer et an overview of the ICARTT campaign.) Our goals were to al., 2001; Brown et al., 2003b]. The removal of N2O5 is quantify the gas-phase chemistry that governed the ob- generally expected to be governed by heterogeneous served nocturnal behavior of NO3 andtoestimatethe chemistry (i.e., reaction of N2O5 on or within aerosol contribution of heterogeneous N2O5 chemistry to NO3 and particles), NOx removal.
ðR5Þ N2O5 þ H2Oaerosol ! 2HNO3 2. Experimental Methods
[7] Routine measurements of CO, NO, and O3 at AI, ME and likely depends strongly on aerosol composition [e.g., 0 00 0 00 Folkers et al., 2003; Hallquist et al., 2003; Thornton and (42°59 13 N, 70°36 55 W) have been made seasonally Abbatt, 2005; Brown et al., 2006a]. Gas-phase reactions of (May–October) from the top story (45 m asl) of a World War II-era surveillance tower since 2002 as part of the UNH N2O5 with water vapor have also been demonstrated [Wahner et al., 1998], AIRMAP Observing Network. The tower was equipped for measurements of an extensive suite of chemical parameters ðR6aÞ N O þ H O ! 2HNO for the ICARTT campaign. The subset of measurements 2 5 2 3 incorporated into this analysis, together with the experi- mental details of the instrumentation deployed during ðR6bÞ N O þ 2H O ! 2HNO þ H O ICARTT, is summarized in Table 1. Additionally, meteoro- 2 5 2 3 2 logical variables were monitored at the National Data Buoy Center Coastal-Marine Automated Network station (IOSN3) although their overall importance in the chemistry of the 0 00 0 00 atmosphere remains a topic of considerable uncertainty on White Island (42°58 00 N, 70°37 24 W, 15 m asl) [Heintz et al., 1996; Martinez et al., 2000; Atkinson et al., [National Data Buoy Center, 2004] located 2.3 km south- 2004; Stutz et al., 2004; Aldener et al., 2006; Brown et al., east of AI. Backward air mass trajectories were simulated at 2006a]. Plymouth State University using the NOAA HYSPLIT model, initiated from AI and run for 24 h (http://pscwx. [4] The reactivity of NO can influence nighttime and 3 plymouth.edu/ICARTT/archive.html). The trajectories were early morning abundances of VOCs and NOx, and thus photochemical production of O . How efficiently NO used to provide qualitative descriptions of the mesoscale 3 3 dynamics accompanying the trace gas observations. mediates the removal of VOCs and NOx from the atmo- sphere depends strongly on the abundance of NOx [Platt 2.1. Trace Gas Measurements and Janssen, 1995] and the sinks for N2O5 [Brown et al., [8] Air samples were collected hourly between 2 July and 2003a; Warneke et al., 2004; Brown et al., 2006b]. Accord- 13 August 2004 for C2 –C10 non-methane hydrocarbons ingly, nighttime NO chemistry is significantly different in 3 (NMHCs), C1 –C2 halocarbons, C1–C5 alkyl nitrates, se- urban versus rural and continental versus marine environ- lected OVOCs, CO2, and CH4. During the sampling period ments, and varies considerably with season of the year a single head metal bellows pump (MB-302MOD, Senior [Platt and Janssen, 1995; Heintz et al., 1996; Geyer and Flexonics, Sharon, MA) continuously drew ambient air Platt, 2002; Vrekoussis et al., 2007]. from 45 m asl through a 20 m  6.35 mm I.D. [5] The New England (NE) coastal marine boundary stainless steel inlet line. Samples were collected in evacuated layer (MBL) is a unique environment for studying nighttime (10À2 mbar) 2 l electropolished stainless steel canisters and NO3 chemistry. It is a corridor for mixing of air masses pressurized to 2.4 bar. Filled canisters were returned to the of both marine and continental origin, and consequently
2of19 D21302 AMBROSE ET AL.: NIGHTTIME NITRATE RADICAL CHEMISTRY D21302
Table 1. Measurement Details for Atmospheric Observations at AI During the ICARTT Campaign Sampling Sample Cycle Variables measured Analytical schemea intervalf period, t Integration time LOD Accuracy Chemical variables b g C2-C10 NMHCs GC-FID 7/2–8/13 1 hr 5 min 0.002–0.003 ppbv ±10% OVOCs GC-MSb C8-C10 aromatics PTR-MS 7/1 – 8/13 10 min 20 s 0.010 ppbv ±15% DMS OVOCs c NO3 DOAS 7/8–8/11 5 min 5 min 3.4 pptv ±1.7 pptv HCHO 15 min 15 min 0.6 ppbv ±0.3 ppbv NO2 15 min 15 min 0.30 ppbv ±0.15 ppbv CO IR spectroscopy (2200 cmÀ1)d 1 min 1 min 10 ppbv ±5 ppbv NO Chemiluminescenced 1 min 1 min 0.075 ppbv ±0.020 ppbv d O3 UV spectroscopy (254 nm) 1 min 1 min 1 ppbv ±1 ppbv
Meteorological variablese Ambient P Barometer 1 hr 2 min 800 mbar ±1 mbar Ambient T Thermistor 1 hr 2 min À40–50 °C±1°C Dew T RH probe 1 hr 2 min À35–30 °C±1°C Wind direction Anemometer 10 min 2 min 0–360° from N ±10° Wind speed 0–62 m sÀ1 ±1 m sÀ1 aThe following sources should be consulted for detailed descriptions of these systems: bSive et al. [2005]; Zhou et al. [2005]. cAlicke et al. [2002]; Pikelnaya et al. [2007]. dMao and Talbot [2004]. eNational Data Buoy Center [2004]. fIntervals are indicated only for systems that were not operational during the entire campaign (4 July–15 August); gValue denotes the time interval over which canister samples were collected.
UNH Climate Change Research Center every 4 days and Environmental, Inc.). Each of the cylinders used in the their contents were analyzed by gas chromatography using calibrations had an absolute accuracy of <±5% for all gases. flame ionization and electron capture detection in con- Using methods similar to those described by Apel et al. junction with quadrupole mass spectrometry. Detailed [1998], standards were diluted to atmospheric mixing ratios discussions of the UNH canister sampling and analysis (ppbv to pptv levels) with catalytic converter prepared zero protocols are provided by Sive et al. [2005] and Zhou et air adjusted to maintain the humidity of the sampled air. al. [2005]. Calibrations were conducted periodically to monitor PTR- [9] In addition to chromatographic analysis, Proton MS performance and quantify the mixing ratios of target Transfer Reaction-Mass Spectrometry (PTR-MS) was used gases. Additionally, mixing ratios for each gas were calcu- to provide high frequency measurements of several groups lated by using the normalized counts per second which were of isomeric NMHCs, certain individual OVOCs, DMS, and obtained by subtracting out the non-zero background signal acetonitrile from 1 July to 12 August (Table 1). The PTR- for each compound. MS sampled air that was continuously drawn through a [11] Long-Path (LP) Differential Optical Absorption 30.5 m  9.525 mm I.D. PFA Teflon tube from the same Spectroscopy (DOAS) was used to measure a suite of height as the canister pump inlet. The flow rate through the inorganic gases, including NO2 and NO3, and formaldehyde sample line was 75 l minÀ1, resulting in a 2 s residence [Alicke et al., 2002]. The retroreflector array was installed time. A sub-stream of fast flowing air off the main sample ( 15 m asl) on the White Island lighthouse, 2.3 km from line was sent directly to the PTR-MS. The PTR-MS was AI. Spectra were recorded from the tower’s third floor operated with a drift tube pressure of 2 mbar and an electric ( 40 m asl) over a 4.6 km path length [Pikelnaya et al., field of 600 V while continuously stepping through a series 2007]. The DOAS system was operated from July 8 to of 30 masses. Of the 30 masses monitored, 6 masses were August 11 although spectra were often not obtainable used for diagnostic purposes while the other 24 masses during periods of persistent fog and/or precipitation. corresponded to the VOCs of interest. The dwell time for each of the 24 masses was 20 s, yielding a total measurement 2.2. Analysis Methods cycle of 10 min. The system was zeroed every 2.5 h for [12] In this paper we use an incremental approach to 4 cycles by diverting the flow of ambient air through a determine the respective importance of gas-phase and heated catalytic converter (0.5% Pd on alumina at 450°C) heterogeneous mechanisms in nighttime NO3 chemistry at to oxidize the VOCs and determine system background AI. This section describes our treatment of gas-phase NO3 signals. chemistry. The potential importance of heterogeneous [10] Calibrations for the PTR-MS system were conducted chemistry is discussed separately in Section 3.4. Table 2 using three different high-pressure cylinders containing lists the measured NO3 reactants along with the kinetic data synthetic blends of selected NMHCs and OVOCs at the for their corresponding reactions. Kinetic data for additional part per billion by volume (ppbv) level (Apel-Reimer reactions considered in this study are also given in Table 2. The pseudo first-order loss rate coefficient (hereinafter
3of19 D21302 AMBROSE ET AL.: NIGHTTIME NITRATE RADICAL CHEMISTRY D21302
a Table 2. Rate Data Applicable to Nighttime Gas-Phase NO3 Chemistry and Pertaining to the Chemical Variables Monitored at AI During the ICARTT Campaign 3 À1 À1 b,c 3 À1 À1 NO3 sink/Reaction A,cm molecule s B (K) k (cm molecule s ) Alkanes Ethane 1 Â 10À17 Propaned 7 Â 10À17 i-Butanee 3.05 Â 10À12 3060 ± 99 1.06 Â 10À16 (±40%) n-Butanef 2.8 Â 10À12 3280 ± 400 4.6 Â 10À17 (±58%) Cyclopentaneh 1.7 Â 10À17 i-Pentaneg 2.99 Â 10À12 2927 ± 173 1.62 Â 10À16 (±35%) n-Pentaneg 8.7 Â 10À17 2,2-Dimethylbutaneh 2.7 Â 10À17 Methylcyclopentaneh 3.2 Â 10À16 Cyclohexaneg 1.4 Â 10À16 2-Methylpentaneg 1.8 Â 10À16 3-Methylpentaneg 2.2 Â 10À16 n-Hexaneg 1.1 Â 10À16 n-Heptaneg 1.5 Â 10À16 2,3-Dimethylpentaneh 3.3 Â 10À16 2,4-Dimethylpentaneg 1.5 Â 10À16 n-Octaneg 1.9 Â 10À16 2,2,4-Trimethylpentaneg 9 Â 10À17 2,3,4-Trimethylpentaneh 4.4 Â 10À16 n-Nonaneg 2.3 Â 10À16 (±40%) n-Decaneg 2.8 Â 10À16
Alkenes and Alkynes Ethenef 3.3 Â 10À12 2880 ± 500 2.1 Â 10À16 (±58%) Ethynef 1 Â 10À16 Propenef 4.6 Â 10À13 1155 ± 300 9.5 Â 10À15 (±58%) 1-Buteneg 3.14 Â 10À13 938 ± 106 1.35 Â 10À14 (±30%) c-2-Buteneg 3.50 Â 10À13 i-Buteneg 3.32 Â 10À13 t-2-Buteneg,i 1.22 Â 10À18 (À)382 ± 28 3.90 Â 10À13 2-Methyl-2-buteneg 9.37 Â 10À12 1-Pentenej 1.20 Â 10À14 c-2-Pentenej 6.55 Â 10À13 t-2-Pentenej 3.78 Â 10À13
Biogenicsb a-Pineneg 1.19 Â 10À12 (À)490 ± 97 6.16 Â 10À12 (±30%) b-Pineneg 2.51 Â 10À12 (±40%) Campheneg 6.6 Â 10À13 (±35%) d-Limoneneg 1.22 Â 10À11 (±35%) DMSk 1.9 Â 10À13 (À)520 ± 200 1.1 Â 10À12 (±41%) Isopreneg 3.03 Â 10À12 446 ± 60 6.78 Â 10À13
Aromatics Toluene 6.8 Â 10À17 Ethylbenzene 5.7 Â 10À16 m + p-Xylenel 3.43 Â 10À16 (±74%) o-Xylene 3.77 Â 10À16 i-Propylbenzenem 1.4 Â 10À16 n-Propylbenzenem 1.4 Â 10À16 2-Ethyltoluenem 7.1 Â 10À16 3-Ethyltoluenem 4.5 Â 10À16 4-Ethyltoluenem 8.6 Â 10À17 1,2,3-Trimethylbenzenen 1.86 Â 10À15 1,2,4-Trimethylbenzenen 1.81 Â 10À15 1,3,5–Trimethylbenzene 8.00 Â 10À16
OVOCs Acetaldehydef 1.4 Â 10À12 1860 ± 500 2.7 Â 10À15 (±58%) Acetonef 3 Â 10À17 Formaldehydef 5.8 Â 10À16 Methanol 9.4 Â 10À12 2650 ± 700 1.30 Â 10À16 (±220%)
Additional R1k 1.4 Â 10À13 2470 ± 150 3.5 Â 10À17 (±15%) R3k 1.8 Â 10À11 (À)110 ± 100 2.6 Â 10À11 (±26%) o À11 3 À1 R4 Keq = (2.8 ± 0.6) Â 10 (cm molecule ) R6a + R6b k,p(9.7 ± 7.6) Â 10À22 R7k 1.19 Â 10À11
4of19 D21302 AMBROSE ET AL.: NIGHTTIME NITRATE RADICAL CHEMISTRY D21302
referred to as the loss efficiency) for removal of NO3 via its The NO3 loss efficiency with respect to the reactions of 0 À1 0 reaction with trace gas i, ki (s ), is given by the product of N2O5 with water vapor, k(H2O(g)+N2O5), was determined as the reaction rate coefficient, k(NO3+i), and the concentration of i: k0 k0 k k H O K NO H O ðÞH2OgðÞþN2O5 indirect ¼ ðÞÁ6a þ 6b Á ½2 eq Á ½2 ½2
0 ð4Þ ki ¼ kðÞNO3þi Á ½i ð1Þ
where the rate coefficients k6a and k6b correspond with N X reactions (R6a) and (R6b), respectively, and [H2O] is the k0 ¼ k0 ð2Þ T i atmospheric water vapor concentration. The NO3 loss rate i¼1 with respect to sink (i.e., loss pathway) j, Lj(NO3), is the 0 À1 product of the loss efficiency for sink j and the NO3 where kT (s ) is the total loss efficiency for the removal of 0 concentration; the total loss rate via gas-phase (homoge- NO3 by all N reactive gases. (The term ki will also be neous) sinks, Lhom(NO3), is given by described as the reactivity of i.) Because N2O5 was not measured, its contribution to NO3 removal was estimated by XK assuming equilibrium with NO2 and NO3 during the 0 0 nighttime hours: LhomðÞ¼NO3 LjðÞ¼NO3 kT þ kindirect Á ½NO3 j¼1 0 ¼ khom Á ½NO3 ð5Þ ½N2O5 kf KeqðÞ¼T ¼ ð3Þ ½NO2 ½NO3 kr where the summation is over all K gas-phase sinks. The total rate of gas-phase nighttime NOx removal is given by The errors incurred by this approximation are expected to be minimal under most conditions observed at AI. For instance, for average conditions encountered at AI (see LðÞNOx night¼ LTðÞþNO3 2 Á LindirectðÞNO3 ð6Þ below) with [O3] 40 ppbv, relative humidity (RH) À3 90%, loss efficiencies for NO3 and N2O5 of 7 Â 10 where the factor of 2 appears in front of the N2O5-mediated À4 À1 and 5 Â 10 s , respectively, and the chemistry NO3 loss rate because N2O5 contains two equivalents of represented as described by Brown et al. [2003b; Equations NOx. Daytime NOx removal results primarily from the [4] and [5]], simple box model calculations demonstrated reaction of NO2 with OH, that ambient N2O5 would have been 95% of its equilibrium value within 15 min of an instantaneous sunset, with ðR7Þ NO2 þ OH þ M ! HNO3ðÞþg M starting concentrations of NO3 and N2O5 equal to zero. Furthermore, recent simultaneous field measurements of O , 3 with a rate given by NO2,NO3, and N2O5 showed good agreement between measured N2O5 mixing ratios and those calculated assum- ing equilibrium among the nitrogen oxides [Brown et al., LðÞNOx day ffi kðÞNO2þOH Á ½NO2 ½OH ð7Þ 2003a]. In our analysis, the uncertainty in kr (average value of 20%) made the dominant contribution to the un- [Brown et al., 2004]. During the ICARTT campaign, OH certainty in Keq [Atkinson et al., 2004]; together the concentrations were calculated following the parameteriza- uncertainties in the measurements of NO2 ( 20%) and tion of Ehhalt and Rohrer [2000] as discussed by Keene et NO ( 40%) made a greater contribution to the overall 3 al. [2007]. The NO3 production rate, P(NO3), was uncertainty ( 60%) in the calculated N2O5 mixing ratios.
Notes to Table 2: aFrom Atkinson [1991] unless otherwise indicated. bCalculated at 298 K; given by: k = AÁexpÀ(B/T). cUncertain by a factor of 2 unless otherwise indicated. dFrom Atkinson et al. [1997]. eFrom Atkinson [1994]. fFrom Atkinson et al. [1999]. gFrom Atkinson [1997]. hDerived from bond-additivity relationships of Atkinson [1991]. iGiven by: k = AÁT2ÁexpÀ(B/T). jFrom Pfrang et al. [2005]. kFrom Atkinson et al. [2004]. lAverage of values for m- and p-xylene from Atkinson [1991]. mFrom Jenkin et al. [2003]. nCalculated at 294 K. oCalculated from parameters given by Atkinson et al. [2004]. p 17 À3 Total rate coefficient with respect to (R6a) and (R6b) calculated at 290 K with [H2O(g)] = (4.0 ± 0.4) Â 10 molecules cm .
5of19 D21302 AMBROSE ET AL.: NIGHTTIME NITRATE RADICAL CHEMISTRY D21302
Figure 1. Time series of (a) NO3 and O3, (b) NO and NO2, (c) ethyne and toluene, (d) isoprene and DMS, and (e) monoterpenes measured between noon on 8 July and noon on 28 July. Segments of thickened lines indicate measurements made during the nighttime hours (21:00–05:00). Uncertainties are described in Table 1.
calculated from the measured mixing ratios of NO2 and O3 where [NO3]0 is the initial NO3 concentration. Since these and the corresponding rate coefficient (Table 2): terms do in fact vary, the time dependence of [NO3]t was obtained by propagating Equation (10) for short intervals, updating the values of P(NO ) and k0 at every time step. PðÞNO3 ¼ kðÞNO2þO3 ðÞT Á ½NO2 ½O3 ð8Þ 3 hom For this investigation, Equation (10) was used to calculate nocturnal NO mixing ratio profiles with the initial The total rate of the NO3 concentration change, d[NO3]/dt, 3 can be approximated as condition [NO3]0=21:00 = 0. For clarity in presentation of our results, we utilized U.S. east coast local time (Eastern d½NO Daylight Time), which is UT–4 h. 3 ¼ PðÞÀNO k0 Á ½NO ð9Þ dt 3 hom 3
0 3. Results and Discussion If the terms P(NO3) and khom in Equation (9) are constant, 3.1. Kinetic Comparison of Gas-Phase Nighttime NO3 the concentration of NO3 at any arbitrary time t,[NO3]t is given by: Loss Processes [13] Time series of NO , its precursors NO and O , 3 2 3 0 0 biogenic VOCs, and selected anthropogenic tracer species PðÞNO3 À PðÞNO3 À khom Á ½NO3 0 Á exp Àkhom Á t ½NO3 ¼ are presented in Figure 1 for 8–28 July 2004, where the t k0 hom thickened lines correspond to measurements during the ð10Þ nighttime hours (21:00–05:00). Highly variable conditions
6of19 D21302 AMBROSE ET AL.: NIGHTTIME NITRATE RADICAL CHEMISTRY D21302
Figure 2. Reactivity of the biogenic compounds relative to that of all the measured VOCs (open circles) calculated for the nighttime hours between 21:00 on 8 July and 05:00 on 28 July. (The NO3 mixing ratios (green trace) measured during the nighttime hours are reproduced for comparison). The average 0 0 uncertainty in the calculated values of kbiogenics/kVOCs was 50%.
were experienced at AI, as the site was frequently impacted maximum of 60 pptv. Geyer et al. [2001] observed similar by continental outflow composed of both anthropogenic and values during July and August, 1998 at a suburban site near biogenic emissions [Chen et al., 2007]. The mixing ratios of Berlin, Germany. During July and August, 1990 at a NO2,NO3, isoprene, DMS, and the monoterpenes varied suburban site in the San Joaquin Valley, CA Smith et al. over wide ranges from below their limit of detection (LOD) [1995] observed average and maximum NO3 mixing ratios to 19.6, 0.073, 0.85, 0.33, and 0.62 ppbv, respectively. The of 2 and 80 pptv respectively. average NO3 mixing ratio was 10 pptv (see below), which [14] The time series shown in Figure 2 presents the ratio 0 appears to agree well with the observations of Warneke et of the total reactivity of biogenic compounds, kbiogenics,to 0 al. [2004] in this region during July and August, 2002. By the total reactivity of all VOCs, kVOCs. The average ratio comparison, the average summertime NO3 mixing ratio on during the time period from 8–28 July was 0.91 ± 0.14, the island of Crete in the East Mediterranean Sea during the emphasizing the dominance of biogenic VOC reactivity. years 2001–2003 was 6 pptv [Vrekoussis et al., 2007]. The alkenes appeared to be the next most important class of Allan et al. [2000] measured NO3 mixing ratios up to 40 VOCs for nighttime NO3 removal, but on average they and 20 pptv in the northeast Atlantic at Mace Head, Ireland accounted for <10% of the VOC reactivity during the night. during July and August, 1996 and at Tenerife Island off the The OVOCs, aromatics and alkanes each accounted for coast of northwestern Africa during June and July, 1997, <1% of the VOC reactivity during the night. respectively. Average NO3 mixing ratios during several [15] The relative contribution of the individual biogenic nights in June, 1995 at a coastal site in north Norfolk, VOCs to the overall NO3 loss rate was evaluated, including 0 0 0 England [Allan et al., 1999] ranged between 4 and 25 pptv. kmonoterpenes, kDMS and kisoprene (Figure 3). Of the biogeni- Average NO3 mixing ratios in summer 1993 at a rural site cally derived compounds measured at AI, DMS appeared to 0 on Ru¨gen Island in the Baltic Sea [Heintz et al., 1996] be the dominant NO3 reactant. Average values of ki/ 0 ranged between 6 and 10 pptv. During several nights in kbiogenics, where i = DMS, isoprene and monoterpenes, were August and September, 2000, at a suburban continental site 0.56 ± 0.36, 0.18 ± 0.21 and 0.26 ± 0.30 respectively. When outside the city center of Houston, TX, Stutz et al. [2004] they were above the LOD, monoterpenes generally observed NO3 mixing ratios typically <10 pptv with a accounted for a large fraction of the NO3 nightly loss, but
Figure 3. Reactivity of DMS, monoterpenes and isoprene calculated for the nighttime hours between 21:00 on 8 July and 05:00 on 28 July. (Note the break in the ordinate between 0.026 and 0.036 sÀ1) The 0 average uncertainties in the values of ki where i = DMS, isoprene and monoterpenes were 60, 60 and 30%, respectively.
7of19 D21302 AMBROSE ET AL.: NIGHTTIME NITRATE RADICAL CHEMISTRY D21302
Figure 4. HYSPLIT trajectories corresponding with the selected nights: (a) 13 July, (b) 14 July, (c) 11 July, (d) 16 July, and (e) 25 July. Units of elevation (Elev.) are in m(Â10À3) above ground level (a.g.l). Red, blue and green trajectories end at AI (star) at 10, 200, and 1000 m a.g.l., respectively.
À1 due to their continental origin they appeared at AI only compared to a NOx loss rate of 0.12 ± 0.11 ppbv hr during time periods influenced by strong offshore flow. This through reactions of NO3 with VOCs. occurred 30% of the time during the ICARTT study period [Chen et al., 2007]. 3.2. Nocturnal Chemistry Case Studies at Appledore Island During ICARTT [16] Assuming equilibrium conditions for the NO2-NO3- 18 N2O5 system, the NO3 and N2O5 mixing ratios should be [ ] The following discussion focuses on trace gas meas- nearly equal when NO2 mixing ratios are several ppbv for urements from several nights between 8–28 July 2004. the average conditions encountered at AI (T = 290 ± 2 K, Backward air mass trajectories were used for qualitative NO3 = 0.011 ± 0.013 ppbv, and NO2 = 4.0 ± 4.2 ppbv). Our source region identification, while trace gas measurements calculations suggest that removal of NO3 via gas-phase and relevant gas-phase kinetics were used to explain im- reactions of N2O5 was frequently comparable in magnitude portant features of nighttime NO3 chemistry. 0 0 19 to NO3 removal with VOCs, indicating kVOCs kindirect. [ ] The selected case studies illustrate the role of night- 0 0 0 0 The average values of kindirect/khom and kVOCs/khom were time NO3 chemistry over a range of conditions at AI, 0.42 ± 0.26 and 0.58 ± 0.27 respectively, indicating that including periods of Atlantic aged marine flow (13 and these two pathways were, in fact, comparable sinks for 14 July), polluted continental outflow (11 and 16 July), and NO3. polluted continental/biogenic outflow (25 July). For each [17] Whereas the reactions of NO3 with VOCs remove case study we examined the relative levels of pollutants and NO3 and NOx at the same rate, hydrolysis of N2O5 is two estimated the transport pathway, calculated the mixing ratio times more efficient than NO3-related mechanisms for NOx of N2O5, and subsequently assessed the NO3/NOx loss rates removal because two equivalents of NOx are associated with due to reaction with marine and terrestrial biogenic VOCs. N2O5. The average rate of NOx loss via reactions of N2O5 3.2.1. Clean Marine Flow on 13 and 14 July À1 20 with H2O(g) was estimated to be 0.36 ± 0.41 ppbv hr [ ] The lowest trace gas levels during the nighttime hours of the ICARTT campaign were measured on 13 and
8of19 D21302 AMBROSE ET AL.: NIGHTTIME NITRATE RADICAL CHEMISTRY D21302
14 July. Backward trajectories for these two nights NO2 mixing ratios, which shifted the nitrogen oxides (Figures 4a and 4b) indicate that prior to their arrival at equilibrium toward N2O5. The reaction of N2O5 with AI, the air masses resided over the Atlantic Ocean for H2O(g) generally becomes more important with increasing >24 h [Keene et al., 2007]. Overall, the trace gas mixing relative humidity or decreasing temperature; however, both ratios were characteristic of relatively clean conditions, of these variables were relatively constant in this case. After 0 exhibiting no recent anthropogenic influences. For example, midnight, khom was apparently dominated by reactions of average mixing ratios of toluene, ethyne and o-xylene on N2O5, such that during the remainder of the evening NO3 the night of 13 July were 0.008 ± 0.002, 0.288 ± 0.051 and chemistry should have been primarily generating HNO3.At 0.015 ± 0.002 ppbv respectively and on the night of 14 July 01:00, NO2 mixing ratios were 11.0 ppbv resulting in a gas- À1 were 0.009 ± 0.003, 0.105 ± 0.011 and 0.015 ± 0.003 ppbv. phase HNO3 production rate of 0.54 ± 0.31 ppbv hr .By By comparison, for the nighttime hours (21:00–05:00) 05:00 when NO2 had increased to 18.4 ppbv, the rate of gas- between July 8 and 28, the average mixing ratios of these phase HNO3 production was increased to 1.1 ± 0.6 ppbv À1 three gases were 0.157 ± 0.151, 0.360 ± 0.231 and 0.031 ± hr . The corresponding rates of gas-phase NOx loss were 0.018 ppbv. 0.58 ± 0.31 and 1.2 ± 0.6 ppbv hrÀ1 respectively. The gas- [21] During both nights NO2 mixing ratios were sub-ppbv phase rates of HNO3 production and NOx loss at 01:00 and and NO3 was below its LOD of 3.4 pptv. The relatively low 05:00 are compared with the corresponding rates from abundances of NO3 presumably resulted from correspond- heterogeneous N2O5 hydrolysis in Table 3. Despite increas- ingly low levels of NOx coupled with high levels of DMS. ing NO2 mixing ratios, the NO3 production rate remained Mixing ratios of DMS were elevated to >0.20 ppbv, while roughly constant at 0.51 ± 0.12 ppbv hrÀ1, resulting from isoprene remained <0.025 ppbv and the monoterpenes were the subsequent decrease in O3 mixing ratios. Thus it appears below their LODs (Figure 1). Throughout both nights, DMS that the slight decreasing trend in the NO3 mixing ratios mixing ratios exhibited increasing trends while oxidant after 01:00 resulted primarily from the increasing trend in 0 concentrations remained low. The average values of kDMS/ the N2O5 loss rate. 0 kVOCs were 0.970 ± 0.004 and 0.948 ± 0.014 on the two 3.2.3. Polluted Continental Outflow on 16 July nights respectively, indicating that reaction of NO3 with [25] Polluted continental air masses characterized by DMS was the dominant gas-phase mechanism for removal average mixing ratios of toluene, ethyne and o-xylene of of NO3. 0.36 ± 0.09, 0.67 ± 0.19 and 0.040 ± 0.009 ppbv respec- 3.2.2. Polluted Continental Outflow on 11 July tively impacted AI on 16 July. For most of this night these [22] Trace gas measurements and backward trajectories anthropogenic marker species had decreasing trends in their on 11 July (Figure 4c) suggest that AI was impacted by mixing ratios (Figure 6a), suggesting a reduced influence of polluted continental air masses characterized by average pollution sources at AI after dark. The 02:00 backward mixing ratios of toluene, ethyne and o-xylene of 0.39 ± trajectory (Figure 4d) showed that the air mass originated 0.13, 0.69 ± 0.21, and 0.057 ± 0.012 ppbv respectively over central New York State during the night of 15 July and (Figure 5a). These levels were 2–5-fold higher compared to was lofted to 1 km altitude over northern Massachusetts the nights of 13 and 14 July. In addition, the average mixing during the day on 16 July. By sunset, the air mass appeared ratios of NO2 and NO3 were elevated by more than an order to be traveling several hundred meters above ground level of magnitude to 12 ± 6 ppbv and 7 ± 3 pptv respectively. and may have had minimal contact with the surface layer Backward trajectories indicate that during daytime on upwind of AI during the night. July 11 the air mass descended to the southeast of AI en [26] Opposite trends in the biogenic reactivity and NO3 route from over northeastern Maine and then circled back to concentrations were observed at both the beginning and end AI from the south (Figure 4c). The air mass appeared to of this night (Figures 6b and 6c). Between 21:00 and 22:00, traverse the Boston metropolitan area at approximately the biogenic reactivity decreased from 6 ± 3 Â 10À3 to 1.5 ± À3 À1 20:00 and was subsequently transported to AI in <6 h. 0.7 Â 10 s while NO3 mixing ratios increased from 4 to The mixing ratios for most of the anthropogenic tracers 21 pptv. These trends corresponded with decreases in increased throughout the night, corresponding to continuous isoprene and DMS mixing ratios from 0.280 to 0.062 ppbv continental outflow of urban pollutants which seemingly and 0.044 to 0.019 ppbv, respectively. From 03:00 to 04:00 fumigated the MBL surrounding AI. the NO3 mixing ratio decreased from 19 to 4 pptv as the À3 [23] The calculated reactivities of biogenic VOCs closely biogenic reactivity increased from 2.3 ± 0.6 Â 10 to 8.1 ± À3 À1 tracked the measured mixing ratios of NO3 around the 1.9 Â 10 s . Increases in the mixing ratios of a-pinene, midnight hours (Figures 5b and 5c). Between 23:00 and b-pinene and camphene (from their LODs to 8, 45 and 01:00, NO3 mixing ratios increased by about a factor of six 44 pptv, respectively) comprised the dominant contribution while biogenic reactivity decreased 8-fold. The loss rate to the biogenic reactivity. In this case a concurrent decrease of isoprene with NO3 was >20 times faster than with O3 in the NO3 production rate, which was 9of19 D21302 AMBROSE ET AL.: NIGHTTIME NITRATE RADICAL CHEMISTRY D21302 Figure 5. (a) Measured mixing ratios of ethyne, toluene, o-xylene, and CO (averaged hourly) for the night of 11 July. (b) Contributions to the total NO3 loss efficiency. For the nighttime hours between 21:00 0 0 0 on 8 July and 05:00 on 28 July, the average uncertainties in the values of kindirect, kVOCs, kbiogenics, and 0 khom were 80, 30, 40, and 40%, respectively. (c) Measured mixing ratios of NO2,NO3, and O3 and calculated mixing ratios of N2O5. The average uncertainty in the calculated mixing ratios of N2O5 was 60%. Uncertainties in the measured NO2,NO3, and O3 mixing ratios are described in Table 1. NO3 measurements below the LOD (Table 1) were set to 0.5 Â LOD. 10 of 19 D21302 AMBROSE ET AL.: NIGHTTIME NITRATE RADICAL CHEMISTRY D21302 Table 3. Comparison Between Rates of NOx Loss and HNO3 Production From NO3 and N2O5 Mechanisms for Selected Times During the ICARTT Campaign a,b Lj(NOx) j = indirect j = het c d a Date j = VOCs f=0 f =2.9 rhet =0.2 rhet =0.4 rhet =3.1 PVOCs(HNO3) 8–28 Julye 0.12 0.36 0.12 0.07 0.14 0.38 0.04 12 July (01:00) 0.08 0.50 0.17 0.10 0.20 0.53 0.03 12 July (05:00) 0.13 1.07 0.37 0.21 0.43 1.13 0.02 16–17 Julyf 0.19 0.85 0.29 0.17 0.34 0.89 0.06 17 Julyg 0.10 0.11 0.04 0.02 0.04 0.12 0.01 26 July (03:00) 0.65 0.18 0.06 0.04 0.07 0.19 0.01 aUnits are ppbv hrÀ1. b NOx loss rate via sink j; For N2O5 mechanisms, L(NOx)=P(HNO3). c Lindirect(NOx) was calculated for k(H2O(g)+N2O5) reduced by factor f (see Section 3.4 for details). d 0 0 Lhet(NOx) is expressed as a function of the ratio, rhet,ofkhet to kindirect (see Section 3.4 for details); rhet = 0.2, 0.4 correspond with f =0;rhet =3.1 corresponds with f = 2.9; Total rates are obtained by summing all j components. eAverage values for the nighttime hours (21:00–05:00). fAverage values for the period from 22:00 on July 16 to 03:00 on July 17. gAverage values for the period from 04:00 to 05:00. average roughly two times larger than the corresponding emissions from the urban coastal corridor stretching from 0 values of kVOCs, while the calculated N2O5 mixing ratios Boston to southern Maine (Figure 4e). Winds were signif- were approximately an order of magnitude greater than the icantly weaker on 25 July (1.3 ± 0.5 m sÀ1) compared to À1 measured NO3 mixing ratios. The average contribution of 11 July (5.7 ± 0.8 m s ), resulting in longer transport times N2O5 hydrolysis to gas-phase NOx removal was 82 ± 7%. to AI on 25 July. The relatively stagnant conditions during The average contributions of gas-phase and heterogeneous the night of 25 July likely led to little influx of air masses mechanisms to NOx removal and HNO3 production are with higher O3 levels, and O3 titration continued after compared in Table 3 for the periods from 22:00 to 03:00 01:00, which prevented the NO3 production rate from and from 04:00 to 05:00. The general decreasing trend in reaching even half the average value for the night of the measured NO3 mixing ratios appeared to result mostly July 11. Relatively weak production, coupled with extraor- from decreasing rates of production as the total gas-phase dinary enhancements of NO3 reactants, particularly mono- loss rate was relatively constant between 22:00 and 03:00 terpenes, suppressed NO3 concentrations to below the LOD and exceeded P(NO3) during the interval 23:00 to 04:00. by 04:00. Conversely, higher levels of monoterpenes on this Because the NO2 mixing ratios remained constant, the NO3 night than on both 11 and 16 July likely resulted in part trend closely followed that of O3 (Figure 6c). from reduced production of NO3. 3.2.4. Polluted and Biogenic Continental Outflow on [30] At 03:00 the rate of gas-phase NOx removal reached July 25 a maximum of 0.83 ± 0.21 ppbv hrÀ1, which was 75% of [28] Trace gas measurements during the night of 25 July the average removal rate after 03:00 on 12 July. However, indicate that AI experienced relatively clean conditions after 03:00 on 26 July the removal rate was apparently prior to 01:00, similar to those observed on the nights of dominated by the reactions of NO3 with monoterpenes and 13 and 14 July. This is supported by the low mixing ratios would have decreased considerably due to the slow rate of of anthropogenic tracers such as CO (<120 ppbv), toluene NO3 production. For instance, if the NO3 mixing ratio was (<0.050 ppbv), ethyne (<0.150 ppbv), o-xylene (